Undergraduate Texts in Physics

Øyvind Grøn

Introduction to
Einstein’s Theory
of Relativity
From Newton’s Attractive Gravity
to the Repulsive Gravity of Vacuum
Energy
Second Edition
Undergraduate Texts in Physics

Series Editors
Kurt H. Becker, NYU Polytechnic School of Engineering, Brooklyn, NY, USA
Jean-Marc Di Meglio, Matière et Systèmes Complexes, Université Paris Diderot,
Bâtiment Condorcet, Paris, France
Sadri D. Hassani, Department of Physics, Loomis Laboratory, University of Illinois
at Urbana-Champaign, Urbana, IL, USA
Morten Hjorth-Jensen, Department of Physics, Blindern, University of Oslo,
Oslo, Norway
Michael Inglis, Patchogue, NY, USA
Bill Munro, NTT Basic Research Laboratories, Optical Science Laboratories,
Atsugi, Kanagawa, Japan
Susan Scott, Department of Quantum Science, Australian National University,
Acton, ACT, Australia
Martin Stutzmann, Walter Schottky Institute, Technical University of Munich,
Garching, Bayern, Germany
Undergraduate Texts in Physics (UTP) publishes authoritative texts covering topics
encountered in a physics undergraduate syllabus. Each title in the series is suitable
as an adopted text for undergraduate courses, typically containing practice
problems, worked examples, chapter summaries, and suggestions for further
reading. UTP titles should provide an exceptionally clear and concise treatment of a
subject at undergraduate level, usually based on a successful lecture course. Core
and elective subjects are considered for inclusion in UTP.
UTP books will be ideal candidates for course adoption, providing lecturers with
a firm basis for development of lecture series, and students with an essential
reference for their studies and beyond.

More information about this series at http://www.springer.com/series/15593

Øyvind Grøn

Introduction to Einstein’s
Theory of Relativity
From Newton’s Attractive Gravity
to the Repulsive Gravity of Vacuum Energy
Second Edition

123
Øyvind Grøn
OsloMet—Oslo Metropolitan University
Oslo, Norway

ISSN 2510-411X ISSN 2510-4128 (electronic)

Undergraduate Texts in Physics
ISBN 978-3-030-43861-6 ISBN 978-3-030-43862-3 (eBook)
https://doi.org/10.1007/978-3-030-43862-3
1st edition: © Springer Science+Business Media, LLC 2009
2nd edition: © Springer Nature Switzerland AG 2020
This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part
of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission
or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar
methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publication does not imply, even in the absence of a specific statement, that such names are exempt from
the relevant protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this
book are believed to be true and accurate at the date of publication. Neither the publisher nor the
authors or the editors give a warranty, expressed or implied, with respect to the material contained
herein or for any errors or omissions that may have been made. The publisher remains neutral with regard
to jurisdictional claims in published maps and institutional affiliations.

This Springer imprint is published by the registered company Springer Nature Switzerland AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
Preface to the Second Edition

These notes are a transcript of lectures delivered by Øyvind Grøn during the spring
of 1997 at the University of Oslo. The manuscript has been revised in 2019. The
present version of this document is an extended and corrected version of a set of
Lecture Notes which were written down by S. Bard, Andreas O. Jaunsen, Frode
Hansen and Ragnvald J. Irgens using LATEX2. Sven E. Hjelmeland has made
many useful suggestions which have improved the text.
The manuscript has been revised in 2019. In this version, solutions to the
exercises have been included. Most of these have been provided by Håkon Enger.
I thank all my good helpers for enthusiastic work which was decisive for the
realization of the book.
I hope that these notes are useful to students of general relativity and look
forward to their comments accepting all feedback with thanks. The comments may
be sent to the author by e-mail to [email protected].

Oslo, Norway Øyvind Grøn

v
Preface to the First Edition

These notes are a transcript of lectures delivered by Øyvind Grøn during the spring
of 1997 at the University of Oslo.
The present version of this document is an extended and corrected version of a
set of Lecture Notes which were typesetted by S. Bard, Andreas O. Jaunsen, Frode
Hansen and Ragnvald J. Irgens using LATEX2. Svend E. Hjelmeland has made
many useful suggestions which have improved the text. I would also like to thank
Jon Magne Leinaas and Sigbjørn Hervik for contributing with problems and Gorm
Krogh Johnsen for help with finishing the manuscript. I also want to thank Prof.
Finn Ravndal for inspiring lectures on general relativity.
While we hope that these typeset notes are of benefit particularly to students of
general relativity and look forward to their comments, we welcome all interested
readers and accept all feedback with thanks.
All comments may be sent to the author by e-mail.

E-mail: [email protected] Øyvind Grøn

vii
Contents

1 Newton’s Theory of Gravitation . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.1 The Force Law of Gravitation . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 Newton’s Law of Gravitation in Local Form . . . . . . . . . . . . . . 4
1.3 Newtonian Incompressible Star . . . . . . . . . . . . . . . . . . . . . . . . 7
1.4 Tidal Forces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.5 The Principle of Equivalence . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1.6 The General Principle of Relativity . . . . . . . . . . . . . . . . . . . . . 17
1.7 The Covariance Principle . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
1.8 Mach’s Principle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
1.9 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2 The Special Theory of Relativity . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.1 Coordinate Systems and Minkowski Diagrams . . . . . . . . . . . . . 23
2.2 Synchronization of Clocks . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.3 The Doppler Effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.4 Relativistic Time Dilation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.5 The Relativity of Simultaneity . . . . . . . . . . . . . . . . . . . . . . . . . 30
2.6 The Lorentz Contraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
2.7 The Lorentz Transformation . . . . . . . . . . . . . . . . . . . . . . . . . . 34
2.8 Lorentz Invariant Interval . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
2.9 The Twin Paradox . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
2.10 Hyperbolic Motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
2.11 Energy and Mass . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
2.12 Relativistic Increase of Mass . . . . . . . . . . . . . . . . . . . . . . . . . . 45
2.13 Lorentz Transformation of Velocity, Momentum,
Energy and Force . . . . . . . . . . . . . . . . . . . . . . . ........... 47
2.14 Tachyons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ........... 50
2.15 Magnetism as a Relativistic Second-Order Effect ........... 51

ix
x Contents

Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
Reference . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
3 Vectors, Tensors and Forms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
3.1 Vectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
3.1.1 Four-Vectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
3.1.2 Tangent Vector Fields and Coordinate Vectors . . . . . . . 62
3.1.3 Coordinate Transformations . . . . . . . . . . . . . . . . . . . . 65
3.1.4 Structure Coefficients . . . . . . . . . . . . . . . . . . . . . . . . . 68
3.2 Tensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
3.2.1 Transformation of Tensor Components . . . . . . . . . . . . 71
3.2.2 Transformation of Basis One-Forms . . . . . . . . . . . . . . 71
3.2.3 The Metric Tensor . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
3.3 The Causal Structure of Spacetime . . . . . . . . . . . . . . . . . . . . . . 76
3.4 Forms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
3.4.1 The Volume Form . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
3.4.2 Dual Forms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
4 Accelerated Reference Frames . . . . . . . . . . . . . . . . . . . . . . ...... 89
4.1 The Spatial Metric Tensor . . . . . . . . . . . . . . . . . . . . . . ...... 89
4.2 Einstein Synchronization of Clocks in a Rotating
Reference Frame . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ...... 92
4.3 Angular Acceleration in the Rotating Frame . . . . . . . . . ...... 95
4.4 Gravitational Time Dilation . . . . . . . . . . . . . . . . . . . . . ...... 98
4.5 Path of Photons Emitted from the Axis in a Rotating
Reference Frame . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ...... 99
4.6 The Sagnac Effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . ...... 99
4.7 Non-integrability of a Simultaneity Curve in a Rotating
Frame . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
4.8 Orthonormal Basis Field in a Rotating Frame . . . . . . . . . . . . . . 102
4.9 Uniformly Accelerated Reference Frame . . . . . . . . . . . . . . . . . 105
4.10 The Projection Tensor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
5 Covariant Differentiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
5.1 Differentiation of Forms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
5.1.1 Exterior Differentiation . . . . . . . . . . . . . . . . . . . . . . . . 119
5.1.2 Covariant Derivative . . . . . . . . . . . . . . . . . . . . . . . . . . 122
5.2 The Christoffel Symbols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
5.3 Geodesic Curves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
5.4 The Covariant Euler–Lagrange Equations . . . . . . . . . . . . . . . . . 127
5.5 Application of the Lagrange Formalism to Free Particles . . . . . . 129
5.5.1 Equation of Motion from Lagrange’s Equations . . . . . . 129
5.5.2 Geodesic World Lines in Spacetime . . . . . . . . . . . . . . 133
Contents xi

5.5.3 Acceleration of Gravity . . . . . . . . . . . . . . . . . . . . . . . . 135

5.5.4 Gravitational Shift of Wavelength . . . . . . . . . . . . . . . . 138
5.6 Connection Coefficients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
5.6.1 Structure Coefficients . . . . . . . . . . . . . . . . . . . . . . . . . 143
5.7 Covariant Differentiation of Vectors, Forms and Tensors . . . . . . 144
5.7.1 Covariant Differentiation of Vectors . . . . . . . . . . . . . . 144
5.7.2 Covariant Differentiation of Forms . . . . . . . . . . . . . . . 145
5.7.3 Covariant Differentiation of Tensors of Arbitrary
Rank . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146
5.8 The Cartan Connection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
5.9 Covariant Decomposition of a Velocity Field . . . . . . . . . . . . . . 151
5.9.1 Newtonian 3-Velocity . . . . . . . . . . . . . . . . . . . . . . . . . 151
5.9.2 Relativistic 4-Velocity . . . . . . . . . . . . . . . . . . . . . . . . 153
5.10 Killing Vectors and Symmetries . . . . . . . . . . . . . . . . . . . . . . . 155
5.11 Covariant Expressions for Gradient, Divergence, Curl,
Laplacian and D’Alembert’s Wave Operator . . . . . . . . . . . . . . . 157
5.12 Electromagnetism in Form Language . . . . . . . . . . . . . . . . . . . . 163
Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
6 Curvature . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
6.1 The Riemann Curvature Tensor . . . . . . . . . . . . . . . . . . . . . . . . 173
6.2 Differential Geometry of Surfaces . . . . . . . . . . . . . . . . . . . . . . 179
6.2.1 Surface Curvature Using the Cartan Formalism . . . . . . 183
6.3 The Ricci Identity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184
6.4 Bianchi’s 1. Identity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 185
6.5 Bianchi’s 2. Identity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186
6.6 Torsion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
6.7 The Equation of Geodesic Deviation . . . . . . . . . . . . . . . . . . . . 188
6.8 Tidal Acceleration and Spacetime Curvature . . . . . . . . . . . . . . . 190
6.9 The Newtonian Tidal Tensor . . . . . . . . . . . . . . . . . . . . . . . . . . 191
6.10 The Tidal and Non-tidal Components of a Gravitational
Field . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192
Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
7 Einstein’s Field Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197
7.1 Newtonian Fluid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197
7.2 Perfect Fluids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
7.2.1 Lorentz Invariant Vacuum Energy—LIVE . . . . . . . . . . 200
7.2.2 Energy–Momentum Tensor of an Electromagnetic
Field . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201
7.3 Einstein’s Curvature Tensor . . . . . . . . . . . . . . . . . . . . . . . . . . . 201
7.4 Einstein’s Field Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202
7.5 The “Geodesic Postulate” as a Consequence of the Field
Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204
xii Contents

7.6 Einstein’s Field Equations Deduced from a Variational

Principle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 206
Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 210
8 Schwarzschild Spacetime . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211
8.1 Schwarzschild’s Exterior Solution . . . . . . . . . . . . . . . . . . . . . . 211
8.2 Radial Free Fall in Schwarzschild Spacetime . . . . . . . . . . . . . . 217
8.3 Light Cones in Schwarzschild Spacetime . . . . . . . . . . . . . . . . . 218
8.4 Analytical Extension of the Curvature Coordinates . . . . . . . . . . 222
8.5 Embedding of the Schwarzschild Metric . . . . . . . . . . . . . . . . . . 225
8.6 The Shapiro Experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226
8.7 Particle Trajectories in Schwarzschild 3-Space . . . . . . . . . . . . . 228
8.7.1 Motion in the Equatorial Plane . . . . . . . . . . . . . . . . . . 229
8.8 Classical Tests of Einstein’s General Theory of Relativity . . . . . 232
8.8.1 The Hafele–Keating Experiment . . . . . . . . . . . . . . . . . 232
8.8.2 Mercury’s Perihelion Precession . . . . . . . . . . . . . . . . . 233
8.8.3 Deflection of Light . . . . . . . . . . . . . . . . . . . . . . . . . . . 236
8.9 The Reissner–Nordström Spacetime . . . . . . . . . . . . . . . . . . . . . 238
Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242
9 The Linear Field Approximation and Gravitational Waves . . . . . . 243
9.1 The Linear Field Approximation . . . . . . . . . . . . . . . . . . . . . . . 243
9.2 Solutions of the Linearized Field Equations . . . . . . . . . . . . . . . 246
9.2.1 The Gravitational Potential of a Point Mass . . . . . . . . . 246
9.2.2 Spacetime Inside and Outside a Rotating Spherical
Shell . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 247
9.3 Inertial Dragging . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 250
9.4 Gravitoelectromagnetism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 251
9.5 Gravitational Waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253
9.5.1 What Sort of Gravitational Waves Is Predicted
by Einstein’s Theory? . . . . . . . . . . . . . . . . . . . . . . . . . 255
9.5.2 Polarization of the Gravitational Waves . . . . . . . . . . . . 256
9.6 The Effect of Gravitational Waves upon Matter . . . . . . . . . . . . 257
9.7 The LIGO-Detection of Gravitational Waves . . . . . . . . . . . . . . 260
9.7.1 Kepler’s Third Law and the Strain
of the Detector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 262
9.7.2 Newtonian Description of a Binary System . . . . . . . . . 265
9.7.3 Gravitational Radiation Emission . . . . . . . . . . . . . . . . . 266
9.7.4 The Chirp . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 267
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 269
Contents xiii

10 Black Holes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 271

10.1 “Surface Gravity”: Acceleration of Gravity
at the Horizon of a Black Hole . . . . . . . . . . . . . . . . . . . . . . . . 271
10.2 Hawking Radiation: Radiation from a Black Hole . . . . . . . . . . . 273
10.3 Rotating Black Holes: The Kerr Metric . . . . . . . . . . . . . . . . . . 275
10.3.1 Zero-Angular Momentum Observers . . . . . . . . . . . . . . 276
10.3.2 Does the Kerr Spacetime Have a Horizon? . . . . . . . . . 277
Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279
11 Sources of Gravitational Fields . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283
11.1 The Pressure Contribution to the Gravitational Mass
of a Static, Spherically Symmetric System . . . . . . . . . . . . . . . . 283
11.2 The Tolman–Oppenheimer–Volkoff Equation . . . . . . . . . . . . . . 285
11.3 An Exact Solution for Incompressible Stars—Schwarzschild’s
Interior Solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 287
11.4 The Israel Formalism for Describing Singular Mass Shells
in the General Theory of Relativity . . . . . . . . . . . . . . . . . . . . . 290
11.5 The Levi-Civita—Bertotti—Robinson Solution of Einstein’s
Field Equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 295
11.6 The Source of the Levi-Civita—Bertotti—Robinson
Spacetime . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 297
11.7 A Source of the Kerr–Newman Spacetime . . . . . . . . . . . . . . . . 299
11.8 Physical Interpretation of the Components
of the Energy–Momentum Tensor by Means
of the Eigenvalues of the Tensor . . . . . . . . . . . . . . . . . . . . . . . 302
11.9 The River of Space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 305
Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 309
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 309
12 Cosmology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 311
12.1 Co-moving Coordinate System . . . . . . . . . . . . . . . . . . . . . . . . 311
12.2 Curvature Isotropy—The Robertson–Walker Metric . . . . . . . . . 312
12.3 Cosmic Kinematics and Dynamics . . . . . . . . . . . . . . . . . . . . . . 314
12.3.1 The Hubble–Lemaître Law . . . . . . . . . . . . . . . . . . . . . 314
12.3.2 Cosmological Redshift of Light . . . . . . . . . . . . . . . . . . 315
12.3.3 Cosmic Fluids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 317
12.3.4 Isotropic and Homogeneous Universe Models . . . . . . . 318
12.3.5 Cosmic Redshift . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 323
12.3.6 Energy–Momentum Conservation . . . . . . . . . . . . . . . . 326
12.4 Some LFRW Cosmological Models . . . . . . . . . . . . . . . . . . . . . 330
12.4.1 Radiation-Dominated Universe Model . . . . . . . . . . . . . 330
12.4.2 Dust-Dominated Universe Model . . . . . . . . . . . . . . . . 331
12.4.3 Transition from Radiation-Dominated
to Matter-Dominated Universe . . . . . . . . . . . . . . . . . . 335
xiv Contents

12.4.4 The de Sitter Universe Models . . . . . . . . . . . . . . . . . . 336

12.4.5 The Friedmann–Lemaître Model . . . . . . . . . . . . . . . . . 337
12.4.6 Flat Universe with Dust and Phantom Energy . . . . . . . 348
12.5 Flat Anisotropic Universe Models . . . . . . . . . . . . . . . . . . . . . . 351
12.6 Inhomogeneous Universe Models . . . . . . . . . . . . . . . . . . . . . . . 355
12.6.1 Dust-Dominated Model . . . . . . . . . . . . . . . . . . . . . . . . 356
12.6.2 Inhomogeneous Universe Model with Dust
and LIVE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 357
12.7 The Horizon and Flatness Problems . . . . . . . . . . . . . . . . . . . . . 358
12.7.1 The Horizon Problem . . . . . . . . . . . . . . . . . . . . . . . . . 358
12.7.2 The Flatness Problem . . . . . . . . . . . . . . . . . . . . . . . . . 360
12.8 Inflationary Universe Models . . . . . . . . . . . . . . . . . . . . . . . . . . 361
12.8.1 Spontaneous Symmetry Breaking and the Higgs
Mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 361
12.8.2 Guth’s Inflationary Model . . . . . . . . . . . . . . . . . . . . . . 363
12.8.3 The Inflationary Models’ Answers to the Problems
of the Friedmann Models . . . . . . . . . . . . . . . . . . . . . . 364
12.8.4 Dynamics of the Inflationary Era . . . . . . . . . . . . . . . . . 366
12.8.5 Testing Observable Consequences of the Inflationary
Era . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 372
12.9 The Significance of Inertial Dragging for the Relativity
of Rotation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 377
12.9.1 The Cosmic Causal Mass in the Einstein-de Sitter
Universe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 378
12.9.2 The Cosmic Causal Mass in the Flat KCDM
Universe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 380
Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 382
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 390

Appendix: Kaluza–Klein Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 393

Solutions to the Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 409
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 511
List of Figures

Fig. 1.1 Newton’s law of gravitation . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

Fig. 1.2 Deduction of Newton’s law of gravitation in local form . . . . . . 4
Fig. 1.3 Solid angle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
Fig. 1.4 Mass shell in a star . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
Fig. 1.5 Tidal forces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
Fig. 1.6 Horizontal tidal force . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
Fig. 1.7 An elastic ring originally circular, falling freely
in the gravitational field of the Earth . . . . . . . . . . . . . . . . . . . . 12
Fig. 1.8 The Earth–Moon system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
Fig. 1.9 Tidal acceleration field . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
Fig. 1.10 A tidal force pendulum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
Fig. 2.1 World lines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
Fig. 2.2 Clock synchronization by the radar method . . . . . . . . . . . . . . . 26
Fig. 2.3 The Doppler effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
Fig. 2.4 Photon clock at rest . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
Fig. 2.5 Moving photon clock . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
Fig. 2.6 Simultaneous events . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
Fig. 2.7 The simultaneous events of Fig. 2.6 in another frame. . . . . . . . 31
Fig. 2.8 Light flash in a moving train . . . . . . . . . . . . . . . . . . . . . . . . . . 32
Fig. 2.9 Length contraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
Fig. 2.10 Space-like, light-like and time-like intervals . . . . . . . . . . . . . . . 38
Fig. 2.11 World line of an accelerating particle . . . . . . . . . . . . . . . . . . . . 39
Fig. 2.12 Twin paradox world lines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
Fig. 2.13 World line of particle with constant rest acceleration . . . . . . . . 43
Fig. 2.14 Light pulse in a box . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
Fig. 2.15 Tachyon paradox . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
Fig. 2.16 Current carrying wire seen from its own rest frame . . . . . . . . . 52
Fig. 2.17 Current carrying wire seen from the frame of a moving
charge . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 53
Fig. 2.18 Light cone due to Cherenkov radiation . . . . . . . . . . . . . . . . . .. 55
Fig. 3.1 Closed polygon (linearly dependent vectors) . . . . . . . . . . . . . .. 60

xv
xvi List of Figures

Fig. 3.2 No finite position vector in curved space . . . . . . . . . . . . . . . . . 63

Fig. 3.3 Vectors in tangent planes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
Fig. 3.4 Basis vectors in Cartesian- and plane polar coordinates . . . . . . 67
Fig. 3.5 Basis vectors in a skew angled coordinate system . . . . . . . . . . 73
Fig. 3.6 Covariant and contravariant components of a vector . . . . . . . . . 75
Fig. 3.7 Causal structure of spacetime . . . . . . . . . . . . . . . . . . . . . . . . . . 77
Fig. 4.1 Simultaneity in a rotating frame . . . . . . . . . . . . . . . . . . . . . . . . 92
Fig. 4.2 Discontinuity of simultaneity in rotating frame . . . . . . . . . . . . . 93
Fig. 4.3 Nonrotating disc with measuring rods . . . . . . . . . . . . . . . . . . . . 96
Fig. 4.4 Lorentz contacted measuring rods on a rotating disc. . . . . . . . . 97
Fig. 4.5 The Sagnac effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
Fig. 4.6 Discontinuous simultaneity surface in a rotating frame . . . . . . . 102
Fig. 4.7 Rigid rotation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
Fig. 4.8 Hyperbolic motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
Fig. 4.9 The velocity of a uniformly accelerated particle . . . . . . . . . . . . 108
Fig. 4.10 Simultaneity in a uniformly accelerated reference frame . . . . . . 109
Fig. 4.11 World lines and simultaneity lines of a uniformly
accelerated reference system . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
Fig. 5.1 Parallel transport . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
Fig. 5.2 Geodesic curve on a flat surface . . . . . . . . . . . . . . . . . . . . . . . . 126
Fig. 5.3 Geodesic curves on a sphere . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
Fig. 5.4 Neighboring geodesics in a Minkowski diagram . . . . . . . . . . . . 127
Fig. 5.5 Time like geodesics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133
Fig. 5.6 Projectiles in 3-space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
Fig. 5.7 The twin paradox . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
Fig. 5.8 Rotating coordinate system . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
Fig. 6.1 Parallel transport . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174
Fig. 6.2 Parallel transport of a vector around a triangle . . . . . . . . . . . . . 174
Fig. 6.3 Curvature and parallel transport . . . . . . . . . . . . . . . . . . . . . . . . 175
Fig. 6.4 Curl as area of a vectorial parallelogram . . . . . . . . . . . . . . . . . . 176
Fig. 6.5 Surface with tangent vectors and normal vector . . . . . . . . . . . . 180
Fig. 6.6 Geodesic deviation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
Fig. 6.7 Spacetime curvature and the tidal force pendulum . . . . . . . . . . 196
Fig. 8.1 Light cones in Schwarzschild spacetime with Schwarzschild
time. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
Fig. 8.2 Light cones in Schwarzschild spacetime with
Eddington-Finkelstein time . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
Fig. 8.3 Embedding of the extended Schwarzschild spacetime . . . . . . . . 226
Fig. 8.4 The Shapiro experiment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
Fig. 8.5 Newtonian centrifugal barrier . . . . . . . . . . . . . . . . . . . . . . . . . . 230
Fig. 8.6 Relativistic gravitational potential outside a spherical body . . . 231
Fig. 8.7 Deflection of light . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237
Fig. 9.1 Deformation of a ring of free particles caused by a
gravitational wave with + polarization . . . . . . . . . . . . . . . . . . . 259
List of Figures xvii

Fig. 9.2 Deformation of a ring of free particles caused

by a gravitational wave with x polarization . . . . . . . . . . . . . . . 260
Fig. 9.3 LIGO gravitational wave detector . . . . . . . . . . . . . . . . . . . . . . . 261
Fig. 9.4 LIGO-gravitational wave signal . . . . . . . . . . . . . . . . . . . . . . . . 262
Fig. 9.5 LIGO gravitational wave discovery signal . . . . . . . . . . . . . . . . 263
Fig. 10.1 Static border and horizon of a Kerr black . . . . . . . . . . . . . . . . . 279
Fig. 11.1 River of space in the Schwarzschild-de Sitter spacetime . . . . . . 308
Fig. 12.1 Cosmological redshift . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 316
Fig. 12.2 Expansion of a radiation dominated universe . . . . . . . . . . . . . . 331
Fig. 12.3 Expansion of matter dominated universe models . . . . . . . . . . . 334
Fig. 12.4 Hubble age . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 334
Fig. 12.5 The scale factor of the De Sitter universe models . . . . . . . . . . . 337
Fig. 12.6 The scale factor of the flat KCDM universe model . . . . . . . . . . 340
Fig. 12.7 The ratio of age and Hubble age of the flat KCDM
universe model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 340
Fig. 12.8 Age-redshift relation of the flat KCDM universe model . . . . . . 341
Fig. 12.9 The Hubble parameter of the flat KCDM universe model . . . . . 342
Fig. 12.10 The deceleration parameter of the flat KCDM universe
model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 343
Fig. 12.11 Point of time for deceleration–acceleration turnover . . . . . . . . . 344
Fig. 12.12 Cosmic redshift at the deceleration–acceleration turnover . . . . . 345
Fig. 12.13 Critical density as function of time . . . . . . . . . . . . . . . . . . . . . . 345
Fig. 12.14 The density parameter of vacuum energy as function
of time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 346
Fig. 12.15 The density of matter as function of time . . . . . . . . . . . . . . . . . 347
Fig. 12.16 The density parameter of matter as a function of time . . . . . . . 347
Fig. 12.17 Evolution of anisotropy in a LIVE-dominated
Bianchi-type I universe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 355
Fig. 12.18 Inflationary potentials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 362
Fig. 12.19 Temperature dependence of Higgs potential . . . . . . . . . . . . . . . 363
Fig. 12.20 Polarization of electromagnetic radiation. . . . . . . . . . . . . . . . . . 372
Fig. 12.21 Causal mass . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 381
List of Definitions

Definition 3.1.1 Four-velocity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60

Definition 3.1.2 Four-momentum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
Definition 3.1.3 Minkowski force . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
Definition 3.1.4 Four-acceleration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
Definition 3.1.5 Reference frame . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
Definition 3.1.6 Coordinate system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
Definition 3.1.7 Co-moving coordinate system . . . . . . . . . . . . . . . . . . . . . 64
Definition 3.1.8 Orthonormal basis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
Definition 3.1.9 Preliminary definition of coordinate basis vector . . . . . . 64
Definition 3.1.10 General definition of coordinate basis vectors . . . . . . . . . 66
Definition 3.1.11 Orthonormal basis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
Definition 3.1.12 Commutator of vectors . . . . . . . . . . . . . . . . . . . . . . . . . . 68
Definition 3.1.13 Structure coefficients . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
Definition 3.2.1 One-form basis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
Definition 3.2.2 Tensors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
Definition 3.2.3 Tensor product . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
Definition 3.2.4 The scalar product . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
Definition 3.2.5 The metric tensor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
Definition 3.2.6 Contravariant components of the metric tensor . . . . . . . . 74
Definition 3.4.1 Antisymmetric tensor . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
Definition 3.4.2 p-form. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
Definition 3.4.3 The wedge product . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
Definition 4.9.1 Born-rigid Motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
Definition 4.10.1 The Projection Tensor. . . . . . . . . . . . . . . . . . . . . . . . . . . 113
Definition 5.2.1 Christoffel symbols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
Definition 5.2.2 Covariant directional derivative . . . . . . . . . . . . . . . . . . . 124
Definition 5.2.3 Parallel transport . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124
Definition 5.3.1 Geodesic curves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
Definition 5.6.1 Koszul’s connection coefficients in an arbitrary
basis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
Definition 5.7.1 Covariant derivative of a vector field . . . . . . . . . . . . . . . 144

xix
xx List of Definitions

Definition 5.7.2 Covariant derivative of a vector component . . . . . . . . . . 144

Definition 5.7.3 Covariant directional derivative of a 1-form field . . . . . . 145
Definition 5.7.4 Covariant derivative of a 1-form . . . . . . . . . . . . . . . . . . . 146
Definition 5.7.5 Covariant derivative of a 1-form component. . . . . . . . . . 146
Definition 5.7.6 Covariant derivative of tensors . . . . . . . . . . . . . . . . . . . . 146
Definition 5.7.7 Covariant derivative of tensor components . . . . . . . . . . . 146
Definition 5.8.1 Exterior derivative of a basis vector . . . . . . . . . . . . . . . . 147
Definition 5.8.2 Connection forms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148
Definition 5.8.3 Scalar product between vector and 1-form . . . . . . . . . . . 148
Definition 5.10.1 Killing vectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
Definition 5.10.2 Invariant basis. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156
Definition 6.1.1 The Riemann curvature tensor . . . . . . . . . . . . . . . . . . . . 175
Definition 6.5.1 Contraction of a tensor component . . . . . . . . . . . . . . . . . 177
Definition 6.6.1 The torsion 2-form . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
Definition 6.9.1 Newtonian tidal tensor . . . . . . . . . . . . . . . . . . . . . . . . . . 191
Definition 8.1.1 Physical singularity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216
Definition 8.1.2 Coordinate singularity . . . . . . . . . . . . . . . . . . . . . . . . . . . 216
Definition 10.3.1 Horizon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 277
List of Examples

Example 1.1 Two particles falling towards each other . . . . . . . . . . . . . 3

Example 1.2 Tidal force on a system consisting of two particles . . . . . 11
Example 1.3 The tidal field on the Earth due to the Moon . . . . . . . . . . 12
Example 2.13.1 The Lever paradox . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
Example 3.1.1 Transformation between Cartesian- and plane polar
coordinates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
Example 3.1.2 Relativistic Doppler effect . . . . . . . . . . . . . . . . . . . . . . . . 67
Example 3.1.3 Structure coefficients in plane polar coordinates . . . . . . . . 69
Example 3.2.1 Tensor product of two vectors . . . . . . . . . . . . . . . . . . . . . 70
Example 3.2.2 A mixed tensor of rank 3 . . . . . . . . . . . . . . . . . . . . . . . . . 71
Example 3.2.3 Cartesian coordinates in a plane . . . . . . . . . . . . . . . . . . . . 73
Example 3.2.4 Plane polar coordinates . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
Example 3.2.5 Non-orthogonal basis-vectors . . . . . . . . . . . . . . . . . . . . . . 73
Example 3.2.6 Line-element in Cartesian coordinates . . . . . . . . . . . . . . . 75
Example 3.2.7 Line element in plane polar coordinates . . . . . . . . . . . . . . 75
Example 3.2.8 The four-velocity identity . . . . . . . . . . . . . . . . . . . . . . . . . 76
Example 3.4.1 Antisymmetric combinations. . . . . . . . . . . . . . . . . . . . . . . 79
Example 3.4.2 A 2-form in a 3-space . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
Example 3.4.3 Duals of basis forms in a spherical coordinate system
in Euclidean 3-space . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 82
Example 4.8.1 The acceleration of a velocity field representing
rigid rotation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
Example 4.9.1 Uniformly accelerated motion through the Milky
Way . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
Example 4.10.1 Covariant condition for uniformly accelerated
motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
Example 4.10.2 Spatial metric and the projection tensor . . . . . . . . . . . . . . 114
Example 5.1.1 Relationship between exterior derivative and curl . . . . . . . 120
Example 5.2.1 The Christoffel symbols in plane polar coordinates . . . . . 123
Example 5.5.1 Vertical free fall in a uniformly accelerated reference
frame . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131

xxi
xxii List of Examples

Example 5.5.2 How geodesics in spacetime can give parabolas

in space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
Example 5.5.3 The Twin Paradox . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
Example 5.5.4 Gravitational redshift or blueshift of light . . . . . . . . . . . . . 139
Example 5.6.1 The connection coefficients in a rotating reference
frame . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
Example 5.6.2 Acceleration in a non-rotating reference frame . . . . . . . . . 142
Example 5.6.3 Acceleration in a rotating reference frame . . . . . . . . . . . . 142
Example 5.8.1 Cartan-connection in an orthonormal basis field
in plane polar coordinates . . . . . . . . . . . . . . . . . . . . . . . . . 150
Example 5.10.1 Killing vectors of an Euclidean plane . . . . . . . . . . . . . . . . 155
Example 5.11.1 Differential operators in spherical coordinates . . . . . . . . . 162
Example 6.1.1 The Riemann curvature tensor of a spherical surface
calculated from Cartan’s structure equations . . . . . . . . . . . 178
Example 6.10.1 Non-tidal gravitational field . . . . . . . . . . . . . . . . . . . . . . . 195
Example 7.1.1 Energy–momentum tensor of a Newtonian fluid . . . . . . . . 198
Example 7.6.1 The energy—momentum tensor of an electric field
in a spherically symmetric spacetime . . . . . . . . . . . . . . . . 209
Example 11.1 Thin dust shell described by the Israel formalism . . . . . . . 291
Example 11.2 Lopez’s source of the Kerr–Newman metric . . . . . . . . . . . 303
Example 12.4.1 Lookback Time for Flat Dust-dominated Universe . . . . . . 335
Example 12.8.1 Polynomial Inflation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 375
List of Exercises

Exercise 1.1 A tidal force pendulum . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

Exercise 1.2 Newtonian potential for a spherically symmetric body. . . . 20
Exercise 1.3 Frictionless motion in a tunnel through the Earth . . . . . . . 20
Exercise 1.4 The Earth—Moon system . . . . . . . . . . . . . . . . . . . . . . . . . 21
Exercise 1.5 The Roche limit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
Exercise 2.1 Robb’s Lorentz invariant spacetime interval formula
(A. A. Robb, 1936) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
Exercise 2.2 The twin paradox . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
Exercise 2.3 Faster than the speed of light? . . . . . . . . . . . . . . . . . . . . . . 55
Exercise 2.4 Time dilation and Lorentz contraction . . . . . . . . . . . . . . . . 55
Exercise 2.5 Atmospheric mesons reaching the surface of the Earth . . . 56
Exercise 2.6 Relativistic Doppler shift . . . . . . . . . . . . . . . . . . . . . . . . . . 56
Exercise 2.7 The velocity of light in a moving medium. . . . . . . . . . . . . 57
Exercise 2.8 Cherenkov radiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
Exercise 2.9 Relativistic form of Newton’s 2. law . . . . . . . . . . . . . . . . . 57
Exercise 2.10 Lorentz transformation of electric and magnetic fields . . . . 57
Exercise 3.1 Four-vectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
Exercise 3.2 The tensor product . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
Exercise 3.3 Symmetric and antisymmetric tensors . . . . . . . . . . . . . . . . 86
Exercise 3.4 Contractions of tensors with different symmetries . . . . . . . 86
Exercise 3.5 Coordinate transformation in an Euclidean plane . . . . . . . . 87
Exercise 4.1 Relativistic rotating disc . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
Exercise 4.2 Uniformly accelerated system of reference . . . . . . . . . . . . . 116
Exercise 4.3 Uniformly accelerated space ship . . . . . . . . . . . . . . . . . . . . 117
Exercise 4.4 Light emitted from a point source in a gravitational
field . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
Exercise 4.5 Geometrical optics in a gravitational field . . . . . . . . . . . . . 118
Exercise 5.1 Dual forms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169
Exercise 5.2 Differential operators in spherical coordinates . . . . . . . . . . 170
Exercise 5.3 Spatial geodesics in a rotating frame of reference . . . . . . . 171

xxiii
xxiv List of Exercises

Exercise 5.4 Christoffel symbols in a uniformly accelerated reference

frame . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
Exercise 5.5 Relativistic vertical projectile motion . . . . . . . . . . . . . . . . . 172
Exercise 5.6 The geodesic equation and constants of motion . . . . . . . . . 172
Exercise 6.1 Parallel transport and curvature . . . . . . . . . . . . . . . . . . . . . 195
Exercise 6.2 Curvature of the simultaneity space in a rotating
reference frame . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195
Exercise 6.3 The tidal force pendulum and the curvature of space . . . . . 196
Exercise 7.1 Newtonian approximation of perfect fluid . . . . . . . . . . . . . 210
Exercise 7.2 The energy—momentum tensor of LIVE. . . . . . . . . . . . . . 210
Exercise 8.1 Non-relativistic Kepler motion . . . . . . . . . . . . . . . . . . . . . . 240
Exercise 8.2 The Schwarzschild solution in isotropic coordinates . . . . . 241
Exercise 8.3 Proper radial distance in the external Schwarzschild
space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241
Exercise 8.4 The Schwarzschild–de Sitter metric . . . . . . . . . . . . . . . . . . 241
Exercise 8.5 The perihelion precession of Mercury and the
cosmological constant . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242
Exercise 8.6 Relativistic time effects and GPS . . . . . . . . . . . . . . . . . . . . 242
Exercise 8.7 The photon sphere . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242
Exercise 10.1 A spaceship falling into a black hole . . . . . . . . . . . . . . . . . 279
Exercise 10.2 Kinematics in the Kerr-spacetime . . . . . . . . . . . . . . . . . . . 280
Exercise 10.3 A gravitomagnetic clock effect . . . . . . . . . . . . . . . . . . . . . . 281
Exercise 11.1 The Schwarzschild-de Sitter metric . . . . . . . . . . . . . . . . . . 309
Exercise 11.2 A spherical domain wall described by the Israel
formalism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 309
Exercise 12.1 Gravitational collapse . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 382
Exercise 12.2 The volume of a closed Robertson–Walker universe . . . . . 383
Exercise 12.3 Conformal time . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 383
Exercise 12.4 Lookback time and the age of the universe . . . . . . . . . . . . 383
Exercise 12.5 The LFRW universe models with a perfect fluid . . . . . . . . 384
Exercise 12.6 Age—density relation for a radiation-dominated
universe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 384
Exercise 12.7 Redshift–luminosity relation for matter-dominated
universe: Mattig’s formula . . . . . . . . . . . . . . . . . . . . . . . . . 385
Exercise 12.8 Newtonian approximation with vacuum energy . . . . . . . . . 385
Exercise 12.9 Universe models with constant deceleration parameter . . . 385
Exercise 12.10 Density parameters as functions of the redshift . . . . . . . . . 386
Exercise 12.11 FRW universe with radiation and matter . . . . . . . . . . . . . . 386
Exercise 12.12 Event horizons in de Sitter universe models . . . . . . . . . . . 386
Exercise 12.13 Flat universe model with radiation and LIVE . . . . . . . . . . 386
Exercise 12.14 De Sitter spacetime . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 387
Exercise 12.15 The Milne Universe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 388
Exercise 12.16 Natural Inflation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 389
Chapter 1
Newton’s Theory of Gravitation

Abstract In this chapter we first deduce Newton’s law of gravitation in its local
form as a preparation for comparing Newton’s and Einstein’s theories, including a
discussion of tidal forces. Then we give a presentation of the main conceptual foun-
dation of the general theory of relativity, emphasizing the principle of equivalence
and the principle of relativity.

In Newton’s theory there is an absolute space and time. They are independent of the
content in the universe. Newton wrote: “Absolute space, in its own nature, without
regard to anything external, remains always similar and immovable.” And further:
“Absolute, true and mathematical time, of itself, and from its own nature flows
equably without regard to anything external.” Thus, every object has an absolute state
of motion in absolute space. Hence an object must be either in a state of absolute
rest or moving at some absolute speed.
Galileo, however, argued for a relativity of rectilinear motion with constant veloc-
ity as least with respect to mechanical phenomena. This principle is obeyed by
Newton’s theory of gravity.
In Newton’s theory an inertial frame is defined as a reference frame moving along
a straight line with constant velocity.
The fundamental laws of Newton’s theory of gravitation are Newton’s three laws
plus the law of gravitation (see below). With reference to an inertial frame Newton’s
three laws take the form:
1. If a body is not acted upon by forces, or if the sum of the forces acting upon a
body is zero, the body is either at rest of moves along a straight line with constant
velocity.
2. The sum of the forces acting upon a body is equal to its (inertial) mass times its
acceleration,


F = m I a . (1.1)

3. If a body A acts upon a body B with a force, then B acts back on A with an
equally large and oppositely directed force.
© Springer Nature Switzerland AG 2020 1
Ø. Grøn, Introduction to Einstein’s Theory of Relativity,
Undergraduate Texts in Physics, https://doi.org/10.1007/978-3-030-43862-3_1
2 1 Newton’s Theory of Gravitation

In a non-inertial frame with acceleration a f one will experience an “artificial

acceleration of gravity” g = − a f , and Newton’s 2. law takes the modified form
  
F = m I a + a f = m I (
a − g). (1.2)

If no forces act on a body, it is said to be freely falling. A freely falling body in a

non-inertial frame will have an acceleration a = g.

1.1 The Force Law of Gravitation

Consider two particles with masses M and m, respectively. They are at a distance
r from each other and act on each other by a gravitational force F. The situation is
shown in Fig. 1.1.
According to Newton’s theory the gravitational force between the particles is
given by

M r
F = −mG 2 er , e = , r = |
r |. (1.3)
r r
Let V be the potential energy of m (see Fig. 1.1). Then

∂V
F = −∇V (
r ), Fi = − i . (1.4)
∂x

For a spherical mass distribution V ( r ) = − mG(M/r) with zero potential

infinitely far from the centre of M. Newton’s law of gravitation is valid for small
velocities, i.e. velocities much smaller than the velocity of light and weak fields.
Weak fields are fields in which the gravitational potential energy of a test particle is
very small compared to its rest mass energy. (Note that here one is interested only in
the absolute values of the above quantities and not their sign).

Fig. 1.1 Newton’s law of gravitation. Newton’s law of gravitation states that the force between two
spherical bodies is attractive, acts along the line joining the centres of the bodies, is proportional
to the product of the masses and inversely proportional to the distance between the centres of the
masses
1.1 The Force Law of Gravitation 3

M GM
mG  mc2 ⇒ r  2 . (1.5)
r c

The Schwarzschild radius for an object with mass M is R S = 2G M/c2 . Hence,

far outside the Schwarzschild radius the gravitational field is weak. To get a feeling
for the magnitudes, you may insert the mass of the earth. Then you find that the
Schwarzschild mass of the Earth is 9 mm. Comparing with the radius of the Earth,
which is R E ≈ 6400 km, we may conclude that the gravitational field is weak on the
surface of the Earth. Similarly the Schwarzschild radius of the Sun is 3 km and the
Earth is about 150 million km from the Sun. Hence the gravitational field of the Sun
is very weak in most parts of the solar system. This explains, in part, the success of
Newtonian gravity for describing the motion of bodies in the gravitational field of
the Earth and the Sun.

Example 1.1 (Two particles falling towards each other) Two point particles with
masses m 1 and m 2 are instantaneously at rest at a distance r0 from each other in
empty space, with no other forces present than the gravitational force between them.

How long time will they fall before they collide?

Newton’s 2. law is valid with reference to an inertial frame. Hence we start by
introducing a coordinate system fixed with respect to the mass centre of the particles.
In this system particles 1 and 2 have coordinates r1 and r2 , respectively. The equations
of motion of the two particles are
m2 m1
r̈1 = G , r̈2 = −G . (1.6)
(r2 − r1 ) 2
(r2 − r1 )2

where a dot denotes differentiation with respect to time. Subtracting the equations
and introducing the distance r = r2 − r2 between the particles as a new coordinate,
we get the differential equation

r 2 r̈ + G(m 1 + m 2 ) = 0. (1.7)

Writing this as

G(m 1 + m 2 )
ṙ r̈ = − ṙ (1.8)
r2
and integrating with the boundary condition that ṙ = 0 for r = r0 , we get the energy
conservation equation


1 1
ṙ = 2G(m 1 + m 2 )
2
− . (1.9)
r r0

Hence, the falling time is given by the integral

4 1 Newton’s Theory of Gravitation

 r0 
r0 r
t= dr. (1.10)
2G(m 1 + m 2 ) r0 − r
0

Performing the integration gives

3/2
π r0
t= √ . (1.11)
2 2G(m 1 + m 2 )

Assume that the particles have a mass equal to 1 kg and starts 1 m from each
other. Then the falling time is 26.5 h. This illustrates that gravity is indeed a very
weak force.

1.2 Newton’s Law of Gravitation in Local Form

Consider a gravitational field due to a mass distribution. Let P be a point in the field
(see Fig. 1.2) with position vector r = x i ei , and let the gravitating mass element be

at r = x i ei . Newton’s law of gravitation for a continuous distribution of mass is

  r − r 3
F = −Gm ρ r d r = −∇V (
r ). (1.12)
|
r − r |3

Note that the ∇ operator acts on the unprimed coordinates, only.

Let us consider Eq. (1.12) term by term

Fig. 1.2 Deduction

of Newton’s law of
gravitation in local form. The
dice is a mass element, and P
is the field point
1.2 Newton’s Law of Gravitation in Local Form 5

1 ∂ 1 ∂  j  −1/2
∇ = ei 
1/2 = ei x − x j x j − x j
|
r − r |
∂ xi  ∂ xi
x j − x j x j − x j
 
1  j  −3/2  k  ∂ xk
= ei − x − x j x j − x j 2 x − x k
2 ∂ xi
 k 
k i
 i 
x − x δk x − x i ei
= −ei  
3/2 = −  
3/2
x j − x j x j − x j x j − x j x j − x j
r − r
=− (1.13)
|
r − r |3

Equations (1.12) and (1.13) imply

  
ρ r 3
r ) = −Gm
V ( dr. (1.14)
|
r − r |

Hence, the gravitational potential at the point P is

  
V (
r) ρ r 3
φ(
r) ≡ = V (
r ) = −G dr. (1.15)
m |
r − r |

It follows that

  r − r 3
∇φ(
r ) = −G ρ r dr, (1.16)
|
r − r |3

and

  r − r 3
∇ 2 φ(
r ) = −G ρ r ∇ · dr. (1.17)
|
r − r |3

We now calculate the divergence in the integrand

r − r ∇ · r   1 3  
∇· = + r − r · ∇ = − r − r
|
r − r |3 |
r − r |3 |
r − r |
3
|
r − r |
3


3 r − r 3 3
· = − = 0, r = r . (1.18)
|
r − r |
5
|
r − r |
3
|
r − r |3

The condition r = r means that the field point is outside the mass distribution.
Hence, the Newtonian potential at a point in a gravitational field outside a mass
distribution satisfies the Laplace equation

∇ 2 φ = 0. (1.19)
6 1 Newton’s Theory of Gravitation

We shall now generalize this to the case where the field point may be inside a mass
distribution. It will then be useful to utilize the Dirac delta function. This function
has the following properties.
 
δ r − r = 0, for r = r , (1.20)
 
  1 if r = r is contained in the integration region.
δ r − r d3r = (1.21)
0 if r = r is not contained in the integration region.

   
f r δ r − r d3r = f (
r ). (1.22)

A calculation of the integral (1.17) which is valid also in the case where the field
point is inside the mass distribution is obtained by utilizing Gauss integral theorem,
 
 3r =
∇ · Ad A · d S,
 (1.23)
V S

where S is the boundary surface of the volume V. We shall also need the definition
of a solid angle,

ds⊥
d ≡ , (1.24)
|
r − r |2

where ds⊥ is the projection of the area normal to the line of sight. It is represented
by absolute value of the component of ds along the line of sight, where ds is the
normal vector of the surface element of the mass distribution subtending the solid
angle d at the field point P (Fig. 1.3).

Fig. 1.3 Solid angle. Solid angle d is defined such that the surface of a sphere subtends the angle
4π at the centre
1.2 Newton’s Law of Gravitation in Local Form 7

Applying the Gauss integral theorem we have

   
r − r 3 r − r ds⊥
∇· dr = · ds = = dΩ (1.25)
|
r − r |3 |
r − r |3 |
r − r |2
V S S S

Hence
 
r − r 3 4π if P is inside the mass distribution
∇· dr = . (1.26)
|
r − r | 3 0 if P is outside the mass distribution
V

This may be written in terms of the Dirac delta function as


r − r  
∇· d3r = 4π δ r − r . (1.27)
|
r − r |3
V

We now have
 

  r − r 3  
r ) = −G ρ r ∇ · ρ r 4π δ r − r d3r
←
∇ φ(
2
dr =G
|
r − r |3
= 4π Gρ(
r ), (1.28)

showing that the Newtonian gravitational potential obeys the Poisson equation. This
means that Newton’s gravitational theory can be expressed in the following way:
• Mass generates a gravitational potential according to

∇ 2 φ = 4π Gρ. (1.29)

• The gravitational potential generates acceleration of gravity g according to

g = − ∇φ. (1.30)

1.3 Newtonian Incompressible Star

We shall apply Eqs. (1.29) and (1.30) to calculate the gravitational field of a New-
tonian incompressible star. Let the gravitational potential be φ(r ). In the spherically
symmetric case Eq. (1.29) then takes the form
8 1 Newton’s Theory of Gravitation
 
1 d 2 dφ
r = 4π Gρ. (1.31)
r 2 dr dr

Assuming that ρ = constant and integrating gives

dφ 4π
r2 = Gρr 3 + K 1 = M(r ) + K 1 . (1.32)
dr 3
where M(r) is the mass inside a sphere with radius r. According to Eq. (1.30) the
gravitational acceleration is

dφ
g = −∇φ = − er , (1.33)
dr
or
M(r ) K1 4π K1
g= + 2 = Gρr + 2 . (1.34)
r2 r 3 r
Finite g in r = 0 gives K 1 = 0

4π dφ 4π
g= Gρr, = Gρr. (1.35)
3 dr 3
Assume that the mass distribution has a radius R. A new integration then leads to

2π
φ= Gρr 2 + K 2 . (1.56)
3
Demanding continuous potential at r = R gives.

2π M(R) 4π
Gρ R 2 + K 2 = = − Gρ R 2 . (1.37)
3 R 3
Hence

K 2 = −2π Gρ R 2 . (1.38)

Thus the potential inside the mass distribution is

2π
φ= Gρ(r 2 − 3R 2 ). (1.39)
3
The star is in hydrostatic equilibrium that is the pressure forces are in equilibrium
with the gravitational forces.
Consider a mass element dm = ρdV = ρdAdr, in the shell drawn in Fig. 1.4.
The pressure force on the mass element is dF = dAd p, and the gravitational
force is
1.3 Newtonian Incompressible Star 9

Fig. 1.4 Mass shell in a star. A shell with thickness dr is affected by both gravitational and pressure
forces

Gm(r )
dG = gdm = dm, (1.40)
r2

where m(r ) is the mass inside the shell. With constant density m(r ) = (4π/3)ρr 3 .
Hence
4π
dG = gdm = Gρ 2 r dAdr. (1.41)
3
Equilibrium, dF = −dG, demands that

4π
dp = − Gρ 2 r dr. (1.42)
3
Integrating gives

2π G 2 2
p = K3 − ρ r . (1.43)
3
Vanishing pressure at the surface of the mass distribution, p(R) = 0, gives the
value of the constant of integration
10 1 Newton’s Theory of Gravitation

2π G 2 2
K3 = ρ R (1.44)
3
which leads to
2π G 2 2
p(r ) = ρ (R − r 2 ). (1.45)
3
No matter how massive the star is, it is possible for the pressure forces to keep
the equilibrium with gravity. In Newtonian theory, gravitational collapse is not a
necessity.

1.4 Tidal Forces

Tidal forces are the difference of gravitational force on two neighbouring particles
in a gravitational field. The tidal force is due to the inhomogeneity of a gravitational
field.
Figure 1.5 shows two point masses, each with mass m, with a separation vector ς
 respectively, where |ς|
and position vectors r and r + ς,  << |
r |. The gravitational
forces on the mass points are F( r ) and F(
 r + ς). By means of a Taylor expansion
to the lowest order in |ς| we get for the i-component of the tidal force
 
∂ Fi
f i = Fi (
r + ς) − Fi (
r) = ς j . (1.46)
∂x j r

The corresponding vector equation is

f = (ς · ∇)r F.
 (1.47)

Using that

Fig. 1.5 Tidal forces. The

separation vector ς between
two mass points 1 and 2
acted upon by gravitational
forces F1 and F2
1.4 Tidal Forces 11

F = −m∇φ, (1.48)

The tidal force may be expressed in terms of the gravitational potential according
to

f = − m(ς · ∇)r ∇φ. (1.49)

It follows that in a local Cartesian coordinate system the i-component of the

relative acceleration of the particles is
 2 
d 2 ςi ∂ φ
=− ς j. (1.50)
dt 2 ∂ x i ∂ x j r

Example 1.2 (Tidal force on a system consisting of two particles) We shall first
consider two test particles with a vertical separation vector in the gravitational field
of a particle with mass M. Let us introduce a small Cartesian coordinate system at a
distance R from the mass (Fig. 1.6). The particles are separated from each other by
a distance z  R.

According to Eq. (1.3) a test particle with mass m at a point (0, 0, z) is acted
upon by a gravitational force

GM
Fz (z) = −m , (1.51)
(R + z)2

While an identical particle at the origin is acted upon by a force

GM
Fz (0) = − m . (1.52)
R2

Fig. 1.6 Horizontal tidal

force. A small Cartesian
coordinate system at a
distance R from a particle
with mass M
12 1 Newton’s Theory of Gravitation

Secondly, we shall first consider two particles in the same height in an

inhomogeneous gravitational field.
If this little system is falling freely towards M, an observer at the origin will say
that the particle at (0, 0, z) is acted upon by a force

2Gm M
f z = Fz (z) − F(0) ≈ z (1.53)
R3
directed away from the origin along the positive z-axis. This is the tidal force.
In the same way particles at the same height in the gravitational field, at positions
(x, 0, 0) and (0, y, 0) are attracted towards the origin by tidal forces

Gm M Gm M
fx = − 3
x, f y = − y. (1.54)
R R3
Note that the tidal force increases faster in strength with decreasing distance from
the mass which produces the gravitational field than the gravitational force.
Equations (1.53) and (1.54) have among others the following consequence: if an
elastic circular ring is falling freely in the gravitational field of the Earth, it will
be stretched in the vertical direction and compressed in the horizontal direction
(Fig. 1.7).
In general tidal forces cause changes of shape.

Example 1.3 (The tidal field on the Earth due to the Moon) The Earth–Moon system
is illustrated in Fig. 1.8. (Actually the distance between the Earth and the Moon is
much greater compared to the magnitude of the Earth.) The tidal force due to the
Moon on the surface of the Earth is the difference between the gravitational force at
A and C in the gravitational field of the Moon.

From the extended Pythagorean law we have, with reference to Fig. 1.8

r12 = R 2 − 2r R cos ψ + r 2 . (1.55)

Fig. 1.7 An elastic ring

originally circular, falling
freely in the gravitational
field of the Earth
1.4 Tidal Forces 13

Fig. 1.8 The Earth-Moon

system

A series expansion of

r 2 − 1/2
1 1 r
= 1 − 2 cos ψ + (1.56)
r R R R

to second order in r/R gives

 
1 1 r 1
r 2  
= 1 + cos ψ + 3 cos2 ψ − 1 . (1.57)
r R R 2 R

Hence, the potential at a point A on the surface of the Earth in the gravitational
field of the Moon is
 
Gm Gm r 1
r 2  
Vm = − =− 1 + cos ψ + 3 cos ψ − 1 .
2
(1.58)
r R R 2 R

The first term, VC = −Gm/R, is the potential in the gravitational  field

 of the
Moon at the centre, C, of the Earth. The second term is V A = − Gm/R 2 r cos ψ =
−gMoon x A , where x A is component of the separation between A and C along the
gravitational field. This is the difference of the potential at C and A in the gravitational
field of the Moon if one neglects the inhomogeneity in the Moon’s field at the Earth.
Hence the sum of the first two terms is then the potential at A in the gravitational
field of the Moon. This means that the third term,

Gm 2  
Vt = − 3
r 3 cos2 ψ − 1 , (1.59)
2R
is the difference between the potential at A in the Moon’s gravitational field if the
field is considered homogeneous with the value at the centre of the Earth and the
actual potential at A. This difference is due to the inhomogeneity of the gravitational
field of the Moon at the Earth, i.e. it is due to the tidal gravitational field. It is therefore
called the tidal potential at A.
14 1 Newton’s Theory of Gravitation

Fig. 1.9 Tidal acceleration field. The Tidal acceleration field (red) at the surface of the Earth due
to the Moon is the acceleration of gravity at the surface (black) of the Earth minus the acceleration
of gravity at the centre (green) of the Earth in the Moon’s gravitational field

The height difference, h, between flood and ebb due to the Moon’s tidal field is
given by

gh = Vt (0) − Vt (π/2), (1.60)

where Vt = Vt (ψ), and g is the acceleration of gravity at the surface of the Earth.
This gives

3 Gm r 2
h = . (1.61)
2 g R3

For a numerical result we need the following values:

MMoon = 7.35 · 1022 kg, g = 9.81 m/s2 , R = 3.85 · 105 km, rEarth = 6378 km

Inserting this into Eq. (1.46) gives the height differences on the ocean of the
Earth due to the Moon, neglecting the effects of ocean currents and coast lines,
h = 53 cm. The tidal field is shown in Fig. 1.9.

1.5 The Principle of Equivalence

Galilei investigated experimentally the motion of freely falling bodies. He found that
they moved in the same way, regardless of what sort of material they consisted of and
what mass they had. In Newton’s theory of gravitation mass appears in two different
ways; as gravitational mass, m G , in the law of gravitation, analogously to charge in
Coulomb’s law, and as inertial mass, m I , in Newton’s second law.
Newton’s 2. law applied to a freely falling body with gravitational mass m G and
inertial mass m I in the field of gravity from a spherical body with mass M then takes
1.5 The Principle of Equivalence 15

the form

d 2 r M
mI 2
= −Gm G 3 r. (1.62)
dt r
The results of Galilei’s measurement indicated that the acceleration is independent
of the constitution of the bodies, and hence, the gravitational and inertial mass must
be the same for all bodies,

mG = m I . (1.63)

Measurements performed by the Hungarian baron Eötvös around the year

1900 indicated that this equality holds with an accuracy better than 10−8 .
A parameter which is often used to specify the accuracy of tests of the equality
of gravitational and inertial mass is
mG
≡ − 1. (1.64)
mI

A very accurate test was published on 18 January 2018 [1]. A team of physi-
cists reported about tests based on 7 years with observational data from the MES-
SENGER space observatory. They deduced from the MESSENGER data that  =
(−4.1 ± 4.6) · 10−15 . This is the most accurate test of the equality of gravitational
and inertial mass to date.
Einstein assumed the exact validity of Eq. (1.63). He considered this as a con-
sequence of a fundamental principle, the principle of equivalence, namely that the
physical effects of a gravitational field due to an acceleration (including rotation) of
the reference frame are equivalent to the physical effects of a gravitational field due
to a mass distribution.
A consequence of this principle is the possibility of removing locally the acceler-
ation of gravity by entering a laboratory in free fall. In order to clarify this, Einstein
considered a homogeneous gravitational field in which the acceleration of gravity,
g, is independent of the position. Using Eq. (1.2) in a freely falling non-rotating
reference frame in this field, with a given by

Gm G M
mIa = , (1.65)
R2
a free particle moves according to

d2 r
mI = m G g − m I g = 0, (1.66)
dt 2
where we have used Eq. (1.63). This means that an observer in such a freely falling
reference frame will experience that a particle which is not acted upon by non-
gravitational forces will move along a straight line with constant velocity.
16 1 Newton’s Theory of Gravitation

According to Newton’s theory the particle is acted upon by a gravitational force.

In Newton’s theory a free particle is a particle which is acted upon by a gravitational
force only. Furthermore, a reference frame which falls freely in a gravitational field
is accelerated according to Newton’s theory.
It is not inertial.
All of this is different according to the general theory of relativity. According
to Einstein’s theory gravitation is not reckoned as a force, and a free particle is not
acted upon by any forces. Furthermore, the general definition of an inertial frame,
valid both in Newton’s and Einstein’s theory, is that an inertial frame is a frame
where Newton’s 1. law is valid. We have seen above that a free particle moves along
a straight line with constant velocity in a freely falling reference frame. According to
Einstein’s theory it is not acted upon by any force. Hence Newton’s 1. law is valid in
the freely falling frame. This means that according to the general theory of relativity
an inertial frame falls freely. Also, there is no acceleration of gravity in an inertial
frame. All of these are consequences of the principle of equivalence according to
Einstein’s theory.
The principle of equivalence has also been formulated in an “opposite way.” An
observer at rest in a homogeneous gravitational field and an observer in an accelerated
reference frame far from any mass distribution will obtain identical results when they
perform similar experiments. The physical effects of a gravitational field caused
by the motion of the reference frame are equivalent to the physical effects of a
gravitational field caused by a mass distribution.
One often hears that there is a connection between gravity and spacetime curvature
according to Einstein’s theory. The concept spacetime curvature will be thoroughly
introduced later, but a few words may be in order already here, so that possible
misunderstanding can be avoided at this initial point.
The experience of acceleration of gravity has nothing to do with spacetime cur-
vature. It depends upon the motion of the observer’s reference frame. Acceleration
of gravity is experienced when the reference frame of the observer is not inertial.
It is independent both of spacetime curvature and whether one is close to a mass
distribution. When we experience acceleration of gravity at the surface of the Earth,
it is because being at rest on this surface means not being in an inertial reference
frame. We accelerate upwards relative to an inertial frame when we are at rest on the
surface of the Earth. Therefore we experience a downwards acceleration of gravity.
The Newtonian force which is related to spacetime curvature is the tidal force
as described mathematically in Eq. (1.50). The relativistic generalization of this
equation is the equation of geodesic deviation (see Chap. 6) which contains the
components of the spacetime curvature.
Tidal forces represent the inhomogeneity of the Newtonian gravitational field. In
order to observe this inhomogeneity by physical measurements, one needs to perform
an experiment that requires some extension in space and time.
The principle of equivalence as formulated above has only a local validity. The
word local here means that the extension in space and time is so small that tidal
effects cannot be measured. Hence the principle of equivalence is valid only in the
limit that the gravitational field can be considered homogeneous. In a geometrical
1.5 The Principle of Equivalence 17

language the principle of equivalence is valid only as far as spacetime curvature

cannot be measured.

1.6 The General Principle of Relativity

The principle of equivalence led Einstein to a generalization of the special principle

of relativity. In his general theory of relativity Einstein formulated a general principle
of relativity, which says that not only velocities are relative, but accelerations, too.
Let us consider two formulations of the special principle of relativity.
S1. All laws of nature are the same (may be formulated in the same way) in all inertial
frames.
S2. Every inertial observer can consider himself to be at rest.
These two formulations may be interpreted as different formulations of a single
principle. But the generalization of S1 and S2 to the general case, which encompasses
accelerated motion and non-inertial frames, leads to two different principles G1 and
G2.
G1. The laws of nature are the same in all reference frames.
G2. Every observer can consider himself to be at rest.
In the literature both G1 and G2 are mentioned as the general principle of relativity.
But G2 is a stronger principle (i.e. stronger restriction on natural phenomena) than
G1. Generally the course of events of a physical process in a certain reference frame
depends upon the laws of physics, the boundary conditions, the motion of the refer-
ence frame and the geometry of spacetime. The two latter properties are described by
means of a metrical tensor. By formulating the physical laws in a metric-independent
way, one obtains that G1 is valid for all types of physical phenomena. Even if the
laws of nature are the same in all reference frames, the course of events of a physical
process will, as mentioned above, depend upon the motion of the reference frame.
As to the spreading of light, for example, the law is that light follows null-geodesic
curves (see Chap. 4). This law implies that the path of a light particle is curved in
non-inertial reference frames and straight in inertial frames.
The question whether G2 is true in the general theory of relativity has been
thoroughly discussed recently, and the answer is not clear yet [2].

1.7 The Covariance Principle

The principle of relativity is a physical principle. It is concerned with physical phe-

nomena. This principle motivates the introduction of a formal principle, called the
covariance principle: the equations of a physical theory shall have the same form
in every coordinate system. This principle is not concerned directly with physical
phenomena. The principle may be fulfilled for every theory by writing the equations
18 1 Newton’s Theory of Gravitation

in a form invariant, i.e. covariant way. This may be done by using tensor (vector)
quantities, only, in the mathematical formulation of the theory.
The covariance principle and the equivalence principle may be used to obtain a
description of what happens in the presence of gravitation. We then start with the
physical laws as formulated in the special theory of relativity. Then the laws are
written in a covariant form, by writing them as tensor equations. They are then valid
in an arbitrary, accelerated system. But the inertial field (fictive force) in the accel-
erated frame is equivalent to a gravitational field. So, starting within a description
referred to an inertial frame, we have obtained a description valid in the presence of
a gravitational field.
The tensor equations have in general a coordinate-independent form. Yet, such
form-invariant, or covariant, equations need not fulfil the principle of relativity. This
is due to the following circumstances. A physical principle, for example the principle
of relativity, is concerned with observable relationships. Therefore, when one is going
to deduce the observable consequences of an equation, one has to establish relations
between the tensor components of the equation and observable physical quantities.
Such relations have to be defined; they are not determined by the covariance principle.
From the tensor equations, that are covariant, and the defined relations between the
tensor components and the observable physical quantities, one can deduce equations
between physical quantities. The special principle of relativity, for example, demands
that the laws which these equations express must be the same in every inertial frame
The relationships between physical quantities and tensors (vectors) are theory
dependent. The relative velocity between two bodies, for example, is a vector within
Newtonian kinematics. However, in the relativistic kinematics of four-dimensional
spacetime, an ordinary velocity, which has only three components, is not a vector.
Vectors in spacetime, so-called 4-vectors, have four components. Equations between
physical quantities are not covariant in general. For example, Maxwell’s equations in
three-vector form are not invariant under a Galilei transformation. However, if these
equations are rewritten in tensor form, then neither a Galilei transformation nor any
other transformation will change the form of the equations.
If all equations of a theory are tensor equations, the theory is said to be given a
manifestly covariant form. A theory which is written in a manifestly covariant form
will automatically fulfil the covariance principle, but it need not fulfil the principle
of relativity.

1.8 Mach’s Principle

Einstein gave up Newton’s idea of an absolute space. According to Einstein all

motion is relative. This may sound simple, but it leads to some highly non-trivial
and fundamental questions. Imagine that there are only two particles connected by a
spring in the universe. What will happen if the two particles rotate about each other?
Will the spring be stretched due to centrifugal forces? Newton would have confirmed
that this is indeed what will happen. However, when there is no longer any absolute
1.8 Mach’s Principle 19

space that the particles can rotate relatively to, the answer is not so obvious. If we,
as observers, rotate around the particles, and they are at rest, we would not observe
any stretching of the spring. But this situation is kinematically equivalent to the one
with rotating particles and observers at rest, which leads to stretching.
Such problems led Mach to the view that all motion is relative. The motion of
a particle in an empty universe is not defined. All motion is motion relatively to
something else, i.e. relatively to other masses. According to Mach this implies that
inertial forces must be due to a particle’s acceleration relatively to the great masses
of the universe. If there were no such cosmic masses, there would not exist inertial
forces, like the centrifugal force. In our example with two particles connected by a
string, there would not be any stretching of the spring if there were no cosmic masses
that the particles could rotate relatively to.
Another example may be illustrated by means of a turnabout. If we stay on this,
while it rotates, we feel that the centrifugal forces lead us outwards. At the same time
we observe that the heavenly bodies rotate. According to Mach identical centrifugal
forces should appear if the turnabout is static and the heavenly bodies rotate.
Einstein was strongly influenced by Mach’s arguments, which probably had some
influence, at least with regards to motivation, on Einstein’s construction of his general
theory of relativity. Yet, it is clear that general relativity does not fulfil all require-
ments set by Mach’s principle. For example there exist general relativistic, rotating
cosmological models, where free particles will tend to rotate relative to the cosmic
masses of the model.
However, some Machian effects have been shown to follow from the equations
of the general theory of relativity. For example, inside a rotating massive shell the
inertial frames, i.e. the free particles, are dragged on and tend to rotate in the same
direction as the shell. This was discovered by Lense and Thirring in 1918 and is
therefore called the Lense–Thirring effect. More recent investigations of this effect
have, among others, led to the following result [3]: “a massive shell with radius
equal to its Schwarzschild radius has often been used as an idealized model of our
universe. Our result shows that in such models local inertial frames near the centre
cannot rotate relatively to the mass of the universe. In this way our result gives an
explanation in accordance with Mach’s principle, of the fact that the fixed stars are
at rest on the heaven as observed from an inertial reference frame.”

1.9 Exercises

1.1 A tidal force pendulum

A tidal force pendulum consists of two points with the same mass that are
connected by a stiff, massless rod as shown in Fig. 1.10.
The length of the rod is 2. The pendulum oscillates with respect to the centre
of the rod, which is fixed at a constant distance from the centre of the Earth. The
pendulum oscillates in a vertical plane. The Earth is considered as a spherically
20 1 Newton’s Theory of Gravitation

Fig. 1.10 A tidal force

pendulum

symmetric mass distribution. The gravitational forces acting on the two mass
points are F1 and F2 as shown in the figure.
Find an expression for the period of the oscillation of the rod. What is the
period of the tidal force pendulum at the surface of the Earth? What happens in
the limit of a homogeneous gravitational field?
1.2 Newtonian potential for a spherically symmetric body

(a) Calculate the Newtonian potential φ(r ) outside and inside a spherical shell
of matter with mass M and radius R.
(b) Let M and R be the mass and radius of the Earth. Assume that the Earth
has constant mass density. Find the potential outside and inside the earth.
(c) Calculate the acceleration of gravity at the surface of the Earth.

1.3 Frictionless motion in a tunnel through the Earth

(a) We neglect the rotation of the Earth and air resistance and friction. Assume
that a hollow straight tube has been drilled through the centre of the Earth.
A small ball is dropped into the tube from the surface of the Earth and will
perform an oscillating motion in the tube.

Find the position of the ball as a function of time. What is the period of the
motion?
References 21

(b) We now consider a straight tube from one point on the surface of the Earth
to another, not passing through the centre of the Earth. A ball is dropped
into the tube from a point on the surface of the Earth.
Find the period of the motion. Does it depend upon the direction of the tube?

(c) How long time does a satellite moving in a circular path around the Earth
at the surface of the Earth use to move around the Earth?

1.4 The Earth–Moon system

(a) Assume that the Earth and the Moon are spherical objects isolated from
the rest of the solar system. Write down the equations of motion for the
Earth–Moon system. Show that there is a solution where the Earth and the
Moon are moving in circular orbits around their common centre of mass.
What are the radii of the orbits in terms of the masses of the Earth and the
Moon and the period of the motion?
(b) Find the Newtonian potential along the line connecting the two bodies.
Show the result graphically and find the point on the line where the
gravitational forces from the bodies cancel each other.
(c) Calculate the difference of the gravitational force from the Moon upon a
1 kg particle on the points on the Earth that are closest to the Moon and
farthest away from the Moon.

1.5 The Roche limit

(a) A spherical Moon with mass m and radius R is orbiting a planet with mass
M. Show that if the Moon is closer to the centre of the planet than

 1/3
2M
r= R,
m

then a stone lying freely on the surface of the Moon will be elevated due to tidal
forces.
(b) The comet Shoemaker–Levy 9, that in July 1994 collided with Jupiter, was
ripped apart already in 1992 after having passed near Jupiter. The comet
had a mass of m = 2.0 · 1012 kg, and the closest passage in 1992 was at a
distance of s = 96 000 km from the centre of Jupiter. The mass of Jupiter
is M = 1.9 · 1027 kg.
Use this information to estimate the size of the nucleus of the comet.
22 1 Newton’s Theory of Gravitation

References

1. Genova, A., et al.: Solar system expansion and strong equivalence principle as seen by the NASA
MESSENGER mission. Nature Commun. 1, 10–11 (2018)
2. Grøn, Ø., Vøyenli, K.: On the foundations of the principle of relativity. Found. Phys. 29, 1695–
1793 (1999)
3. Brill, D.R., Cohen, J.M.: Rotating masses and their effect on inertial frames. Phys. Rev. 143,
1011–1015 (1966)
Chapter 2
The Special Theory of Relativity

Abstract This chapter gives a concise and yet rather complete introduction to the
special theory of relativity. Minkowski diagrams are used to illustrate several con-
cepts such as the relativity of simultaneity. Special relativity is a theory of flat space-
time admitting accelerated and rotating reference frames. In this chapter we also
show how magnetism appears as a 2 order effect in v/c of electricity due to the
Lorentz transformation.

In this chapter we shall give a short introduction to the fundamental principles of the
special theory of relativity and deduce some of the consequences of the theory.
The special theory of relativity was presented by Albert Einstein in 1905. It was
founded on two postulates:
1. The laws of physics are the same in all Galilean frames.
2. The velocity of light in empty space is the same in all Galilean frames and
independent of the motion of the light source.
Einstein pointed out that these postulates are in conflict with Galilean kinemat-
ics, in particular with the Galilean law for the addition of velocities. According to
Galilean kinematics two observers moving relative to each other cannot measure the
same velocity for a certain light signal. Einstein solved this problem by a thorough
discussion of how two distant clocks should be synchronized.

2.1 Coordinate Systems and Minkowski Diagrams

The most simple physical phenomenon that we can describe is called an event. This
is an incident that takes place at a certain point in space and at a certain point in time.
A typical example is the flash from a flashbulb.
A complete description of an event is obtained by giving the position of the
event in space and time. Assume that our observations are made with reference to a
reference frame. We introduce a coordinate system into our reference frame. Usually
it is advantageous to employ a Cartesian coordinate system. This may be thought
of as a cubic lattice constructed by measuring rods. If one lattice point is chosen as
© Springer Nature Switzerland AG 2020 23
Ø. Grøn, Introduction to Einstein’s Theory of Relativity,
Undergraduate Texts in Physics, https://doi.org/10.1007/978-3-030-43862-3_2
24 2 The Special Theory of Relativity

origin, with all coordinates equal to zero, then any other lattice point has three spatial
coordinates equal to the distances of that point along the coordinate axes that pass
through the origin. The spatial coordinates of an event are the three coordinates of the
lattice point at which the event happens. It is somewhat more difficult to determine
the point of time of an event. If an observer is sitting at the origin with a clock, then
the point of time when he catches sight of an event is not the point of time when the
event happened. This is because the light takes time to pass from the position of the
event to the observer at the origin.
Since observers at different positions have to make different such corrections, it
would be simpler to have (imaginary) observers at each point of the reference frame
such that the point of time of an arbitrary event can be measured locally. But then
a new problem appears. One has to synchronize the clocks, so that they show the
same time and go at the same rate. This may be performed by letting the observer
at the origin send out light signals so that all the other clocks can be adjusted (with
correction for light travel time) to show the same time as the clock at the origin.
These clocks show the coordinate time of the coordinate system, and they are called
coordinate clocks.
By means of the lattice of measuring rods and coordinate clocks, it is now easy to
determine four coordinates (ct, x, y, z) for every event. (We have multiplied the time
coordinate t by the velocity of light c in order that all four coordinates shall have
the same dimension.) This coordinatization makes it possible to describe an event
as a point P in a so-called Minkowski diagram. In this diagram we plot ct along the
vertical axis and one of the spatial coordinates along the horizontal axis.
In order to observe particles in motion, we may imagine that each particle is
equipped with a flashlight and that they flash at a constant frequency. The flashes
from a particle represent a succession of events. If they are plotted into a Minkowski
diagram, we get a series of points that describe a curve in the continuous limit. Such
a curve is called a world line of the particle. The world line of a free particle is a
straight line, as shown to the left of the time axis in Fig. 2.1.
A particle acted upon by a net force has a curved world line as the velocity of the
particle changes with time. Since the velocity of every material particle is less than
the velocity of light, the tangent of a world line in a Minkowski diagram will always
make an angle less than 45° with the time axis. A flash of light gives rise to a light
front moving onwards with the velocity of light. If this is plotted in a Minkowski
diagram, the result is a light cone. In Fig. 2.1 we have drawn a light cone for a flash
at the origin. It is obvious that we could have drawn light cones at all points in the
diagram. An important result is that the world line of any particle at a point is inside
the light cone of a flash from that point. This is an immediate consequence of the
special principle of relativity and is also valid locally in the presence of a gravitational
field.
2.2 Synchronization of Clocks 25

Fig. 2.1 World lines. World lines of particles moving slower than light are inside the light cone

2.2 Synchronization of Clocks

There are several equivalent methods that can be used to synchronize clocks. We
shall here consider the radar method. Then a mirror is placed on the x-axis and emits
a light signal from the origin at time t A . This signal is reflected by the mirror at t B
and received again by the observer at the origin at time t C . According to the second
postulate of the special theory of relativity, the light moves with the same velocity
in both directions, giving

1
tB = (t A + tC ). (2.1)
2
When this relationship holds we say that the clocks at the origin and at the mirror
are Einstein synchronized. Such synchronization is presupposed in the special theory
of relativity. The situation corresponding to synchronization by the radar method is
shown in Fig. 2.2.
The radar method can also be used to measure distances. The distance L from the
origin to the mirror is given by
c
L= (tC − t A ). (2.2)
2
26 2 The Special Theory of Relativity

Fig. 2.2 Clock synchronization by the radar method

2.3 The Doppler Effect

Consider three observers (1, 2 and 3) in an inertial frame. Observers 1 and 3 are at
rest, while 2 moves with constant velocity along the x-axis. The situation is shown
in Fig. 2.3.
Each observer is equipped with a clock. If observer 1 emits light pulses with a
constant period τ 1 , then observer 2 receives them with a longer period τ 2 according
to his or her clock. The fact that these two periods are different is a well-known
phenomenon, called the Doppler effect. The same effect is observed with sound; the
frequency of a receding vehicle is lower than that of an approaching one.
We are now going to deduce a relativistic expression for the Doppler effect. Firstly,
we see from Fig. 2.3 that the two periods τ 1 and τ 2 are proportional to each other,

τ2 = K τ1 . (2.3)

The constant K(v) is called Bondi’s K-factor. Since observer 3 is at rest, the period
τ 3 is equal to τ 1 so that

1
τ3 = τ2 . (2.4)
K
These two equations imply that if 2 moves away from 1, so that τ 2 > τ 1 , then τ 3
< τ 2 . This is because 2 moves towards 3.
2.3 The Doppler Effect 27

Fig. 2.3 The Doppler effect

The K-factor is most simply determined by placing observer 1 at the origin, while
letting the clocks show t1 = t2 = 0 at the moment when 2 passes the origin. This
is done in Fig. 2.3. The light pulse emitted at the point of time t A is received by 2
when his clock shows τ2 = K t A . If 2 is equipped with a mirror, the reflected light
pulse is received by 1 at a point of time tC = K τ2 = K 2 t A . According to Eq. (2.1)
the reflection event then happens at a point of time

1 1 
tB = (tC + t A ) = K 2 + 1 t A . (2.5)
2 2
The mirror has then arrived at a distance x B from the origin, given by Eq. (2.2),
c c 
xB = (tC − t A ) = K 2 − 1 t A . (2.6)
2 2
Thus, the velocity of observer 2 is

xB K1 − 1
v= = 1 c. (2.7)
tB K +1

Solving this equation with respect to K gives

 1/2
c+v
K = . (2.8)
c−v
28 2 The Special Theory of Relativity

This result is relativistically correct. The special theory of relativity was included
through the tacit assumption that the velocity of the reflected light is c. This is a
consequence of the second postulate of special relativity; the velocity of light is
isotropic and independent of the velocity of the light source.
Since the wavelength λ of the light is proportional to the period τ , Eq. (2.3) gives
the observed wavelength λ for the case when the observer moves away from the
source,
 1/2
 c+v
λ = Kλ = λ. (2.9)
c−v

This Doppler effect represents a redshift of the light. If the light source moves
towards the observer, there is a corresponding blueshift given by K −1 .
It is common to express this effect in terms of the relative change of wavelength,

λ − λ
z= = K −1 (2.10)
λ
which is positive for redshift. If v  c, Eq. (2.9) gives

λ v
= K ≈1+ (2.11)
λ c
To this order the redshift is

z = v/c. (2.12)

This expression of the Doppler shift is well known in non-relativistic physics.

2.4 Relativistic Time Dilation

Every periodic motion can be used as a clock. A particularly simple clock is called
the photon clock. This is shown in Fig. 2.4.
The clock consists of two parallel mirrors that reflect a light pulse back and forth.
If the period of the clock is defined as the time interval between each time the light
pulse hits the lower mirror, then t  = 2L 0 /c.
Assume that the clock is at rest in an inertial reference frame   where it is placed
along the y-axis, as shown in Fig. 2.4. If this system moves along the ct-axis with a
velocity v relative to another inertial reference frame , the light pulse of the clock
will follow a zigzag path as shown in Fig. 2.5.
The light signal follows a different path in  than in   . The period t of the
clock as observed in  is different from the period t  which is observed in the
rest frame. The period t is easily found in Fig. 2.5. Since the pulse takes the time
2.4 Relativistic Time Dilation 29

Fig. 2.4 Photon clock at rest

Fig. 2.5 Moving photon clock

(1/ 2)t from the lower to the upper mirror and since the light velocity is always the
same, we find
 2  2
1 1
ct = L 20 + vt . (2.13)
2 2

i.e.
30 2 The Special Theory of Relativity

2L 0 1
t = γ , γ = . (2.14)
c 1 − v 2 /c2

The γ factor is a useful shorthand notation for a term which is often used in
relativity theory. It is commonly known as the Lorentz factor.
Since the period of the clock in its rest frame is t  = 2L 0 /c, we get

t = γ t  . (2.15)

Thus, we have to conclude that the period of the clock when it is observed to move
(t) is greater that its rest period (t  ). In other words: a moving clock goes slower
than a clock at rest. This is called the relativistic time dilation. The period t  of
the clock as observed in its rest frame is called the proper period of the clock. The
corresponding time t  is called the proper time of the clock.
One might be tempted to believe that this surprising consequence of the special
theory of relativity has something to do with the special type of clock that we have
employed. This is not the case. If there had existed a mechanical clock in  that did
not show the time dilation, then an observer at rest in  might measure his velocity
by observing the different rates of his light clock and this mechanical clock. In this
way he could measure the absolute velocity of . This would be in conflict with the
special principle of relativity.

2.5 The Relativity of Simultaneity

Events that happen at the same point of time are said to be simultaneous events.
We shall now show that according to the special theory of relativity, events that are
simultaneous in one reference frame are not simultaneous in another reference frame
moving with respect to the first. This is what is meant by the expression “the relativity
of simultaneity”.
Consider again two mirrors connected by a line along the x  -axis, as shown in
Fig. 2.6.
Halfway between the mirrors there is a flash-lamp emitting a spherical wavefront
at a point of time tC .
The points at which the light front reaches the left-hand and the right-hand mirrors
are denoted by A and B, respectively. In the reference frame   of Fig. 2.6 the events
A and B are simultaneous.
If we describe the same course of events from another reference frame , where
the mirror moves with constant velocity v in the positive x-direction, we find the
Minkowski diagram shown in Fig. 2.7. Note that the light follows world lines making
an angle of 45° with the axes. This is the case in every inertial frame.
In  the light pulse reaches the left mirror, which moves towards the light, before
it reaches the right mirror, which moves in the same direction as the light. In this
reference frame the events when the light pulses hit the mirrors are not simultaneous.
2.5 The Relativity of Simultaneity 31

Fig. 2.6 Simultaneous events A and B

Fig. 2.7 The simultaneous events of Fig. 2.6 in another frame

As an example illustrating the relativity of simultaneity, Einstein imagined that

the events A, B and C happen in a train which moves past the platform with a velocity
v. The event C represents the flash of a lamp at the mid-point of a wagon. A and B
are the events when the light is received at the back end and at the front end of the
wagon, respectively. This situation is shown in Fig. 2.8.
32 2 The Special Theory of Relativity

Fig. 2.8 Light flash in a moving train

As observed in the wagon, A and B happen simultaneously. As observed from the

platform the rear end of the wagon moves towards the light which moves backwards,
while the light moving forwards has to catch up with the front end. Thus, as observed
from the platform A will happen before B.
The time difference between A and B as observed from the platform will now be
calculated. The length of the wagon, as observed from the platform, will be denoted
by L. The time coordinate is chosen such that tC = 0. The light moving backwards
hits the rear wall at a point of time t A . During the time t A the wall has moved a
distance vt A forwards, and the light has moved a distance ct A backwards. Since the
distance between the wall and the emitter is L/2, we get

L
= vt A + ct A . (2.16)
2
Thus
L
tA = . (2.17)
2(c + v)

In the same manner one finds

L
tB = . (2.18)
2(c − v)

It follows that the time difference between A and B as observed from the platform
is
vL
t = t B − t A = γ 2 . (2.19)
c2
As observed from the wagon A and B are simultaneous. As observed from the
platform the rear event A happens at a time interval t before the event B. This is
the relativity of simultaneity.
2.6 The Lorentz Contraction 33

2.6 The Lorentz Contraction

During the first part of the nineteenth century the so-called luminiferous ether was
introduced into physics to account for the propagation and properties of light. After
J. C. Maxwell showed that light is electromagnetic waves and the ether was still
needed as a medium in which electromagnetic waves propagated.
It was shown that Maxwell’s equations do not obey the principle of relativity, when
coordinates are changed using the Galilean transformations. If it is assumed that the
Galilean transformations are correct, then Maxwell’s equations can only be valid in
one coordinate system. This coordinate system was the one in which the ether was at
rest. Hence, Maxwell’s equations in combination with the Galilean transformations
implied the concept of “absolute rest”. This made the measurement of the velocity
of the Earth relative to the ether of great importance.
An experiment sufficiently accurate to measure this velocity to order v 2 /c2 was
carried out by Michelson and Morley in 1887. A simple illustration of the experiment
is shown in Fig. 2.9.
Our earlier photon clock is supplied by a mirror at a distance L along the x-axis
from the emitter. The apparatus moves in the x-direction with a velocity v. In the
rest frame   of the apparatus, the distance between A and B is equal to the distance
between A and C. This distance is denoted by L 0 and is called the rest length between
A and B.
Light is emitted from A. Since the velocity of light is isotropic and the distances to
B and C are equal in   , the light reflected from B and that reflected from C have the
same travelling time. This was the result of the Michelson–Morley experiment, and
it seems that we need no special effects such as the Lorentz contraction to explain
the experiment.
However, before 1905 people believed in the physical reality of absolute velocity.
The Earth was considered to move through the “ether” with a velocity that changed

Fig. 2.9 Length contraction

34 2 The Special Theory of Relativity

with the seasons. The experiment should therefore be described under the assumption
that the apparatus is moving.
Let us therefore describe an experiment from our reference frame , which may
be thought of as at rest in the “ether”. Then according to Eq. (2.14) the travel time
of the light being reflected at C is

2L 0
tC = γ . (2.20)
c
For the light moving from A to B we may use Eq. (2.18) and for the light from B
to A Eq. (2.17). This gives

L L 2L
t B = + = γ2 . (2.21)
c−v c+v c

If length is independent of velocity, then L = L 0 . In this case the travelling times

of the light signals will be different. The travelling time difference is

2L 0
t B − tC = γ (γ − 1) . (2.22)
c

To the lowest order in v/c we have γ ≈ 1 + (1/2)(v/c)2 , so that

1  v 2
t B − tC ≈ . (2.23)
2 c
which depends upon the velocity of the apparatus.
According to the ideas involving an absolute velocity of the Earth through the
ether, if one lets the light reflected at B interfere with the light reflected at C (at the
position A), then the interference pattern should vary with the season. This was not
observed. On the contrary, observations showed that t B = tC .
Assuming that length varies with velocity, Eqs. (2.20) and (2.21), together with
this observation, gives

v2
L= 1− L 0. (2.24)
c2
Hence, L < L 0 , i.e. the length of a rod is less when it moves than when it is at
rest. This is called the Lorentz contraction.

2.7 The Lorentz Transformation

 
An event P has coordinates t  , x  , 0, 0 in a Cartesian coordinate system associated
with a reference frame   . Thus the distance from the origin of   to P measured
2.7 The Lorentz Transformation 35

with a measuring rod at rest in   is x  . If the distance between the origin of   and
the position at the x-axis where P took place is measured with measuring rods at
rest in a reference frame moving with velocity v in the x-direction relative to   , one
finds the length γ −1 x due to the Lorentz contraction. Assuming that the origin of 
and   coincided at the point of time t = 0, the origin of   has an x-coordinate vt
at a point of time t. The event P thus has an x-coordinate

x = vt + γ −1 x  (2.25)

or

x  = γ (x − vt). (2.26)

The x-coordinate may be expressed in terms of x  and y  by letting v → −v,

 
x = γ x  + vt  . (2.27)

The y- and z-coordinates are associated with axes directed perpendicular to the
direction of motion. Therefore, they are the same in the two-coordinate systems
 
y = y, z z . (2.28)

Substituting x  from Eq. (2.26) into Eq. (2.27) reveals the connection between the
time coordinates of the two-coordinate systems,
 vx 
t = γ t − 2 (2.29)
c
and
 
vx 
t = γ t + 2 . (2.30)
c

The latter term in this equation is nothing but the deviation from simultaneity in
 for two events that are simultaneous in   .
The relations (2.27)–(2.30) between the coordinates of  and   represent a
special case of the Lorentz transformations. The above relations are special since
the two-coordinate systems have the same spatial orientation, and the x- and x  -
axes are aligned along the relative velocity vector of the associated frames. Such
transformations are called boosts.
For non-relativistic velocities, v  c, the Lorentz transformations (2.27)–(2.30)
pass over into the corresponding Galilei transformations.
The Lorentz transformation gives a connection between the relativity of simul-
taneity and the Lorentz contraction. The length of a body is defined as the difference
36 2 The Special Theory of Relativity

between the coordinates of its end points, as measured by simultaneity in the rest
frame of the observer.
Consider the wagon of Sect. 2.5. Its rest length is L 0 = x B − x A . The difference
between the coordinates of the wagon’s end points, x A − x B as measured in , is
given implicitly by the Lorentz transformation

x B − x A = γ [x B − x A − v(t B − t A )]. (2.31)

According to the above definition the length L of the moving wagon is given by
L = x B − x A with t B = t A .
From Eq. (2.3) we then get

L0 = γ L . (2.32)

which is equivalent to Eq. (2.24).

The Lorentz transformation will now be used to deduce the relativistic formulae
for velocity addition. Consider a particle moving with velocity u along the x  -axis of
  . If the particle was at the origin at t  = 0, its position at t  is x  = u t  . Using this
relation together with Eqs. (2.27) and (2.28) we find the velocity of the particle as
observed in 

x u + v
u= = . (2.33)
t 1 + u  v/c2

A remarkable property of this expression is that by adding velocities less than c

one cannot obtain a velocity greater than c. For example, if a particle moves with a
velocity c in   then its velocity in  is also c regardless of the velocity of  relative
to   .
Equation (2.33) may be written in a geometrical form by introducing the so-called
rapidity η defined by

u
tanh η = (2.34)
c

for a particle with velocity u  . Similarly the rapidity, η̄, of   relative to  is given
by
v
tanh η̄ = . (2.35)
c
Since
  tanh η + tanh θ
tanh η + η̄ = , (2.36)
1 + tanh η tanh θ

the relativistic velocity addition formula, Eq. (2.33), may be written as

2.7 The Lorentz Transformation 37

η = η + η̄. (2.37)

Since rapidities are additive, their introduction simplifies some calculations and
they have often been used as variables in elementary particle physics.
With these new hyperbolic variables we can write the Lorentz transformation in
a particularly simple way. Using Eq. (2.35) in Eqs. (2.27) and (2.30) we find

x = x  cosh η̄ + ct  sinh η̄, ct = x  sinh η̄ + ct  cosh η̄. (2.38)

2.8 Lorentz Invariant Interval

Let two events be given. The coordinates of the events, as referred to two different ref-
erence frames  and   , are connected by a Lorentz transformation. The coordinate
differences are therefore connected by
 v   
t = γ t  + 2 x  , x = γ x  + vt  , y = y  , z = z  .
c
(2.39)

This leads to
 2  2  2  2
−(ct)2 + (x)2 + (y)2 + (z)2 = − ct  + x  + y  + z  ,
(2.40)

showing that the quantity

(s)2 = −(ct)2 + (x)2 + (y)2 + (z)2 (2.41)

is invariant under a Lorentz transformation. The quantity s is called a spacetime

interval, or only an interval. Due to the minus sign in Eq. (2.40), the square of the
interval between two events may be positive, zero or negative. These three types of
intervals are termed space-like if (s)2 > 0, xero or light-like if (s)2 = 0 and
time-like if (s)2 < 0.
The reasons for these names are the following. Given two events with a space-
like interval (A and B in Fig. 2.10), there exists a Lorentz transformation to a new
reference frame where A and B happen simultaneously. In this frame the distance
between the events is purely spatial. Two events with a light-like interval (C and D
in Fig. 2.10) can be connected by a light signal, i.e. one can send a photon from C to
D. The events E and F have a time-like interval between them and can be observed
from a reference frame in which they have the same spatial position, but occur at
different points of time.
38 2 The Special Theory of Relativity

Fig. 2.10 Space-like, light-like and time-like intervals. Interval between A and B is space-like,
between C and D light-like and between E and F time-like

Since all material particles move with a velocity less than that of light, the points
on the world line of a particle are separated by time-like intervals. The curve is then
said to be time-like. All time-like curves through a point pass inside the light cone
from that point.
For a particle with velocity u = x/t Eq. (2.40) gives
 
u2
(s) = − 1 − 2 (ct)2 .
2
(2.42)
c

In the rest frame   of the particle x  = 0, giving

(s)2 = −(ct)2 . (2.43)

The time t  in the rest frame of the particle is the same as the time measured on a
clock carried by the particle. It is called the proper time of the particle and denoted
by τ . From Eqs. (2.42) and (2.43) it follows that

u2
τ = 1− t = γ −1 t, (2.44)
c2
which is an expression of the relativistic time dilation.
Equation (2.43) is important. It gives the physical interpretation of a time-like
interval between two events. The interval is a measure of the proper time interval
2.8 Lorentz Invariant Interval 39

Fig. 2.11 World line of an accelerating particle

between the events. This time is measured on a clock that moves such that it is present
at both events. In the limit u → c (the limit of a light signal), τ = 0. This shows
that (s)2 = 0 for a light-like interval.
Consider a particle with a variable velocity, u(t), as indicated in Fig. 2.11. In
this situation we can specify the velocity at an arbitrary point of the world line.
Equation (2.44) can be used with this velocity, in an infinitesimal interval around
this point,

u 2 (t)
dτ = 1− dt. (2.45)
c2
This equation means that the acceleration has no local effect upon the proper time
of the clock. Here the word “local” means as measured by an observer at the position
of the clock. Such clocks are called standard clocks.
If a particle moves from A to B in Fig. 2.11, the proper time as measured on a
standard clock following the particle is found by integrating Eq. (2.45)

t B
u 2 (t)
τB − τ A = 1 − 2 dt. (2.46)
c
tA

The relativistic time dilation has been verified with great accuracy by observations
of unstable elementary particles with short lifetimes [1].
An infinitesimal spacetime interval
40 2 The Special Theory of Relativity

ds 2 = −c2 dt 2 + dx 2 + dy 2 + dz 2 . (2.47)

is called a line element. The physical interpretation of the line element between two
infinitesimally close events on a time-like curve is

ds 2 = −c2 dτ 2 , (2.48)

where dτ is the proper time interval between the events, measured with a clock
following the curve. The spacetime interval between two events is given by the
integral (2.46). It follows that the proper time interval between two events is path-
dependent. This leads to the following surprising result: A time-like interval between
two events is greatest along the straightest possible curve between them.

2.9 The Twin Paradox

Rather than discussing the lifetime of elementary particles, we may as well apply
Eq. (2.46) to a person. Let her name be Eva. Assume that Eva is rapidly accelerating
from rest at the point of time t = 0 at origin to a velocity v along the x-axis of a (ct,
x) coordinate system in an inertial reference frame  (Fig. 2.12).
At a point of time t P she has come to a position x P . She then rapidly decelerates
until reaching a velocity v in the negative x-direction. At a point of time t Q , as
measured on clocks at rest in , she has returned to her starting location. If we

Fig. 2.12 Twin paradox world lines. World lines of the twin sisters Eva and Elisabeth
2.9 The Twin Paradox 41

neglect the brief periods of acceleration, Eva’s travelling time as measured on a

clock which she carries with her is
 1/2
v2
tEva = 1− 2 tQ . (2.49)
c

Now assume that Eva has a twin sister named Elizabeth who remains at rest at
the origin of .
Elizabeth has become older by τElisabeth = t Q during Eva’s travel, so that
 1/2
v2
τEva = 1 − 2 tElisabeth . (2.50)
c

For example, if Eva travelled to Alpha Centauri (the Sun’s nearest neighbour at
four light years) with a velocity v = 0.8c, she would be gone for 10 years as measured
by Elizabeth. Therefore Elizabeth has aged 10 years during Eva’s travel. According
to Eq. (2.50), Eva has only aged 6 years. According to Elizabeth, Eva has aged less
than herself during her travels.
The principle of relativity, however, tells that Eva can consider herself as at rest
and Elizabeth as the traveller. According to Eva it is Elizabeth who has only aged by
6 years, while Eva has aged by 10 years during the time they are apart.
What happens? How can the twin sisters arrive at the same prediction as to how
much each of them age during the travel? In order to arrive at a clear answer to these
questions, we shall have to use a result from the general theory of relativity. The twin
paradox will be taken up again in Chap. 5.

2.10 Hyperbolic Motion

With reference to an inertial reference frame it is easy to describe relativistic acceler-

ated motion. The special theory of relativity is in no way limited to describe motion
with constant velocity.
Let a particle move with a variable velocity u(t) = dx/dt along the x-axis in
. The frame   moves with velocity
  v in the same direction relative to . In this
frame the particle velocity is u  t  = dx  /dt  . At every moment the velocities u and
u are connected by the relativistic formula for velocity addition, Eq. (2.33). Thus,
according to Eq. (2.30), a velocity change du in   and the corresponding velocity
change du in  are related by
 
dt  + v/c2 dx  1 + u  v/c2 
dt =  = dt . (2.51)
1 − v 2 /c2 1 − v 2 /c2
42 2 The Special Theory of Relativity

Combining these expressions we obtain the relationship between the acceleration

of the particle as measured in  and  
 3/2
du 1 − v 2 /c2 
a= =  3 a . (2.52)
dt 
1 + u v/c 2

Let   be the rest frame of the particle at a point of time t. Then u  = 0 at this
moment, giving
 3/2
u2
a = 1− 2 a. (2.53)
c

Here a is the acceleration of the particle as measured in its instantaneous rest

frame. It is called the rest acceleration of the particle. Equation (2.53) can be
integrated if we know how the rest acceleration of the particle varies with time.
We shall now focus on the case where the particle has uniformly accelerated
motion and moves along a straight path in space. The rest acceleration of the particle
is constant, say a  = g. Integration of Eq. (2.53) with u(0) = 0 then gives
 −1/2
g2
u = 1 + 2 t2 gt. (2.54)
c

Integrating once more gives

 1/2
c2 g2 2 c2
x= 1+ 2t + x0 − , (2.55)
g c g

where the integration constant x0 is equal to the position at t = 0.

Equation (2.55) can be given the form
 2
c2 c4
x − x0 + − c2 t 2 = . (2.56)
g g2

This is the equation of a hyperbola in the Minkowski diagram (Fig. 2.13).

Since the world line of a particle with uniformly accelerated, rectilinear motion
has the shape of a hyperbola, this type of motion is called hyperbolic motion.
Using the proper time τ of the particle as a parameter, we may obtain a simple
parametric representation of its world line. Substituting Eq. (2.54) into Eq. (2.45)
we get

dt
dτ =    . (2.57)
1 + c2 /g t 2
2.10 Hyperbolic Motion 43

Fig. 2.13 World line of

particle with constant rest
acceleration

Integration with τ (0) = 0 gives

 
c gt
τ = arc sinh . (2.58)
g c
or
 
c gt
t= sinh . (2.59)
g c

Inserting this expression into Eq. (2.55) we get

 
c2 gt c2
x= cosh + x0 − . (2.60)
g c g

These expressions shall be used later when describing uniformly accelerated

reference frames.
Note that hyperbolic motion results when the particle moves with constant rest
acceleration. Such motion is usually called uniformly accelerated motion. Motion
with constant acceleration as measured in the “laboratory frame”  gives rise to the
usual parabolic motion.
44 2 The Special Theory of Relativity

2.11 Energy and Mass

The existence of an electromagnetic radiation pressure was well known before Ein-
stein formulated the special theory of relativity. In black body radiation with mass
density ρ there is an isotropic pressure p = (1/3)ρc2 . If the radiation moves in a
certain direction (laser), then the pressure in this direction is p = ρc2 . Einstein gave
several deductions of the famous equation connecting the inertial mass of a body
with its energy content. A deduction he presented in 1906 is as follows.
Consider a box with a light source at one end. A light pulse with radiation energy
E is emitted to the other end where it is absorbed (see Fig. 2.14).
The box has a mass M and a length L. Due to the radiation pressure of the shooting
light pulse the box receives a recoil. The pulse is emitted during a time interval t.
During this time the radiation pressure is

E E
p = ρc2 = = , (2.61)
V Act
where V is the volume of the radiation pulse and A the area of a cross section of the
box.
The recoil velocity of the box is
  
F pA E At E
v = −at = − t = − t = − =− (2.62)
M M Act M Mc

The pulse takes the time L/c to move to the other side of the box. During this
time the Box moves a distance

Fig. 2.14 Light pulse in a box

2.11 Energy and Mass 45

L EL
x = v =− . (2.63)
c Mc2
Then the box is stopped by the radiation pressure caused by the light pulse hitting
the wall at the other end of the box.
Let m be the mass of the radiation. Before Einstein one would put m = 0. Einstein,
however, reasoned as follows. Since the box and its contents represent an isolated
system, the mass centre has not moved. The mass centre of the box with mass M has
moved a distance x to the left, and the radiation with mass m has moved a distance
L to the right. Thus

m L + Mx = 0 (2.64)

which gives
  
M M EL E
m=− x = − − 2
= 2 (2.65)
L L Mc c
or

E = mc2 . (2.66)

Here we have shown that radiation energy has a mass given by Eq. (2.65). Einstein
derived Eq. (2.66) using several different methods showing that it is valid in general
for all types of systems.
The energy content of even small bodies is enormous. For example, by transform-
ing 1 g of matter to heat, one may heat 300,000 metric tons of water from room
temperature to the boiling point. (The energy corresponding to a mass m is enough
to change the temperature by T of an object of mass M and specific heat capacity
cv : mc2 = Mcv T ).

2.12 Relativistic Increase of Mass

In the special theory of relativity, force is defined as rate of change of momentum. We

consider a body that gets a change of energy dE due to the work performed on it by
a force F. According to Eq. (2.66) and the definition of work (force times distance)
the body gets a change of mass dm, given by

c2 dm = dE = Fds = Fvdt = vd(mv) = mvdv + v 2 dm, (2.67)

which gives
46 2 The Special Theory of Relativity

m
v
dm vdv
= , (2.68)
m c2 − v 2
m0 0

where m 0 is the rest mass of the body—i.e. its mass as measured by an observer
co-moving with the body—and m its mass when its velocity is equal to v. Integration
gives
m0
m= = γ m0. (2.69)
1 − v 2 /c2

In the case of small velocities compared to the velocity of light we may use the
approximation

v2 1 v2
1− 2
=1+ . (2.70)
c 2 c2
With this approximation Eqs. (2.66) and (2.69) give

1
E ≈ m 0 c2 + m 0 v 2 . (2.71)
2
This equation shows that the total energy of a body encompasses its rest energy
m 0 c2 and its kinetic energy. In the non-relativistic limit the kinetic energy is
(1/2)m 0 v 2 . The relativistic expression for the kinetic energy is

E K = E − m 0 c2 = (γ − 1)m 0 c2 . (2.72)

Note that E K → ∞ when v → c.

According to Eq. (2.33), it is not possible to obtain a velocity greater than that
of light by adding velocities. Equation (2.72) gives a dynamical reason that material
particles cannot be accelerated up to and above the velocity of light.
Using Eq. (2.69) the energy and momentum of a particle with rest mass m 0 and
velocity v are

E = γ m 0 c2 , p = γ m0v (2.73)

Using the identity

v2
γ2 = γ2 + 1, (2.74)
c2
It follows from the expressions (2.73) that

E 2 = p 2 c2 + m 20 c4 , (2.75)
2.12 Relativistic Increase of Mass 47

or

m 20 c4 = E 2 − p 2 c2 . (2.76)

Since the rest mass m 0 of a particle is a velocity-independent quantity and hence

Lorentz invariant, it follows that the quantity E 2 − p 2 c2 of a particle is a Lorentz
invariant having the same value in all inertial reference frames even if the energy and
momentum of the particle depend upon the frame.

2.13 Lorentz Transformation of Velocity, Momentum,

Energy and Force

We shall write down the Lorentz transformation of velocity, momentum, energy and
force. (Detailed derivations are given by W. G. V. Rosser in An Introduction to the
Theory of Relativity, Butterworths, 1964). Let  and   be two inertial frames with
a relative velocity v in the x-direction and with co-moving Cartesian coordinates. A
particle is moving with velocity components u x , u y , u z in  and u x , u y , u z in   .
The transformation formulae for the velocity component are
 
ux − v u y 1 − v 2 /c2 u z 1 − v 2 /c2
u x = 
, uy = 
, uz = . (2.77)
1 − vu x /c2 1 − vu x /c2 1 − vu x /c2

It follows from these formulae that

  
2 (u x − v)2 + u 2 − u 2x 1 − v 2 /c2
u = u 2
x + u 2y + u 2
z =  2 . (2.78)
1 − vu x /c2

As shown by Rosser this leads to

1 1 − vu x /c2
 =  . (2.79)
1 − u 2 /c2 1 − v 2 /c2 1 − u 2 /c2

In the frames  and   the momentum of a particle with rest mass m 0 has
components

m0ui m 0 u i
pi =  , pi =  , (2.80)
1 − u 2 /c2 1 − u 2 /c2

and energies

m 0 c2 m 0 c2
E= , E =  . (2.81)
1 − u 2 /c2 1 − u 2 /c2
48 2 The Special Theory of Relativity

Using the first of Eqs. (2.77) and (2.79) gives

px = γ (mu x − mv), (2.82)

where γ and m is given in Eq. (2.79). Since mu x = px and m = E/c2 we get

 
px = γ px − Ev/c2 . (2.83)

Using the second and third of Eq. (2.77) we get

p y = p y , pz = pz . (2.84)

Inserting Eq. (2.79) in the second of Eq. (2.81) we obtain

E  = γ (E − vpx ). (2.85)

At this stage we define force as a 3-vector

dp
f= , (2.86)
dt
or
d px
fx = etc. (2.87)
dt
where p is the momentum as given in Eq. (2.73). Similarly we have

d px
f x = etc. (2.88)
dt 
Using the transformation Eq. (2.83) we then have
   
d vE dt d px v dE
f x = γ  px − 2 = γ  − 2 . (2.89)
dt c dt dt c dt
 
From the Lorentz transformation t  = γ t − vx/c2 we have

dt 1 1 1

= dt  =  vx
=  vu x
. (2.90)
dt dt
d
dt
γ t− c2
λ 1− c2

Furthermore we insert d px /dt = f x in Eq. (2.89) and

dE
= f · u = fx u x + f y u y + fz uz . (2.91)
dt
2.13 Lorentz Transformation of Velocity, Momentum, Energy … 49

Inserting this in Eq. (2.89) and rearranging gives

vu y vu z
f x = f x − fy − 2 fz . (2.92)
c2 − vu x c − vu x

Since p y = p y we get

d p y d py dt d p y dt
f y = = =  =  fy. (2.93)
dt  dt  dt dt dt
Using Eq. (2.90) we obtain the transformations equations for the y- and
z-components of the force

fy fz
f y =  , f z =  . (2.94)
γ 1 − vu x /c2 γ 1 − vu x /c2

The inverse transformation is

vu y vu z f y
f x = f x + 2 f  + f , fy =  ,
c + vu x y c2 + vu x z γ 1 + vu x /c2
f z
fz =  . (2.95)
γ 1 + vu x /c2

If   is the rest system of the particle, so that u x = u y = u z = 0, the

transformation reduces to
 
f x = f x , f y = 1 − v 2 /c2 f y , f z = 1 − v 2 /c2 f z . (2.96)

Example 2.13.1 (The Lever paradox) A right-angled lever with arms of equal length
L  is at rest at the origin of a Cartesian coordinate system co-moving with an inertial
reference frame   . It is oriented with the arms along the x  - and y  -axes. At the
end of the x  -arm there acts a force f  in the y-direction, and at the end of the y-arm
there acts an equal force in the x  -direction. Hence the torques acting in the opposite
directions are equal, so the lever is in rotational equilibrium and will not start rotating.

An inertial frame  moves in the negative x-direction with velocity  v relative to  .

In this frame the length of the x-arm is Lorentz contracted to L = 1 − v /c L , but2 2

the arm in the y-direction is unchanged. Furthermore, according to the transformation

Eq. (2.95) the force in the  x-direction is unchanged, but the force in the y-direction
is diminished to f y = 1 − v 2 /c2 f y . Hence in this frame there is a net torque
   
f  L  − 1 − v 2 /c2 f  1 − v 2 /c2 L  = v 2 /c2 f  L  , so one is inclined to conclude
that the lever will start rotating in the clockwise direction.
This is in conflict with the analysis in the rest frame of the lever, and this conflict
is what is called the lever paradox.
50 2 The Special Theory of Relativity

What will happen? Shall we trust the analysis in the rest frame of the lever or in
the laboratory frame where the lever moves?
The answer is related to the relativity of simultaneity and what we mean by a
physical object. If an extended physical object exists at a certain moment, it is made
up of a set of simultaneous events. Due to the relativity of simultaneity these events
are not simultaneous in a frame moving relatively to the first one. Hence the lever
in the laboratory frame and the rest frame of the lever consists of different sets of
events.
The set of simultaneous events in the rest frame of an object determines what will
happen to the object. We therefore have the rule: In order to determine what happens
to an extended body acted upon by forces one has to perform the calculations in the
rest frame of the body.
This is not in conflict with the principle of relativity. The rule does not introduce
any absolute velocity. If one transforms the description in the rest frame of the body
to the laboratory frame, there will appear unusual terms depending upon the relative
velocity of the object and the laboratory frame. This does not mean that the laws of
nature are different in different inertial reference frames. But it means that the laws
contain this relative velocity, due to the relativity of simultaneity, because the objects
that are described by simultaneity in different reference frames do not consist of the
same sets of events.

2.14 Tachyons

Particles cannot pass the velocity barrier represented by the velocity of light. How-
ever, the special theory of relativity permits the existence of particles that have always
moved with a velocity v > c. Such particles are called tachyons.
Tachyons have special properties that have been used in the experimental searches
for them. There is currently no observational evidence for the physical existence of
tachyons.
There are also certain theoretical difficulties with the existence of tachyons. The
special theory of relativity applied to tachyons leads to the following paradox. Using
a tachyon telephone a person, A emits a tachyon to B at a point of time t1 . B moves
away from A. The tachyon is reflected by B and reaches A before it was emitted; see
Fig. 2.15. If the tachyon could carry information it might bring an order to destroy
the tachyon emitter when it arrives back at A.
To avoid similar problems in regards to the energy exchange between tachyons and
ordinary matter, a reinterpretation principle is introduced for tachyons. For certain
observers a tachyon will move backwards in time, i.e. the observer finds that the
tachyon is received before it was emitted. Special relativity tells us that such a tachyon
is always observed to have negative energy.
According to the reinterpretation principle, the observer will interpret his observa-
tions to mean that a tachyon with positive energy moves forward in time. In this way,
2.14 Tachyons 51

Fig. 2.15 Tachyon paradox. A emits a tachyon at a point of time t1 . It is reflected at B and arrives
back at A at a point of time t2 . Note that the arrival event at A is later than the reflection event as
measured by B

one finds that the energy exchange between tachyons and ordinary matter proceeds
in accordance with the principle of causality.
However, the reinterpretation principle cannot be used to remove the problems
associated with the exchange of information between tachyons and ordinary matter.
The tachyon telephone paradox cannot be resolved by means of the reinterpreta-
tion principle. The conclusion is that if tachyons exist, they cannot be carriers of
information in our slowly moving world.

2.15 Magnetism as a Relativistic Second-Order Effect

Electricity and magnetism are described completely by Maxwell’s equations of the

electromagnetic field,

1
∇ ·E= ρq , (2.97)
ε0

∇ · B = 0, (2.98)

∂B
∇ ×E=− , (2.99)
∂t
52 2 The Special Theory of Relativity

1 ∂E
∇ × B = μ0 j + , (2.100)
c2 ∂t
together with Lorentz’s force law

F = q(E + v × B). (2.101)

However, the relation between the magnetic and the electric force was not fully
understood until Einstein had constructed the special theory of relativity. Only then
could one clearly see the relationship between the magnetic force on a charge moving
near a current-carrying wire and the electric force between charges.
We shall consider a simple model of a current-carrying wire in which we assume
that the positive ions are at rest while the conducting electrons move with the velocity
v. The charge per unit length for each type of charged particle is λ̂ = Sne where S
is the cross-sectional area of the wire, n the number of particles of one type per unit
length and e the charge of one particle. The current in the wire is

J = Snev = λ̂v. (2.102)

 

The wire is at rest in an inertial frame  . As observed in  it is electrically neutral.

Let a charge q move with a velocity u along the wire in the opposite direction of
the electrons. The rest frame of q is . The wire will now be described from  (see
Figs. 2.16 and 2.17).
Note that the charge per unit length of the particles as measured in their own rest
frames,  0 , is
 1/2
v2
λ0− = 1 − 2 λ̂, λ0+ = λ̂ (2.103)
c


Since the distance between the electrons is Lorentz contracted in  compared to

their distance in  0 .
The velocities of the particles as measured in  are

Fig. 2.16 Current carrying

wire seen from its own rest
frame
2.15 Magnetism as a Relativistic Second-Order Effect 53

Fig. 2.17 Current carrying

wire seen from the frame of a
moving charge

v+u
v− = − , v+ = −u. (2.104)
1 + uv/c2

The charge per unit length of the negative particles as measured in  is

 −1/2
v2
λ− = 1 − −2 λ0− . (2.105)
c

Substitution from Eqs. (2.103) and (2.104) gives

 uv 
λ− = γ 1 + 2 λ̂, (2.106)
c
 −1/2
where γ = 1 − u 2 /c2 . In a similar manner the charge per unit length of the
positively charged particles as measured in  is

λ+ = γ λ̂. (2.107)

Thus as observed in the rest frame of q the wire has a net charge per unit length
uv
λ = λ− − λ + = γ λ̂. (2.108)
c2
As a result of the different Lorentz contractions of the positive and negative ions
when we transform from their respective rest frames to , a current-carrying wire
which is electrically neutral in the laboratory frame is observed to be electrically
charged in the rest frame of the charge q.
As observed in this frame there is a radial electrical field with field strength

λ
E= . (2.109)
2π ε0 r

Then a force F acts on q this is given by

54 2 The Special Theory of Relativity

qλ λ̂v
F = qE = = γ qu. (2.110)
2π ε0 r 2π ε0 c2 r


If a force acts upon q as observed in  then a force also acts on q as observed in

. According to the relativistic transformation of a force component normal to the


direction of the relative velocity between  and , this force is


λ̂v
F = γ −1 F = qu. (2.111)
2π ε0 c2 r

Inserting J = λ̂v from Eq. (2.102) and using c2 = (ε0 μ0 )−1 , where μ0 is the
permeability of vacuum, we obtain

μ0 J
F= qu. (2.112)
2πr


This is exactly the expression obtained if we calculate the magnetic flux density
B around the current-carrying wire using Ampere’s circuit law

J
B = μ0 (2.113)
2πr
and use the force law Eq. (2.101) for a charge moving in a magnetic field
 

F = qu B . (2.114)

We have seen here how a magnetic force appears as a result of an electrostatic force
and the special theory of relativity. The considerations above have also demonstrated
that a force which is identified as electrostatic in one frame of reference is observed as
a magnetic force in another frame. In other words, the electric and the magnetic force
are really the same. What an observer names it depends upon his state of motion.

Exercises

2.1. Robb’s Lorentz invariant spacetime interval formula (A. A. Robb, 1936)

Show that the spacetime interval between the emission event at the√ point of time
t A and the reflection event at t B in Fig. 2.2 can be expressed as s = c t A tC , where
tC is the point of time when the reflected light signal arrives back at the emitter.
2.2. The twin paradox
On New Year’s day 2004, an astronaut (A) leaves Earth on an interstellar journey.
He is travelling in a spacecraft at the speed of v = (4/5)c heading towards Alpha
Exercises 55

Centauri. This star is at a distance of 4 ly (ly = light years) measured from the
reference frame of the Earth. As A reaches the star, he immediately turns around
and heads home. He reaches the Earth New Year’s day 2016 (in Earth’s time frame).
The astronaut has a brother (B), who remains on Earth during the entire journey. The
brothers have agreed to send each other a greeting every New Year day with the aid
of radio telescope.
(a) Show that A only sends 6 greetings (including the last day of travel), while B
sends 10.
(b) Draw a Minkowski diagram where A’s journey is depicted with respect to the
Earth’s reference frame. Include all the greetings that B is sending. Show with
the aid of the diagram that while A is outbound, he only receives one greeting,
while on his way home he receives nine.
(c) When does B receive signals from A?
(d) Show how the results from (b) and (c) can be deduced from the Doppler effect.
2.3. Faster than the speed of light?
The quasar 3C273 emits a jet of matter that moves with the speed v0 towards Earth
making an angle φ to the line of sight (see Fig. 2.18).
(a) Assume that two signals are sent towards the Earth simultaneously, one from
A and one from B. How much earlier will the signal from B reach the Earth
compared to that from A?
(b) Find an expression of the transverse distance that the emitted part has moved
when it reaches B. How much time (relative to the Earth) has this part been
travelling?
(c) Calculate the velocity v0 of the light source in terms of v and θ , and find the
value of v0 if v = 10c and θ = 10◦ . How large must v0 be in order that the
observed transverse velocity shall be larger than c?
2.4. Time dilation and Lorentz contraction

Fig. 2.18 Light cone due

to Cherenkov radiation
56 2 The Special Theory of Relativity

(a) At what speed does a clock move if its rate of time is 0.6 times the rate when it
is at rest?
(b) A rod moves in the x-direction. An observer following the rod measures that it
makes an angle π/4 with the x-axis. What is the speed of the rod if an observer
at rest on the x-axis finds that it makes an angle π/3 with the x-axis, due to the
Lorentz contraction of its length component in the x-direction?

2.5. Atmospheric mesons reaching the surface of the Earth

Atmospheric muons are formed when molecules in the Earth’s atmosphere are hit
by particles in the cosmic rays about L 0 = 10 km above the surface. The half-life of
a muon as measured by an observer co-moving with the muon is t0 = 1.56 × 10−6 s.
The average velocity of the muons are v = 0.98 c.
(a) According to a non-relativistic calculation, how many of ten million muons
formed at 10 km height reach the Earth’s surface?
(b) Taking the relativistic time dilation into account, how many will then reach the
surface of the Earth?
(c) How is this explained by an observer following the muons?

2.6. Relativistic Doppler shift

The relativistic formula for the Doppler effect of an object moving along the
direction of sight with a velocity v away from the observer is

1 + v/c
λr = λe ,
1 − v/c

where λe is the frequency measured by an observer co-moving with the object, and
λr is the frequency measured for the light received by an observer at rest. If the object
moves towards the observer the signs are interchanged.
For an object moving away from an observer there is an increase in the measured
wavelength—a redshift. The redshift of an object is defined as

λr − λe
z= .
λe

Hence the redshift due to the Doppler effect is

1 + v/c
z= − 1.
1 − v/c

Positive value of z means redshift and negative value blueshift.

The measured value of z for the centre of our neighbour galaxy, the Andromeda
Galaxy, is z = − 0.0004.
Exercises 57

Determine the velocity of the Andromeda Galaxy along the line of sight from this
measurement. Which way does the Andromeda Galaxy move relative to the Milky
Way?
2.7. The velocity of light in a moving medium
Light moves more slowly in a material medium than in empty space. The index
of refraction, n, of the medium is defined as the ratio of the velocity of light in the
medium when it is at rest, u 0 , and the velocity of light in empty space. Then
u0
n= .
c
We now consider a medium moving with a velocity v in the same direction as the
light. The speed of light in this medium relative to the laboratory frame is related to
the speed of light in a frame co-moving with the water by the relativistic velocity
addition law,

u0 + v
u= .
1 + uc02v

Find the velocity of light in a moving medium in terms of its index of refraction
and velocity.
2.8. Cherenkov radiation
When a particle moves through a medium with a velocity v greater than the
velocity of light in the medium, it emits a cone of radiation with a half-angle θ given
by cos θ = c/nv.
From Wikipedia: https://en.wikipedia.org/wiki/Cherenkov_radiation#/media/
File:Cherenkov.svg.
(a) What is the threshold kinetic energy (in MeV) of an electron moving through
water in order that it shall emit Cherenkov radiation? The index of refraction of
water is n = 1.3. The rest energy of an electron is m e c2 = 0.5.11 MeV.
(b) What is the limiting half-angle of the cone for high-speed electrons moving
through water?

2.9. Relativistic form of Newton’s 2 law

In order to simplify the calculation we shall consider motion along the x-direction
only. Let the particle have rest mass m 0 and velocity v. Its momentum is p = γ m 0 v,
 −1/2
where γ = 1 − v 2 /c2 .
The relativistic form of Newton’s 2 law is F = d p/dt, where F is the force acting
on the particle, and t is the coordinate time.
Calculate the form of this law as expressed in terms of v and dv/dt.
2.10. Lorentz transformation of electric and magnetic fields
58 2 The Special Theory of Relativity

It follows from the Lorentz transformations (2.27), (2.28) and (2.30) that the
partial derivatives transform as
   
∂ ∂ ∂ ∂ ∂ v ∂ ∂ ∂ ∂ ∂
=γ +v  , =γ + 2  , = , = .
∂t ∂t  ∂x ∂x ∂x c ∂t ∂y ∂y ∂z ∂z

Deduce the transformation equations for electric and magnetic fields by using
the transformation equations for the partial derivatives together with the requirement
that Maxwell’s equations shall be Lorentz invariant.

Reference

1. Frisch, D.H., Smith, J.H.: Measurement of the relativistic time dilation using µ-mesons. Am. J.
Phys. 31, 342–355 (1963)
Chapter 3
Vectors, Tensors and Forms

Abstract In this chapter we develop the main mathematical concepts used in this
book. First vectors, not only as quantities with length and direction, but as differential
operators. Then tensors of arbitrary rank are introduced. As a preparation for using
Cartan’s formalism we introduce forms, i.e. antisymmetric covariant tensors. This
antisymmetric tensor formalism is most effective when we introduce an orthonormal
basis field. Using an orthonormal basis co-moving with an observer, i.e. where the
time-like vector is equal to the 4-velocity of the observer, simplifies the physical
interpretation of the calculations.

3.1 Vectors

An expression of the form a μ eμ , where a μ , μ = 1, 2, …, n are real numbers, is

known as a linear combination of the vectors eμ .
The vectors e1 , . . . , en are said to be linearly independent if there does not exist
real numbers a μ = 0 such that a μ eμ = 0 (Fig. 3.1).
This has a geometrical interpretation: A set of vectors are linearly independent if
it is not possible to construct a closed polygon of the vectors (even by adjusting their
lengths).
A set of vectors e1 , . . . , en are said to be maximally linearly independent if
e1 , . . . , en , v are linearly dependent for all vectors v = eμ , i.e. if there exist real
numbers a μ such that v + a μ eμ = 0. Hence the vector v may be written as a linear
combination of the vectors eμ ,

v = v μ eμ . (3.1)

We define the dimension of a vector-space as the number of vectors in a maximally

linearly independent set of vectors of the space. The vectors eμ in such a set are known
as the basis vectors of the space. The numbers v μ are called the components of the
vector v in this basis.

© Springer Nature Switzerland AG 2020 59

Ø. Grøn, Introduction to Einstein’s Theory of Relativity,
Undergraduate Texts in Physics, https://doi.org/10.1007/978-3-030-43862-3_3
60 3 Vectors, Tensors and Forms

Fig. 3.1 Closed polygon (linearly dependent vectors)

3.1.1 Four-Vectors

Four-vectors (also denoted 4-vectors) are vectors which exist in (4-dimensional)

spacetime. A 4-vector equation represents four independent component equations.
As a first example of a 4-vector we shall consider the 4-velocity, i.e. the velocity
of a particle through spacetime.

Definition 3.1.1 (Four-velocity) The four-velocity of a particle is a 4-vector with

components equal to the derivatives of the coordinates with respect to the proper
time τ of the particle,

dx μ
u = u μ eμ = eμ . (3.2)
dτ
One often writes the 4-velocity in the following way
 
u = u 0 , u 1 , u 2 , u 3 , (3.3)

where u 0 is the time component of the 4-velocity, and u i , i = 1, 2, 3 are the spatial
components.
The proper time of the particle is the time measured by a standard clock carried
by the particle. From Eq. (2.45) we have
3.1 Vectors 61

dt 1
= = γ, (3.4)
dτ 1 − v 2 /c2

where v is the ordinary velocity through 3-dimensional space. Note that ν is a rel-
ative velocity which can be transformed away by referring to the rest frame of the
particle. On the other hand the 4-velocity, which is a vector in spacetime, cannot be
transformed away. In the rest frame of the particle it still has a non-vanishing time
component.
 0With reference to a Cartesian coordinate system with coordinates
x = ct, x 1 = x, x 2 = y, x 3 = z the 4-velocity has the component form

dt dx dy dz
u = c et + ex + ey + ez . (3.5)
dτ dτ dτ dτ
This may be written
 
dt dx dy dz
u = cet + ex + ey + ez . (3.6)
dτ dt dt dt

The ordinary velocity through 3-space is

dx dy dz
v = ex + ey + ez . (3.7)
dt dt dt
Using Eqs. (3.4) and (3.7) the 4-velocity can be written

u = γ (cet + v), (3.8)

or
 
u = γ c, vx , v y , vz . (3.9)

In a co-moving reference frame where the particle is at rest, the 3-velocity vanishes
and γ = 1. In this frame the four-velocity is

u = cet . (3.10)

In its rest frame the particle moves only in the time direction.

Definition 3.1.2 (Four-momentum) The four-momentum of a particle is equal to the

rest mass of the particle times its four-velocity,

P = m 0 u, (3.11)

where m 0 is the rest mass of the particle. The components of the 4-momentum are
62 3 Vectors, Tensors and Forms

P = (E/c, p), p = γ m 0 v = m v, (3.12)

where E is the relativistic energy of the particle.

Definition 3.1.3 (Minkowski force) The Minkowski force acting on a particle is equal
to the derivative of the 4-momentum with respect to the proper time of the particle,

F = d P/dτ.
 (3.13)

Its components are

 
 1  
F =γ f · v, f , (3.14)
c

where

d p
f = (3.15)
dt
is the ordinary force.

Definition 3.1.4 (Four-acceleration) The four-acceleration of a particle is equal to

the derivative of its four-velocity with respect to its proper time,

dU
A = . (3.16)
dτ
In the general theory of relativity gravitation is not considered a force. A particle
in free fall is in Newtonian gravitational theory said to be influenced by a gravitational
force. According to the general theory of relativity the particle is not influenced by
any force. Such a particle has no 4-acceleration. The equation A = 0 implies that
the particle is not in free fall. It is then influenced by non-gravitational forces. It
is important to distinguish between the ordinary 3-acceleration, which represents
the acceleration of a particle through 3-space relative to an observer, and the 4-
acceleration which represents deviation from free fall. In the context of the general
theory of relativity the words “a non-accelerating particle” usually mean “a particle
which is in free fall”.

3.1.2 Tangent Vector Fields and Coordinate Vectors

In a curved space position vectors with finite length do not exist (see Fig. 3.2).
Different points in a curved space have different tangent planes. Finite vectors do
only exist in these tangent planes (see Fig. 3.3). However, infinitesimal position
vectors d
r do exist.
3.1 Vectors 63

Fig. 3.2 No finite position vector in curved space. Vectors can only exist in tangent planes. The
vectors in the tangent plane of N do not contain the vector NP (dashed line)

Fig. 3.3 Vectors in tangent planes. In curved space vectors can only exist in tangent planes. The
figure shows the tangent plane of a point P in a curved surface
64 3 Vectors, Tensors and Forms

We shall now define the concepts reference frame, coordinate system, co-moving
coordinate system, orthonormal basis and coordinate basis vector.
Definition 3.1.5 (Reference frame) A reference frame is a set of particles with spec-
ified motion. An inertial reference frame is a frame in which Newton’s first law is
valid. In the general theory of relativity, where gravity is not a force, this means that
an inertial reference frame is a non-rotating set of free particles.
Definition 3.1.6 (Coordinate system) A coordinate system is a continuum of 4-tuples
giving a unique set of coordinates for events in spacetime.
Definition 3.1.7 (Co-moving coordinate system) A co-moving coordinate system
in a frame is a coordinate system where the particles in the reference frame have
constant spatial coordinates.
 
Definition 3.1.8 (Orthonormal basis) An orthonormal basis eμ̂ in spacetime is
defined by

etˆ · etˆ = −1(c = 1), eî · e ĵ = δî ĵ , (3.17)

where î and ĵ are space indices.

Definition 3.1.9 (Preliminary definition of coordinate basis vector) Our preliminary
definition is: A coordinate basis vector is the partial derivative of the position vector
with respect to a coordinate,

∂ r
eμ = . (3.18)
∂xμ
A vector field is a continuum of vectors, where the components are continuous
and differentiable functions of the coordinates. Let u be a tangent vector field to a
curve with parameter λ (coordinate along the curve). Then

d
r
u = . (3.19)
dλ
The position vector of a point on the curve is a function of the coordinates
which are again functions of the curve parameter, r = r[x μ (λ)]. The chain rule
for differentiation then yields

∂ r dx μ dx μ
u = = eμ = u μ eμ . (3.20)
∂ x μ dλ dλ
Thus, the components of the tangent vector field along a curve parameterized by
λ are given by

dx μ
uμ = . (3.21)
dλ
3.1 Vectors 65

In the theory of relativity the invariant parameter of the world line of a particle is
usually chosen to be the proper time of the particle,

dx μ
uμ = . (3.22)
dτ
This means that the tangent vector field of the world line of a particle is made up
of the four-velocity of the particle.

3.1.3 Coordinate Transformations

 
Let there be given two coordinate systems, {x μ } and x μ covering a region of
spacetime with basis vectors

∂ r ∂ r
eμ = , eμ = μ . (3.23)
∂xμ ∂x
Suppose there exists a coordinate transformation such that the primed coordi-

nates a functions of the unprimed, x μ (x μ ), and an
 inverse transformation such

that the unprimed are functions of the primed, x μ x μ . Applying the chain rule
of differentiation we then obtain

∂ r ∂ r ∂ x μ ∂xμ
eμ =  =  = 
eμ . (3.24)
∂xμ ∂xμ ∂xμ ∂ x μ

This is the transformation equation for the basis vectors. The quantities ∂ x μ /∂ x μ
are the elements of the transformation matrix. Indices that are not summation indices
are called free indices. We have the following rule: In all terms on each side of an
equation the free indices should appear in the same way (as an upper index or a lower
index).
Applying this rule we can now find the inverse transformation

∂xη
eμ = eμ μ . (3.25)
∂x
A vector itself is invariant. Only its components and the basis vectors transform.
Hence,

  ∂xμ
v = v μ eμ = v μ eμ = v μ eμ . (3.26)
∂ x μ
Thus, the components of a vector transform as follows
66 3 Vectors, Tensors and Forms


μ ∂xμμ μ μ ∂x
μ
v =v  , v = v . (3.27)
∂xμ ∂xμ
The directional derivative along a curve parameterized by λ is

d ∂ dx μ ∂
= μ = vμ μ , (3.28)
dλ ∂ x dλ ∂x
where v μ = dx μ /dλ are the components of the tangent vector of the curve. The
directional derivative along an x ν -axis is found by inserting λ = x ν . This gives

∂ ∂xμ ∂ ∂
= δνμ μ = ν . (3.29)
∂xμ ∂xν ∂x ∂x
Hence, the directional derivative along a coordinate axis is equal to the par-
tial derivative with respect to the coordinate of the axis. From the chain rule of
differentiation we get in the primed coordinate system

∂ ∂xμ ∂
 = . (3.30)
∂x μ ∂ x μ ∂ x μ
Comparing with Eq. (3.18) shows that the partial derivatives transform in the same
way as the basis vectors.
A weakness of the preliminary definition (3.18) of basis vectors is that it involves
the position vectors that are not defined in a curved space, only in the tangent plane
of the space. We would therefore like to have a general definition of basis vectors
not involving the position vector. The transformation (3.30) of the partial derivatives
motivates to the following definition.

Definition 3.1.10 (General definition of coordinate basis vectors) We define the

coordinate basis vectors as partial derivatives,

∂
eμ = . (3.31)
∂xμ
This is the general definition of coordinate basis vectors. It applies in curved as
well as in flat spaces.
Since an arbitrary vector can be written as a linear combination of basis vectors,
we now have a new way of thinking about vectors. From now on we can think of a
vector as a differential operator, not only a quantity with magnitude and direction,
i.e. an arrow.

Example 3.1.1 (Transformation between Cartesian- and plane polar coordinates)

The basis vectors of a Cartesian coordinate system and those of plane polar coordi-
nates are shown in Fig. 3.4. We see from the figure that the coordinates are related
by the transformation
3.1 Vectors 67

Fig. 3.4 Basis vectors in

Cartesian- and plane polar
coordinates

x = r cos θ, y = r sin θ (3.32)

The transformation of the basis vectors is found from the chain rule for
differentiation
∂ ∂x ∂ ∂y ∂
er = = + = cos θ ex + sin θ ey ,
∂r ∂r ∂ x ∂r ∂ y
∂ ∂x ∂ ∂y ∂
eθ = = + = −r sin θ ex + r cos θ ey . (3.33)
∂θ ∂θ ∂ x ∂θ ∂ y

The basis vectors in the Cartesian coordinate system are unit vectors. The
magnitudes of the basis vectors in the plane polar coordinate system are
 
|er | = er · er =
cos2 θ + sin2 θ = 1,
 
|eθ | = eθ · eθ = r 2 cos2 θ + r 2 sin2 θ = r. (3.34)

This shows that the coordinate basis vectors need not be unit vectors.
Definition 3.1.11 (Orthonormal basis) An orthonormal basis is a vector basis
consisting of unit vectors that are normal to each other.
We shall denote orthonormal basis by writing a hat over the indices. Note that an
orthonormal basis will not in general be a coordinate basis, i.e. the basis vectors of
an orthonormal basis are not just partial derivatives.
The orthonormal basis of the plane polar coordinate system is

1
er̂ = er , eθ̂ = eθ . (3.35)
r
Example 3.1.2 (Relativistic Doppler effect) Consider a photon with energy E and
momentum p in the rest frame of the emitter and the observer. The energy of the
68 3 Vectors, Tensors and Forms

photon as observed in a frame moving with velocity v away from the observer is found
from the transformation Eq. (2.85). Inserting the energy E = hν and momentum
p = hν/c in the rest frame of the emitter gives
v

ν = γ ν − ν , (3.36)
c
Giving

c−v
ν = ν. (3.37)
c+v

This change of observed frequency depending upon the observed velocity of the
emitter is often called the Doppler shift.

3.1.4 Structure Coefficients

Let us first define a mathematical quantity which is essential in connection with the
structure coefficients.

Definition 3.1.12 (Commutator of vectors) The commutator of two vectors is


u, v = uv − vu, (3.38)

where uv is defined as

uv = u μ eμ (v ν eν ). (3.39)

Hence, uv means that u acts upon v as a differential operator. The commutator of
two vectors is itself a vector.
ρ
Definition 3.1.13 (Structure coefficients) The structure coefficients cμν are the
components of the commutators of the basis vectors,
 ρ
eμ , eν = cμν eρ . (3.40)

It follows that the structure coefficients are antisymmetric in their lower indices,
ρ ρ
cνμ = −cμν . (3.41)

Calculating the structure coefficients in a coordinate basis we get

 
∂ ∂ ∂2 ∂2
μ
, ν = − = 0. (3.42)
∂x ∂x ∂ x μ∂ x ν ∂xν∂xμ
3.1 Vectors 69

Hence the structure coefficients vanish in coordinate basis.

Example 3.1.3 (Structure coefficients in plane polar coordinates) We shall calcu-

late the structure coefficients of an orthonormal basis in a system of plane polar
coordinates. Using Eq. (3.35) we have
   
 ∂ 1 ∂ ∂ 1 ∂ 1 ∂
er̂ , eθ̂ = , = −
∂r r ∂θ ∂r r ∂θ r ∂θ
1 ∂ 1 ∂2 1 ∂2 1 1
=− 2 + − = − 2 eθ = − eθ̂ . (3.43)
r ∂θ r ∂r ∂θ r ∂θ ∂r r r
Since
 ρ̂
er̂ , eθ̂ = cr̂ θ̂ eρ̂ , (3.44)

and using the antisymmetry (3.41), we get the only non-vanishing structure
coefficients in the orthonormal basis field of a plane polar coordinate system

1
cr̂θ̂θ̂ = −cθ̂θ̂ r̂ = − . (3.45)
r

3.2 Tensors
 
Definition 3.2.1 (One-form basis) A one-form basis ω1 , . . . , ωn is defined by
letting the basis forms act upon the basis vectors in an operation called contraction,
according to the rule

ωμ (eν ) = δνμ . (3.46)

An arbitrary one-form can be expressed as a linear combination of the basis forms

α = αμ ωμ , (3.47)

where αμ are the components of α is the given basis. Using Eqs. (3.46) and (3.47)
we find

α(eν ) = αμ ωμ (eν ) = αμ δνμ = αν . (3.48)

This means that the contraction of a one form with a basis vector gives the compo-
nent of the one-form corresponding to the basis vector. The contraction of a one-form
α with a vector v gives
70 3 Vectors, Tensors and Forms
   
v ) = α v μ eμ = v μ α eμ = v μ αμ .
α( (3.49)

This shows that the contraction of a one-form with a vector corresponds to the
scalar product of two vectors.
We shall now need the concept multilinear function which has the following
meaning. A multilinear function is a function which is linear in all its arguments.
Then we are ready to define tensors.

Definition 3.2.2 (Tensors) A tensor is a multilinear function which maps one-forms

and vectors into real numbers.
We have three different types of tensors:
• A covariant tensor maps vectors
• A contravariant tensor maps forms
• A mixed tensor maps tensors and forms.
 
N
A tensor of rank maps N one-forms and N  vectors into real numbers. It
N
is, however, usual to say that the tensor has a rank N + N  . A tensor of rank zero
is called a scalar quantity and may be given in terms of real numbers. A vector is a
contravariant tensor of rank one, and a one-form is a covariant tensor of rank one.

Definition 3.2.3 (Tensor product) Let T and S be two covariant tensors of rank m
and n. The tensor product of T and S is

u 1 , . . . , um , v1 , . . . , vn ) = T (

T ⊗ S( u 1 , . . . , um )S(
v1 , . . . , vn ). (3.50)

T ⊗ S is a tensor of rank m + n. Let R = T ⊗ S. Then we have

R = Rμ1 ...μq ωμ1 ⊗ ωμ2 ⊗ · · · ⊗ ωμq . (3.51)

Notice that S ⊗ T = T ⊗ S. The components of a tensor are found by contracting

the tensor with the basis elements, i.e. basis vectors, in the case of a covariant tensor,
 
Rμ1 ...μq = R eμ1 , . . . , eμq . (3.52)

The indices of the components of a covariant tensor are written as lower indices
as in Eq. (4.52), and the indices of the components of a contravariant tensor as upper
indices.

Example 3.2.1 (Tensor product of two vectors) Let u and v be two vectors and α
and β two one-forms

u = u μ eμ , v = v μ eμ , α = αμ ωμ , β = βμ ωμ . (3.53)

3.2 Tensors 71

From these we can construct a contravariant tensor of rank 2 by R = u ⊗ v with

components
       
R μ1 μ2 = R ωμ1 , ωμ2 = u ⊗ v ωμ1 , ωμ2 = u ωμ1 v ωμ2 = u μ1 v μ2 . (3.54)

3.2.1 Transformation of Tensor Components

We shall not limit the discussion to coordinate

 transformations
   but will consider
transformations between any type of bases, eμ → eμ . The elements of the
μ
transformation matrices are denoted by Mμ . Then the transformations of the basis
vectors are written
μ 
eμ = eμ Mμ , eμ = eμ Mμμ . (3.55)

It follows that
μ 
Mμ Mνμ = δνμ . (3.56)

For a coordinate transformation the elements of the transformation matrix are


 ∂xμ
Mμμ = . (3.57)
∂xμ

3.2.2 Transformation of Basis One-Forms

The basis 1-forms transform inversely relatively to the basis vectors

  μ 
ωμ = Mμμ ωμ , ωμ = Mμ ωμ . (3.58)

The components of a tensor of higher rank transform such that every contravariant
index (upper) transforms as a basis 1-form and every covariant index (lower) as a
basis vector. Also, all elements of the transformation matrix are multiplied with one
another.

Example 3.2.2 (A mixed tensor of rank 3) Tensor components transform homoge-

neously, which means that the transformed tensor components are linear combina-
tions of the original ones. The components of a mixed tensor of rank 3, for example,
transform as follows.
  μ
Tμα ν  = Mαα Mμ Mνν Tμν
α
. (3.59)
72 3 Vectors, Tensors and Forms

Tensor transformation of components means that tensors have a basis-independent

existence. That is, if a tensor has non-vanishing components in a given basis then it
has non-vanishing components in all bases. This means that tensor equations have
a basis-independent form. Tensor equations are invariant. A basis transformation
might result in the vanishing of one or more tensor components. Equations written
in component form will differ from one basis to another. An equation expressed in
terms of tensor components can be transformed from one basis to another using the
tensor component transformation rules. An equation that is expressed only in terms
of tensor components is said to be a covariant equation. An equation expressed in
terms of the tensors themselves has the same form in every coordinate system and is
said to be an invariant equation.

3.2.3 The Metric Tensor

Definition 3.2.4 (The scalar product) The scalar product of two vectors u and v is
denoted by g(u , v) and is defined as a symmetric linear mapping which for each pair
of vectors gives a scalar, i.e. a number. The symmetry means that g(v , u) = g(
u , v).

Definition 3.2.5 (The metric tensor) The metric tensor, g = gμν ωμ ⊗ ων , is a

covariant symmetric tensor of rank 2 with components made up of the scalar products
of the basis vectors,
 
gμν = g eμ , eν = eμ · eν . (3.60)
 
The values of the scalar products g eμ , eν are given by specifying the scalar
products of each pair of basis vectors in a basis. The symmetry of the metric tensor
means that its components obey

gνμ = gμν . (3.61)

The components of the metric tensor are often called the metric and are written
as a 2 × 2 matrix. Those with equal indices are on the diagonal of this matrix. If
they are the only non-vanishing indices, the metric is said to be diagonal. It follows
from Eq. (3.60) that the metric is diagonal if the basis vectors are orthogonal to each
other.
The scalar product of two arbitrary vectors u and v is
 
u , v) = g(u μ e, v ν eν ) = u μ v ν g eμ , eν = u μ v ν gμν .
u · v = g( (3.62)

The usual notation is

u · v = gμν u μ v ν . (3.63)
3.2 Tensors 73

The absolute value of a vector is

  
|
v| = v , v) =
g( gμν v μ v ν . (3.64)

The scalar product of two vectors is an invariant, meaning that is has the same
value in every coordinate system.
Example 3.2.3 (Cartesian coordinates in a plane) In the present case the scalar
products of the basis vectors are

ex · ex = ey · ey = 1, ex · ey = ey · ex = 0. (3.65)

Hence

gx x = g yy = 1, gx y = g yx = 0. (3.66)

This is often written in matrix form

 
10
gμν = . (3.67)
01

Example 3.2.4 (Plane polar coordinates) For this coordinate system the scalar
product of the basis vectors is

er · er = 1, eθ · eθ = r 2 , er · eθ = 0. (3.68)

Hence, in plane polar coordinates the metric tensor has the components
 
1 0
gμν = . (3.69)
0 r2

Example 3.2.5 (Non-orthogonal basis vectors) We consider a skew angled coordi-

nate system where the basis vectors make an angle θ with each other as shown in
Fig. 3.5.
For this coordinate system the scalar products of the basis vectors are

Fig. 3.5 Basis vectors in a

skew angled coordinate
system. Basis vectors e1 and
e2
74 3 Vectors, Tensors and Forms

e1 · e1 = e2 · e2 = 1, e1 · e2 = e2 · e1 = cos θ. (3.70)

Hence, the metric tensor has the components

 
1 cos θ
gμν = . (3.71)
cos θ 1

Definition 3.2.6 (Contravariant components of the metric tensor) The contravariant

components of the metric tensor, g μν , are defined by

g μα gαν = δνμ . (3.72)

The contravariant components of the metric tensor make up the inverse matrix
relative to the matrix made up of the covariant components. It follows from Eq. (3.72)
that the mixed metric tensor can be thought of as a unit tensor of rank 2,

I = δνμ ων ⊗ eμ = ωμ ⊗ eμ . (3.73)

 
Note that I( u ) = ωμ ⊗ eμ ( u ) = ωμ (
u )eμ = u μ eμ = u, i.e. the unit tensor
applied to a vector gives out the vector.
It is possible to define a mapping between tensors of different type, i.e. covariant
or contravariant, using the metric tensor. An example with tensors of rank one is
shown in Fig. 3.6.
We can for example map a vector on a one-form
     
vμ = g v, eμ = g v α eα , eμ = v α g eα , eμ = v α gαμ = gαμ v α . (3.74)

This is known as lowering an index. Raising of an index is made according to

v μ = g μα vα . (3.75)

We shall now define distance in spacetime along a curve. Let the curve be
parametrized by λ (proper time for a time-like curve) with a tangent vector field
v. The squared distance ds 2 between the points along the curve is defined as

ds 2 = g(
v , v)dλ2 . (3.76)

Using Eq. (3.56) we get

ds 2 = gμν v μ v ν dλ2 . (3.77)

The tangent vectors have the components v μ = dx μ /dλ, which gives

ds 2 = gμν dx μ dx ν . (3.78)
3.2 Tensors 75

Fig. 3.6 Covariant and contravariant components of a vector

The expression (3.78) is known as the line-element.

Example 3.2.6 (Line element in Cartesian coordinates) Inserting the metric com-
ponents (3.66) into Eq. (3.78) we obtain the line element of a plane in a Cartesian
coordinate system

ds 2 = dx 2 + dy 2 . (3.79)

Example 3.2.7 (Line element in plane polar coordinates) With the metric compo-
nents (3.61) we get

ds 2 = dr 2 + r 2 dθ 2 . (3.80)

The line element in (flat) four-dimensional Minkowski spacetime with Cartesian

spatial coordinates is required to be Lorentz invariant. It may be shown that the
line-element then takes the form
76 3 Vectors, Tensors and Forms

ds 2 = −c2 dt 2 + dx 2 + dy 2 + dz 2 . (3.81)

In this case the metric tensor is often called the Minkowski metric, and its com-
ponents are denoted by ημν . Introducing a time coordinate x0 = ct we obtain the
Minkowski metric,
⎛ ⎞
−1 0 0 0
⎜ 0 1 0 0⎟
ημν =⎜
⎝ 0
⎟. (3.82)
0 1 0⎠
0 0 0 1

The Minkowski line-element is then written

ds 2 = ημν dx μ dx ν . (3.83)

An orthonormal basis field can be used in curved as well as in flat spacetime.

Often one uses a so-called co-moving orthonormal basis with an observer. This
means that we have an orthonormal basis field along the world line of the observer
with time-like vector equal to the four-velocity of the observer divided by the velocity
of light, etˆ = u/c. In every orthonormal basis field there is Minkowski metric with
components (3.82) whether spacetime is flat or curved, and the line-element takes
the form

ds 2 = ημ̂ν̂ dx μ̂ dx ν̂ . (3.84)

Example 3.2.8 (The four-velocity identity) Let us calculate the scalar product of the
four-velocity by itself, using Eqs. (3.9) and (3.82),
 
u · u = ημ̂ν̂ u μ̂ u ν̂ = −γ 2 c2 − v 2 = −c2 . (3.85)

This is called the four-velocity identity. In an arbitrary basis it takes the form

gμν u μ u ν = −c2 . (3.86)

3.3 The Causal Structure of Spacetime

The causal structure of spacetime can be illustrated by considering the light cone
(Fig. 3.7).
Let us recapitulate some important points. The world lines of material particles
or an observer, moving slower than light, are inside the light cone. Such curves are
called time-like. The invariant parameter of a time-like curve is usually chosen to be
3.3 The Causal Structure of Spacetime 77

Fig. 3.7 Causal structure of spacetime From: https://en.wikipedia.org/wiki/Light_cone#/media/

File:World_line.svg

the proper time τ of an observer following the curve. Then a tangent vector of the
curve is the 4-velocity of the observer.
A point in spacetime represents an event. The distance in spacetime between two
infinitesimally nearby points in spacetime is called an interval. A time-like interval
is the interval between two points on a time-like curve. It has dstime2
< 0.
We shall now give a general physical interpretation of the line element for time-
like intervals. For this purpose it is sufficient to consider the Minkowski line-element
which can be written
    2  2 
1 dx 2 dy dz
ds = −c 1 − 2
2 2
+ + dt 2 . (3.87)
c dt dt dt

Consider a particle moving with a coordinate velocity

dx dy dz
v = v x ex + v y ey + v z ez = ex + ey + ez . (3.88)
dt dt dt
Then
78 3 Vectors, Tensors and Forms
   
(v x )2 + (v y )2 + (v z )2 2 2 v2 2 2
ds = − 1 −
2
c dt = − 1 − 2 c dt . (3.89)
c2 c

From the special theory of relativity we know that the time measured by a standard
clock following the particle, i.e. the proper time of the particle, is

v2
dτ = 1− dt. (3.90)
c2

Hence we obtain the general physical interpretation of the line-element for a

time-like interval

ds 2 = −c2 dτ 2 . (3.91)

This means that a time-like interval in spacetime is measured by a clock, it is

essentially a time interval.
We shall later define geodesic curves as the straightest possible curves between
two events in spacetime. In flat Minkowski spacetime they are straight. It will also
be shown that geodesic curves have extremal length between two events.
For light the velocity is v = c. Then the proper time vanishes. The world line of
light moving freely is called light-like, and an interval along the world line of light
2
is called light-like. Hence the interval along a light-like curve vanishes, dslight = 0.
It is therefore also called a zero-interval.
A spacelike curve represents the world line of a particle moving faster than light.
The interval between two events on such a curve is called a space-like interval and
2
has dsspace > 0.

3.4 Forms

Definition 3.4.1 (Antisymmetric tensor) An antisymmetric tensor is a tensor whose

sign changes under an arbitrary exchange of two arguments,

A(. . . , u, . . . , v, . . .) = A(. . . , v, . . . , u, . . .). (3.92)

The components of an antisymmetric tensor changes sign under exchange of two

indices,

A . . . μ . . .ν . . . . . . = A . . .ν . . . μ . . . (3.93)

Definition 3.4.2 (p-form) A p-form is defined to be an antisymmetric, covariant

tensor of rank p.
3.4 Forms 79

Definition 3.4.3 (The wedge product) The wedge product is an antisymmetric tensor
product is defined by

( p + q)! [μ1
ω[μ1 ⊗ · · · ⊗ ωμ p ] ∧ ω[ν1 ⊗ · · · ⊗ ων p ] = ω ⊗ · · · ⊗ ωνq ] , (3.94)
p!q!

where [ ] antisymmetric combination, which is defined by

⎛ ⎞
the sum of terms with all possible permutations
1⎝
ω[μ1 ⊗ · · · ⊗ ωμ p ] = of indices with plus for even and minus for an odd ⎠.
p!
number of permutations
(3.95)

Example 3.4.1 (Antisymmetric combinations) The antisymmetric combination of a

tensor product of two basis form is

1  μ1 
ω[μ1 ⊗ ωμ2 ] = ω ⊗ ω μ2 − ω μ2 ⊗ ω μ1 . (3.96)
2
The antisymmetric combination of a tensor product of three basis form is

1 μ1
ω[μ1 ⊗ ωμ2 ⊗ ωμ3 ] = (ω ⊗ ωμ2 ⊗ ωμ3 + ωμ3 ⊗ ωμ1 ⊗ ωμ2 + ωμ2 ⊗ ωμ3 ⊗ ωμ1
6
− ωμ2 ⊗ ωμ1 ⊗ ωμ3 + ωμ3 ⊗ ωμ2 ⊗ ωμ1 + ωμ1 ⊗ ωμ3 ⊗ ωμ2 ).
(3.97)

Example 3.4.2 (A 2-form in a 3-space) In a 3-dimensional space a 2-form may be

written in component form as

α =α12 ω1 ⊗ ω2 + α21 ω2 ⊗ ω1 + α13 ω1 ⊗ ω3 + α31 ω3 ⊗ ω1

α23 ω2 ⊗ ω3 . (3.98)

The antisymmetry of α means that

α21 = −α12 , α31 = −α13 , α32 = −α23 . (3.99)

Hence
   
α = α12 ω1 ⊗ ω2 − ω2 ⊗ ω1 + α13 ω1 ⊗ ω3 − ω3 ⊗ ω1
 
+ α23 ω2 ⊗ ω3 − ω3 ⊗ ω2 , (3.100)

Using Einstein’s summation convention this may be written as

α = α|μν| 2ω[μ ⊗ ων] , (3.101)

80 3 Vectors, Tensors and Forms

where |μν| means summation with μ < ν, and the bracket denotes antisymmetriza-
tion. We now use the definition (3.94) of the wedge product ∧ with p = q = 1. This
gives

α = α|μν| ωμ ∧ ων , (3.102)

which is often written as

1
α= αμν ωμ ∧ ων . (3.103)
2
A tensor of rank 2 can always be split up into a symmetric and an antisymmetric
part,

1  1 
Tμν = Tμν − Tνμ + Tμν + Tνμ = Aμν + Sμν , (3.104)
2 2
where
1  1 
Aμν = Tμν − Tνμ , Sμν = Tμν + Tνμ . (3.105)
2 2
We thus have
1 
Sμν Aμν = Tμν + Tνμ (T μν − T νμ )
4
1 
= Tμν T μν − Tμν T νμ + Tνμ T μν − Tμν T νμ = 0. (3.106)
4
This shows that summation over the indices of a product of a symmetric and an
antisymmetric quantity vanishes. In a summation Tμν Aμν , where Tμν has no symme-
try, and Aμν is antisymmetric, only the antisymmetric part of Tμν contributes. Hence,
in the expression (3.103) only the antisymmetric combinations α[μν] contribute to
the summation. These antisymmetric combinations are the form-components.
Forms are antisymmetric covariant tensors. Because of this antisymmetry a form
with two equal component indices vanishes, α[μ...μ...] = 0. This implies that on an n-
dimensional space all p-forms with p > n vanish. Hence on a 2-dimensional surface
there exist only up to 2-forms, and in four-dimensional spacetime there exist only
up to 4-forms.

3.4.1 The Volume Form

The antisymmetric Levi-Civita symbol is defined by

3.4 Forms 81
⎧
⎨ 1 if μ1 . . . μn is an even permutation of 1 . . . n
εμ1 ...μn = −1 if μ1 . . . μn is an odd permutation of 1 . . . n . (3.107)
⎩
0 otherwise

It follows that εμ1 ...μn if two indices are equal.

The determinant of an n × n-matrix A with elements Aμν may be written

A = det(A) = εμ1 ...μn A1μ1 A2μ2 . . . Anμn . (3.108)

For example for n = 2 this equation gives

A = εμ1 η2 A1μ1 A2μ2 = ε12 A11 A22 + ε21 A12 A21 = A11 A22 − A12 A21 . (3.109)

We shall now consider

 an n-dimensional space with a metric tensor having com-
ponents gμν . Let ωμ̂ be an orthonormal form basis. The volume form V is defined
by

V = ω1̂ ∧ · · · ∧ ωn̂ . (3.110)

μ
Let M μ̂ be the elements of the transformation matrix to an arbitrary basis ωμ =
μ μ̂
Mμ̂ ω . Then

V = Mμ1̂ 1 . . . Mμn̂ n ωμ1 ∧ · · · ∧ ωμn = Mμ1̂ 1 . . . Mμn̂ n εμ1 ...μn ω1 ∧ · · · ∧ ωn

= Mω1 ∧ · · · ∧ ωn (3.111)

where M is the determinant of the transformation matrix.

It follows from the general Eq. (3.59) for the components of a tensor that the
transformation from the components of the metric tensor in an orthonormal basis to
the components in an arbitrary basis is

gμν = Mμμ̂ Mνν̂ gμ̂ν̂ . (3.112)

Since the determinant of a matrix is equal to the determinant of the transposed

matrix (rows and columns interchanged), it follows from Eq. (3.113) that the corre-
sponding transformation of the determinant made up of the components of the metric
tensor is

g = M 2 ĝ. (3.113)
82 3 Vectors, Tensors and Forms

where ĝ is the determinant made by the components of the metric tensor in an

orthonormal basis. In usual 3-space ĝ = 1, and in 4-dimensional spacetime ĝ = − 1.
Inserting the positive square root of (3.113) into Eq. (3.111) gives

V = |g|ω1 ∧ · · · ∧ ωn . (3.114)

where |g| is the absolute value of the determinant made by the components of the
metric tensor. The tensor components of the volume form are

Vμ1 ...μn = |g|εμ1 ...μn . (3.115)

The volume form represents an invariant volume-element. The corresponding

invariant distance in the μ-direction is
 
Lμ = gμμ ωμ . (3.116)

3.4.2 Dual Forms

The dual of a p-form α in an n-dimensional space is denoted by α, where  is called

Hodge’s star operator, and is defined by

1
α = Vν ...ν μ ...μ α ν1 ...ν p ωμ1 ∧ · · · ∧ ωμn− p . (3.117)
p!(n − p)! 1 p 1 n− p

The dual of an orthogonal basis p-form is

√
  1 |g|
 ω ν1 ∧ · · · ∧ ω ν p = εν ...ν μ ...μ ωμ1 ∧ · · · ∧ ωμn− p , (3.118)
(n − p)! g p 1 p 1 n− p

where g p is the determinant of the metric tensor associated with the space of the
p-form α, and g is the determinant made up of the components of the metric tensor
in the n-dimensional space.

Example 3.4.3 (Duals of basis forms in a spherical coordinate system in Euclidean

3-space) The transformation from spherical coordinates (r, θ, φ) to Cartesian
coordinates (x, y, z) is

x = r cos φ sin θ, y = r sin φ sin θ, z = r cos θ. (3.119)

3.4 Forms 83
 
The coordinate basis vectors er = ∂/∂r, eθ = ∂/∂θ, eφ = ∂/∂φ are

er = sin θ cos φex + sin θ sin φe y + cos θ ez ,

eθ = r cos θ cos φex + r cos θ sin φe y − r sin θ ez ,
eφ = −r sin θ sin φex + r sin θ cos φe y , (3.120)

Inserting these expressions into Eq. (3.60) we find the non-vanishing components
of the metric tensor

grr = 1, gθθ = r 2 , gφφ = r 2 sin2 θ. (3.121)

The line-element takes the form

ds 2 = dr 2 + r 2 dθ 2 + r 2 sin2 θ dφ 2 . (3.122)

The volume form is

V = r 2 sin θ ωr ∧ ωθ ∧ ωφ . (3.123)

The dual of a basis form is

√
1 g
ων = ενμ1 μ2 ωμ1 ∧ ωμ2 , (3.124)
2 gν

where

g = r 4 sin2 θ, gr = 1, gθ = r 2 , gφ = r 2 sin θ. (3.125)

This gives

ωr = r 2 sin θ ε123 ω2 ∧ ω3 = r 2 sin θ ωθ ∧ ωφ , (3.126)

ωθ = sin θ ωφ ∧ ωr , (3.127)

ωφ = ωr ∧ ωθ . (3.128)

The double dual is given by

α = ĝ(−1) p(n− p) α. (3.129)

Hence the double dual is the identity up to a sign. The dual of the volume form is

1
V = εμ ...μ εμ1 ...μn = ĝ = ±1. (3.130)
n! 1 n
84 3 Vectors, Tensors and Forms

Table 3.1 Right column shows the components of the forms dual to the forms in 3-space with
components shown in the left column
0-form: 3-form:
√
φ=φ φ = gφ
1-form: 2-form:
E : [E 1 , E 2 , E 3 ] ⎡ ⎤
0 E3 −E 2
√ ⎢ ⎥
E : g⎢
⎣ −E 0
3 E1 ⎥
⎦
E 2 −E 1 0

2-form: 1-form: 
⎡ ⎤ √
B : g B 23 , B 31 , B 12
0 B12 −B31
⎢ ⎥
B:⎢ ⎣ −B12 0 B23 ⎥⎦
B31 −B23 0

3-form: 0-form:
G : G 123 = G G : g − 1/2 G

Let α and β be p-forms with corresponding vectors A and B, respectively. Then

1 μ1 ...μ p
(α) ∧ β = α βμ1 ...μ p ε1···n ω1 ∧ · · · ∧ ωn = (A · B)V . (3.131)
p!

Furthermore

(α) ∧ β = α∧(β). (3.132)

The following relationship valid for n = 3 between the wedge product of 1-forms
and the vector product of vectors should be noted

1
 α ∧ β = ενλμ (A ∧ B)νλ ωμ = (A × B)μ ωμ . (3.133)
2
Examples with dual forms in 3-space and 4-dimensional spacetime are shown in
Tables 3.1 and 3.2.
3.4 Forms 85

Table 3.2 Right column shows the components of the forms dual to the forms in 4-dimensional
spacetime with components shown in the left column
0-form: 4-form:
√
φ=φ φ = −gφ
1-form: 3-form: √
A : [A0 , A1 , A2 , A3 ] A : (A)012 = − −g A3 etc.
2-form: 2-form:
⎡ ⎤ ⎡ ⎤
0 F01 F02 F03 0 F 23 −F 13 F 12
⎢ ⎥ ⎢ ⎥
⎢ −F01 0 F12 F13 ⎥ √ ⎢ −F 23 0 F 03 −F 02 ⎥
F :⎢⎢
⎥
⎥ F : −g ⎢
⎢ 13
⎥
⎥
⎣ −F02 −F12 0 F23 ⎦ ⎣F −F 03 0 F 01 ⎦
−F03 −F13 −F23 0 −F 12 F 02 −F 01 0

3-form: 1-form:
√ 
G : G αβγ G : −g −G 123 , G 230 , −G 301 , G 012
4-form: 0-form:
H : H0123 = H H = −(−g)−1/2 H

Exercises

3.1. Four-vectors
(a) Given three four-vectors

A = 4et + 3ex + 2e y + ex , B = 5et + 4ex + 3e y , C = et + 2ex + 3e y + 4ex
et · et = −1, ex · ex = ey · ey = ez · ez = 1 (3.134)

Show that A is time-like A · A < 0 , B is light-like B · B = 0 and C is

space-like C · C > 0 .
(b) Assume that A and B are two non-vanishing orthogonal vectors, A · B = 0.
Show the following
• If A is light-like, then B is space-like or light-like.
• If A and B are light-like, then they are proportional.
• If A is space-like, then B is time-like, light-like or space-like.

Illustrate this in a 3-dimensional Minkowski diagram.

(c) A change of basis is given by

et  = cosh θ et + sinh θ ex , ex  = sinh θ et + cosh θ ex , ey  = ey , ez = ez
(3.135)

Show that this describes a Lorentz transformation along the x-axis, where the
relative velocity v between the reference frames is given by v = tanh θ . Draw
86 3 Vectors, Tensors and Forms

the vectors in a 2-dimensional Minkowski diagram and find what type of curves
et  and ex  describe as θ varies.
(d) The 3-vector v describing the velocity of a particle is defined with respect to an
observer. Explain why the 4-velocity u is defined independent of any observer.
The 4-momentum of a particle, with rest mass m, is defined by p = m u =
md r /dτ , where τ is the co-moving time of the particle. Show that p is timelike
and that p · p = − m 2 . Draw, in a Minkowski diagram, the curve to which p
must be tangent to and explain how this is altered as m → 0. Assume that the
energy of the particle is being observed by an observer with 4-velocity u. Show
that the energy he measures is given by

E = − p · u. (3.136)

This is an expression which is very useful when one wants to calculate the
energy of a particle in an arbitrary reference frame.
3.2. The tensor product
(a) Given two 1-forms α = (1, 1, 0, 0) and β = (−1, 0, 1, 0). Show—by using the
vectors e0 and e1 as arguments—that α ⊗ β = β ⊗ α.
(b) Find the components of the symmetric and antisymmetric parts of α ⊗ β.
3.3. Symmetric and antisymmetric tensors
(a) A tensor T of rank 2 in four-dimensional spacetime with Minkowski metric
ηαβ = diag(−1, 1, 1, 1) has contravariant components
⎛ ⎞
0 1 0 0
⎜1 −1 0 2⎟
T αβ =⎜
⎝2
⎟.
0 0 1⎠
1 0 −2 0

Find
1. The components of the symmetric tensor T (αβ) and the antisymmetric tensor
T [αβ] .
2. The mixed components Tβα .
3. The covariant components Tαβ .
(b) Does it make sense to talk about the symmetric and the antisymmetric parts of
a mixed tensor, i.e. a tensor with both vector- and form-arguments? Explain!
3.4. Contractions of tensors with different symmetries
 
2
Let A be an antisymmetric tensor of rank , B a symmetric tensor of rank
0
     
0 0 2
, C an arbitrary tensor of rank and D an arbitrary tensor of rank .
2 2 0
Exercises 87

Show that

Aαβ Bαβ = 0, Aαβ Cαβ = Aαβ C[αβ] , Bαβ D αβ = Bαβ D (αβ) . (3.137)

3.5. Coordinate transformation in an Euclidean plane

 
In this exercise we shall consider vectors in an Euclidean plane. Let ex , ey be
an orthonormal basis in the plane,

ei · e j = δi j .

A position vector as decomposed in this basis is

x = x i ei = x ex + ye y

 
A new coordinate system x  , y  is related to the {x, y}-system by the transfor-
mation

x  = 2x − y, y = x + y

(a) Find ex  and ey  expressed in terms of ex and ey  
(b) Find the basis vectors in the x  , y  -system in terms   of  ex , ey .

(c) Find the components of the metric  tensor
 
 in the x , y -system.
(d) Calculate the line-element in the x , y -system.
 
(e) We now define a set of basis vectors ω  i by ω  i = Mii ei with summation over i.
The scalar products of these vectors define the contravariant components of the
metric tensors.
 Use  this to find the contravariant components of the metric tensor in the
x , y -system.
Chapter 4
Accelerated Reference Frames

Abstract This chapter begins with an introduction to the formalism used to project
four-dimensional spacetime into a 3-dimensional spatial 3-space. Then we apply this
formalism to deduce the spatial geometry in a rotating reference frame and discuss
Ehrenfest’s paradox. Also we show that it is impossible to Einstein synchronize
clocks around a closed path in a rotating frame because this leads to a contradiction
in a non-rotating frame. Gravitational time dilation and frequency shift, and also
the Sagnac experiment are discussed. Finally we give an introduction to special
relativistic kinematic in a uniformly accelerated reference frame in flat spacetime. It
is pointed out that an observer experiences an acceleration of gravity in such a frame.

4.1 The Spatial Metric Tensor

Let e0̂ (x0 = ct) be the 4-velocity field of the reference particles in a reference frame
R. We are going to find the metric tensor γij in a tangent space orthogonal to e0
expressed by the metric tensor gμν of spacetime.
The spatial basis  vectors {ei } are not in general orthogonal to e0 in an arbitrary
coordinate basis eμ . We choose e0 e0̂ . Let ei⊥ be the component of ei orthogonal
to e0 , that is ei⊥ · e0 = 0. The spatial metric tensor is defined by

γij = ei⊥ · ej⊥ , γ00 = γi0 = γ0i = 0. (4.1)

It follows that the spatial metric tensor is symmetric. It is the projection of the
metric tensor of spacetime onto the surface orthogonal to the 4-velocity field of
the reference particles in the given reference frame. Hence it describes the spatial
geometry in the simultaneity 3-space of the reference frame.
The component of ei along e0 is

ei · e0 gi0

ei = e0 = e0 . (4.2)
e0 · e0 g00

Hence, the component of ei orthogonal to e0 is

© Springer Nature Switzerland AG 2020 89

Ø. Grøn, Introduction to Einstein’s Theory of Relativity,
Undergraduate Texts in Physics, https://doi.org/10.1007/978-3-030-43862-3_4
90 4 Accelerated Reference Frames

gi0
ei⊥ = ei − ei = ei − e0 . (4.3)
g00

Thus, the non-vanishing components of the spatial metric tensor are

   
gi0 gj0
γij = ei − e0 · ej − e0 = ei · ej
g00 g00
gj0 gi0 gi0 gj0
− e0 · ei − e0 · ej + 2 e0 · e0
g00 g00 g00
gi0 gj0 gi0 gj0 gi0 gj0 gi0 gj0
= gij − − + = gi j − . (4.4)
g00 g00 g00 g00

The spatial line-element is

 
gi0 gj0
dl 2 = γij dxi dxj = gij − dxi dxj . (4.5)
g00

This line-element gives the distance between simultaneous events in a reference

frame where the metric tensor of spacetime in a co-moving coordinate system is gμν .
Consider a transformation of the form
   
x0 = x0 xμ , xi = xi xi (4.6)

Applying the transformation Eq. (3.113) to each term in Eq. (4.4) and noting that
M0i = 0 for the transformation (4.6) we get

j
γi j = Mii Mj γij . (4.7)

This shows that γij transform as tensor components under a transformation of the
form (4.6).
From the transformation (4.6) we have

∂ ∂xμ ∂
 = . (4.8)
∂x0 ∂x0 ∂xμ

Since ∂xi /∂x0 = 0. Thus

∂x0
e0 = e0 , (4.9)
∂x0

showing that e0 is parallel to e0 . It follows that the 4-velocity field of particles
with fixed coordinates in two coordinate systems connected by a transformation of
4.1 The Spatial Metric Tensor 91

the form (4.6) is identical. This means that Eq. (4.6) represents coordinate trans-
formations between different coordinate systems that are both co-moving with the
same reference frame. Hence transformations of the form (4.6) are called internal
coordinate transformations.
Equation (4.7) shows that the components of the spatial metric tensor transform as
tensor components and the spatial line-element are invariant under internal transfor-
mations, i.e. under transformations between different coordinate systems co-moving
with one and the same reference frame.
The line-element of spacetime can be expressed as

ds2 = −c2 dt̂ 2 + dl 2 , (4.10)

where dt̂ = 0 represents the simultaneity defining the spatial line-element. The
temporal part of the line-element may be expressed as

dt̂ 2 = dl 2 − ds2 = γμν − gμν dxμ dxν

= γij − gij dxi dxj + 2(γi0 − gi0 )dxi dx0 + (γ00 − g00 )dx0 dx0
 
gi0 gj0  2
= gij − − gij dxi dxj − 2gi0 dxi dx0 − g00 dx0
g00


 0 2 gi0 0 i gi0 gj0 i j
= −g00 dx + 2 dx dx + 2 dx dx
g00 g00

 2
gi0 1
= (−g00 )1/2 dx0 + dx . (4.11)
g00

It follows that
 
gi0 i
dt̂ = (−g00 ) 1/2
dx +
0
dx . (4.12)
g00

The 3-space orthogonal to the world lines of the reference particles in R, defined
by dt̂ = 0, corresponds to a coordinate time interval
gi0 i
dt = − dx . (4.13)
g00

This is not an exact differential, that is, dt is not integrable, which means that
one cannot in general define a 3-space orthogonal to the world lines of the reference
particles, i.e. a “simultaneity space”, in an arbitrary reference frame. We must also
conclude that unless gi0 /g00 is constant, it is not possible to Einstein synchronize
clocks around closed curves.
92 4 Accelerated Reference Frames

4.2 Einstein Synchronization of Clocks in a Rotating

Reference Frame

We shall show that it is not possible to Einstein synchronize clocks around a closed
curve in a rotating reference frame. If this is attempted, contradictory boundary
conditions in the non-rotating laboratory frame will arise, due to the relativity of
simultaneity (See Fig. 4.1).
The distance in the laboratory frame between two points is

2π r
x = . (4.14)
n

We now make a Lorentz transformation from the instantaneous rest frame t  , x
of the circumference of the rotating frame R to the non-rotating laboratory frame
 rω   1
t = γ t  + 2 x , x = γ x + rωt  , γ =  , (4.15)
c 1− r 2 ω2
c2

where r is the radios of the circumference and ω is the angular velocity of R.

The proper distance between two points with distance (4.14) in the laboratory
frame is x = γ x. Hence we get a time difference

Fig. 4.1 Simultaneity in a rotating frame. Events that are simultaneous in the rotating frame are
not simultaneous in the stationary frame
4.2 Einstein Synchronization of Clocks in a Rotating Reference Frame 93

rω rω 2π r
t = γ 2 2
x = γ 2 2 (4.16)
c c n
In the laboratory frame for simultaneous events in the rotating frame separated
by a proper distance x . This is due to the relativity of simultaneity. Around the
circumference the time difference is accumulated to

2π r 2 ω
nt = γ , (4.17)
c2
and we get a discontinuity of simultaneity as shown in Fig. 4.2.
Let IF be an inertial frame with cylinder coordinates (T , R, Θ, Z). The line-
element has then the form

ds2 = −c2 dT 2 + dR2 + R2 dΘ 2 + dZ 2 . (4.18)

In a reference frame RF rotating with constant angular velocity ω we have co-

moving cylinder coordinates (t, r, θ, z). The two coordinate systems are related by
the coordinate transformation

t = T , r = R, θ = Θ − ωT , z = Z. (4.19)

Fig. 4.2 Discontinuity of simultaneity in rotating frame

94 4 Accelerated Reference Frames

The first equation means that the coordinate clocks in RF go with a position
independent rate equal to that of the clocks in the non-rotating frame IF, i.e. equal to
that of the inertial rest frame of the rotational axis.
The line-element in the co-moving coordinates in RF is

ds2 = − c2 dt 2 + dr 2 + r 2 (dθ + ω dt)2 + dz 2


= − 1 − r 2 ω2 /c2 dt 2 + dr 2 + r 2 dθ 2 + dz 2 + 2r 2 ω dθ dt. (4.20)

The metric tensor has the following components


gtt = − 1 − r 2 ω2 /c2 , grr = gzz = 1, gθθ = r 2 , gθt = gtθ = r 2 ω. (4.21)

Putting dt = 0 in Eq. (4.20) gives

ds2 = dr 2 + r 2 dθ 2 + dz 2 . (4.22)

This represents the Euclidean geometry of the 3-space in IF.

The non-vanishing components of the spatial metric tensor (4.4) in the co-moving
coordinate system of the rotating frame are
2
gθ0 r 2 ω2 /c2 r2
γrr = γzz = 1, γθθ = gθθ − = r2 −  = . (4.23)
g00 − 1 − r 2 ω2 /c2 1 − r 2 ω2 /c2

Inserting this into Eq. (4.5) gives the spatial line-element in the co-moving
coordinate system of the rotating frame

r 2 dθ 2
dl 2 = dr 2 + + dz 2 . (4.24)
1 − r 2 ω2 /c2

It describes the geometry of a local 3-space orthogonal to the world line of a

reference particle in RF. This 3-space cannot be extended to a finite 3-dimensional
space in RF since Einstein synchronization is not integrable in RF. From the line
element (4.20) it is seen that the geometry of this local simultaneity space in RF is
non-Euclidean. The circumference of a circle with radius r is
2π r
lθ = > 2π r. (4.25)
1 − r 2 ω2 /c2

This means that the spatial geometry is hyperbolic in the rotating frame.
We shall now explain this result first from the point of view of observers at rest
in the non-rotating frame F, and then from the point of view of observes co-moving
with the “rotating” frame, R.
4.2 Einstein Synchronization of Clocks in a Rotating Reference Frame 95

We shall first define the concept standard measuring rod. A standard measuring
rod has by definition a constant rest length even if it is accelerated. It is not allowed by
a standard measuring rod to be compressed or strained. Hence a standard measuring
rod will have a Lorentz contraction according to its velocity.
As observed from F the measuring rods along a circle about the origin have a
velocity v = rω. Hence they will be Lorentz contracted by the factor 1 − r 2 ω2 /c2 .
Hence there is place for more standard measuring rods around the circle the faster
the frame R rotates. Therefore the measured length of the circle will be larger by
this factor. This is the reason for the result (4.25) from the point of view of an F-
observer. Hence according to the F-observers there is no question of a non-Euclidean
geometry. The result (4.25) is explained by the Lorentz contraction of the standard
measuring rods.
It may further be noted that since the material of a rotating disc cannot Lorentz
contract an engraved scale on the disc cannot be used as a set of standard measuring
rods. When the disc is put into rotation the material tries to Lorentz contract in the
tangential direction, but is not allowed to do so. Hence a tangential strain will develop
in the material of a disc that is put into rotation.
We shall now assume the validity of the principle of relativity for rotating motion.
Then the observers in R can think of themselves as at rest and the environment as rotat-
ing. From this point of view the standard measuring rods are not Lorentz contracted.
Hence the explanation of the F-observers does not work for the R-observers.
According to Einstein’s interpretation of the general theory of relativity the expla-
nation of the R-observes is as follows. The R-observer experiences what in Newton’s
theory is called a centrifugal force field. According to the principle of equivalence
this is reckoned as a gravitational field in the theory of relativity. The R-observer
will say that there is a non-Euclidean spatial geometry in the R-frame, and that this
is connected with the gravitational field which is present in this frame.
In general an experimental result—in the present case that the measured length
of a rotating disc with radius r is larger than 2π r—is independent of the reference
frame that the experiment is described from, but the explanation of the result depends
upon the motion of the observer’s reference frame.

4.3 Angular Acceleration in the Rotating Frame

We will now investigate what happens when we give RF an angular acceleration.

Then we consider a rotating circle made of standard measuring rods, as shown in
Fig. 4.3.
All points on a circle are accelerated simultaneously in IF (the laboratory system).
We let the angular velocity increase from ω to ω + dω, measured in IF. Lorentz
transformation to an instantaneous rest frame for a point on the circumference then
gives an increase in velocity in this system:
96 4 Accelerated Reference Frames

Fig. 4.3 Nonrotating disc with measuring rods. The standard measuring rods are fastened with
nails at one end so that they are free to contract when the disc is put into rotation

rdω
rdω = . (4.26)
1 − r 2 ω2 /c2

Due to the relativity of simultaneity the points on the disc at the front end of the
rods are accelerated before the points on the disc at the rear ends, in the rotating
frame. The time difference of the accelerations at the front and rear ends is

rωL0 /c2
t  = , (4.27)
1 − r 2 ω2 /c2

where L0 = 2π r/n is the rest length of the rods. In IF all points of the circumference
are accelerated simultaneously. However, in RF the front points are accelerated before
the rear points, so the distances between the points will increase, i.e. the length of
the circumference of the disc will increase.
It is a defining property of standard measuring rods that they shall move in such
a way that their proper length is preserved. They are not allowed to be stretched
or compressed. The standard measuring rods must therefore move under a different
acceleration program than the points on the disc. All the points of the standard rods
are accelerated simultaneously in RF. Hence the measuring rods will separate from
each other during a period with angular acceleration of the disc as shown in Fig. 4.4.
In IF the separation of the rods is interpreted as a consequence of the Lorentz
contraction of the standard rods.
As observed in RF the distance between the rods is increased by
4.3 Angular Acceleration in the Rotating Frame 97

Fig. 4.4 Lorentz contacted measuring rods on a rotating disc. Separation of the measuring rods
has different explanations as referred to IF and RF. The standard measuring rods have been Lorentz
contracted as observed in IF, while the length of the periphery has increased as observed in RF

r 2 ωL0 d ω/c2
ds = rd ω t  =  3/2 . (4.28)
1 − r 2 ω2 /c2

The increase of distance during the acceleration period is

ω  
r 2 L0 ωd ω 1
s = 2  3/2 = − 1 L0 . (4.29)
c 1 − r 2 ω2 /c2 1 − r 2 ω2 /c2
0

Hence, after the acceleration period there is a proper distance s between the rods.
In the laboratory system, IF, the distance between the rods is
 
1
s= 1− r 2 ω2 /c2 s = 1 − r 2 ω2 /c2 − 1 L0
1 − r 2 ω2 /c2

= L0 − L0 1 − r 2 ω2 /c2 . (4.30)

We now have the situation shown in Fig. 4.4. There is room for more standard
rods around the periphery the faster the disc rotates. This means that as measured
with measuring rods at rest in the rotating frame the measured length of the periphery
(number of standard rods) gets larger with increasing angular velocity. This is how
an inertial observer would explain the measuring result of the rotating observer.
According to the rotating observer, however, the disc material has been stretched in
the tangential direction. Note that as measured by the inertial observer the length of
98 4 Accelerated Reference Frames

the periphery is 2π r independent of the angular velocity of the disc, since the inertial
observer uses measuring rods at rest in the non-rotating reference frame.

4.4 Gravitational Time Dilation

We consider a standard clock moving along a circle about the rotational axis. This
clock has constant r and z. Along the world line of the clock the line element (4.20)
can be written as

  
r 2 ω2 r2 r2ω
ds2 = c2 dt 2 − 1 − 2 + 2 ωc2 + 2 2 ωc . (4.31)
c c c

where ωc = dθ/dt is the angular velocity of the clock in RF. Utilizing the physical
interpretation of the line-element (3.91) for a time-like interval we conclude that the
proper time interval d τ of the clock is related to the coordinate time interval in RF
by

r 2 ω2 r2 r2ω
dτ = dt 1 − 2
− 2 ωc2 − 2 2 ωc . (4.32)
c c c
which may be written

dτ = dt 1 − r 2 (ω + ωc )2 /c2 . (4.33)

Here t represents proper time in RF at the axis, r = 0, which is equal to the proper
time in IF, and τ represents proper time at an arbitrary point in RF. Since the rate of
coordinate time is position independent, it follows that the rate of proper time in RF
decreases with increasing distance from the axis. Also it decreases with increasing
angular velocity ω of RF relative to IF, and it depends upon the angular velocity ωc
of the clock in RF, both its magnitude and sign. The rate of proper time of a clock
moving in RF compared to the time in IF, dτ/dt, is maximal for ωc = −ω. Such a
clock is at rest in IF which is non-rotating relative to the large scale cosmic masses.
For this clock dτ = dt. As considered in RF such a clock moves together with the
large-scale cosmic masses. Hence a clock at rest relative to the large scale cosmic
masses goes at a maximal rate.
A standard clock at rest in RF has ωc = 0. The proper time interval of these clocks
is

r 2 ω2
dτ = dt 1 − 2 . (4.34)
c
Seen from IF, the non-rotating laboratory system, Eq. (4.34) represents the
velocity-dependent time dilation from the special theory of relativity.
4.4 Gravitational Time Dilation 99

But how is Eq. (4.34) interpreted in RF? The clock does not move relative to
an observer in this system, hence what happens cannot be interpreted as a velocity-
dependent phenomenon. According to Einstein, the fact that standard clocks slow
down the farther away from the axis of rotation they are is due to a gravitational
effect.
We will now find the gravitational potential at a distance r from the axis. The
centripetal acceleration is rω2 so
 r  r
1
Φ=− g(r)dr = − rω2 dr = − r 2 ω2 . (4.35)
0 0 2

We then get
 
r 2 ω2 2Φ
dτ = dt 1 − 2 = dt 1 + 2 . (4.36)
c c

In RF the position-dependent time dilation is interpreted as a gravitational time

dilation: Time flows slower further down in a gravitational field.

4.5 Path of Photons Emitted from the Axis in a Rotating

Reference Frame

Let us first describe the photon paths in the inertial frame IF. In this frame the photon
paths are radial. Consider a photon path with  = 0, R = T with light source at
R = 0.
Transforming to RF the equation of the path is,

r = t, θ = −ωt. (4.37)

The orbit equation is thus θ = −ωr which is the equation for an Archimedean
spiral. The time used by a photon out to distance r from axis is t = r/c.

4.6 The Sagnac Effect

The Sagnac effect appears when a beam of light is split and the two beams are made
to follow the same path, but in opposite directions. Returning to the point of entry the
two light beams are allowed to undergo interference. The phases of the two beams,
and thus the position of the interference fringes, are shifted depending upon the
angular velocity of the apparatus.
Let us first describe the Sagnac effect in the inertial rest frame of the axis of the
apparatus. In this frame the velocity of light is isotropic, but the emitter/receiver
100 4 Accelerated Reference Frames

Fig. 4.5 The Sagnac effect.

Experiment demonstrates the
anisotropic velocity of light
as observed by a non-local
measurement in a rotating
reference frame

moves due to the disc’s rotation as shown in Fig. 4.5. Let t1 be the travel time around
the disc of photons that move with the rotation.
Then

2π r + rωt1 = ct1 , (4.38)

giving

2π r
t1 = . (4.39)
c − rω

For light travelling in the opposite direction the travel time is shorter because the
light meets the emitter/receiver,

2π r
t2 = . (4.40)
c + rω

The difference in travel time is

 
1 1 2π r2rω 4Aω
t = t1 − t2 = 2π r − = 2 = γ2 2 , (4.41)
c − rω c + rω c − r 2 ω2 c

where A is the area enclosed by the photon path or orbit.

We shall now describe the same experiment from the point of view of an observer
at rest in a frame rotating together with the apparatus. In this case the line-element is
given by Eq. (4.31) with ωC replaced by the angular velocity of the light, ωL . Since
ds2 = 0 along the world line of the light, the equation of motion of the light takes
the form
4.6 The Sagnac Effect 101

r 2 (ωL + ω)2 − c2 = 0, (4.42)

which has the solutions

ωL = −ω ± c/r. (4.43)

Hence, the speed of the light is

vL± = −rω ± c. (4.44)

We see that in the rotating frame RF the measured round trip velocity of light is
not isotropic. The difference in travel time of the two beams is
 
1 1 4Aω
t = t1 − t2 = 2π r − = γ2 . (4.45)
c − rω c + rω c2

In agreement with the time difference (4.41) observed in IF. In IF this time dif-
ference is explained as a consequence of the motion of the apparatus, while in RF,
where the apparatus is at rest, it is explained as due to the anisotropic velocity of
light.

4.7 Non-integrability of a Simultaneity Curve in a Rotating

Frame

We shall here give some supplementary remarks to the treatment of Einstein syn-
chronization in a rotating reference frame in Sect. 4.1. There we made a separation
of the spacetime line-element, ds2 , in a spatial part, dl 2 , and a temporal part, c2 dt̂ 2 ,
according to ds2 = dl 2 − c2 dt̂ 2 , where
   
gi0 gj0 gi0 i
dl 2 = gij − dxi dxj , d dt̂ = −g00 dx0 + dx , x0 = ct. (4.46)
g00 g00

As applied to the rotating reference frame R this gives

  
r2 r2ω r2ω
dl = dr +
2 2
dθ 2 + dz 2 , dt̂ = 1 − 2 dt − dθ .
1 − r 2 ω2 /c2 c 1 − r 2 ω2 /c2
(4.47)

Here dt = 0 means simultaneity in the non-rotating laboratory system, IF, and

dt̂ = 0 simultaneity in the rotating frame, RF. The simultaneity of the laboratory
frame is defined globally, but simultaneity in the rotating frame, RF, is only defined
locally. With dt̂ = 0 we get
102 4 Accelerated Reference Frames

Fig. 4.6 Discontinuous simultaneity surface in a rotating frame

r2ω
dt = dθ (4.48)
1 − r 2 ω2 /c2

which is not a total differential. This means that simultaneity in the rotating frame
RF cannot be defined around a closed curve about the axis. If define simultaneous
events in RF along a circle about the axis, we come to progressively later events in
IF as given by Eq. (4.48). Going around the circle we arrive at the point of departure
at a later event than the one we started from. This means that the 3-space defined
by simultaneity in RF does not represent a simultaneity space in IF. In a Minkowski
diagram with reference to IF the 3-space is shaped as shown in Fig. 4.6. It has a
discontinuity.

4.8 Orthonormal Basis Field in a Rotating Frame

We saw in Sect. 4.1.1 how the spatial metric representing a simultaneity space of an
observer with 4-velocity u was defined in terms of orthogonal basis vectors, where
the time-like basis vector was chosen to be the 4-velocity of the observer. It has a
magnitude c.
Let us define an orthonormal basis vector field, also called a tetrad field, co-
moving with an observer at rest in an arbitrary reference frame. The 4-velocity of
4.8 Orthonormal Basis Field in a Rotating Frame 103

the observer is u . As time-like unit basis vector we choose

1 dt 1
e0̂ = u = e0 = √ e0 . (4.49)
c dτ −g00

We shall express the spatial vectors of this co-moving orthonormal basis in terms
of the coordinate basis vectors in a coordinate basis {e0 , e1 , e2 , e3 } where e0 is parallel
to u , and the spatial coordinate vectors need not be orthogonal to the time-like basis
vector.
As shown in Sect. 4.1.1 a space-like basis vector ei may be separated in one
component
gi0
ei = e0 (4.50)
g00

along e0 and one ei⊥ = ei − ei orthogonal to e0 , i.e.

gi0
ei⊥ = ei − e0 . (4.51)
g00
√ √
Since this vector has a magnitude |ei⊥ | = ei⊥ · ei⊥ = γii , the corresponding
unit vector is
 
gi0
eî = (γii )−1/2 ei − e0 . (4.52)
g00

The second and third space-like vectors in the orthonormal basis are then given
by

eĵ · eî = eĵ · e0̂ = 0, ek̂ = eî × eĵ . (4.53)

Let us now consider the rotating reference frame, RF. The coordinate transforma-
tion is

T = t, R = r, Θ = θ + ωt, Z = z. (4.54)

Hence the transformation from the coordinate basis vectors in IF to those in RF

is
∂T ∂Θ
et = eT + eΘ = eT + ωeΘ , er = eR , eθ = eΘ , ez = eZ . (4.55)
∂t ∂t

Note that even if T = t the basis vectors eT and et have different directions.
The vector field eT is directed along the world lines of the reference particles in IF
that are parallel to the cylinder axis in the figure above while the vector field et is
directed along the world lines of the reference particles in RF which has the spiral
104 4 Accelerated Reference Frames

shape given by θ = constant shown in Fig. 4.6. The simultaneity space in IF is the
horizontal planes orthogonal to eT , and the simultaneity space in RF is a succession
of simultaneity spaces locally orthogonal to et .
In order to find the orthonormal basis carried by an observer in RF by means of the
formulae above, we must first find the components of the 4-velocity in the co-moving
coordinate system in RF. Since the observer is at rest in RF, the time component is
the only non-vanishing component. It follows from the line-element in RF as applied
to an observer at rest that the 4-velocity is

dt c
u = c et = et . (4.56)
dτ 1 − r 2 ω2 /c2

Inserting Eq. (4.56) into Eq. (4.49) and the expressions (4.21) and (4.23) for the
components of the metric tensor and the spatial metric tensor in RF into Eq. (4.52)
then give the orthonormal basis carried by an observer in RF

1 1 − r 2 ω2 /c2 rω/c2
et̂ = √ et , er̂ = er , eθ̂ = eθ + et . (4.57)
c2 − r 2 ω 2 r 1 − r 2 ω2 /c2

Example 4.8.1 (The acceleration of a velocity field representing rigid rotation) The
velocity field is

v = rωeθ̂ = ωeθ . (4.58)

Note that the coordinate component of the velocity is not equal to the physical
velocity component. The physical velocity components are those in an orthonormal
basis (Fig. 4.7).

We shall calculate the acceleration field, a = dv

dt
. Using the chain rule of
differentiation we get

Fig. 4.7 Rigid rotation

4.8 Orthonormal Basis Field in a Rotating Frame 105

∂v dxμ ∂v

a = = vμ μ . (4.59)
∂xμ dt ∂x

Since the only non-vanishing velocity component is vθ = ω we get

∂v
a = ω . (4.60)
∂θ
This gives

∂(ωeθ ) ∂eθ
a = ω = ω2 . (4.61)
∂θ ∂θ
We have found earlier that

er = cos θ ex + sin θ ey , eθ = −r sin θ ex + r cos θ ey . (4.62)

Differentiation gives

∂eθ
= −r cos θ ex − r sin θ ey = −rer = −rer̂ . (4.63)
∂θ
Hence the acceleration is

a = −rω2 er̂ . (4.64)

This is the centripetal acceleration for circular motion.

4.9 Uniformly Accelerated Reference Frame

Consider a particle moving along a straight line with velocity u and acceleration
a = du/dT . The acceleration in the instantaneous inertial rest frame of the particle,
its proper acceleration, is denoted by â and is given by
 3/2
a = 1 − u2 /c2 â. (4.65)

Assume that the particle has constant proper acceleration â = g, that is

du  3/2
= 1 − u2 /c2 g, (4.66)
dt
which on integration with u(0) = 0 gives
106 4 Accelerated Reference Frames

dX gT
u= = 2 2 . (4.67)
dT (1 + g cT2 )1/2

Integrating once more gives

c2 g2T 2
X = (1 + 2 )1/2 + k, (4.68)
g c

where k is a constant of integration. This may be written

(X − k)2 − c2 T 2 = c4 /g 2 . (4.69)

This describes a hyperbola in the Minkowski diagram as shown in Fig. 4.8.

The proper time interval as measured by a clock which follows the particle is
 1/2
u2
dτ = 1 − 2 dT . (4.70)
c

Substitution for u(T ) and integration with τ (0) = 0 give

 
c gT
τ = arcsinh , (4.71)
g c
or

Fig. 4.8 Hyperbolic motion.

Graph is the world line of a
reference particle in a
uniformly accelerated
reference frames as drawn
with reference to the inertial
frame in which the particle is
at rest at the point of time
T =0
4.9 Uniformly Accelerated Reference Frame 107

c  gτ 
T= sinh . (4.72)
g c

Inserting this into Eq. (4.68) gives

c2  gτ 
X = cosh + k. (4.73)
g c

We now use this particle as the origin of space in an uniformly accelerated

reference frame, AF. With the initial condition X (0) = 0 we get k = −c2 /g and

gX gτ
1+ = cosh . (4.74)
c2 c
Example 4.9.1 (Uniformly accelerated motion through the Milky Way) Let a traveller
traverse the Milky Way, say a distance of a hundred thousand years, with a constant
proper acceleration equal to 9.8 m/s2 , to the centre and then reversing the acceleration
so that the traveller stops at the other side. The travel would take approximately a
hundred thousand years as measured by the stationary observer. It has been said that
the traveller would only become around 20 years older during the travel.
Let us see how this comes about. Let X be the distance and T the time as measured
by the stationary observer, τ the proper time of the traveller and g = 9.8 m/s2 the
proper acceleration. For uniformly accelerated motion, i.e. motion with constant
proper acceleration, we saw in Eqs. (4.72) and (4.74) that

gX gτ gT gτ
1+ = cosh , = sinh (4.75)
c2 c c c

From the identity cosh2 (gτ/c) − sinh2 (gτ/c) = 1 it then follows that
   2
gX 2 gT
1+ 2 − =1 (4.76)
c c
or
    
gT gX 2 gX gX 2
= 1+ 2 −1= 2 2 + . (4.77)
c c c c2

In the present case we have 1 year = 3.16 × 107 s, 1 light year = ct = 9.48 × 1015
m, and the distance from the initial position of the traveller to the centre of the Milky
Way is X = 5 × 104 light years = 4.74 × 1020 m. Hence gX /c2 = 5.16 · 104  1,
so we can approximate the expression (4.73) by T ≈ X /c. The reason that this is a
good approximation in the present case is that the traveller travels with approximately
the velocity of light during nearly all the time. This can be seen by calculating the
velocity as measured by the stationary observer by taking the differentials of the
expressions (4.75)
108 4 Accelerated Reference Frames

Fig. 4.9 The velocity of a uniformly accelerated particle

dX gτ
U = = c tanh . (4.78)
dT c
The velocity as a function of the proper time is shown in Fig. 4.9.
We see that already at gτ/c = 2 corresponding to gX /c2 = 2.8 or a travelled
distance of about 2.7 light years, the velocity of the traveller is close to that of light.
Then the travel time to the centre of the Milky Way as measured by the stationary
observer is close to T = 50 thousand years, giving gT /c = 5.16 × 104 . Inserting this
into the second expression (4.75) written as

c gT
τ= arsinh (4.79)
g c

gives τ = 11.2 years. So the whole travel takes a hundred thousand years as measured
by the stationary observer, while the traveller only gets 22.4 years older. The traveller
ages extremely slowly because she travels with a velocity which is so close to the
velocity of light.
From her point of view she ages by only 22.4 years because the Milky Way moves
with nearly the velocity of light and is shaped like a disc which is only 22.4 light
years thick due to the Lorentz contraction.

Definition 4.9.1 (Born-rigid Motion) Born-rigid motion of a system is a motion

such that every element of the system has constant rest length.
The uniformly accelerated reference frame is required to move in a Born-rigid
way. Let the inertial frame has coordinates (T , X , Y , Z) and the accelerated frame
has coordinates (t, x, y, z). We now denote the X -coordinate of the “origin particle”
by X0 . From Eq. (4.73) we then have

gX0 gτ0
1+ 2
= cosh , (4.80)
c c

where τ0 is the proper time for this particle and k is set to −c2 /g. (These are Møller
coordinates. Putting k = 0 gives Rindler coordinates).
4.9 Uniformly Accelerated Reference Frame 109

Fig. 4.10 Simultaneity in a

uniformly accelerated
reference frame. The vector


X lies along the

“simultaneity line” which
makes the same angle with
the x-axis as does et̂ with the
cT-axis

The coordinate time at an arbitrary point in AF is defined by t = τ0 . That is

coordinate clocks in AF run identically with the standard clock at the “origin particle”.
Hence, the coordinate time has a position independent rate. Let X0 be the position
4-vector of the “origin particle”. Decomposed in the laboratory frame, this becomes
   
c2 gt c2 gt
X0 = sinh , cosh − 1 , 0, 0 . (4.81)
g c g c

P is chosen such that P and P0 are simultaneous in the accelerated frame AF.
The distance (see Fig. 4.10) vector from P0 to P, decomposed into an orthonormal
co-moving basis of the “origin particle”, is X̂ = (0, x̂, ŷ, ẑ) where x̂, ŷ and ẑ are
physical distances measured simultaneously in AF. The space coordinates in AF are
defined by

x ≡ x̂, y ≡ ŷ, z ≡ ẑ. (4.82)



The position vector of P is X = X0 + X . The relationship between basis vectors

in IF and the co-moving orthonormal basis is given by a Lorentz transformation in
the x-direction,
⎛ ⎞
cosh η sinh η 0 0
∂xμ ⎜ sinh η cosh η 0 0 ⎟
eμ̂ = eμ μ̂ = (eT , eX , eY , eZ , )⎜
⎝ 0
⎟, (4.83)
∂x 0 1 0⎠
0 0 01

where η is the rapidity defined by

110 4 Accelerated Reference Frames

U0
tanh η ≡ . (4.84)
c
Here U0 is the velocity of the “origin particle”,

dX0 gt
U0 = = c tanh . (4.85)
dT0 c

Hence, in the present case the rapidity is

gt
η= . (4.86)
c
So the basis vectors can be written as follows,

gt gt
et̂ = eT cosh + eX sinh ,
c c
gt gt
ex̂ = eT sinh + eX cosh ,
c c
eŷ = eY , eẑ = eZ . (4.87)


The equation X = X0 + X can now be decomposed in IF,

cT eT + X eX + Y eY + ZeZ

 
c2 gt c2 gt gt
= sinh eT + cosh − 1 eX + x sinh eT
g c g c c
gt
+ x cosh eX + yeY + zeZ . (4.88)
c
This then gives the coordinate transformations:

T = gc sinh gtc + xc sinh gtc ,

2
X = cg cosh gtc − 1 + x cosh gt
c
. (4.89)
Y = y, Z = z

The transformations of T and X may be written

gT  gx  gt gX  gx  gt
= 1 + 2 sinh , 1 + 2 = 1 + 2 cosh . (4.90)
c c c c c c
It follows from these equations that

gT gX gt
= (1 + 2 ) tanh , (4.91)
c c c
4.9 Uniformly Accelerated Reference Frame 111

showing that the coordinate curves t = constant are straight lines in the T, X-frame
passing through the point T = 0, X = −c2 /g.
Using the identity cosh2 θ − sinh2 θ = 1 we get
   2 
gX 2 gT gx 2
1+ 2 − = 1+ 2 , (4.92)
c c c

showing that the coordinate curves x = constant are hyperbolae in the T, X-diagram
(Fig. 4.11).
Taking the differentials of the coordinates in Eq. (4.89) gives the form of the
line-element in the co-moving coordinates of the uniformly accelerated reference
frame,
 gx 2
ds2 = −c2 dT 2 + dX 2 + dY 2 + dZ 2 = − 1 + 2 c2 dt 2 + dx2 + dy2 + dz 2 .
c
(4.93)

Fig. 4.11 World lines and simultaneity lines of a uniformly accelerated reference system.
Minkowski diagram of the world lines and simultaneity lines of the uniformly accelerated reference
frame AF with reference to the inertial frame IF in which AF is at rest at T = 0
112 4 Accelerated Reference Frames

The line-element is an invariant quantity, but looks different in different coordinate

systems. Note that the metric is diagonal, i.e. only components with equal indices
are different from zero when the basis vectors are orthogonal.
We now consider standard clocks at rest in the accelerated reference frame, i.e.
dx = dy = dz = 0 along the world lines of the clocks. Then the line-element (4.93)
reduces to
 gx 2
ds2 = − 1 + 2 c2 dt 2 . (4.94)
c
Utilizing the physical interpretation (3.91) of the line element for a time-like
interval we obtain the relationship between the proper time and the coordinate time
 gx 
dτ = 1 + 2 dt. (4.95)
c
Since the rate of coordinate time is position independent, this equation tells how
the rate of proper time depends upon the position.
An observer in the accelerated frame AF experiences a gravitational field in the
negative x-direction. When x < 0 then dτ < dt. Equation (4.95) says that time
passes slower further down in a gravitational field. This is called the gravitational
time dilation.
Consider a standard clock moving in the x-direction with velocity v = dx/dt.
Then

 
gx 2 2 2 gx 2 v2 2 2
− c dτ = − 1 + 2 c dt + dx = − 1 + 2 − 2 c dt
2 2 2
(4.96)
c c c

Hence

 gx 2 v2
dτ = 1+ − 2 dt. (4.97)
c2 c
This expresses the combined effect of the position dependent gravitational time
dilation and the velocity-dependent kinematic time dilation.
Let us consider how light moves in the uniformly accelerated reference frame.
As a simple example we consider light moving in the y-direction in the laboratory
frame,

X = X0 , Y = Y0 + cT , Z = 0. (4.98)

Inserting this in the coordinate transformation above we obtain

gT g  gx 2  g 2  
gX0 2
= 2 (y − y0 ), 1 + 2 + 2 (y − y0 ) = 1 + 2 . (4.99)
c c c c c
4.9 Uniformly Accelerated Reference Frame 113

The last equation shows that the light moves along a circular path with radius
X0 + c2 /g. Differentiating the equation of the trajectory with respect to x we obtain

dy x + c2 /g
= . (4.100)
dx y − y0

Hence dy/dx = 0 at the horizon. In other word the light moves in the vertical
direction at the horizon. At that position the light has no motion in the y-direction.
The reason is that the time does not progress at the horizon.
Note also from the line-element that
dx  gx 
= 1+ 2 c (4.101)
dt c
for light moving in the x-direction. Light moves slower the further down it is in the
gravitational field, and the coordinate velocity of the light approaches zero as the
light approaches the horizon.

4.10 The Projection Tensor

An observer moves through spacetime with a 4-velocity u . An arbitrary 4-vector a

shall be decomposed into one component, a , along u and one, a⊥ , orthogonal to u ,
so that a = a + a⊥ . Using units so that c = 1 the 4-velocity is then a time-like unit
vector, u · u = −1, and

(a · u )u
a = = −(a · u )u, a⊥ = a + (a · u )u. (4.102)
u · u
In the following we shall need the corresponding 4-velocity form,

u = uμ ωμ . (4.103)

Definition 4.10.1 (The Projection Tensor) The projection tensor, P, is a mixed

tensor of rank 2 which is given in terms of the unit tensor I = ωμ ⊗ eμ of rank 2, the
4-vector u and the corresponding 4-form u as

P = I + u ⊗ u . (4.104)

The mixed components of the projection tensor are

Pμν = δμν + uμ uν . (4.105)

114 4 Accelerated Reference Frames

Note that the last term is not a scalar product. The covariant components of the
projection tensor are

Pμν = gμν + uμ uν . (4.106)

Applying the projection tensor to the vector a and noting from the line below
Eq. (3.73), that I(a) = a , we get

P(a) = I(a) + u ⊗ u (a) = a + u(a)u = a + (a · u )u = a⊥ . (4.107)

This means that when one applies the projection tensor to a vector a one gets out
the component a⊥ of the vector orthogonal to the vector u of the projection tensor.
In other words one gets the projection onto the simultaneity space orthogonal to the
time direction of the observer with 4-velocity u .

Example 4.10.1 (Covariant condition for uniformly accelerated motion) The covari-
ant condition for uniformly accelerated motion of a particle is that the proper acceler-
ation of the particle is constant, which means that the projection of the rate of change
of the acceleration vector vanishes in an instantaneous rest system of the particle. The
covariant mathematical expression of this is: P(d a /d τ ) = 0, where τ is the proper
time of the particle. The component form of this condition is: Pνμ daν /d τ = 0.

Example 4.10.2 (Spatial metric and the projection tensor ) We shall here generalize
Formula (4.4) for the spatial metric tensor. Consider a coordinate interval dxμ . It
has a component orthogonal to a 4-velocity u given by dl μ = Pαμ dxα . The invariant
spatial line-element is

dl 2 = dlα dl α = gαβ dl α dl β = gαβ Pμα Pνβ dxμ dxν = Pμβ Pνβ dxμ dxν . (4.108)

The components of the spatial metric tensor in a simultaneity space orthogonal

to u may be defined by

dl 2 = γμν dxμ dxν . (4.109)

This gives
 
γμν = Pμβ Pνβ = gμβ + uμ uβ δνβ + uν uβ = gμν + uν uμ + uμ uν
+ uμ uν uβ uβ = gμν + uμ uν . (4.110)

The right-hand side is the covariant components of the projection tensor. Hence
the spatial metric tensor is equal to the projection tensor,

γμν = Pμν . (4.111)

4.10 The Projection Tensor 115

We shall now consider the simultaneity space of an observer at rest in the

coordinate system. Then the contravariant components of the 4-velocity are

uμ = (−g00 )−1/2 (1, 0, 0, 0). (4.112)

The covariant components are

u0 = −(−g00 )1/2 , ui = (−g00 )−1/2 gi0 . (4.113)

Substituting these expressions into Eq. (4.110) gives

gi0 gj0
γ00 = γi0 = 0, γij = gij − , (4.114)
g00

in agreement with Eq. (4.4).

Exercises

4.1. Relativistic rotating disc

A disc rotates with constant angular velocity in ω in its own plane and around a fixed
axis A. The axis is chosen to be the origin in a non-rotating Cartesian coordinate
system (x, y) with coordinate clocks showing t. (The z-coordinate is kept constant
from now on). The motion of a given point on the disc can be expressed as

x = r cos(ωt + φ), y = r sin(ωt + φ), (4.115)

where (r, φ) are coordinates specifying the point at the disc.

(a) An observer is able to move on the disc and performs measurements of distance
between neighbouring points at different locations on the disc. The measure-
ments are performed when the observer is stationary with respect to the disc.
The result is assumed to be the same as that measured in an inertial frame with
the same velocity as the observer at the time of the measurement. The lengths
measured by the observer are given by

d2 = f1 (r, φ)dr 2 + f2 (r, φ)dφ 2 . (4.116)

Find f1 (r, φ) and f2 (r, φ).

We now assume that the observer measures the distance from the axis A to a
point (R, 0) along the line φ = 0, by adding the result of measurements between
neighbouring points. What is the result the observer finds? Furthermore the observer
116 4 Accelerated Reference Frames

measures the distance around the circle r = R. What is then found? In what way,
based on this result, is it possible to deduce that the metric considered by the observer
is non-Euclidean? Will the observer find a negative or positive curvature of the disc?
We introduce coordinates (t̃, x̃, ỹ) that follow the rotating disc. They are given by

t̃ = t, x̃ = r cos φ, ỹ = r sin φ. (4.117)

(b) Find the invariant interval ds2 = dx2 + dy2 − c2 dt 2 in terms of the coordinates
(t̃, x̃, ỹ).
(c) Light signals are sent from the axis A. How will the paths of the light signals
be as seen from the (x̃, ỹ) system? Draw a figure that illustrates this. A light
signal with the frequency ν0 is received by the observer in r = R, φ = 0. Which
frequency ν will be measured by the observer?
(d) We now assume that standard clocks measuring proper time are tightly packed
around the circle r = R. The clocks are at rest on the disc. We now want to
synchronize the clocks and start out with a clock at the point (R, 0). The clocks
are then synchronized in the direction of increasing φ in the following way:
When a clock is tuned at a point φ, the clock at the neighbouring point φ + d φ
is also tuned so that it shows the same time at simultaneity in the instantaneous
rest frame of the two clocks.

Show that there is a problem with synchronization when this process is per-
formed around the entire circle, by the fact that the clock we started out with is no
longer synchronous with the neighbouring clock which is tuned according to the
synchronization process. Find the time difference between these two clocks.
(e) Locally around a point (r, φ, t) we can define an inertial system being an instan-
taneous rest frame of the point (r, φ) on the disc. We introduce an orthonormal
set of basis vectors eλ̂ , eη̂ and eξ̂ in this frame. The vector eλ̂ points along the
time axis of the system, eξ̂ points radially and eη̂ tangentially. Find the vectors
expressed by et , ex and ey .

4.2. Uniformly accelerated system of reference

We will now study a coordinate system (t, x) co-moving with a uniformly accelerated
reference frame, AF, in a 2-dimensional Minkowski space. The connection with a
Cartesian coordinate system (T , X ) co-moving with an instantaneous inertial rest
frame, IF, of AF at the point of time T = 0 is given by the coordinate transformation

T = x sinh(at), X = x cosh(at) (4.118)

where a is a constant.
(a) Draw the coordinate lines t = constant and x = constant in a (T , X )-diagram.
(b) Find the line-element ds2 = −dT 2 + dX 2 expressed by t and x.
(c) We now assume that a particle has a path in spacetime so that it follows one of
the curves x = constant. Such a motion is called hyperbolic motion. Why?
Exercises 117

Show that the particle has constant acceleration when the acceleration is
measured in the instantaneous rest frame of the particle. Find the acceleration
of the rod has constant acceleration and the rod g. Find also the velocity and
acceleration of the particle in the system (T , X ).
(d) Show that at any point on the trajectory of a reference particle in AF the direction
of the coordinate axes in the (T , X )-system will overlap with the time and spatial
axis of the instantaneous rest frame of the particle. Explain why it is possible
to see from the line element that the X -coordinate measures length along the
spatial axis, whereas the T -coordinate, which is the coordinate time, is in general
not the proper time of the particle. For what value of X is the coordinate time
equal to the proper time?
The (t, x)-coordinate system can be considered as an attempt to construct,
from the instantaneous rest frames along the path, a coordinate system covering
the entire spacetime. Explain why this is not possible for the entire space. (Hint:
There is a coordinate singularity at a certain distance from the trajectory of the
particle).
(e) A rod is moving in the direction of its own length. At the time T = 0 the rod
is at rest, but still accelerated. The length of the rod measured in the stationary
system is L at this time. The rod moves so that the forwards point of the rod has
constant rest acceleration measured in the instantaneous rest frame.
We assume that the acceleration of the rod finds place so that the infinitesimal
distance d between neighbouring points on the rod measured in the instantaneous
rest frame is constant. Find the motion of the rear point of the rod in the stationary
reference system. Why is there a maximal length of the rod, Lmax ?
If the rear point of the rod has constant acceleration and the rod is accelerated as
previously in this exercise, then is there a maximal value of L ?
4.3. Uniformly accelerated space ship
(a) A spaceship leaves the Earth at the time T = 0 and moves with a constant
acceleration g, equal to the gravitational constant at the Earth, into space.
Find how far the ship has travelled during 10 years of proper time of the
ship.
(b) Radio signals are sent from the Earth towards the spaceship. Show that
signals that are sent after a given time T will never reach the ship (even
if the signals travel with the speed of light). Find T . At what time are the
signals sent from the Earth if they reach the ship after 10 years (proper
time of the ship)?
(c) Calculate the frequency of the radio signals received by the ship, given
by the frequency ν0 (emitter frequency) and the time t0 (emitter time).
Investigate the behaviour of the frequency when the proper time on the
space ship τ → ∞.
118 4 Accelerated Reference Frames

4.4. Light emitted from a point source in a gravitational field

A point-like light source is at the position x = x1 , y = 0 in a uniformly accelerated
reference frame AF. A photon is emitted from the source at a point of time t = 0. It
is emitted in the (x, y)− plane in a direction making an angle θ0 with the x-axis.
Find the equation for the path followed by the photon, and identify the nature of
the trajectory.
4.5. Geometrical optics in a gravitational field
It was shown in Eq. (4.99) that light does not move along a straight path in the
gravitational field experienced in a uniformly accelerated reference frame AF, but
along a circular trajectory.
(a) A sphere seen from above. The observer P is at a height x = x1 . The centre
of the sphere is at a distance b vertically beneath P. Light is emitted from the
surface of the body. It moves along a circular path with radius R = x1 / sin θ ,
where θ is the angle between the light path and the x-axis at the position of
the observer. The corresponding angle without a gravitational field would be θ0
given by sin θ0 = r/b, where r is the radius of the sphere. The acceleration of
gravity at the position of the observer is g/c2 = 1/x1 .
Calculate how the angle θ depends upon g, r and b.
(b) A sphere seen from below. Same as (a) but with the observer below the sphere.
(c) An experiment. Place a camera one metre above the centre of a sphere of radius
10 cm and another one a metre below in the gravitational field of the Earth.
Calculate the difference of θ in photographs taken with the two cameras.
Chapter 5
Covariant Differentiation

Abstract The theory of forms is a theory of antisymmetric tensors. In such a theory

we need an antisymmetric version of the covariant derivative such that the derivative
of a form is a form. Hence in this chapter we first introduce the covariant derivative and
then the antisymmetric exterior derivative. The relativistic Euler–Lagrange equations
are introduced and applied to deduce the equation of motion of free particles in curved
spacetime—the geodesic equation. Since gravity is not reckoned as a force in the
general theory of relativity “free particles” in this theory correspond to particles acted
upon by gravity in Newton’s theory. The fundamental equations of electromagnetism
in form language are also presented in this chapter.

5.1 Differentiation of Forms

We must have a method of differentiation which maintains the antisymmetry so that

differentiating a form gives out a form.

5.1.1 Exterior Differentiation

The exterior derivative of a 0-form, i.e. a scalar function, f, is given by

∂f μ
df = ω = f μ ωμ , (5.1)
∂xμ
where ωμ are coordinate basis forms
 
∂
ωμ = δνμ . (5.2)
∂xν

We then get

© Springer Nature Switzerland AG 2020 119

Ø. Grøn, Introduction to Einstein’s Theory of Relativity,
Undergraduate Texts in Physics, https://doi.org/10.1007/978-3-030-43862-3_5
120 5 Covariant Differentiation

∂xμ ν
ωμ = δνμ ων = ω = dx μ . (5.3)
∂xν
This shows that we can write the basis forms as exterior derivatives of the
coordinates in an arbitrary coordinate basis. The differential dx μ is given by

dx μ (d
r ) = dx μ , (5.4)

where d r is an infinitesimal position vector. Note that dx μ are not infinitesimal

quantities. It follows that in a coordinate basis the exterior derivative of a p-form

1
α= αμ ...μ dx μ1 ∧ · · · ∧ dx μ p (5.5)
p! 1 p

has the following component form:

1
dα= αμ ...μ ,μ dx μ0 ∧ dx μ1 ∧ · · · ∧ dx μ p , (5.6)
p! 1 p 0

where μ0 ≡ ∂ ∂x μ . This shows that the exterior derivative of a p-form is a (p + 1)-form.

0
Let (dα)μ0 ...μ p be the form components of dα. They must, by definition, be
antisymmetric under an arbitrary interchange of indices:

1 1
dα = (dα)μ0 ...μ p dx μ0 ∧ · · · ∧ dx μ p = α[αμ1 ...μ p ,μ0 ] dx μ0 ∧ · · · ∧ dx μ p .
( p + 1)! p!
(5.7)

Hence

(dα)μ0 ...μ p = ( p + 1)α[μ1 ...μ p ,μ0 ] . (5.8)

The component form of the form equation dα = 0 is

α[μ1 ...μ p ,μ0 ] = 0. (5.9)

Example 5.1.1 (Relationship between exterior derivative and curl) We consider a

1-form in a 3-space,

α = αi dx i , x i = (x, y, z). (5.10)

Its exterior derivative is

dα = αi, j dx j ∧ dx i . (5.11)
5.1 Differentiation of Forms 121

Also, assume that dα = 0. The corresponding component equation is α[i, j] = 0

or αi, j − α j, i = 0. Writing it out we have

∂α y ∂αx ∂αz ∂αx ∂α y ∂αz

− = 0, − = 0, − = 0. (5.12)
∂x ∂y ∂x ∂z ∂z ∂y

which corresponds to

∇ × α = 0. (5.13)

The outer product of an outer product is

d2 α ≡ d(dα), (5.14)

with component form

1 ∂2
d2 α = αμ1 ...μ p ,ν1 ν2 dx ν2 ∧ dx ν1 ∧ · · · ∧ dx μ p , ν1 ν2 ≡ ν ν . (5.15)
p! ∂x 1∂x 2

Since

∂2 ∂2
,ν1 ν2 ≡ =,ν2 ν1 ≡ (5.16)
∂ x ν1 ∂ x ν2 ∂ x ν2 ∂ x ν1
the quantities αμ1 ...μ p ,ν1 ν2 are symmetric in ν1 and ν2 . On the other hand, the basis is
antisymmetric in ν1 and ν2 . We saw in Eq. (3.107) that the product of a symmetric
and an antisymmetric quantity vanishes. Hence, we have

d2 α = 0, (5.17)

which means that the second exterior derivative of a p-form (with scalar components,
see later) vanishes. This is sometimes called Poincaré’s lemma. For a 1-form, it
corresponds to the vector equation

 = 0.
∇ · (∇ × A) (5.18)

Let α be a p-form and β a q-form. Then

d(α ∧ β) = dα ∧ β + (−1) p α ∧ dβ. (5.19)

122 5 Covariant Differentiation

5.1.2 Covariant Derivative

The general theory of relativity contains a covariance principle which states that
all equations expressing laws of nature must have the same form irrespective of the
coordinate system in which they are derived. This is achieved by writing all equations
in terms of tensors. Let us see if the partial derivative of vector components transform
as tensor components.
Consider a vector A = Aμ eμ = Aμ eμ . The basis transforms as


∂ ∂xν ∂
 = . (5.20)
∂xv ∂ x v ∂ x ν
Hence
 
∂   ∂xν ∂  

 ∂xν ∂ ∂xμ μ
Aμ,ν  ≡ ν  Aμ = ν  ν Aμ = ν  ν A . (5.21)
∂x ∂x ∂x ∂x ∂x ∂xμ

Performing the differentiation of the product gives

 
μ ∂xν ∂xμ μ ∂ xν μ ∂2xμ
A,ν  =  A ,ν + A . (5.22)
∂xν ∂xμ ∂ x ν ∂xν∂xμ
The first term corresponds to a tensor transformation. However, the presence of the
last term shows that Aμ,ν does not, in general, transform as the components of a tensor.
Note that Aμ,ν will transform as a components of a tensor under linear transformations
such as the Lorentz transformations.
In order to obtain a proper tensor formalism, the partial derivative must be gener-
alized so as to ensure that when its generalization is applied to tensor components it
produces tensor components.

5.2 The Christoffel Symbols

The covariant derivative of the components of a vector field was introduced by Elwin
Christoffel to be able to differentiate tensor fields. It is defined in coordinate basis
μ
by generalizing the partial derivative Aμ,ν to a derivative written as A;ν and which
transforms tensorially,

 ∂xμ ∂xν μ
Aμ;ν  ≡ · A . (5.23)
∂ x μ ∂ x ν  ;ν
Definition 5.2.1 (Christoffel symbols) The Christoffel symbols are defined by writing
the covariant derivative of the vector components in the form
5.2 The Christoffel Symbols 123

μ
A;ν ≡ Aμ,ν + Aα αν
μ
. (5.24)

μ
The Christoffel symbols αν are also called the “connection coefficients in coor-
dinate basis”. From the transformation formulae of the left-hand side and the first
term on the right-hand side it follows that the Christoffel symbols transform as
 
 ∂xν ∂xμ ∂xα α ∂ xα ∂2xα
μα  ν  =  + . (5.25)
∂ x ν ∂ x μ ∂ x α μν
 
∂ x α ∂ x μ ∂ x ν 
Due to the last term, the Christoffel symbols do not transform as tensor compo-
nents. It is possible to make all Christoffel symbols vanish by transforming into a
locally Cartesian coordinate system which is co-moving in a locally non-rotating
reference frame in free fall. Such coordinates are known as Gaussian coordinates.
As discussed in Sect. 1.4 according to the general theory of relativity an inertial
frame is a non-rotating frame in free fall. The Christoffel symbols are 0 (zero) in a
locally Cartesian coordinate system which is co-moving in a local inertial frame Local
Gaussian coordinates are indicated with a hat over the indices since the coordinate
vector basis is orthonormal in such a coordinate system, giving
α̂
μ̂ν̂ = 0. (5.26)

A transformation from local Gaussian coordinates to any coordinates leads to


 ∂ x α ∂ 2 x α̂
μα  ν  =   . (5.27)
∂ x α̂ ∂ x μ ∂ x ν
This equation shows that the Christoffel symbols are symmetric in the two lower
indices, i.e.
 
μα  ν  = να μ . (5.28)

Example 5.2.1 (The Christoffel symbols in plane polar coordinates) The transfor-
mation between plane polar coordinates and Cartesian coordinates is

x =
r cos θ, y = r sin θ,
(5.29)
r = x 2 + y 2 , θ = arctan xy ,

We need the derivatives of the coordinates with respect to each other

∂x ∂x ∂r ∂r
∂r
= cos θ, ∂θ
= −r sin θ, ∂x
= rx = cos θ, ∂y
= sin θ,
∂y ∂y ∂θ (5.30)
∂r
= sin θ, ∂θ
= r cos θ, ∂x
= − sinr θ , ∂θ
∂y
= cosr θ ,
124 5 Covariant Differentiation

Inserting these expressions into Eq. (5.27) gives the Christoffel symbols in the
plane polar coordinates

∂r ∂ 2 x ∂r ∂ 2 y
θθ
r
= + = cos θ (−r cos θ ) + sin θ (−r sin θ )
∂ x ∂θ 2 ∂ y ∂θ 2
= −r (cos θ 2 + sin θ 2 ) = −r, (5.31)

∂θ ∂ 2 x ∂θ ∂ 2 y sin θ cos θ 1
rθθ = θr
θ
= + =− (− sin θ ) + (cos θ ) = .
∂ x ∂θ ∂r ∂ y ∂θ ∂r r r r
(5.32)

The geometrical interpretation of the covariant derivative was given by Levi-

Civita. Consider a curve S in any (e.g. curved) space. It is parameterized by λ, i.e.
x μ = x μ (λ). The parameter λ is invariant and chosen to represent the arc length.
The tangent vector field of the curve is

u = (dx μ /dλ)eμ . (5.33)


The curve passes through a vector field A.

Definition 5.2.2 (Covariant directional derivative) The covariant directional

derivative of the vector field along the curve is defined as

d A μ dx
ν
μ
∇u A = ≡ A;ν eμ = A;ν u ν eμ . (5.34)
dλ dλ
Definition 5.2.3 (Parallel transport) The vectors are said to be connected by parallel
transport along the curve if
μ
A;ν u ν = 0. (5.35)

According to the geometrical interpretation of Levi-Civita, the covariant direc-

tional derivative is

μ

A (λ + λ) − A(λ)
∇u A = A;ν u ν eμ = lim , (5.36)
λ→0 λ

where A (λ + λ) means the vector A parallel transported from Q to P (Fig. 5.1).
5.3 Geodesic Curves 125

μ
Fig. 5.1 Parallel transport from P to Q. The vector B = A;ν u ν λeμ

5.3 Geodesic Curves

Definition 5.3.1 (Geodesic curves) A geodesic curve is defined by the requirement

that the vectors of the tangent vector field of the curve are connected by parallel
transport.
This definition says that geodesic curves are “as straight as possible”. If vectors

in a vector field A(λ) are connected by parallel transport by a displacement along a
μ
vector u, we have A;ν u ν = 0. For geodesic curves, we then have

μ
u ;ν u ν = 0 (5.37)

which is the geodesic equation

(u μ,ν + αν
μ α ν
u )u = 0. (5.38)

Then we are using that

d dx ν ∂ ∂
≡ ν
= uν ν , (5.39)
dλ dλ ∂ x ∂x
which gives

du μ ∂u μ
= u ν ν = u ν u μ,ν . (5.40)
dλ ∂x
Hence, the geodesic equation can be written as
126 5 Covariant Differentiation

du μ μ α ν
+ αν u u = 0. (5.41)
dλ
A usual notation is to represent the derivative with respect to an invariant curve
parameter by an over dot, · = dλ d
. Then the expression for the components of the
tangent vector of the curve takes the form

dx μ
uμ = = ẋ μ , (5.42)
dλ
and the geodesic equation is written as

ẍ μ + αν
μ α ν
ẋ ẋ = 0. (5.43)

Geodesic curves on a flat surface and on a spherical surface are shown in Figs. 5.2
and 5.3, respectively.

Fig. 5.2 Geodesic curve on

a flat surface. On a flat
surface, the geodesic curve is
the minimal distance
between P and Q

Fig. 5.3 Geodesic curves on

a sphere
5.4 The Covariant Euler–Lagrange Equations 127

5.4 The Covariant Euler–Lagrange Equations

Let a particle have a world line between two points in spacetime (events) P1 and
P2 . The curve is described by an invariant parameter λ (proper time τ is used for
particles with a rest mass, i.e. for time-like world lines).
The Lagrange function is a function of the coordinates and their derivatives,

dx μ
L = L(x μ , ẋ μ ), ẋ μ ≡ . (5.44)
dλ
Note that if λ = τ then ẋ μ are the 4-velocity components.
The action integral is S = L(x μ , ẋ μ )dλ. The principle of extremal action
(Hamiltons principle) says that the world line of a particle is determined by the
condition that S shall be extremal for all infinitesimal variations of curves which
keep P1 and P2 fixed, i.e.

λ2
δ L(x μ , ẋ μ )dλ = 0, (5.45)
λ1

where λ1 and λ2 are the parameter values at P1 and P2 (Fig. 5.4).

For all the variations the following condition applies,

δx μ (λ1 ) = δx μ (λ2 ) = 0. (5.46)

We write Eq. (5.45) as

Fig. 5.4 Neighboring

geodesics in a Minkowski
diagram
128 5 Covariant Differentiation

λ2
λ2  
∂L μ ∂L μ
δ Ldλ = δx + δ ẋ dλ. (5.47)
∂xμ ∂ ẋ μ
λ1 λ1

Partial integration of the last term gives

λ2  
λ2  
∂L μ ∂ L μ λ2 d ∂L
δ ẋ dλ = δx − δx μ dλ. (5.48)
∂ ẋ μ ∂ ẋ μ λ1 dλ ∂ ẋ μ
λ1 λ1

Due to the conditions δx μ (λ1 ) = δx μ (λ2 ) = 0, the first term becomes zero. Then
we have

λ2  
∂L d ∂L
δS = − δx μ dλ. (5.49)
∂xμ dλ ∂ ẋ μ
λ1

The world line the particle follows is determined by the condition δS = 0 for any
variation δx μ . Hence, the world line of the particle must be given by
 
∂L d ∂L
− = 0. (5.50)
∂xμ dλ ∂ ẋ μ

These are the covariant Euler–Lagrange equations.

The canonical momentum pμ conjugated to a coordinate x μ is defined as

∂L
pμ ≡ . (5.51)
∂ ẋ μ
The Lagrange equations can now be written as

d pμ ∂L ∂L
= μ or ṗμ = μ . (5.52)
dλ ∂x ∂x
A coordinate which the Lagrange function does not depend on is known as a cyclic
coordinate. Hence, ∂∂xLμ = 0 for a cyclic coordinate. From Eq. (5.52) then follows:
The canonical momentum conjugated to a cyclic coordinate is a constant of motion.
That is, pμ = constant if x μ is cyclic. Note that only the covariant component of the
momentum is constant, not the corresponding contravariant component since raising
an index introduces metric functions.
A free particle in spacetime has the Lagrange function

1 1 1
L= u · u = ẋμ ẋ μ = gμν ẋ μ ẋ ν . (5.53)
2 2 2
5.4 The Covariant Euler–Lagrange Equations 129

An integral of the Lagrange equations is obtained readily from the 4-velocity

identity:

ẋμ ẋ μ = −c2 for a particle with rest mass
. (5.54)
ẋμ ẋ μ = 0 for light

The line element is

ds 2 = gμν dx μ dx ν = gμν ẋ μ ẋ ν dλ2 = 2Ldλ2 . (5.55)

Thus the Lagrange function of a free particle is obtained from the line element.

5.5 Application of the Lagrange Formalism to Free

Particles

To describe the motion of a free particle, we start by setting up the line element
of the spacetime in the chosen coordinate system. There are coordinates on which
the metric does not depend. For example, given axial symmetry we may choose the
angle θ which is a cyclic coordinate here and the conjugate (covariant) impulse Pθ
is a constant of the motion (the orbital spin of the particle). If, in addition, the metric
is time independent (stationary metric) then t is also cyclic and pt is a constant of
the motion (the mechanical energy of the particle).
A static metric is time independent and unchanged under time reversal (i.e. t →
−t). A stationary metric is independent of time, but changes under time reversal.
Examples of static metrics are the Minkowski metric and the metric in a uniformly
accelerated reference frame. The rotating cylindrical coordinate system is stationary.

5.5.1 Equation of Motion from Lagrange’s Equations

The Lagrange function for a free particle is

1
L= gμν ẋ μ ẋ ν , (5.56)
2

where gμν = gμν (x λ ), and the Lagrange equations are

 
∂L d ∂L
− = 0. (5.57)
∂xβ dτ ∂ ẋ β

Differentiation of L with respect to the coordinates and velocity components give

130 5 Covariant Differentiation

∂L 1 ∂L
β
= gμν,β ẋ μ ẋ ν , = gβν ẋ ν . (5.58)
∂x 2 ∂ ẋ β
Differentiation of the latter quantities with respect to the proper time leads to
   ·
d ∂L ∂L
≡ = ġβν ẋ ν + gβν ẍ ν = gβν,μ ẋ μ ẋ ν + gβν ẍ ν (5.59)
dτ ∂ ẋ β ∂ ẋ β

Equations (5.58) and (5.59) give

1
gμν,β ẋ μ ẋ ν − gβν,μ ẋ μ ẋ ν − gβν ẍ ν = 0. (5.60)
2
The second term on the left-hand side of Eq. (5.60) may be rewritten making use
of the fact that ẋ μ ẋ ν is symmetric in μν, as follows:

1
gβν,μ ẋ μ ẋ ν = (gβμ,ν + gβν,μ )ẋ μ ẋ ν . (5.61)
2
Hence, the equation of motion of the free particle may be written,

1
gβν ẍ ν + (gβμ,ν + gβν,μ − gμν,β )ẋ μ ẋ ν = 0. (5.62)
2

Finally, since we are free to multiply (5.62) through by g αβ , we can isolate ẍ α to

get the equation of motion in the form

1
ẍ α + g αβ (gβμ,ν + gβν,μ − gμν,β )ẋ μ ẋ ν = 0. (5.63)
2
This may be written as

ẍ α + μν
α μ ν
ẋ ẋ = 0, (5.64)

where the symbols  αμν in (5.64) are given by

α 1 αβ
μν ≡ g (gβμ,ν + gβν,μ − gμν,β ). (5.65)
2
Comparison with Eq. (5.43) shows that Eq. (5.64) describes a geodesic curve and
that  αμν are Christoffel symbols. Hence, free particles follow geodesic curves in
spacetime.
The equation of a time-like geodesic curve has been deduced from a variational
principle which says that if there are two fixed points P1 and P2 in spacetime, then
there exists an open subset of spacetime containing these two points such that among
all curves contained in this open subset, the geodesic will be the curve of longest
length between these two points. Note that for time-like curves, the length of a curve
5.5 Application of the Lagrange Formalism to Free Particles 131

between the two events P1 and P2 which the curve passes through is the proper time
used by a particle travelling from P1 to P2 along the curve.
So, the variational principle says that time-like geodesics maximizes the proper
time along the curve among all curves in spacetime close to the geodesic.
There exist non-geodesic curves between two events far away from the geodesic
curve between the events, along which a particle following the curve may have larger
proper time between the events than a particle following the geodesic curve. Consider
for example a clock at rest outside the Earth compared to a clock moving freely along
a circular path, and calculate the proper time between two meetings of the clocks. It
turns out that a standard clock on the non-geodesic particle at rest measures a larger
proper time between departure and arrival of a clock following the particle which
moves freely along a circular path in 3-space. (An exact calculation requires the
Schwarzschild spacetime. See Chap. 9) Hence in general the variational principle
has a local character.

Example 5.5.1 (Vertical free fall in a uniformly accelerated reference frame) The
Lagrange function of the particle is

1 gx 2 1 ẋ 2
L=− 1 + 2 t˙2 + , (5.66)
2 c 2 c2
where the dot denotes differentiation with respect to the proper time τ of the freely
falling particle. This gives

∂L g gx  ∂L ẋ
= − 2 1 + 2 t˙2 , = 2. (5.67)
∂x c c ∂ ẋ c
Hence the Euler–Lagrange equation
 ·
∂L ∂L
− =0 (5.68)
∂x ∂ ẋ

takes the form

 gx 
ẍ + g 1 + 2 t˙2 = 0. (5.69)
c
Furthermore we have
  ·
∂L ∂L gx 2 ∂L  gx  g  gx 2
= 0, = − 1 + 2 t˙, = −2 1 + 2 2 ẋ t˙ − 1 + 2 t¨.
∂t ∂ t˙ c ∂ t˙ c c c
(5.70)

Hence, the Lagrange equation

 ·
∂L ∂L
− =0 (5.71)
∂t ∂ t˙
132 5 Covariant Differentiation

takes the form

2g  gx 
ẋ ˙+ 1+
t t¨ = 0. (5.72)
c2 c2
Combining Eqs. (5.69) and (5.72) leads to
 gx  g  gx 2
−c2 1 + 2 2 ẋ t˙2 − c2 1 + 2 t˙t¨ + ẋ ẋ = 0. (5.73)
c c c
Integrating this equation gives the 4-velocity identity
 gx 2
−c2 1 + 2 t˙2 + ẋ 2 = −c2 . (5.74)
c
Since the metric is static the momentum
 gx 2
pt = ∂ L/∂ t˙ = − 1 + 2 t˙ (5.75)
c
is a constant of motion. The value of pt is determined by the initial condition. Inserting
the expression for t˙ into the 4-velocity identity gives

pt2 ẋ 2
− 2
+ = −1. (5.76)
1 + gx c2
c2

Assume that the particle is falling from rest at an initial position x = x0 , i.e.
ẋ(x0 ) = 0. Then
 gxo 
pt = − 1 + 2 . (5.77)
c
Inserting this into the 4-velocity identity leads after a short calculation to

x
τ
1+ gx
c2
 dx = cdτ . (5.78)
gx0 2  gx 2
x0 1+ c2
− 1+ c2 0

Performing the integration gives


gx0 2  gx 2 gτ
1+ 2
− 1+ 2 = . (5.79)
c c c

This expression shows that the particle uses a finite proper time
c gx0 
τmax = 1+ 2 (5.80)
g c
5.5 Application of the Lagrange Formalism to Free Particles 133

to arrive at the horizon at x = −c2 /g. The position of the particle as a function of
its proper time is
 
c2  gx0 2  gτ 2
x= 1+ 2 − −1 . (5.81)
g c c

5.5.2 Geodesic World Lines in Spacetime

Consider two time-like curves between two events in Minkowski spacetime. In

Fig. 5.5 they are drawn in a Minkowski diagram which refers to an inertial reference
frame.
The general interpretation of the line element for a time-like interval is: The
spacetime distance between O and P in Fig. 5.5 equals the proper time interval
between two events O and P measured on a clock moving in a such way that it is
present both at O and P.

ds 2 = −c2 dτ 2 (5.82)

The proper time interval between two events at coordinate times T 0 and T 1 are

T1 
υ 2 (T )
τ0−1 = 1− dT. (5.83)
c2
T0

Fig. 5.5 Time like geodesics

134 5 Covariant Differentiation

Fig. 5.6 Projectiles in 3-space. The particle moves between two events O and P at fixed points in
time

We can see that τ0−1 is maximal along the geodesic curve with υ(T ) = 0. Time-
like geodesic curves in spacetime have maximal distance between two points.

Example 5.5.2 (How geodesics in spacetime can give parabolas in space) A

geodesic curve between two events O and P has maximal proper time. The proper
time interval of a particle with position x and velocity v in a gravitational field with
acceleration of gravity g is

 gx 2 υ 2
dτ = dt 1 + 2 − 2 . (5.84)
c c
This expression shows that the proper time of the particle proceeds faster the
higher up in the field the particle is, and it proceeds slower the faster the particle
moves.
In Fig. 5.6, we have drawn several paths between to events O and P with the same
height in a gravitational field.
The path chosen by the particle between O and P is such that the proper time taken
by the particle between these two events is as large as possible. Thus the particle will
follow a path such that its co-moving standard clock goes as fast as possible. If the
particle follows the horizontal line between O and P it goes as slowly as possible and
the kinematical time dilation is as small as possible. Then the slowing down of its
co-moving standard clock due to the kinematical time dilation is as small as possible,
favoring a fast rate of the clock, but the particle is far down in the gravitational field
and its standard clock goes slowly for that reason. Paths further up lead to a greater
rate of proper time due to a smaller gravitational time dilation. But above the curve
drawn as a thick line, the kinematical time dilation will dominate, and the proper
time proceeds more slowly.
We shall now deduce the mathematical expression of what has been said above.
Time-like geodesic curves are curves with maximal proper time, i.e.
5.5 Application of the Lagrange Formalism to Free Particles 135

τ1

τ= −gμν ẋ μ ẋ ν dτ (5.85)
0

is maximal for a geodesic curve. However, the action

τ1
τ1
S = −2 Ldτ = − gμν ẋ μ ẋ ν dτ (5.86)
0 0

is maximal for the same curves, and this gives an easier calculation.
In the case of a vertical curve in a hyperbolically accelerated reference frame, the
Lagrangian is
  
1 gx 2 2 ẋ 2
L= − 1 + 2 t˙ + 2 . (5.87)
2 c c

Using the Euler–Lagrange equations now gives Eq. (5.69). Since spacetime is flat,
the equation represents straight lines in spacetime. The projection of such curves
into the three space of arbitrary inertial frames gives straight paths in 3-space, in
accordance with Newton’s 1st law. However, projecting it into an accelerated frame
where the particle also has a horizontal motion, and taking the Newtonian limit, one
finds the parabolic path of projectile motion.

5.5.3 Acceleration of Gravity

A free particle has vanishing 4-acceleration and moves along a time-like geodesic
curve. The i-component of the geodesic equation is

ẍ i + μν
i
ẋ μ ẋ ν = 0. (5.88)

We define the acceleration of gravity as the 3-acceleration of a free particle instan-

taneously at rest. Since the spatial components of the 4-velocity of a particle at rest
vanish, the acceleration of gravity is given by

ẍ i = − tti t˙2 . (5.89)

Hence the acceleration of gravity is given by the Christoffel symbols tti . They
vanish in a local inertial reference frame, i.e. in a freely falling non-rotating reference
frame. There is an acceleration of gravity in any non-inertial laboratory independent
of the geometrical properties of spacetime.
In the Newtonian limit dτ = dt, t˙ = 1 and the components of the acceleration of
gravity are written ẍ i = g i . It follows that
136 5 Covariant Differentiation

Fig. 5.7 The twin paradox.

The twins Eartha and Starry
each travel between two
fixed events in spacetime

g i = −tti . (5.90)

Example 5.5.3 (The Twin Paradox) Consider two twins, Starry and Eartha. Starry
travels to Proxima Centauri, four light years from the Earth, with a velocity v = 0.8c
 − 1/2
so that γ = 1 − v 2 /c2 = 5/3. The trip takes five years out and five years back.
This means that Eartha, who remains at the Earth, is ten years older when she meets
Starry at the end of her journey. Starry, on the other hand, is 10/γ = 6 years older
(Fig. 5.7).
According to the general principle of relativity, Starry can consider herself as
being at rest and Eartha as the one whom undertakes the long journey. In this picture
it seems that Starry and Eartha must be ten and six years older, respectively, upon
their return.
Let us accept the principle of general relativity as applied to accelerated reference
frames and review the twin “paradox” in this light.
Starry’s description of the trip when she sees herself as stationary is as follows.
She perceives a Lorentz-contracted distance between the Earth and Proxima Centauri,
namely, four light years ×1/γ = 2.4 light years. The Earth and Eartha travel with
v = 0.8c. Her travel time in one direction is then 2.4 light
0.8c
years
= 3 years. So the round
trip takes six years according to Starry. This means that Starry is six years older when
they meet again. This is in accordance with the result arrived at by Eartha. According
to Starry, Eartha ages by only six years ×1/γ = 3.6 years during the round trip,
not ten years as Eartha found. It is this conflict which constitutes the twin paradox.
Note that the formulation of the twin paradox makes use of the general principle of
relativity.
On turning about Starry experiences a force which reduces her velocity and accel-
erates her towards the Earth and Eartha. This means that she experiences a gravita-
tional force directed away from the Earth. Eartha is higher up in this gravitational field
and ages faster than Starry, because of the gravitational time dilation. We assume that
5.5 Application of the Lagrange Formalism to Free Particles 137

Starry has constant proper acceleration and is stationary in a uniformly accelerated

frame as she turns about. Let us follow Starry’s calculation of the ageing of Eartha
while Starry experiences this gravitation field.
The canonical momentum pt for Eartha is then (see Eq. (5.75))
 gx 2
pt = − 1 + 2 ct˙. (5.91)
c
Inserting this into the 4-velocity identity gives
 gx 2  gx 2
pt2 − c2 1 + 2 = 1 + 2 ẋ 2 (5.92)
c c
or

1 + gxc2
dτ =   dx. (5.93)
gx 2
pt − c 2 1 +
2
c2

Since ẋ = 0 for x = x2 (x2 is Eartha’s turning point according to Starry), it

follows that
 gx2 
pt = c 1 + 2 . (5.94)
c
Let x1 be Eartha’s position according to Starry at the moment that Starry begins
to notice the gravitational field, that is when Eartha begins to slow down in Starry’s
frame. Integration from x1 to x2 and inserting the value of pt gives

c  gx2 2  gx1 2
τ1−2 = 1+ 2 − 1+ 2 . (5.95)
g c c

Eartha neglected the time used by Starry to change from an outwards to a return
velocity at Proxima Centauri when she calculated the aging of herself and Starry.
This means that she took the limit of an infinitely large acceleration. Hence Starry
must do the same. In this limit Starry gets

1
lim τ1−2 = x 2 − x12 . (5.96)
g→∞ c 2

Now setting x2 = 4 years and x1 = 2.4 light years, respectively, we get

lim g→∞ τ1−2 = 3.2 years. Hence, Eartha’s aging as she turns about is, according
to Starry,

τEartha = 2 lim τ1−2 = 6.4 years.

g→∞
138 5 Covariant Differentiation

So all in all Eartha has aged by τEartha = 3.6+6.4 = 10 years, according to Starry,
which is just what Eartha herself found.
We have now seen: The formulation of the twin paradox makes use of the gen-
eral principle of relativity for accelerated motion. The paradox arises if one tries to
describe the ageing of both twins from the point of view of each twin without taking
into account the effect of gravity upon the rate of time. Hence, both the formulation
and the solution of the twin paradox involves the general theory of relativity.

5.5.4 Gravitational Shift of Wavelength

We shall consider the shift of wavelength of light moving up or down in a gravitational

field. The 4-momentum of a particle with relativistic energy E and spatial velocity v
(3-velocity) is given by

E
P = 2 (c, v). (5.97)
c

Let U be the 4-velocity of an observer. In a co-moving orthonormal basis of the

observer we have U = (c, 0, 0, 0). This gives

U · P = − Ê. (5.98)

Hence, the energy of a particle with 4-momentum P measured by an observer

with 4-velocity U is

Ê = −U · P.
 (5.99)

Let E e = −(U · P)
 e and E 0 = − (U · P) 0 be the energy of a photon, measured
locally by observers at rest in the emitter and observer positions, respectively. This
gives1

Ee E0
= . (5.100)
 
(U · P)e   0
(U · P)

Let the wavelength of the light, measured by the emitter and observer, be λe and
λ0 , respectively. We then have

hc hc
λe = , λ0 = (5.101)
Ee E0

·
1A B = A0 B 0 + A1 B 1 + · · · = g00 A0 B 0 + g11 A1 B 1 + · · · , an orthonormal basis gives A · B =
−A0 B 0 + A1 B 1 + · · · .
5.5 Application of the Lagrange Formalism to Free Particles 139

which gives

(U · P)
 e
λ0 = λ. (5.102)
 0 e
(U · P)

This formula may be applied to observers with arbitrary motion.

We shall here restrict ourselves to an observer at rest in a coordinate system with
time-independent diagonal metric. Then we have

dt
U · P = U t Pt = c Pt , (5.103)
dτ
where Pt is a constant of motion (since t is a cyclic coordinate) for photons and hence
has the same value in emitter and observer positions. The line element is

ds 2 = gtt c2 dt 2 + gii (dx i )2 . (5.104)

Using the physical interpretation (3.82) of the line element for a time-like interval,
we obtain for the proper time of an observer at rest
√
dτ 2 = −gtt dt 2 ⇒ dτ = −gtt dt. (5.105)

Hence
dt 1
=√ , (5.106)
dτ −gtt

which gives
c
U · P = √ Pt . (5.107)
−gtt

Inserting this into the expression for the wavelength (5.102) gives the formula for
the gravitational shift of the wavelength of light emitted and observed by an emitter
and an observer at rest in the reference frame,

(gtt )0
λ0 = λe . (5.108)
(gtt )e

Example 5.5.4 (Gravitational redshift or blueshift of light) Inserting the metric of

a uniformly accelerated reference frame with
 gx 2
gtt = − 1 + 2 (5.109)
c
140 5 Covariant Differentiation

gives

1+ gx0
c2
λ0 = λe . (5.110)
1+ gxe
c2

The gravitational redshift (light moving upwards in a gravitational field) or

blueshift (light moving downwards) is

λ0 − λe 1+ gx0
c2
g
(x
c2 0
− xe ) gh
zG = = −1= ≈ , (5.111)
λe 1+ gxe
c2
1+ gxe
c2
c2

where h = x0 − xe is the difference in height between transmitter and receiver. In

the case of motion upwards the observed wavelength increases with the height. By
motion of radiation downwards there is a blueshift.
This effect was first measured in 1959 in the Pound–Rebka experiment with a
height difference between emitter and receiver of h = 22.5 m giving z G = 2.5 ×
10−15 .

5.6 Connection Coefficients

The covariant directional derivative of a scalar field f in the direction of a vector u

is defined as

∇u f ≡ u( f ). (5.112)

Here the vector u should be taken as a differential operator. (In coordinate basis,
u = u μ ∂ ∂x μ .) The directional derivative along a basis vector eν is written as

∇ν ≡ ∇eν . (5.113)

Hence ∇μ ( ) = ∇eμ ( ) = eμ ( ).

Definition 5.6.1 (Koszul’s connection coefficients in an arbitrary basis.) In an

α
arbitrary basis the Koszul connection coefficients, μν , are defined by

α
∇ν eμ ≡ μν eα , (5.114)

α α
which may also be written eν (eμ ) = μν eα . In coordinate basis μν is reduced to
α α
Christoffel symbols and one often writes eμ,ν = μν eα . In an arbitrary basis μν has
no symmetry.
5.6 Connection Coefficients 141

Fig. 5.8 Rotating coordinate system. The non-rotating coordinate system (X, Y ) and the rotating
system (x, y), rotating with angular velocity ω

Example 5.6.1 (The connection coefficients in a rotating reference frame) In an iner-

tial reference frame, there is a co-moving Cartesian coordinate system with coordi-
nates X, Y and corresponding plane polar coordinates R, . In a reference frame
rotating with angular velocity ω, the corresponding co-moving coordinates are x, y
and r, θ (Fig. 5.8). The coordinate clocks in the rotating frame shows the same time
as those in the inertial frame, t = T .
The transformation between the coordinates is

t = T, r = R, θ =  − ωT,
X = R cos , Y = R sin ,
X = r cos(θ + ωt), Y = r sin(θ + ωt). (5.115)

The time-like coordinate basis vector in the rotating frame is calculated from

∂ ∂X ∂ ∂Y ∂ ∂T ∂
et = = + + , (5.116)
∂t ∂t ∂ X ∂t ∂Y ∂t ∂ T
Using this formula together with the transformation (5.115), and corresponding
formulae for the spatial basis vectors, give

et = −r ω sin(θ + ωt)e X + r ω cos(θ + ωt)eY + eT ,

∂X ∂ ∂Y ∂
er = + = cos(θ + ωt)e X + sin(θ + ωt)eY ,
∂r ∂ X ∂r ∂Y
∂X ∂ ∂Y ∂
eθ = + = −r sin(θ + ωt)e X + r cos(θ + ωt)eY . (5.117)
∂θ ∂ X ∂θ ∂Y
142 5 Covariant Differentiation

We are going to find the Christoffel symbols, which involves differentiation of

the basis vectors. The transformation (5.117) makes this easy, since eX , eY , eT are
constant. Differentiation gives

∇t et = −r ω2 cos(θ + ωt)e X − r ω2 sin(θ + ωt)eY . (5.118)

The connection coefficients are calculated from Eq. (5.114). Note that to calculate
α
μν , the right-hand side of Eq. (5.118) has to be expressed by the basis which we are
differentiating. Comparing the right-hand side of Eq. (5.118) with the expression for
er we see that ∇t et = −r ω2 er giving ttr = −r ω2 . The other nonzero Christoffel
symbols are

ω 1
rθt = trθ = θ
, θr = rθθ = , θt
r
= tθ
r
= −r ω, θθ
r
= −r. (5.119)
r r
Example 5.6.2 (Acceleration in a non-rotating reference frame) The covariant
expression for the acceleration is

r¨ = v˙ = v;i j v j ei = (v̇ i + αβ

i
v α v β )ei , (5.120)

where · ≡ d/dt. Here i, j and k are space indices. Inserting the Christoffel symbols
(5.31), (5.32) for plane polar coordinates gives
 
2  
ainert = (r̈ − r θ̇ 2 )er + θ̈ + ṙ θ̇ eθ = (r̈ − r θ̇ 2 )er̂ + r θ̈ + 2ṙ θ̇ eθ̂ . (5.121)
r

Example 5.6.3 (Acceleration in a rotating reference frame) Inserting the Christoffel

symbols from Example 5.6.1 into Eq. (5.119) gives
 
2
r ˙
arot = (r̈ − r θ̇ 2 + ttr t˙2 + θt r ˙
θ̇ t + tθ t θ̇ )er + θ̈ + ṙ θ̇ + rθt ṙ t˙ + trθ t˙ṙ eθ
r
= (r̈ − r θ̇ 2 − r ω2 − 2r ωθ̇)er̂ + (r θ̈ + 2ṙ θ̇ + 2ṙ ω)eθ̂
= ainert − (r ω2 + 2r ωθ̇ )er̂ + 2ṙ ωeθ̂ . (5.122)

The angular velocity of the reference frame, is ω  = ωez . We also introduce

r = r er . The velocity in a rotating reference frame is then

r˙ = ṙ er + r e˙ r . (5.123)

Furthermore

der ∂ er dx i
e˙ r = = i = v i er,i . (5.124)
dt ∂ x dt
Using Definition 5.6.1 in a coordinate basis, this may be written as
5.6 Connection Coefficients 143

e˙ r = v i ri e j .
j
(5.125)

Applying the expressions for the Christoffel symbols in Example 5.6.1 and
introducing orthonormal basis, we get

1
e˙ r = v θ rθθ eθ = θ̇ eθ = θ̇ eθ̂ . (5.126)
r
Hence

v = r˙ = ṙ er̂ + r θ̇ eθ̂ . (5.127)

Inserting this into the expression for the acceleration gives

r¨ rot = r¨ inert + ω

 × (ω
 × r) + 2ω
 × v. (5.128)

We see that the centrifugal acceleration (the term in the middle) and the Coriolis
acceleration (last term) are contained in the expression for the covariant derivative.

5.6.1 Structure Coefficients

The commutator of two vectors, u and v, expressed by covariant directional

derivatives is given by

u , v] = ∇u v − ∇v u.

[ (5.129)

Let u = eμ and v = eν . We then have

[eμ , eν ] = ∇μ eν − ∇ν eμ . (5.130)

Using the definitions of the connection and structure coefficients we get

α α α
cμν eα = (νμ − μν )eα . (5.131)

Thus (in a torsion free space)

α α α
cμν = νμ − μν . (5.132)

In coordinate basis we have

∂ ∂
eμ = , eν = ν , (5.133)
∂xμ ∂x
144 5 Covariant Differentiation

and therefore
     
∂ ∂ ∂ ∂ ∂ ∂ ∂2 ∂2
[eμ , eν ] = μ
, ν
= μ ν
− ν μ
= μ ν − ν μ = 0.
∂x ∂x ∂x ∂x ∂x ∂x ∂x ∂x ∂x ∂x
(5.134)
α
Equation (5.134) shows that cμν = 0, and Eq. (5.132) then implies that the
connection coefficients in Eq. (5.98) are symmetrical in a coordinate basis:
α α
νμ = μν . (5.135)

5.7 Covariant Differentiation of Vectors, Forms

and Tensors

5.7.1 Covariant Differentiation of Vectors

The covariant directional derivative of a vector field was defined in Eq. (5.34). It
should be noted, however, that there are several ways of defining the covariant deriva-
tive that are consistent with each other. Here we shall formulate some definitions that
are alternative to those given in Sect. 5.2.
Definition 5.7.1 (Covariant derivative of a vector field) The covariant derivative of
a vector in an arbitrary basis is defined by

∇ν A = ∇ν (Aμ eμ ) = ∇ν Aμ eμ + Aα ∇ν eα . (5.136)

Using the definition (5.114) of the connection coefficients, this may be written

∇ν A = [eν (Aμ ) + Aα αν

μ
]eμ . (5.137)

The covariant derivative of the components of a vector field was introduced in

Eq. (5.23) by the requirement that the derivative of a tensor component shall transform
as a tensor component. There exists, however, another point of view. The ordinary
partial derivative of the vector components does only describe the change of the
components of a vector, not of the vector itself. The covariant derivatives of the
vector components, on the other hand, describe the change of the vector itself. Let
us see how that comes about.
Definition 5.7.2 (Covariant derivative of a vector component) The covariant
μ
derivative of a vector component Aμ is written as A;ν and is defined by

∇ν A = Aμ;ν eμ . (5.138)

5.7 Covariant Differentiation of Vectors, Forms and Tensors 145

Comparison with Eq. (5.137) shows that

μ
A;ν = eν (Aμ ) + Aα αν
μ
. (5.139)

Note that the term with the connection coefficients come from the second term in
Eq. (5.136). Hence, it represents the change of the basis vector field with position and
time. The first term describes the changes of the vector component with position and
time. The sum describes the change of the vector itself, as seen from the definition
(5.138).

5.7.2 Covariant Differentiation of Forms

Definition 5.7.3 (Covariant directional derivative of a 1-form field) Given a vector

field A and a 1-form field α, the covariant directional derivative of α in the direction
of the vector u is defined by

 ≡ ∇u [α( A)
(∇u α)( A)  ] − α(∇u A).
 (5.140)
  
αμ Aμ

Let the 1-form α be a basis field, α = ωμ . Then ωμ (eν ) ≡ δνμ . Furthermore let

A = eν and u = eλ . We then have

(∇λ ωμ )(eν ) = ∇λ [ωμ (eν )] − ωμ (∇λ eν ). (5.141)

  
μ
δν

The covariant directional derivative ∇λ of a constant scalar field is zero, ∇λ δνμ = 0.

We therefore get
μ
(∇λ ωμ )(eν ) = −ωμ (∇λ eν ) = −ωμ (νλ
α α μ
eα ) = −νλ α μ
ω (eα ) = −νλ δα = −νλ .
(5.142)

The contraction of a 1-form and a basis vector gives the components of the 1-form,
α(eν ) = αν . Equation (5.142) gives the directional derivatives of the basis forms and
μ
tells that the ν-component of ∇λ ωμ is equal to −νλ . Hence
μ ν
∇λ ωμ = −νλ ω . (5.143)

Using the rule for differentiation of a product gives

μ
∇λ α = ∇λ (αμ ωμ ) = ∇λ (αμ )ωμ + αμ ∇λ ωμ = eλ (αμ )ωμ − αμ νλ ων . (5.144)

This motivates the following definition.

146 5 Covariant Differentiation

Definition 5.7.4 (Covariant derivative of a 1-form) The covariant derivative of a

1-form α = αμ ωμ is given by
μ
∇λ α = [eλ (αν ) − αμ νλ ]ων . (5.145)

Definition 5.7.5 (Covariant derivative of a 1-form component) The covariant

derivative of the 1-form components αμ is denoted by αν;λ and defined by

∇λ α ≡ αν;λ ων . (5.146)

It follows that
μ
αν;λ = eλ (αν ) − αμ νλ . (5.147)
μ
Note that νλ in Eq. (5.147) are not in general the same as the Christoffel symbols.
However, in coordinate basis we get
μ
αν;λ = αν,λ − αμ λν , (5.148)
μ μ
where λν = νλ are Christoffel symbols.

5.7.3 Covariant Differentiation of Tensors of Arbitrary Rank

Definition 5.7.6 (Covariant derivative of tensors) Let A and B be two tensors of

arbitrary rank. The covariant directional derivative along a basis vector eλ of a tensor
A ⊗ B of arbitrary rank is defined by

∇λ (A ⊗ B) ≡ (∇λ A) ⊗ B + A ⊗ (∇λ B). (5.149)

We will use Eq. (5.149) to find the formula for the covariant derivative of the
components of a tensor of rank 2:

∇α S = ∇α (Sμν ωμ ⊗ ων )
= (∇α Sμν )ωμ ⊗ ων + Sμν (∇α ωμ ) ⊗ ων + Sμν ωμ ⊗ (∇α ων )
β β
= (Sμν,α − Sβν μα − Sμβ να )ωμ ⊗ ων , (5.150)

where Sμν,α = eα (Sμν ).

Definition 5.7.7 (Covariant derivative of tensor components) The covariant deriva-

tive Sμν;α of the covariant components of a tensor of rank 2 is defined by

∇α S = Sμν;α ωμ ⊗ ων (5.151)
5.7 Covariant Differentiation of Vectors, Forms and Tensors 147

we get
β β
Sμν;α = Sμν,α − Sβν μα − Sμβ να . (5.152)

For the metric tensor we get

β β
gμν;α = gμν,α − gβν μα − gμβ να . (5.153)

From

gμν = eμ · eν (5.154)

we get
 
gμν,α = ∇α gμν = ∇α eμ · eν = (∇α eμ ) · eν + eμ (∇α eν )
β β β β
= μα eβ · eν + eμ · να eβ = gβν μα + gμβ να . (5.155)

This means that

gμν;α = 0. (5.156)

So the metric tensor is a constant tensor.

5.8 The Cartan Connection

Definition 5.8.1 (Exterior derivative of a basis vector)

ν
deμ ≡ μα eν ⊗ ωα . (5.157)

ν
Note that this can be thought of as a 1-form with vector components μα eν , a
vectorial form. The exterior derivative of a vector field is

d A = d(eμ Aμ ) = eν ⊗ dAν + Aμ deμ . (5.158)

In arbitrary basis the exterior derivative of the vector components are

dAν = eλ (Aν )ωλ (5.159)

∂
In coordinate basis eλ (Aν ) = ∂xλ
(Aν ) = Aν,λ , giving

d A = eν ⊗ [eλ (Aν )ωλ ] + Aμ μλ

ν
eν ⊗ ωλ = (eλ (Aν ) + Aμ μλ
ν
)eν ⊗ ωλ . (5.160)
148 5 Covariant Differentiation

Together with Eq. (5.139), this gives

d A = Aν;λ eν ⊗ ωλ . (5.161)

Definition 5.8.2 (Connection forms μν ) The connection forms μν are 1-forms,
defined by

deμ ≡ eν ⊗ νμ . (5.162)

From Eqs. (5.157) and (5.162) we have

ν
deμ = μα eν ⊗ ωα = eν ⊗ μα
ν
ωα = eν ⊗ νμ . (5.163)

This shows that

νμ = μα
ν
ωα . (5.164)

The exterior derivatives of the components of the metric tensor:

dgμν = d(eμ · eν ) = eμ · deν + eν · deμ , (5.165)

where the meaning of the dot is defined as follows.

Definition 5.8.3 (Scalar product between vector and 1-form) The scalar product
between a vector u and a vectorial 1-form A = Aμν eμ ⊗ ων is defined by

u · A ≡ u α Aμν (eα · eμ )ων . (5.166)

Using this definition, we get

dgμν = (eμ · eλ )λν + (eν · eγ )γμ = gμλ λν + gνγ γμ . (5.167)

Lowering an index gives

dgμν = μν + νμ . (5.168)

In an orthonormal basis field there is Minkowski metric: gμ̂ν̂ = ημ̂ν̂ which is

constant. Then we have dgμ̂ν̂ = 0 which implies

ν̂ μ̂ = −μ̂ν̂ . (5.169)

The connection forms with lower indices are antisymmetric in an orthonormal

basis. It follows from (5.164) and (5.169) that

ν̂ μ̂α̂ = −μ̂ν̂ α̂ . (5.170)

5.8 The Cartan Connection 149

Note also that

îtˆ ĵ = −tˆî ĵ = î tˆ ĵ = tîˆ ĵ ,  îĵ k̂ = −î k̂ .

ĵ
(5.171)

We shall now consider Cartan’s 1 structure equation which gives a relationship

between the structure coefficients and the connection coefficients. As a preparation
for deducing Cartan’s 1 structure equation we shall first deduce a mathematical
identity valid for 1-forms.
The commutator of two vectors is itself a vector, and the contraction of a 1-form
with a vector is essentially the same quantity as a scalar product of two vectors.
Hence for a 1-form α we have
   ν  
α u, v = αν u, v = αν u μ v ν ,μ −v μ u ν ,μ . (5.172)

From this we get

   
v ) = u μ eμ (αν v ν ) = u μ αν ,μ v ν + αν v ν ,μ ,
u α( (5.173)

and
   
u ) = v μ αν ,μ u ν + αν u ν ,μ ,
v α( (5.174)

Furthermore
 
u , v) = αμ ,ν −αν ,μ u ν v μ .
d α( (5.175)

It follows from these expressions that

     
d α(
u , v) = u α( u ) − α u, v .
v ) − v α( (5.176)

This identity is valid in an arbitrary basis.

Let α = ωρ , u = eμ , v = eν . Then
        
d ωρ eμ , eν = eμ ωρ (eν ) − eν ωρ eμ − ωρ eμ , eν
     α 
= eμ δνρ − eν δμρ − ωρ cμν α ρ
eα = −cμν ρ
δα = −cμν . (5.177)

A 2-form applied to two basis vectors is equal to its components. Hence we get
Cartan’s 1 structure equation

1 ρ μ
dωρ = − cμν ω ∧ ων . (5.178)
2
Combining this with Eq. (5.132) and utilizing the antisymmetry of the wedge
product we have
150 5 Covariant Differentiation

1 ρ
dωρ = − (νμ ρ
− μν )ωμ ∧ ων = μν
ρ
ωμ ∧ ων . (5.179)
2
Together with Eq. (5.164) this gives

dωρ = −ρν ∧ ων . (5.180)

In coordinate basis, we have ωρ = dx ρ . Thus,

dωρ = d2 x ρ = 0. (5.181)

The exterior derivatives of the basis forms vanish in coordinate basis. We also have
ρ
cμν = 0, and Cartan’s 1 structure equation is reduced to an identity. This formalism
cannot be used in coordinate basis! But due to the antisymmetry (5.169) the Cartan
formalism is particularly useful in orthonormal basis.
Example 5.8.1 (Cartan connection in an orthonormal basis field in plane polar
coordinates.) Here the line-element is

ds 2 = dr 2 + r 2 dθ 2 .

Introducing basis forms in an orthonormal basis field (where the metric is gr̂ r̂ =
gθ̂ θ̂ = 1), the metric tensor takes the form

g = gr̂ r̂ ωr̂ ⊗ ωr̂ + gθ̂ θ̂ ωθ̂ ⊗ ωθ̂ = ωr̂ ⊗ ωr̂ + ωθ̂ ⊗ ωθ̂ .

Hence

ωr̂ = dr, ωθ̂ = r dθ

Exterior differentiation gives

1 r̂
dωr̂ = d2 r = 0, dωθ̂ = dr ∧ dθ = ω ∧ ωθ̂ .
r
From Eq. (5.180) we then have

μ̂ μ̂ μ̂
dωμ̂ = −ν̂ ∧ ων = −r̂ ∧ ωr̂ − θ̂ ∧ ωθ̂ .

Furthermore dωr̂ = 0, which gives

r̂θ̂ = θ̂r̂ θ̂ ωθ̂ (5.182)

since ωθ̂ ∧ ωθ̂ = 0. Note that r̂r̂ = 0 because of the antisymmetry ν̂ μ̂ = −μ̂ν̂ .
We also have
5.8 The Cartan Connection 151

1
dωθ̂ = − ωθ̂ ∧ ωr̂ ,
r
and from Eq. (5.180) it follows that

dωθ̂ = − r̂θ̂ ∧ ωr̂ − θ̂θ̂ ∧ωθ̂ ,


=0

From Eq. (5.164) we have

r̂θ̂ = r̂θ̂θ̂ ωθ̂ + r̂θ̂r̂ ωr̂ , (5.183)

giving

1
r̂θ̂θ̂ = .
r

Furthermore r̂θ̂ = −r̂θ̂ . Using Eqs. (5.182) and (5.183) we get

1
r̂θ̂r̂ = 0, θ̂r̂ θ̂ = −
r
giving

1
r̂θ̂ = −r̂θ̂ = − ωθ̂ .
r

5.9 Covariant Decomposition of a Velocity Field

I order to develop some intuition about the kinematics of fluids we shall first consider
an ordinary 3-velocity field in Newtonian hydrodynamics.

5.9.1 Newtonian 3-Velocity

The total or material derivative of the velocity is

v
D ∂ v
= + (
v · ∇)
v. (5.184)
Dt ∂t
152 5 Covariant Differentiation

Here ∂ v/∂t is the local derivative which represents the change of the velocity field
at a certain position, for example the increase with time of the velocity at certain place
in a river due to rain. The term ( v · ∇) v is called the convective derivative and is
sometimes written d v /dt. It represents the change of velocity with position at a fixed
point of time, for example the slowing down of the velocity further downwards due to
a widening of the river. The total derivative represents the change of the velocity field
following a fluid particle, due to both local changes with time and the inhomogeneity
of the velocity field. The component version of Eq. (5.184) is

Dv i ∂v i ∂v i
= + vj j . (5.185)
Dt ∂t ∂x
Introducing a Cartesian coordinate system with coordinates (x, y, z), the convec-
tive derivative may be written in matrix form as
⎛ dv x ⎞ ⎛ ∂v x ∂v x ∂v x
⎞⎛ ⎞
∂ x ∂ y ∂z vx
⎠=⎜ ⎟⎝ y ⎠
dt
⎝ dv y
⎝
∂v ∂v y ∂v y
y
⎠ v . (5.186)
dt ∂ x ∂ y ∂z
dv z ∂v z ∂v z ∂v z
dt ∂ x ∂ y ∂z
vz

The 3 × 3 matrix may be separated into three parts: The trace

∂v x ∂v y ∂v z
θ= + + , (5.187)
∂x ∂y ∂z

The antisymmetric part

⎛ ⎞
∂v x ∂v y ∂v x ∂v z
0 ∂y
− ∂ x ∂z
− ∂x
1⎜ ∂v y ∂v z ⎟
ωi j = ⎝ 0 ∂z
− ∂y ⎠, (5.188)
2 minus
the same 0

and the symmetric, trace-free part,

⎛  x  ⎞
2 ∂v − ∂v y ∂v z ∂v x
+ ∂v ∂v x ∂v z
y
1
− +
⎜ 3 ∂x ∂y ∂z  y ∂ y ∂ x  ∂z ∂x ⎟
⎜ ⎟
2 ∂v − ∂v − ∂v ∂v x ∂v z
x z
σi j =⎜ 1
∂y ∂x ∂z ∂z
+ ∂x ⎟.
⎝ 3  z  ⎠
2 ∂v ∂v x ∂v
y
1
3 ∂z
− ∂x
− ∂y
(5.189)

Here θ represents the volume expansion (or contraction), ωi j rotation and σi j

the shear. The component form of these expressions is

vi, j = θi j + ωi j + σi j , (5.190)
5.9 Covariant Decomposition of a Velocity Field 153

1 k
θi j = v ,k δi j , (5.191)
3
1  1
ωi j = vi , j −v j ,i = (∇ × v)i j , (5.192)
2 2
1  1
σi j = vi , j +v j ,i − v k ,k δi j . (5.193)
2 2

5.9.2 Relativistic 4-Velocity

We shall now find the relativistic generalizations of these expressions, i.e. we shall
find covariant expressions that represent expansion, rotation and shear as measured
by an observer following the fluid.
The covariant derivative may be thought of as a generalization of the total deriva-
tive. We first separate the covariant directional derivative of a 4-velocity field in a
component along the 4-velocity and a component orthogonal to it. The directional
derivative of the 4-velocity along itself is the 4-acceleration. It is orthogonal to the
4-velocity and has covariant components
  μ
u˙ α = u α;μ u μ = u α,μ − u ν αμ
ν
u . (5.194)

Putting them equal to zero gives the geodesic equation describing the world line
of a freely falling particle.
The projection of the tensor with components u α;β into the spatial simultaneity
space orthogonal to the 4-velocity, is
 
u α;β ⊥
= u μ;ν Pαμ Pβν . (5.195)

where Pαμ are the mixed components of the projection tensor defined in Eq. (4.104).
In the same way as we did in the Newtonian case, this may be separated in three
parts.
Expansion:

1 μ
θαβ = θ Pαβ , θ = u ;μ . (5.196)
3
Shear:
1  1 μ
σαβ = u μ;ν + u ν;μ Pαμ Pβν − u ;μ Pαβ . (5.197)
2 3
Rotation:
154 5 Covariant Differentiation

1 
ωαβ = u μ;ν − u ν;μ Pαμ Pβν . (5.198)
2
The covariant derivative of a 4-velocity field can now the be separated in
expansion, shear, rotation and a 4-acceleration-term as follows

u α;β = θαβ + σαβ + ωαβ − u̇ α u β (5.199)

In the relativistic literature, the expressions for shear and rotation are written in
different ways that are found by inserting the components of the projection tensor
from Eq. (4.106) into Eq. (5.195),
  
u μ;ν Pαμ Pβν = u μ;ν δ ηα + u μ u α δβν + u ν u β . (5.200)

Using that

u μ;ν δαμ δβν = u α;β , u μ;ν δαμ u ν u β = u α;ν u ν u β = u̇ α u β , (5.201)

we get

u μ;ν Pαμ Pβν = u α;β + u̇ α u β + u μ;β u μ u α + u̇ μ u μ u α u β . (5.202)

It follows from the 4-velocity identity, u μ u μ = −1, that the 4-acceleration is

orthogonal to the 4-velocity, u̇ μ u μ = 0, and that u μ u μ ; β = 0, giving u μ; β u μ = 0.
Hence Eq. (5.202) simplifies to

u μ;ν Pαμ Pβν = u α;β + u̇ α u β . (5.203)

Furthermore
μ μ
u α;μ Pβ = u α;μ δβ + u α;μ u μ u β = u α;β u̇ α + u β , (5.204)

or

u α;β = u α;μ Pβμ − u̇ α u β . (5.205)

We then get
μ
u α;μ Pβ = θαβ + σαβ + ωαβ , (5.206)

with the alternative expressions for shear and rotation

1 μ
 1
μ
σαβ = u α;μ Pβ + u β;μ Pαμ − u ;μ Pαβ , (5.207)
2 3
5.9 Covariant Decomposition of a Velocity Field 155

1 μ

ωαβ = u α;μ Pβ − u β;μ P αμ . (5.208)
2

5.10 Killing Vectors and Symmetries

Killing vectors are useful for describing the symmetry properties of space in an
invariant way.
Definition 5.10.1 (Killing vectors) We will here define a Killing vector as a vector
ξ which obeys Killing’s equation

ξμ;ν + ξν;μ = 0. (5.209)

In this form Killing’s equation is valid in an arbitrary basis. In a coordinate basis

the equation reduces to
α
ξμ,ν + ξν,μ = 2ξα μν . (5.210)

One may show that if ξ (1) and ξ (2) are two Killing vectors, and a, b are con-
stants, then a ξ (1) +bξ (2) is a Killing vector. Furthermore, the commutator ξ (1) , ξ (2)
between two Killing vectors is also a Killing vector.
In an n-dimensional space, there are maximally n(n + 1)/2 linearly independent
Killing vectors. In four-dimensional spacetime, there may be up to ten such vectors.
A metric and the corresponding space that admit the maximum number of Killing
vectors is said to be maximally symmetric.
Example 5.10.1 (Killing vectors of an Euclidean plane) We consider a two-
dimensional Euclidean plane with Cartesian coordinates. Then the line-element has
the form

dl 2 = dx 2 + dy 2 , (5.211)

and the Christoffel symbols vanish. Hence, the Killing equations reduce to

ξx,x = ξ y,y = 0, ξx,y + ξ y,x = 0. (5.212)

These equations have the following solutions,

∂ ∂ ∂ ∂
ξ1 = , ξ2 = , ξ1 = x −y . (5.213)
∂x ∂y ∂y ∂x

The first vector represents invariance by translation in the x-direction, the second
in the y-direction, and the third vector represents invariance by rotation about an
156 5 Covariant Differentiation

axis orthogonal to the plane. A two-dimensional space with three Killing vectors is
maximally symmetric. Hence the Euclidean space is maximally symmetric.

Definition 5.10.2 (Invariant basis) An invariant basis is defined as a basis where

the basis vectors commute with a Killing vector,

ξ , eμ = 0. (5.214)

Since the commutator of two coordinate basis vectors vanish, a consequence of

this definition is that if a basis vector of a coordinate basis is a Killing vector, then
this coordinate basis is invariant.  
The commutator of two vectors, u, v is also called the Lie derivative of v with
respect to u, which is writte

(5.215)

The Lie derivative of a scalar function with respect to a vector is the directional
derivative of the scalar function along the vector,

(5.216)

We have the rule

(5.217)

From this we get

(5.218)

Equations (5.214) and (5.215) give

(5.219)

It follows from Eq. (5.215) for an invariant basis, together with Eqs. (5.218) and
(5.219), that
 
ξ gμν = 0 (5.220)

for an invariant basis. Hence the components of the metric tensor are constants along
Killing vectors in a space with an invariant basis field.
We shall finally find the relation between the Killing vectors of a space and the
constants of motion for a particle moving freely in the space. A free particle moves
along a geodesic curve with the equation
5.10 Killing Vectors and Symmetries 157

∇u u = 0 (5.221)

where u is the 4-velocity of the particle. Consider the scalar product u · ξ , where ξ is
a Killing vector field. The covariant directional derivative of this product along the
geodesic curve is
   
∇u u · ξ = u σ u μ ξμ ;σ = u σ u μ;σ ξμ + u σ u μ ξμ;ν . (5.222)

Here the first term vanishes because of Eq. (5.221), and the second vanishes since
u σ u μ is symmetric in σ and μ while ξμ;ν is antisymmetric in σ and μ due to the
Killing equation (5.209). Thus
 
∇u u · ξ = 0. (5.223)

We then have the result that u · ξ is constant along a geodesic curve. For a particle
with constant rest mass, this may also be expressed as: p · ξ is constant along a
geodesic curve, where p is the 4-momentum of the particle.
Assume that ξα is a time-like Killing vector associated with a cyclic coordinate x α .
Then a coordinate system can be chosen such that ξα = (−1, 0, 0, 0), i.e. ξα = −δα0 e0 .
In this case

p · ξα = − pμ δα0 = − pα0 = constant. (5.224)

which is the energy of the particle. Correspondingly, if ξα is a space-like Killing

vector associated with a cyclic coordinate, and one chooses a coordinate system so
that ξα = (0, 1, 0, 0), one obtains

pα1 = constant. (5.225)

This represents conservation of momentum in the x-direction for a free particle and
is in accordance with the Lagrangian dynamics that the covariant momenta conjugate
to cyclic coordinates are constants of motion for freely falling particles. The Killing
vectors describe symmetries of spacetime. Hence the equation p · ξα = constant
relates constants of motion of a free particle to symmetries of spacetime.

5.11 Covariant Expressions for Gradient, Divergence,

Curl, Laplacian and D’Alembert’s Wave Operator

The exterior derivative d of a p-form was defined in Eq. (5.6). It is an antisymmetric

derivative giving a (p + 1)-form. The so-called Hodge dual of the exterior derivative
δ is called the codifferential. In an n-dimensional space, it is defined by
158 5 Covariant Differentiation

δ = (−1)n( p+1)+1 d. (5.226)

where the Hodge star operator acting on a form gives the dual form defined in Eq.
(3.118). The codifferential acting on a p-form gives out a ( p − 1)-form. For example
the codifferential of a 1-form is a scalar function. If the dimension of the space is
even, for example in four-dimensional spacetime, then

δ = −d. (5.227)

Since there only exist forms with p − 1 ≥ 0 the codifferential of a scalar func-
tion f is not defined as a form equation, but according to a separate definition the
codifferential of a scalar function vanishes,

δ f = 0. (5.228)

In the same manner as the exterior derivative as applied to a p-form with scalar
components, satisfies Poincare’s lemma, d2 = 0, the codifferential satisfies

δ δ = δ 2 = 0. (5.229)

A generalization of the Laplacian operator in 3-space

∂2 ∂2 ∂2
= + + , (5.230)
∂x2 ∂ y2 ∂z 2

is the form operator called the Hodge Laplacian or alternatively the Laplace–Beltrami
operator, defined by

 = δ d + d δ. (5.231)

Due to Eq. (5.228) the Hodge Laplacian as applied to a scalar function takes the
form

φ = δ dφ = ddφ. (5.232)

The Hodge Laplacian of a 1-form is

 α = (dd + dd)α. (5.233)

Let us find the covariant form of the Hodge Laplacian in four-dimensional space-
time as applied to a scalar function f. Using Eq. (3.118) and the definition (5.6) of
the exterior derivative, Eq. (5.232) gives
 
1 αβγ δ ∂ ∂f
f = V 4 α Vλβγ δ
4! ∂x ∂ xλ
5.11 Covariant Expressions for Gradient, Divergence, Curl, … 159
 
1 −1 αβγ δ ∂ √ λρ ∂ f
= √ ε −gε λβγ δ g
3! −g ∂xα ∂xρ
 
3! 1 ∂ √ ∂f
= − √ δλα α −gg λρ ρ
3! −g ∂ x ∂x
 
1 ∂ √ λρ ∂ f
= −√ −gg . (5.234)
−g ∂ x λ ∂xρ
√
where Vαβγ δ = −gεαβγ δ , are the components of the volume form as given in
Eq. (3.116), and the minus sign comes from

1 1√ 1 1
V αβγ δ = Vαβγ δ = −gεαβγ δ = − √ εαβγ δ = − √ εαβγ δ . (5.235)
g g −g −g

In this case, the Hodge Laplacian reduces to minus the d’Lambertian wave
operator,  = −. Hence we have the covariant form of the wave operator
 
1 ∂ √ λρ ∂ f
f = √ −gg . (5.236)
−g ∂ x λ ∂xρ

Performing the same calculation for a scalar field in 3-space, we get the covariant
form of the Laplacian
 
1 ∂ √ ij ∂ f
f = √ gg . (5.237)
g ∂xi ∂x j

We shall finally look at the relationship between the exterior derivative and the
gradient, divergence and curl. The gradient is defined as the vector corresponding to
the 1-form dφ. Hence we get the covariant expression for the gradient as decomposed
in a coordinate basis

 f = g i j ∂ f ei .
∇ (5.238)
∂x j
Considering an orthogonal coordinate basis, the physical components of the gra-
√
dient are obtained by transforming to an orthonormal basis according to ei = gii eî .
Hence the physical components of the gradient are given by

1 ∂f
∇f = √ e . (5.239)
gii ∂ x i î

The divergence of a vector is given by

|g|div A = dA, (5.240)

160 5 Covariant Differentiation

where the components of the 1-form A are found by lowering the components of the
 From Eqs. (3.116) and (3.118) we have
vector A.

1 
A = |g|εν·μ1 ...μn−1 Aν dx ν1 ∧ dx μ1 ∧ · · · ∧ dx μn−1 . (5.241)
(n − 1)!

The exterior derivative of this form is

√ 
1 ∂ |g|Aν
dA = ενμ ...μ dx α ∧ dx μ1 ∧ · · · ∧ dx μn−1 . (5.242)
(n − 1)! 1 n− p ∂xα

To avoid multicounting we use the notation ε|ν1 ·μ1 ...μn−1 | meaning that the sum-
mation over the indices is performed under the condition ν < μ1 < · · · < μn−1 .
Then
√ 
∂ |g|Aν
dA = ε|ν·μ1 ...μn−1 | dx α ∧ dx μ1 ∧ · · · ∧ dx μn−1 . (5.243)
∂xα
The dual of this form is
√ 
 ∂ |g|Aν
dA = |g| . (5.244)
∂xν
Inserting this into Eq. (5.240) gives the covariant expression for the divergence
of a vector
√ 
 1 ∂ |g|Aν
div A = √ . (5.245)
|g| ∂xν

We shall find the covariant expressions for the components of the curl of a vector
A in the special case of a coordinate system with an orthogonal vector basis. The
corresponding 1-form is A. The covariant expressions of the components of the curl
 × A is obtained from
∇

 × A = (d A)# .
∇ (5.246)

This involves the following operations: The vector is originally given in an

orthonormal basis associated with a chosen coordinate system. We call the cor-
responding vector components for the physical components of the vector. One wants
to express the physical components of ∇  × A in an arbitrary orthogonal coordinate
system in terms of the physical components of the vector and the components of
the metric tensor in the coordinate system. Since the expression (5.246) involves the
exterior derivative of the 1-form A, which is most easily performed in a coordinate
basis because then the exterior derivatives of the basis form vanish due to Poincare’s
lemma, one first transforms to the coordinate basis of the coordinate system. Then
the (contravariant) indices of the components of the vector A is lowered, and one
5.11 Covariant Expressions for Gradient, Divergence, Curl, … 161

obtains the (covariant) components of the 1-form A in the arbitrary coordinate basis
in terms of the physical components of the vector and the components of the metric
tensor (Eq. 5.250 below). Then one takes the exterior derivative of this form and
gets the 2-form d A as decomposed in the coordinate basis. One now transforms to
the form basis corresponding to the orthonormal vector basis. Further one takes the
dual of this form and gets a 1-form d A. The symbol # means that one converts this
1-form to a vector by raising the indices of the components of d A. This is automat-
ically taken account of when we decompose d A in a 1-form basis corresponding to
an orthonormal vector basis where the covariant and contravariant components are
equal. Hence we get the vector as decomposed in the orthonormal basis. In this way
we have found the physical components of ∇  × A. 
We shall demonstrate this procedure in the case of an arbitrary orthogonal vector
basis, so that the metric tensor is diagonal, i.e. the only non-vanishing components
of the metric tensor are those with equal indices.
In a coordinate system with orthogonal basis vectors, the line element is

dl 2 = gii dx i dx i . (5.247)

The coordinate basis vectors are ei = ∂/∂ x i , and the corresponding coordinate
basis forms are dx i . Since the metric is diagonal, the transformation of the basis
vectors to a coordinate basis has the form eî = (gii )−1/2 ei . The corresponding basis
√
forms dual to the orthonormal vector basis are ωî = gii dx i . In this space we
have a vector A = Aî eî , î = 1, 2, 3 as decomposed in an orthonormal basis. As
decomposed in the coordinate basis the vector components are

Ai = (gii )−1/2 Aî . (5.249)

So that

A = Ai ei = (gii )−1/2 Aî ei . (5.249)

The corresponding 1-form is

√
A = gii Ai dx i = gii Aî dx i . (5.250)

The exterior derivative of this form is

1 √  √ 
dA = gii Aî , j − g j j A ĵ ,i dx j ∧ dx i
 2
1 1 √  √ 
= √ gii Aî , j − g j j A ĵ ,i ω ĵ ∧ ωî . (5.251)
2 gii g j j

The dual of this form is the 1-form

162 5 Covariant Differentiation

1 √  √ 
d A = √ gii Aî , j − g j j A ĵ ,i ωk̂ , k̂ = î, ĵ. (5.252)
gii g j j

Thus the curl is

1 √  √ 
∇ × A = √ gii Aî , j − g j j A ĵ ,i ek̂ . (5.253)
gii g j j

Example 5.11.1 (Differential operators in spherical coordinates) The components

of the metric tensor in spherical coordinates in Euclidean 3-space are given in
Eq. (3.122), grr = 1, gθθ = r 2 , gφφ = r 2 sin2 θ . Inserting this into Eq. (5.239)
we get the expression for the gradient in spherical coordinates

∂f 1 ∂f 1 ∂f
∇f = er̂ + eθ̂ + e . (5.254)
∂r r ∂θ r sin θ ∂φ φ̂

The curl is
! "
1 ∂   ∂ Aθ̂
∇ × A = φ̂
sin θ A − er̂
r sin θ ∂θ ∂φ
! "
1 ∂ Ar̂ ∂  φ̂ 
+ − sin θ rA eθ̂
r sin θ ∂φ ∂r
! "
1 ∂  θ̂  ∂ Ar̂
+ rA − eφ̂ . (5.255)
r ∂r ∂θ

The divergence is
 
 
1 ∂ r 2 Ar̂ 1 ∂ sin θ Aθ̂ 1 ∂ Aφ̂
∇ · A = 2 + + . (5.256)
r ∂r r sin θ ∂θ r sin θ ∂φ

The Laplacian is
   
1 ∂ 2∂f 1 ∂ ∂f 1 ∂2 f
∇ f = 2
2
r + 2 sin θ + . (5.257)
r ∂r ∂r r sin θ ∂θ ∂θ r sin2 θ ∂φ 2

The d’Alembertian wave operator is

   
1 ∂ 2∂f 1 ∂ ∂f 1 ∂2 f 1 ∂2 f
f = 2 r + 2 sin θ + − .
r ∂r ∂r r sin θ ∂θ ∂θ r sin2 θ ∂φ 2 c2 ∂t 2
(5.258)
5.12 Electromagnetism in Form Language 163

5.12 Electromagnetism in Form Language

We shall consider electromagnetism in flat spacetime decomposing the vectors and

forms in a coordinate basis. In this section it will be shown that the simplest possible
form equations for electromagnetism in four-dimensional spacetime leads by means
of Poincare’s lemma to Maxwell’s equations.
Let us introduce an electromagnetic potential form

A = Aν dx ν = −φcdt + c Ai dx i . (5.259)

Hence

A0 = −A0 = φ, (5.260)

where φ is the electric scalar potential. The spatial component of A is the

electromagnetic vector potential.
The exterior derivative of A is

F = d A, (5.261)

and is called the electromagnetic field form. In coordinate basis it is written in

component form as
 
1 1 ∂ Aν ∂ Aμ
F= Fμν dx μ ∧ dx ν = dA = − dx μ ∧ dx ν . (5.262)
2 2 ∂xμ ∂xν

Hence, the components of the electromagnetic field form are

∂ Aν ∂ Aμ
Fμν = μ
− . (5.263)
∂x ∂xν

The electric field strength E and the magnetic flux density B are given by

1
F = E i dx i ∧ cdt + cBi j dx i ∧ dx j , B|i j| = Bk , k = i, j. (5.264)
2
Here |i j| means that i < j. The components of the electromagnetic field form is
given in terms of the electric field strength and the magnetic flux density by
⎛ ⎞
0 −E x −E y −E z
⎜ Ex 0 cBz −cB y ⎟
Fμν =⎜
⎝ E y −cBz 0
⎟. (5.265)
cBx ⎠
E z cB y −cBx 0

The electromagnetic field form can now be written as

164 5 Covariant Differentiation
 
1 ∂ Aj ∂ Ai
F = Fi t dx i ∧ dt + c − dx i ∧ dx j
2 ∂xi ∂x j
   
∂ At ∂ Ai 1 ∂ Aj ∂ Ai
=c − dx i
∧ dt + c − dx i ∧ dx j
∂xi ∂t 2 ∂xi ∂x j
   
∂ Ai 1 ∂ Aj ∂ Ai
= −c (∇φ)i − dx i ∧ dt + c − dx i ∧ dx j . (5.266)
∂t 2 ∂xi ∂x j

It follows from Eqs. (5.264) and (5.266) that the electric field strength and the
magnetic flux density is given in terms of the electric scalar potential and the
electromagnetic vector potential by

 − ∂ Ā ,
E = −∇φ B = ∇ 
 × A. (5.267)
∂t
Consider a transformation of the form

A = A + cd, (5.268)

where  is a scalar function. Then, using Poincare’s lemma, we get

F  = d A = d A + cd d = d A = F. (5.269)

Hence the field form F, and thereby the electromagnetic field strengths E and
B are invariant against a transformation of the form (5.268). This transformation is
called a gauge transformation. It has the component form

∂ ∂
−φ  cdt + c Ai dx i = −φcdt + c Ai dx i + c dt + c i dx i . (5.270)
∂t ∂x
Hence the gauge transformation takes the form

∂
φ = φ − , A = A + ∇φ.
 (5.271)
∂t
As a consequence of Poincare’s lemma the electromagnetic field is invariant
against this transformation. This freedom of gauge has been utilized to introduce
a condition upon the potentials as follows. It follows from Eq. (5.268) that

δ A = δ A − cδ d f = δ A − c  f, (5.272)

where  is the d’Alembertian wave operator in Cartesian coordinates, i.e.

1 ∂2 ∂2 ∂2 ∂2
=− + + + , (5.273)
c2 ∂t 2 ∂x2 ∂ y2 ∂z 2
5.12 Electromagnetism in Form Language 165

and f is a scalar gauge function. Introducing a gauge function which satisfies the
Lorenz condition

c  f = δ A , (5.274)

Equation (5.272) reduces to

δ A = 0. (5.275)

This is called the Lorenz gauge (Lorenz is a Danish physicist who died in 1892).
In this gauge the 4-divergence of A vanishes in vacuum. This corresponds to the
equation

 · A + 1 ∂φ = 0.
∇ (5.276)
c2 ∂t
It follows from Poincare’s lemma that

d F = d d A = 0. (5.277)

The component form of the field form is given in Eq. (5.264). Writing it out gives

F = E x dx ∧ cdt + E y dy ∧ cdt + E z dz ∧ cdt + cBz dx ∧ dy

+ cB y dz ∧ dx + cBx dy ∧ dz. (5.278)

The exterior derivative of F is the 3-form

∂ F|i j| k
dF = dx ∧ dx i ∧ dx j
∂x k 
∂ Bx ∂ By ∂ Bz
=c + + dx ∧ dy ∧ dz
∂x ∂y ∂z
 
∂ By ∂ Ex ∂ Ez
=c + − dt ∧ dy ∧ dz
∂t ∂z ∂x
 
∂ Bx ∂ Ez ∂ Ey
=c + − dt ∧ dz ∧ dx
∂t ∂y ∂z
 
∂ Bz ∂ Ey ∂ Ex
=c + − dt ∧ dx ∧ dy
∂t ∂x ∂y
 
  ∂ B
 
= c ∇ · B dx ∧ dy ∧ dz + c  
+ ∇ × E dt ∧ dx i ∧ dy j , k = i, j.
∂t
k
(5.279)

This shows that the form equation

166 5 Covariant Differentiation

dF = 0 (5.280)

corresponds to Maxwell’s source free equations

 · B = 0
∇ (5.281)

and

∂ B
+ ∇ × E = 0. (5.282)
∂t
The electromagnetic current form J is the 1-form

J = −ρcdt + (1/c) ji dx i . (5.283)

where ρ is the charge density and ji the i-component of the current density. The
source of the electromagnetic field form is the current form

1
δF =− J, (5.284)
ε0

where δ is the codifferential defined in Eq. (5.227), and ε0 is the permittivity of empty
space. Writing out this equation gives
   
−δ F = dF = ∂ ν Fνλ dx λ = ∂ 0 F00 + ∂ i Fi0 cdt + ∂ 0 F0 j + ∂ i Fi j dx j .
(5.285)

Using the component form (5.278) of F we obtain

 
  1 ∂ E
−δ F = ∇ · E cdt +  × B
− c∇ dx j . (5.286)
c ∂t
j

Equations (5.283), (5.284) and (5.285) give

 · E = ρ
∇ (5.287)
ε0

and


 × B − 1 ∂ E = μ0 j.
∇ (5.288)
c2 ∂t

where μ0 = 1/ε0 c2 is the permeability of empty space. Equations (5.287) and

(5.288) are the Maxwell source equations. Hence the form equation (5.284) contains
5.12 Electromagnetism in Form Language 167

the Maxwell source equations. The displacement current ∂ E/∂t is automatically

included in the formulation of Maxwell’s equation in the form language.
Let α be a p-form. In Minkowski spacetime

α = −(−1) p(4− p) α. (5.289)

Since dF is a 3-form we have

dF = −(−1)3 dF = dF. (5.290)

Hence, taking the dual of Eq. (5.284) gives

J = −ε0 dF. (5.291)

Taking the exterior derivative we get

dJ = −ε0 d dF = 0 (5.292)

due to Poincare’s lemma. Taking the dual of this equation we get

δ J = 0. (5.293)

The divergence of the current form vanishes. This is the mathematical expression
of the conservation of charge in the form language. Using Eq. (5.283) the vector
version of this equation is

∂ρ  · j = 0,
+∇ (5.294)
∂t
which is the equation of continuity in the electromagnetism.
It follows from Eqs. (5.261) and (5.284) that

1
δdA = − J. (5.295)
ε0

Applying the Laplace–Betrami operator defined in Eq. (5.231) to A gives

 A = δ d A + d δ A. (5.296)

It follows that
1
A=− J + d δ A. (5.297)
ε0
168 5 Covariant Differentiation

Asuming that A fulfills the Lorenz gauge condition (5.275) the last term in this
equation vanishes, so that

1
A=− J. (5.298)
ε0

This is the equation for electromagnetic waves. The time component of this
equation is
ρ
φ = − , (5.299)
ε0

and the space component is

 A = −μ0 j. (5.300)

The equation of motion and the energy equation of a charged particle in an

electromagnetic field is combined in the following equation
q
M= u ).
F( (5.301)
c

Here F( u ) is the contraction as defined in Eq. (3.46) of the field form F with the
4-velocity u of the charge, and M is the Minkowski force form,

dE d pi i
M = −γ dt + γ dx , (5.302)
dt dt

where E and p are the energy and momentum of the particle, and u = γ (c, v).
Writing out Eq. (5.301) gives

dE d pi i  
−γ dt + γ dx = −E x u x − E y u y − E z u z cdt
dt dt  
+ E x u t cBz u y − cB y u z dx
 
+ E x u t − cBz u x cBx u z dy
 
+ E z u t cB y u x − cBx u y dz
   
= −γ E · v cdt + γ c E + v × B dx i . (5.303)
i

The time component of this equation is the equation representing energy

conservation
dE
= q E · v, (5.304)
dt
5.12 Electromagnetism in Form Language 169

and the space component is the equation of motion of a charge in an electromagnetic

field

d p  
= q E + v × B . (5.305)
dt
The covariant form of the equation of motion is found from Eq. (5.301),

d pμ
= q Fνμ u ν . (5.306)
dτ
For a particle with constant rest mass m 0 , this gives

m 0 a μ = q Fνμ u ν , (5.307)

where a μ is the μ-component of the particle’s 4-acceleration.

Exercises

5.1 Dual forms

Let {eî } be a Cartesian basis in the three-dimensional Euclidean space. Using a

vector a = a î eî there are two ways of constructing a form:

(i) By constructing a 1-form from its covariant components a j = g ji a i :

A = aî dx î .

(ii) By constructing a 2-form from its dual components, defined by aî ĵ = εî ĵ k̂ a k̂ :

1
a= a dx î ∧ dx ĵ .
2 î ĵ
We write this form as a = A where  means to take the dual form.

(a) Given the vectors a = ex + 2e y − ez and b = 2ex − 3e y + ez .
Find the corresponding 1-forms A and B, and the dual 2-forms a = A and
b = B, and also the dual form θ to the 1-form σ = dx − 2dy.
(b) Take the exterior product A ∧ B and show that

θî ĵ = εî ĵ k̂ C k̂ ,

where θ = A ∧ B and C = a × b.

170 5 Covariant Differentiation

(c) Show that the exterior product A ∧ B is given by the 3-form

A ∧ B = ( 
a · b)dx ∧ dy ∧ dz.

(d) Show that the exterior derivative of a 1-form, dA, corresponds to the curl ∇ × A
of the corresponding vector.
(e) Finally show that

d ∗ d f = ∇ 2 f dx ∧ dy ∧ dz

for a scalar field f.

5.2 Differential operators in spherical coordinates

We consider an Euclidean three-dimensional space with Cartesian coordinates

(x, y, z) and spherical coordinates (r, θ, φ). The transformation from the spherical
to the Cartesian coordinates is

x = r sin θ cos φ, y = r sin θ sin φ, z = r cos θ

(a) Find the components of the metric tensor and the form of the line-element in
spherical coordinates.
(b) Let f be a scalar field. The gradient of f is given by

∂f ∂f ∂f
∇f = er̂ + eθ̂ + e ,
∂r ∂θ ∂φ φ̂

where eî are the orthonormal basis vectors formed from the coordinate basis
vectors in the spherical coordinate system.
Find the expressions for the gradient of f in coordinate basis.
(c) In a coordinate system with orthogonal coordinate basis vectors, the curl of a
vector field is given by
   
1∂ A 3̂
∂ A 2
1 ∂ A 1̂
∂ A 3̂
∇ × A = √ − e1 + √ − e2
g22 g33 ∂ x 2 ∂x3 g11 g33 ∂ x 3 ∂x1
 
1 ∂ A2̂ ∂ A1̂
+√ − e3 .
g11 g22 ∂ x 1 ∂x2

Find an expression for the curl in spherical coordinates. (The division by the
√
factors gii g j j is a normalization of the area of a surface element normal to the
basis vector ek , k = i, j.)
(d) The divergence of a vector field can in general (in an arbitrary basis) be defined
by
Exercises 171
 
∇ · A ε = d ∗ A,

where

ε= |g|ω1 ∧ ω2 ∧ ω3 ,

Is the volume form, and |g| is the determinant of the matrix formed by
the components of the metric tensor. The volume form represents an invariant
volume element.
Find the expression for the divergence of A in spherical coordinate.
Finally find the expression for the Laplacian of f in the spherical coordinate
system.

5.3 Spatial geodesics in a rotating frame of reference

Our point of departure is the line-element (4.20) for 3-space in a rotating reference
frame,

r 2 dθ 2
d2 = dr 2 + + dz 2 .
1 − r 2 ω2 /c2

We shall consider geodesics in the two-dimensional surface with z = constant.

The task is to calculate the shortest curve between two points with the same dis-
tance from the axis using the Lagrangian equations with the Lagrange function
L = (1/2)˙2 , where the dot denotes differentiation with respect to an invariant
parameter representing the arc length along the curve.

(a) Find the form of the 2-vector identity for the tangent vectors of the curve.
(b) Find an expression for the momentum pθ conjugate to the cyclic coordinate θ
of L.
(c) Find the differential equation for the geodesic curves.
(d) Use the boundary condition that the point on the curve closest to the axis has a
distance r0 from the axis, to show that
r0
pθ =  .
1 − r02 ω2 /c2

(e) Show that the differential equation of the curve can be written as

dr ω2 r dr dθ
 − 2 = .
r r 2 − r02 c r 2 − r02 r0

Integrate this equation and find the equation of the curve. Finally draw the
curve.
172 5 Covariant Differentiation

5.4 Christoffel symbols in a uniformly accelerated reference frame

(a) Use the coordinate transformation (4.80)–(4.82) and the formula (5.27) to calcu-
late the non-vanishing Christoffel symbols in the coordinate system of Chap. 4
co-moving with a uniformly accelerated reference frame.
(b) Use Eq. (5.65) to calculate the same Christoffel symbols as in (a).

5.5 Relativistic vertical projectile motion

A particle is thrown vertically upwards with velocity v from the origin of the
coordinate system in the gravitational field of a uniformly accelerated reference
frame.
Calculate the maximal height of the particle.

5.6 The geodesic equation and constants of motion

(a) Show that the geodesic equation can be written in the following form: dudsα −
1 ∂gβγ β γ
2 ∂xα
u u = 0.
(b) Assume that the metric is static and the space is cylindrically symmetric with
cylindrical coordinates (r, θ, z). What constants of motion are there then for a
free particle?
Chapter 6
Curvature

Abstract The Riemann curvature tensor is introduced, and the expression of its
components in terms of the derivatives of the metric and the structure coefficients is
deduced. Tidal forces are discussed in a relativistic context, and it is pointed out that
the relativistic gravitational field has both a non-tidal component due to the motion
of the reference frame and a tidal component due to spacetime curvature.

6.1 The Riemann Curvature Tensor

 along a vector v was defined

The covariant directional derivative of a vector field A
and interpreted geometrically in Sect. 5.2 as follows:

  
 = dA = Aμ vν eμ = lim AQP (λ + λ) − A(λ) .
∇υ A (6.1)
;ν
dλ λ→0 λ

 QP be the vector A
Let A  parallel transported from Q to P in Fig. 6.1.
In Fig. 6.2 we have illustrated that a vector parallel transported around a closed
curve on a curved surface changes direction during the round trip.
Then to first order in λ we have A  P λ and
 P + (∇v A)
 QP = A

A  Q λ.
 Q − (∇v A)
 PQ = A (6.2)

To second order in λ we have

 
 PQ = 1 − ∇υ λ + 1 ∇υ ∇υ (λ)2 A
A Q. (6.3)
2

We shall now parallel transport a vector around the closed polygon shown in
Fig. 6.3.
 PQ is parallel transported further onto R we get
If A
  
 PQR = 1 − ∇u λ + 1 ∇u ∇u (λ)2 1 − ∇υ λ + 1 ∇υ ∇υ (λ)2 A
A  R , (6.4)
2 2
© Springer Nature Switzerland AG 2020 173
Ø. Grøn, Introduction to Einstein’s Theory of Relativity,
Undergraduate Texts in Physics, https://doi.org/10.1007/978-3-030-43862-3_6
174 6 Curvature


Fig. 6.1 Parallel transport of A

Fig. 6.2 Parallel transport of a vector around a triangle

where A  Q is replaced by A R because the differential operator always shall be applied

 around the whole polygon
to the vector in the first position. If we parallel transport A
in Fig. 6.3 we get
  
 PQRSTP = 1 + ∇u λ + 1 ∇u ∇u (λ)2 1 + ∇υ λ + 1 ∇υ ∇υ (λ)2
A
2 2
 
  1
· 1 − ∇[u,υ]
 (λ)
2
1 − ∇u λ + ∇u ∇u (λ)2
2
 
1 P.
1 − ∇υ λ + ∇υ ∇υ (λ)2 A (6.5)
2
6.1 The Riemann Curvature Tensor 175

Fig. 6.3 Curvature and parallel transport

Calculating to second order in λ gives

 
A  P + [∇u , ∇υ ] − ∇[u,υ]
 PQRSTP = A 2
 (λ) AP . (6.6)

There is a variation of the vector under parallel transport around the closed
polygon,
 
=A
δA  P = [∇u , ∇υ ] − ∇[u,υ]
 PQRSTP − A 
 AP (λ) .
2
(6.7)

Definition 6.1.1 (The Riemann curvature tensor) The Riemann’s curvature tensor
is defined as
 
R( , A, 
 u , v) ≡ [∇u , ∇v ] − ∇[u,v] (A). (6.8)

The components of the Riemann curvature tensor are defined by applying the
tensor on basis vectors
μ  
Rναβ eμ ≡ [∇α , ∇β ] − ∇[eα ,eβ ] (eν ). (6.9)

It follows from Definition (6.8) that the Riemann tensor is antisymmetric in its last
two arguments. Hence the components are antisymmetric in their last two indices,

Rμνβα = −Rμναβ . (6.10)

 under parallel transport around the polygon,

The expression for the change of A
Eq. (6.7), can now be written as
176 6 Curvature

Fig. 6.4 Curl as area of a

vectorial parallelogram.
Parallelogram defined by the
vectors u λ and v λ is
represented mathematically
by the vector
S = u × v(λ)2

 = R( , A,
A  u , v)(λ)2
= R( , Aν eν , uα eα , vβ eβ )(λ)2
μ
= eμ Rναβ Aν uα vβ · (λ)2
1
= eμ Rμναβ Aν (uα vβ − uβ vα )(λ)2 . (6.11)
2

The area of the parallelogram defined by the vectors u λ and v λ is

S = u × v (λ)2 . (6.12)

These vectors are shown in Fig. 6.4.

Using that

(u × v)αβ = uα vβ − uβ vα , (6.13)

we can write Eq. (6.11) as

= 1 ν μ
A A Rναβ S αβ eμ . (6.14)
2
The components of the Riemann tensor expressed by the connection coefficients
(5.114) and structure coefficients (3.40) are given by
μ
eμ Rναβ = [∇α , ∇β ]eν − ∇[eα ,eβ ] eν
ρ
= (∇α ∇β − ∇β ∇α − cαβ ∇ρ )eν
ρ
= ∇α ∇β eν − ∇β ∇α eν − cαβ ∇ρ eν
μ μ ρ μ
= ∇α
νβ eμ − ∇β
να eμ − cαβ
νρ eμ
μ μ μ μ ρ
μ
= (∇α
νβ )eμ +
νβ ∇α eμ − (∇β
να )eμ −
να ∇β eμ − cαβ
νρ eμ
μ ρμ μ ρ μ ρ μ
= eα (
νβ )eμ +
νβ
ρα eμ − eβ (
να )eμ −
να
ρβ eμ − cαβ
νρ eμ . (6.15)

This gives the components of the Riemann curvature tensor in an arbitrary basis
6.1 The Riemann Curvature Tensor 177

μ μ μ ρ
μ ρ μ μ ρ
Rναβ = eα (
νβ ) − eβ (
να ) +
νβ
ρα −
να
ρβ − cαβ
νρ . (6.16)

In coordinate basis Eq. (6.16) is reduced to

μ μ μ ρμ ρ μ
Rναβ =
νβ,α −
να,β +
νβ
ρα −
να
ρβ , (6.17)

μ μ
where
νβ =
βν are the Christoffel symbols.
Since the basis vectors are derivative operators, the first two terms is a linear com-
bination of derivatives of the connection coefficients. In a local Cartesian coordinate
system co-moving with a local inertial reference frame all the connection coeffi-
cients vanish, and only the first two terms in the expression of the components of
the Riemann curvature tensor remain. As we have seen in Sect. 5.5.3 this means that
in such a system there is no acceleration of gravity. But in general the derivatives of
the connections coefficients will not vanish. Hence in general spacetime is curved.
This shows that the acceleration of gravity does not depend upon the curvature of
spacetime. It depends instead upon the motion of the reference frame. The curvature
of spacetime is given by a tensor and is an invariant property of spacetime at the con-
sidered position. The acceleration of gravity is, however, not an invariant property
of spacetime since it is given by certain connection coefficients that are not tensor
components. They can be transformed away. This is the mathematical expression of
the fact that you can transform away the acceleration of gravity locally by going into
a local inertial frame.

Definition 6.5.1 (Contraction of a tensor component) Contraction of a tensor

component is an operation defined by
μ
Rνβ ≡ Rνμβ . (6.18)

The summation is over μ. In this way a new tensor is constructed from another
tensor with a rank 2 lower than the original tensor.
The tensor with components Rνβ is called the Ricci curvature tensor. Another
contraction gives the Ricci curvature scalar R = Rμμ .
Due to the antisymmetry (6.10) we can define a matrix of curvature forms,

1 μ α
Rμν = R ω ∧ ωβ . (6.19)
2 ναβ
Inserting the components of the Riemann tensor from Eq. (6.14) gives
 
μ ρ 1 ρ μ
Rμν = eα (
νβ ) +
νβ
ρα − cαβ
νρ ωα ∧ ωβ .
μ
(6.20)
2

We shall now introduce the curvature form. Then we need the connection forms,

μν =
να
μ α
ω . (6.21)
178 6 Curvature

The exterior derivatives of the basis forms are,

1 ρ
dωρ = − cαβ ωα ∧ ωβ = −ρα ∧ ωα (6.22)
2
The exterior derivatives of the connection forms can now be written
μ
dμν = d
νβ ∧ ωβ +
νρ μ
dωρ
μ 1 ρ μ α
= eα (
νβ )ωα ∧ ωβ − cαβ
νρ ω ∧ ωβ . (6.23)
2
Combining Eqs. (6.20)–(6.22) the curvature forms take the form
μ
Rμν = dμν + λ ∧ λν . (6.24)

This is Cartans 2. structure equation.

Example 6.1.1 The Riemann curvature tensor of a spherical surface calculated from
Cartan’s structure equations
Let r = R be the radius of the spherical surface. The calculation is performed in
5 steps.
1. Write down the metric tensor and introduce a form basis dual to an orthonormal
vector basis.

g = ωθ̂ ⊗ ωθ̂ + ωϕ̂ ⊗ ωϕ̂ = R2 dθ ⊗ dθ + R2 sin2 θ dϕ ⊗ dϕ,

giving

ωθ̂ = Rdθ, ωφ̂ = R sin θ dϕ.

2. Use Cartan’s 1. structure equation, dωμ = ων ∧ μν , to calculate the structure
forms. Since R is constant and using Poincare’s lemma and the antisymmetry of
the connection forms, exterior differentiation of the basis forms lead to

1 cos θ ϕ̂ ϕ̂
dωθ̂ = 0 = ωϕ̂ ∧ θ̂ϕ̂ , dωϕ̂ = R cos θ dθ ∧ dϕ = ωθ̂ ∧ ω = ωθ̂ ∧ θ̂
R sin θ
giving

ϕ̂ 1 cos θ ϕ̂
θ̂ϕ̂ = f (θ, ϕ)ωϕ̂ , θ̂ = g(θ, ϕ)ωθ̂ + ω
R sin θ

where the functions f (θ, ϕ) and g(θ, ϕ) are determined from the antisymmetry of
ϕ̂
the connection forms. Using that θ̂ϕ̂ = θ̂ ϕ̂ = −ϕ̂ θ̂ = −θ̂ , we get
6.1 The Riemann Curvature Tensor 179

1 cos θ
f (θ, ϕ) = , g(θ, ϕ) = 0.
R sin θ
Hence,

ϕ̂ 1 cos θ ϕ̂
θ̂ = −θ̂ϕ̂ = ω = cos θ dϕ.
R sin θ
The reason for going back to coordinate basis here is that then it is easier to
calculate the exterior derivative dϕ̂θ̂ .
3. Calculate the Riemann curvature forms from Cartan’s 2. structure equation,

ϕ̂   1 ϕ̂
Rθ̂ϕ̂ = dθ̂ϕ̂ + θ̂ϕ̂ ∧ ϕ̂ = dθ̂ϕ̂ = d cos θ dϕ = − sin θ dθ ∧ dϕ = − 2 ωθ̂ ∧ ωϕ̂ = −R .
R θ̂

4. Calculate the non-vanishing components of the Riemann tensor from

Rμ̂ν̂ = (1/2)Rμναβ ωα ∧ ωβ .

This gives

ϕ̂ 1 ϕ̂
Rθ̂ϕ̂ θ̂ ϕ̂ = Rθ̂ ϕ̂ θ̂ = −Rθ̂ϕ̂ ϕ̂ θ̂ = −Rθ̂ θ̂ ϕ̂ =
.
R2
5. Calculate the components of the Ricci curvature tensor and the Ricci curvature
scalar,

1 ϕ̂ 2
Rθ̂ θ̂ = Rϕ̂ ϕ̂ = 2
, R = Rθ̂θ̂ + Rφ̂ = 2 .
R R

6.2 Differential Geometry of Surfaces

Imagine an arbitrary surface embedded in an Euclidean 3-dimensional space

(Fig. 6.5).
The coordinate basis vectors on the surface are
∂ ∂
eu = , ev = , (6.25)
∂u ∂v
where u and v are coordinates on the surface. The directional derivatives of the basis
vectors are written as
α
eμ,ν =
μν eα + Kμν N , α = 1, 2. (6.26)
180 6 Curvature

Fig. 6.5 Surface with tangent vectors and normal vector

Greek indices run through the surface coordinates, and N is a unit vector
orthogonal to the surface. It follows from Eq. (6.26) that

Kμν = eμ,ν · N . (6.27)

Equation (6.26) is called Gauss’ equation. In coordinate basis we have

eμ,ν = ∂xμ∂ ∂xν = ∂xν∂∂xμ = eν,μ . Hence
2 2

Kμν = Kνμ . (6.28)

Let u be the unit tangent vector to a curve on the surface, parameterized by λ .

Differentiating u along the curve, we get

du μ ν
= u;ν u eμ + Kμν uμ uν N . (6.29)
dλ  
2nd fundamental form

We define two scalar quantities κg and κN by

du
= κg e + κN N . (6.30)
dλ
6.2 Differential Geometry of Surfaces 181

Here e = N ×u is a unit vector in the surface which is orthogonal both to N and to u .
κg is called the geodesic curvature, and κN the normal curvature (external curvature).
Note that κg = 0 for geodesic curves on the surface. The geodesic curvature is given
by the covariant directional derivative of the tangent vector field of a curve along the
curve,
μ
κg e = u;ν uν eμ = ∇u u . (6.31)

It follows from Eqs. (6.29) that

du 
· N = Kμν uμ uν , (6.32)
dλ
and from Eq. (6.30) that

du 
κN = · N. (6.33)
dλ

We also have that u · N = 0 along the whole curve. Differentiation gives

du  dN
· N + u · = 0. (6.34)
dλ dλ
It follows from the last three equations that

dN
κN = Kμν uμ uν = −u · , (6.35)
dλ
which is called Weingarten’s equation.
The geodesic curvature κg and normal curvature κN together give a complete
description of the geometry of a surface in a flat 3-dimensional space. We are now
going to consider geodesic curves through a point on the surface. The point of depar-
ture is the tangent vector u = uμ eμ with u · u = gμν uμ uν = 1 . The directions with
maximum and minimum values for the normal curvatures are found by extremalizing
κN under the condition gμν uμ uν = 1. We then solve the variation problem δF = 0
for arbitrary uμ where F = Kμν uμ uν − k(gμν uμ uν − 1). Here k is the Lagrange
multiplicator. Variation with respect to uμ gives

δF = 2(Kμν − kgμν )uν δuμ . (6.36)

The requirement that δF = 0 for arbitrary uμ demands that

(Kμν − kgμν )uν = 0. (6.37)

For this system of equations to have nonzero solutions, we must have

182 6 Curvature

det(Kμν − kgμν ) = 0, (6.38)

or

K11 − kg11 K12 − kg12

K21 − kg21 K22 − kg22
= 0. (6.39)

This gives the following quadratic equation for k:

k 2 det(gμν ) − (g11 K22 − 2g12 K12 + g22 K11 )k + det(Kμν ) = 0

(K symmetric K12 = K21 ). (6.40)

The equation has two solutions, k1 and k2 . These are the extremal values of k.
In order to find the meaning of k, we multiply Eq. (6.37) by uμ , which gives

0 = (Kμν − kgμν )uμ uν

= Kμν uμ uν − kgμν uμ uν
= κN − k ⇒ k = κN . (6.41)

The extremal values of κN are called the principal curvatures of the surface. Let
the directions of the geodesics with extreme normal curvature be given by the tangent
vectors u and v. Equation (6.37) gives

Kμν uν = kgμν uν . (6.42)

Multiplying by vμ we get

Kμν uν vμ = k1 gμν uν vμ = k1 uμ vμ = k1 (u · v),

Kμν vν uμ = k2 gμν vν uμ = k2 (u · v), (6.43)

which gives

(k1 − k2 )(u · v) = Kμν (uν vμ − vν uμ ) = 2Kμν u[ν vμ] . (6.44)

Kμν is symmetric in μ and ν. So we get (k1 − k2 )(u · v) = 0 . For k1 = k2 we have to

demand u · v = 0. So the geodesics with extremal normal curvature are orthogonal
to each other.
6.2 Differential Geometry of Surfaces 183

The Gaussian curvature (at a point) is defined as

K = κN 1 · κN 2 (6.45)

Since κN 1 and κN 2 are solutions of the quadratic equation above, we get

det(Kμν )
K= . (6.46)
det(gμν )

6.2.1 Surface Curvature Using the Cartan Formalism

When we use the Cartan formalism, we introduce an orthonormal set of basis vectors
at each point of the surface. Greek indices run through the surface coordinates (2-
dimensional) and Latin indices through the space coordinates (3-dimensional):

eâ = (e1̂ , e2̂ , N ), eμ̂ = {e1̂ , e2̂ }. (6.47)

where N is a unit vector field orthogonal to the surface. Using the exterior derivative
and form formalism, we find how the unit vectors on the surface change:

deν̂ = eâ ⊗ âν̂ = eα̂ ⊗ α̂ν̂ + N ⊗ 3̂ν̂ , (6.48)

μ̂ μ̂
where ν̂ =
ν̂ α̂ ωα̂ are the connection forms on the surface, i.e. the intrinsic
connection forms. The extrinsic connection forms are
μ̂ μ̂
3̂ν̂ = Kν̂ α̂ ωα̂ , 3̂ = Kα̂ ωα̂ . (6.49)

We let the surface be embedded in an Euclidean (flat) 3-dimensional space. This

means that the curvature forms of the 3-dimensional space are zero:

Râ3̂b̂ = 0 = dâb̂ + âk̂ ∧ k̂b̂ , (6.50)

which gives

Rμ̂3̂ν̂ = 0 = dμ̂ν̂ + μ̂α̂ ∧ α̂ν̂ + μ̂3̂ ∧ 3̂ν̂ = Rμ̂ν̂ + μ̂3̂ ∧ 3̂ν̂ , (6.51)

μ̂
where Rν̂ are the curvature forms of the surface. We then have

1 μ̂ μ̂
R ωα̂ ∧ ωβ̂ = −3̂ ∧ 3̂ν̂ . (6.52)
2 ν̂ α̂β̂
184 6 Curvature

Inserting the components of the extrinsic connection forms, we get (using the
μ
antisymmetry of α and β in Rναβ )

μ̂ μ̂ μ̂
Rν̂ α̂β̂ = Kα̂ Kν̂ β̂ − Kβ̂ Kν̂ α̂ . (6.53)

We now lower the first index:

Rμ̂ν̂ α̂β̂ = Kμ̂α̂ Kν̂ β̂ − Kμ̂β̂ Kν̂ α̂ . (6.54)

Rμ̂ν̂ α̂β̂ are the components of a curvature tensor which only refer to the dimensions
of the surface. In particular

R1̂2̂1̂2̂ = K1̂1̂ K2̂2̂ − K1̂2̂ K2̂1̂ = det K. (6.55)

We then have the following connection between this component of the Riemann
curvature tensor of the surface and the Gaussian curvature of the surface:
det Kμ̂ν̂ R
K = κN 1 κ N 2 = = 1̂2̂1̂2̂ = R1̂2̂1̂2̂ , (6.56)
det gμ̂ν̂ det gμ̂ν̂

where we have used that det gμ̂ν̂ = 1 in orthonormal basis. Since the right-hand side
refers to the intrinsic curvature and the metric on the surface, we have proved that
the Gaussian curvature of a surface is an intrinsic quantity. It can be measured by
observers on the surface without embedding the surface in a 3-dimensional space.
This is the contents of Gauss’ theorema egregium.

6.3 The Ricci Identity

Applying the Riemann tensor to a vector gives

−
→ μ
e μ Rναβ Aν = (∇α ∇β − ∇β ∇α − ∇[− 
→ e β ] )(A).
e α ,−
→ (6.57)

In coordinate basis this is reduced to

e μ Rναβ Aν = (A;βα − Aμ;αβ )−

−
→ μ μ →
e μ, (6.58)

where
μ μ
A;αβ ≡ (A;β );α . (6.59)

The Ricci identity on component form is

6.3 The Ricci Identity 185

μ μ μ
Aν Rναβ = A;βα − A;αβ . (6.60)

We can write this as

1 μ ν−
=
d2 A R A →e μ ⊗ ωα ∧ ωβ . (6.61)
2 ναβ
This shows that the 2. exterior derivative of a vector is equal to zero only in a flat
space. Equations (6.60) and (6.61) both represent the Ricci identity.

6.4 Bianchi’s 1. Identity

We shall here need Cartan’s 1. structure equation,

dωμ = −μν ∧ ων , (6.62)

and Cartan’s 2. structure equation,

Rμν = dμν + μλ ∧ λν . (6.63)

Exterior differentiation of Eq. (6.62) and use of Poincaré’s lemma (5.16) give
d2 ωμ = 0. Hence,
μ
0 = dμν ∧ ων − λ ∧ dωλ . (6.64)

Use of (6.62) gives

dμν ∧ ων + μλ ∧ λν ∧ ων = 0. (6.65)

From this we see that

μ
(dμν + λ ∧ λν ) ∧ ων = 0. (6.66)

We now get Bianchi’s 1. identity,

Rμν ∧ ων = 0 (6.67)

The component form of Bianchi’s 1. identity is

1 μ α
Rναβ ω ∧ ωβ ∧ων = 0. (6.68)
2
 
Rμν
186 6 Curvature

The component equation is

μ
R[ναβ] = 0 (6.69)

or
μ μ μ
Rναβ + Rαβν + Rβνα = 0, (6.70)

μ μ
where the antisymmetry Rναβ = −Rνβα has been used.

6.5 Bianchi’s 2. Identity

Exterior differentiation of Eq. (6.63) gives

μ ρ μ μ
dRμν = Rλ ∧ λν − μρ ∧ λ ∧ λν − λ ∧ Rλν + λ ∧ λρ ∧ ρν
= Rμλ ∧ λν − μλ ∧ Rλν . (6.71)

We now have Bianchi’s 2. identity as a form equation:

μ μ
d Rμν + λ ∧ Rλν − Rλ ∧ λν = 0 (6.72)

As a component equation Bianchi’s 2. identity is given by

μ
Rν[αβ;γ ] = 0. (6.73)

The Riemann curvature tensor has four symmetries.

I. The definition of the Riemann tensor implies that: Rμναβ = −Rμνβα .
μ
II. Bianchi’s 1. identity:R[ναβ] = 0.
III. From Cartan’s 2. structure equation follows:

Rμναβ = −Rνμαβ . (6.74)

By choosing a locally Cartesian coordinate system in an inertial frame we get the

following expression for the components of the Riemann curvature tensor:

1
Rμναβ = (gμβ,να − gμα,νβ + gνα,μβ − gνβ,μα ), (6.75)
2
from which we get:
IV. The fourth symmetry of the Riemann curvature tensor is Rμναβ = Rαβμν .
6.5 Bianchi’s 2. Identity 187

These four symmetries reduce the number of independent components of the

Riemann tensor in 4-dimensional spacetime from 256 to 20. Contraction of μ and α
leads to

Rνβ = Rβν , (6.76)

that is the Ricci tensor is symmetric. In 4-dimensional spacetime the Ricci tensor
has 10 independent components.

6.6 Torsion

Definition 6.6.1 (The torsion 2-form) The torsion 2-form is defined by

 
T (u ∧ v) = ∇u v − ∇v u − u , v , (6.77)

where u and v are arbitrary vectors. It is a vectorial form, which means that the
torsion form has vector components. The contraction of the torsion with u ∧ v has
the component form,
 ρ ρ ρ
 μ ν
T (u ∧ v) = −
μν −
νμ + cμν u v eρ . (6.78)

It follows that the torsion form has the component form

1 ρ ρ ρ

T=
νμ −
μν − cμν eρ ⊗ ωμ ∧ ων . (6.79)
2
ρ
Introducing the scalar torsion components Tμν by

ρ μ ν
T (u ∧ v) = Tμν u v eρ , (6.80)

we get
ρ ρ ρ ρ
Tμν =
νμ −
μν − cμν , (6.81)

So that
1 ρ
T= T eρ ⊗ ωμ ∧ ων .. (6.82)
2 μν
ρ
In coordinate basis cμν = 0, giving

ρ ρ ρ
Tμν =
νμ −
μν . (6.83)
188 6 Curvature

Hence torsion induces an antisymmetric part of the connection coefficients in a

coordinate basis.
The spacetime of the general theory of relativity is assumed to be torsion free.
This is called a Riemannian space. Then the connection coefficients are related to
the structure coefficients by
ρ ρ ρ
cμν =
νμ −
μν . (6.84)

This shows that the structure coefficients represent the antisymmetric parts of the
connection coefficients. In a coordinate basis in a Riemannian space the connection
coefficients are symmetric and the structure coefficients vanish.
It follows from Eqs. (5.164), (5.78) and (6.79) that the vectorial torsion form
(form with vector components) may be written
 
T = eρ ⊗ dωρ + μν ∧ ωρ . (6.85)

The torsion two forms, T ρ , with scalar components are defined by

T = eρ ⊗ T ρ , (6.86)

hence,

T ρ = dωρ + μν ∧ ωρ . (6.87)

6.7 The Equation of Geodesic Deviation

Consider two nearby geodesic curves (Fig. 6.6), both parametrized by a parameter
λ. Let s be a vector connecting two curves with the same value of λ. The connecting
vector s is said to measure the geodesic deviation of the curves
In order to deduce an equation describing how the geodesic deviation varies along
the curves, we consider the covariant directional derivative of s along the curves, ∇u s,
where u is the tangent vector field of the curves.  
Let u and s be coordinate basis vectors of a coordinate system. Then s, u = 0
so that

∇u s = ∇s u , (6.88)

giving

∇u ∇u s = ∇u ∇s u . (6.89)

6.7 The Equation of Geodesic Deviation 189

Fig. 6.6 Geodesic deviation.

Two neighbouring geodesic
curves with a vector s
connecting points on the
curves with the same
parameter value

Furthermore

R(u, s) u = [∇u , ∇s ] − ∇[u, s] u = [∇u , ∇s ]u. (6.90)

Thus

∇u ∇u s = ∇s ∇u u + R(u, s) u . (6.91)

Since the curves are geodesics ∇u u = 0, and R(u, s) = −R(s, u ) due to the
antisymmetry of the Riemann tensor, the equation reduces to

∇u ∇u s + R(s, u ) u = 0. (6.92)

This is the equation of geodesic deviation. The component form of the equation
is
 μ
D2 s
+ Rμανβ uα sν uβ = 0. (6.93)
dλ2

where D/dλ is the covariant derivative with respect to an invariant curve parameter
λ. In co-moving geodesic normal coordinates with u = (1, 0, 0, 0) and vanishing
Christoffel symbols, the covariant derivative reduces to the ordinary derivative, and
the equation for geodesic deviation reduces to
 i
d2 s
+ Ri0j0 sj = 0. (6.94)
dλ2
190 6 Curvature

6.8 Tidal Acceleration and Spacetime Curvature

In Chap. 1 we found Eq. (1.50) for the tidal acceleration, i.e. the relative acceleration
between two nearby particles,

d2 ζ k i ∂ φ
2
= −ζ , (6.95)
dt 2 ∂xi ∂xk

where ζ j is the j-component of the separation vector, and φ is the Newtonian grav-
itational potential. Comparing these equations we see that in the Newtonian limit
the non-vanishing components of the Riemann curvature tensor of spacetime are the
second derivatives of the Newtonian potential,

∂ 2φ
Ri0j0 = . (6.96)
∂xi ∂xj

In Newtonian physics the acceleration of gravity is given by

g = −∇φ, (6.97)

or in component form

∂φ
gi = − . (6.98).)
∂xi

Comparing with Eq. (5.90) we see that with a locally Cartesian coordinate system
the non-vanishing Christoffel symbols are

∂φ

00
i
= . (6.99)
∂xi

The Christoffel symbols are the first derivatives of the Newtonian gravitational
potential. According to Eq. (6.96) the second derivatives are the components of the
Riemann curvature tensor. Hence in the Newtonian approximation the non-vanishing
components of the curvature tensor are

∂
00
i
Ri0j0 = . (6.100)
∂xj
6.9 The Newtonian Tidal Tensor 191

6.9 The Newtonian Tidal Tensor

There are several definitions of the Newtonian tidal tensor that are mathematically
equivalent. One is as follows.

Definition 6.9.1 (Newtonian tidal tensor) The Newtonian tidal tensor is a symmet-
rical tensor of rank 2 with components

∂gi
Ei j = − , (6.101)
∂xj
i.e. Ei j is minus the change of the i-component of the acceleration of gravity due
to a displacement in the j-direction. Since

∂φ
gi = − , (6.102)
∂xi
the components of the Newtonian tidal tensor may be written

∂ 2φ
Ei j = . (6.103)
∂xi ∂xj
It follows that the Newtonian tidal tensor is symmetrical.
The Newtonian gravitational field equation

∂ 2φ
∇2φ = = 4π Gρ (6.104)
∂xi ∂xi

can now be written

Eii = 4π Gρ. (6.105)

Also it follows that the equation of tidal acceleration can be written

d2 ζ k
= −Eik ζ i , (6.106)
dt 2
and that in the Newtonian limit the tidal tensor is related to the Riemann curvature
tensor of spacetime by

Ri0j0 = Eji . (6.107)

192 6 Curvature

6.10 The Tidal and Non-tidal Components

of a Gravitational Field

In Newton’s theory of gravitation all gravitational fields are caused by masses. A grav-
itational field may be described mathematically in terms of a gravitational potential
φ. The acceleration of gravity can be expressed as the negative gradient of the gravita-
tional potential as in Eq. (6.97). The gravitational field is an acceleration field which
has a gravitational field strength equal to the acceleration of gravity at each point in
space. The gravitational field strength is a local quantity, while the gravitational field
itself is a global concept.
Einstein generalized Newton’s 1 law to the form: The 4-acceleration of a free
particle vanishes when no non-gravitational forces act upon it,

dU μ μ
Aμ = +
αβ U α U β = 0. (6.108)
dτ
This equation holds in arbitrary coordinate systems in all frames of reference,
inertial or accelerated. But only in coordinate systems in which the coefficients of
the metric tensor are constant, do the Christoffel symbols vanish, so that the equation
takes the Newtonian three-vector form a = 0. In an arbitrary reference frame the
term dU i /dτ represents the i-component of the acceleration of the particle relative
to the reference frame, and according to Eq. (6.108) it is given by

dU i
= −
αβ
i
U αU β. (6.109)
dτ
The acceleration of gravity may be defined in two mathematically identical ways,
either in terms of the 3-acceleration of a free particle instantaneously at rest, or in
terms of the 4-acceleration
dU μ μ
Aμ = +
αβ U α U β (6.110)
dτ
of a particle permanently at rest in a reference frame. Let the gravitational field point
in the i-direction at a point. Then the quantity
αβi
U α U β determines the acceleration
of gravity.
This has an important conceptual consequence: Experiencing a gravitational field
strength, i.e. that there is a non-vanishing acceleration of gravity, in other words that
a free particle falls, has nothing to do with the curvature of spacetime. It depends upon
certain Christoffel symbols. The fact that the Christoffel symbols are not tensor com-
ponents implies that all of them may be transformed away at a point by introducing
locally Cartesian coordinates co-moving with a freely falling local reference frame.
This is the mathematical expression of one of the properties of gravity expressed in
connection with the principle of equivalence, that gravity may be transformed away
locally by going into a freely falling room. So when do we experience a gravitational
6.10 The Tidal and Non-tidal Components of a Gravitational Field 193

field strength? The answer is: We experience a gravitational field strength when we
are in a room which is not freely falling. Note that this is valid whether spacetime is
curved of flat.
As seen from Eq. (6.93) of geodesic deviation spacetime curvature is connected
to inhomogeneity in a gravitational field—the difference between acceleration at two
nearby points. In Newton’s theory this is associated with the phenomenon of tidal
forces.
Inertial effects upon physical phenomena may be defined as those effects that
depend upon the state of acceleration or rotation of the reference frame. In acceler-
ated or rotating reference frames there are non-vanishing Christoffel symbols that
represent inertial effects, such as the Coriolis and centrifugal acceleration in a rotating
reference frame. We shall take a closer look at this connection.
It should be noted that not all of the Christoffel symbols represent a deviation from
uniform motion of free particles. For example, there are non-vanishing Christoffel
symbols in a system of polar coordinates in an inertial reference frame that only
tell about the geometrical properties of the coordinate system, but do not have any
kinematical significance. On the other hand the Christoffel symbols
00 i
represent the
acceleration of a free particle instantaneously at rest, and
j 0 represent the Coriolis
i

acceleration.
The gravitational field strength depends upon the chosen frame of reference and
vanishes in a local freely falling reference frame, i.e., an inertial reference frame.
Generally a gravitational field will have both a non-tidal tidal component which can
be transformed away, and a tidal component which cannot be transformed away. We
shall now consider the mathematical representation of these components.
Let us start by considering gravity in Newton’s theory. Introducing the gravita-
tional potential φ and the separation vector s between P0 and a nearby point P, the
acceleration of gravity at P is given by the first two terms of a Taylor expansion about
P0 ,
     2 
∂φ ∂φ ∂ φ
gi = − = − − sj = (gNT )i + (gT )i , (6.111)
∂xi P ∂xi P0 ∂xi ∂xj

where the non-tidal component of the gravitational field strength is

 
∂φ
(gNT )i = − , (6.112)
∂xi P0

and the tidal component is

 2 
∂ φ
(gT )i = − sj . (6.113)
∂xi ∂xj P0
194 6 Curvature

Noting that (gT )i = d2 si /dt 2 and comparing with Eq. (6.94) we see that in the
weak field limit there is a simple connection between the tidal component of the
gravitational field strength and spacetime curvature,
 2 
∂ φ
(gT )i = −Ri0j0 sj =− sj . (6.114)
∂xi ∂xj P0

This shows that the relativistic counterpart to inhomogeneity of a gravitational

field, i.e. to a tidal gravitational field, is spacetime curvature.
We now proceed to consider gravity according to Einstein’s theory. Making a
Taylor expansion of Eq. (6.109) for a free particle instantaneously at rest in a reference
frame with a stationary metric, we find the relativistic expression for the i-component
of the gravitational field strength
 i  
gRi = −
00 P0
−
00,j P0 s .
i j
(6.115)

In this case Eq. (6.17) gives

α i

00,j
i
= Ri0j0 +
0j
α0 −
00
k

kji . (6.116)

Inserting this into Eq. (6.115) leads to

 i   
α i
gRi = −
00 P0
+
k

i
00 kj −

0j α0 s j
− Ri
0j0 sj . (6.117)
P0 P0

The first term of Eq. (6.117) represents the acceleration of gravity at point P0 , i.e.,
it represents the uniform part of the gravitational field. The second term represents
the non-uniform part of the gravitational field which is also present in a non-inertial
reference frame in flat spacetime, for example, the non-uniformity of the centrifugal
field in a rotating reference frame. The last term represents the tidal effects, which
in the general theory is proportional to the spacetime curvature. This suggests the
following separation of a gravitational field into a non-tidal part and a tidal part:

gRi = gNT
i
+ gTi , (6.118)

where the non-tidal component of the gravitational field is given by

 i  
α i
i
gNT = −
00 P0
+
00
k

kji −
0j
α0 sj , (6.119)
P0

and the tidal part by


gTi = − Ri0j0 sj . (6.120)
P0
6.10 The Tidal and Non-tidal Components of a Gravitational Field 195

As in the Newtonian case, the non-tidal part of the gravitational field can be
transformed away locally by going into a local inertial frame. The tidal part cannot
be transformed away.

Example 6.10.1 Non-tidal gravitational field. Let us as a simple illustrationdp con-

sider the gravitational field strength in a rotating reference frame in flat spacetime.
In this case the Riemann curvature tensor vanishes, so the gravitational field has no
tidal component. The non-vanishing Christoffel symbols are given in Eq. (5.119).
Inserting these into Eq. (6.119) gives for the non-tidal component of the gravitational
field
θ
r
gNT = rω2 + ω2 sr , gNT = ω 2 sθ . (6.121)

The term rω2 is the centrifugal acceleration at the point P0 , and the other terms
are due to the inhomogeneity of the centrifugal gravitational field.

Exercises

6.1 Parallel transport and curvature

(a) A curve P(λ) runs through a point P = P(0), and a vector A  is defined at
this point. The vector is parallel transported along the curve so that in each

point P(λ) there is a well-defined vector A(λ) . Express the condition that
the vectors along the curve are parallel as an equation of the components
of the vector Aμ (λ). Show that the change of the components of the vector
by an infinitesimal displacement dxμ is
μ
dAμ = −
λν (x)Aλ dxν .

(b) A closed curve has the shape of a parallelogram with the sides da and db.
The corners of the parallelogram are denoted by A, B, C and D, respectively.
 is parallel transported from A and C along the two curves ABC
A vector A
and ADC. Show that the result in these two cases is in general not the same.
 by parallel transporting it
Then use this fact to show that the change of A,
along the closed curve ABCDA, is

δAα = −Rαβγ δ Aβ daγ dbδ ,

where Rαβγ δ is the Riemann curvature tensor.

6.2 Curvature of the simultaneity space in a rotating reference frame
Calculate the curvature scalar R of a 2-dimensional simultaneity space in a rotating
reference frame with the line-element
196 6 Curvature

Fig. 6.7 Spacetime

curvature and the tidal force
pendulum. Here
a = Rφ̂ t̂ φ̂ t̂ φ̂ cos θ −
Rr̂ t r̂ t̂ r̂ sin θ,  =
 sin θ eφ̂ +  cos θ er̂

r 2 dθ 2
dl 2 = dr 2 + .
1 − r 2 ω2 /c2

6.3 The tidal force pendulum and the curvature of space

We will again consider the tidal force pendulum. Here we shall use the equation for
geodesic deviation to find the period of the pendulum.

(a) Why can the equation for geodesic equation be used to find the period of the
pendulum in spite of the fact that the particles do not move along geodesics?
Assume that the centre of the pendulum is fixed at a distance R from the centre
of mass of the Earth. Introduce an orthonormal basis {eâ } with the origin at the
centre of the pendulum (see Fig. 6.7).
(b) Show to first order in v/c and φ/c2 , where v is the 3-velocity of the masses and
φ the gravitational potential at the position of the pendulum, that the equation
of geodesic deviation takes the form

d2 î
+ Rî0̂ĵ0̂ ĵ = 0.
dt 2
(c) Find the period of the pendulum expressed in terms of the components of
Riemann’s curvature tensor.
Chapter 7
Einstein’s Field Equations

Abstract This chapter starts with a hydrodynamical description of energy–momen-

tum conservation in a Newtonians context in order to give some intuition about
the relativistic formulation of energy–momentum conservation as represented by
a vanishing divergence of the energy–momentum tensor. Einstein demanded that
energy–momentum conservation should follow from the field equations, and hence
he needed a divergence-free curvature tensor. This is deduced from Bianchi’s 2. iden-
tity. It is shown that one need not postulate that free particles follow geodesic curves,
but that it follows from the field equations.

7.1 Newtonian Fluid

We shall begin this chapter by giving a mathematical formulation of the law of

energy–momentum conservation. Again (like we did in the Sect. 5.9.1) we shall start
by considering a Newtonian fluid.
The total derivative of a velocity field was presented in Sect. 5.9.1 and is

Dv ∂v  v.
≡ + (v · ∇) (7.1)
Dt ∂t
The component notation for the expression of the total derivative of the velocity
takes the form

Dvi ∂vi ∂vi

≡ + vj j . (7.2)
Dt ∂t ∂x
In Newtonian hydrodynamics mass conservation is represented mathematically
by the continuity equation,

© Springer Nature Switzerland AG 2020 197

Ø. Grøn, Introduction to Einstein’s Theory of Relativity,
Undergraduate Texts in Physics, https://doi.org/10.1007/978-3-030-43862-3_7
198 7 Einstein’s Field Equations

∂ρ ∂ρ ∂(ρvi )
+ ∇ · (ρv ) = 0 or + = 0. (7.3)
∂t ∂t ∂xi
Conservation of momentum is represented by the Euler’s equation of motion
(ignoring gravity),
 i i
D υ  ∂υ j ∂υ ∂p
ρ = −∇ p or ρ +v =− i. (7.4)
Dt ∂t ∂x j ∂x

The energy–momentum tensor is a symmetric tensor of rank 2 which describes

material properties,
⎛ ⎞
T 00 T 01 T 02 T 03
⎜ T 10 T 11 T 12 T 13 ⎟
T μν =⎜
⎝ T 20
⎟, (7.5)
T 21 T 22 T 23 ⎠
T 30 T 31 T 32 T 33

Here T 00 represents mass–energy density, T i0 represents momentum density, T ii

represents pressure for T ii > 0 and tension for T ii < 0, and T i j represents shear
forces for i = j.

Example 7.1.1 (Energy–momentum tensor of a Newtonian fluid) In the case of a

Newtonian fluid the components of the energy–momentum tensor are

T 00 = ρc2 , T i0 = ρcvi , T i j = ρvi v j + pδ i j , (7.6)

where ρ is the mass density, p the pressure, assumed isotropic here and vi the i-
component of the velocity. We choose a locally Cartesian coordinate system in an
inertial frame such that the covariant derivatives are reduced to partial derivatives.
μν
The divergence of the momentum–energy tensor, T;ν , has four components, one for
each value of μ.
The zeroth component is

∂ρ ∂(ρvi )
T;ν0ν = T,ν0ν = T,000 + T,i0i = + , (7.7)
∂t ∂xi
which in comparison with Newtonian hydrodynamics shows that

T;ν0ν = 0 (7.8)

is the continuity equation. This equation represents the conservation of mass–energy.

The i- component of the divergence is

ij ∂(ρvi ) ∂(ρvi v j + pδ i j )
T,νiν = T,0i0 + T, j = +
∂t ∂x j
7.1 Newtonian Fluid 199

∂vi ∂ρ ∂ρv j j ∂v
i
∂p
=ρ + vi + vi + ρv + i, (7.9)
∂t ∂t ∂x j ∂x j ∂x
According to the continuity equation

∂(ρυ i ) ∂ρ
=− , (7.10)
∂x i ∂t
which implies that

∂vi ∂ρ ∂ρ ∂vi ∂p Dvi ∂p

T ,νiν = ρ + vi − vi + ρv j j + i = ρ + i. (7.11)
∂t ∂t ∂t ∂x ∂x Dt ∂x
Hence the Euler equation of motion, which represents momentum conservation,
takes the form

T;νiν = 0. (7.12)

This is Euler’s equation of motion. It expresses the conservation of momentum.

It follows from Eqs. (7.8) and (7.12) that the equations
μν
T;ν = 0 (7.13)

are the general expressions for energy and momentum conservation.

7.2 Perfect Fluids

We now turn to the general relativistic case. A perfect fluid is a fluid with no viscosity.
The components of the energy–momentum tensor of a perfect fluid are

p
Tμν = ρ + 2 u μ u ν + pgμν , (7.14)
c
where ρ and p are the mass–density and pressure of tension, respectively, measured
in the rest frame of the fluid, and u μ are the components of the 4-velocity of the fluid.
In a co-moving orthonormal basis the components of the 4-velocity are u μ̂ =
(c, 0, 0, 0). Then the energy–momentum tensor is given by
⎛ ⎞
ρc2 0 0 0
⎜ 0 p 0 0⎟
Tμ̂ν̂ =⎜
⎝ 0
⎟, (7.15)
0 p 0⎠
0 0 0 p
200 7 Einstein’s Field Equations

where p > 0 is pressure and p < 0 is tension.

There are three different types of perfect fluids that are particularly useful:
1. Dust or non-relativistic gas is given by p = 0 and the energy–momentum tensor
Tμν = ρu μ u ν .
2. Radiation or ultra-relativistic gas is given by a traceless energy–momentum
tensor, i.e. Tμμ = 0. It follows that p = (1/3)ρc2 .
3. The third type is a sort of vacuum energy of particular significance for
construction of relativistic universe models.

7.2.1 Lorentz Invariant Vacuum Energy—LIVE

Let us follow a thought presented by the Belgian cosmologist Georges Lemaître

around 1935. Assume a particle is alone in the universe. It is not possible to define
motion for such a particle. Hence all motion is relative.
Quantum mechanics implies that particle–antiparticle pairs are created and then
annihilated again in a very short time restricted by the Heisenberg uncertainty rela-
tionships. Averaging over a macroscopic time and region in space this implies the
existence of a quantum mechanical vacuum energy on a macroscopic scale. If it is
possible to measure velocity relative to this energy, it would act as a sort of ether and
re-establish absolute motion into the physics.
According to the special theory of relativity, which has been experimentally con-
firmed in several ways, this cannot be the case. Hence it must be impossible to
measure velocity relative to the vacuum energy. This implies that all the components
of the energy–momentum tensor of the vacuum energy must be Lorentz invariant. It
is shown in Exercise 7.2 that Lorentz invariance of all the components of an energy–
momentum tensor implies that the energy–momentum tensor is proportional to the
metric tensor.
Assume now that the Lorentz invariant vacuum energy, LIVE, can be described
as a perfect fluid.

Tμν = ρLIVE + pLIVE /c2 u μ u ν + ρLIVE gμν (7.16)

Lorentz invariance then requires that the vacuum energy obeys the equation of
state

pLIVE = −ρLIVE c2 . (7.17)

Hence LIVE is in a state of strain.

7.2 Perfect Fluids 201

7.2.2 Energy–Momentum Tensor of an Electromagnetic Field

Given an electric field E = E i ei and a magnetic field B = B i ei . The electromagnetic
field tensor is an antisymmetric tensor of rank 2 given by (using units so that c = 1)
⎡ ⎤
0 E1 E2 E3
⎢ −E 1 0 −B 3 B 2 ⎥
Fην =⎢ ⎥
⎣ −E 2 B 3 0 −B 1 ⎦. (7.18)
−E 3 −B 2 B 1 0

The electromagnetic energy–momentum tensor is a symmetric tensor of rank 2

with components

1
Tμν = Fμα Fνα − gμν Fαβ F αβ . (7.19)
4
This tensor is used when one is going to find solutions of Einstein’s field equations
for spacetimes with electromagnetic fields.

7.3 Einstein’s Curvature Tensor

Einstein assumed that the field equations representing the relativistic generalization
of Newton’s law of gravitation have the form: spacetime curvature ∝ momentum–
energy tensor. Also, he demanded that energy and momentum conservation should
follow as a consequence of the field equation. This puts the following constraints on
the curvature tensor: It must be a symmetric, divergence-free tensor of rank 2.
A good candidate is the Ricci tensor introduced in Eq. (6.18). It is a symmetric
curvature tensor of rank 2. Let us see whether it is divergence free. In order to calculate
its divergence we start with Bianchi’s 2. identity which was deduced in Sect. 6.5,
μ μ μ
Rναβ;σ + Rνσ α;β + Rνβσ ;α = 0. (7.20)

Contraction of μ and α and using Eq. (6.73) gives

μ
Rνβ;σ − Rνσ ;β + Rνβσ ;μ = 0. (7.21)

Further contraction of ν and σ leads to

σ
2Rβ;σ = R;β . (7.22)

Thus, the divergence of the Ricci tensor is

202 7 Einstein’s Field Equations

σ 1
Rβ;σ = R;β . (7.23)
2
It is not vanishing. Hence, the Ricci tensor is not the curvature tensor to be put
into the left-hand side of the field equations. However, we can use this expression to
construct a new divergence-free curvature tensor.
Since the metric tensor is covariant divergence free, we have that (gβσ R);σ =
σ
gβ R;σ . Now we multiply Eq. (7.23) by gαβ to get

 1 β
gαβ Rβσ − g R = 0. (7.24)
;σ 2 α ;β
Interchanging σ and β in the first term and using that the mixed components of
the metric tensor are gαβ = δαβ , we get
 
1
Rαβ − δαβ R = 0. (7.25)
2 ;β

Hence the tensor Rαβ − (1/2)δαβ R is divergence-free. This tensor is called the
Einstein tensor, and its covariant components are denoted by E αβ , that is

1
E αβ = Rαβ − gαβ R. (7.26)
2
In 4. dimensional spacetime the metric tensor has ten independent components.
Due to the identities
μν
E ;ν = 0, (7.27)

which represent four equations that any Einstein tensor fulfils as a consequence
of Bianchi’s 2. identity, the field equations give only six independent equations to
determine the components of the metric tensor. Since there are ten independent
components, this leaves four free metric functions. This secures a free choice of
coordinate system.

7.4 Einstein’s Field Equations

Einstein’s field equations are

E μν = κ Tμν (7.28)

where κ is a constant called Einstein’s constant of gravity. Inserting the expression

(7.26) for the components of the Einstein tensor we have
7.4 Einstein’s Field Equations 203

1
Rμν − gμν R = κ Tμν . (7.29)
2
Contraction and using that gμμ = δμμ = 4 gives

R = −κ T, (7.30)

where T = Tμμ . Thus the field equations may be written in the form
 
1
Rμν = κ Tμν − gμν T . (7.31)
2

In the Newtonian limit the metric may be written as

 
2φ 2 2
ds = − 1 + 2 c dt + (1 + h ii )(dx 2 + dy 2 + dz 2 ),
2
(7.32)
c

where the Newtonian potential |φ|  c2 , and h ii is a perturbation of the metric

satisfying |h ii |  1. We also have T00  Tkk and T ≈ −T00 . Then the 00-component
of the field equations may be approximated by
κ
R00 ≈ T00 . (7.33)
2
Furthermore we have

μ ∂ 00
k
1
R00 = R0μ0 = R0i0
i
= 00,i
i
− 0i,0
i
= = 2 ∇ 2 φ· (7.34)
∂xk c

Since T00 ≈ ρc2 , Eq. (7.33) can be written as

1 4
∇2φ = κc ρ. (7.35)
2
Comparing this equation with the Newtonian law of gravitation on local form,

∇ 2 φ = 4π Gρ, (7.36)

we see that Einstein’s constant of gravity is

8π G
κ= . (7.37)
c4
204 7 Einstein’s Field Equations

It has the value κ = 2.077 × 10−43 s2 /m kg.

In classical empty space we have Tμν = 0, which gives

E μν = 0, (7.38)

or

Rμν = 0. (7.39)

These are the field equations for empty space. Note that Rμν = 0 does not imply
Rμναβ = 0. In general curvature is non-vanishing in empty space. In the words of J.
A. Wheeler: “Mass there curves spacetime here”.
It was shown by D. Hilbert that the field equations may be deduced from a
variational principle with action

√
R −gd4 x, (7.40)

√
where R −g is the Lagrange density. One may also include a so-called cosmological
constant :

√
(R + 2) −gd4 x. (7.41)

The field equations with cosmological constant are

1
Rμν − gμν R + gμν = κ Tμν . (7.42)
2
The field equations of empty space with a cosmological constant are

1
Rμν − gμν R + gμν = 0. (7.43)
2
Solutions of these equations are sometimes called Einstein spaces.

7.5 The “Geodesic Postulate” as a Consequence of the Field

Equations

The principle that free particles follow geodesic curves has been called the “geodesic
postulate”. We shall now show that the “geodesic postulate” follows as a consequence
of the field equations.
7.5 The “Geodesic Postulate” as a Consequence of the Field Equations 205

Consider a system of free particles in curved spacetime. This system can be

regarded as a pressure-free gas. Such a gas is called dust. It is described by an
energy–momentum tensor

T μν = ρu μ u ν , (7.44)

where ρ is the rest density of the dust as measured by an observer at rest in the dust,
and u μ are the components of the 4-velocity of the dust particles.
Einstein’s field equations as applied to spacetime filled with dust take the form

1
R μν − g μν R = κρu μ u ν . (7.45)
2
Because the divergence of the left-hand side is zero, the divergence of the right-
hand side must be zero, too,

(ρu μ u ν );ν = 0 (7.46)

or

(ρu ν u μ );ν = 0. (7.47)

We now regard the quantity in the parenthesis as a product of ρu ν and u μ . By the

rule for differentiating a product we get
μ
(ρu ν );ν u μ + ρu ν u ;ν = 0. (7.48)

Since the 4-velocity of any object has a magnitude equal to the velocity of light
we have

u μ u μ = −c2 . (7.49)

Differentiation gives

(u μ u μ );ν = 0 . (7.50)

Using, again, the rule for differentiating a product, we get

μ
u μ;ν u μ + u μ u ;ν = 0. (7.51)

From the rule for raising an index and the freedom of changing a summation index
from α to μ, say, we get

u μ;ν u μ = u μ u μ;ν = g μα u α u μ;ν = u α g μα u μ;ν = u α u α;ν = u μ u μ;ν . (7.52)

206 7 Einstein’s Field Equations

Thus the two terms of Eq. (7.51) are equal. It follows that each of them are equal
to zero. So we have
μ
u μ u ;ν = 0. (7.53)

Multiplying Eq. (7.48) by u μ , we get

μ
(ρu ν );ν u μ u μ + ρu ν u μ u ;ν = 0. (7.54)

Using Eq. (7.47) in the first term and Eq. (7.51) in the last term, which then
vanishes, we get

(ρu ν );ν = 0. (7.55)

Thus the first term in Eq. (7.48) vanishes and we get

μ
ρu ν u ;ν = 0. (7.56)

Since ρ = 0 we must have

μ
u ν u ;ν = 0. (7.57)

This is just the geodesic equation. Hence, it follows from Einstein’s field equations
that free particles follow geodesic curves of spacetime.

7.6 Einstein’s Field Equations Deduced from a Variational

Principle

It was shown by David Hilbert how Einstein’s field equations can be deduced from
a variational principle. A detailed discussion of this is given in for example Ø. Grøn
and S. Hervik: “Einstein’s General Theory of Relativity”, Chap. 8. We shall here
only give a brief summary as a preparation for the presentation of the Kaluza–Klein
theory in the Appendix.
Hilbert’s variational principle has the form

√
δ I = 0, I = L −g d4 x, (7.58)

where I is the action, L the Lagrange function—also called the Lagrangian—and g

the determinant of the metric tensor. Einstein’s field equations for empty space are
obtained from the Lagrange function

L = R, (7.59)
7.6 Einstein’s Field Equations Deduced from a Variational Principle 207

where R is the Ricci curvature scalar. This gives

  
μν 1 μν √
δI = R − Rg δ gμν −g d4 x. (7.60)
2

Requiring that δ I = 0 for arbitrary variations of gμν gives the field equations for
empty space

1
R μν − R g μν = 0. (7.61)
2
Einstein’s field equations (7.29) for a spacetime filled with energy and matter,
described by an energy–momentum tensor with mixed components Tνμ , are now
written
1
Rνμ − R δ μ = κ Tνμ . (7.62)
2 ν
According to the Lagrangian formalism the energy–momentum tensor of an
electromagnetic field is given by

∂L
μ0 T̄νμ = Aρ, ν − Lδνμ , (7.63)
∂ Aρ, μ

where Aρ are the components of the electromagnetic vector potential, and μ0 is the
permeability of empty space. The electromagnetic field tensor is given in terms of
the electromagnetic vector potential in Eq. (5.263) which is here written in the form

Fμν = Aν, μ − Aμ, ν . (7.64)

The Lagrange function of the electromagnetic field is

1 μν
L= F Fμν . (7.65)
4
In order to calculate the first term at the right-hand side of Eq. (7.3) this is written
as
1 μν 
L= F Aν, μ − Aμ, ν . (7.66)
4
Due to the antisymmetry of F μν this can be written as

1 μν
L= F Aν, μ . (7.67)
2
208 7 Einstein’s Field Equations

Hence
∂L
= Fμρ, (7.68)
∂ Aρ, μ

so
1
μ0 T̄νμ = F ρ μ Aρ, ν − Fρσ F ρσ δνμ . (7.69)
4
The covariant components are

1
μ0 T̄μν = Fμρ Aρ, ν − Fρσ F ρσ gμν . (7.70)
4
The energy–momentum tensor should be symmetric and divergence free in order
to fit into Einstein’s field equations. But the first term in the expression (7.70) is
not symmetric in μ and ν. The usual procedure for obtaining a symmetric energy–
λμ
momentum tensor is the following. Add a term of the form K ν,λ to the expression
λμ
(7.69), where K ν is antisymmetric in the first two indices. The mathematical expres-
sion of energy–momentum conservation is that the divergence of the energy–momen-
λμ
tum tensor vanishes. The divergence of the added term is K ν, λμ , which is symmetric
in the lower indices λ and μ and antisymmetric in the upper ones. So the summation
over λ and μ makes this term vanish. This means that the tensor
λμ
Tνμ = T̄νμ + K ν,λ (7.71)

is equally good for describing energy–momentum conservation as T̄νμ .

For electromagnetism we choose

K λμν = F μλ Aν . (7.72)

Hence
λμ  μλ
K ν ,λ = F μλ Aν , λ = F, λ Aν + F μλ Aν, λ . (7.73)

The component form of Maxwell’s source free equations (5.280) is

μλ
F, λ = 0, (7.74)

giving
λμ
K ν ,λ = F μλ Aν, λ . (7.75)

From Eqs. (7.70), (7.71) and (7.75) we get

7.6 Einstein’s Field Equations Deduced from a Variational Principle 209

 1
μ0 Tμν = Fμρ Aν, ρ − Aρ, ν − Fρσ F ρσ gμν . (7.76)
4
Using Eq. (7.64) we finally get the symmetric energy–momentum tensor of an
electromagnetic field

1
μ0 Tμν = Fμρ Fρν − Fρσ F ρσ gμν . (7.77)
4
Example 7.6.1 (The energy–momentum tensor of an electric field in a spherically
symmetric spacetime) We shall consider a spherically symmetric space with a charge
at the centre of the coordinate system. Then there is a static, radial electric field in
this spacetime, and the electromagnetic field tensor has only two non-vanishing
components

1 Q
F01 = −F10 = Er = . (7.78)
c 4π ε0 cr 2

In the present case the only non-vanishing components of the electromagnetic

energy–momentum tensor (7.77) are

1
μ0 Tμ μ = gμ0 Fμ1 F 01 + gμ1 Fμ0 F 10 − gμ μ F01 F 01 , (7.79)
2
where we have summarized two equal terms in the last term due to the antisymmetry
F01 = −F10 .
We use spherical coordinates so the angular components are

gθθ = r 2 , gφφ = r 2 sin2 θ (7.80)

as given in Eq. (3.122). The non-vanishing components of the energy–momentum

tensor are
 
1 1 1 Q 2 g00
T00 = g00 F01 F − g00 F01 F
01 01
= g00 F01 F 01 = , (7.81)
μ0 2 2μ0 32π 2 ε0 r 4
 
1 1 1 Q 2 g11
T11 = g11 F01 F − g11 F01 F
01 01
= g11 F01 F 01 = , (7.82)
μ0 2 2μ0 32π 2 ε0 r 4
1 1 2 Q2 1
T22 = g22 F01 F 01 = r F01 F 01 = . (7.83)
2μ0 2μ0 32π 2 ε0 r 2
1 1 2 Q 2 sin2 θ
T33 = g33 F01 F 01 = r sin2 θ F01 F 01 = . (7.84)
2μ0 2μ0 32π 2 ε0 r 2
210 7 Einstein’s Field Equations

Exercises

7.1. Newtonian approximation of perfect fluid

Let

T αβ = p ηαβ + ρ + p/c2 u α u β

be the components of the energy momentum tensor of a perfect fluid in flat

spacetime with Minkowski metric ημν . Here p is the pressure and ρ the mass
density of the fluid, and u α the components of its 4-velocity.
αβ αβ
(a) Explain why the conservation law T;β = 0 in this case reduces to T,β = 0.
(b) We shall consider the Newtonian limit where p/c2 can be neglected com-
pared to ρ in the second term of T αβ , and the components of the 4-velocity
of the fluid are u α ≈ (c, v) where v is the ordinary velocity of a fluid
element. Show that in this case the conservation law (a) implies mass
conservation as represented by the equation of continuity,

∂ρ
+ ∇ · (ρ v) = 0,
∂t
(c) And momentum conservation as represented by the Euler equation of
motion,
 
∂v
ρ + (v · ∇)v = −∇ p.
∂t

7.2. The energy–momentum tensor of LIVE

(a) Show that the energy–momentum tensor of a Lorentz invariant medium
is proportional to the metric tensor.
(b) Show that a Lorentz invariant perfect fluid has equation of state p = −ρc2 .
(c) How is the density of Lorentz invariant vacuum energy, LIVE, related to
the cosmological constant?
Chapter 8
Schwarzschild Spacetime

Abstract The Schwarzschild solution describing spacetime outside a spherical mass

distribution is deduced. In this deduction we give a detailed prescription of how one
calculates the components of Einstein’s curvature tensor using differential forms
as decomposed in an orthonormal basis. The predictions for the classical tests
of Einstein’s theory—gravitational frequency shift and time dilation, deflection of
light passing the Sun and the perihelion shift of Mercury—are deduced. Finally
the Reissner–Nordström solution describing spacetime outside a charged particle is
deduced.

8.1 Schwarzschild’s Exterior Solution

This is a solution of the vacuum field equations E μν = 0 for a static spherically

symmetric spacetime. One can then choose the following form of the line element
(employing units so that c = 1):

ds 2 = −e2α(r ) dt 2 + e2β(r ) dr 2 + r 2 dΩ 2 ,
dΩ 2 = dθ 2 + sin2 θ dφ 2 . (8.1)

These coordinates are chosen so that the area of a sphere with radius r is 4πr 2 .
They are often called “curvature coordinates”.
The physical distance in the radial direction, corresponding to a coordinate
distance dr, is
√
dlr = grr dr = eβ(r ) dr. (8.2)

We shall now determine the components of the Einstein tensor by using the Cartan
formalism and give the procedure in eight steps.

© Springer Nature Switzerland AG 2020 211

Ø. Grøn, Introduction to Einstein’s Theory of Relativity,
Undergraduate Texts in Physics, https://doi.org/10.1007/978-3-030-43862-3_8
212 8 Schwarzschild Spacetime

1. We first express the basis forms in orthonormal basis in terms of the coordinate
basis forms,

tˆ
ω
−
= eα(r ) d− t ω
−
r̂
= eβ(r ) d− r, ω
−
θ̂
= r d− θ ω
−
φ̂
= r sin θ d− φ. (8.3)

2. Then we compute the connection forms by applying Cartan’s 1. structure

equations

μ̂
dωμ̂ = −Ων̂ ∧ ω
−
v̂
. (8.4)

dωtˆ = eα α  d− r ∧ d− t
tˆ
= eα α  e−β ω
−
r̂
∧ e−α ω
−

tˆ
= −e−β α  ω
−
∧ω
−
r̂

tˆ
= −Ω
−
∧ ωr̂ (8.5)
r̂

Hence
tˆ tˆ
Ω
−
= e−β α  ω
−
+ f1ω
−
r̂
. (8.6)
r̂

3. To determine the f-functions, we apply the antisymmetry

μ̂ν̂ = −
ν̂ μ̂ . (8.7)

This gives the non-zero connection forms

φ̂

r̂φ̂ = −
r̂ = − r1 e−β ωφ̂ ,
φ̂

θ̂φ̂ = −
θ̂ = − r1 cot θ ωφ̂ ,
(8.8)

r̂tˆ = +
r̂tˆ = e−β α  ωtˆ,

r̂θ̂ = −
r̂θ̂ = − r1 e−β ωθ̂ .

4. We then proceed to determine the curvature forms by applying Cartan’s 2.

structure equations

μ̂ μ̂
μ̂ α̂
R
−
= dΩ ν̂ + Ω
−
∧Ω
−
, (8.9)
ν̂ α̂ ν̂

which gives
8.1 Schwarzschild’s Exterior Solution 213

tˆ tˆ
R
−
= −e−2β (α  + α 2 − α  β  )ω
−
∧ω
−
r̂
,
r̂
tˆ tˆ
R
−
= − r1 e−2β α  ω
−
∧ω
−
θ̂
,
θ̂
tˆ tˆ
R
−
= − r1 e−2β α  ω
−
∧ω
−
φ̂
,
φ̂
(8.10)
R
−
r̂
= r1 e−2β β  ω
−
r̂
∧ω
−
θ̂
,
θ̂
R
−
r̂
= r1 e−2β β  ω
−
r̂
∧ω
−
φ̂
,
φ̂
θ̂
R
−
= 1
r2
(1 − e−2β )ω
−
θ̂
∧ω
−
φ̂
.
φ̂

5. By applying the relation

1 μ̂ α̂
R μ̂ν̂ = R ω ∧ ωβ̂ , (8.11)
2 ν̂ α̂β̂
we find the components of Riemann’s curvature tensor.
6. Contraction gives the components of Ricci’s curvature tensor,

Rμ̂ν̂ ≡ Rμ̂α̂ α̂ν̂ . (8.12)

7. A new contraction gives Ricci’s curvature scalar,

R ≡ Rμ̂μ̂ . (8.13)

8. The components of the Einstein tensor can then be found,

1
E μ̂ν̂ = Rμ̂ν̂ − ημ̂ν̂ R, (8.14)
2
where ημ̂ν̂ = diag(−1, 1, 1, 1). Hence

E tˆtˆ = r2 e−2β β  + r12 (1 − e−2β ),

Er̂ r̂ = r2 e−2β α  − r12 (1 − e−2β ),  (8.15)
 β
E θ̂ θ̂ = E φ̂ φ̂ = e−2β α  + α 2 − α  β  + αr − r
.

We want to solve the equations E μ̂ν̂ = 0. In the present case there are only two
independent equations,

E tˆtˆ = 0 and Er̂ r̂ = 0. (8.16)

Adding the two equations we get

2 −2β 
e (β + α  ) = 0. (8.17)
r
214 8 Schwarzschild Spacetime

Integration gives

α + β = K1, (8.18)

where K 1 is an integration constant.

We now have

ds 2 = −e2α dt 2 + e2(K 1 −α) dr 2 + r 2 d

2 . (8.19)

Letting r → ∞ the line element should describe flat spacetime in spherical

coordinates with α = β = 0. This requires that K 1 = 0, and hence, that β = − α.
Since we have ds 2 = −e2α dt 2 + e−2α dr 2 + r 2 d
2 , this means that grr = − 1/gtt .
We must solve one more equation to get the complete solution and choose the equation
E tˆtˆ = 0, which gives

2 −2β  1
e β + 2 (1 − e−2β ) = 0. (8.20)
r r
This equation can be written as

1 d  
2
r 1 − e−2β = 0. (8.21)
r dr
Integration gives

r (1 − e−2β ) = K 2 . (8.22)

If we choose K 2 = 0 we get β = 0 giving α = 0 and

ds 2 = −dt 2 + dr 2 + r 2 d
2 , (8.23)

which is the Minkowski spacetime described in spherical coordinates. In general,

K 2 = 0 and 1 − e−2β = Kr2 ≡ Kr , giving

K
e2α = e−2β = 1 − , (8.24)
r
and

K dr 2
ds = − 1 −
2
dt 2 + + r 2 d
2 . (8.25)
r 1 − Kr

We can find K by going to the Newtonian limit and compare with a purely New-
tonian calculation. According to Newton’s theory the acceleration of gravity at a
distance r from a spherical mass M is
8.1 Schwarzschild’s Exterior Solution 215

d2 r GM
g= =− 2 . (8.26)
dt 2 r
Let us now calculate the corresponding acceleration in the Newtonian limit of
the general theory of relativity. In this limit the proper time τ of a particle will be
approximately equal to the coordinate time t. The acceleration of a freely falling
particle in 3-space is given by the geodesic equation,

d2 x μ μ
+ αβ u α u β = 0. (8.27)
dτ 2
For a particle instantaneously at rest in a weak field, we have dτ ≈ dt. Using
u μ = (1, 0, 0, 0), we get

d2 r
g= = − ttr . (8.28)
dt 2
This equation gives a physical interpretation of ttr as the gravitational accelera-
tion. This is a mathematical way to express the principle of equivalence: The grav-
itational acceleration can be transformed to 0 since the Christoffel symbols always
can be transformed into 0 locally, by going into a freely falling non-rotating frame,
i.e. a local inertial frame. In the Newtonian approximation we have
⎛ ⎞
1 rα ⎜ ∂gαt ∂gαt ∂gtt ⎟ 1 ∂gtt
ttr = g ⎜⎝ + − α⎟ ⎠ = − 2grr ∂r . (8.29)
2


∂t ∂t

 ∂ x
1
gr α =0 =0

Inserting

K ∂gtt K
gtt = − 1 − , =− 2 (8.30)
r ∂r r

into Eq. (8.29) and then calculating the acceleration of gravity from Eq. (8.28) we
get

K
g = − ttr = − . (8.31)
2r 2
Comparing with the Newtonian expression for the acceleration of gravity,
Eq. (8.26), and inserting the velocity of light, then lead to

2GM
K = . (8.32)
c2
Then we have the line element of the exterior Schwarzschild metric,
216 8 Schwarzschild Spacetime

2GM 2 2 dr 2
ds = − 1 − 2
2
c dt + + r 2 d
2 . (8.33)
c r 1 − 2GM
c2 r

We now introduce the Schwarzschild radius of a mass M,

2GM
RS ≡ . (8.34)
c2
Hence, the line element of the exterior Schwarzschild spacetime as expressed in
curvature coordinates, takes the form

RS 2 2 dr 2
ds 2 = − 1 − c dt + + r 2 d
2 . (8.35)
r 1 − RrS

The Schwarzschild radius of the Earth is R S ∼ 0.9 cm and of the Sun, R S ∼ 3 km.
Far from a localized mass distribution the gravitational field is weak. The definition
is that in a region where r  R S , there is a weak gravitational field.
A standard clock at rest in the Schwarzschild spacetime shows a proper time τ :

RS
dτ = 1− dt. (8.36)
r

It follows from the time independence of the metric that the coordinate clocks
are adjusted to go at the same rate independent of their position. Hence Eq. (8.36)
shows that the rate of proper time is slower for decreasing value of r, i.e. farther
down in the gravitational field. Time is not running at the Schwarzschild radius.

Definition 8.1.1 (Physical singularity) A physical singularity is a point where the

curvature is infinitely large.

Definition 8.1.2 (Coordinate singularity) A coordinate singularity is a point (or a

surface) where at least one of the components of the metric tensor is infinitely large,
but where the curvature of spacetime is finite.

The Kretschmann’s curvature scalar is Rμναβ R μναβ . From the Schwarzschild

metric, we get

R S2
Rμναβ R μναβ = 12 , (8.37)
r6
which diverges only at the origin. Since there is no physical singularity at r = R S ,
the singularity here is just a coordinate singularity and can be removed by a transfor-
mation to a coordinate system falling inwards (Eddington–Finkelstein coordinates,
Kruskal–Szekeres analytical extension of the description of Schwarzschild spacetime
to include the region inside RS ).
8.2 Radial Free Fall in Schwarzschild Spacetime 217

8.2 Radial Free Fall in Schwarzschild Spacetime

The Lagrangian function of a particle moving radially in Schwarzschild spacetime

is

1 RS 2 2 1 ṙ 2 d
L =− 1− c t˙ +  , • ≡ , (8.38)
2 r 2 1− r R S dτ

where τ is the time measured on a standard clock which the particle is carrying. The
momentum pt conjugate to the cyclic coordinate t is a constant of motion,

∂L RS 2
pt = =− 1− c t˙. (8.39)
∂ t˙ r

The 4-velocity identity, u μ u μ = −c2 , takes the form


RS 2 2 ṙ 2
− 1− c t˙ + = −c2 . (8.40)
r 1 − RrS

Inserting the expression for t˙ from Eq. (8.39) gives


pt2 RS 2
ṙ 2 − = − 1 − c . (8.41)
c2 r

The boundary condition is that the particle is falling from rest at r = r0 , giving

pt2 RS 2
= 1 − c . (8.42)
c2 r0

Inserting this into Eq. (8.41) gives

  
dr pt2 RS 2  1 1
ṙ = =− 2
− 1− c = − RS − . (8.43)
dτ c r r r0

With the initial condition τ (r0 ) = 0 we have

r  
r RS
dr = −c τ. (8.44)
r0 − r r0
r0

Performing the integration gives

218 8 Schwarzschild Spacetime
    
3/2
r0 π r r r
τ= √ + 1− − arcsin . (8.45)
c RS 2 r0 r0 r0

The proper time that a particle spends falling from r0 to R S is

    
3/2
r0 π RS RS RS
τ= √ + 1− − arcsin , (8.46)
c RS 2 r0 r0 r0

which is finite. Assuming that r0  R S and calculating to 1. order in R S /r0 we

obtain

π r0 r0
τ (R S ) ≈ . (8.47)
2 c RS

Let us calculate the corresponding travelling time as measured by a stationary

observer. From Eqs. (7.39) and (7.42) we have

1− RS
r0
dt = dτ. (8.48)
1− RS
r

From (8.43) we have


r0 r
dτ = − dr. (8.49)
R S (r0 − r )

Hence the time taken by the particle to fall down to the Schwarzschild radius, as
measured by the stationary observer, is
 r0
1 1 r 3/2
t(R S ) = − √ dr , (8.50)
RS r0 (r − R S ) r0 − r
RS

which diverges. Hence it takes an infinitely long coordinate time to fall down to the
Schwarzschild radius. But the proper time of the falling object is finite.

8.3 Light Cones in Schwarzschild Spacetime

The Schwarzschild line element (with c = 1) is


RS dr 2
ds 2 = − 1 − dt 2 +   + r 2 d
2 . (8.51)
r 1 − RrS
8.3 Light Cones in Schwarzschild Spacetime 219

We will look at radially moving photons. Then ds 2 = d

2 = 0, giving
1 √
r 2 dr r − RS
√ =± dt, (8.52)
r − RS
1
r2
or
r dr
= ±dt (8.53)
r − RS

with + for outward motion and − for inward motion. For inwardly moving photons,
integration yields
 
 r 

r + t + R S ln − 1 = k = constant. (8.54)
RS

We now introduce a new time coordinate t  such that the equation of motion for
photons moving inwards takes the form

r + t  = k. (8.55)

Hence in this coordinate system the coordinate velocity of light is equal to the
invariant velocity of light,

dr
= −1. (8.56)
dt
It follows from Eqs. (8.54) and (8.55) that
 
 r 
 
t = t + R S ln − 1. (8.57)
R S

The coordinate t  is called an ingoing Eddington–Finkelstein coordinate. For

photons moving outwards we have
 
 r 
r + R S ln − 1 = t + k. (8.58)
RS

Substituting for t from Eq. (8.57) we get

 
 r 
r + 2R S ln − 1 = t  + k. (8.59)
RS

Differentiating we find the coordinate velocity of outgoing light in this coordinate

system,
220 8 Schwarzschild Spacetime

dr r − RS

= . (8.60)
dt r + RS

Making use of curvature coordinates we get the following coordinate velocities

for inwardly and outwardly moving photons,

dr RS
vL = =± 1− , (8.61)
dt r

which shows how light is decelerated in a gravitational field. Figure 8.1 illustrates how
this is viewed by a non-moving observer located far away from the mass. In Fig. 8.2
we have instead used the time coordinate t  of the ingoing Eddington–Finkelstein
coordinate system.
Note that since the special theory of relativity is valid locally, all material particles
have world lines inside the light cone formed by the light they emit. From the shape
of the light cone inside r = R S we see that nothing emitted from a position inside
the spherical surface r = R S can escape from this region. An observer outside this
surface cannot see anything from the inside region. Hence this surface is a horizon
for an external observer.

Fig. 8.1 Light cones in Schwarzschild spacetime with Schwarzschild time. At a radius r = R S the
light cones collapse, and nothing can any longer escape, when we use the Schwarzschild coordinate
time
8.3 Light Cones in Schwarzschild Spacetime 221

Fig. 8.2 Light cones in Schwarzschild spacetime with Eddington-Finkelstein time. Using the ingo-
ing Eddington–Finkelstein time coordinate there is no collapse of the light cone at r = R S . Instead
we get a collapse at the singularity at r = 0. The angle between the left part of the light cone and the
t  -axis is always 45◦ . We also see that once the emitter gets inside the horizon at r = R S , nothing
can escape

The region inside the horizon is called a black hole since it is invisible because radi-
ation cannot come out of this region according to classical (non-quantum mechanical)
general relativity. A time-reversed black hole is called a white hole.
Let us compare the coordinate velocity with curvature coordinates of a parti-
cle falling vertically with the velocity of light moving vertically. It follows from
Eqs. (8.48) and (8.49) that a particle falling from rest at r = r0 has a coordinate
velocity

 
dr ṙ RS R S (r0 − r )
v= = =− 1− . (8.62)
dt t˙ r (r0 − R S )r

Comparing with the coordinate velocity of light as given in Eq. (8.61) we see that

R S (r0 − r )
v= vL . (8.63)
(r0 − R S )r
222 8 Schwarzschild Spacetime

The coordinate velocity of a particle falling from rest at an infinitely far position
is 
RS
lim v = v L . (8.64)
r →∞ r

Both Eqs. (8.63) and (8.64) give v(R S ) = v L . Hence, a particle falling freely from
any distance moves with the velocity of light through the horizon of a black hole.

8.4 Analytical Extension of the Curvature Coordinates

The curvature coordinates are co-moving with a static reference frame outside a
spherical mass distribution. If the mass has collapsed to a black hole, there exists a
horizon at the Schwarzschild radius. As we have seen in Sect. 8.3 there do not exist
static observers at finite radii inside the horizon. Hence, the curvature coordinates
are well defined only outside the horizon.
Also the rr-component of the metric tensor has a coordinate singularity at the
Schwarzschild radius. The curvature of spacetime is finite here. Kruskal and Szekeres
have introduced new coordinates that are well defined inside as well as outside the
Schwarzschild radius, and with the property that the metric tensor is non-singular
for all r > 0.
In order to arrive at these coordinates we start by considering a photon moving
radially inwards. From Eq. (8.54) we then have
 
 r 

t = −r − R S ln − 1 + v, (8.65)
R S

where v is a constant along the world line of the photon. We introduce a new radial
coordinate
 
 r 
∗
r ≡ r + R S ln  − 1. (8.66)
RS

Then the equation of the world line of the photon takes the form

t + r ∗ = v. (8.67)

The value of the constant v does only depend upon the point of time when the
photon was emitted. We may therefore use v as a new time coordinate.
For an outgoing photon we get in the same way

t − r ∗ = u, (8.68)
8.4 Analytical Extension of the Curvature Coordinates 223

where u is a constant of integration, which may be used as a new time coordinate for
outgoing photons. The coordinates u and v are the generalization of the light cone
coordinates of Minkowski spacetime to the Schwarzchild spacetime.
From Eqs. (8.67) and (8.68) we get

1
dt = (dv + du), (8.69)
2
1
dr ∗ = (dv − du), (8.70)
2
and from Eq. (8.66),

Rs
dr = 1 − dr ∗ . (8.71)
r

Inserting these differentials into Eq. (8.51) we arrive at a new form of the
Schwarzschild line element,

Rs
ds = − 1 −
2
du dv + r 2 d
2 . (8.72)
r

The metric is still not well behaved at the horizon. Kruskal and Szekeres found
coordinates that are well behaved at the horizon.
Introducing the coordinates

U = −e− 2Rs ,
u
(8.73)

v
V = e 2Rs , (8.74)

gives
 
r∗  Rs  r
U V = −e 2Rs = −e Rs = − − 1e Rs
v−u
(8.75)
r

and
dU dV
du dv = −4Rs2 . (8.76)
UV
The line element (8.72) then takes the form

4Rs3 − r
ds 2 = − e Rs dU dV + r 2 d
2 . (8.77)
r
224 8 Schwarzschild Spacetime

This is the first form of the Kruskal–Szekeres line element. Here there is no
coordinate singularity, only a physical singularity at r = 0.
We may furthermore introduce two new coordinates:
 1
1 r 2 r t
T = (V + U ) =  − 1 e 2Rs sinh , (8.78)
2 Rs 2Rs
 1
1 r 2 r t
Z = (V − U ) =  − 1 e 2Rs cosh . (8.79)
2 Rs 2Rs

Hence

V = T + Z, (8.80)

U = T − Z, (8.81)

giving

dU dV = dT 2 − dZ 2 . (8.82)

Inserting this into Eq. (8.72) we arrive at the second form of the Kruskal–Szekeres
line element

4Rs3 − r  2 
ds 2 = − e Rs dT − dZ 2 + r 2 d
2 . (8.83)
r
The inverse transformations of Eqs. (8.74) and (8.75) are
 
r  r
 − 1 e Rs = Z 2 − T 2 , (8.84)
R 
s

t T
tanh = . (8.85)
2Rs Z

Note from Eq. (8.83) that with the Kruskal–Szekeres coordinates T and Z the
equation of the radial null geodesics has the same form as in flat spacetime:

Z = ±T + constant. (8.86)
8.5 Embedding of the Schwarzschild Metric 225

8.5 Embedding of the Schwarzschild Metric

We will now look at a static, spherically symmetric space. A simultaneity surface

dt = 0 through the equatorial plane, dθ = 0, has the line element

ds 2 = grr dr 2 + r 2 dφ 2 (8.87)

with a radial coordinate such that a circle with radius r has a circumference of length
2πr .
Now we embed this surface in a flat three-dimensional space with cylinder
coordinates (z, r, φ) and line element

ds 2 = dz 2 + dr 2 + r 2 dφ 2 . (8.88)

The surface described by the line element in Eq. (8.87) has the equation z = z(r ).
The line element in (8.88) can therefore be written as
  2 
dz
ds = 1 +
2
dr 2 + r 2 dφ 2 . (8.89)
dr

Demanding that (8.89) is in agreement with (8.87) we get

 2
dz
grr = 1 + , (8.90)
dr
or
dz 
= ± grr − 1. (8.91)
dr
Choosing the positive solution gives

dz = grr − 1dr. (8.92)

In the Schwarzschild spacetime we have

1
grr = . (8.93)
1− RS
r

Making use of this we find the equation of the intersection of the simultaneity
surface dt = 0 through the equatorial plane, dθ = 0, and the paper plane,
 r 
dr
z=  = 4R S (r − R S ). (8.94)
RS r
RS
−1
226 8 Schwarzschild Spacetime

Fig. 8.3 Embedding of the

extended Schwarzschild
spacetime. It represents a
worm-hole connecting a
black and a white hole

Rotating this about the symmetry axis and including negative values of z give the
surface shown in Fig. 8.3, which consists of two so-called Flamm paraboloids glued
together at the Schwarzschild radius.

8.6 The Shapiro Experiment

The radial speed of light in curvature coordinates as measured by an observer far

from a mass distribution, say the Sun, is given in Eq. (8.61). The formula shows that
the speed of light slows down farther down in a gravitational field.
To measure this effect, one can look at how long it takes for light to get from, for
example, Mercury (Shapiro used Venus in the first experiment in 1967) to the Earth
[1]. This is illustrated in Fig. 8.4.
The travel time from z1 to z2 is
 z2  z2   z2 
dz RS RS
t = ≈ 1 + dz = 1 + √ dz
z1 1 − r
RS
z1 r z1 b2 + z 2

z 22 + b2 + z 2
= z 2 + |z 1 | + R S ln  , (8.95)
z 12 + b2 − |z 1 |
8.6 The Shapiro Experiment 227

Fig. 8.4 The Shapiro

experiment. General
relativity predicts that light
travelling from Mercury to
the Earth will be delayed due
to the effect of the Sun’s
gravity field on the speed of
light. This effect has been
measured by Shapiro et al.
[1] and is therefore called the
Shapiro effect

where RS is the Schwarzschild radius of the Sun.

The deceleration is greatest when Earth and Mercury (where the light is reflected)
are on nearly opposite sides of the Sun.The impact parameter b is then small. A
series expansion to the lowest order of b z gives

4|z 1 |z 2
t = z 2 + |z 1 | + R S ln . (8.96)
b2
The last term represents the extra travelling time due to the effect of the Sun’s
gravity field on the speed of light. The main effect is slowing down of the velocity
of light farther down in the gravitational field of the Sun. (Also the path is a little
longer because of the bending due to gravity, but the delay due to this is smaller
than the velocity effect and has been neglected in the calculation). Let us insert the
magnitudes of the quantities in Eq. (8.96) and calculate the magnitude of the Shapiro
effect.
R S = the Schwarzschild radius of the Sun = ∼ 3 km
|z 1 | = the radius of Earth’s orbit = 15 × 1010 m
z 2 = the radius of Mercury’s orbit = 5.8 × 1010 m
b = R = 7 × 108 m
give a delay of 1.1 × 10−4 s. In addition to this one must also take into account the
effect of the Earth‘s atmosphere upon the travelling time of the light.
228 8 Schwarzschild Spacetime

8.7 Particle Trajectories in Schwarzschild 3-Space

The Lagrange function of a free particle in the Schwarzschild spacetime is

1
L= gμν Ẋ μ Ẋ ν
2
 1 2
1 Rs 2 ṙ 1 1
=− 1− t˙ + 2 Rs + r 2 θ̇ 2 + r 2 sin2 θ φ̇ 2 . (8.97)
2 r 1− r 2 2

Since t is a cyclic coordinate


∂L Rs
− pt = − = 1− t˙ = constant = E, (8.98)
∂ t˙ r

where E is the particle’s energy per unit rest mass as measured by an observer “far
away” (r  Rs ). Also φ is a cyclic coordinate so that

∂L
pφ = = r 2 sin2 θ φ̇ = constant, (8.99)
∂ φ̇

where pφ is the particle’s orbital angular momentum per unit rest mass, with units
so that c = 1 the angular velocity φ̇ has dimension length−1 , and pφ has dimension
length.
−
→
In the present case the 4-velocity identity U 2 = gμν Ẋ μ Ẋ ν = −1 takes the form

Rs 2 ṙ 2
− 1− t˙ + + r 2 θ̇ 2 + r 2 sin2 θ φ̇ 2 = −1, (8.100)
r 1 − Rrs

which on substitution for t˙ =

pφ
E
1− Rrs
and φ̇ = r 2 sin2 θ
becomes

E2 ṙ 2 pφ2
− + + r 2 θ̇ 2 + = −1. (8.101)
1− Rs
r
1− Rs
r
r 2 sin2 θ

Now, referring back to the Lagrange equation


d ∂L ∂L
− =0 (8.102)
dτ ∂ Ẋ μ ∂ Xμ

we get

pφ2 cos θ
(r 2 θ̇ )· = r 2 sin θ cos θ φ̇ 2 = . (8.103)
r 2 sin3 θ
8.7 Particle Trajectories in Schwarzschild 3-Space 229

Multiplying this by r 2 θ̇ leads to

cos θ θ̇ 2
(r 2 θ̇ )(r 2 θ̇)· = p , (8.104)
sin3 θ φ
which, on integration, gives
 p 2
φ
(r 2 θ̇ )2 = k − , (8.105)
sin θ
where k is the constant of integration.
Because of the spherical geometry we are free to choose a coordinate system such
that the particle moves in the equatorial plane and along the equator at a given time
t = 0. That is θ = π/2 and θ̇ = 0 at time t = 0. This determines the constant of
integration, giving k = pφ2 such that

1
(r 2 θ̇)2 = pφ2 1 − . (8.106)
sin2 θ

The right-hand side is negative for all θ = π/2. It follows that the particle cannot
deviate from its original (equatorial) trajectory. Also, since this particular choice of
trajectory was arbitrary, we can conclude, quite generally, that any motion of free
particles in a spherically symmetric gravitational field is planar.

8.7.1 Motion in the Equatorial Plane

We now consider motion in the equatorial plane. With θ = π/2 Eq. (8.101) reduces
to

E2 ṙ 2 pφ2
− + + = −1, (8.107)
1− RS
r
1− RS
r
r2

that is
  
RS pφ2
ṙ = E − 1 −
2 2
1+ 2 , (8.108)
r r

or

RS pφ2 R S pφ2
ṙ 2 = E 2 − 1 + − 2 + . (8.109)
r r r3
230 8 Schwarzschild Spacetime

A particle falling from rest infinitely far from the mass distribution, ṙ (∞) = 0,
has E = 1, i.e. its energy at the starting point is equal to it rest mass (when the units
are chosen so that c = 1). Hence the energy equation takes the form

1 2
ṙ + VR = 0 (8.110)
2
with a relativistic effective potential

GM pφ2 GM pφ2
VR = − + 2− , (8.111)
r 2r r3
where we have used that R S = 2 GM. Note that the Newtonian “mechanical energy”
of a free particle falling from rest at the zero level of the effective potential vanishes,
since the rest mass energy is not included in this energy. It is, however, included in
the total relativistic energy E.
The Newtonian potential V N is

GM pφ2
VN = − + 2. (8.112)
r 2r
Hence the last term in Eq. (8.111) is a relativistic effect. The potential (8.112) is
plotted in Fig. 8.5.
The Newtonian potential has a centrifugal barrier preventing a particle with pφ =
0 to arrive at the origin, r = 0.
The relativistic potential (8.111) is compared to the Newtonian potential in
Fig. 8.6.
The centrifugal barrier is not infinitely high according to the theory of relativity
as it is in Newton’s theory. This means that relativistic gravity is stronger than
Newtonian gravity. The reason is that the increase of the moving particle’s kinetic
energy gives it a larger mass, and hence, increases the gravity.

Fig. 8.5 Newtonian

centrifugal barrier.
Newtonian potential as
function of the radius. Note
the centrifugal barrier. Due
to this, particles with pφ = 0
cannot arrive at r = 0
8.7 Particle Trajectories in Schwarzschild 3-Space 231

Fig. 8.6 Relativistic

gravitational potential
outside a spherical body. V R
and VN plotted as a function
of r. When relativistic effects
are included, there is no
longer an infinitely high
potential barrier, and a
particle with pφ = 0 can fall
down to r = 0

An orbit equation is one which connects r and φ. For motion in the equatorial
plane for weak fields we have

dφ pφ • d pφ d
= , ≡ = . (8.113)
dt mr 2 dt mr 2 dφ

Introducing the new radial coordinate u ≡ 1/r we get

du 1 dr 1 mr 2 dr m
=− 2 =− 2 = − ṙ . (8.114)
dφ r dφ r pφ dt pφ

Hence,

pφ du
ṙ = − . (8.115)
m dφ

Substitution from Eq. (8.115) for ṙ in the energy equation yields the orbit equation
 2  
du 1 E2
+ (1 − 2GMu) u + 2 2
= . (8.116)
dφ pφ pφ2

Differentiating this, we find

d2 u GM RS 3
2
+ u = 2 + 3GMu 2 = 2
+ RS u2. (8.117)
dφ pφ 2 pφ 2

The last term on the right-hand side is a relativistic correction term.

232 8 Schwarzschild Spacetime

8.8 Classical Tests of Einstein’s General Theory

of Relativity

8.8.1 The Hafele–Keating Experiment

In 1971 Hafele and Keating measured the difference in time shown on moving and
stationary atomic clocks at different height in a gravitational field [2]. This was
performed by flying around the Earth in the East–West direction, comparing the time
on the clock in the plane with the time on a clock on the ground.
The proper time interval measured on a clock moving with a velocity vi = dx i /dt
in an arbitrary coordinate system with metric tensor gμν is

 g 1  1
μν μ ν 2 vi v2 2
dτ = − 2 d x d x = −g00 − 2gi0 − 2 dt, v2 = gi j vi v j . (8.118)
c c c

For a diagonal metric tensor, gi0 = 0, we get

 1
v2 2
dτ = −g00 − 2 dt, v2 = gii (vi )2 . (8.119)
c

We now look at an idealized situation where a plane flies at constant altitude h

and with constant speed along the equator,
 1
RS v2 2
dτ = 1 − − 2 dt, r = R + h. (8.120)
r c

where R is the radius of the Earth. To lowest order in R S /r and v2 /c2 , we get

RS 1 v2
dτ = 1 − − dt. (8.121)
2r 2 c2

The speed of the moving clock is

v = (R + h)
+ u, (8.122)

where
is the angular velocity of the Earth and u is the speed of the plane. A series
expansion and use of the expression (8.122) for v give

GM 1 R 2
2 gh 2R
u + u 2 GM
τ = 1 − 2
− 2
+ 2
− 2
t, g = 2 − R
2 .
Rc 2 c c 2c R
(8.123)

u > 0 when flying in the direction of the Earth’s rotation, i.e. eastwards.
8.8 Classical Tests of Einstein’s General Theory of Relativity 233

Let us find the velocity of a clock with maximal rate of ageing. Differentiating
τ with respect to u gives

dτ R
+ u
=− t. (8.124)
du c2
Putting the derivative equal to zero shows that the maximal rate of ageing happens
for a clock moving along the equator on the Earth with a velocity u = −R
. This may
be understood as a consequence of the fact that this clock is at rest in a non-rotating
reference frame.
For the clock that was left on the airport (stationary, h = u = 0) we get

GM 1 R 2
2
τ0 = 1 − − t. (8.125)
Rc2 2 c2

To lowest order, the relative difference in travel time is

τ − τ0 ∼ gh 2R
u + u 2
k= = 2 − . (8.126)
τ0 c 2c2

Inserting approximate values,

g = 10 m/s2 , h = 10 km, R = 6400 km,
= 7.27 × 10−5 rad/s, u =
±250 m/s gives a
Travel time: τ0 = 1.2 × 105 s (a little over 24 h);
For travelling eastwards: ke = −1.0 × 10−12 ;
For travelling westwards: kw = 2.1 × 10−12 .
Hence, (τ − τ0 )e = −1.2 × 10−7 s ≈ −120 ns and (τ − τ0 )w = 2.5 ×
−7
10 s ≈ 250 ns.
These relativistic predictions were verified with about 15% accuracy in the
experiment.

8.8.2 Mercury’s Perihelion Precession

We shall now calculate Mercury’s perihelion shift. The point of departure is the orbit
Eq. (8.117). This will here be slightly generalized to

d2 u RS
+u = + ku 2 , (8.127)
dφ 2 2 pφ2

where k is a theory-dependent or situation-dependent constant. The general theory

of relativity gives k = (3/2)R S , and Newton’s theory of gravitation gives k = 0.
234 8 Schwarzschild Spacetime

Here pφ = r 2 dφ/dτ . For light dτ → 0 and pφ → ∞. Hence the orbit equation

for light reduces to

d2 u
+ u = ku 2 . (8.128)
dφ 2

The orbit equation for a material particle has a circular solution with an inverse
radius fulfilling

RS
u0 = + ku 20 . (8.129)
2 pφ2

The corresponding Newtonian radius is

RS
u 0N = . (8.130)
2 pφ2

We shall now apply the procedure used in stability analysis to calculate the per-
ihelion precession of Mercury. The equilibrium solution will be perturbed, and the
equation of motion of the perturbation is calculated to 1. order in the perturbation.
This equation will tell whether the equilibrium solution is stable or not.
Hence, the circular motion is perturbed so that u = u 0 + u 1 with u 1  u 0 . To 1.
order in u 1 , we have

d2 u 1 RS
+ u0 + u1 = + ku 20 + 2ku 0 u 1 . (8.131)
dφ 2 2 pφ2

Using the equation for the circular solution we get the equation of motion for the
perturbation

d2 u 1
+ (1 − 2ku 0 )u 1 = 0. (8.132)
dφ 2

If 1−ku 0 > 0, this is the equation of harmonic oscillations, and the circular motion
is stable. If 1 − ku 0 < 0, the solution of the equation is exponential functions, and
the equilibrium solution is unstable. In the present case we assume that k is so small
that the solution is stable. In this case, and with the initial condition u 1 (0) = eu 0 ,
where e is an integration constant, the solution of this equation can be written

u 1 = u 0 e cos( f φ), f = 1 − 2ku 0 . (8.133)

The constant e is called the eccentricity of the orbit and tells how elongated it is.
Using that
8.8 Classical Tests of Einstein’s General Theory of Relativity 235

u = u 0 + u 1 = u 0 [1 + e cos( f φ)], (8.134)

we have
r0
r= . (8.135)
1 + e cos( f φ)

In the Newtonian case with k = 0 we get f = 1, and then this expression describes
a (non-precessing) ellipse. However in the relativistic case k > 0 and f < 1. Then
φ has to increase by 2π/ f > 2π in order that r shall return to the initial value. The
precession angle per orbit is
 
1 1
φ = 2π − 1 = 2π √ − 1 ≈ 2π ku 0 . (8.136)
f 1 − 2ku 0

Inserting the Newtonian value of u 0 we have

π k RS
φ = . (8.137)
pφ2

On the other hand, the general relativistic value k = (3/2)R S leads to

 2
3 RS
φ = π . (8.138)
2 pφ

The angular momentum per unit mass can be expressed in terms of the period by
means of Kepler’s 2. law. This law says that the planet has a constant areal velocity,
dA/dt, where dA is the area swept out by the radius vector from the Sun to the planet
during a time dt. In a small time dt the planet sweeps out a small triangle with base
line r and height r dφ and area dA = (1/2)r 2 dφ, and so the constant areal velocity
is
dA 1 dφ
= r2 . (8.139)
dt 2 dt
The area enclosed by the elliptical orbit is πab where a and b are the semi major
and semi minor axes of the ellipse. Hence integrating over one period, T, we get

1 2 dφ 1 
r T = pφ T = πab = πa 2 1 − e2 . (8.140)
2 dt 2
Hence the angular per unit mass can be written
√
2πa 2 1 − e2
pφ = . (8.141)
T
236 8 Schwarzschild Spacetime

Inserting this in the general formula for the precession angle per orbit gives

RS T 2k
φ =  . (8.142)
4πa 4 1 − e2

According to Kepler’s 3. law the square of the period is proportional to the cube
of the semi major axis,

4π 2 a 3 4π 2 a 3 8π 2 a 3
T2 =  ≈ = . (8.143)
G MSun + m Mercury GM RS

Inserting this into Eq. (8.142) gives

2π k
φ =  . (8.144)
a 1 − e2

Finally, substituting the general relativistic value of k leads to

3π R S
φ =  . (8.145)
a 1 − e2

Inserting the Schwarzschild radius of the Sun, R S = 3 km, the semi major axis
of Mercury’s orbit, a = 5.8 × 107 km and the eccentricity e = 0.2, and using that
the period of Mercury’s orbital motion is 88 days, give a precession angle 43 arc
seconds per hundred years.
This solved an old problem, namely that observations showed that Mercury’s
elliptical orbit precesses by 575 arc seconds per hundred years, while only 532 arc
seconds per hundred years could be accounted for by gravitational forces from the
other planets in the solar system. So there was a discrepancy of 43 arc seconds per
hundred years between observations and the Newtonian prediction.

8.8.3 Deflection of Light

The orbit Eq. (8.128) for light, i.e. for a free particle with mass m = 0, reduces to

d2 u
+ u = ku 2 . (8.146)
dφ 2

If light is not deflected, it will follow the straight line

b
cos φ0 = = bu 0 , (8.147)
r0
8.8 Classical Tests of Einstein’s General Theory of Relativity 237

Fig. 8.7 Deflection of light. Light travelling close to a massive object is deflected

where b is the impact parameter of the path. This represents the horizontal dashed
line in Fig. 8.7. The zeroth order solution (8.147) fulfils

d2 u 0
+ u 0 = 0. (8.148)
dφ 2

Hence it is a solution of (8.146) with k = 0.

The perturbed solution is

u = u 0 + u 1 , |u 1 |  u 0 . (8.149)

Inserting this into the orbit equation gives

d2 u 0 d2 u 1
+ + u 0 + u 1 = ku 20 + 2ku 0 u 1 + ku 21 . (8.150)
dφ 2 dφ 2

The first and third terms at the left-hand side cancel each other due to Eq. (8.148),
and the last term at the right-hand side is small to second order in u1 and will be
neglected. Hence we get

d2 u 1
+ u 1 = ku 20 + 2ku 0 u 1 . (8.151)
dφ 2

The last term at the right-hand side is much smaller than the first and will also be
neglected. Inserting for u0 from (8.147) we then get

d2 u 1 k
+ u 1 = 2 cos2 φ. (8.152)
dφ 2 b

This equation has a particular solution of the form

238 8 Schwarzschild Spacetime

u 1 p = A + B cos2 φ. (8.153)

Inserting this into (8.152) we find

2k k
A= , B=− . (8.154)
3b2 3b2
Hence
k  
u1 p = 2 − cos2 φ (8.155)
3b2
giving

1 cos φ k  
= u = u0 + u1 = + 2 2 − cos2 φ . (8.156)
r b 3b
π
The deflection of the light θ is assumed to be small. We therefore put φ = 2
+ θ
2
where θ  π (see Fig. 8.7). Hence

π θ θ θ
cos φ = cos + = − sin ≈− . (8.157)
2 2 2 2

Thus, the term cos2 φ in (8.155) can be neglected. Furthermore, the deflection of
the light is found by letting r → ∞, i.e. u → 0. Then we get

4k
θ = . (8.158)
3b
For motion in the Schwarzschild spacetime outside the Sun, k = (3/2)R S where
RS is the Schwarzschild radius of the Sun, and for light passing the surface of the
Sun b = R , where R is the actual radius of the Sun. The deflection is then

RS
θ = 2 = 1.75 . (8.159)
R

This general relativistic prediction was verified in two British expeditions utilizing
a solar eclipse in 1919.

8.9 The Reissner–Nordström Spacetime

We shall now solve Einstein’s field equations for a static spherically symmetric space-
time with a radial electric field outside a body with mass M and charge Q. According
to Eqs. (7.81)–(7.84) the non-vanishing components of the energy–momentum tensor
8.9 The Reissner–Nordström Spacetime 239

in an orthonormal basis are

Q2 1 Q2 1
Ttˆtˆ = Tθ̂ θ̂ = Tφ̂ φ̂ = , Tr̂ r̂ = − . (8.160)
32π ε0 r
2 4 32π 2 ε0 r 4

From Eq. (8.15) we get by adding the tˆtˆ− and r̂ r̂ −field equations and integrating

α(r ) = −β(r ). (8.161)

Inserting this into the tˆtˆ field equation we get

1  
− 2β 
R 2Q G Q2
r 1 − e = , R 2
= . (8.162)
r2 r4 Q
4π ε0 c4

Integrating this equation and determining the integration constant by the require-
ment that the solution shall reduce to the Schwarzschild metric with vanishing charge,
we obtain

RS R 2Q 2GM
e−β = 1 − + 2 , RS = . (8.163)
r r c2
Hence, the line element of the spacetime outside a massive changed body takes
the form
 
RS R 2Q 2 2 dr 2
ds = − 1 −
2
+ 2 c dt + R 2Q
+ r 2 d
2 . (8.164)
r r 1− RS
+ r r2

This is the Reissner–Nordström solution. There are two coordinate singularities

at
 2
RS RS
r± = ± − R 2Q . (8.165)
2 2

The exterior singularity is a horizon for an observer outside the mass and charge
distribution.
The maximum allowed charge of the body is given by R Q = R S /2 which leads
to

Q max = 2 π ε0 GM. (8.166)

One may note that the “Reissner–Nordström length” as given in Eq. (8.162),
corresponding to the elementary charge, is
240 8 Schwarzschild Spacetime

e G
Re = 2 . (8.167)
2c π ε0

Inserting the values of the constants gives Re = 1.38 × 10−35 m, which is a little
smaller than the Planck length.

Exercises

8.1. Non-relativistic Kepler motion

In the first part of this exercise we will consider the gravitational potential at a distance
r from the Sun, V (r ) = −GM/r , where M is the mass of the Sun.
(a) Write down the classical Lagrangian in spherical coordinates (r, θ, φ) for a
planet with mass m moving in this field. The Sun is assumed to be stationary.
What is the physical interpretation of the canonical momenta pφ = ?
How is it possible, by just looking at the Lagrangian, to state that pφ is a
constant of motion?
Find the Euler equation for ϑ and show that it can be written into the form

d 2
mr θ̇ +
4 2
= 0. (8.168)
dt m sin2 θ

Based on the above equation, show that the planet moves in a plane by
choosing a direction of the z-axis so that at a given time, t = 0, we have that
θ = π/2 and θ̇ = 0.
(b) Write down the Euler equation for r and use this equation to find u = 1/r as a
function of φ. Show that the orbits that describe bound states are elliptic. Find
the period T 0 for a circular orbit in terms of the radius R of the circle.
(c) If the Sun is not entirely spherical, but rather a little deformed (i.e. more flat
near the poles), the gravitational field in the plane where the Sun has its greatest
extension will be modified into
GM S
V (r ) = − − 3, (8.169)
r r
where S is a small constant. We now assume that the motion of the planet takes
place in the plane where the expression of is V (r) valid. Show that a circular
motion is still possible. What is the period T now, expressed by the radius R?
We now assume that the motion deviates slightly from a pure circular orbit, that
is u = R1 + u 1 , where u 1  R1 . Show that u1 varies periodically around the orbit,

u 1 = k sin( f φ). (8.170)

Exercises 241

(d) Find f and show that the path rotates in space. What is the size of the angle φ
that the planetary orbit rotates per round trip?
The constant S can be written as S = 21 J2 GMRSun 2
where J 2 is a parameter
describing the quadrupole moment and RSun is the radius of the Sun. Observational
data indicate that J2 3 · 10−5 . Calculate how large the rotation φ of the orbit of
Mercury this can cause. Is this sufficient to explain the discrepancy between the
observed perihelion motion of Mercury and that predicted by Newtonian theory?
8.2. The Schwarzschild solution in isotropic coordinates

(a) We introduce a new radial coordinate ρ so that the Schwarzschild metric (with
units so that c = 1),

 
RS R S −1 2
ds = − 1 −
2
dt + 1 −
2
dr + r 2 d
2 , (8.171)
r r

gets the following form


RR
ds = − 1 −
2
dt 2 + f 2 (ρ)(dρ 2 + ρ 2 d
2 ), (8.172)
r (ρ)

where d
2 = dθ 2 + sin2 θ dφ 2 . Find the functions r (ρ) and f (ρ), and write down
the explicit expression of the line element with ρ as the radial coordinate.
(b) What is the value of ρ at the Schwarzschild horizon r = R S and at the origin,
r = 0? The Schwarzschild coordinates t and r interchange their roles as r < R S .
What is the behaviour of ρ inside the horizon?

8.3. Proper radial distance in the external Schwarzschild space

Calculate the proper radial distance from a coordinate position r to the horizon RS
in the external Schwarzschild space.
8.4. The Schwarzschild–de Sitter metric
The Einstein equations for empty space with a cosmological constant Λ are

1
Rμν − Rgμν + gμν = 0. (8.173)
2
(a) Use curvature coordinates and solve the Einstein field equations with a cos-
mological constant for a static spacetime with a spherically symmetric 3-space
outside a spherical body with mass M.
(b) The solution of Einsteins field equations with a cosmological constant in glob-
ally empty space, i.e. with√M = 0, is called the de Sitter spacetime. Introduce
a de Sitter radius R = 3/. Give a physical interpretation of this radius
242 8 Schwarzschild Spacetime

and calculate how large it is in a universe where the value of the cosmological
constant corresponds to a density of LIVE equal to the average density of the
masse and vacuum energy of the universe,  = 10−52 m−2 .

8.5. The perihelion precession of Mercury and the cosmological constant

(a) Show that the orbit equation for free particles moving outside a spherically
symmetric body with mass M has the form

d2 u M 
2
+ u = 2 + 3Mu 2 − , (8.174)
dφ L 3L 2 u 3

where u = 1/r , and L is the angular momentum per unit mass for the particle.
(b) Assume that the orbit can be described as a perturbation of a circle and calculate
the precession angle per round trip.
(c) Estimate the contribution to the precession of Mercury’s perihelion from the
cosmological constant if we assume that the value of the cosmological constant
is  ≈ 10−52 m−2 .

8.6. Relativistic time effects and GPS

Calculate the magnitude of the kinematical and gravitational time effects upon the
GPS satellite clocks. Are standard clocks on the GPS satellites going slower or faster
than a standard clock at rest on the surface of the Earth.
To compute the position of an object by means of the GPS system with a precision
of 1 m, the GPS satellite clocks must measure time with a precision of one part in
1013 .
Are the relativistic effects so small that they can be neglected?
8.7. The photon sphere
The photon sphere is defined as a spherical shell made up of light moving horizontally
in the Schwarzschild spacetime. Calculate the radius of the photon sphere.

References

1. Shapiro, I.I., Ingalls, R.P., Smith, W.B., Campbell, D.B., Dyce, R.E., Jurgens, R.B., Pettengill,
G.H.: Fourth test of general relativity—new radar result. Phys. Rev. Lett. 26, 1132–1135 (1971)
2. Hafele, J., Keating, R.: Around the world atomic clocks: predicted relativistic time gains. Science
177, 166–168 (1972)
Chapter 9
The Linear Field Approximation
and Gravitational Waves

Abstract The linear field approximation of Einstein’s field equations is presented.

The solutions of these equations inside and outside a rotating spherical shell are
deduced. It is shown that inertial dragging is a consequence of the general theory of
relativity. The equations of gravitomagnetism are deduced. Gravitational waves are
described and related to their detection by the LIGO-detectors.

9.1 The Linear Field Approximation

In this chapter we shall describe weak gravitational fields, i.e. fields far from a black
hole, meaning that r  R S , where R S is the Schwarzschild radius of the black
hole. Normalizing the gravitational potential, φ, to zero far from the black hole, this
means that |φ|  c2 , and that the curvature of spacetime is small. It is then possible
to introduce a coordinate system such that the metric deviates very little from the
Minkowski metric. Then it will be a good approximation in order to linearize the
field equations.
We now assume that the gravitational field is weak and introduce a near-Cartesian
coordinate system. The components of the metric tensor can then be written as
gμν = ημν + h μν where ημν is the Minkowski metric, and h μν  1. Also the
derivatives h μν,λ are small.
Einstein’s field equations are

1
Rμν − gμν R = κ Tμν , (9.1)
2
β α
Here R ≡ Rβ is the Ricci curvature scalar and Rμν = Rμαν the components of
α
the Ricci curvature tensor, where Rμαν are the components of the Riemann curvature
tensor. According to Eq. (6.17) they are given by the expression
α α α α λ α λ
Rμβν = μν,β − μβ,ν + λβ μν − λν μβ (9.2)

© Springer Nature Switzerland AG 2020 243

Ø. Grøn, Introduction to Einstein’s Theory of Relativity,
Undergraduate Texts in Physics, https://doi.org/10.1007/978-3-030-43862-3_9
244 9 The Linear Field Approximation and Gravitational Waves

α
in coordinate basis. Here μν are the Christoffel symbols. They are calculated from
the expression (5.65),

α 1 αβ  
μν = g gβμ,ν + gβν,μ − gμν,β . (9.3)
2
In the linear approximation we will only calculate to first order in the metric
perturbation h μν . Then the expression for the Christoffels symbols becomes

α 1 αβ  
μν = η h βμ,ν + h βν,μ − h μν,β . (9.4)
2
Inserting this into Eq. (9.2) gives

α 1 αγ  
Rμβν = η h γ ν,μβ + h μβ,γ ν − h μν,γβ − h γβ,μν . (9.5)
2
Calculating the Ricci tensor by contracting the 1. and 3. index we find

1 α 
Rμν = h μα,ν + h ανα,μ − h αμν,α − h ,μν , (9.6)
2

where h ≡ h αα = ηαβ h αβ . Hence the Ricci scalar is

R = Rμμ = h αβ ,αβ −h,αα . (9.7)

Then the Einstein’s field equations in the linear approximation take the form
αβ β
h αμα,ν + h ανα,μ − h αμν,α − h ,μν − ημν (h αβ, − h ,β ) = 16π GTμν . (9.8)

This equation can be simplified by introducing

1
h̄ μν = h μν − ημν h. (9.9)
2

We assume that h̄ satisfies the so-called Lorenz gauge condition

h̄ αμ,α = 0. (9.10)

Then the field equations take the form

h̄ αμν,α = −2κ Tμν . (9.11)

Introducing the d’Alembert’s wave operator in Minkowski spacetime

9.1 The Linear Field Approximation 245

 = ηαβ ∂α ∂β = −∂ 2 /∂t 2 + ∂ 2 /∂ x 2 + ∂ 2 /∂ y 2 + ∂ 2 /∂z 2 = −∂ 2 /∂t 2 + ∇ 2 ,

(9.12)

Einstein’s field equations in the linear field approximation can be written as

h̄ μν = −2κ Tμν . (9.13)

Hence, the linearized field equations for empty space take the form

h̄ μν = 0. (9.14)

This is d’Alembert’s wave equation for waves moving with the velocity of light.
It here describes gravitational waves. Hence, it follows from Einstein’s theory that
gravitational waves move with the velocity of light. This reveals a deep relation-
ship between electromagnetism and gravity which is not yet fully understood. The
explanation may be hidden in the Kaluza–Klein theory (see the Appendix).
In the case of a static spacetime in matter with density ρ the tt-component of
Einstein’s field equations takes the form

∇ 2 h̄ tt = −2κρ. (9.15)

Comparing with the Newtonian gravitational field Eq. (1.29) we get

h̄ tt = −4φ/c2 . (9.16)

where φ is the Newtonian gravitational potential.

In order to find the line-element in the linear field approximation we need the
reverse of Eq. (9.9). Taking the trace of h̄ μν and utilizing that ημμ = δμμ = 4 we get
h̄ = −h. Hence,

1
h μν = h̄ μν − ημν h̄. (9.17)
2
Inserting Eq. (9.16) gives

h tt = h ii = (1/2)h̄ tt = −2φ. (9.18)

Hence the line element takes the form

   
2φ 2 2 2φ  2 
ds = − 1 + 2 c dt + 1 − 2 dx + dy 2 + dz 2 ,
2
(9.19)
c c
or
   
2φ 2φ  
ds 2 = − 1 + 2 c2 dt 2 + 1 − 2 dr 2 + r 2 d2 . (9.20)
c c
246 9 The Linear Field Approximation and Gravitational Waves

9.2 Solutions of the Linearized Field Equations

9.2.1 The Gravitational Potential of a Point Mass

The space time outside a spherical mass distribution with given mass does not depend
upon the radius of the mass distribution. Hence the spacetime is the same as that
outside a particle. We choose to consider spacetime outside a particle in order to
simplify the calculation.
A point mass with the mass m is situated in the origin of a Cartesian coordinate
system. Its energy–momenta tensor has only one non-vanishing component,

T00 = mδ(
r ), (9.21)

where δ(
r ) is the 3-dimensional δ-function.
In this case Einstein’s field equation (9.13) takes the form

∇ 2 h̄ 00 = −2κmδ(
r ). (9.22)

Using that

r )d3r = 1,
δ( (9.23)

and integrating Eq. (9.22) gives


∇ 2 h̄ 00 d3r = −2κm. (9.24)

Utilizing Gauss’s integral theorem we can transform the left-hand side to an

integral over the surface of a spherical surface with radius r and get

dh̄ 00
∇ h̄ 00 · d S = 4πr 2 = −2κm. (9.25)
dr

where r = x 2 + y 2 + z 2 . Hence,

dh̄ 00 4Gm
=− 2 2. (9.26)
dr c r

Integration with the boundary condition h̄ 00 = 0 gives

lim r →∞

4Gm
h̄ 00 = . (9.27)
c2 r
9.2 Solutions of the Linearized Field Equations 247

Hence,

4Gm
h̄ = h̄ μμ = − . (9.28)
c2 r
Inserting Eqs. (9.27) and (9.28) in Eq. (9.17) gives

2G M RS
h 00 = h ii = 2
= . (9.29)
c r r
Hence the line element of the spacetime outside the particle is
   
RS 2 2 RS  2 
ds 2 = − 1 − c dt + 1 + dx + dy 2 + dz 2 . (9.30)
r r

In terms of spherical coordinates the line element takes the form

   
RS 2 2 RS  2 
ds = − 1 −
2
c dt + 1 + dr + r 2 d2 . (9.31)
r r

This is the same as the linearized line element, (S8.50), of the Schwarzschild
spacetime as expressed in isotropic coordinates.

9.2.2 Spacetime Inside and Outside a Rotating Spherical

Shell

We shall solve the linearized field equations, (9.13), inside and outside a rotating
spherical shell with radius R and mass M consisting of dust particles. The energy–
momentum tensor of the shell is
M
Tμν = ρu μ u ν , ρ = δ(r − R), u μ = (−1, −Rω sin θ sin φ, Rω sin θ cos φ, 0).
4π R 2
(9.32)

where ω is the angular velocity of the shell. In this case Einstein’s field equations
(9.13) take the form

2RS
∇ 2 h̄ 00 = − 2 δ(r − R), ∇ 2 h̄ ii = 0
R
2RS ω 2RS ω .
∇ 2 h̄ 0x = sin θ sin φδ(r − R), ∇ 2 h̄ 0y = sin θ cos φδ(r − R)
R R
(9.33)

These equations are integrated in the same manner as Eq. (9.22). For h̄ 00 we get
248 9 The Linear Field Approximation and Gravitational Waves

dh̄ 00 −2RS /r 2 , r ≥ R
= . (9.34)
dr 0, r<R

Integrating once more and demanding that h̄ 00 is continuous at the shell give

2RS /r, r ≥ R
h̄ 00 = . (9.35)
2RS /R, r ≤ R

The second Eq. (9.33) and the requirement that there shall be Minkowski metric
infinitely far from the shell gives h̄ ii = 0.
Because of the spherical symmetry the two last equations in (9.33) will have the
same solution, so it is sufficient to solve one of then, say the last one. Due to the form
of the right-hand side we assume that h̄ 0y has the form

h̄ 0y = f (r ) sin θ cos φ. (9.36)

This gives

1 ∂ 2 ∂ 1 ∂2 1 ∂ ∂
∇ 2 h̄ 0y = r + + sin θ f (r ) sin θ cos φ
r ∂r ∂r  θ ∂φ r 2 sin2 θ ∂θ ∂θ
2 2 2
r 2 sin ,
2 2
= f  + f  − 2 f sin θ cos φ
r r
(9.37)

where f  = d f /dr . Inserting this into the last Eq. (9.33) leads to

2  2 2RS ω
f  + f − 2 f =− δ(r − R). (9.38)
r r R
This equation may be written as


1  2  2RS ω
r f =− δ(r − R). (9.39)
r2 R

Using that δ(r − T ) = −δ(R − r ) we have


1  2  2RS ω
r f = δ(R − r ). (9.40)
r2 R

Integration gives

1  2  0, r>R
r f = 2RS ω . (9.41)
r2 R
, r<R
9.2 Solutions of the Linearized Field Equations 249

Hence,

 2  0, r>R
r f = 2RS ω 2 . (9.42)
R
r , r<R

Integrating once more gives


K out , r>R
r f =
2
2RS ω 3 . (9.43)
3R
r + K in r < R
,

Hence,

K out /r 2 , r>R
f = 2RS ω . (9.44)
3R
r + K in
r2
, r <R

The requirement that the metric is well defined at r = 0 implies that K in = 0.

Furthermore the requirement that the metric is continuous at the shell then implies
that K out = 2RS R 2 ω/3. Hence the solution of Eq. (9.33) outside and inside the shell
is
  
2RS ω R 2
,r >R
f (r ) = 2RS ω r r
3 . (9.45)
3 R
, r<R

Inserting the expressions for h̄ μν from Eqs. (9.35), (9.36) and (9.45) into Eq. (9.17)
gives the line element
⎧    
⎪
⎪ RS 2 2 RS  2  4R ω
S
⎨ − 1 − R c dt + 1 + R
⎪ dr + r 2 d2 − r 2 sin θdφdt, r < R
2 3R
ds =     .
⎪
⎪
⎪ RS 2 2 RS  2  4R R 2 ω
S
⎩− 1 − c dt + 1 + dr + r 2 d2 − sin θdφdt, r > R
r r 3r
(9.46)

The angular momentum of the shell is

S = (1/3)RS R 2 ω. (9.47)

Hence the line-element (9.46) can be written

⎧    
⎪ RS 2 2 RS  2  4S
⎪
⎨ − 1 − c dt + 1 + dr + r 2 d2 − 3 r 2 sin θdφdt, r < R
R R R
ds 2 =     .
⎪
⎪ RS 2 2 RS  2  4S
⎩ − 1− c dt + 1 + dr + r 2 d2 − sin θdφdt, r > R
r r r
(9.48)
250 9 The Linear Field Approximation and Gravitational Waves

Let us see how the shell modifies space inside it.

9.3 Inertial Dragging

We shall first find a formula for the inertial dragging effect inside the shell. Consider
a free observer moving in the equator plane, θ = π/2, in the space described by the
line-element (9.46) for r < R. The Lagrange function of the observer is
     
1 RS 2 2 1 RS 2 1 RS 2 2 2RS ω 2
L=− 1− c t˙ + 1+ ṙ + 1+ r φ̇ − r φ̇ t˙.
2 R 2 R 2 R 3R
(9.49)

The momentum pφ conjugates to the cyclic coordinate φ,

 
∂L RS 2 2RS ω 2
pφ = = 1+ r φ̇ − r t˙. (9.50)
∂ φ̇ R 3R

An observer with pφ = 0 is called a zero-angular-momentum-observer, ZAMO.

The coordinate angular velocity of the ZAMO is

dφ φ̇ 2 RS
IN = = = ω. (9.51)
dt t˙ 3 R + RS

The metric is time independent. This means that the physical distance between
arbitrary coordinate points is independent of the time. Hence the coordinates are
co-moving with a stiff reference frame. At the origin the dφdt—term of the line-
element (9.46) vanishes, and close to the origin the line element approaches the
linearized line-element of the Schwarzschild spacetime. This means that as seen by
a non-rotating observer at the origin the reference frame of the coordinate system is
non-rotating.
Imagine that the observer at the origin throws out a particle. It is a free particle
with zero angular momentum, a ZAMO. This particle will however not move along
a radial line. It has a constant coordinate angular velocity given by Eq. (9.51) and
moves along a spiral path. The particle is dragged in the same direction at the shell
rotates. This phenomenon has several names. It was originally called the Lense–
Thirring effect because it was first described in a published article by the Austrian
Physicist Lense and Thirring in 1918. Later it has been called inertial dragging.
The reason for the latter name is that a free particle with vanishing angular momen-
tum represents a local inertial frame. Hence inside the shell inertial frames are
dragged around by the rotation of the shell—inertial dragging.
There is a similar effect outside the shell. Calculating the angular velocity of a
ZAMO outside the shell in the same manner as above leads to
9.3 Inertial Dragging 251
 2
2 RS R
IN = ω. (9.52)
3 r + RS r

The dragging angular velocity outside the shell decreases from the value (9.51)
at the shell to zero infinitely far from the shell.
Note that the expressions (9.51) and (9.52) are valid only in the linear field
approximation, i.e. for R >> R S . Corresponding expressions for arbitrarily strong
gravitational fields will be found in the next chapter.

9.4 Gravitoelectromagnetism

In the context of gravitoelectromagnetism the Newtonian gravitational potential

GM RS
φ=− =− (9.53)
r 2r
of a rotating body with mass M is called the “gravito-electric” potential. One also
introduces a “gravitomagnetic” vector potential A with components (including the
velocity of light in this section)

G S j xk
Ai = εi jk 3 , (9.54)
c r

where εi jk is the antisymmetric Levi-Civita symbol, and S j is the j-component of

the angular momentum of the rotating body. In 3-vector notation this takes the form

G S × r
A = . (9.55)
c r3
The angular momentum of the rotating mass distribution is related to the energy–
momentum tensor by

Si = 2 εijk x j j k d3r, (9.56)

where

j k = T 0k /c (9.57)

is the mass current density. Then the line element of the linear field approximation
can be written
   
2φ 2φ 4
ds 2 = − 1 + 2 c2 dt 2 + 1 − 2 δi j dx i dx j − Ai dx i dt. (9.58)
c c c
252 9 The Linear Field Approximation and Gravitational Waves

In terms of the gravitoelectromagnetic potentials the Lorenz gauge condition

(9.10) takes the form

1 ∂φ 1
+ ∇ · A = 0. (9.59)
c ∂t 2
The “gravito-electric” field is

1 ∂ A
EG = −∇φ − , (9.60)
2c ∂t
and the gravitomagnetic field is

1 
BG = ∇ × A. (9.61)
2
It follows from the field Eqs. (9.13) together with Eqs. (9.59)–(9.61) that the
gravito-electric and -magnetic fields satisfy equations of similar form as Maxwell’s
equations for electromagnetic fields,

∇ · E G = −4π Gρ, (9.62)

∇ · BG = 0, (9.63)

∂ BG
∇ × EG = − , (9.64)
∂t
1 ∂ EG 4π G
∇ × BG = + 2 j. (9.65)
c ∂t
2 c
The gravitoelectromagnetic analog of the Lorentz force is
 
F = m EG + 4
v × BG . (9.66)

In the linear field approximation the gravitomagnetic field can be written as a

dipole field,
⎡   ⎤
G S · r r
BG = − ⎣ S − 3 ⎦. (9.67)
2cr 3 r2

If a body with an angular momentum L is in a gravitomagnetic field BG , the field

will cause a torque
9.4 Gravitoelectromagnetism 253

1 
τ = L × BG . (9.68)
2c
The torque is equal to the time derivative of the angular momentum

d L
τ = . (9.69)
dt
The rate of change of angular momentum is

d L 
 × L,
= (9.70)
dt

 is the angular velocity of the inertial dragging. Equations (9.63)–(9.65) give

where 

1 
 =−
 BG . (9.71)
2c
Outside a body, for example the Earth, producing the gravitomagnetic field (9.67),
the dragging angular velocity is
⎡  ⎤
 · r r
S
G ⎣ ⎦.
 =
 S−3 (9.72)
4c2 r 3 r2

At the equatorial plane S is orthogonal to r so S · r = 0 and the equation reduces

to
G 
 =
 S. (9.73)
4c2 r 3
In the case of a rotating body, for example a gyroscope, this dragging field rep-
resents the rate of angular precession of the rotation axis (which was observed in
the Gravity probe B experiment), and in the case of orbital motion it represents the
perihelion precession of the orbit (for example the orbit of the planet Mercury), or
the precession of the orbital plane (which was observed for the LAGEOS satellites).

9.5 Gravitational Waves

We will here consider gravitational waves in the weak field approximation of Ein-
stein’s equations using the Maxwell like equations for the gravitoelectromagnetic
fields.
Inserting Eq. (9.60) into Eq. (9.62) gives
254 9 The Linear Field Approximation and Gravitational Waves

1 ∂ 
−∇ 2 φ − ∇ · A = −4π Gρ. (9.74)
2c ∂t
From the Lorenz gauge condition we have

2 ∂φ
∇ · A = − . (9.75)
c ∂t
Hence
 
1 ∂2
− 2 2 + ∇ φ = φ = 4π Gρ.
2
(9.76)
c ∂t

Furthermore, inserting Eqs. (9.60) and (9.61) into Eq. (9.65) gives

1   1 ∂φ 1 ∂ 2 A 4π G
∇ × ∇ × A = − ∇ − 2 2 + 2 j. (9.77)
2 c ∂t 2c ∂t c
Using the identity
   
∇ × ∇ × A = ∇ ∇ · A − ∇ 2 A (9.78)

we get
   
1 ∂φ 1 1 1 ∂2 4π G 1 4π G
∇ + ∇ · A = − 2 2 + ∇ 2 A + 2 j =  A + 2 j.
c ∂t 2 2 c ∂t c 2 c
(9.79)

Due to the Lorenz gauge condition (9.59) the left-hand side vanishes. Hence we
obtain
8π G 
 A = − j. (9.80)
c2
In empty space Eqs. (9.76) and (9.80) reduce to

φ = 0,  A = 0. (9.81)

These are equations of gravitational waves moving through empty space with the
velocity of light.
9.5 Gravitational Waves 255

9.5.1 What Sort of Gravitational Waves Is Predicted

by Einstein’s Theory?

We shall now investigate the nature of these waves. In order to simplify the treatment
we shall consider plane gravitational waves. There are three types of such gravita-
tional waves, transverse (T), shear (S) and longitudinal (L). We shall here investigate
what type Einstein’s theory predicts the existence of and follow a procedure given
by Eddington in an article published in 1922 [1].
We consider plane gravitational waves moving in the x-direction with velocity v.
Hence h μν are periodic functions of x + vt. Differentiation with respect to x + vt
will be denoted by ’, and the coordinates by (x, y, z, t) = (x1 , x2 , x3 , x4 ). Then the
only non-vanishing derivatives are

∂ 2 gμν  ∂ 2 gμν  ∂ 2 gμν

= h μν , = vh μν , = v 2 h μν . (9.82)
∂ x12 ∂ x1 ∂ x4 ∂ x42

The covariant components of the linearized Riemann tensor are

 
1 ∂ 2 gμν ∂ 2 gαβ ∂ 2 gμα ∂ 2 gνβ
Rμναβ = + − − . (9.83)
2 ∂ xα ∂ xβ ∂ xμ ∂ xν ∂ xβ ∂ xν ∂ xμ ∂ xα

For a spacetime with plane gravitational waves the Riemann tensor has 21 inde-
pendent components, six of them vanishing. The 15 non-vanishing components were
given by Eddington as follows for i and j equal to 1, 2 and 3,

1 2  1 1
Ri4j4 = v h ij , R1ij4 = −R1ji4 = − vh ij , R1i1j = h ij ,
2 2 2 . (9.84)
v  1 1
R14i4 = −v R1i14 = vh 1i − h i4 , R1414 = v 2 h 11 − vh 14 + h 44
2 2 2
The corresponding non-vanishing components of the Einstein tensor are

v    1  1 
G 11 = vh 11 − 2h 14 + h 44 − h 22 − h 33 , G kk = v 2 − 1 h kk , k = 2, 3
2 2 2
1  1  v 
G 44 = h 44 − 2vh 14 − h 11 + h 22 + h 33 , G 1k = vG k4 = vh 1k − h k4 ,
2 2 2
1 2   v   

G 23 = v − 1 h 23 , G 14 = − h 22 + h 33 , (9.85)
2 2
Einstein’s field equations for empty space are G μν = 0, which is a set of second-
order differential equations for h μν . Since h μν are periodic functions, the second
derivatives of the functions are equal to minus the functions themselves. This leads
to field equations

v(vh 11 − 2h 14 ) + h 44 − h 22 − h 33 = 0, h kk = 0, k = 2, 3
256 9 The Linear Field Approximation and Gravitational Waves

h 44 − 2vh 14 − h 11 − h 22 − h 33 = 0,
vh 1k − h k4 = 0, h 23 = 0, v(h 22 + h 33 ) = 0. (9.86)

These equations can be reduced to the following seven conditions,

 
h 22 + h 33 = 0, 1 − v 2 (h 22 , h 33 , h 23 ) = 0,
h 24 = vh 12 , h 34 = vh 13 , v 2 h 11 − 2vh 14 + h 44 = 0. (9.87)

Eddington pointed out that for T-waves h 22 , h 33 , h 23 cannot all vanish. Hence for
these waves Eq. (9.87) implies v = ±1, meaning that according to the general the-
ory of relativity transverse gravitational waves move with a coordinate independent
velocity equal to the velocity of light in empty space. For S- and L-waves h 22 , h 33 , h 23
are zero, and there is no coordinate independent equation determining v. The value
of v found from the three last relationships in (9.87) depends upon the metric com-
ponents and is hence coordinate dependent. Furthermore from (9.84) it is seen that
for S- and L-waves the relationships (9.87) imply that the Riemann curvature tensor
vanishes so that spacetime is flat and the periodic changes of the metric components
are coordinate artifacts.
The conclusion is that the general theory of relativity predicts the existence of
transverse gravitational waves travelling with the velocity of light.

9.5.2 Polarization of the Gravitational Waves

The gravitational waves may be represented by the metric functions

h̄ μν = Aμν cos(kα x α ), (9.88)

where Aμν are the components of a symmetric tensor of rank 2, and kα are the
components of a constant wave vector. Inserting (9.88) into the field Eqs. (9.13)
leads to

kα k α = 0. (9.89)

Hence kα is a null-vector, meaning again that the gravitational waves move with
the velocity of light. An observer with four-velocity U μ would observe the wave to
have an angular frequency

ω = −kα U α . (9.90)

In the co-moving frame of the observer, where U μ = (1, 0, 0, 0), the so-called
transverse traceless gauge condition takes the form
9.5 Gravitational Waves 257

j
h μ0 = 0, h k, j = 0, h ii = 0. (9.91)

The first of these equations tells that only the spatial components of the metric
perturbations are nonzero. The second says that the spatial components are divergence
free, and the third says that they are trace free. It should also be noted that since
h = h μμ = 0 there is no difference between h̄ μν and h μν in this gauge.
We now choose the orientation of the coordinate system such that the gravitational
wave travels along the z-axis. The metric perturbation then takes the form
⎡ ⎤
0 0 0 0
⎢0 hxx hxy 0⎥
h μν =⎢
⎣0
⎥. (9.92)
hxy −h x x 0⎦
0 0 0 0

There are only two free metric functions. This corresponds to the fact that there
are only two different polarizations of gravitational waves according to Einstein’s
theory.

9.6 The Effect of Gravitational Waves upon Matter

We shall investigate physical effects of gravitational waves upon systems they pass.
A gravitational wave is a “curvature wave”. We shall therefore study the physical
effects of gravitational waves by utilizing the Eq. (6.94) of geodesic deviation,

d2 s i
= −R0i j0 s j . (9.93)
dt 2

Here s i are the components of the separation vector between two nearby geodesics.
From the expression (9.5) for the components of the Riemann curvature tensor in the
linear field approximation we get in the transverse traceless gauge,

R0i j0 = (1/2)h ij,00 . (9.94)

Note that the mixed components of the metric perturbations are not equal to the
Kronecker symbols, so the derivatives do not in general vanish. Equation (9.93) now
takes the form

d2 s i 1
= − h ij,00 s j . (9.95)
dt 2 2
Inserting the components of the metric perturbation from Eq. (9.87) we obtain
258 9 The Linear Field Approximation and Gravitational Waves

d2 s x 1 1 d2 s y 1 y 1 d2 s z
= s x h xx,tt + s y h xy,tt , = s x h x,tt − s y h xx,tt , = 0. (9.96)
dt 2 2 2 dt 2 2 2 dt 2
These equations show that only the s x and s y components of the separation vector
between two nearby free particles will be disturbed by a gravitational wave travelling
in the z-direction.
A gravitational antenna is a system able to shape form, and according to Eq. (9.96)
it is only disturbed in directions perpendicular to the wave propagation. Let us use
the above equations to describe what happens to a ring in the xy-plane of free test
particles at rest as a gravitational wave passes in the z-direction. To lowest order we
can then neglect the terms with s y at the right-hand side of the two first Eqs. (9.96)
so that

d2 s x 1 d2 s y 1
= − h xy,tt s x , = − h xx,tt s y . (9.97)
dt 2 2 dt 2 2
Suppose a wave propagating in the z-direction with

h x x = h sin[ω(t − z)], h x y = 0 (9.98)

hits the particles. Let us consider two particles that are hit by the wave. One is at the
origin, and the other has initially the position x = ε cos θ, y = ε sin θ, z = 0. The
initial separation vector has the components s x (0) = ε cos θ, s y (0) = ε sin θ . Then
Eqs. (9.97) and (9.98) give

d2 s x 1 d2 s y 1
= − εhω2 cos θ sin ωt, = εhω2 sin θ sin ωt. (9.99)
dt 2 2 dt 2 2
These equations have the solutions
   
h h
s = ε cos ω 1 + sin ωt , s = ε sin ω 1 − sin ωt ,
x y
(9.100)
2 2

satisfying the initial conditions. It follows from these expressions that

   
(s x )2 (s y )2 h h
+ 2 = 1, a = ε 1 + sin ωt , b = ε 1 − sin ωt . (9.101)
a2 b 2 2

This shows that an originally circular ring is deformed to an oscillating ellip-

tical shape by the gravitational wave, varying between being stretched in the y-
direction and compressed in the x-direction and then stretched in the x-direction
and compressed in the y-direction. This is called the + polarization and is shown in
Fig. 9.1.
We now consider the case when the wave has
9.6 The Effect of Gravitational Waves upon Matter 259

Fig. 9.1 Deformation of a ring of free particles caused by a gravitational wave with + polarization

h x x = 0, h x y = h sin[ω(t − z)]. (9.102)

In this case Eqs. (9.97) and (9.98) give

d2 s x 1 d2 s y 1
2
= − εhω2 sin θ sin ωt, 2
= − εhω2 cos θ sin ωt. (9.103)
dt 2 dt 2
These equations have the solutions

1 1
s x = ε cos θ + εh sin θ sin ωt, s y = ε sin θ + εh cos θ sin ωt. (9.104)
2 2
Rotating the coordinate axes through an angle π/4 we obtain new components of
the separation vector

 1    1  
sx = √ sx − s y , s y = √ sx + s y . (9.105)
2 2

Inserting the expressions (9.104) gives

    
x π h y
 π h
s = ε cos θ + 1 − sin ωt , s = ε cos θ + 1 + sin ωt .
4 2 4 2
(9.106)

It follows from these expressions that

 x  2  y  2    
s s h h
+ = 1, a = ε 1 − sin ωt , b = ε 1 + sin ωt . (9.107)
a2 b2 2 2

In this case the gravitational wave produces elliptical deformations rotated an

angle π/4 relative to those of Fig. 9.1, as shown in Fig. 9.2. This is called the x
polarization.
Since h x x and h x y are independent, Figs. 9.1 and 9.2 show the existence of two
different states of polarization that are oriented at an angle π/4 relative to each other.
260 9 The Linear Field Approximation and Gravitational Waves

Fig. 9.2 Deformation of a ring of free particles caused by a gravitational wave with x polarization

9.7 The LIGO-Detection of Gravitational Waves

As noted in Sect. 9.6 the effect of a gravitational wave, GW, upon two freely falling
nearby particle is to change the distance, L, between them. If the induced change of
distance is L, the fractional change in length h = L/L is called the GW-strain.
This is a dimensional measure of the amplitude of the gravitational wave.
LIGO consists of two detectors, one at the West coast of USA and one at the East
coast. They are separated by about 3000 km. Hence light uses about 10 ms to move
the distance between them. This means that signals of a certain type separated in
time by less than 10 ms can be a sign that a gravitational wave has passed through
the detectors.
The detectors are Michelson interferometers as illustrated in Fig. 9.3. The inter-
ferometer is constructed so that in normal modus the light that has been reflected
from the upper mirror and the right hand mirror is in opposite phase and there is
destructive interference when it meets after the reflections. Hence no light arrives at
the detector in normal modus.
The two mirrors of the Fabry-Pérot cavities are hanged up so that they are free
to move along the 4 km long L’s. When a gravitational wave, say from a system
of black holes with some tens of solar masses a billion light years away that spiral
towards each other and collide, passes the detector, the distances between the two
mirrors of the Fabry–Pérot cavities will get a length change with opposite signs, one
is shortened, and the other gets longer. Hence the phases of the reflected light change,
and a light signal arrives at the detector. In this situation typically the GW-strain is
10−21 . Hence with L = 4 km one has to be able to detect a change of length of the
order 10−18 m.
The shape of the signal has been calculated and is as shown in Fig. 9.4.
The signal has three phases. The first phase is the inspiral when the objects are
outside the innermost stable orbit. In this phase the frequency and the amplitude
increase. It is said to chirp. The second phase is the merger when they are inside and
collide, and the third phase is the ringdown when the object formed by the collision,
vibrates and falls to rest. The shape of the signal in the inspiral phase can be calculated
analytically using post-Newtonian theory, the merger using numerical calculations
and the ringdown using perturbation theory.
The first detection of gravitational waves by LIGO came 14 September 2015 [2].
The most recent was detected by LIGO and Virgo. The detected signals are shown
in Fig. 9.5.
9.7 The LIGO-Detection of Gravitational Waves 261

Fig. 9.3 LIGO gravitational wave detector

Order of magnitude estimates for the amplitudes of the waves emitted at different
phases are given in [3], and a simple, but quantitative analysis has recently been
given by Mathur et al. [4]. The main points when it comes to extracting information
of the physical properties of the source of gravitational waves from the signals in an
interferometric gravitational wave detector such as LIGO when a gravitational wave
is detected, will be presented below.
The LiGO–Virgo-team detected a signal of this type the 14 August 2017.
Figure 9.5 is from this report. An analysis of the signal that lasted for about 0.2 s
showed that they had detected a gravitational wave emitted by a system of two black
holes spiralling towards each other and merging at a distance of 1,8 billion light
years. The frequency increased from 35 to 250 Hz meaning that the number of times
the black holes went around each other per second increased from 17 to 125 during
less than the fifth of a second. The masses of the black holes were detected to be
31M and 25 M , where 1 M is the mass of the Sun. The mass of the final black
hole was 54M . Hence an energy corresponding to 3M was emitted in the form of
gravitational waves. The relative velocity of the black holes increased from 0.32c to
0.57c in this brief time interval.
262 9 The Linear Field Approximation and Gravitational Waves

Fig. 9.4 LIGO-gravitational wave signal. LIGO-signal when a gravitational wave from a system
of two compact objects, for example two black holes, passes the detector. There are three phases,
the inspiral when the objects are outside the innermost stable orbit, the merger when they are inside
and collide and the ringdown when the object formed by the collision vibrates and falls to rest.
(Nobel.org)

9.7.1 Kepler’s Third Law and the Strain of the Detector

For a system of two compact objects with masses M1 and M2 one defines the so-called
chirp mass,
 1/5
(M1 M2 )3
Mch = (9.108)
M

where M = M1 + M2 is the total mass of the system, and the reduced mass is

M1 M2
μ= . (9.109)
M
One also defines the dimensionless mass ratio for the system,

μ M1 M2
η= = . (9.110)
M M2
The expressions for the chirp mass and the total mass can be solved for the masses
of the compact objects, giving
 9.7 The LIGO-Detection of Gravitational Waves

Fig. 9.5 LIGO gravitational wave discovery signal. Discovery signals registered at Hanford, Livingston and Virgo of a gravitational wave from colliding black
holes. Reprinted from [2] by The Author(s) licensed under CC BY 3.0
263
264 9 The Linear Field Approximation and Gravitational Waves
 2  2
M M 5/3 M M 5/3
M1 = + − M 1/3 Mch , M2 = − − M 1/3 Mch ,
2 2 2 2
(9.111)

which requires Mch < 2−6/5 M.

In the inspiral phase there will be a maximal orbital angular frequency, , given
by Kepler’s third law

22 R 3 = c2 RS . (9.112)

when two objects with equal masses are about to merger, where R = R1 + R2 is
the sum of the radial coordinates of the compact objects, and RS = 2G M/c2 is the
Schwarzschild radii of the total mass.
At a distance r from the source of two inspiralling compact objects, the strain due
to a gravitational wave is

4RS R 2 2
h= . (9.113)
c2 r
It follows from the last two equations that

RS2
h=2 . (9.114)
rR
Let us use this formula to make an estimate of the expected magnitude of the strain
for a gravitational wave detector on the Earth. The maximal value of h is obtained
just before the objects merger, when R ≈ 2R S . Then

h max ≈ RS /r. (9.115)

The Schwarzschild radius of the Sun is RS = 3 km = 3 · 10−13 l · y. Hence for

a system of for example black holes with total mass equal to 50 solar masses one
billion light years from the Earth the strain is h = 1.5 · 10−20 . For an interferometer
with 4 km long arms this means that the distance between two mirrors at the ends of
the arms will change by L = h L = 6 · 10−17 m, which is less than the magnitude
of an atomic nucleus.
Combining Eq. (9.114) with Eq. (9.110) and introducing the Schwarzschild radii
R S1 and R S2 of the two objects, the strain can be expressed as

2 RS1 RS2
h= . (9.116)
η rR

Sometimes the strain is expressed by the total mass and the chirp mass. Introducing
the Schwarzschild radius of the total mass, RS = 2G M, and the chirp mass, RSch =
2G Mch /c2 , and using Eq. (9.108) the strain can be expressed as
9.7 The LIGO-Detection of Gravitational Waves 265

1/3 5/3
2 RS RSch
h= . (9.117)
η rR

Using Eq. (9.112) the strain as measured at a distance r far away from the colliding
objects can then be expressed as

5/3 24/3
h = 2/3 RSch . (9.118)
c2/3 ηr

9.7.2 Newtonian Description of a Binary System

We use a coordinate system co-moving with the mass centre of the binary system,
M1 R1 = M2 R2 . Using this together with R = R1 + R2 we get

M2 M1
R1 = R, R2 = R. (9.119)
M M
Furthermore we introduce an angular coordinate, θ = θ1 = θ2 − π . The kinetic
energy of the system is

1 1
EK = M1 v12 + M2 v22 . (9.120)
2 2
where the square of the velocities of the compact objects are
 2
M2
v12 = Ṙ12 + R12 θ̇12 = v2 ,
M
 2
M1
v22 = Ṙ22 + R22 θ̇22 = v2 ,
M
v 2 = Ṙ 2 + R 2 θ̇ 2 . (9.121)

Inserting these expressions into Eq. (9.120) and using Eq. (9.109) gives

1 2
EK = μv . (9.122)
2
The objects move along elliptical paths. In the Newtonian theory of binary systems
it is usual to write the equation of the elliptic orbit as

p = R(1 + e cos θ ), (9.123)

266 9 The Linear Field Approximation and Gravitational Waves

where p is called the semi-latus rectum, and e is the eccentricity of the ellipse, i.e. the
ratio of the semi-minor and semi-major half axis of the ellipse, e = b/a. The specific
angular momentum, L = R 2 θ̇, is a constant of motion. The relation L 2 / p = G M
will be needed below. Differentiation of Eq. (9.123) with respect to time gives

R 2 e sin θ Le R 2 θ̇ L
Ṙ = = sin θ, R θ̇ = = (1 + e cos θ). (9.124)
R(1 + e cos θ) p R p
 
Using these expressions and that p = a 1 − e2 one finds that the kinetic energy
is
 
1 2 1 G M1 M2 G M1 M2
EK = μG M − = − . (9.125)
2 R a R 2a

The potential energy of the system is

G M1 M2
EP = − . (9.126)
R
It follows that the mechanical energy of the system, E = E K + E P , has the value

G M1 M2
E =− , (9.127)
2a
which is constant. Approximating the distance between the objects by the semi-major
half axis of the ellipse, R ≈ a we can use Eq. (9.112) to express the mechanical
energy in terms of the angular frequency,
 
G M1 M2  2/3
E =− 1/3
. (9.128)
RS 2

9.7.3 Gravitational Radiation Emission

According to the general theory of relativity the binary system will emit gravita-
tional waves and thereby loose mechanical energy. The emitted power is given by
Eq. (9.121) in Ref. [5] for a system of two compact objects with equal masses. In
the general case allowing different masses this expression is generalized to

32
P= Gcμ2 R 4 6 . (9.129)
5
Using Eq. (9.108) this may be written as
9.7 The LIGO-Detection of Gravitational Waves 267

32 cG 4 M12 M22 M
P= . (9.130)
5 R5
The emission of gravitational waves causes the compact object to move towards
each other along a spiral path. The orbital energy may be approximated by

G M1 M2
E =− . (9.131)
2R
The rate of change of the orbital energy,

G M1 M2 Ṙ
Ė = , (9.132)
2 R2
is equal to the power given in Eq. (9.130). Hence

8
R 3 Ṙ = − c RS1 RS2 RS . (9.133)
5
where R S = R S1 + R S2 . The distance between the objects at a point of time t is
 1/4
32
R= R04 − R S1 R S2 R S ct , (9.134)
5

where R0 = R(0). This shows that it takes a finite time

5R04
tGW = (9.135)
32c RS1 RS2 RS

before two objects moving around an elliptical orbit with initial distance R0 , collides
with each other. If a detection signal lasts for a time tD until the sign that a merger
has happened appears, then the objects had an initial distance
 1/4
32
R0 = RS1 RS2 RS ctD (9.136)
5

when the initial part of the signal was emitted.

9.7.4 The Chirp

I order to extract information from the measured data, and it will be useful to express
the emitted effect in terms of the frequency f of the observed radiation. As shown in
Eq. (9.110) of Ref. [5] the frequency of the radiation is twice the orbital frequency.
Hence  = π f. Inserting this into Eq. (9.112) gives
268 9 The Linear Field Approximation and Gravitational Waves

 1/3
c 2 RS
R= . (9.137)
2π 2 f 2

Hence Eqs. (9.129) and (9.131) take the form

 2 4/3
32 10/3 c RS
P= π cGμ2 f 10/3 , (9.138)
5 2
 1/3
G M1 M2 2π 2
E =− f 2/3 . (9.139)
2 c 2 RS

Differentiation gives
 1/3
G M1 M2 2
Ė = −π 2/3 f −1/3 f˙. (9.140)
3 c 2 RS

Equating (9.138) and (9.140) leads to

96 π 8/3
f −11/3 f˙ = (G Mch )5/3 , (9.141)
5c5
where Mch is the chirp mass defined in Eq. (9.108). Integration gives
   −3/8
64 1/3 8/3 R Sch 5/3 −8/3
f = f0 1− 2 π f0 t . (9.142)
5 c

where f 0 = f (0). Solving this equation with respect to the Schwarzschild radius of
the chirp mass leads to
 3/5
5 f 1 1
RSch = c , f ≡ − , t = t2 − t1 , (9.143)
64 · 21/3 π 8/3 t 8/3
f1
8/3
f2

which may be written

 3/5
f
Mch = 7.8 · 10 H z 2
M . (9.144)
t

The objects begin to coalesce when their separation is equal to the sum of their
Schwarzschild radii, R = 2RS . Inserting this into Eq. (9.112) gives
c
RS = , (9.145)
4π f c
9.7 The LIGO-Detection of Gravitational Waves 269

where f c is the frequency of the emitted gravitational radiation when the objects
begin to coalesce. This may be written

2.0 · 104 H z
M= M , (9.146)
fc

where M is the mass of the Sun.

Let us apply Eqs. (9.144) and (9.146) to the signals shown in Fig. 9.5. We shall
only illustrate the method and therefore do not include the uncertainties. The figure
shows that the initial frequency is f 1 = 30 Hz at the point of time t1 = 0.46 s and the
frequency is f 2 ≈ 150 Hz at t2 = 0.50 s. This gives f = 1.1 · 10−4 , t = 0.04 s
with a rather large uncertainty. Hence Mch ≈ 23M . The measured peak frequency
was around f p = 200 Hz. This gives from Eq. (9.146) a total mass M = 100M .
More accurate relativistic calculations give a somewhat lower mass. Also there is
a rather large uncertainty in the value of the maximal frequency, which may be
somewhat higher than the measured peak frequency due to the very rapid merger.
Assuming that f c ≈ 350 Hz, we get M ≈ 57M . Inserting these values for the
total mass and the chirp mass into Eq. (9.111) gives for the masses of the two black
holes, M1 ≈ 38M , M2 ≈ 19M . The masses determined by the more accurate
calculations of the LIGO–VIRGO-teams were Mch ≈ 24M , M ≈ 56M , M1 ≈
31M , M2 ≈ 25M . The Schwarzschild radii of the black holes are RS1 = 91 km
and RS2 = 74 km. Inserting these radii and the duration of the signal before merger,
tD ≈ 0.07 s, into Eq. (9.136) gives the distance between the black holes when they
emitted the first detected gravitational wave, R0 = 630 km.
On 17 August 2017 LIGO and Virgo detected a gravitational wave from colliding
neutron stars, and 1.7 s later ESA’s INTEGRAL telescope and NASA’s Fermi gamma
ray space telescope detected a short gamma ray burst from the same source. The
next days and weeks the afterglow of the burst was observed by a large number
of telescopes. This was the first time electromagnetic signals were observed from
the source of directly detected gravitational waves. It opened a new window for
observing the universe and gave a great stimulus to the new research area called
multi-messenger astronomy.

References

1. Eddington, A.S.: The propagation of gravitational waves. Proc. Roy. Soc. Lond. 268–281 (1922)
2. Abbott, P.B., et al.: Observation of gravitational waves from a binary black hole merger. PRL
116, 061102 (2016). https://doi.org/10.1103/PhysRevLett.116.061102
3. Satyaprakash, B.S., Schutz, B.F.: Physics, astrophysics and cosmology with gravitational waves.
Living Rev. Relativity 12, 2 (2009)
4. Mathur, H., Brown, K., Lowenstein, A.: An analysis of the LIGO discovery based on introductory
physics. Am. J. Phys. 85(9) (2017)
5. Grøn, Ø., Hervik, S.: Einstein’s General Theory of Relativity. Springer (2007)
Chapter 10
Black Holes

Abstract Spacetime outside black holes with and without rotation—i.e. the Kerr and
Schwarzschild spacetimes—is studied. By considering the motion of free particles
in the Kerr spacetime we find an exact expression for the angular velocity of the
inertial dragging. Hawking radiation from a non-rotating black hole is also studied.

10.1 “Surface Gravity”: Acceleration of Gravity

at the Horizon of a Black Hole

The quantity which is called surface gravity is equal to the acceleration of gravity at
the horizon of a black hole. The acceleration of gravity is equal to the acceleration
of a freely falling particle instantaneously at rest as observed by a static observer in
the coordinate system. However, it has become usual to express the acceleration of
gravity measured by an observer in terms of the acceleration scalar of the observer.
Note that the 4-acceleration, and hence, the acceleration scalar of a free particle
vanishes, so this cannot be used in the same way.
It is tempting to define the acceleration of gravity mathematically as equal to the
acceleration scalar a of the observer, since this is an invariant quantity representing
the acceleration of the observer relatively to a freely falling particle, as measured by
the standard measuring rods and clocks of the observer. But we saw in Eq. (8.36)
that the standard clocks do not proceed at the horizon of a black hole. Therefore the
acceleration scalar of the observer diverges there. For this reason it has become usual
to define the acceleration of gravity as

dτ a
g=a = t, (10.1)
dt u

where u t is the time component of the 4-velocity of the observer. This quantity has
a finite value at the horizon of a black hole.
In this chapter we use units so that c = 1. Surface gravity is denoted by κ and is
defined by

© Springer Nature Switzerland AG 2020 271

Ø. Grøn, Introduction to Einstein’s Theory of Relativity,
Undergraduate Texts in Physics, https://doi.org/10.1007/978-3-030-43862-3_10
272 10 Black Holes

a 
κ = lim , a = aμ a μ , (10.2)
r →r+ ut

where r+ is the horizon radius, r+ = R S for the Schwarzschild spacetime, and u t is

the time component of the 4-velocity.
The 4-velocity of an observer permanently at rest in the Schwarzschild spacetime
is

dt 1 et
u = u t et = et = √ et =  . (10.3)
dτ −gtt 1− RS
r

The only component of the 4-acceleration different from zero is ar . The

4-acceleration of the observer is
μ
a = ∇u u = u ;ν u ν eμ = (u μ,ν + αν
μ α ν
u )u eμ . (10.4)

The covariant, radial component of the 4-acceleration is

r tt
ar = (u r,ν + r αν u α )u ν = u r,ν u ν +r tt (u t )2 = . (10.5)
   1 − RrS
=0

The Christoffel symbol is

1 ∂gtt RS
r tt = − = − 2. (10.6)
2 ∂r 2r
Inserting this into Eq. (10.5) gives

R S /2r 2
ar = . (10.7)
1− RS
r

The contravariant component of the 4-acceleration is

ar RS RS
a r = grr ar = = (1 − )ar = 2 . (10.8)
grr r 2r

The positive sign of a r means that an observer permanently at rest in the curvature
coordinate system accelerates outwards relative to an inertial (freely falling) reference
frame which has vanishing 4-acceleration.
The acceleration scalar is

√ r R S /2r 2
a= ar a =  . (10.9)
1 − RrS
10.1 “Surface Gravity”: Acceleration of Gravity at the Horizon of a Black Hole 273

This represents the acceleration as measured with standard instruments at the

position of the particle. The acceleration of gravity as defined in Eq. (10.1) is

RS
g= . (10.10)
2r 2
The surface gravity is equal to the acceleration of gravity at the horizon of a black
hole,

1
κ = lim g = . (10.11)
r →R S 2R S

Inserting the velocity of light we get

c2
κ= . (10.12)
2R S

On the horizon of a black hole with one solar mass κ = 2 × 1013 m/s2 .

10.2 Hawking Radiation: Radiation from a Black Hole

The radiation from a black hole has a thermal spectrum. Following Hawking we shall
write down an expression for the temperature of a Schwarzschild black hole of mass
M. The Planck spectrum has an intensity maximum at a wavelength given by Wien’s
displacement law,

N c
= . (10.13)
kT
where k is Boltmann’s constant and N = 0.2014. For radiation emitted from a black
hole Hawking derived the following expression for the wavelength at a maximum
intensity,

8π N G M
 = 4π N R S = . (10.14)
c2
Substituting for  from Wien’s displacement law leads to

c3 c
T = = κ. (10.15)
8π Gk M 2π k

Inserting values for , c and k gives

274 10 Black Holes

2 × 10−4 m
T ≈ K. (10.16)
RS

For a black hole with one solar mass, we have T ≈ 10−7 . When the mass
decreases because it radiates, the temperature increases. So a black hole has a negative
heat capacity. The energy loss of a black hole because of radiation is given by the
Stefan–Boltzmann law,

dM A
− = σT4 2, (10.17)
dt c
where A is the area of the horizon

16π G 2 M 2
A = 4π R S2 = , (10.18)
c4
In Eq. (10.17)

π 2k4
σ = (10.19)
603 c2
is Stefan–Boltzmann’s constant. Combining these equations we get

dM KH c4
= − 2 , KH = = 2.2 · 1015 kg3 /s. (10.20)
dt M 15360π G 2
Hence

M
M 2 dM = − K H t, (10.21)
M0

giving

1/3
M(t) = M03 − 3K H t . (10.22)

The time taken for a black hole to evaporate is

 3
M03 M0
tev = = tev , (10.23)
3K H M

where tev = 2.1 · 1067 years is the evaporation time of the Sun. The evaporation
time of a black hole with one Planck mass is

G
tev P = 5120π = 5120π t P . (10.24)
c5
10.2 Hawking Radiation: Radiation from a Black Hole 275

where t P = 5.4·10− 44 s is the Planck time. Hence tev P = 8.67·10−40 s. A primordial

black hole made at the Big Bang and exploding at the present time when the age of
the universe is t0 = 13.7 · 109 years, is

M0 = (3K H t0 )1/3 , (10.25)

giving M0 ≈ 1011 kg which is around the mass of Mount Everest. It is called a

“mini-black hole”.

10.3 Rotating Black Holes: The Kerr Metric

The solution of Einstein’s field equations describing spacetime outside a rotating

mass distribution was found by Roy Kerr in 1963 and is therefore called the Kerr
spacetime.
A time-independent, time-orthogonal metric is known as a static metric. A time-
independent metric is known as a stationary metric. A stationary metric allows
rotation.
Consider a stationary metric which describes an axially symmetric space

ds 2 = −e 2ν dt 2 + e2μ dr

2+ e 2μ(dφ2− ωdt)
2 2ψ 2
+ e2λ dθ 2
, (10.26)
= − e − e ω dt + e dr + e dφ 2 + e2λ dθ 2 − 2e2ψ ω dφ dt
2ν 2ψ 2 2ψ

where ν, μ, ψ, λ and ω are functions of r and θ . The covariant components of the

metric tensor are

gtt = e2ψ ω2 − e2ν , grr = e2μ , gφφ = e2ψ , gθθ = e2λ , gtφ = − e2ψ ω.
(10.27)

Hence,
gtφ
ω=− , gtt = gφφ ω2 − e2ν . (10.28)
gφφ

By solving the vacuum field equations for this line element, Kerr found the
solution:

ρ2 ρ2 Σ2
e2ν = , e 2μ
= , e 2ψ
= sin2 θ, e2λ = ρ 2 ,
Σ2  ρ2
R S ar
ω= , where ρ 2 = r 2 + a 2 cos2 θ,
Σ2
 = r 2 + a 2 − R S r,
Σ 2 = (r 2 + a 2 )2 − a 2  sin2 θ. (10.29)
276 10 Black Holes

This is the Kerr solution expressed in Boyer–Lindquist coordinates. The function

ω is an angular velocity. As mentioned above the Kerr solution is the metric for
spacetime outside a rotating mass distribution. The constant a represents spin per
mass unit for the mass distribution and R S its Schwarzschild radius. With the units
we use, a has dimension length.
The line element has the form
 
RS r ρ2 2R S ar
ds 2 = − 1 − 2 dt 2 + dr 2 − sin2 θ dtdφ + ρ 2 dθ 2
ρ  ρ2
 
R S a 2r
+ r 2 + a2 + sin 2
θ sin2 θ dφ 2 . (10.30)
ρ2

Light emitted from the surface, r = r0 , where gtt (r0 ) = 0, is infinitely redshifted
further out. Observed from the outside time does not proceed at this surface. The
radial coordinate r0 is given by

ρ 2 = R S r0 , (10.31)

or

r02 + a 2 cos2 θ = R S r0 , (10.32)

giving

r0 = R S /2 ± (R S /2)2 − a 2 cos2 θ. (10.33)

This is the equation of two surfaces with the property that light emitted from the
surfaces is infinitely redshifted. In the equatorial plane, θ = π/2, the radius of the
surface with infinitely redshifted light has a coordinate distance from the origin equal
to the Schwarzschild radius of the mass distribution,

r0 = R S . (10.34)

10.3.1 Zero-Angular Momentum Observers

The Lagrange function of a free particle in the equator plane, θ = π/2, is

1 1 1 1
L = − (e2ν − ω2 e2ψ )t˙2 + e2μ ṙ 2 + e2ψ φ̇ 2 + e2λ θ̇ 2 − ωe2ψ t˙φ̇. (10.35)
2 2 2 2

Here θ̇ = 0. The constant momentum pφ of the cyclic coordinates φ is

10.3 Rotating Black Holes: The Kerr Metric 277

∂L dt dφ
pφ ≡ = e2ψ (φ̇ − ωt˙), t˙ = , φ̇ = . (10.36)
∂ φ̇ dτ dτ

The angular velocity of the particle relative to the coordinate system is

dφ φ̇
= = . (10.37)
dt t˙
Hence

pφ = e2ψ t˙( − ω). (10.38)

Thus a zero-angular momentum observer, a ZAMO, has an angular velocity in the

Boyer–Lindquist coordinate system,  = ω. This gives the physical interpretation
of ω. It is the angular velocity of the ZAMOs in the Kerr spacetime relative to an
observer at rest infinitely far from the mass distribution. Hence ω is the angular
velocity of the inertial dragging in the Kerr spacetime. It follows from Eqs. (10.28)
and (10.29) that

gtφ RS a
ω=− = 3 . (10.39)
gφφ r + a 2 (r + R S )

Note that ω → 0 when r → ∞. The Kerr spacetime approaches the

Minkowski spacetime for large r. Calculating the inertial dragging angular veloc-
ity at the surface of the Earth one finds ωEarth = 10− 14 rad/s. At the infinite redshift
surface with radius (10.34) the dragging angular velocity is
a
ω(r0 ) = . (10.40)
R S2 + 2a 2

10.3.2 Does the Kerr Spacetime Have a Horizon?

Definition 10.3.1 (Horizon) A horizon is a one-way surface through which anything

can enter, but nothing can exit, not even light. Hence the region inside the horizon is
invisible for an outside observer.
Consider a particle in an orbit with constant r and θ . Its 4-velocity is

d
x dt d
x
= (−gtt − 2gtφ  − gφφ 2 )− 2 (1, ).
1
u = = (10.41)
dτ dτ dt
The existence of a stationary orbit requires

gφφ 2 + 2gtφ  + gtt < 0. (10.42)

278 10 Black Holes

This implies that

min <  < max , (10.43)

where

gtt gtt
min = ω − ω2 − and max = ω + ω2 − . (10.44)
gφφ gφφ

Outside the surface with infinite redshift gtt < 0. That is  can be negative, zero
and positive. Inside the surface r = r0 with infinite redshift gtt > 0. Here min > 0
and static particles,  = 0, cannot exist. This is due to the inertial dragging effect.
The surface r = r0 is therefore known as “the static border”. The interval of ,
where stationary orbits are allowed, is reduced to zero when min = max , that is

gtt = ω2 gφφ , (10.45)

which is the equation of the horizon. For the Kerr metric we have

gtt = ω2 gφφ − e2ν . (10.46)

Comparing with the second of Eq. (10.28), we see that the equation of the horizon
becomes

e2ν = 0, (10.47)

which requires  = 0 or

r 2 − R S r + a 2 = 0. (10.48)

The exterior solution is


r+ = R S /2 + (R S /2)2 − a 2 , (10.49)

and this is the equation for a spherical surface. Note that the radius of the horizon
is smaller than the Schwarzschild radius of the rotating mass. The dragging angular
velocity at the horizon is
a
ω(r+ ) = . (10.50)
R S r+

The static border and horizon of a Kerr black hole are shown in Fig. 10.1.
Going from a region outside the static border and inwards we have the following
situation. In the outside region it is possible for an observer to stay at rest if a suitable
non-gravitational force acts upon the observer. The surface r = r0 has two roles;
10.3 Rotating Black Holes: The Kerr Metric 279

Fig. 10.1 Static border and horizon of a Kerr black

light emitted from sources at rest on it are infinitely redshifted, and inside the surface
the inertial dragging is so strong that it is impossible to stay at rest even by trying
to do so by means of a motor. But it is possible to move along a path with constant
r , i.e. there exist stationary orbits. Also it is possible to come out of this region. But
inside the horizon, everything fall inwards, and it is not possible neither to move with
constant r nor to move outwards.

Exercises

10.1. A spaceship falling into a black hole

(a) In this problem we will consider a spaceship (A) falling radially from the
Earth (neglecting the gravitational field of the earth) into a Schwarzschild
black hole with mass the mass of the Sun, M = MSun at the position of
the Sun, 150 million km from the Earth.
What is the Schwarzschild radius of the black hole?
Find the equations of motion of the spaceship in curvature coordinates
r and t, using the proper time τ as time parameter.
Solve the equations of motion with the initial condition that the space
ship falls from rest at the Earth at proper time τ = 0.
When (in terms of τ ) does the spaceship reach the Schwarzschild
radius? And the singularity?
280 10 Black Holes

(b) The spaceship (A) has radio contact with a stationary space station (B)
at r B = 1 light years. The radio signals are sent with intervals T and
with frequency ω from both A and B. The receivers at A and B receive
signals with frequency ω A and ω B , respectively. Find ω A and ω B as a
function of the position of the spaceship. Hint: Perform the calculation
in two steps. At first find the change in frequency between two stationary
inertial systems in the points r A (the position of the spaceship) and r B .
Then calculate the change in frequency due to a transfer into an inertial
system with the velocity of the spaceship.
10.2. Kinematics in the Kerr spacetime
A Kerr black hole is an electrically neutral, rotating black hole. When
spacetime outside a Kerr black hole is described in Boyer–Lindquist coordi-
nates, the line element is the following,

ds 2 = −e2ν dt 2 + e2μ dr 2 + e2λ dθ 2 + e2ψ (dφ − ωdt)2 ,

where

ρ 2  2μ ρ 2 2λ
e2ν = , e = , e = ρ2,
Σ2 
 2
Σ gtφ 2Mar
e2ψ = sin2 θ, ω = − = ,
ρ 2 gφφ Σ2
ρ 2 = r 2 + a 2 cos2 θ,  = r 2 + a 2 − 2Mr, Σ 2 = (r 2 + a 2 )2 − a 2  sin2 θ.

Here M is the mass of the hole and a its spin per unit mass.
(a) Consider light moving in negative and positive direction of φ in the
equatorial plane, θ = π/2. What is the coordinate velocity cφ = dφ/dt
of light?
We now want to investigate the Sagnac effect in the Kerr space. An
emitter–receiver is attached to a point in the BL coordinate system. Light
signals with the frequency ν are sent by means of mirrors in both direc-
tions along the circle r = r0 , θ = π/2. Find the travel time difference of
light travelling in opposite directions, when the signals reach the receiver.
(b) A ZAMO is an observer with vanishing angular momentum. In the follow-
ing a ZAMO in the Kerr spacetime will be describing particles with fixed
r- and θ -coordinates. Introduce an orthonormal basis (etˆ
, er̂
, eθ̂
, eφ̂
),
where etˆ
is the 4-velocity of a ZAMO. The dual basis 1-forms are

tˆ

ω
−
= eν ω
−
t
, ω
−
r̂
= eμ ω
−
r
,

θ̂
ω
−
= eλ ω
−
θ
, ω
−
φ̂
= eψ (ω
−
φ
− ωω
−
t
). (10.51)
Exercises 281

Show that

etˆ
= e−ν (et + ωeφ ), er̂
= e−μ er ,
eθ̂
= e−λ eθ , eφ̂
= e−ψ eφ , (10.52)

where (et , er , eθ , eφ ) are the coordinate basis vectors in the BL coordinate
system.
Given a particle with 4-velocity components

dφ
U μ = t˙, φ̇ = t˙(1, ), = (10.53)
dt
in the Boyer–Lindquist coordinate system.
Show that the physical velocity of the particle, measured by a ZAMO,
is

v φ̂ = eψ−ν ( − ω). (10.54)

φ̂

What is the velocity v0 of a fixed coordinate point measured by a

ZAMO?
(c) Introduce an orthonormal basis field given by the expressions

e0̂ = (−g00 )−1/2 e0 , eî = (γii )−1/2 [ei − (gi0 /g00 )e0 ], (10.55)

where

γii = gii − gi0

2
/g00 .

Show that

etˆ = γ̂ e−ν et , er̂ = e−μ er ,

φ̂

eθ̂ = e−λ er , eφ̂ = γ̂ −1 e−ψ eφ + γ̂ e−ν v0 et , (10.56)

φ̂

where γ̂ = (1 − (v0 )2 )−1/2 . Find the dual basis 1-forms.

10.3. A gravitomagnetic clock effect
This problem is concerned with the difference of proper time shown by
two clocks moving freely in opposite directions in the equatorial plane of the
Kerr spacetime outside a rotating body. The clocks move along a path with
r = constant and θ = π/2.
(a) Show that in this case the radial geodesic equation reduces to

Γttr dt 2 + 2Γφtr dφdt + Γφφ

r
dφ 2 = 0. (10.57)
282 10 Black Holes

Calculate the Christoffel symbols and show that the equation takes
the form
 2
dt dt r3
− 2a + a2 − = 0, (10.58)
dφ dφ M

where M is the mass of the rotating body and a its angular momentum
per unit mass, a = J/M.
(b) Show that the time difference for one closed orbit in (φ → φ + 2π ) the
direct and the retrograde direction is t+ − t− ≈ 4πa = 4π J/M, or in
S.I. units,

t+ − t− = 4πa = 4π J/mc2 . (10.59)

Estimate this time difference for clocks in satellites moving in the

equatorial plane of the Earth. (The mass of the Earth is m = 6 · 1026 kg
and its angular momentum J = 1034 kg m2 /s.)
Chapter 11
Sources of Gravitational Fields

Abstract In this chapter we shall first find a general expression of the acceleration
of gravity due to a mass distribution. Then we shall deduce the solution of Einstein’s
field equations inside an incompressible star—the internal Schwarzschild solution.
Furthermore we shall present Israel’s formalism for describing singular mass shells
in the general theory of relativity, and apply this first to a shell consisting of dust
particles, and then to find a source of the conformally flat, spherically symmetric
Levi-Civita—Bertotti—Robinson metric, and finally to the Kerr spacetime. Lastly
we shall introduce a river model of space.

11.1 The Pressure Contribution to the Gravitational Mass

of a Static, Spherically Symmetric System

According to the Definition (10.1) of the acceleration of gravity (including a minus

sign denoting inwards acceleration, here),
a 
g=− t
, a = aμ a μ . (11.1)
u

where a is the acceleration scalar and u t the time component of the 4-velocity of an
observer permanently at rest in the coordinate system. We have the line element,

ds 2 = −e2α(r ) dt 2 + e2β(r ) dr 2 + r 2 d2 . (11.2)

Hence

gtt = −e2α , grr = e2β , (11.3)

which gives for the acceleration of gravity,

g = −eα−β α  . (11.4)

© Springer Nature Switzerland AG 2020 283

Ø. Grøn, Introduction to Einstein’s Theory of Relativity,
Undergraduate Texts in Physics, https://doi.org/10.1007/978-3-030-43862-3_11
284 11 Sources of Gravitational Fields

From the expressions (8.15) for the components of the Einstein tensor
E tˆtˆ, Er̂ r̂ , E θ̂ θ̂ , E φ̂ φ̂ , it follows that
 
φ̂ 2α 
E ttˆˆ − Er̂r̂ − E θ̂θ̂ − E φ̂ = −2e−2β + α  + α 2 − α  β  . (11.5)
r

We also have
 
2α 
(r 2 eα−β α  ) = r 2 eα−β + α  + α 2 − α  β  , (11.6)
r

which gives

1 φ̂
g= 2 (E ttˆˆ − Er̂r̂ − E θ̂θ̂ − E φ̂ )r 2 eα+β dr. (11.7)
2r

By applying Einstein’s field equations

E ν̂μ̂ = 8π GTν̂μ̂ (11.8)

we get

4π G φ̂
g= (Ttˆtˆ − Tr̂r̂ − Tθ̂θ̂ − Tφ̂ )r 2 eα+β dr. (11.9)
r2

This is the Tolman–Whittaker expression for gravitational acceleration.

The corresponding Newtonian expression is

4π G
gN = − ρr 2 dr . (11.10)
r2

The relativistic gravitational mass–density is therefore defined as

φ̂
ρG = −Ttˆtˆ + Tr̂r̂ + Tθ̂θ̂ + Tφ̂ . (11.11)

For an isotropic fluid with

φ̂
Ttˆtˆ = −ρ, Tr̂r̂ = Tθ̂θ̂ = Tφ̂ = p (11.12)

we get ρG = ρ + 3 p (with c = 1), which becomes

3p
ρG = ρ + . (11.13)
c2
11.1 The Pressure Contribution to the Gravitational Mass of a Static … 285

It follows that in relativity, pressure has a gravitational effect. Greater pressure

gives increasing gravitational attraction. Strain ( p < 0) decreases the gravitational
attraction. In the Newtonian limit, c → ∞, pressure has no gravitational effect.
Inserting the equation of state (7.17) of LIVE into Eq. (11.13) gives ρG = −2ρ <
0. Hence the assumption that it is not possible to measure velocity relative to the
vacuum energy, and that the vacuum energy can be described as a perfect fluid,
together with the general theory of relativity, imply that the vacuum energy produces
repulsive gravitation.

11.2 The Tolman–Oppenheimer–Volkoff Equation

We are going to solve Einstein’s field equations inside a spherically symmetric,

static mass distribution with density ρ(r ). Then the line-element can be given the
form (11.2). Using Eq. (8.15) for the components of the Einstein curvature tensor
the field equation for E tˆtˆ can be written

1 d
[r (1 − e−2β )] = 8π Gρ. (11.14)
r 2 dr
Integration gives
 r
r (1 − e−2β ) = 2G 4πρr 2 dr. (11.15)
0

The mass inside a spherical surface with coordinate radius r is

 r
m(r ) = 4πρr 2 dr. (11.16)
0

Hence, Eq. (11.15) can be written as

2Gm(r ) 1
e−2β = 1 − = . (11.17)
r grr

From the field equation for Er̂ r̂ , we have

2 dα −2β 1
e − 2 (1 − e−2β ) = 8π Gp. (11.18)
r dr r
This leads to
 
2 dα 2Gm(r ) 2Gm(r )
1− − = 8π Gp, (11.19)
r dr r r3
286 11 Sources of Gravitational Fields

or

dα m(r ) + 4πr 3 p(r )

=G . (11.20)
dr r (r − 2Gm(r ))

The relativistic equation of hydrostatic equilibrium is

T;νr̂ ν̂ = 0. (11.21)

Written out

T,ν̂r̂ ν̂ + Γα̂νν̂ T r̂ α̂ + Γα̂r̂ν̂ T α̂ν̂ = 0. (11.22)

The first term is

T,ν̂r̂ ν̂ = T,r̂r̂ r̂ = p,r̂ = er̂ ( p). (11.23)

√  √ 
Since er · er = grr we have |er | = grr . Hence, er̂ = 1/ grr er . This gives

1 dp
T,ν̂r̂ ν̂ = er̂ ( p) = √ er ( p) = e−β . (11.24)
grr dr

We here use ordinary derivatives instead of partial derivatives since p only depends
upon r. The second term is

α̂ν̂ ν̂ T r̂ α̂ = r̂ν̂ν̂ T r̂ r̂ = r̂ν̂ν̂ p = r̂tˆtˆ p + r̂î î p. (11.25)

The third term is

α̂r̂ ν̂ T α̂ν̂ = ν̂r̂ ν̂ T ν̂ ν̂ = tr̂ˆtˆρ + îr̂î p. (11.26)

In orthonormal basis, we have

ν̂ μ̂ = −μ̂ν̂ ⇒ ν̂ μ̂α̂ = μ̂ν̂ α̂ and r̂î î = î r̂ î = −r̂ î î = −r̂ˆ . (11.27)

i î

Hence the last terms in the expressions for α̂ν̂ ν̂ T r̂ α̂ and α̂r̂ ν̂ T α̂ν̂ cancel each other,
and the hydrostatic equation then takes the form

dp
e−β + r̂tˆtˆ p + tr̂ˆtˆρ = 0. (11.28)
dr
Furthermore

r̂tˆtˆ = −tˆr̂ tˆ = r̂ tˆtˆ = tr̂ˆtˆ, (11.29)

11.2 The Tolman–Oppenheimer–Volkoff Equation 287

and
dα
tr̂ˆtˆ = e− β . (11.30)
dr
The hydrostatic equation then takes the form

dp dα
+ ( p + ρ) = 0. (11.31)
dr dr
Inserting Eq. (11.20) into Eq. (11.31) gives

dp m(r ) + 4πr 3 p(r )

= −G(ρ + p) . (11.32)
dr r (r − 2Gm(r ))

This is the Tolman–Oppenheimer–Volkov (TOV) equation which can be used to

construct relativistic star models. It can be written in the form
 
p(r ) 4πr 3 p(r )
dp Gρm(r ) 1 + ρ
1 + m(r )
=− )
. (11.33)
dr r 2
1− r 2Gm(r

In Newtonian theory, only the first factor appears. The three last terms are relativis-
tic corrections. They show that pressure, p > 0, makes relativistic gravity stronger
than Newtonian gravity. This is significant for the possibility of collapse to black
holes.

11.3 An Exact Solution for Incompressible

Stars—Schwarzschild’s Interior Solution

The metric component gtt = −e2α(r ) can now be calculated from Eq. (11.31) in the
form
dp
dα = − (11.34)
ρ+p

We now consider an incompressible star, say a neutron star, with ρ = constant

and radius R. Integration of eq. (11.34) with vanishing pressure at the surface then
gives

α(r ) p  
dp p
dα = − ⇒ α(r ) = α(R) − ln 1 + . (11.35)
ρ+p ρ
α(R) 0
288 11 Sources of Gravitational Fields

Requiring continuity of the metric at r = R, and using the exterior Schwarzschild

metric, we get

gtt (R) 1 − RRS

gtt (r ) = e2α(r ) =  2 = −  2 . (11.36)
1 + ρp 1 + ρp

For an incompressible star, the mass inside a radius r is

4π 3
m(r ) = ρr . (11.37)
3
We then get

2Gm(r ) 8π Gρ 2
e−2β = 1 − =1− r . (11.38)
r 3
Defining a constant a with dimension length by

3
a2 = , (11.39)
8π Gρ

this may be written as

r2
e−2β = 1 − . (11.40)
a2
It may be noted that

8π Gρ 3 R3
R S = 2Gm(R) = R = 2. (11.41)
3 a
Hence

R3 R2 RS
a2 = and 2 = . (11.42)
RS a R

The TOV-equation may now be written

dp 1 r
= − 2 (ρ + 3 p)(ρ + p) . (11.43)
dr 2a ρ 1− r2
a2

So
p r
dp 1 r
=− 2 dr. (11.44)
(ρ + 3 p)(ρ + p) 2a ρ 1− r2
a2
0 R
11.3 An Exact Solution for Incompressible Stars—Schwarzschild’s … 289

which leads to

ρ+p a2 − R2
= , (11.45)
ρ + 3p a2 − r 2

giving the pressure distribution

 
√ √
a −r − a − R
2 2 2 2 1 − Rs 2
R 3 r − 1 − RRS
p(r ) = √ √ ρ=   ρ, r < R.
3 a2 − R2 − a2 − r 2 3 1 − RRS − 1 − RRs3 r 2
(11.46)

This gives

p 2 1 − RRS
1+ =   . (11.47)
ρ 3 1 − RRS − 1 − Rs 2
r
R3

Combining this with Eq. (11.36) we get

 2
1 RS RS 2
gtt = − 3 1 − − 1 − 3r . (11.48)
4 R R

Hence the line-element of the spacetime inside an incompressible star is

 2
1 R S R S dr 2
ds 2 = − 3 1 − − 1 − 3 r 2 dt 2 + + r 2 d2 . (11.49)
4 R R 1 − RRS3 r 2

The spacetime described by this line-element is called the internal Schwarzschild

spacetime.
In Newtonian gravity, there is no limit to how large a star can be if there exists a
sufficiently effective mechanism for generating a pressure gradient which may resist
gravity. We shall see that it is not so in relativistic gravity.
In order to satisfy the condition of hydrostatic equilibrium, the pressure must be
positive. This must be valid at every distance from the centre of the star, also at the
centre, p(0) > 0. It follows from the expression for the pressure distribution that

1 − 1 − RRS
p(0) =  > 0, (11.50)
3 1 − RRS − 1
290 11 Sources of Gravitational Fields

which requires that R S < (8/9)R. Including here the velocity of light, this leads to
the condition
√

c2 2 2 c2 3
M< √ = a, a = c . (11.51)
G 3π Gρ 3 G 8π Gρ

A star with a larger mass will collapse to a black hole independent of the
mechanism that generates a pressure gradient in the star.
Let us consider an incompressible neutron star as an illustrating example. A typical
density is then ρn = 5 · 1017 kg/m3 . This gives an = 18 km and M < 22 · 1030 kg =
11 M .

11.4 The Israel Formalism for Describing Singular Mass

Shells in the General Theory of Relativity

We shall here follow the descriptions of this formalism as presented by Israel [1],
Misner, Thorne Wheeler [2], and Lightman, Press, Price and Teukolsky [3].
Let Si j be the components of the energy–momentum tensor of a singular shell,
and Tαβ the components of the energy–momentum tensor of the medium filling the
region at each side of the shell. Here the indices i, j mark the coordinates in the shell
and α, β the coordinates of the spacetime outside the shell. Israel has shown that the
equation of continuity can be written
 
u j S mj|m = − Tαβ u α n β , (11.52)

where u j is the j-component of the 4-velocity of a point on the shell, n β the β−

component of its normal vector, the vertical line denotes covariant derivative with
respect to the intrinsic metric on the shell, and the bracket denotes the difference
between the values of the quantity inside the bracket at the two sides of the shell.
The energy–momentum tensor of the shell has the form

Si j = σ u i u j + ti j , (11.53)

where σ is the mass per unit area of the shell, and ti j are the components of the stress
tensor of the shell.
The components of the extrinsic curvature tensor of the shell are ([2] p. 513)

K i j = −n i; j = −n i, j + n α iαj , (11.54)

where iαj are the Christoffel symbols of the spacetime. The equation of motion of
the shell is ([3] exercise 21.8)
11.4 The Israel Formalism for Describing Singular Mass Shells … 291
 
  1
K i j = κ Si j − S (3) gi j , S = Skk = g i j Si j . (11.55)
2

Contraction gives

[K ] = −(κ/2)S. (11.56)

Hence, solving Eq. (11.55) with respect to S ij gives the energy–momentum tensor
of the shell in terms of the discontinuity of the extrinsic curvature tensor of the shell,
as evaluated inside and outside the shell
 
κ S ij = K ij − K δ ij . (11.57)

Example 11.1 (Thin dust shell described by the Israel formalism) We shall describe
a spherical shell consisting of dust particles in empty space. The shell will collapse,
and we shall first investigate whether the rest mass of the shell is constant and then
find the equation of motion of the shell.

In empty space, the energy–momentum tensor at the right-hand side of equation

of continuity (11.52) vanishes. Then it takes the form,

u j S mj|m = 0. (11.58)

Also the stress tensor of the shell vanishes. Then the energy–momentum tensor
of the shell as given in Eq. (11.53) has the form

Si j = σ u i u j . (11.59)

Hence,
 
u j S mj|m = u j σ u j u m |m = u j σ|m u j u m + u j σ u j|m u m + σ u j u j u m
|m . (11.60)

We now use the 4-velocity identity which here takes the form u j u j = −1, and
that the 4-acceleration is

a j = u m u j|m . (11.61)

Equation (11.60) then takes the form

j
u j S mj|m = −u m σ|m + σ u j a j − σ u | j . (11.62)

Here
dσ
u m σ|m = (11.63)
dτ
292 11 Sources of Gravitational Fields

is the derivative of the mass density of the shell with respect to its proper time.
Furthermore, since the 4-acceleration is orthogonal to the 4-velocity, u j a j = 0.
Thus we get

dσ
u j S mj|m = − − σ u j u| j . (11.64)
dτ
The continuity equation of the shell then takes the form

dσ
= −σ u j u | j . (11.65)
dτ
Writing this out we have

dσ  − 1/2 (3) 1/2 j 

= −σ (3) g g u  , (11.66)
dτ j

where (3) g is the determinant of the three-dimensional metric on the shell.

The motion of the shell is given by r = R(τ ) where 4π R 2 (τ ) is the area of the
shell at the point of time τ . Let the intrinsic geometry of the shell be described in
terms of the coordinates (τ, θ, φ). Then the line-element of the intrinsic geometry of
the shell may be written

ds 2 = −dτ 2 + R 2 (τ )d2 . (11.67)

Since
(3)
g = −R 4 (τ ) sin2 θ, (11.68)

and the only non-vanishing component of the 4-velocity in the intrinsic coordinate
system is

dτ
uτ = = 1, (11.69)
dτ
the continuity equation of the shell reduces to

1 dσ 1 dR 2 2 dR
=− 2 =− . (11.70)
σ dτ R dτ R dτ
Integration gives

σ R 2 = constant. (11.71)

Hence, the rest mass of the shell

11.4 The Israel Formalism for Describing Singular Mass Shells … 293

M S = 4π σ R 2 (11.72)

is constant.
We shall now find the equation of motion of the shell. For a shell consisting of
dust particles, Eq. (11.55) reduces to
 
  1 (3)
Ki j = κσ u i u j + gi j . (11.73)
2

The equation of motion will be found from the θ θ − component of this equation.
Since u θ = 0 and (3) gθθ = R 2 this gives

[K θθ ] = (κ/2)σ R 2 = R SS /2, (11.74)

where R SS is the Schwarzschild radius of the shell. As applied to a spherical surface

with the line-element (11.67), Eq. (11.54) gives
α
K θθ = n α θθ = n r r θθ = −Rn r . (11.75)

Hence, Eq. (11.74) takes the form

R SS
nr + − nr − = − . (11.76)
2R
We now utilize that

n · n̄ = 1, u · u = −1, u · n = 0. (11.77)

In the exterior region with Schwarzschild metric, these equations take the form
   r + 2
R S  t+ 2 n
− 1− n + = 1, (11.78)
r 1 − RrS
 
R S  t 2 (u r )2
− 1− u + = −1, (11.79)
r 1 − RrS
 
R S t t+ ur nr +
− 1− un + = 0, (11.80)
r 1 − RrS

where R S is the constant Schwarzschild radius of the shell. Equations (11.78–11.80)

give
 
ur r+
 r 2 R S 1/2
n t+
= , n = 1+ u − . (11.81)
1 − R S /R R
294 11 Sources of Gravitational Fields

The radial component of the 4-velocity of a particle on the shell is

dR
ur = ≡ Ṙ. (11.82)
dτ
Hence,
 
r+ R S 1/2
n = 1 + Ṙ −
2
. (11.83)
R

Inside the shell R S = 0, giving

 1/2
n r − = 1 + Ṙ 2 . (11.84)

Inserting the expressions (11.83) and (11.84) into Eq. (11.76) gives the equation
of motion of the shell
 
R S 1/2  1/2 R SS
1 + Ṙ −
2
− 1 + Ṙ 2 = . (11.85)
R 2R

Solved with respect to M = R S /2 (using units so that Newton’s gravitational

constant and the velocity of light are equal to one), we get
 1/2
M = M S 1 + Ṙ 2 − M S2 /2R. (11.86)

Let T be the time measured on a standard clock at rest just inside the shell. Then
 − 1/2
dT = 1 − (dR/dT )2 dτ. (11.87)

This gives
 −1
1 + Ṙ 2 = 1 − (dR/dT )2 . (11.88)

Hence Eq. (11.86) takes the form

MS M2
M= − S. (11.89)
1 − (dR/dT )2 2R

This equation expresses energy conservation. The mass M of the external

Schwarzschild spacetime represents the total energy of the shell. The quantity
 −1/2
M S 1 − (dR/dT )2 is the relativistic energy of the shell as measures by an
observer at rest in the coordinate system at the position of the shell, i.e. the sum of
rest mass energy and kinetic energy, not including any gravitational potential energy.
11.4 The Israel Formalism for Describing Singular Mass Shells … 295

The term M S2 /2R is the Newtonian expression for the gravitational self energy—or
potential energy—of a spherical shell.
Differentiation of Eq. (11.86) with constant M gives
 1/2
R̈ = − 1 + Ṙ 2 M S /2R 2 . (11.90)

The acceleration of a shell instantaneously at rest is

R̈ = −M S /2R 2 . (11.91)

It accelerates as if half of its mass is inside the shell and half is outside.

11.5 The Levi-Civita—Bertotti—Robinson Solution

of Einstein’s Field Equations

We shall here derive and give a physical interpretation of a conformally flat, static,
spherically symmetric solution of Einstein’s field equation which was found inde-
pendently by Levi-Civita in 1917 ([4]), and by B. Bertotti and I. Robinson in 1959
([5] and [6]), and which is therefore called the Levi-Civita—Bertotti—Robinson
solution of Einstein’s field equations.
A spacetime is said to be conformally flat if it admits a coordinate system such that
the line-element can be written as a function times the line element of the Minkowski
spacetime. In the static spherically symmetric case the line element of a conformally
flat spacetime can be written
 
ds 2 = e2α(r ) −c2 dt 2 + dr 2 + r 2 d2 (11.92)

Einstein’s field equations take the form

 
4 G Q2
G tt = e−2α 2α  + α 2 + α  = −e−4α , (11.93)
r 4π ε0 c4 r 4
 
−2α 4  G Q2
Gr = e
r
3α + α = −e−4α
2
, (11.94)
r 4π ε0 c4 r 4
 
θ φ −2α 2  G Q2
Gθ = Gφ = e 2α + α + α = e−4α
 2
. (11.95)
r 4π ε0 c4 r 4

Here Q = Q(r ) is the charge inside a spherical shell with radius r. Subtracting
Eq. (11.93) from (11.95) we obtain

G Q2
e2α α  = − . (11.96)
4π ε0 c4 r 3
296 11 Sources of Gravitational Fields

Inserting this into Eq. (11.94) leads to two different solutions, either

Q(r ) = 0, (11.97)

or

4π ε0 α
Q(r ) = ±c 2
re . (11.98)
G

The first solution gives α(r ) = constant, which can be chosen to be zero with
a suitable choice of coordinates. Hence this solution represents the Minkowski
spacetime.
Inserting Eq. (11.98) into (11.96) gives

1
α = − . (11.99)
r
Integration gives

RQ
eα = , (11.100)
r
where R Q is a constant of integration with dimension length. Inserting this into
Eq. (11.64) leads to

4π ε0
Q(r ) = ±c 2
RQ . (11.101)
G

This means that the charge inside a spherical surface with radius r is constant.
Hence the charge density vanishes in the considered region. Thus the assumption of
a conformally flat, static and spherically symmetric spacetime implies that there is
no charge in this region, although there is a radial electric field there. So there must
be a charge inside the region described by the solution (11.100). The line element of
the spacetime outside the charge distribution has the form

R 2Q  2 2 
ds 2 = −c dt + dr 2 + r 2 d2 . (11.102)
r2
Since the metric is static the coordinate clocks go at the same rate everywhere,
equal to the rate of standard clocks at r = R Q . There is continuous metric at r = R Q
with Minkowski spacetime inside this radius. Hence the charge is situated on a
spherical shell with radius r = R Q .
The line-element (11.102) shows that in this spacetime standard clocks at rest
slow down for increasing r. Hence, the gravitational field points in the direction of
increasing r. This means that the mass and charge distribution inside the region with
this geometry causes repulsive gravitation.
11.5 The Levi-Civita—Bertotti—Robinson Solution of Einstein’s … 297

The Kretschmann curvature scalar of this spacetime is

8
R μναβ Rμναβ = , (11.103)
R 4Q

which is constant. Hence there is no physical singularity in this spacetime.

Let us introduce a new radial coordinate equal to the physical radial distance,

RQ
dr̂ = dr. (11.104)
r

Integration with the boundary condition that r = R Q corresponds to r̂ = R Q

gives
 
r
r̂ = R Q 1 + ln . (11.105)
RQ

The inverse transformation is

r = R Q e(r̂ −R Q )/R Q . (11.106)

With the radial coordinate r̂ the line-element takes the form

ds 2 = −e−2(r̂ −R Q )/R Q c2 dt 2 + dr̂ 2 + R 2Q d2 . (11.107)

The geometry of space has a very strange property. The physical area of a spherical
surface with radius r̂ is independent of the radius and equal to the area of a surface
with radius r̂ = R Q .
We shall now find the physical properties of source of this spacetime [7]. These
properties will presumably provide an explanation for the phenomenon of repulsive
gravity outside the charged shell.

11.6 The Source of the Levi-Civita—Bertotti—Robinson

Spacetime

We shall now find the energy–momentum tensor of a static spherical shell as given
by Eq. (11.57) with Minkowski spacetime inside the shell and a curved spacetime
outside. In order to provide a rather general result, we shall first consider a line-
element,

ds 2 = −eα c2 dt 2 + eβ dr 2 + eγ d2 , (11.108)

298 11 Sources of Gravitational Fields

describing spacetime outside the shell, where α, β and γ are functions of r, and then
specialize to the Levi-Civita—Bertotti—Robinson metric.
The unit normal vector to a spherical surface about the origin is

n = e−β/2 er . (11.109)

The covariant components of the extrinsic curvature tensor are given by

Eq. (11.54). Inserting the expression (5.65) for the Christoffel symbols with the
line element (11.108) gives

1 1
K i j = − n r gi j,r = − e−β/2 gi j,r . (11.110)
2 2
The mixed components are

1 φ 1
K tt = − e−β/2 α,r , K θθ = K φ = − e−β/2 γ,r . (11.111)
2 2
Inserting these expressions into Eq. (11.57) gives the mixed components of the
energy–momentum tensor of the shell
   
κ Stt = −2 K θθ = e−β/2 γ,r ,
φ     1  
κ Sθθ = κ Sφ = − K tt − K θθ = e−β/2 α,r + γ,r . (11.112)
2
We shall now specialize to the Levi-Civita—Bertotti—Robinson spacetime
  with
the line element (11.102) outside the shell. Then α+ = β+ = 2 ln R Q /r and
γ+ = 2 ln R Q outside the shell, and α− = β− = 0, γ− = 2 ln r inside it. This gives

φ 2c2
κ Stt = κ Sθθ = κ Sφ = − . (11.113)
RQ

Surprisingly, the mass density and strain of the shell are independent of its radius.
It follows from Eq. (11.113) that the components of the energy–momentum tensor
of the singular shell may be written as

2c2
S ij = −σ δ ij , σ = , (11.114)
κ RQ

where σ is the mass density of the shell. Hence, the shell has a mass

M = 4π σ R 2Q = R Q c2 /G. (11.115)

Note that the radius of the shell is only half its Schwarzschild radius.
11.6 The Source of the Levi-Civita—Bertotti—Robinson Spacetime 299

Equation (11.114) shows that the shell is a domain wall. Furthermore the Levi-
Civita—Bertotti—Robinson spacetime is a solution of Einstein’s field equations with
a radial electric field. Hence√the domain wall is charged. Using Eqs. (11.101) and
(11.115) can be written Q = 4π ε0 G M, which shows that the charge of the domain
wall is proportional to its mass.
From Eq. (11.113) we can understand why there is repulsive gravity in the Levi-
Civita—Bertotti—Robinson spacetime. This equation shows that there is a strain
in the charged domain wall equal to minus its energy density. Hence, according
to Eq. (11.11) is has negative gravitational mass, and according to the Tolman–
Whittaker formula (11.9) the strain of the domain wall causes repulsive gravity in
the space outside the charged domain wall.

11.7 A Source of the Kerr–Newman Spacetime

The most general black holes can have three properties: mass, charge and angular
momentum. The spacetime outside a black hole with all three properties is a gener-
alization of the Kerr spacetime with an electric field. It is called the Kerr–Newman
spacetime. The line element of this spacetime in Boyer–Lindquist coordinates may
be written
 2 
dr   r2  2
ds 2 = −c2 dt 2 + ρ 2 + dθ 2 + r 2 + a 2 sin2 θ dφ 2 + eff2 cdt − a sin2 θ dφ ,
 ρ
2
reff = R S r − 2R 2Q , ρ 2 = r 2 + a 2 cos2 θ,  = r 2 + a 2 − reff
2
, (11.116)

where R S is the Schwarzschild radius of the mass distribution, R Q the radius given
in Eq. (11.101) corresponding to its charge Q, and a = J/Mc a length representing
its angular per unit mass.
We shall need both the covariant and the contravariant components of the metric
tensor,
⎡  r2

ar 2 sin2 θ
⎤
− 1 − ρeff2 0 0 − effρ 2
⎢ ⎥
⎢ 0 ρ2
0 0 ⎥
gμν = ⎢⎢  ⎥, (11.117)
0 0 ρ 2
0 ⎥
⎣ ⎦
ar sin θ
2 2
a r sin θ
2 2 2
− effρ 2 0 0 r 2 + a 2 + effρ 2 sin2 θ
⎡ ar 2
⎤
− ρ2  0 0 − ρ 2eff
2

⎢ ⎥
⎢ 0 ρ2 0 0 ⎥
μν
g =⎢ ⎢ ⎥,
1 ⎥ .
⎣ 0 0 ρ2 0 ⎦ (11.118)
ar 2
− ρ 2eff 0 0 −a sin θ
2 2

ρ 2  sin2 θ
 2
 2 = r 2 + a 2 − a 2  sin2 θ
300 11 Sources of Gravitational Fields

The unit normal vector of the shell is n = n r er . Hence, n ·

n = 1 gives grr (n r )2 = 1.
With the metric (11.117) this leads to
√

n = er . (11.119)
ρ

From Eq. (11.110) we now find the components of the external curvature tensor
of a surface with r = constant,
√
1 r ∂gθθ r 
K θθ = n r θθ = − n
r
=− ,
2 ∂r ρ
√    √
 ∂gφφ 2re2f f r a 2  2
K φφ = − = − 2r + R S − sin θ
2
sin θ,
2ρ ∂r ρ2 2ρ 2 ρ
√  √
 ∂gtt 2re2f f r 
K tt = − = − RS − ,
2ρ ∂r ρ 2 2ρ 3
√   √
 ∂gφt 2re2f f r a  2
K φt = − = − RS − sin θ. (11.120)
2ρ ∂r ρ2 2ρ 3

Lopez [8] has considered a shell with radius

2R 2Q Q2
r0 = = , (11.121)
RS 4π ε0 Mc2

so that reff = 0. In the case that Q is equal to the elementary charge, and M is the
electron mass, r0 is equal to the classical electron radius, 2.8 × 10−15 m. At the shell
the line-element (11.82) reduces to

ρ02 2
ds 2 = −c2 dt 2 + dr + ρ02 dθ 2 + r12 sin2 θ dφ 2 ,
r12
ρ02 = r02 + a 2 cos2 θ, r12 = r02 + a 2 . (11.122)

This line-element describes Minkowski spacetime at the shell in oblate spher-

ical coordinates. One can introduce ordinary spherical coordinates (R, ) by the
transformation

R cos  = r cos θ, R sin  = r 2 + a 2 sin θ. (11.123)

The ordinary coordinates are given explicitly in terms of the oblate ones by

a2 a2
R = r 1 + 2 sin2 θ , tan  = 1+ tan θ. (11.124)
r r2
11.7 A Source of the Kerr-Newman Spacetime 301

The inverse transformation is



 1 + a 22 cos2 θ sin 
r = R R
a2
, sin θ =  . (11.125)
1 + R2 2
1 + Ra 2

Taking the differential and inserting them into the line element (11.88) gives
 
ds 2 = −c2 dT 2 + dR 2 + R 2 d2 + sin2 dφ 2 , (11.126)

which describes Minkowski spacetime in ordinary spherical coordinates.

We now assume that there is Minkowski spacetime inside the shell, so that the
metric is continuous at the shell. However the derivatives are not continuous. At the
shell the expressions (11.120) for the components of the external curvature tensor
then reduce to
 
r0 r1 RS a2 r1
K θθ = − , K φφ = − r0 + sin 2
θ sin2 θ,
ρ0 2 ρ0 2 ρ0
R S r1 R S ar1
K tt = − , K φt = sin2 θ, (11.127)
2 ρ0 3 2 ρ03

and the metric components (11.117) and (11.118) are

   
ρ2 r2 1 1
gμν = diag −1, 20 , ρ02 , r12 sin2 θ , g μν = diag −1, 12 , 2 , 2 2 .
r1 ρ0 ρ0 r1 sin θ
(11.128)

Hence,
 
r0 r1 φ+ RS a2 1
K θθ+ = − , K φ = − r 0 + sin 2
θ ,
ρ03 2 ρ02 ρ0 r1
R S r1 φ+ RS a
K tt+ = , Kt = (11.129)
2 ρ0 3 2 r1 ρ03

outside the shell. In the oblate spherical coordinates inside the shell a
= 0, but there
M = 0. Thus the discontinuities of the mixed components of the exterior curvature
tensor at the shell are

    RS a2
φ
K θθ = 0, K φ = − sin2 θ,
2 r1 ρ03
 t R S r1   RS a
φ
Kt = , Kt = . (11.130)
2 ρ03 2 r1 ρ03
302 11 Sources of Gravitational Fields

The components of the energy–momentum tensor are now calculated from

Eq. (11.57) with the result
    RS a2
φ
κ Stt = − K θθ + K φ = sin2 θ, (11.131)
2r1 ρ03
    RS     R S r1
φ φ
κ Sθθ = − K tt + K φ = − , κ Sφ = − K tt + K θθ = − 3 ,
2r1 ρ0 2ρ0
(11.132)
  RS a
φ φ
κ St = K t = , (11.133)
2r1 ρ03
φ R S r1 a
κ Sφt = κg tt Stφ = κg tt Sφt = κg tt gφφ St = − sin2 θ. (11.134)
2ρ03

The energy–momentum tensor of the shell may be written

RS  i  1  2 
2κ S ij = − u u j + δ ij , u i = r , 0, 0, a . (11.135)
r1 ρ0 r1 ρ0 1

This is Lopez’s source of the Kerr–Newman spacetime. The first term represents
dust with negative rest mass density, and the second term a domain wall. The four
velocity of the dust particles show that they move in the φ− direction. Thus the
source may be described as a domain wall with “bubbles” of negative rest energy
within it rotating around in the wall.

11.8 Physical Interpretation of the Components

of the Energy–Momentum Tensor by Means
of the Eigenvalues of the Tensor

Let a system be described by an energy–momentum tensor with mixed components

S ij . The eigenvalues, λ, of the tensor are defined by the matrix equation
 i 
 S − λδ i  = 0. (11.136)
j j

In an n-dimensional space (11.136) is an n-degree equation for λ so that S ij has n

eigenvalues. Furthermore S ij has n eigenvectors, u(k) , given by

S ij u (k) j = λ(k) u (k)i . (11.137)

Let u and v be two different eigenvectors of S ij associated with two different

eigenvalues λ1 and λ2 . Then
11.8 Physical Interpretation of the Components … 303

S ij u i = λ1 u i , S ij vi = λ1 vi . (11.138)

giving

Si j u i v j − Si j u j vi = (λ2 − λ1 )u i vi . (11.139)

Interchanging the summation indices in the last term and using that the tensor S
is symmetrical, Si j = S ji , we get

(λ2 − λ1 )(
u · v) = 0. (11.140)

Since in general λ2
= λ1 this requires

u · v = 0. (11.141)

Hence, the eigenvectors of S are orthogonal.

For the physical problems considered, Eq. (11.137) will give one time like eigen-
vector, and the other are space like. We now require that all of the eigenvectors shall
be unit vectors. The eigenvectors of S will form an orthonormal basis. The time like
eigenvector is the 4-velocity of a point of the system described by the tensor S. This
means that the eigenvectors form an orthonormal basis co-moving with the system.
Let this four-velocity be u. Then the eigenvectors of S have the following physical
interpretation,

λ(t) = ρ0 , λ(k) = − pk , (11.142)

where ρ0 is the proper energy density measured by an observer at rest with the physical
system, and pk is the stress in the k-direction, i.e. the stress towards a surface with
normal vector in the k-direction. The components of the energy–momentum tensor
S may in general be written as

Si j = ρ0 u i u j + pk vi(k) v(k)
j , (11.143)
k

where v(k) is a space like eigenvector of S.

Example 11.2 (Lopez’s source of the Kerr–Newman metric) The energy–momen-
tum tensor of Lopez’s source of the Kerr–Newman tensor has components given in
Eqs. (11.132–11.134). Equation (11.136) here takes the form
 
 St − λ 0 St 
φ
 t
 
 0 Sθθ − λ 0  = 0. (11.144)
 φ 
 Sφt 0 Sφ − λ 

Thus
304 11 Sources of Gravitational Fields

   φ
φ  
Stt − λ Sθθ − λ Sφ − λ − St Sφt Sθθ − λ = 0. (11.145)

One solution of this equation is

λθ = Sθθ . (11.146)

The other two solutions are found from


φ φ φ
λ2 − Stt + Sφ λ + Stt Sφ − St Sφt = 0, (11.147)

giving

1 t 1 2
φ φ φ
λt = St + Sφ − Stt − Sφ + St Sφt ,
2 4
1 t 1 2
φ φ φ
λφ = St + Sφ + Stt − Sφ + St Sφt . (11.148)
2 4

Inserting the components (11.133–11.134) of the energy–momentum tensor of

Lopez’s source of the Kerr–Newman spacetime we get

RS
λt = 0, λθ = λφ = − . (11.149)
2r1 ρ0

The eigenvectors are now found from Eq. (11.137). The time like eigenvector, i.e.
the 4 velocity field of the source is given by
φ φ
Stt u t + Sφt u φ = 0, St u t + Sφ u φ = 0. (11.150)

The 4-velocity identity gives

 2  2
gtt u t + gφφ u φ + 2gφt u φ u t = −1. (11.151)

Since gφt = 0 at the shell this reduces to

 2  2
gtt u t + gφφ u φ = −1. (11.152)

Equations (11.150) and (11.152) give

⎡  2 ⎤−1/2 ⎡  2 ⎤−1/2
Stt Sφt
u t = ⎣−gtt − gφφ ⎦ , u φ = ⎣−gtt − gφφ ⎦ . (11.153)
Sφt Stt

Inserting the values of the metric tensor at the shell,

11.8 Physical Interpretation of the Components … 305

gtt = −1, gφφ = r12 sin2 θ, (11.154)

and
 2
Stt a2
= (11.155)
Sφt r14

gives the eigenvector

1  2 
u = r1 et + aeφ . (11.156)
r1 ρ0

This is the 4-velocity of particles following the shell. According to the interpre-
tation at the end of the previous section, this means that the bubbles with negative
rest energy within the shell rotate rigidly with an angular velocity

uφ a a
= = 2c = 2 c, (11.157)
u t r1 r0 + a 2

where we have used the last of Eq. (11.122) and inserted the velocity of light.

11.9 The River of Space

When teaching the theory of relativity one sometimes gets the question: What is
space? It is here understood that the question is concerned with ordinary three-space
and not the four-dimensional spacetime. The first part of the answer is to make clear
that space is a theory dependent concept. The second is to try to explain what we
mean by “space” according to the general theory of relativity.
One definition is to say that space is a set of simultaneous events. Even if this is
an essential part of what we mean by space, this definition is not sufficient to give
us a picture of space which makes us understand, for example, why light cannot be
emitted from the horizon of a black hole, and what is meant by an “expanding space”.
We must demand from the properties of space that it provides an interpretation of
the expression the “expanding space” according to the general theory of relativity.
Furthermore, the definition should capture the phenomenon of inertial dragging.
We shall here define the concept of “physical space” as a continuum of freely
moving reference particles with specified initial conditions. In a homogeneous and
isotropic universe, the only motion of the reference particles is that due to a universal
change of the distances between the particles. This may be described by a single
scale factor and defines the Hubble flow which obeys the Hubble–Lemaître law.
These particles make up what we call the “river of space” ([10]).
306 11 Sources of Gravitational Fields

We shall here describe the river of space in the Schwarzschild–de Sitter space-
time with the line element (11.172). One may think of the cosmological constant as
representing the constant density of LIVE (see 7.2.1) causing repulsive gravity.
Let

RS r2 2G M 3
f (r ) = 1 − − 2 , RS = , RH = . (11.158)
r RH c2 

The value of the cosmological constant representing the density of LIVE in our
universe is  ∼ 1052 m −2 . Hence R H ∼ 1, 7 · 1026 m ∼ 1.7 · 1010 l.y. The horizons
of this spacetime are given by f (r ) = 0, giving
√
2 RH 3 3 RS
r1 = √ R H sin θ, r2 = R H cos θ − √ sin θ, sin(3θ ) = . (11.159)
3 3 2 RH

We shall assume that 0 < θ < π/6.

Let us first calculate the 4-acceleration of a particle at rest in the coordinate
system. Hence we consider a particle with dr = dθ = dφ = 0 in the line element
(11.172). Then the only non-vanishing component of the particle’s 4-velocity is
u t = dt/dτ = f (r )−1/2 , where τ is the proper time of the particle. Since u t is
independent of the time, the components of the particle’s 4-acceleration are given by
 2
u α = ttα u t , where the only non-vanishing Christoffel symbol is ttr . Hence the
only nonzero component of the 4-acceleration is

2r 3 − R S R 2H
ar = . (11.160)
2R 2H r 2

Accordingly, the 4-acceleration vanishes on a 3-surface with radius

 1/3
R S R 2H
r0 = . (11.161)
2

Inserting, for example, the Schwarzschild radius of the Sun and the density of
dark energy in the universe we get r0 ≈ 200l.y.. It is clear that R H r0 for all
localized systems we are aware of in the local part of the universe. A particle at rest
at r = r0 is a free particle. Note further that a r < 0 for r < r0 and a r > 0 for r > r0 .
In this spacetime we define the river of space by a continuum of free particles
that are at rest at r = r0 . Hence the river of space flows inwards towards the central
mass for r < r0 and outwards for r > r0 . The reason for the outwards flow outside
r = r0 is the repulsive gravity of the LIVE.
Let us now calculate the velocity of the river of space in the Schwarzschild–de
Sitter spacetime. It consists of falling particles with a Lagrangian
11.9 The River of Space 307

1 1 ṙ 2
L = − f (r )c2 t˙2 + , (11.162)
2 2 f (r )

where the dots represent differentiation with respect to the proper time of a parti-
cle. Since L does not depend upon the time, the conjugate momentum to the time
coordinate is a constant of motion,

∂L
pt = = − f (r )c2 t˙ = constant. (11.163)
∂ t˙
From the four-velocity identity, we obtain

pt2 ṙ 2
− + = −c2 . (11.164)
f (r ) f (r )

Inserting the boundary condition ṙ (r0 ) = 0 gives

 
 R S 2/3
pt = −c f (r0 ), f (r0 ) = 1 − 3 . (11.165)
2R H

Hence

ṙ = ±c f (r0 ) − f (r ), (11.166)

with + for r > r0 and – for r < r0 . Inserting the expression (11.165) into Eq. (11.163)
gives
√
f (r0 )
t˙ = . (11.167)
f (r )

The coordinate velocity of the river of space is then

dr ṙ f (r )
= = ± f (r ) 1 − . (11.168)
dt t˙ f (r0 )

We now introduce an orthonormal basis vector field associated with observers at

rest in the coordinate system. The relationship between the coordinate basis vectors
and those of the orthonormal basis vector field is
1 
etˆ = √ et , er̂ = f (r )er . (11.169)
f (r )

Hence, the velocity of the river of space as measured by an observer at rest in the
static reference frame is
308 11 Sources of Gravitational Fields

Fig. 11.1 River of space in the Schwarzschild–de Sitter spacetime. The centre vertical line marks
the equilibrium radius, where the river of space is at rest, and the left and right lines mark the
horizons where the river of space flows with the velocity of light (From Braeck and Grøn [10])

dr̂ 1 dr f (r ) r 2 + 2r03 /r − 3r02

= =± 1− =± . (11.170)
dtˆ f (r ) dt f (r0 ) R 2H − 3r02

Inserting the horizon radii (11.159) gives dr̂ /dtˆ = ±c with minus at the
Schwarzschild horizon and plus at the de Sitter horizon. Hence, the river of space
flows at the velocity of light into a black hole around the origin of the coordinate
system and there is an outwards flow with the velocity of light at the outer horizon.
The function (11.170) is plotted in Fig. 11.1.
The river of space flows with superluminal velocity inside the Schwarzschild
horizon and outside the de Sitter horizon.
Note that for r r0 the expression (11.170) reduces to

dr̂ c c 
≈ H0 r, H0 =  ≈ =c . (11.171)
dtˆ R 2H − 3r02 R H 3

Hence, far outside the equilibrium radius, the velocity of the river of space is
proportional to the distance. This is the Hubble–Lemaître expansion law which will
be discussed in the next chapter. Thus, the flow of the river approaches the Hubble
flow far from the mass distribution.
Exercises 309

Exercises

11.1 The Schwarzschild–de Sitter metric

ShowSchwarzschild–de Sitter metric that the solution of Einstein’s field equations

with a cosmological constant in a static, spherically symmetric space is
 
RS r2 dr 2
ds = − 1 −
2
+ 2 c2 dt 2 + r2
+ r 2 d2 , (11.172)
r RH 1 − RrS + R 2H

where
√ R S = 2G M/c2 is the Schwarzschild radius of the central mass, and R H =
3/ is the de-Sitter horizon radius which is the horizon radius in the case that there
is no central mass, R S = 0.
11.2 A spherical domain wall described by the Israel formalism
Consider a static, spherically symmetric domain wall in empty space with mass
density σ and radius R. Show that the mass M of the Schwarzschild spacetime
outside the wall is

M = (1 − 2π σ R)4π σ R 2 . (11.173)

References

1. Israel, W.: Singular hypersurfaces and thin shells in general relativity. Nuovo Cimento 44B,
1–14 (1966)
2. Misner C.W., Thorne, K.S., Wheeler, J.A.: Gravitation. Princeton University Press (1973)
3. Lightman, A.P., Press, W.H., Price, R.H., Teukolsky, S.A.: Problem Book in Relativity and
Gravitation. Princeton University Press (1975)
4. An English translation of the 1917-article is: T. Levi-Civita, The physical reality of some
normal spaces of Bianchi. Gen. Rel. Grav. 43, 2307 (2011)
5. Bertotti, B.: Uniform electromagnetic field in the theory of general relativity. Phys. Rev. 116,
1331–1333 (1959)
6. Robinson, I.: A solution of the Maxwell-Einstein equations. Bull. Acad. Pol. Sci. Ser. Sci.
Math. Astr. Phys. 7, 351 (1959)
7. Grøn, Ø., Johannesen, S.: A solution of the Einstein-Maxwell equations describing conformally
flat spacetime outside a charged domain wall. Eur. Phys. J. Plus. 126, 89 (2011)
8. Lopez, C.A.: Extended model of the electron in general relativity. Phys. Rev. D30, 313–316
(1984)
9. Grøn, Ø.: New derivation of Lopez’s source of the Kerr-Newman field. Phys. Rev. D32, 1588–
1589 (1985)
10. Braeck, S., Grøn, Ø.: A river model of space. Eur. Phys. J. Plus 128, 24 (2013)
Chapter 12
Cosmology

Abstract The Lemaître–Friedmann–Robertson–Walker (LFRW) universe models

are deduced as solutions of Einstein’s field equations, and the Hubble–Lemaître
expansion law is found as a general property of these models. It is shown that the
cosmic redshift due to the expansion of the space contains both the kinematic Doppler
effect due to the velocity of the emitter relative to the observer and the gravitational
shift of wavelength for light moving vertically in a gravitational field. Several obser-
vational properties of a flat LFRW universe model with dust and Lorentz invariant
vacuum energy (LIVE) are deduced. Also anisotropic and inhomogeneous universe
models are considered. Finally some inflationary universe models are discussed, and
their predictions for some observational properties are confronted with observed
data.

12.1 Co-moving Coordinate System

In this chapter we will first consider expanding homogeneous and isotropic models
of the universe. They are called Lemaître-Friedmann–Robertson–Walker universe
models. We introduce an expanding frame of reference with the galactic clusters
as reference particles. Then we introduce a “co-moving coordinate system” in this
reference frame with spatial coordinates χ , ϑ, φ. Time measured on standard clocks
carried by the galactic clusters is used as coordinate time.
The line element can then be written in the form

ds2 = −c2 dt 2 + a(t)2 [R20 dχ 2 + r(χ )2 d 2 ], (12.1)

Here χ is a dimensionless radial coordinate, and R0 represents the present value

of the curvature radius of the 3-space. The coordinate time t is shown on standard
clocks co-moving with the reference particles of the expanding system, dχ = d =
0 and ds2 = − c2 dτ 2 = − c2 dt 2 . The function a(t) is called the scale factor, and t is
called cosmic time. The age of the universe is the present value of the cosmic time,
t0 . The scale factor tells how the reference particles move radially. It is normalized
to have the value 1 now, a(t 0 ) = 1. Hence the scale factor represents the ratio of the

© Springer Nature Switzerland AG 2020 311

Ø. Grøn, Introduction to Einstein’s Theory of Relativity,
Undergraduate Texts in Physics, https://doi.org/10.1007/978-3-030-43862-3_12
312 12 Cosmology

distance between the reference particles at an arbitrary point of time and their present
distance.
We shall now investigate whether the reference particles are freely moving, i.e.
whether they obey the geodesic equation. Since a reference particle is permanently
at rest in the coordinate system, and the derivative of the coordinate time with respect
to its proper time is dt/dτ = 1, its 4-velocity has components

dxμ dxμ
uμ = = = (c, 0, 0, 0). (12.2)
dτ dt
duμ
This applies at an arbitrary time, so dt
= 0. Hence, the geodesic equation

duμ μ
+ αβ uα uβ = 0 (12.3)
dt
reduces to
μ
tt = 0. (12.4)

Since
0 0 0
μ 1   
tt = g μν ( gνt,t + gtν,t + gtt,ν ) = 0, (12.5)
2
The geodesic equation is fulfilled for the reference particles. Hence they are freely
falling.

12.2 Curvature Isotropy—The Robertson–Walker Metric

We introduce an orthonormal form basis,

ωt̂ = dt, ωχ̂ = R0 a(t)dχ , ωθ̂ = a(t)r(χ )dθ,

ωφ̂ = a(t)r(χ ) sin θ dφ. (12.6)

Then we use Cartan’s 1. and 2. structure equations,

dωμ̂ = −μ̂v̂ ∧ ωv̂ , Rμ̂ν̂ = dμ̂ν̂ + λ̂μ̂ ∧ λ̂ν̂ (12.7)

to find the connection forms and the curvature forms. Calculations give (notation:
·
= dtd ,  = dχ
d
)
12.2 Curvature Isotropy—The Robertson–Walker Metric 313

Rt̂ î = aä ωt̂ ∧ ωî , ωî = ωχ̂ , ωθ̂ , ωφ̂ ,

 2 
χ̂ 
R ĵ = aȧ2 − rar 2 ωχ̂ ∧ ωĵ , ωĵ = ωθ̂ , ωφ̂ , (12.8)
 2 2

Rθ̂φ̂ = aȧ2 + r 21a2 − rr2 a2 ωθ̂ ∧ ωφ̂ .

The present curvature of 3-space (dt = 0) can be found by putting a = 1. That is

χ̂ r 
3 R ĵ = − ωχ̂ ∧ ωĵ ,
r

θ̂ 1 r 2
3 R φ̂ = − ωθ̂ ∧ ωφ̂ . (12.9)
r2 r2

The 3-space is assumed to be isotropic and homogeneous. This demands

r  1 − r 2 k
− = 2
= 2, (12.10)
r r R0

where R0 is the present value of the curvature radius of the 3-space. Here k is a
dimensionless constant which has the value 1 for a positively curved (spherical)
space, 0 for Euclidean (flat) space, and −1 for negatively curved (hyperbolic) space.

k
r  + r = 0 and r  = 1 − k(r/R0 )2 . (12.11)
R20

The solutions with the boundary conditions r(0) = 0, r  (0) = 1 are

r = R0 sinh χ , dr = R20 + r 2 dχ (k = −1),

r = R0 χ , dr =

R0 dχ (k = 0), (12.12)
r = R0 sin χ , dr = R20 − r 2 dχ (k = 1) .

In all three cases one may write

dr = R0 1 − k(r/R0 )2 dχ . (12.13)

We now substitute

dr 2
R20 dχ 2 = (12.14)
1 − k(r/R0 )2

into the line element (12.1). In standard coordinates it then takes the form
 
ds2 = −c2 dt 2 + R20 a2 (t) dχ 2 + Sk2 (χ )d 2 . (12.15)
314 12 Cosmology

where
⎧
⎨ sin χ , k = 1
Sk (χ ) = χ , k=0 . (12.16)
⎩
sinh χ , k = −1

With the radial coordinate r the line-element takes the form


dr 2
ds2 = −c2 dt 2 + a2 (t) + r 2
d 2
. (12.17)
1 − k(r/R0 )2

This is called the Robertson–Walker line element.

The 3-space has constant curvature and is spherical for k = 1, Euclidean for k =
0 and hyperbolic for k = −1. Universe models with k = 1 are known as “closed”,
and models with k = −1 are known as “open”. Models with k = 0 are called “flat”
even though also these models have curved spacetime.

12.3 Cosmic Kinematics and Dynamics

12.3.1 The Hubble–Lemaître Law

The physical distance at a point of time t, to a particle with an instantaneous coordinate

distance χ from an observer at the origin, is
√
l= gχχ χ = R0 a(t)χ . (12.18)

The value of χ determines which reference‚ particle’ (galactic cluster) we are

observing, and a(t) how it is moving. The velocity of the particle relative to the
observer at the origin is

ȧ
v = R0 (ȧχ + aχ̇ ) = R0 aχ + aχ̇ . (12.19)
a

The Hubble parameter is defined as

ȧ
H= . (12.20)
a
The present value of the Hubble parameter, H0 = H (t0 ) is called the Hubble
constant. A universe expanding with constant velocity has an age equal to the inverse
value of the Hubble constant. This is called the Hubble age of the universe,
12.3 Cosmic Kinematics and Dynamics 315

1
tH = . (12.21)
H0

We shall call lH = c tH the Hubble length.

A large number of observations have been performed in order to determine the
age of the universe and the value of the Hubble constant. Favoured values are t0 =
13.8 × 109 years, and H0 = 21.5 km/s per million light years corresponding to a
Hubble age tH = 13.85 × 109 years. It may be noted that the astronomers use the
length unit parsec which is equal to 3.26 light years. So they give the Hubble constant
in units km/s per Megaparsec, H0 = 70.1 km/s per Mpc.
Equation (12.19) may be written as

v = H l + aR0 χ̇ = vH + vP . (12.22)

Here

vH = H l (12.23)

is the velocity of the Hubble flow, which represent the expansion of the universe.
It says that the velocity of the Hubble flow is proportional to the distance from the
observer. This is the Hubble–Lemaître law. The general relativistic interpretation of
this law is that space expands.
Furthermore

vP = aR0 χ̇ (12.24)

is called the peculiar velocity of the considered particle. It is a velocity peculiar to

the considered particle due to a local gravitational field at the position of the particle.
In other words: vH is the velocity of space, and vP is a velocity through space. The
velocity of space is permitted to be larger than the velocity of light, which is the case
farther away from the observer than c/H , and the velocity though space is always
smaller than the velocity of light.

12.3.2 Cosmological Redshift of Light

We consider light emitted from a position with standard radial coordinate χe and
received by the observer at χ = 0. Let te be the period of the light as measured
in the emitter position at the emission time, and t 0 the period as measured in the
receiver position at the receiving time (Fig. 12.1).
Light follows curves with ds2 = 0, with dϑ = dφ = 0, and we have

cdt = −a(t)R0 dχ . (12.25)

316 12 Cosmology

Fig. 12.1 Cosmological redshift

Integration from emitter event to receiver event gives

t0 0
c
R0
dt
a(t)
=− dχ = χe ,
te χe
(12.26)
 0
t0 +t 0
c
R0
dt
a(t)
=− dχ = χe ,
te +te χe

which gives

t0
+ t0 t0
dt dt
− =0 (12.27)
a a
te + te te

or
t0
+t0 te
+te
dt dt
− = 0. (12.28)
a a
t0 te

During the integration time interval from t e to t e + t e the expansion factor a (t)
can be considered a constant with value a(t e ) and during the integration time interval
12.3 Cosmic Kinematics and Dynamics 317

from t 0 to t 0 + t 0 with value a(t 0 ), giving

te t0
= . (12.29)
a(te ) a(t0 )

t 0 and t e are periods of the light at the receiving and emitting time. Since the
wavelength of the light is λ = ct we have

λ0 λe
= . (12.30)
a(t0 ) a(te )

This can be interpreted as a “stretching” of the electromagnetic waves due to the

expansion of space [1]. The cosmological redshift is denoted by z and is given by

λ0 − λe a(t0 )
z= = − 1. (12.31)
λe a(te )

The scale factor is usually normalized to have a present value equal to one, a(t0 ) =
1, so that

1
1+z = . (12.32)
a

12.3.3 Cosmic Fluids

The energy–momentum tensor for a perfect fluid with mass density ρ as measured
by a co-moving observer in the fluid, and pressure p (no viscosity and no thermal
conductivity), is

Tμν = (ρ + p/c2 )uμ uν + pgμν . (12.33)

In an orthonormal basis

Tμ̂ν̂ = (ρ + p/c2 )uμ̂ uν̂ + pημ̂ν̂ , (12.34)

where ημ̂ν̂ is the Minkowski metric. We consider three types of cosmic fluid.
1. Dust: p = 0,

Tμ̂ν̂ = ρuμ̂ uν̂ . (12.35)

2. Radiation: p = 13 ρc2 ,
318 12 Cosmology

4
Tμ̂ν̂ = ρuμ̂ uν̂ + pημ̂ν̂
3
ρ
= (4uμ̂ uν̂ + ημ̂ν̂ ). (12.36)
3
The trace vanishes for radiation,

μ̂ ρ  μ̂ 
T = Tμ̂ = 4u uμ̂ + δμμ = 0. (12.37)
3

3. Lorentz invariant vacuum energy, LIVE, pL = −ρL c2 ,

Tμ̂ν̂ = −ρημ̂ν̂ . (12.38)

The density of LIVE can be related to the cosmological constant by

 = κ ρL . (12.39)

One has also introduced a more general type of vacuum energy given by the
equation of state

pφ = wρφ c2 , (12.40)

where φ represents a scalar field, and w is a factor which is often assumed to be

constant. In a homogeneous universe the pressure and density are given by

1 2 1
pφ = φ̇ − V (φ), ρφ = φ̇ 2 + V (φ), (12.41)
2 2
where V (φ) is the potential for the scalar field. Then we have

φ̇
1 2
− V (φ)
w= 2
. (12.42)
φ̇
1 2
2
+ V (φ)

The special case φ̇ = 0 gives LIVE with w = −1. The more general vacuum is
called “quintessence”.

12.3.4 Isotropic and Homogeneous Universe Models

We will discuss isotropic and homogenous universe models with perfect fluid and a
non-vanishing cosmological constant . Calculating the components of the Einstein
tensor from the line-element (12.17) we find in an orthonormal basis
12.3 Cosmic Kinematics and Dynamics 319

3ȧ2 3kc2 2ä ȧ2 kc2

Et̂ t̂ = + 2
, Em̂ m̂ = − − − . (12.43)
a2 R0 a2 a a2 R20 a2

Hence, the trace of the Einstein tensor is

μ̂ 6 
Eμ̂ = − 2
a ä + ȧ2 + kc2 /R20 . (12.44)
a
The components of the energy–momentum tensor of a perfect fluid in a co-moving
orthonormal basis are

Tt̂ t̂ = ρ, Tm̂m̂ = p. (12.45)

Hence the t̂ t̂ component of Einstein’s field equations is

ȧ2 + kc2 /R20

3 = κρ +  (12.46)
a2

where κ is Einstein’s gravitational constant as given in Eq. (7.37). The m̂m̂ component
of Einstein’s field equations is

ä ȧ2 kc2
−2 − 2 − 2 2 = κpc2 − , (12.47)
a a R0 a

where ρ is the energy density and p is the pressure. These equations are called the
Friedmann–Lemaître equations.
There are three cases of empty universe models, i.e. models with ρ = 0.
1. Empty, flat universe model with vanishing cosmological constant: ρ = p =
k =  = 0. Then Eq. (12.44) gives ȧ = 0. Integrating with the normalization
a(t0 ) = 1 gives a(t) = 1. The line-element then takes the form

ds2 = − c2 dt 2 + dr 2 + r 2 d 2 (12.48)

This represents the Minkowski spacetime in spherical coordinates.

2. Empty universe model with curved 3-space and vanishing cosmological constant:
ρ = p =  = 0, k = 0. Then Eq. (12.44) gives ȧ2 + kc2 /R20 = 0. This requires
k = − 1, i.e. hyperbolic 3-space. For an expanding universe model we then get
ȧ = 1. Integrating with the normalization a(t0 ) = 1 gives a(t) = t/t0 . The
line-element then takes the form
 2 
t dr 2
ds = −c dt +
2 2 2
+ r d .
2 2
(12.49)
t0 1 + (r/R0 )2

The universe model represented by this line-element is called the Milne

universe model.
320 12 Cosmology

In this universe model the Hubble parameter (12.20) is

1
H= . (12.50)
t
Hence the age of this universe model is related to the Hubble constant by

1
t0 = = tH . (12.51)
H0

Applying the coordinate transformation

 ct0 R
ct = c2 T 2 − R2 , r = √ (12.52)
c2 T 2 − R2

transforms the line-element to the form (12.48) which represents the Minkowski
spacetime. When there exists a coordinate transformation between two line-
elements they represent the same spacetime in two coordinate systems, which
may be co-moving with different reference frames. The coordinate R is co-moving
with a static reference frame, SR. The coordinate r is co-moving with another
reference frame, RF. We can find the motion of the reference particles of RF
relative to those of SR as follows.
Solving the last of the two transformation equations with respect to R gives

rcT
R=
(12.53)
c2 t02 + r 2

The reference particles of RF have r = constant. Hence for these particles

R increases linearly with T. This means that the frame in which r is co-moving
is expanding with a constant expansion velocity. Hence the Milne universe is
nothing but the Minkowski spacetime as described in an expanding reference
frame.
3. Expanding, flat, empty universe model with positive cosmological constant, ρ =
p = k = 0 ,  > 0 For this universe model Eq. (12.46) reduces to

ȧ 
H= = . (12.54)
a 3

Hence the Hubble parameter is constant. Integration with a(t0 ) = 1 gives

a(t) = eH (t−t0 ) . (12.55)

The line-element takes the form

 
ds2 = − c2 dt 2 + e2H (t−t0 ) dr 2 + r 2 d 2 . (12.56)
12.3 Cosmic Kinematics and Dynamics 321

The spacetime represented by this line-element is called the De Sitter

spacetime.
It was represented by De Sitter in 1917 as a static and spherically symmetric
solution of Einsteins’s equations with a cosmological constant for empty space.
Five years later it was shown that the reference particles of the static frame were
not freely moving, and that when transforming the solution to a coordinate system
co-moving with freely moving reference particles, one obtained the line-element
above. Also Georges Lemaître showed in 1933 that the cosmological constant
could be interpreted to represent the constant energy density of Lorentz Invariant
Vacuum Energy, LIVE.
Equations (12.46) and (12.47) then give

κ 
ä = − a(ρ + 3p/c2 ) + . (12.57)
6 3
Inserting the gravitational mass–density ρG from Eq. (11.13) this equation takes
the form
κ 
ä = − aρG + . (12.58)
6 3

Inserting p = wρc2 into Eq. (11.13) gives

ρG = (1 + 3w)ρ, (12.59)

which is negative for w < −1/3, i.e. according to Eq. (12.42), for φ̇ 2 < V (φ).
Special cases:
• matter in the form of dust: w = 0, ρG = ρm ,
• radiation: w = 13 , ρG = 2ρR ,
• LIVE: w = −1, ρG = −2ρL .
In a universe dominated by LIVE the acceleration of the cosmic expansion is
κ
ä = aρL > 0, (12.60)
6
that is accelerated expansion. This means that LIVE acts upon itself with repulsive
gravitation.
The field equations can be combined into
 2
ȧ κ  kc2
H ≡
2
= ρm + − 2 2 , (12.61)
a 3 3 R0 a

where ρm is the density of matter in the form of dust,

322 12 Cosmology

 = κρL , (12.62)

and ρL is the constant density of LIVE. Then we may write

κ kc2
H2 = ρ − 2 2, (12.63)
3 R0 a

where ρ = ρm +ρL . The critical density ρcr is the density in a universe with Euclidean
space-like geometry, k = 0, which gives

κρcr = 3H 2 . (12.64)

The present value of the Hubble parameter, i.e. the Hubble constant, is

H0 = (κ/3)ρcr 0 (12.65)

We introduce the density parameters

ρm ρL
m = , Λ = . (12.66)
ρcr ρcr

Note that these quantities are only well defined for expanding universe models
with H = 0. Furthermore we introduce a dimensionless parameter which represents
the curvature of 3-space

kc2
k = − . (12.67)
R20 a2 H 2

with present value

kc2
k0 = − . (12.68)
R20 H02

Note that k < 0 means positive spatial curvature, and k > 0 means negative
spatial curvature. Equation (12.63) can now be written as

m + L + k = 1. (12.69)

This equation shows that an empty, expanding universe has k = 1; i.e. an

empty, expanding universe has negative spatial curvature. It may be noted that since
the present value of the scale parameter is normalized to a(t0 ) = 1 the curvature
parameter may be written

k0 H02 1 − 0 H02

k = 2 2
= , 0 = m0 + L0 . (12.70)
a H a2 H 2
12.3 Cosmic Kinematics and Dynamics 323

Hence a sufficiently large mass density 0 > 1 gives spherical spatial geometry,
0 = 1 gives Euclidean spatial geometry, and 0 < 1 gives hyperbolic spatial
geometry. Inserting k0 = 1−0 into Eq. (12.68) and solving the resulting equation
with respect to R0 gives

lH
R0 = √ . (12.71)
|1 − 0 |

The Hubble length in our universe is 13.8 billion light years. Observations of tem-
perature fluctuations in the CMB-radiation show that 0 ≈ 1. Hence the curvature
radius of the 3-space is extremely large, much larger than the Hubble length of the
universe.

12.3.5 Cosmic Redshift

We consider a homogeneous and isotropic universe with standard coordinates so

that the line-element has the form (12.1). An observer is at χ = 0 and an emitter at
χ = χe . A light signal is emitted at a point of time te and received by the observer
at the present time t0 . The scale factor is normalized to have the value a(t0 ) = 1
at the present time. The emitter is relatively close to the observer in the sense that
the expansion velocity of the emitter relatively to the observer is much smaller than
the velocity of light, vH  c. Due to the Hubble–Lemaître law that the expansion
velocity is proportional to the distance, this condition takes the form

vH = H l = R0 H (te )a(te )χe  c, (12.72)

The equation of motion of a light signal is given in Eq. (12.25). Integrating between
the emitter and the observer gives

χe
c(t0 − te ) = R0 a(t)dχ. (12.73)
0

To second order in c(t0 − te ) this gives

t0
c dt 1
R0 χe = = c(t0 − te ) + H0 c2 (t0 − te )2 , (12.74)
a(t) 2
te

where we have used that a(t0 ) = 1 and ȧ(t0 ) = H0 . In the condition (12.72) it is suffi-
cient to keep the first term. Hence the condition takes the form H (te )a(te )(t0 − te ) 
1. In this inequality we can approximate the values of the Hubble parameter and the
scale factor at the emission point of time by their values at the time that the light
324 12 Cosmology

signal is received by the observer. Hence we arrive at the condition H0 (t0 − te )  1.

In the following we shall include only terms up to 2 order in H0 (t0 − te ).
The cosmic redshift is given in Eq. (12.32). Introducing the deceleration
parameter,

a ä
q=− , (12.75)
ȧ2
we have

ȧ H ȧ2 ä H2 H 2q
ż = − = − , z̈ = 2 − = 2 + , (12.76)
a2 a a3 a2 a a

giving ż(t0 ) = − H0 , z̈(t0 ) = H02 (2 + q0 ). Hence a Taylor expansion of z about

t = t0 to 2 order in H0 (t0 − te ) gives

1
z(te ) = H0 (t0 − te ) + 1 + q0 H02 (t0 − te )2 . (12.77)
2

We now introduce an observer G instantaneously at the position of the emitter at

the point of time te , but permanently at rest relative to the observer O at the origin.
The velocity of the emitter relative to this observer is

Ve = ȧ(te ) χe . (12.78)

Due to the Doppler effect G measures a redshift

zD = [(1 + Ve /c)/(1 − Ve /c)]1/2 − 1 (12.79)

of the emitted signal. Since |Ve |  c we can use the two first terms of a Mc Laurin
expansion, giving

Ve 1 Ve 2
zD ≈ + . (12.80)
c 2 c

A Taylor expansion of ȧ about t0 gives to 2 order in H0 (t0 − te ),

ȧ(te ) ≈ ȧ(t0 ) + ä(t0 ) (t0 − te ) = H0 + q0 H02 (t0 − te ). (12.81)

Inserting the expressions (12.74) and (12.81) in Eq. (12.78) we get


1
Ve ≈ H0 c(t0 − te ) + + q0 H02 c2 (t0 − te )2 . (12.82)
2

Substituting this into Eq. (12.80) gives

12.3 Cosmic Kinematics and Dynamics 325

zD = H0 (t0 − te ) + (1 + q0 )H02 (t0 − te )2 . (12.83)

Inserting an observer in the homogeneous LFRW universe model breaks the homo-
geneous symmetry. Hence the gravitational potential depends upon the distance from
the observer. From Eq. (12.57) with a vanishing cosmological constant we see that in
the theory of relativity it is natural to generalize the Newtonian gravitational potential
of Eq. (1.29) and define a relativistic gravitational potential φG by
 
∇ 2 φG = 4π G ρ + 3p/c2 (12.84)

Using Eq. (12.57) we then have

∇ 2 φG = 3qH 2 . (12.85)

The potential at the present time is given by

∇ 2 φG = 3q0 H02 . (12.86)

Integrating this equation with the conditions (dφG /dr)r=0 = φG (0) = 0 gives

1
φG = q0 H02 r 2 . (12.87)
2
Note that there is accelerated expansion for q0 < 0 and decelerated expansion for
q0 > 0. Hence the gravitational potential decreases in the outwards direction when
there is accelerated expansion and increases for decelerated expansion.
The distance from the emitter to the observer at a time close to the present time,
when the scale factor is equal to one, is

r = c(t0 − te ). (12.88)

Hence the gravitational potential at the emitter is

1
φG = q0 H02 c2 (t0 − te )2 . (12.89)
2
We now consider the Newtonian limit where the contribution of the pressure to
the acceleration of gravity is negligible and the relativistic gravitational potential
reduces to the Newtonian gravitational potential, φG → φ, and we can neglect the
expansion of the region between the emitter and the observer. Then the gravitational
shift of the wavelength of light is given by Eqs. (5.108) and (5.111),

λ0 (gtt )0
zG = −1= − 1. (12.90)
λe (gtt )e
326 12 Cosmology

According to Eq. (9.20) the tt-component of the metric tensor is then


2φ
gtt = − 1 + 2 , (12.91)
c

where 2|φ|/c2  1. We then get


1 + 2φ0 /c2
zG = − 1. (12.92)
1 + 2φe /c2

Noting that φ0 = 0 and making a series expansion to 1 order in φ/c2 we have

zG = −φe /c2 ≈ −φG /c2 . (12.93)

Inserting the potential (12.89) gives

1
zG = − q0 H02 (t0 − te )2 . (12.94)
2
It follows from Eqs. (12.78), (12.84) and (12.95) that in the limit of small cosmic
redshifts,

z = zD + z G . (12.95)

In this limit the cosmic redshift due to the expansion of the universe is equal to
the sum of the kinematic redshift due to the Doppler effect and the gravitational shift
of wavelength due to the gravitational field between the emitter and the observer.
The Doppler effect is a redshift since the emitter moves away from the observer,
but the gravitational effect is a blueshift in a universe with retarded expansion due
to attractive gravitation, since the light moves downwards in this gravitational field.
In a universe with accelerated expansion due to the repulsive gravity of the vacuum
energy, light moves upwards in the gravitational field between the emitter and the
observer, and the gravitational effect is a redshift.
Hence the cosmic redshift due to the expansion of the space contains both the
kinematic Doppler effect due to the velocity of the emitter relative to the observer
and the gravitational shift of wavelength for light moving vertically in a gravitational
field.

12.3.6 Energy–Momentum Conservation

From the 2 Bianchi identity and Einstein’s field equations, it follows that the energy–
momentum density tensor is covariantly divergence free. The time component
12.3 Cosmic Kinematics and Dynamics 327

expresses the equation of continuity and takes the form

[(ρ + p/c2 )ut̂ uν̂ ];ν̂ + (pηt̂ ν̂ );ν̂ = 0. (12.96)

Since ut̂ = 1, um̂ = 0 and ηt̂ t̂ = −1, ηt̂ m̂ = 0, we get

(ρ + p/c2 ). + (ρ + p/c2 )uν̂;ν̂ − ṗ = 0 (12.97)

or

ρ̇ + (ρ + p/c2 )(uν̂,ν̂ + uα̂  ν̂α̂ν̂ ) = 0. (12.98)

Here uν̂,ν̂ = 0 and  t̂ t̂ t̂ = 0. Calculating  m̂t̂ m̂ from Cartan’s 1. structure equation

in the form
μ̂
d ωμ̂ =  α̂ β̂
ωα̂ ∧ ωβ̂ (12.99)

with the basis forms (12.6) we get

φ̂ ȧ
 m̂t̂ m̂ =  r̂ t̂ r̂ +  θ̂t̂ θ̂ +  t̂ φ̂
=3 . (12.100)
a
Inserting this into Eq. (12.98) gives

(ρc2 a3 ). + p(a3 ). = 0. (12.101)

Let V = a3 be a co-moving volume in the universe and U = ρc2 V the energy in

the co-moving volume. Then we may write

dU + pdV = 0. (12.102)

This is the first law of thermodynamics for an adiabatic expansion. It follows that
the universe expands adiabatically. The adiabatic equation can be written as

ρ̇ ȧ
= −3 . (12.103)
ρ + p/c 2 a

With p = wρc2 we get

dρ da
= −3(1 + w) . (12.104)
ρ a

Assuming that w = constant this equation can be integrated to give

328 12 Cosmology

 −3(1+w)
ρ a
ln = ln . (12.105)
ρ0 a0

It follows that
 −3(1+w)
a
ρ = ρ0 or ρa3(1+w) = constant (12.106)
a0

We have three particularly important special cases:

• Matter in the form of dust: w = 0 gives

ρm a3 = constant. (12.107)

Thus, the mass in a co-moving volume is constant.

• Radiation: w = 1/3 gives

ρrad a4 = constant. (12.108)

Thus, the radiation energy density decreases faster than the density of the dust
when the universe is expanding. The energy in a co-moving volume is decreasing
because of the thermodynamic work on the surface of a co-moving volume. In a
remote past, the density of radiation was greater than the density of dust.
• LIVE: w = −1 gives ρL = constant. The vacuum energy in a co-moving volume
is increasing proportionally to a3 . In spite of this energy is conserved locally
because of the negative work performed at a co-moving, expanding surface. This
work transfers energy from the region outside the surface to the region inside it,
maintaining the constant value of the energy density of LIVE in the expanding
universe. Since we can choose the surface to have an arbitrarily large radius this
amounts to extracting energy from an infinitely remote region to the region in
a finite distance from the observer at the origin. Hence there is a difficulty with
global energy conservation in cosmology.
The Friedmann–Lemaître Eq. (12.63) can be written

H2  0  k0
2
= rad
4
+ m3 0 + 2 + L0 . (12.109)
H0 a a a

where rad 0 and m 0 are the present values of the radiation density and the density
of dust, respectively, and 0 is given in Eq. (12.70). Since H = (1/a)(da/dt), the
scale factor can be found as a function of time by integrating the equation

da rad0 
= H0 + m 0 + k0 + L0 a2 . (12.110)
dt a2 a
12.3 Cosmic Kinematics and Dynamics 329

By means of Eq. (12.32) this equation can be expressed in terms of the redshift of
the source. In this way we obtain an equation for the Hubble parameter at the time
of emission of light emitted from a source with redshift z and received at the present
time,

H2
= rad 0 (1 + z)4 + m0 (1 + z)3 + k0 (1 + z)2 + L0 . (12.111)
H02

Similarly this equation can be integrated to find the redshift, and hence the scale
factor as a function of time by utilizing that

1 dz
H =− . (12.112)
1 + z dt

In combination with Eq. (12.111) this gives an expression for age of the universe
at the time a source with observed redshift z emits the observed light, i.e. the emission
point of time,

∞
dz
tE (z) = tH  .
(1 + z) rad0 (1 + z) + m0 (1 + z)3 + k0 (1 + z)2 + L0
4
z
(12.113)

where the Hubble age tH is defined in Eq. (12.21). This is called the age-redshift
relationship. The age of the universe is found by letting the redshift of the source be
zero, t0 = t(0). Taking the integral from 0 to z gives the time taken for the radiation
to move from the source to the observer and arrive at the present time. This is called
the lookback time.
Equation (12.113) is in general an elliptic integral, but it can be integrated in
terms of elementary functions in the case that the cosmological constant vanishes, so
that  = 0. This gives the age-redshift relationship for a curved, dust-dominated
universe.
 √  
1 + m0 z m0 1 − m0
tE (z) = tH − arc sinh . (12.114)
(1 − m0 )(1 + z) (1 − m0 )3/2 m0 (1 + z)

The age of this universe is

  
tH m0 1 − m0
t0 = tE (0) = 1− √ arc sinh . (12.115)
1 − m0 1 − m0 m0
330 12 Cosmology

12.4 Some LFRW Cosmological Models

12.4.1 Radiation-Dominated Universe Model

The energy–momentum tensor for radiation is trace free. According to the Einstein’s
field equations the Einstein tensor must then be trace free. Using Eq. (12.44) we then
have

aä + ȧ2 + kc2 /R20 = 0,

(12.116)
(aȧ + ktc2 /R20 )· = 0.

Integration gives

aȧ + ktc2 /R20 = B. (12.117)

Integrating once more gives

1 2 1 kc2 2
a + t = Bt + C. (12.118)
2 2 R20

The initial condition a(0) = 0 gives C = 0. Hence

a= 2Bt − kt 2 c2 /R20 . (12.119)

For k = 0 we have

√ B
a= 2Bt, ȧ = . (12.120)
2t

The expansion velocity reaches infinity at t = 0, lim ȧ = ∞ (Fig. 12.2).

t→0
According to the Stefan–Boltzmann law we have

ρR = σ T 4 . (12.121)

Combining this with Eq. (12.108) we have

T 4 a4 = constant, (12.122)

or
T0
T= , (12.123)
a
12.4 Some LFRW Cosmological Models 331

Fig. 12.2 Expansion of a radiation dominated universe. In a radiation-dominated universe the

expansion velocity reaches infinity at t = 0

where T0 is the temperature of the radiation at the present time. Equations (12.120)
and (12.123) give

t0
T = T0 . (12.124)
t

12.4.2 Dust-Dominated Universe Model

From the first of the Friedmann–Lemaître equations we have

kc2 κ
ȧ2 + 2
= ρa2 . (12.125)
R0 3

We now introduce a so-called conformal time η by

dt d 1 d
= a(η) ⇒ = . (12.126)
dη dt a dη

Hence,

da 1 da
ȧ = = . (12.127)
dt a dη
332 12 Cosmology

We also introduce
κ
A≡ ρm0 . (12.128)
3
The first Friedmann–Lemaître equation then gives

kc2 κ κ
aȧ2 + a = ρm a3 = ρm0 = A. (12.129)
R20 3 3

Using η we get
 2
c2
1
a
da
dη
= R20
(A − ka),
(12.130)

√

da
dη
= ca
R0
A
a
−k = c
R0
aA 1 − ka
A
,

where we have chosen the positive root. We now introduce u by


a
a = Au , u = 2
, (12.131)
A

and get

da du
= 2Au , (12.132)
dη dη

which together with Eq. (12.131) gives

du c
√ = dη. (12.133)
1 − ku 2 2R 0

This equation will first be integrated for k = −1. Then


du cη
√ = + K, (12.134)
1+ u2 2R0
or
cη
arsinh(u) = + K. (12.135)
2R0

The condition u(0) = 0 gives K = 0. Hence


1 cη 1 cη
a = sinh2 = cosh −1 (12.136)
A 2R0 2 R0
12.4 Some LFRW Cosmological Models 333

or

A cη
a= cosh −1 . (12.137)
2 R0

From Eqs. (12.128), (12.65) and (12.66) we have

κ ρm0
A= ρm0 = H02 = H02 m0 . (12.138)
3 ρcr0

From Eqs. (12.67) and (12.69) we get

R20 H02
k= (m0 − 1). (12.139)
c2
Hence, the scale factor of the negatively curved, dust-dominated universe model
is

1 m0 cη
a(η) = cosh −1 . (12.140)
2 1 − m0 R0

Inserting this into Eq. (12.126) and integrating with t(0) = η(0) leads to

m0 cη cη
t(η) = sinh − . (12.141)
2H0 (1 − m0 )3/2 R0 R0

Integrating Eq. (12.118) for k = 0 leads to an Einstein–de Sitter universe

 23
t
a(t) = . (12.142)
t0

Finally integrating Eq. (12.125) for k = 1 gives


1 m0 cη
a(η) = 1 − cos , (12.143)
2 m0 − 1 R0

m0 cη cη
t(η) = − sin . (12.144)
2H0 (m0 − 1)3/2 R0 R0

This is a parametric representation of a cycloid (Fig. 12.3).

In the Einstein–de Sitter model the Hubble parameter is

21
H= . (12.145)
3t
The age of this universe model is
334 12 Cosmology

Fig. 12.3 Expansion of matter dominated universe models. For k = 1 the density is larger than the
critical density, and the universe is closed. For k = 0 we have ρ = ρ cr and the expansion velocity
of the universe will approach zero as t → ∞. For k = −1 we have ρ < ρcr . The universe is then
open and will continue expanding forever

t0 = (2/3) tH . (12.146)

With a Hubble age tH = 13.85 billion years the age of the flat, dust-dominated
universe becomes t0 = 9.2 billion years, in conflict with the age t0 = 13.8 years
based upon a large set of different observations (Fig. 12.4).

Fig. 12.4 Hubble age. Here t H is the Hubble age, i.e. the age of the universe if the expansion had
been constant. But in the dust-dominated universe the expansion rate was faster closer to the Big
Bang, so the age is lower
12.4 Some LFRW Cosmological Models 335

Example 12.4.1 (Lookback Time for Flat Dust-dominated Universe) In the case of a
flat universe 0 = 1. With L0 = 0 Eq. (12.113) with modified limits corresponding
to the lookback time reduces to
z  
dz 2 1
tLB = tH 5
= tH 1 − 3 . (12.147)
(1 + z) 2 3 (1 + z) 2
0

Using Eq. (12.146) for the age of a flat, dust-dominated universe the expression
for the lookback time can be written
 
1
tLB = t0 1 − . (12.148)
(1 + z)3/2

Hence, the redshift of an object with lookback time tLB is

1
z= 2/3 − 1. (12.149)
1− tLB
t0

12.4.3 Transition from Radiation-Dominated

to Matter-Dominated Universe

We consider the early universe filled with radiation and matter, but where vacuum
energy can be neglected. The universe is assumed to be flat. Then Friedmann’s 1.
equation takes the form
κ
ȧ2 = (ρm + ρr )a2 . (12.150)
3
For matter,

ρm a3 = ρm0 . (12.151)

For radiation,

ρr a4 = ρr0 . (12.152)

Hence
κ
a2 ȧ2 = (ρm0 a + ρr0 ). (12.153)
3
The present values of the critical density and the density parameters are
336 12 Cosmology

ρm0 ρr0
κρcr0 = 3H02 , m0 = , r0 = , (12.154)
ρcr0 ρcr0

giving

aȧ = H0 (m0 a + r0 )1/2 . (12.155)

Integration with a(0) = 0 leads to

3/2
4 r0 2 (m0 a − 2r0 )(m0 a + r0 )1/2
H0 t = + . (12.156)
3 2m0 3 2m0

From Eqs. (12.155) and (12.156) it follows that at the transition time teq when
ρm = ρr , the scale factor has the value

ρr0 r0
aeq = = . (12.157)
ρm0 m0

Inserting this into Eq. (12.156) gives

2 √  3/2
teq = 2 − 2 r0 tH . (12.158)
3 2m0

The microwave background radiation has a temperature 2.73 K corresponding to

a density parameter r0 = 8.4 × 10−5 . In a flat universe without vacuum energy
m0 = 1 − r0 . Inserting tH = 13.85 × 109 years then leads to teq = 47 × 103 years.

12.4.4 The de Sitter Universe Models

These are universe models without radiation and matter, containing only LIVE.
There are three models depending upon the spatial curvature. For these models the
Friedmann Eq. (12.46) takes the form

 2 kc2
ȧ2 − a =− 2. (12.159)
3 R0

This equation has the following solution

⎧


⎪ R 3
⎪ c0  cosh

t ,k=1
⎨ √
3
a(t) = e
/3 t .
, 
 k = 0
(12.160)
⎪
⎪
⎩ R0 3 sinh 
t , k = −1
c  3
12.4 Some LFRW Cosmological Models 337

Fig. 12.5 The scale factor of

the De Sitter universe models

The scale factor as a function of time is shown for these universe models in
Fig. 12.5. Note that the flat De Sitter universe is infinitely old. It had no Big Bang.

12.4.5 The Friedmann–Lemaître Model

The dynamics of galaxies and clusters of galaxies has made it clear that far stronger
gravitational fields are needed to explain the observed motions than those produced
by visible matter [2]. At the same time it has become clear that the density of this
dark matter is only about 30% of the critical density, although it is a prediction by the
usual versions of the inflationary universe models that the total density of all that is
contained in the universe ought to be equal to the critical density [3]. Also the recent
observations of the temperature fluctuations of the cosmic microwave radiation have
shown that space is either flat or very close to flat [4–6]. The energy which fills up
the universe to the critical density must be evenly distributed in order not to affect
the dynamics of the galaxies and the clusters.
Furthermore, in 1998 observations of supernovae of type-I a with high cosmic
redshifts indicated that the expansion of the universe is accelerating [7, 8]. This was
explained as a result of repulsive gravitation due to some sort of vacuum energy.
Thereby the missing energy needed to make space flat was identified as vacuum
energy. Hence, it seems that we live in a flat universe with vacuum energy having a
338 12 Cosmology

density around 70% of the critical density and with matter having a density around
30% of the critical density.
Until the discovery of the accelerated expansion of the universe the standard
model of the universe was assumed to be the Einstein–de Sitter model, which is a
flat universe model dominated by cold matter. Now it seems that we must replace
this model with a new “standard model” containing both dark matter and vacuum
energy [9].
Recently several types of vacuum energy or so-called quintessence energy have
been discussed [10, 11]. However, the most simple type of vacuum energy is the
Lorentz invariant vacuum energy, LIVE, which has constant energy density during
the expansion of the universe [12, 13]. This type of energy can be mathematically
represented by including a cosmological constant in Einstein’s gravitational field
equations. The flat universe model with cold dark matter and this type of vacuum
energy is the Friedmann–Lemaître model. This universe model is usually denoted
the Lambda-Cold-Dark-Matter ( CDM) model, but a better name would be the
LIVE-Cold-Dark-Matter (LCDM) model.
The field equations for the flat Friedmann–Lemaître universe model are found by
putting k = p = 0 in Eq. (12.62). This gives

ä ȧ2
2 + 2 = ,  = κρL (12.161)
a a
Integration leads to

 3
aȧ2 = a + K, (12.162)
3
where K is a constant of integration. Since the amount of matter in a volume co-
moving with the cosmic expansion is constant, ρm a3 = ρm0 a03 , where the index 0
refers to values at the present time. Normalizing the expansion factor so that a0 =
1 and comparing Eqs. (12.46) and (12.162) then give K = (κ/3)ρm0 . Introducing
a new variable x by a3 = x 2 and integrating once more with the initial condition
a(0) = 0, we obtain

3K t
a3 = sinh2 , (12.163)
 tL

where
2 2 2
tL = √ = √ = √ tH , (12.164)
3 3H0 L0 3 L0

where L denotes LIVE. Since there is, at the present time (July 2019), a rather large
disagreement between the value of H0 as determined from supernova observations
and from observations of temperature fluctuations in the cosmic microwave back-
ground radiation, we shall here determine the parameters of the universe model
12.4 Some LFRW Cosmological Models 339

from the observed values of the age of the universe and the value of the density
parameters of LIVE. From a large number of different types of observations we have
t0 = (13.8 ± 0.02) × 109 years and L0 = 0.694 ± 0.007 [14].
Since the present universe model has flat space, the total density is equal to the
critical density, i.e. m +L = 1. Equation (12.162) with the normalization a(t0 ) = 1
gives 3H02 = 3K + . Equation (12.46) with k = 0 gives κρ0 = 3H02 − . Hence
K = κρ0 /3 and

3K κρ0 ρ0 m0
= = = . (12.165)
  ρL L0

In terms of the values of the relative densities at the present time the expression
for the scale factor then takes the form

t m0 1 − L0
a = A1/3 sinh2/3 , A= = . (12.166)
tL L0 L0

√ Inserting L0 = 0.694 gives A = 0.44. Using the identity sinh(x/2) =

(cosh x − 1)/2 this expression may be written as
  
A 2t
a =
3
cosh −1 . (12.167)
2 tL

The age t 0 of the universe

√ is found from a(t0 ) = 1, which, by use of the formula
arctanh x = arsinh (x/ 1 − x2 ), leads to the expression

L0 
t0 = tL arsinh = tL artanh L0 . (12.168)
m0

We use this to determine tL , writing

t0
tL = √ . (12.169)
artanh L0

Inserting t0 = 13.8 × 109 years and L0 = 0.694 gives tL = 11.5 × 109 years.
Substituting the values A = 0.44 and tL = 11.5 × 109 years into Eq. (12.166) gives

t
a = 0.76 sinh2/3 1.2 . (12.170)
t0

This function is plotted in Fig. 12.6.

A consistency requirement on the Hubble constant H0 follows from Eqs. (12.164)
and (12.158). If the general theory of relativity is correct, and if it is correct to describe
the universe as homogeneous and isotropic on a large scale, and dominated by dust
and LIVE, then the Hubble constant must obey the equation
340 12 Cosmology

Fig. 12.6 The scale factor of the flat CDM universe model

√
2 artanh L0 1
H0 = √ . (12.171)
3 L0 t0

Inserting t0 = 13.8 × 109 years and L0 = 0.694 gives H0 = 68.3 km/s per Mpc.
This is the predicted value of the Hubble constant. The corresponding Hubble age is
tH = (977/H0 ) × 109 years, giving tH = 14.3 × 109 years.
The ratio of the age of the universe and its Hubble age, t0 /tH = H0 t0 , is plotted
in Fig. 12.7. The age of the universe increases with increasing density of vacuum
energy. In the limit that the density of the vacuum approaches the critical density,
there is no dark matter, and the universe model approaches the de Sitter model with

Fig. 12.7 The ratio of age and Hubble age of the flat CDM universe model
12.4 Some LFRW Cosmological Models 341

exponential expansion and no Big Bang. This model behaves in the same way as the
steady-state cosmological model and is infinitely old.
Invoking the phenomenon of perfect inertial dragging it has been argued that the
validity of the principle of relativity for rotational motion requires that the age of the
universe must be equal to its Hubble age, H0 t0 = 1 [15]. According to Eq. (12.171)
this requires that
  
3
tanh L0 = L0 . (12.172)
2

The positive, real solution of this equation is L0 = 0.737.

Using Eqs. (12.166) and (12.168) we obtain a nice form of the age-redshift
relationship, giving the emission point of time of an object with redshift z,


L0 1
arsinh m0 (1+z)3/2
tE = 
 t0 . (12.173)
L0
arsinh m0

Inserting L0 = 0.7 and m0 = 0.3 with t 0 = 13.8·109 years gives

1.53
tE = 12.5 × 109 arsinh years. (12.174)
(1 + z)3/2

This is plotted in Fig. 12.8.

Fig. 12.8 Age-redshift relation of the flat CDM universe model. The emission time is given in
billion years
342 12 Cosmology

Fig. 12.9 The Hubble parameter of the flat CDM universe model

The time which the radiation has taken in moving from the source to the observer
and being observed at the present time, i.e. the lookback time, is tLB = t0 − tE . The
distance which the radiation has travelled is tLB light years.
The Hubble parameter as a function of time is

2 t
H= coth . (12.175)
3 tL tL

Inserting t0 = 1.2 tL we get H t0 = 0.8 coth(1.2 t/t0 ), which is plotted in Fig. 12.9.
It may be noted that the Hubble parameter is given as a function of the redshift
in Eq. (12.111) with RAD 0 = 0 and 0 = 1. The graph in Fig. 12.8 shows that
the Hubble parameter decreases all the time and approaches a constant value H∞ =
2/3 tL in the infinite future. The Hubble age is

tH = (3/2)tL L0 . (12.176)

Inserting numerical values gives tH = 14.4 · 109 years. In this universe model the
age of the universe is nearly as large as the Hubble age, while in the Einstein–de
Sitter model the corresponding age is t0ED = (2/3)tH = 9.5 × 109 years. The reason
for this difference is that in the Einstein–de Sitter model the expansion is decelerated
all the time, while in the Friedmann–Lemaître model the repulsive gravitation due
to the vacuum energy has made the expansion accelerate lately (see below). Hence,
for a given value of the Hubble parameter the previous velocity was larger in the
Einstein–de Sitter model than in the Friedmann–Lemaître model.
A dimensionless quantity representing the rate of change of the cosmic expansion
velocity is the deceleration parameter, which is defined as
12.4 Some LFRW Cosmological Models 343

Fig. 12.10 The deceleration parameter of the flat CDM universe model

q = − ä/aH 2 . (12.177)

For the present universe model the deceleration parameter as a function of time is

1 
q= 1 − 3 tanh2 (t/tL ) , (12.178)
2
which is shown graphically in Fig. 12.10.
From Eqs. (12.169) and (12.178) the deceleration parameter at the present point
of time may be written as

1
q0 = − (3 L0 − 1). (12.179)
2
With  L0 = 0.7 we get q0 = − 0.55.
The inflection point of time t 1 when deceleration turned into acceleration is given
by q = 0. This leads to
 √ 
t1 = t artanh 1/ 3 (12.180)

or is expressed in terms of the age of the universe

 √ 
artanh 1/ 3
t1 = √ t0 . (12.181)
artanh 0

The corresponding cosmic redshift is

344 12 Cosmology
 1/3
1 2L0
z(t1 ) = −1= − 1. (12.182)
a(t1 ) 1 − L0

Inserting L0 = 0.7 gives t1 = 0.54t0 and z(t1 ) = 0.67.

The results of analyzing the observations of supernova SN 1997 at z = 1.7, cor-
responding to an emission time tE = 0.30t0 = 4.5 × 109 years, have provided evi-
dence that the universe was decelerated at that time [16]. Turner and Riess [17] have
recently argued that the other supernova data favour a transition from deceleration
to acceleration for a redshift around z = 0.5.
It may be noted from Eq. (12.57) that in a flat universe with dust and LIVE and
vanishing cosmological constant, the transition from deceleration to acceleration
happens when the gravitational mass density vanishes, i.e. when ρm + ρL + 3pL = 0.
Since pL = − ρL , this gives ρm (t1 ) = 2ρL (t1 ). Hence the constant density of LIVE
is half the density of the dust at this transition.
Note that the expansion velocity given by Hubble’s law, v = H l, always decreases
as seen from Fig. 11.8. This is the velocity away from the Earth of the cosmic fluid
at a fixed physical distance l from the Earth ȧ. The quantity ȧ, on the other hand, is
the velocity of a fixed fluid particle co-moving with the expansion of the universe. If
such a particle accelerates, the expansion of the universe is said to accelerate. While
Ḣ tells how fast the expansion velocity changes at a fixed distance from the Earth, the
quantity ä represents the acceleration of a free particle co-moving with the expanding
universe. The connection between these two quantities is ä = a(Ḣ + H 2 ).
The ratio of the inflection point of time and the age of the universe, as given in
Eq. (12.168), is depicted graphically as a function of the present relative density of
vacuum energy in Fig. 12.11. The turnover point of time happens earlier the greater
the vacuum density is. The change from deceleration to acceleration would happen
at the present time if L0 = 1/3.

Fig. 12.11 Point of time for deceleration–acceleration turnover

12.4 Some LFRW Cosmological Models 345

Fig. 12.12 Cosmic redshift at the deceleration–acceleration turnover

The redshift of the inflection point given in Eq. (12.181) as a function of vacuum
energy density is plotted in Fig. 12.12 Note that the redshift of future points of time is
negative, since then a > a0 . If L0 < 1/3 the transition to acceleration will happen
in the future.
The critical density is

ρcr = ρL tanh−2 (t/tL ). (12.183)

This is plotted in Fig. 12.13. The critical density decreases with time.
Equation (12.182) shows that the density parameter of LIVE is

Fig. 12.13 Critical density as function of time

346 12 Cosmology

L = tanh2 (t/tL ), (12.184)

which is plotted in Fig. 12.14.

The density of LIVE approaches the critical density. Since the density of LIVE
is constant, this is better expressed by saying that the critical density approaches the
density of the vacuum energy. Furthermore, since the total energy density is equal to
the critical density all the time, this also means that the density of matter decreases
faster than the critical density. The density of matter as function of time is

ρm = ρ sinh−2 (t/tL ), (12.185)

which is shown graphically in Fig. 12.15.

The density parameter of the dust as function of time is

m = cosh−2 (t/tL ), (12.186)

which is shown in Fig. 12.16. Adding the density parameters of the expressions
(12.183) and (12.185) we get the total density parameter TOT = m +  = 1.
The universe became vacuum dominated at a point of time t 2 when ρL (t2 ) =
ρm (t2 ). From Eq. (12.184) it follows that this point of time is given by sinh(t2 /tL ) = 1.
Using Eq. (12.142) we get

arsinh(1)
t2 = √ t0 . (12.187)
artanh( L0 )

It follows that the corresponding redshift is

Fig. 12.14 The density parameter of vacuum energy as function of time

12.4 Some LFRW Cosmological Models 347

Fig. 12.15 The density of matter as function of time

Fig. 12.16 The density parameter of matter as a function of time

z(t2 ) = A−1/3 − 1. (12.188)

Inserting L0 = 0.7 gives t2 = 0.73 t0 and z(t2 ) = 0.32. Hence, the transition to
accelerated expansion happens before the universe becomes vacuum dominated.
As mentioned above, many different observations indicate that we live in a uni-
verse with critical density, where cold matter contributes with about 30% of the
density and vacuum energy with about 70%. Such a universe is well described by
the Friedmann–Lemaître universe model which has been presented above.
However, this model is not quite without problems in explaining the observed
properties of the universe. In particular there is now much research directed at solving
348 12 Cosmology

the so-called coincidence problem. As we have seen, the density of the vacuum energy
is constant during the expansion, while the density of the matter decreases inversely
proportional to a volume co-moving with the expanding matter. Yet, one observes that
the density of matter and the density of the vacuum energy are of the same order of
magnitude at the present time. This seems to be a strange and unexplained coincidence
in the model. Also just at the present time the critical density is approaching the
density of the vacuum energy.
There is now a great activity in order to try to explain these coincidences by
introducing more general forms of vacuum energy called quintessence and with a
density determined dynamically by the evolution of a scalar field [18].
However, the simplest type of vacuum energy is LIVE. One may hope that a
future theory of quantum gravity may settle the matter and let us understand the
vacuum energy. In the meantime we can learn much about the dynamics of a vacuum-
dominated universe by studying simple and beautiful universe models such as the
Friedmann–Lemaître model.

12.4.6 Flat Universe with Dust and Phantom Energy

We shall here consider a flat LFRW universe model with dust and phantom
energy [18]. Differentiating Eq. (12.63) with k = 0 and using Eq. (12.103) we
get

2Ḣ + 3H 2 = − κp/c2 . (12.189)

The pressure is due to the phantom energy which can be described as due to a
scalar field φ with a potential V (φ). The density and pressure of the phantom energy
are

κρφ = −φ̇ 2 /2 + V (φ), κpφ /c2 = −φ̇ 2 /2 − V (φ). (12.190)

Equation (12.189) then takes the form

1 2
2Ḣ + 3H 2 = φ̇ + V (φ). (12.191)
2
The evolution equation for the phantom energy field is

φ̈ + 3H φ̇ = V  (φ), (12.192)

where the prime denotes differentiation with respect to φ.

It is now assumed that the time derivative of the phantom field is proportional to
the Hubble parameter,
12.4 Some LFRW Cosmological Models 349

ȧ
φ̇ = −αH = −α . (12.193)
a
where α is a positive constant. Integration gives

φ = φ0 − α ln a. (12.194)

Hence the phantom field decreases during the expansion of the universe.
We shall now deduce a differential equation for the relationship between the
Hubble parameter and
the scalar field itself (not its derivative). We use that

d d
= φ̇ . (12.195)
dt dφ

Differentiating Eq. (12.191) and substituting for V  from Eq. (12.192) we then
get
    
2H − 3H φ̇ 2 + 2 H  − φ̇ φ̈ + 6HH  φ̇ = 0. (12.196)

Using the relationship (12.193) in this equation gives


3 3
HH  + H 2 + α − HH  − H 2 = 0. (12.197)
α 2

Introducing y = H 2 this equation can be written as


3 
y + α − y − 3y = 0. (12.198)
α

This is a 2 order linear differential equation with constant coefficients. The general
solution is

H 2 (φ) = Ae3φ/α + Be− αφ . (12.199)

It follows from Eqs. (12.191), (12.193) and (12.199) that the potential of the
phantom field is

α2 α2
V (φ) = 3 + Be− αφ − Ae3φ/α . (12.200)
2 2

Using Eqs. (12.190), (12.193) and (12.200) we find the energy density of the
phantom field

κρφ = 3Be− αφ − α 2 Ae3φ/α . (12.201)

350 12 Cosmology

Similarly the pressure is

 
κpφ = − 3 + α 2 Be− αφ . (12.202)

The mass density of the dust is

   
κρm = κ ρcr − ρφ = 3H 2 − κρφ = 3 − α 2 Ae3φ/α . (12.203)

Hence the constant A represents the dust. Note that there is no future singularity
in this model.
Inserting the expression (12.199) for H into Eq. (12.193) gives a differential
equation for the time evolution of the phantom field, but this equation has no analytical
solution. The equations simplify if we assume that there is no dust, only the phantom
field. Then A = 0, and the equation of state parameter of the phantom field is

pφ α2
w= = −1 − , (12.204)
ρφ 3

which is less than minus one. This is characteristic of phantom energy. In the present
case the potential of the phantom field is

α2
V (φ) = 3 + Be− αφ , (12.205)
2

and the Hubble parameter is

√
H (φ) = B e− (α/2)φ . (12.206)

Letting H0 = H (0) we have B = H02 . Inserting Eq. (12.206) into Eq. (12.193)
gives

e(α/2)φ φ̇ = −α H0 . (12.207)

Integration with φ(0) = φ0 gives

 2/α
(α/2)φ0 α2
φ(t) = ln e − H0 t . (12.208)
2

The field vanishes at a point of time

2  α t0 
t0 = e 2 − 1 tH , (12.209)
α 2

where tH = 1/H0 . After this point of time the field is negative and diverges at
12.4 Some LFRW Cosmological Models 351

2 (α/2)φ0
t1 = e tH . (12.210)
α2
Inserting Eq. (12.208) into (12.206) gives

ȧ H0
H= = (α/2)φ   . (12.211)
a e 0 − α 2 /2 H0 t

Hence H (0) = e− (α/2)φ0 H0 which is finite, H (t0 ) = H0 and H (t1 ) = ∞, which

is called the Big Rip. Integration with a(t0 ) = 1 gives
 − 2/α2
α2
a(t) = e(α/2)φ0 − H0 t . (12.212)
2

Hence at the Big Rip a(t1 ) = ω. It should be noted that the presence of dust
removes the Big Rip.

12.5 Flat Anisotropic Universe Models

We shall consider anisotropic world models with flat 3-space of Bianchi type-I
following Ref. [19]. The line-element has the form
 2  2  2
ds2 = −c2 dt 2 + a12 dx1 + a22 dx2 + a32 dx3 . (12.213)

The directional Hubble parameters are defined by

ȧi
Hi = . (12.214)
ai

The mean scale factor is defined as

a = (a1 a2 a3 )1/3 . (12.215)

A “volume scale factor”,

V = a3 = a1 a2 a3 (12.216)

will also be useful. The average Hubble parameter is

1
H= (H1 + H2 + H3 ). (12.217)
3
These definitions give
352 12 Cosmology

1
H= (ln V )· = (ln a)· . (12.218)
3
The anisotropy parameter is defined by

3 
1  Hi − H 2
A= . (12.219)
3 i=1 H

Einstein’s field equations with a cosmological constant for this class of universe
models filled by a perfect fluid may be written

· κ 3p
(ln V ) + +
H12 +
H22 = ρ + 2 + ,
H32 (12.220)
2 c
1 κ p
(V Hi )· = ρ − 2 + . (12.221)
V 2 c
We shall consider some special cases.
1. Anisotropic empty universe with vanishing cosmological constant
In this case Eqs. (12.220) and (12.221) reduces to

(ln V )· + H12 + H22 + H32 = 0, (12.222)

1
(V Hi )· = 0. (12.223)
V
The last equation gives
αi
Hi = , (12.224)
V
This gives


3


3 αi
i=1
3H = Hi = . (12.225)
i=1
V

Combining this with Eq. (12.218) gives


3
V̇ = αi . (12.226)
i=1

Integration with V (0) = 0 gives

12.5 Flat Anisotropic Universe Models 353


3
V = αi t. (12.227)
i=1

Inserting this into Eq. (12.224) gives

ȧi αi 1 pi αi 3
Hi = = 3 = , pi = 3 , pi = 1. (12.228)
i=1 αi i=1 αi
ai t t i=1

Integration gives

ai = ai0 t pi . (12.229)

Inserting this into Eq. (12.222) gives


3
pi2 = 1. (12.230)
i=1

The directional and average Hubble parameters are

pi 1
Hi = , H= . (12.231)
t 3t
Inserting this into Eq. (12.219) gives

1
3
A= (3pi − 1)2 = 2. (12.232)
3 i=1

The universe described in this subsection is called the Kasner universe. It repre-
sents the Minkowski universe as described from a reference frame with anisotropic
expansion. Due to the conditions (12.228) and (12.230) it is not possible to have
equal values for the three constants pi . Hence there is no isotropic special case of
the Kasner universe. In particular, the Milne universe is not a special case since the
Kasner universe is spatially flat, while the Milne universe has curved 3-space. It may
be noted that the Kasner universe has a constant anisotropy parameter equal to 2.
2. Anisotropic universe dominated by LIVE
In this case Eqs. (12.220) and (12.221) take the form

(ln V )· · + H12 + H22 + H32 = , (12.233)

1
(V Hi )· = , (12.234)
V
354 12 Cosmology

where the cosmological constant represents the constant density of LIVE. It follows
from Eqs. (12.217), (12.218) and (12.234) that

Ḣ + 3H 2 = . (12.235)

The solution of this equation with the boundary condition H (0) = H0 is


H = H0 coth(3HL t), HL = /3. (12.236)

Inserting this into Eq. (12.218) gives

(ln V )· = 3H0 coth(3HL t). (12.237)

The general solution of this equation is

V = V0 sinhH0 /HL (3HL t). (12.238)

In order to have a simple illustrating example we choose H0 = HL giving

V = V0 sinh(3HL t). (12.239)

Equation (12.234) can be written

V̇
Ḣi + Hi = 3HL2 . (12.240)
V
Inserting (12.239) gives

Ḣi + 3HL coth(3HL t) Hi = 3HL2 . (12.241)

The general solution of this equation is

ȧi Ci
Hi = = + HL coth(3HL t), (12.242)
ai sinh(3HL t)

where Ci are integration constants. Integration with the initial condition ai (0) = 0
gives
 
3 3
HL t cosh 3 − pi
2
ai = 21/3 sinhpi HL t , (12.243)
2 2
 
where pi = 1
3
1+ Ci
. Inserting (12.236) and (12.242) into Eq. (12.219) and using
HL
 
that Ci = (3pi − 1)HL together with the conditions 3i=1 pi = 1 and 3i=1 pi2 = 1,
give the anisotropy parameter
12.5 Flat Anisotropic Universe Models 355

Fig. 12.17 Evolution of

anisotropy in a
LIVE-dominated
Bianchi-type I universe.
Initially the anisotropy
parameter A has the maximal
value of the empty Kasner
universe, but it decreases
rapidly towards zero

2
A= . (12.244)
cosh (3HL t)
2

This is shown graphically in Fig. 12.17. For 3HL t 1 the anisotropy param-
eter approaches zero exponentially. Hence the LIVE causes the universe model to
isotropize. At the beginning, 3HL t → 0, and the expression (12.244) gives the value
A = 2 of the Kasner universe.

12.6 Inhomogeneous Universe Models

We shall here consider the Lemaître–Tolman–Bondi (LTB) class of universe mod-

els following [20–23]. These universe models are inhomogeneous and spherically
symmetric. The line-element has the form

F 2 (r, t) 2
ds2 = −c2 dt 2 + dr + F 2 (r, t)d 2 , (12.245)
1 − k(r)

where k(r) is an arbitrary function of r which represents the curvature of 3-space.

We use the notation  = d/dr and · = d/dt.
356 12 Cosmology

12.6.1 Dust-Dominated Model

The Einstein equations for the dust-dominated LTB-universe models can be written
as

k k
H⊥2 + 2H H⊥ + + = κρm , (12.246)
F2 FF 
k k
(1 − 3q)2H⊥2 − 2H H⊥ + 2 − = −κρm . (12.247)
F2 FF 
where the Hubble parameters and the deceleration parameter are

Ḟ Ḟ  F F̈
H⊥ = , H =  , q = − 2 . (12.248)
F F Ḟ
Adding Eqs. (12.246) and (12.247) and using the Definition (12.248) gives

2F F̈ + Ḟ 2 = −k, (12.249)

Integration of this equation leads to

α k
H⊥2 = 3
− 2, (12.250)
F F
where α is a function of r. The functions α and k are not determined by the field equa-
tions. They must be specified as boundary conditions. Differentiating Eq. (12.250)
with respect to r and inserting the result into Eq. (12.247) gives

α
κρm = . (12.251)
F 2F 
Substituting Eqs. (12.249) and (12.250) into the expression (12.248) for the
deceleration parameter yields
α
q= . (12.252)
2(α − kF)

Let F(r, t = 0) = F0 (r). We now introduce a conformal time η by k 1/2 dt =

− (cF/R0 ) d η, where R0 is the present value of the curvature radius of the 3-space.
The solution of Eqs. (12.246) and (12.247) with k < 0 can then be written
  
α cη cη α cη
F =− cosh − 1 + F0 cosh + 1− sinh , (12.253)
2k R0 R0 kF0 R0
12.6 Inhomogeneous Universe Models 357
    
√ α cη cη cη α cη
−k t = − sinh − + F0 sinh + 1− cosh −1
2k R0 R0 R0 kF0 R0
(12.254)

The dust-dominated solution given in Eqs. (12.140) and (12.141) for a homoge-
neous dust-dominated LFRW universe with negative spatial curvature is found by
choosing F0 = 0 and α = H02 m0 r 3 , k = −H02 (1 − m0 )r 2 . Note that in this case
F = r a(η).

12.6.2 Inhomogeneous Universe Model with Dust and LIVE

We shall here consider an inhomogeneous generalization of the LCDM-universe

which was presented in Sect. 12.4.5. In this case Eq. (12.246) is generalized to

k k
H⊥2 + 2H H⊥ + + = κ(ρm + ρL ), (12.255)
F 2 FF 
and the acceleration equation is

F̈ k
2 + H⊥2 + 2 = κρL . (12.256)
F F
Here ρL = ρL0 is the constant density of LIVE. Equation (12.250) now takes the
form
α k κ
H⊥2 = 3
− 2 + ρL . (12.257)
F F 3
Equation (12.251) is still valid. Combining Eqs. (12.255) and (12.256) we get

F̈ F̈  κ
2 +  = (2ρL − ρm ). (12.258)
F F 2
The present values of the density parameters are defined by the equations

α = H⊥0
2
F02 m , (12.259)

k = H⊥0
2
F02 (m + L − 1), (12.260)

where F0 = F0 (r) = F(r, t0 ), H⊥0 = H⊥0 (r) = H (r, t0 ) and L0 = L0 (r) =
κρL /3H⊥0
2
(r). With these definitions Eq. (12.257) of the Hubble parameter H⊥ takes
the form
358 12 Cosmology
  3  2 1/2
Ḟ F0 F0
= H⊥ (r, t) = H0 m0 + L0 + (1 − m0 − L0 ) .
F F F
(12.261)

It follows that the scale factor F is given implicitly as a function of time by


dF
H0 t =  F0 3  F0 2 !1/2 . (12.262)
m0 F
+ L0 + (1 − m0 − L0 ) F

We now consider an inhomogeneous universe corresponding to the LCDM-

universe where the density parameters of dust and LIVE obey the condition
m0 + L0 = 1. Then Eq. (12.262) reduces to

dF
H0 t =  F0 3 !1/2 . (12.263)
m0 F + L0

Integrating this equation with the initial condition F(r, 0) = 0 gives

  3 
2 L0 F 2
H0 t = √ arsinh . (12.264)
3 L0 m0 F0

Hence
 1/3  
m0 23
F = F0 sinh 2 L0 H0 t . (12.265)
L0 2

In the present universe model m0 , L0 and H0 are functions of r, while in the
LCDM-universe they are constants.

12.7 The Horizon and Flatness Problems

12.7.1 The Horizon Problem

The cosmic microwave background (CMB) radiation from two points A and B in
opposite directions has the same temperature. This means that it has been radiated
by sources of the same temperature at these points. Thus, the universe must have
been in thermic equilibrium at the decoupling time, td = 3 × 105 years. This implies
that points A and B, “at opposite sides of the universe” as seen by an observer, had
been in causal contact already at that time. That is, a light signal must have had time
12.7 The Horizon and Flatness Problems 359

to move from A to B during the time from t = 0 to t = 3 × 105 years. The points A
and B must have been within each other’s horizons at the decoupling.
Consider a photon moving radially in space described by the Robertson–Walker
metric (12.17) with k = 0. Light follows a null-geodesic curve; i.e. the curve is
defined by ds2 = 0. We get (using units so that c = 1),

dt
dr = . (12.266)
a(t)

The particle horizon (also called the cosmological horizon) is the maximum dis-
tance from which light from particles could have travelled to the observer during the
time which the universe has existed. It represents the boundary between the observ-
able and the unobservable regions of the universe. Its distance at the present epoch
defines the size of the observable part of the universe. The coordinate distance from
an observer at the origin of the coordinate system to the particle horizon at the time
t is

t
dt
r = . (12.267)
a(t)
0

The physical radius of the particle horizon is

t
dt
lPH = a(t)r = a(t) . (12.268)
a(t)
0

To find a quantitative expression for the “horizon problem”, we may consider a

model with critical mass–density (Euclidean space-like geometry). Using p = wρ
and k = 0, Eq. (12.63) takes the form

ȧ κρ0 − 3 (1+w)
= H0 a− 2 (1+w)
3
= a 2 (12.269)
a 3

Integration with the condition a(t 0 ) = 1 gives

 3(1+w)
2
t 2 1 2
a= , t0 = = tH . (12.270)
t0 3(1 + w) H0 3(1 + w)

where tH = 1/H0 is the Hubble age of the universe. Inserting this into the expression
(12.268) and integrating gives

3w + 3
lPH = t. (12.271)
3w + 1
360 12 Cosmology

Hence, the radius of the particle horizon increases proportionally to the cosmic
time. Equation (12.270) also shows that the age of a flat radiation-dominated universe
is only half of the Hubble age.
Let us call the volume inside the horizon the “horizon volume” and denote it by
VPH . From Eq. (12.271) it follows that VPH ∝ t 3 . At the decoupling time, the horizon
volume may therefore be written as
 3
td
(VPH )d = V0 , (12.272)
t0

where V 0 is the size of the present horizon volume. Events within this volume are
causally connected, and a volume of this size may be in thermal equilibrium at the
decoupling time.
Let (V0 )d be the size, at the decoupling, of the part of the universe that corre-
sponds to the present horizon volume, i.e. the observable universe. For our Euclidean
universe, Eq. (12.271) holds, giving
 w+1
2
a3 (td ) td
(V0 )d = V0 = V0 . (12.273)
a3 (t0 ) t0

From Eqs. (12.272) and (12.273), we get

 − 3w+1
(V0 )d td w+1
= . (12.274)
(VPH )d t0

Using that td = 10−4 t0 and inserting w = 0 for dust, we find

(V0 )d
= 104 . (12.275)
(VPH )d

Thus, there was room for 104 causally connected areas at the decoupling time
within the region which represents our observable universe. Points at opposite sides
of our observable universe were therefore not causally connected at the decoupling,
according to the Friedmann models of the universe. These models therefore cannot
explain that the temperature of the radiation from such points is the same.

12.7.2 The Flatness Problem

According to Eqs. (12.67) and (12.68), the total mass parameter  = ρ/ρcr is given
by
12.7 The Horizon and Flatness Problems 361

kc2
−1= . (12.276)
R20 a2 H 2

Using the expansion factor (12.269) for a universe near critical mass–density, we
get
 2( 3w+1
3w+3 )
−1 t
= . (12.277)
0 − 1 t0

For a radiation-dominated universe, we get

−1 t
= . (12.278)
0 − 1 t0

Measurements indicate that 0 − 1 is of the order of magnitude 1. The age of the

universe is about t0 = 1017 s. When we stipulate initial conditions for the universe,
it is natural to consider the Planck time, tP = 10−43 s, since this is the limit of
the validity of general relativity. At earlier time, quantum effects will be important,
and one cannot give a reliable description without using quantum gravitation. The
stipulated initial condition for the mass parameter then becomes that  − 1 is of
order 10−60 at the Planck time. Such an extreme fine tuning of the initial value of the
universe’s mass–density cannot be explained within standard Big Bang cosmology.
Since a universe with critical density has Euclidean spatial geometry and is called
flat, this fine tuning requirement is called the “flatness problem”.

12.8 Inflationary Universe Models

12.8.1 Spontaneous Symmetry Breaking and the Higgs

Mechanism

The particles responsible for the electroweak force are the W ± and Z 0 bosons. They
are massive, causing the weak force to only have short-distance effects. This was
originally a problem for the quantum field theory describing this force, since it made
it difficult to create a renormalizable theory. This was solved by Higgs and Kibble
in 1964 by introducing the so-called Higgs mechanism.
The main idea is that the massive bosons W ± and Z 0 are originally massless, but
are given a mass by interacting with a Higgs field φ. The effect causes the mass of the
particles to be proportional to the value of the Higgs field in vacuum. It is therefore
necessary that the Higgs field has a value different from zero in the vacuum (the
vacuum expectation value must be nonzero).
Let us see how the Higgs field can get a nonzero vacuum expectation value.
The important thing for our purpose is that the potential of the Higgs field may be
362 12 Cosmology

temperature dependent. Let us assume that the potential of the Higgs field is described
by the function

1 2 2 1 4
V (φ) = μ φ + λφ , (12.279)
2 4

where the sign of μ2 depends on whether the temperature is above or below a critical
temperature Tcr . This sign has an important consequence on the shape of the potential
V. The potential is shown in Fig. 12.18 for two different temperatures. For T >
Tcr , μ2 > 0, and the shape is like in Fig. 12.18a, and there is a stable minimum for
φ = 0. However, for T < Tcr , μ2 < 0, the shape is like in Fig. 12.18b. In this case
|μ|
the potential has stable minima for φ = ±φ0 = ± √ λ
and an unstable maximum at
φ = 0.
The “real” vacuum state of the system is at a stable minimum of the potential.
For T > Tcr , the minimum is in the “symmetric” state φ = 0. On the other hand, for
T < Tcr this state is unstable. It is therefore called a “false vacuum”. The system will
move into one of the stable minima at φ = ±φ0 . When the system is in one of these
states, it is no longer symmetric under the change of sign of φ. Such a symmetry,
which is not reflected in the vacuum state, is called spontaneously broken. From
Fig. 12.16b we see that the energy of the false vacuum is larger than for the real
vacuum.
The central idea, from which the “inflationary cosmology” originated, was to take
into consideration the consequences of the unified quantum field theories, the gauge
theories, at the construction of relativistic models for the early universe. According
to the Friedmann models, the temperature was extremely high in the early history
of the universe. If one considers Higgs fields associated with GUT models (grand

Fig. 12.18 Inflationary potentials. The shape of the potential depends on the sign of μ2 . a Higher
temperature than the critical, with μ2 > 0. b Lower temperature than the critical, with μ2 < 0
12.8 Inflationary Universe Models 363

Fig. 12.19 Temperature dependence of Higgs potential

unified theories), one finds a critical temperature Tcr corresponding to the energy
kTcr = 1014 GeV, where k is the Boltzmann’s constant. Before the universe was
about t1 = 10−35 s old, the temperature was larger than this. Thus, the Higgs field
was in the symmetric ground state. According to most of the inflation models, the
universe was dominated by radiation at this time.
When the temperature decreases, the Higgs potential changes. This could happen
as shown in Fig. 12.19. Here, there is a potential barrier at the critical temperature,
which means that there cannot be a classical phase transition. The transition to the
stable minimum must happen by quantum tunnelling. This is called a first-order
phase transition. Alan Guth’s original inflation model [23] was based on a first-order
phase transition.

12.8.2 Guth’s Inflationary Model [24]

According to most of the inflationary models, the universe was dominated by radia-
tion during the time before 10−35 s. The universe was then expanding so fast that there
was no causal contact between the different parts of the universe that became our
observable universe. Probably, the universe was rather inhomogeneous, with consid-
erable space-like variations in temperature. There were also regions of false vacuum,
with energy densities characteristic of the GUT energy scale, which also controlled
the critical temperature. While the energy density of the radiation decreased quickly,
as a−4 , the energy density of vacuum was constant. At the time t = 10−35 s, the
energy density of the radiation became less than that of the vacuum.
364 12 Cosmology

At the same time the potential started to change such that the false vacuum went
from being stable to unstable. Thus, there was a first-order phase transition to a
real vacuum. Because of the inhomogeneity of the universe’s initial condition, this
happened with different speeds at differing places. The potential barrier slowed down
the process, which happened by tunnelling, and the universe was at several places
considerably under cooled. Then there appeared “bubbles” dominated by the energy
of the false vacuum. These regions acted upon themselves with repulsive gravity.
By integrating the equation of motion for the expansion factor in such a vacuum-
dominated bubble, one gets

8π Gρcr
a=e , H =
Ht
. (12.280)
3

By inserting the GUT value above, we get H = 6.6 × 1034 s−1 , i.e.H −1 = 1.5 ×
−35
10 s. With reference to field theoretical works by Sidney Coleman and others,
Guth reasoned that a realistic duration of the nucleation process happening during
the phase transition is 10−33 s. During this time, the expansion factor increases by a
factor of 1028 . This vacuum-dominated epoch is called the inflationary era.
Let us look closer at what happened with the energy of the universe in the course
of its development, according to the inflationary models. To understand this we first
have to consider what happened at the end of the inflationary era. When the Higgs
field reached the minimum corresponding to the real vacuum, it started to oscillate.
According to the quantum description of the oscillating field, the energy of the false
vacuum was converted to radiation and particles. In this way the equation of state
for the energy dominating the development of the expansion factor changed from
p = −ρ, characteristic of LIVE, to p = (1/3)ρ, characteristic of radiation.
The energy density and the temperature of the radiation then increased enor-
mously. Before and after this short period around the time t = 10−33 s the radiation
energy increased adiabatically, such that ρa4 = constant. According to Stefan–Boltz-
mann law of radiation, ρ ∝ T 4 . Therefore, aT = constant during adiabatic expan-
sion. This means that during the inflationary era, while the scale factor increased
exponentially, the energy density and temperature of radiation decreased exponen-
tially. At the end of the inflationary era, the radiation was reheated so that it returned
to approximately the energy it had when the inflationary era started.
It may be interesting to note that the Newtonian theory of gravitation does not
allow an inflationary era, since stress has no gravitational effect according to it.

12.8.3 The Inflationary Models’ Answers to the Problems

of the Friedmann Models

The horizon problem will be investigated here in the light of the inflationary universe
models. The problem was that there was room for about 10,000 causally connected
12.8 Inflationary Universe Models 365

regions inside the region defined by our presently observable universe at the decou-
pling time when the universe became transparent to the background radiation. Let
us calculate the horizon radius lh and the radius a of the region presently within the
particle horizon, lPH = 15 × 109 ly = 1.5 × 1026 cm, at the time t1 = 10−35 s when
the inflation started. From Eq. (12.120) for the radiation-dominated period before
the inflationary era, one gets

lPH = 2t1 = 6 × 10−25 cm. (12.281)

The radius, at time t 1 , of the region corresponding to our observable uni-

verse is found by using a ∝ eHt during the inflation era from t1 =
10−35 s to t2 = 10−33 s, a ∝ t 1/2 in the radiation-dominated period from t2 to t3 =
1011 s and a ∝ t 2/3 in the matter-dominated period from t 3 until now, t0 = 1017 s.
This gives
 1/2  2/3
eHt1 t2 t3
a1 = Ht lPH (t0 ) = 1.5 × 10−28 cm. (12.282)
e 2 t3 t0

Hence at the beginning of the inflationary era the horizon radius, lPH , was larger
than the radius a1 of the region corresponding to our observable universe. The whole
of this region was then causally connected, and thermic equilibrium was established.
This equilibrium was preserved, and there was thermal equilibrium at the decoupling
time about 400,000 years later. This explains the observed isotropy of the cosmic
background radiation and solves the horizon problem.
We will now consider the flatness problem. This problem was the necessity, in the
Friedmann models, of fine tuning the initial density in order to obtain the closeness
of the observed mass–density to the critical density. Again, the inflationary models
give another result, making fine tuning unnecessary.
Inserting the scale factor (12.280) into Eq. (12.276), we get

k −2Ht
−1= e , (12.283)
H2
where H is constant and given in Eq. (12.280). The ratio of  − 1 at the end of the
inflationary era to the beginning of the inflationary era becomes

2 − 1
= e−2H (t2 −t1 ) = 10−56 . (12.284)
1 − 1

Contrary to the Friedmann models, where the mass–density moves away from
the critical density as time increases, the density approaches the critical density
exponentially during the inflationary era. Within a large range of initial conditions,
this means that according to the inflation models the universe should still have a
density very close to the critical density.
366 12 Cosmology

12.8.4 Dynamics of the Inflationary Era

During the inflationary era the evolution of the universe is assumed to be dominated
by a scalar field φ which is called the inflaton field. The first Friedmann equation is

κ κ 1 2
H = ρ=
2
φ̇ + V , (12.285)
3 3 2

where ρ is the energy and V = V (φ) the potential of the inflaton field. The continuity
equation is

ρ̇ + 3H ρ = −3Hp. (12.286)

It follows from these equations that


ρ̇ = − 3κρ(ρ + p). (12.287)

The equation for the evolution of the inflaton field which generates the dark energy
causing repulsive gravity during the inflationary era is

φ̈ + 3H φ̇ = −V  , (12.288)

where V  = dV /dφ.
It follows from the second Friedmann equation that the acceleration of the cosmic
expansion is given by

ä κ
= − (ρ + 3p). (12.289)
a 6
The inflaton field is often described as a perfect fluid with density and pressure

1 2 1
ρ= φ̇ + V, p = φ̇ 2 − V. (12.290)
2 2
Hence, the fluid obeys the equation of state

(1/2)φ̇ 2 − V
p = wρ, w = . (12.291)
(1/2)φ̇ 2 + V

It interpolates between a Lorentz invariant vacuum energy (LIVE) with w = −1

for a constant inflaton field and a Zel’dovich fluid with w = 1 for a flat potential with
V = 0. Solved with respect to φ̇ 2 the second of these equations gives

1+w
φ̇ 2 = 2V. (12.292)
1−w
12.8 Inflationary Universe Models 367

This equation shows that V > 0 for |w| < 1.

The acceleration Eq. (12.289) of the scale factor then takes the form

ä κ 
= − φ̇ 2 − V . (12.293)
a 3
Differentiating Eq. (2.285) and inserting Eq. (2.289) gives

Ḣ = −(κ/2)φ̇ 2 , (12.294)

or

φ̇ = −(2/κ)H  , (12.295)

where H  = dH /d φ = Ḣ /φ̇. Equation (2.294) shows that H is a decreasing function

of time. It follows from Eqs. (2.285) and (2.295) that

κ 2 V = 3κH 2 − 2H 2 . (12.296)

Equation (2.294) shows that the Hubble parameter is constant and there is expo-
nential expansion for a constant inflaton field. This represents the case where the
inflaton field behaves like LIVE with a constant density, which may be represented
by a cosmological constant. Equation (2.294) implies that the Hubble parameter is a
decreasing function of time for a variable scalar field.
During most of the inflationary era, i.e. except during the transient phases at the
beginning and the end of the era, the scalar field changes very slowly so that φ̈  H φ̇.
If the potential V is not too small, the condition φ̇  V may also be satisfied. Then
w ≈ −1 which means that the inflaton field behaves like LIVE with approximately
constant energy density, and with exponential expansion of the space during most of
the inflationary era.
In the so-called slow-roll approximation we shall assume that φ̈  H φ̇, but not
in general that φ̇ 2  V . Then Eq. (12.288) reduces to

V
φ̇ ≈ − . (12.297)
3H
Equations (12.291) and (12.294) give

1+w
Ḣ = −κ V . (12.298)
1−w

It follows from Eqs. (12.285) and (12.294) that

κ V = Ḣ + 3H 2 . (12.299)
368 12 Cosmology

Hence

Ḣ = −(3/2)(1 + w)H 2 . (12.300)

Integration of this equation for constant w = −1 gives

 3(1+w)
2
t
a = a1 . (12.301)
t1

Hence, power law expansion corresponds to a constant equation of state parameter

w = −1 during the inflationary era, and exponential expansion to w = − 1. Inserting
the first of Eq. (2.291) into Eq. (12.287) gives

ρ̇ = − 3κρ(1 + w)ρ. (12.302)

Integrating this equation for w = −1 with ρ(0) = ρ0 gives

ρ0
ρ= √ 2 . (12.303)
1 + (1/2)(1 + w) 3κρ0 t
√
Hence for ρ0 t MP the energy density of an inflaton field with constant equa-
tion of state parameter, w = −1, decreases approximately inversely proportionally
to the square of time.
We define the slow-roll parameters ε , η by
 2
1 V 1 V 
ε≡ , η≡ . (12.304)
2κ V κ V

The absolute values of the slow-roll parameters are much less than one during the
slow-roll period. This means that during a slow-roll period the graph of V (φ) is very
flat and has small curvature.
Alternatively one defines “Hubble slow-roll parameters”, εH , ηH in terms of the
Hubble parameter and its derivatives with respect to the inflaton field

2 H 2 2 H 
εH = , ηH = . (12.305)
κ H κ H

Inserting the first of these expressions into Eq. (12.296) we get for the inflaton
potential

κ V = (3 − εH )H 2 . (12.306)

It follows from Eq. (12.295) that during the slow-roll era differentiation with
respect to time and with respect to the inflaton field are related by
12.8 Inflationary Universe Models 369

d 2 d
= − H . (12.307)
dt κ dφ

Using this in the Definition (12.305) we get simple expressions for εH and ηH

Ḣ 1 Ḧ
εH ≡ − 2
, ηH = − . (12.308)
H 2 Ḣ H

Using that H = ȧ/a the first equation takes the form

aä
εH = 1 − = 1 + q, (12.309)
ȧ2
where q is the deceleration parameter defined in Eq. (12.75). A requirement for
inflation is that there is accelerated expansion, ä > 0. Hence a necessary condition
for inflation is that εH < 1.
It follows from Eq. (12.294) that

φ̈ 1 Ḧ
= , (12.310)
φ̇ 2 Ḣ

giving

φ̈
ηH = − . (12.311)
H φ̇

This equation may be written

 ·
(1/2)φ̇ 2 = −ηH H φ̇ 2 . (12.312)

Hence the sign of the parameter ηH decides whether the kinetic energy of the
inflaton field increases, ηH < 0, or decreases, ηH > 0. The kinetic energy is constant
for ηH = 0.
To lowest order

εH = ε, ηH = η − ε. (12.313)

Equations (2.57), (12.63) and (2.309) give

w = −1 + (2/3)εH . (12.314)

It follows that a universe with εH = 0 is dominated by LIVE with equation of

state parameter w = − 1 and a constant energy density.
From Eqs. (12.285), (2.294) and (2.308) we get
370 12 Cosmology

φ̇ 2 /2
εH = 3 . (12.315)
φ̇ 2 /2 + V

Hence the parameter εH represents the ratio of the kinetic energy and the
total energy of the inflaton field. This is exact. It does not require the slow-roll
approximation.
The ratio of the final value af of the scale factor during the inflationary era and
the initial value a(N ) is
af
= eN , (12.316)
a(N )

where N is called the number of e-folds of the slow-roll era. Hence

 
N = ln af /a . (12.317)

Note that N = 0 at the end of inflation, so that N counts the number of e-folds until
inflation ends and increases as we go backward in time. It follows from Eq. (12.317)
that

Ṅ = −H , (12.318)

or
d 1 d
=− . (12.319)
dN H dt

If Ḣ  H 2 Eq. (12.299) can be approximated by

κ V = 3H 2 . (12.320)

Hence

V H  
= 2 , V  = (6/κ) H 2 + HH  (12.321)
V H

Using this together with Eqs. (2.319) and Ṅ = N  φ̇, we have

H κ H V
dN = − dφ = dφ = κ  dφ. (12.322)
φ̇ 2 H V

This equation can be used to relate derivative with respect to N and derivative
with respect to φ as

d 2ε d
= , (12.323)
dN κ dφ
12.8 Inflationary Universe Models 371

which may be written


κ dφ 2
ε= , (12.324)
2 dN

showing that ε > 0. From Eqs. (12.288) and (12.319) we have

H  (N )
εH = . (12.325)
H

Hence H  (N ) > 0. From the Definition (12.304) and Eq. (12.323) we get

1 V  (N )
ε= . (12.326)
2 V
Integration of Eq. (12.322) gives

φ φ  
V κ κ  
N ≈κ dφ = dφ < φ − φf , (12.327)
V 2ε 2εmin
φf φf

where εmin is the minimum value of ε. Note that if V  > 0 we must have φf < φ in
order that N > 0, and if V  < 0 we must have φf > φ. Equation (12.327) implies a
bound on the change of the value of the scalar field during the inflationary era,

2εmin 
φ > N = N MP 2εmin . (12.328)
κ

This is called the Lyth bound.

There exists a third type of slow-roll parameters. They are defined by

d ln|εn |
ε1 = εH , εn+1 = − . (12.329)
dN
Using Eq. (12.319) we have

1 ε̇n
εn+1 = . (12.330)
H εn
372 12 Cosmology

12.8.5 Testing Observable Consequences of the Inflationary

Era

It has turned out that observations of the polarization of the cosmic microwave back-
ground radiation (CMB) can be used to test predictions made by different inflationary
models [25].
The polarization of the CMB may at every point be described by an amplitude of
the oscillation and a direction. The field of polarization is decomposed in two modes:
the E-mode and the B-mode. The E-mode is curl free like the electrical field of a
charged particle. The B-mode is divergence free like the magnetic field of a current
(Fig. 12.20).
One has classified the polarization in three types:
Scalar perturbation: Energy density fluctuations in the plasma (resulting in hotter
and colder regions) cause velocity distributions that are out of phase with the acoustic
density mode. The fluid velocity from hot to colder regions causes blueshift of the
photons, resulting in E-mode polarization.
Vector perturbation: Vorticity in the plasma causes Doppler shifts resulting in the
quadrupole lobes in the figure. However, vorticity would be damped by inflation and
is expected to be negligible.

Fig. 12.20 Polarization of electromagnetic radiation. Illustration of the E-mode and the B-mode
of polarization of electromagnetic radiation
12.8 Inflationary Universe Models 373

Tensor perturbation: Gravity waves stretch and squeeze space in orthogonal direc-
tions. This also stretches the wavelength of radiation, therefore creating both E-mode
and B-mode polarization in the radiation temperature field.
Gravity waves from inflation produce tensor perturbations.
The power spectra of scalar and tensor fluctuations are represented by
 nS −1+(1/2)αS ln(k/k∗ )+...  nT +(1/2)αT ln(k/k∗ )+...
PS = AS (k∗ ) kk∗ , PT = AT (k∗ ) kk∗ ,
.
AS = V
24π 2 εMP4
, AT = 2V
3π 2 MP4
(12.331)

Here k is the wave number of the perturbation which is a measure of the average
spatial extension for a perturbation with a given power. One often writes k = ȧ = aH ,
where a is the scale factor representing the ratio of the physical distance between
reference particles in the universe relative to their present distance. The quantities
AS and AT are amplitudes, and nS and nT are the spectral indices of the scalar and
tensor fluctuations. The quantities −δns ≡ nS − 1 and nT are called the tilt of the
power spectra of curvature perturbations and tensor modes, respectively, because
they represent the deviation from the values δns = nt = 0 that represent a scale
invariant spectrum.
Furthermore αS and αT are factors representing the k-dependence of the spectral
indices. They are called the running of the spectral indices and are defined by

dnS dnT
αS = , αT = . (12.332)
d ln k d ln k
If nS = 1 the spectrum of the scalar fluctuations is said to be scale invariant. An
invariant mass–density power spectrum is called a Harrison-Zel’dovich spectrum.
One of the predictions of the inflationary universe models is that the cosmic mass
distribution has a spectrum that is nearly scale invariant, but not exactly. The observa-
tions and analysis of the Planck team [26] have given the result nS = 0.968 ± 0.006.
Hence we may use nS = 0.968 as the preferred value of nS , corresponding to
δns = 0.032. Furthermore they have obtained αS = − 0.003 ± 0.007.
The tensor-to-scalar ratio r is defined by

PT (k∗ ) AT
r≡ = , (12.333)
PS (k∗ ) AS

which is a positive quantity. From Eqs. (12.331) and (12.333) we have

r = 16ε. (12.334)

The tensor-to-scalar ratio can be determined from observations of the B-mode of

polarization of the CMB. In the measured wavelength region this B-mode pattern
is partly due to radiation from galactic dust and partly to imprints on the CMB at
374 12 Cosmology

the time 380,000 years after the Big Bang, when the universe became transparent
for the CMB, from relic gravitational waves produced by quantum fluctuations in
the inflationary era. At the present time (November 2019) the best restriction on the
tensor-to-scalar ratio obtained from CMB-measurements is r < 0.04.
It follows from Eqs. (12.334) and (12.314) that the equation of state parameter
during the slow-roll era is given in terms of the tensor-to-scalar ratio as

1 + w = r/24. (12.335)

With 0 ≤ r < 0.04 this gives − 0.9983 ≤ w ≤ −1 during the slow-roll era,
which is equal to or very close to the equation of state of LIVE.
We shall now find how the spectral indices depend upon the slow-roll parameters.
From Eq. (12.331) it follows that they are given by
   
d ln PS (k) d ln PT (k)
δns = − , nT = . (12.336)
d ln k k= aH d ln k k= aH

The quantities inside the brackets are evaluated at the horizon crossing where
k = k· , and the wave number is equal to the scale factor times the Hubble parameter.
It will be useful to write
d d dN
= × . (13.337)
d ln k dN d ln k

Hence, using that AS ∝ H 2 /ε, the scalar spectral indices may be written as

d ln ε d ln H dN d ln H dN
δns = −2 , nT = 2 . (12.338)
dN dN d ln k dN d ln k

Using Eqs. (12.318) and (12.308) we get in the slow-roll approximation

d ln H Ḣ
= − 2 = εH . (12.339)
dN H
From the condition that the spectral indices are calculated at the horizon crossing
we have k = aH . Equation (12.317) gives dN = − d ln a. Hence d ln k = d ln a +
d ln H = − dN + d ln H . Since H is approximately constant during the slow-roll
inflationary era, it follows that

dN
≈ −1. (12.340)
d ln k
It follows from the Definition (12.304) that the derivative of the slow-roll
parameter ε with respect to the scalar field is given by
12.8 Inflationary Universe Models 375

2ε 
ε (φ) = 2ε(η − 2ε) . (12.341)
κ

Using Eq. (12.323) this may be written as

dε
= 2ε(η − 2ε). (12.342)
dN
Inserting this together with Eqs. (12.313) and (12.339) into Eq. (12.338) leads to

δns = 2(3ε − η) (12.343)

It follows from Eqs. (12.313) and (12.343) that the spectral tilt can be expressed
in terms of the Hubble slow-roll parameters as

δns = 2(2εH − ηH ). (12.344)

Equations (12.338), (12.339), (12.340) and (12.313) give

nT ≈ −2ε (12.345)

A consistency relation between r and nT follows from Eqs. (12.334) and (12.245)
r
nT = − . (12.346)
8
Example 12.8.1 Polynomial Inflation As a simple illustration we shall here use the
formalism above to calculate the optical parameters δns , nT and r for the class of
inflationary models called polynomial inflation.
The so-called chaotic inflation models are a class of polynomial models. The
potential of the inflaton field in this type of inflationary models is

V = M 4 φ̂ p , (12.347)
√
where φ̂ = κφ. M is the energy scale of the potential when the inflaton field has
Planck mass, and it is assumed that p is constant and φ > 0.

We shall now deduce expressions for the spectroscopic parameters of this model
in terms of the number of e-folds and find the restrictions that the observational
results δns = 0.032, r < 0.04 put on this class of inflationary models.
Differentiating the potential we get

V p V  p(p − 1)
= , = . (12.348)
V φ̂ V φ̂ 2

Hence, the slow-roll parameters for this model are

376 12 Cosmology

1 p 2 1 p(p − 1)
ε= , η= , (12.349)
2κ φ κ φ2

Hence for this class of inflationary universe models we have

2(p − 1)
η= ε . (12.350)
p

Inserting the expressions (12.349) into Eqs. (12.343) and (12.334) we get

p(p + 2) 8p2
δns = κ , r = κ . (12.351)
φ2 φ2

This gives the δns , r− relation

8p
r= δns , (12.352)
p+2

which may be written

2r
p= . (12.353)
8δns − r

It follows from this equation that the observational results δns = 0.032, r < 0.04
give requirement p < 0.37.
From Eqs. (12.322) and (12.347) we have
κ
dN = φ d φ. (12.354)
p

Integrating through the slow-roll inflationary era, we get

κ  2 
N= φ − φf2 . (12.355)
2p

where φ = φ(N ) is the value of the field strength when the slow-roll   era with N
e-folds begins. It is usual to define the end of the inflationary era by ε φf = 1. From
Eq. (12.349) we then get
p
φf = √ . (12.356)
2κ

Inserting this expression for φf into Eq. (12.356) gives


p
φ= (p + 4N ) . (12.357)
2κ
12.8 Inflationary Universe Models 377

Inserting this into Eq. (12.349) shows that for an inflationary era in which the
potential of the dark energy is a power of the scalar field, the slow-roll parameters
are
p 2(p − 1)
ε= , η= . (12.358)
p + 4N p + 4N

From Eqs. (12.343), (12.345) and (12.334) we then get

2(p + 2) 2p 16p
δns = , nT = − , r= . (12.359)
p + 4N p + 4N p + 4N

Hence
8 − r/2
N= . (12.360)
8δns − r

The observational results δns = 0.032, r < 0.04 then requires that for these
models the number of e-folds is restricted to N < 37.
There is a consensus that in order to solve the horizon and flatness problems the
number of e-folds of the universe during the inflationary era must be larger than 50.
Hence polynomial inflation does not give a satisfactory solution of these problems.
Confrontations of observable consequences of the inflationary era and observa-
tional data have been thoroughly discussed for several inflationary universe models
in [24].

12.9 The Significance of Inertial Dragging

for the Relativity of Rotation

The first published paper on inertial dragging inside a rotating shell based on the
general theory of relativity was published by H. Thirring in 1918. He calculated the
angular velocity, , of a Zero Angular Momentum Observer (ZAMO) inside a shell
with Schwarzschild radius, RS , and radius, r0 , rotating slowly with angular velocity,
ω, in the weak field approximation, and found the inertial dragging angular velocity,

8RS
= ω. (12.361)
3r0

Both the angular velocity of the shell and that of the ZAMO are defined with
respect to a system that is non-rotating in the far away region with asymptotic
Minkowski spacetime.
In 1966 Brill and Cohen [26] presented a calculation of the angular velocity of an
inertial frame inside a rotating shell valid for arbitrarily strong gravitational fields, but
still restricted to slow rotation. The calculation of Brill and Cohen gave the dragging
378 12 Cosmology

angular velocity  inside a massive shell with radius r0 , Schwarzschild radius RS

and rotating with angular velocity ω,

4RS (2r0 − RS )
= ω. (12.362)
(r0 + RS )(3r0 − RS )

A detailed calculation of this expression is found in [26]. For weak fields, i.e.
for r0 RS , this expression reduces to that of Thirring. But if the shell has so
great mass that its Schwarzschild radius is equal to the radius of the shell, r0 = RS ,
the expression above gives  = ω. Then there is perfect dragging. In this case the
inertial properties of space inside the shell no longer depend on the properties of the
ZAMO at infinity, but are completely determined by the shell itself. Brill and Cohen
further wrote that a shell of matter with radius equal to its Schwarzschild radius
together with the space inside it can be taken as an idealized cosmological model,
and proceeded: “Our result shows that in such a model there cannot be a rotation of
the local inertial frame in the centre relative to the large masses in the universe. In
this sense our result explains why the “fixed stars” are indeed fixed in our inertial
frame”. This means that rotation is relative to the motion of the large scale cosmic
masses, and hence that the principle of relativity is valid for rotational motion in a
universe with perfect inertial dragging.
When we look outwards in space, we look backwards in time, because we see an
object the way it was when it emitted the light that we receive. Remarkably, gravi-
tational waves move at the velocity of light. Although it has a quantum mechanical
explanation in the fact that both photons and gravitons are massless, it is a strange
coincidence from a classical point of view, possibly indicating a deep connection
between gravity and electromagnetism. It means that when we search for sources of
gravitational effects that have propagated undisturbed from a changing source to an
observer, neglecting tales of gravitational waves that can be contributions from the
inside of the light cone, we must look at events along the past light cone.

12.9.1 The Cosmic Causal Mass in the Einstein-de Sitter

Universe

We search for cosmic sources of inertial dragging here and now. Hence, we introduce
the concept causal mass, i.e. the mass which produces gravitational effects here and
now. When the causal mass at the point of time t0 of an observer at r = 0 is calculated
by performing an integral with a mass element formed as a spherical shell about the
observer with coordinate radius and thickness r and dr, respectively, the mass of the
element is calculated by inserting the value of the density at the emission time of the
considered mass element on the past light cone.
If the causal mass inside the particle horizon of the universe is so great that its
Schwarzschild radius is equal to or larger than the radius of the horizon, there will
12.9 The Significance of Inertial Dragging for the Relativity of Rotation 379

be perfect inertial dragging [26]. In this case the principle of relativity is valid for
rotational motion in such a universe.
In order to give a simple illustration we first consider an example permitting
analytical expressions in terms of elementary functions; i.e. we shall first consider
the Einstein–de Sitter universe. This is a flat universe containing only dust. It has
scale factor

a(t) = (t/t0 )2/3 , (12.363)

where t0 is the present age of the universe, and the scale factor has been normalized
to a(t0 ) = 1.
The Hubble age is tH = 1/H0 , where H0 is the present value of the Hubble param-
eter. Inserting the most recent value of the Hubble parameter gives tH = 13.9 Gy.
The age of this universe is

t0ED = (2/3)tH . (12.364)

where ED means Einstein-de Sitter.

The physical radius of the particle horizon is (we are using units so that c = 1),

1
1
RPH (t) = a(t) dt. (12.365)
a(t)
0

Inserting the scale factor (12.363) gives the horizon radius of the Einstein–de
Sitter universe,
2/3
RPHED (t) = 3 t0 t 1/3 . (12.366)

The present horizon radius is

RPHED (t0 ) = 3t0ED = 2tH . (12.367)

Hence, the present radius of the particle horizon is 27.8 Gly in an Einstein–de
Sitter universe with the measured value of the Hubble parameter.
We shall now calculate the Schwarzschild radius of the causal mass inside the
particle horizon. It is calculated by integrating along the past light cone; i.e. the
density is evaluated at retarded points of time,

r0PH t0PH
 √   3 2
rSED = 2GM = 8π G ρ g t - r dr = ρa t - r r dr. (12.368)
0 0
380 12 Cosmology

Here, g is the determinant of the spatial part of the metric, and r0PH is the present
radius of the particle horizon. Since the density of the dust is ρ = ρ0 a−3 , we get

rSED = (8π G/3)ρ0 r0PH

3
. (12.369)

Using Eq. (12.367) and given that the present density is equal to the present value
of the critical density,

3H02 3
ρ0 = = , (12.370)
8π G 8π GtH2

we get

rSED = 8 tH . (12.371)

Inserting Planck observational data gives rSED = 108.8 Gly. Hence for this universe
model, rSED = 4RPHED , showing that the Schwarzschild radius of the causal mass
inside the horizon is larger than the horizon radius. This indicates that the causal mass
inside the horizon is so great that there is perfect inertial dragging in this universe,
and hence that the principle of relativity is valid for rotational motion in this universe.
The lookback distance is the radius of a surface S around an observer equal to
the velocity of light times the age of the universe, rLED = t0 . Inserting this as the
upper limit in the integral (12.215), we find that the Schwarzschild radius of the mass
inside S is equal to the lookback distance, rSHED = rLED . This corresponds to the
condition for perfect inertial dragging used in Sect. 12.7. However, from a causal
point of view, the relevant surface is the particle horizon.

12.9.2 The Cosmic Causal Mass in the Flat ΛCDM Universe

We now consider the flat CDM universe which has scale factor [9]

1 − 0 2 2tH
a(t) = A1/3 sinh2/3 (t/t ), A = , t = √ = √ . (12.372)
0 3 3 0

The present radius of the particle horizon in this universe is

√
 1/A
arsinh
2 tH dx
r0PH = rPH (t0 ) = . (12.373)
3 1/6
0 (1 − 0 )
1/3 sinh2/3 (x)
0

Inserting the Planck values 0 = 0.68 and t H = 13.9 Gy and calculating the
integral numerically gives R0PH = 45 Gly.
12.9 The Significance of Inertial Dragging for the Relativity of Rotation 381

The mass which acts causally at the present time t0 upon the observer located at
r = 0 in the CDM universe is
r0PH
  
M = 4π ρM 0 + ρ a(te (r))3 r 2 dr. (12.374)
0

Here ρM0 and ρ denote the present density of the dust and the vacuum energy,
respectively, and te denotes the emission time of a signal emitted at the coordinate
distance r and received at r = 0 at the time t0 . Accordingly te is a function of the
coordinate distance r the signal travels and is given implicitly by the relation

dr t
= (12.375)
dt a(t)

with the initial condition r(−t0 ) = 0. The present Schwarzschild radius of this mass
is RS = 2 GM. Solving these equations numerically and plotting the Schwarzschild
radius of the causal mass inside the particle horizon and the present radius of the
particle horizon as functions of 0 we obtain the result shown in Fig. 12.21 which
is taken from the article [27].
We see from this that the Schwarzschild radius of the causal mass inside the
present particle horizon in a flat CDM universe is larger than the present radius
of the particle horizon. This means that there is perfect inertial dragging in the
universe. Hence the motion of inertial frames in the universe is determined by the

Fig. 12.21 Causal mass. The upper curve represents the Schwarzschild radius of the causal mass
inside the present particle horizon in a flat CDM universe, and the lower curve represents the
present radius of the particle horizon
382 12 Cosmology

average motion of the causal mass in the universe. If this is a realistic model of our
universe, we may conclude that rotational and accelerated motion is relative in our
universe according to the general theory of relativity.

Exercises

12.1. Gravitational collapse

In this problem we shall find a solution to Einstein’s field equations describing a
spherically symmetric gravitational collapse. The solution shall describe the space-
time both exterior and interior to the star. To connect the exterior and interior solu-
tions, the metrics must be expressed in the same coordinate system. We will assume
that the interior solution has the same form as a Friedmann solution. The Friedmann
solutions are expressed in co moving coordinates, thus freely falling particles have
constant spatial coordinates.
Let (ρ, τ ) be the infalling coordinates, i. co-moving coordinates with freely falling
particles. τ is the proper time of a freely falling particle starting at infinity with
zero velocity. These coordinates are connected to the curvature coordinates via the
requirements

ρ = r, for τ = 0,
(12.376)
τ = t, for r = 0.

(a) Show that the transformation between the in falling coordinates and the curvature
coordinates is given by

2  3 3

τ= √ ρ2 − r2 ,
3c RS
⎡ 1 ⎤
2
RS ⎢
r
+1
2 1 RS ⎥
t = τ − (RS r) 2 + ln⎣   1 ⎦, (12.377)
c c r 2
RS
−1

where RS is the Schwarzschild radius of the star.

(b) Show that the Schwarzschild metric in these coordinates takes the form

 − 23
3 1
− 23
ds = −c dτ + 1 − (RS ) cτρ
2 2 2 2 dρ 2
2
  43
3 1
− 23
 
+ 1 − (RS ) cτρ
2 ρ 2 dθ 2 + sin2 θ dφ 2 . (12.378)
2
Exercises 383

Assume the star has a position-dependent energy density ρ(τ ), and that the pres-
sure is zero. Assume further that the interior spacetime can be described with a
Friedmann solution with Euclidean geometry (k = 0).
(c) Find the solution when the radius of the star is R0 at τ = 0.
12.2. The volume of a closed Robertson–Walker universe
Show that the volume of the region contained inside a radius r = aχ = a arcsin r is

1
V = 2π a χ − sin 2χ .
3
(12.379)
2

Find the maximal volume. Find also an approximate expression for V when χ 
R.
12.3. Conformal time
Find the form of the line-element (12.1) if the cosmic time t is replaced by conformal
time η defined by Eq. (12.126).
What is the equation of light moving radially when we use conformal time?
12.4. Lookback time and the age of the universe
The lookback time of an object is the time required for light to travel from an emitting
object to the receiver. Hence, it is tL ≡ t0 − te , where t 0 is the point of time the object
was observed and t e is the point of time the light was emitted.
(a) Show that the lookback time is given by

2
dy
tL = , (12.380)
(1 + y)H (y)
0

where z is the redshift of the object and the Hubble parameter H (y) is given in
Eq. (12.111).
(b) Show that the lookback time in the Milne universe model with a(t) =
(t/t0 ), k < 0, is

1 z
tL = . (12.381)
H0 1 + z

and find the age of this universe.

 
(c) Show that tL = t0 1 − (1 + z)−3/2 , where t0 = 2/(3H0 ), in a flat, matter-
dominated universe.
384 12 Cosmology

What is the age of this universe.?

(d) Find the lookback time of an object with redshift z and the age of the matter-
dominated universe models with positive and negative spatial curvature.
(e) Find the age of a matter-dominated universe from the parametric solutions
(12.140)–(12.144).
Are the resulting expressions in agreement with those found in d)?
(f) Find the lookback time–redshift relation and the age of a flat universe with dust
and LIVE.
(g) Find the lookback time—redshift relation for a flat, LIVE-dominated universe.
12.5. The LFRW universe models with a perfect fluid

In this problem we will investigate FRW models with a perfect fluid. We will assume
that the perfect fluid obeys the equation of state

p = wρ, (12.382)

where −1 ≤ w ≤ 1.
(a) Write down the Friedmann equations for a LFRW universe model with a w-law
perfect fluid. Express the equations in terms of the scale factor a only.
(b) Assume that a(0) = 0. Show that when −1/3 < w ≤ 1, the closed model will
recollapse. Explain why this does not happen in the flat and open models.
(c) Solve the Friedmann equation for a general w = −1 in the flat case. What is the
Hubble parameter and the deceleration parameter? Also write down the time
evolution for the matter density.
(d) Find the particle horizon distance in terms of H 0 , w and z.
(e) Specialize the above to the dust-dominated, radiation-dominated and LIVE-
dominated universe models.
(f) Find a general formula in terms of the density parameters for the present value
of the deceleration parameter of a LFRW universe model.
12.6. Age—density relation for a radiation-dominated universe

Show that the age of a radiation-dominated universe model is given by

tH
t0 = √ (12.383)
1 + rad0

for all values of k.

Exercises 385

12.7. Redshift–luminosity relation for matter-dominated universe: Mattig’s for-

mula
The luminosity distance of an object with redshift z is
⎡ ⎤
 z
1+z dz
dL = √ Sk ⎣H0 |k0 | ⎦, (12.384)
H0 |k0 | H (z)
0

where the function Sk is defined in Eq. (12.17) and H (z) is given in Eq. (12.111).
Show that the luminosity distance of an object with redshift z in a matter-
dominated universe with relative density m0 and Hubble constant H 0 is

2c  !
dL = m0 z + (m0 − 2)( 1 + m0 z − 1) . (12.385)
H0 2m0

This is called Mattig’s formula. Find the corresponding formula for the Einstein–
de Sitter universe, with m0 = 1.

12.8. Newtonian approximation with vacuum energy

(a) Show that Einstein’ linearized field equation for a static spacetime containing
dust with density ρ and vacuum energy with density ρ takes the form of a
modified Poisson equation

∇ 2 φ = 4π G(ρ − 2ρ ). (12.386)

(b) Assume there is a particle with mass m at the origin. Solve Eq. (12.386) in the
space outside the particle, and find the acceleration of gravity at a function of
the distance from the origin.

Find the radius of a spherical surface where the acceleration of gravity vanishes.
How large is the mass inside this surface compared to the mass of the particle at
the origin?
Evaluate the importance of LIVE for gravitational phenomena in the solar system.

12.9. Universe models with constant deceleration parameter

(a) Show that the universe with constant deceleration parameter q has
expansion factor

 1+q
1
t
a= , q = −1, and a ∝ eHt , q = −1. (12.387)
t0
386 12 Cosmology

This is the scale factor of a flat universe with a perfect fluid. Find the equation of
state of the fluid.
12.10. Density parameters as functions of the redshift

Show that the relative densities of LIVE and matter as functions of a are

L0 a3
L = ,
L0 + (1 − L0 − m0 )a + m0
a3
m0
m = . (12.388)
L0 a + (1 − L0 − m0 )a + m0
3

What can you conclude from these expressions concerning the universe at early
and late times?

12.11. FRW universe with radiation and matter

Show that the scale factor and the cosmic time as functions of conformal time of a
universe with radiation and matter are [28].
(
a = a0 [α(1 − cos η) + β sin η]
k>0: (12.389)
t = a0 [α(η − sin η) + β(1 − cos η)],
(  
a = a0 21 αη2 + βη 
k=0: (12.390)
t = a0 16 αη3 + 21 βη2 ,
(
a = a0 [α(cosh η − 1) + β sinh η]
k<0: (12.391)
t = a0 [α(sinh η − η) + β(cosh η − 1)],

where α = a02 H02 m0 /2 and β = (a02 H02 γ 0 )1/2 , and γ 0 and m0 are the present
density parameters of radiation and matter, and H 0 is the present value of the Hubble
parameter.

12.12. Event horizons in de Sitter universe models

Find the coordinate distances to the event horizons of the de Sitter universe models
with k > 0, k = 0 and k < 0 as function of time.

12.13. Flat universe model with radiation and LIVE

(a) Find the scale factor as a function of time for a flat universe with radi-
ation and Lorentz invariant vacuum energy represented by a cosmo-
logical constant , and with present density parameter of radiation
rad0 .
(b) Calculate the Hubble parameter, H, as a function of time, and show that
the model approaches a de Sitter model in the far future. Find also the
deceleration parameter, q(t).
Exercises 387

(c) When is the inflection point, t 1 , for which the universe went from decel-
eration to acceleration? What is the corresponding redshift observed at
the time t 0 ?

12.14 De Sitter spacetime

Consider a De Sitter spacetime with coordinates (t, r) and line element

 
ds2 = −c2 dt 2 + e2Ht dr 2 + r 2 d2 , (12.393)

where the Hubble parameter H is constant.

(a) Find the redshift of light emitter from a coordinate r as measured at a point of
time t0 by an observer at the origin. The Hubble parameter H is assumed to be
known.
(b) What is the 4-acceleration of a reference particle at rest in the coordinate system?
What does your result tell about the reference frame in which these coordinates
are co-moving?
Will an observer with constant radial coordinate r experience an acceleration of
gravity?
Introducing coordinates (T , R) by the transformation
 
R = reHt , T = t − ln 1 − H 2 r 2 e2Ht (12.394)

or
R 
r=  , eHt = eHT 1 − H 2 R2 /c2 , (12.395)
eHT 1 − H 2 R2 /c2

the line element takes the form (you need not show this)

  dR2
ds2 = − c2 − H 2 R2 dT 2 + + R2 d 2 . (12.396)
1 − H 2 R2 /c2

(c) Find the redshift of light emitted from a coordinate R as measured by an observer
at the origin. Why is your result different to the one in a)?
(d) What is the 4-acceleration of a reference particle at rest in the coordinate system?
What does your result tell about the reference frame in which these coordinates
are co-moving? Will an observer with constant radial coordinate r experience
an acceleration of gravity?
(e) How does a reference particle with r = r0 = constant move in the (T , R)-
coordinate system.
(f) How is the redshift of light explained in the (T , R)-coordinate system? How
is it explained in the (t, r)-system?
388 12 Cosmology

12.15. The Milne Universe

(a) The Milne Universe has a line element

 2 
t dr 2
ds = −cdt +
2 2
+ r d .
2 2
(12.397)
t0 1 + r 2 /c2 t02

The t , r coordinates are co-moving in a reference frame E. Give a physical

interpretation of the line-element.
(b) The Hubble parameter is H = ȧ/a where ȧ is the derivative of the scale factor
with respect to time. Calculate the Hubble parameter of this universe model at
an arbitrary point of time. Find the age of the universe in terms of the present
value of the Hubble parameter.
(c) The so-called deceleration parameter is given by q ≡ − aä/ȧ2 .
What is the value of q for this universe model? What does this tell about the
expansion?
(d) Introduce new coordinates T and R by the transformation

  2
r rt
T =t 1+ , R= . (12.398)
ct0 t0

Show that the inverse transformation is

 ct0 R
ct = c2 T 2 − R2 , r=√ , (12.399)
c2 T 2 − R2

and that
rcT
R=
. (12.400)
c2 t02 + r 2

(e) Make a Minkowski diagram with reference to the cT , R-system, and draw the
world lines of the reference particles of E, i.e. those with r = constant and the
simultaneity curves of E, i.e. those with t = constant.
(f) Use the transformation Eqs. (3) to show that the differentials of the coordinates
co-moving in the reference frame E are

c2 T dT − R dR c3 t0 T (T dR − R dT )
dt = √ , dr =  3/2 . (12.401)
c c2 T 2 − R2 c2 T 2 − R2
Exercises 389

Use these differentials and the expressions for t and r in Eq. (3) to calculate the
line element in the T , R coordinate system. What does your result tell about the
spacetime described by the line element you just have found and the line element
(1)?
(g) Calculate the cosmic redshift, z, of a star with r = r1 = constant in terms of r1
and the point of time t0 of the observation. Then calculate the redshift of a star
with R = constant. Explain the results you found.
(h) Einstein’s field equations as applied to an isotropic and homogeneous universe
model lead to the Friedmann equations

ȧ2 + kc2 8π G ä 4π G  p
2
= ρ , =− ρ+3 2 (12.402)
a 3 a 3 c
where a is the scale factor, ρ and p are the density and pressure of the cosmic fluid,
respectively, and k is the spatial curvature index.
Apply these equations to the line-element (1). What does your result tell about
this universe model?

12.16. Natural Inflation

The natural inflation model has potential

 
V (φ) = V0 1 + cos φ̃ , (12.403)

where φ̃ = φ/M , and M is the spontaneous symmetry breaking scale. We shall here
write Einstein’s gravitational constant as κ = 1/MP2 , where MP is the Planck mass.
(a) Show that for this model the spectral parameters are


3 − cos φ̃ 1 − cos φ̃ 1 − cos φ̃ MP 2
δns = b , nT = − b , r = 8b , b=
1 + cos φ̃ 1 + cos φ̃ 1 + cos φ̃ M
(12.404)

Observations have given the results δns = 0.032 and r < 0.04.
(b) Show that

r
b = δns − , (12.405)
4
and use this to calculate the requirement from the observations upon the symmetry
breaking scale. Is there any problem with the result?
390 12 Cosmology

(c) Show that for this model the number of e-folds is

 
1 1 − cos φ̃f
N = ln  . (12.406)
b 1 − cos φ̃

Use this to express the spectral parameters as

(2 + b)eb N + 2 2b 16b
δns = b , nT = − , r= ,
(2 + b)e − 2
b N (2 + b)e − 2
b N (2 + b)eb N − 2
(12.407)

In order to solve the horizon- and flatness problems the number of e-folds must
be larger than 50. Insert N = 50, r = 0.04 and make a judgement of this model.

References

1. Grøn, Ø., Elgarøy, Ø.: Is space expanding in the Friedmann universe models. Am. J. Phys. 75,
151–157 (2007)
2. McGaugh, S.: Constraints on the radial mass distribution of dark matter halos from rotation
curves. In: Natarajan, P. (ed.) The Shapes of Galaxies and their Dark Halos, pp. 186–193. World
Scientific (2001)
3. Linde, A.:. Inflation and string cosmology. Int. J. Mod. Phys. A17SI, 89–104 (2001)
4. de Bernadis, P., et al.: Multiple peaks in the angular power spectrum of the cosmic microwave
background: significance and consequences for cosmology. Astrophys. J. 584, 559–566 (2001)
5. Stompor, R., et al.: Cosmological implications of the MAXIMA-1 high resolution cosmic
microwave background anisotropy measurement. Astrophys. J. 561, L7–L10 (2001)
6. Pryke, C., et al.: Cosmological parameter extraction from the first season of observations with
DASI. Astrophys. J. 568, 46–51 (2001)
7. Riess, A.G., et al.: Observational evidence from supernovae for an accelerating universe and a
cosmological constant. Astron. J. 116, 1009–1038 (1998)
8. Perlmutter, S., et al.: Measurements of omega and lambda from 42 high-redshift supernovae.
Astrophys. J. 517, 565–586 (1999)
9. Grøn, Ø.: A new standard model of the universe. Eur. J. Phys. 23, 135–144 (2002)
10. Zlatev, I., Wang, L., Steinhardt, P.J.: Quintessence, cosmic coincidence, and the cosmological
constant. Phys. Rev. Lett. 82, 896–899 (1999)
11. Carroll, S.M.: Quintessence and the rest of the world. Phys. Rev. Lett. 81, 3067–3070 (1998)
12. Zeldovich, Y.: The cosmological constant and the theory of elementary particles. Sov. Phys.
Usp. 11, 381–393 (1968)
13. Grøn, Ø.: Repulsive gravitation and inflationary universe models. Am. J. Phys. 54, 46–52
(1986)
14. Lahav. O.: Cosmological Parameters. Particles data group. Updated 2017. http://pdg.lbl.gov/
2017/reviews/rpp2017-rev-cosmological-parameters.pdf. (2017)
15. Grøn, Ø., Jemterud, T.: An interesting consequence of the principle of relativity. Euro. Phys.
J. Plus. 131, 91 (2016). https://doi.org/10.1140/epjp/i2016-16091-9
16. Riess, A.G.: The farthest known supernova: support for an accelerating universe and a glimpse
of the epoch of deceleration. Astrophys. J. 560, 49–71 (2001)
References 391

17. Wen-Fui, W., Zeng-Wei, S., Bin, T.: Exact solution of phantom dark energy model. Chin. Phys.
B 19, 119801 (2010)
18. Grøn, Ø.: Expansion isotropization during the inflationary era. Phys. Rev. D 32, 2522–2527
(1985)
19. Alnes, H., Amarzguioui, M., Grøn, Ø.: An inhomogeneous alternative to dark energy? Phys.
Rev. D73, 083519 (2006)
20. Enqvist, K.: Lemaître-Tolman-Bondi model and accelerating expansion. Gen. Rel. Grav. 40,
451–466 (2008)
21. McGaugh, S.:Constraints on the radial mass distribution of dark matter halos from rotation
curves. In: P. Natarajan (ed.) The Shapes of Galaxies and their Dark Halos, pp. 186–193.
World Scientific (2001)
22. Turner, M.S.: Dark Energy and the New Cosmology, astro-ph/0108103 (2001)
23. Guth, A.H.: The inflationary universe: a possible solution to the horizon and flatness problems.
Phys. Rev. D 23, 347–356 (1981)
24. Grøn, Ø.: Predictions of spectral parameters by several inflationary universe models in light of
the Planck results. Universe 4(2), 15 (163 pages) (2018)
25. Ade, P., et al.: A joint analysis of BICEP2/Keck array and Planck Data. Phys. Rev. Lett. 114,
101301 (2015)
26. Brill, D.R., Cohen, J.M.: Rotating masses and their effect on inertial frames. Phys. Rev.143,
1011–1015 (1966)
27. Braeck, S., Grøn, Ø.G., Farup, I.: The cosmic causal mass. Universe. 3, 38 (2017). https://doi.
org/10.3390/universe3020038
28. Johannesen. S.: Smooth Manifolds and Fibre Bundles with Application to Theoretical Physics.
CRC Press (2017)
Appendix
Kaluza–Klein Theory

In the general theory of relativity gravitation is a property of 4-dimensional space-

time and of the motion of the reference frame. Kaluza–Klein theory, [1, 2], is a
unified theory of gravitation and electromagnetism in which electromagnetism is the
projection of certain elements of gravitation in 5-dimensional spacetime into our
4-dimensional spacetime. The fifth dimension is spatial. According to this theory
there is no electromagnetism in the 5-dimensional world, only gravitation which is
described by the general theory of relativity in 5-dimensional spacetime.

A.1 The Structure of the Kaluza–Klein Theory

The Kaluza–Klein theory is constructed in the following way. Let the line-element
in a Riemannian 5-dimensional space be

ds 2 = gμν dx μ dx ν , μ, ν = 1, 2, 3, 4, 5 (A.1)

with signature (−, +, +, +, +).

We experience a 4-dimensional world. We are flat-landers in a 4-dimensional
hyper-surface R4 —our spacetime—in the 5-dimensional universe. It has become
usual to consider the fifth dimension as a cylinder with axis orthogonal to 4-
dimensional spacetime. Hence, the intersection of the fifth cylinder dimension with
our 4-dimensional spacetime is a circle. Also it is assumed that this circle has an
extremely small radius, since we have no direct experience of this fifth dimension.
This means that what we ordinarily think of as a point in 4-dimensional spacetime
is a small circle according to the Kaluza–Klein theory.
Let e5 be a basis vector orthogonal to our 4-dimensional spacetime. Einstein and
Bergmann (1938) have shown that if one requires that closed geodesic curves in the
fifth dimension—around the cylinder—shall be continuous, then it is necessary that

gμν,5 = 0. (A.2)

© Springer Nature Switzerland AG 2020 393

Ø. Grøn, Introduction to Einstein’s Theory of Relativity,
Undergraduate Texts in Physics, https://doi.org/10.1007/978-3-030-43862-3
394 Appendix: Kaluza–Klein Theory

Furthermore g55 is a constant along the curve. One can then require that g55 is
generally constant, and by a suitable choice of coordinate one can normalize g55 so
that

g55 = 1. (A.3)

The conditions (A.2) and (A.3) are called the cylinder conditions.
In the same way that the spatial metric in the line-element (4.5) represents a
projection of the metric in 4-dimensional spacetime into the 3-space orthogonal to the
time direction, the metric of 4-dimensional spacetime is obtained in the Kaluza–Klein
theory by projecting the metric of 5-dimensional spacetime onto the 4-dimensional
spacetime orthogonal to the a basis vector pointing along the fifth dimension. Hence
the line element of 4-dimensional spacetime is,
g5i g5 j
dl 2 = γi j dx i dx j , γi j = gi j − = gi j − g5i g5 j , (A.4)
g55

where we have used Eq. (A.3).

It follows from Eqs. (4.6)–(4.9) that dl 2 is invariant against a coordinate
transformation of the form
     
xi = xi xi , x5 = x5 + f xi . (A.5)

Here the first equation represents an arbitrary transformation in 4-dimensional

spacetime, and the second a gauge-transformation in the fifth dimension: the choice
of origin on a circle x 1 , x 2 , x 3 , x 4 = constant depends upon the position in 4-
dimensional spacetime. Note that the cylinder conditions (A.2) and (A.3) are invariant
against a transformation of the form (A.5). Furthermore the transformation of g5i
against (A.5) has the form
      
∂xμ ∂xν ∂x5 ∂xν ∂x5 ∂x5 ∂xν ∂f
g5ν = gμ ν = g5ν + g5 5 = g5 ν  + ν .
∂x5 ∂xi ∂x5 ∂xν ∂x5 ∂xν ∂xν ∂x
(A.6)

Hence one can introduce a one-form in the 5-dimensional spacetime

g5 = g5 + d f. (A.7)

The physical interpretation of this form from the 4-dimensional point of view is
that it represents the electromagnetic vector potential form,

g5 = A. (A.8)

This leads to the gauge transformation of the electromagnetic vector potential,

Appendix: Kaluza–Klein Theory 395

A = A + d f. (A.9)

From this equation and Poincare’s lemma it follows that the electromagnetic field
form

F = dA (A.10)

is invariant against the gauge transformation (A.9). Also this invariance is here seen
as a consequence of the invariance of the metric 1-form g5 against a transformation
representing a free choice of the origin on the fifth cylinder dimension.

A.2 Calculation of the 5-dimensional Curvature Scalar

We shall now calculate the curvature scalar of the Riemann tensor representing the
curvature of the 5-dimensional spacetime using Cartan’s structure equations.
The line-element of 5-dimensional spacetime may be written as
 2
ds 2 = γi j dx i d x j + dx 5 + Ai dx i , Ai = g5i , (A.11)

where γi j are the components of the metric tensor of our 4-dimensional spacetime.
We introduce a form basis

ωi = dx i , ω5 = dx 5 + Ai dx i . (A.12)

Exterior differentiation gives

 
1 ∂ Ai ∂ Aj
dω = 0, dω =
i 5
− dx i ∧ dx j . (A.13)
2 ∂x j ∂xi

Hence the electromagnetic field form has components

∂ Ai ∂ Aj ∂g5i ∂g5 j
Fi j = − = − . (A.14)
∂x j ∂x i ∂x j ∂xi
From Cartan’s 1. structure equation in the form (5.180) we have

dω5 = −5j ∧ ω j . (A.15)

According to Eq. (5.164) the components of the connection form are

5j =  5jν ων . (A.16)

It follows from Eqs. (A.3) and (A.13)–(A.16) that

396 Appendix: Kaluza–Klein Theory

1  1 1 μ
ikj =4 ikj − Ai F jk + A j Fik , 5 ji =  5ji = − Fi j = F ji , 55 = 0.
2 2 2
(A.17)

Hence
1
5i j = i5j = Fi j . (A.18)
2
With the basis (A.12) the metric has the form
 
γi j 0
gμν = . (A.19)
0 1

Hence

dg5 j = dg j5 = dg55 = 0. (A.20)

It follows from Eq. (5.169) and this equation that

i5 = −5i , (A.21)

giving

1
i5 j = −5i j = − Fi j . (A.22)
2
Thus
1
5i j = − F ji . (A.23)
2
It follows from Eq. (5.132) that

 ij5 = 5i j + c5i j , (A.24)

where the structure coefficients are defined by Eq. (3.40) which here gives
 μ
e5 , ei = c5i eμ . (A.25)

Here
∂ ∂ ∂
e5 = , ei = i − Ai 5 . (A.26)
∂x 5 ∂x ∂x

Since Ai is independent of x 5 we get

Appendix: Kaluza–Klein Theory 397


∂ ∂ ∂
, − A i = 0. (A.27)
∂x5 ∂xi ∂x5

It follows that
μ
c5i = 0. (A.28)

Equations (A.23) and (A.24) then give

1
 ij5 = 5i j = − F ji . (A.29)
2
We have

ij =  ijk ωk +  ij5 ω5 . (A.30)

The curvature forms are given in Cartan’s 2. structure equation (6.24). In the
present case they give for the curvature of the 5-dimensional spacetime,
μ
R ij = dij + iμ ∧  j = dij + ik ∧ kj + i5 ∧ 5j
= d  ijk ωk + ik ∧ kj + d  ij5 ω5 + i5 ∧ 5j . (A.31)
= 4
R ij +  ij5,k ωk ∧ω +5
 ij5 dω5 + i5 ∧ 5j

From Eqs. (A.15), (A.16) and (A.19) we have

1
dω5 = Fi j ωi ∧ ω j . (A.32)
2
and
1 i
i5 ∧ 5j = F Fn j ωm ∧ ωn , (A.33)
4 m
which leads to
1 i k 1 1
R ij =4 R ij − F j,k ω ∧ ω5 − F ji Fmn ωm ∧ ωn + Fmi Fn j ωm ∧ ωn . (A.34)
2 4 4
The components of the Riemann curvature tensor are given as the components of
the curvature forms by

1 i
R ij = R ωμ ∧ ων . (A.35)
2 jμν

We need the components, R ijmn , of these curvature forms for the 5-dimensional
spacetime in 4-dimensional spacetime. Then the second term in the right-hand side of
398 Appendix: Kaluza–Klein Theory

Eq. (A.34) does not contribute. Also the components of the 2-forms are antisymmet-
ric. This is automatically taken care of in the third term due to the antisymmetry of the
electromagnetic field tensor, but in the fourth term we have to take the antisymmetric
combination when we write down the form-components. This gives

1 i 1 1
R ijmn =4 R ijmn − F Fmn + Fmi Fn j − Fni Fm j . (A.36)
2 j 4 4
We are going to calculate the Ricci curvature scalar in the 5-dimensional
spacetime,
j
R = R j + R55 , (A.37)

where the terms at the right-hand side are the mixed components of the Ricci curvature
tensor. These are calculated as follows.
The components of the Riemann curvature tensor are

R jn = R μjμn = R ijin + R 5j5n = R ijin + R5 j5n = R ijin + R j5n5 . (A.38)

Using Eq. (A.36) and (A.48) we get

1 i 1 1 1
R jn =4 R jn − F j Fin + Fii Fn j − Fni Fi j − F ji Fni . (A.39)
2 4 4 4

The electromagnetic field tensor is trace free, so Fii = 0. Hence the third term at the
right-hand side vanishes. Also since the electromagnetic field tensor is antisymmetric,
the two last terms cancel each other. Hence we get

1 i
R jn =4 R jn − F Fin . (A.40)
2 j
Thus

j j 1 ij j 1
R j =4 R j − F Fi j =4 R j − Fi j F i j . (A.41)
2 2

In order to calculate R55 we need the curvature forms

μ j
R i5 = di5 + iμ ∧ 5 = di5 + ij ∧ 5 . (A.42)

It follows from Eqs. (5.164) and (A.29) that

1
i5 = 5i j ω j = − F ji ω j . (A.43)
2
Hence
Appendix: Kaluza–Klein Theory 399

1 i k
di5 = − F j,k ω ∧ ωj. (A.44)
2
Using Eqs. (A.30) and (A.43) we get
j  i m  j
Ω ij ∧ Ω 5 = Γ jm ω + Γ j5
i
ω5 ∧ Γ5n ωn
j j
. (A.45)
= Γ jm
i
Γ5n ωm ∧ ωn + Γ j5
i
Γ5n ω5 ∧ ωn

j
In the first term at the right-hand side  ijm 5n is symmetric in m and n while the
basis is antisymmetric. Hence this term vanishes. Using Eq. (A.29) we then obtain

j 1 i j 5
ij ∧ 5 = F F ω ∧ ωm . (A.46)
4 j m
Inserting the expressions (A.44) and (A.46) into Eq. (A.42) gives

1 i k 1
R i5 = − F j,k ω ∧ ω j + F ji Fmj ω5 ∧ ωm . (A.47)
2 4
This gives

1 i 
R5i jk = Fk, j − F j,k
i
(A.48)
2
and
1 i j
i
R55m = F F (A.49)
4 j m
or
1
i
R5m5 = − F ji Fmj . (A.50)
4
It follows that
1 j 1
R55 = R55 = R5i5
i
= − F ji Fi = − Fi j F i j . (A.51)
4 4
Inserting Eqs. (A.41) and (A.51) into Eq. (A.37) finally gives the Kaluza–Klein
expression for the Ricci curvature scalar of the 5-dimensional spacetime

j 1 1 1
R =4 R j − Fi j F i j + Fi j F i j =4 R j − Fi j F i j (A.52)
2 4 4
Comparing with Eq. (7.65) the last term is recognized as the Lagrangian of an
electromagnetic field. The variational principle (7.58) with the Lagrangian equal to
the Ricci curvature scalar of 5-dimensional spacetime gives Einstein’s gravitational
400 Appendix: Kaluza–Klein Theory

field equations in this 5-dimensional word. But as interpreted from our 4-dimensional
perspective, the right-hand side of Eq. (A.52) shows that we have the Lagrangian
leading to the Einstein’s gravitational equations and Maxwell’s source free equations.
So the Einstein’s theory in the 5-dimensional world is interpreted as a unified theory
of gravity and electromagnetism in our 4-dimensional subspace of the 5-dimensional
world.

A.3 Field Equations for Kaluza–Klein Theory with g55 = 1

The theory as developed so far has one serious problem. The Einstein equations
in 5-dimensional spacetime, (marking here the curvature tensors of 5-dimensional
spacetime by an index 5)

15
5
E =5 Rμν − Rgμν = κ Tμν , (A.53)
2
represent in general 15 equations. Since there are 5 Bianchi identities in 5-
dimensional spacetime, there are 10 independent field equations. In general the
5-dimensional metric tensor has 15 independent components. But when the the-
ory is projected down to our 4-dimensional spacetime g55 appears as an unwanted
scalar field, i.e. a scalar field which is foreign to Einstein’s theory of 4-dimensional
spacetime and gravitation. This is eliminated by the cylinder condition (A.3). With
this condition there are 14 functions to be determined. Because of the transforma-
tion (A.5) which contains 5 arbitrary functions, there are only 9 independent metric
functions. Hence the system is over determined. Thus, with the cylinder condition
(A.3) it is necessary to reduce the number of independent equations by 1. This can
be performed by adding a term in the equations in such a way that a sixth algebraic
identity results. In this way there will only be 9 independent equations.
We shall therefore require that the curvature tensor at the left-hand side of the
field equations in 5-dimensional spacetime is both divergence free, trace free and
symmetric. The trace of the 5-dimensional Einstein tensor is

35
5
E =− R. (A.54)
2
So in order to obtain a trace-free curvature tensor we must add a symmetric,
divergence-free tensor with trace (A.54). The tensor with components

35
Nμν = Rg5μ g5ν (A.55)
2
fulfil these conditions. Thus a symmetric, divergence-free and trace-free curvature
tensor in 5-dimensional spacetime is
Appendix: Kaluza–Klein Theory 401

↔ 1 5
5
E μν =5 Rμν − gμν − 3g5μ g5ν R. (A.56)
2
It was originally proposed by Leibowitz and Rosen [3] and may therefore be called
the Leibowitz-Rosen tensor. Hence, the modified field equations have the form

1 5
5
Rμν − gμν − 3g5μ g5ν R = κ 5 Tμν , (A.57)
2

where 5 Tμν are the components of the energy–momentum tensor in 5-dimensional

spacetime. Since the left-hand side of this equation is trace free, the right-hand side
must be so, too. From an arbitrary energy–momentum tensor Tμν we can always
construct a corresponding trace-free tensor
5
Tμν = Tμν − T g5μ g5ν . (A.58)

Since the electromagnetic energy–momentum tensor is trace free 5 Tμν = Tμν for
electromagnetic fields.

A.4 The 5-dimensional Counterpart of Electric Charge

From the tensor 5 Tμν we can form a quantity with components


2κ i
Ji = − T (A.59)
μ0 5

that transform as the components of a 4-vector under the transformation (A.5). Here
κ is Einstein’s gravitational constant given in Eq. (7.37), and μ0 is the permeability
of empty space. The conservation equation
μν
5
T;ν = 0, (A.60)

following from the 5-dimensional field equations, then implies that

J;ii = 0, (A.61)

which is the electromagnetic equation of continuity

∂ρ
∇ · j + = 0. (A.62)
∂t
Hence conservation of charge follows from a geometric identity, the Bianchi
identity, in this theory.
402 Appendix: Kaluza–Klein Theory

Let us consider a gas of dust particles in 5-dimensional spacetime. Its trace-free

energy–momentum tensor has the components
 
5
Tμν = ρ u μ u ν + g5μ g5ν . (A.63)

where u μ are the components of the 5-velocity of the dust particles. For this gas the
conservation Eq. (A.60) takes the form

(ρu ν );ν u μ + ρa μ = 0, (A.64)

where

a μ = u ν u μ;ν (A.65)

are the components of the 5-acceleration of the dust particles. The 5-velocity is a
unit vector. It follows that the 5-velocity and the 5-acceleration are orthogonal. Thus,
multiplying Eq. (A.64) by u μ gives

(ρu ν );ν = 0 (A.66)

Inserting this into Eq. (A.64) gives

a μ = 0, (A.67)

which shows that the dust particles follow geodesic curves in 5-dimensional
spacetime. Written out this equation takes the form

du μ μ
+ Γαβ u μ u ν = 0. (A.68)
ds
The k-component of this equation is

du k  5 2
+ ikj u i u j + 25ik u 5 u i + 55
k
u = 0. (A.69)
ds
Inserting the expressions of the 5-dimensional Christoffel symbols from Eq.
(A.17) gives

du k 4 k i j 1 
+ i j u u = Ai F jk + A j Fik u i u j + Fik u 5 u i . (A.70)
ds 2

Since u i u j is symmetric in i and j this can be written as

du k 4 k i j
+ i j u u = A j Fik u i u j + Fik u 5 u i . (A.71)
ds
Appendix: Kaluza–Klein Theory 403

The Lagrange function of a free particle in the 5-dimensional spacetime is obtained

from the line-element (A.11) and is

1 1 2
L= γi j ẋ i ẋ j + ẋ 5 + Ai ẋ i , (A.72)
2 2
where the dot denotes differentiation with respect to the proper arc length along the
world line of the particle. Since x 5 is a cyclic coordinate the momentum p5 conjugate
to x 5 is a constant of motion,
 
p5 = m 0 u 5 + A j u j = m 0 u 5 . (A.73)

where m 0 is the rest mass of the particle. From this equation we get

u 5 = p5 − A j u j . (A.74)

Inserting this into Eq. (A.71) gives

 
du k 4 k i j
m0 + i j u u = p5 Fik u i , (A.75)
ds
or

m 40 a k = p5 Fik u i , (A.76)

where 4 a k is the k-component of the particle’s 4-acceleration. It was shown by Lei-

bowitz and Rosen [1] that the physical components of the electromagnetic field tensor
are
1
F̂ i j = √ Fij . (A.77)
c ε0 κ

where ε0 is the permittivity of empty space, and we have used S.I. units here. Hence,
Eq. (A.76) takes the form
√
m 40 a k = c ε0 κ p5 F̂ik u i . (A.78)

Comparing with the equation of motion of a charged particle in an electromagnetic

field, Eq. (5.307) we get
√
q = c ε0 κ p5 . (A.79)

Hence the conserved covariant momentum of the particle in the fifth direction,
i.e. around the cylindrical fifth dimension, is interpreted as the charge of the particle
from our 4-dimensional point of view. Inserting the expression (7.37) for Einstein’s
gravitational constant gives
404 Appendix: Kaluza–Klein Theory

1
q= 8π ε0 G p5 . (A.80)
c
The Kaluza–Klein theory may provide an explanation of the fact that gravita-
tional waves move with the velocity of light. According to this theory all of electro-
magnetism comes from projection of gravity in a 5-dimensional world into our 4-
dimensional spacetime. According to this theory electromagnetic waves are the pro-
jection of certain gravitational waves that propagate isotropically. Hence the projected
waves, both those projections that are interpreted as electromagnetic waves from
our 4-dimensional point of view, and the gravitational waves in our 4-dimensional
spacetime, move with the same velocity. This should be expressed by saying that
electromagnetic waves move with the velocity of gravity, i.e. of gravitational waves.
In this connection there appears a problem which has not, so far, been solved.
According to the general theory of relativity the source of a gravitational wave must
change its quadrupole moment; i.e. it must change its shape. A particle cannot emit
gravitational waves. But if the particle is electrically charged, it can emit electromag-
netic waves. According to Larmor’s formula it will emit electromagnetic radiation
when it is accelerated.
In the 5-dimensional world a particle which we call charged is a neutral particle
moving around the fifth spatial cylinder dimension. Hence in order that a charged
particle in our world should emit electromagnetic radiation when it is accelerated,
a neutral particle moving around the compact fifth dimension should emit those
gravitational waves that are interpreted as electromagnetic waves in our world, when
it is accelerated in a direction orthogonal to its circular motion in the fifth direction.
Whether this is really the case is still an unsolved problem.

A.5 Quantization of Charge as a Consequence

of Quantization of Momentum Along a Closed Path
Around the Fifth Cylinder Dimension

Due to the closed character of the fifth cylinder dimension the momentum p5 is
periodic. As shown by Klein [2] this implies a quantization of p5 according to

r5 p5 = n, (A.81)

where r5 is the radius of the cylindrical fifth dimension. Thus

1 
q= 8π ε0 Gn. (A.82)
r5 c

This means that in the Kaluza–Klein theory quantization of charge need not be
postulated. It follows as a consequence of quantization of momentum around the
closed fifth dimension. This is again a consequence of the de Broglie relationship
Appendix: Kaluza–Klein Theory 405

associating a wave with the momentum. Then quantization follows from the require-
ment of constructive interference, demanding a whole number of wavelengths around
the fifth dimension.
The quantum of charge is the elementary charge e. Inserting q = e into Eq. (A.82),
solving the resulting equation with respect to r5 and putting n = 1gives the smallest
allowed radius of the cylinder dimension,
√
 8π ε0 G
r5 = . (A.83)
ce
The Planck length is

G
lP = . (A.84)
c3

Inserting the values of the constants gives l P ≈ 1.6 × 10−35 m. The fine-structure
constant is a dimensionless number

e2 1
α= ≈ . (A.85)
4π ε0 c 137

Thus, in terms of the Planck length the radius of the fifth cylinder dimension is

2
r5 = lP , (A.86)
α

giving r5 ≈ 16.6l P ≈ 2.6 × 10−34 m. This may represent a quantum of length in

5-dimensional spacetime.

A.6 Electric Field from Inertial Dragging in the Fifth

Dimension

In Kaluza–Klein theory electric charge in the projection of momentum in the fifth

dimension. It is natural then to wonder. What in the 5-dimensional world with gravity
only is the electrical field of a charged particle the projection of?
In order to answer this question we shall consider a solution of the field equations
for empty 5-dimensional space that was found some years ago [4],
 
b 3a 2 2 2 dr 2 a  2
ds = − 1 + − 2 c dt +
2
a2
+ r 2 d2 − dtdx 5 + dx 5 .
r r 1 + br + r2
r
(A.87)
406 Appendix: Kaluza–Klein Theory

where a and b are constants. Projecting this line-element into ordinary 4-dimensional
spacetime we obtain
 
b a2 2 2 dr 2
ds = ds = − 1 + + 2 c dt +
2 2
a2
+ r 2 d2 . (A.88)
r r 1 + br + r2

With the identification b = R S and a = Rq as given in Eq. (8.164), this repre-

sents the Reissner–Nordström solution of Einstein’s field equations in 4-dimensional
spacetime with an electric field.
It remains to identify which part of the gravitational field in 5-dimensional space-
time is interpreted as an electric field from our 4-dimensional point of view. Let us
first identify the physical meaning of the constant a in the line-element (A.87) from
the 5-dimensional point of view.
Consider a test particle with p5 = 0 corresponding to a zero-angular momentum
particle in the Kerr spacetime as was described in Sect. 10.3.1. According to Eq.
(A.74) it has

u 5 = ẋ 5 = −A j u j . (A.89)

Assume the particle is at rest in the spatial 3-space of the 4-dimensional spacetime.
Then the only non-vanishing component of the 4-velocity is u 0 = ct˙. Hence

ẋ 5 = −A0 t˙. (A.90)

According to Eq. (A.11) we have

A0 = g50 = g05 . (A.91)

With the line-element (A.87)

a
g05 = − . (A.92)
r
Thus
b
A0 = − , (A.93)
r
and
a
ẋ 5 = t˙. (A.94)
r
This gives
Appendix: Kaluza–Klein Theory 407

dx 5 ẋ 5 a
= = . (A.95)
dt t˙ r
This shows that a particle with vanishing conserved momentum in the 5-direction
will drift through the coordinate system in the 5-direction. This is inertial dragging.
Thus from the 5-dimensional point of view the constant a represents inertial drag-
ging due to the motion of the central body moving in the 5-direction. Hence the
central body moving around the fifth direction causes an inertial dragging field. The
projection of this field into 4-dimensional spacetime is interpreted from our point of
view as the electric field due to the charge of this central body.
Let us conclude with a speculation. One may wonder whether it is possible to
use of our knowledge of quantum electrodynamics within the frame of Kaluza–
Klein theory to shed some light upon how we should quantize gravity. In what way
should one quantize gravity in order that the projection of quantized gravity into
our 4-dimensional spacetime shall give us the quantum electrodynamic theory? The
inertial dragging field may play a role in this programme.

References

1. Kaluza T.: Zum Unitätsproblem in der Physik. Sitzungsber. Preuss. Akad. Wiss.
Berlin. Math. Phys. 966–972 (1921)
2. Klein, Q.: Quantentheorie und fünfdimensionale Relativitätstheorie. A 37(12),
895–906 (1926)
3. Leibowitz, E., Rosen, N.: Five-dimensional relativity theory. Gen. Rel. Grav. 4,
449–474 (1973)
4. Grøn, Ø.: Classical Kaluza-Klein description of the hydrogen atom. Nuovo
Cimento B91, 57–66 (1986)
Solutions to the Exercises

Solutions for Chapter 1

1.1 A tidal force pendulum

Two particles each with mass m are connected by a rigid rod of length 2 l. The system
is free to oscillate in any vertical plane about its centre of mass. The mass of the rod is
negligible relative to m. The pendulum is at a distance R from the centre of a spherical
distribution of a spherical distribution of matter with mass M (see Fig. 1.10).

|l × (F1 − F2 )| = I θ̈

where I = 2ml 2 is the moment of inertia of the pendulum.

By Newton’s law of gravitation

R+l R−l
F1 = −G Mm , F2 = −G Mm .
|R + l| 3
|R − l|3

Thus
l×R l×R
G Mm − G Mm = 2ml 2 θ̈.
|R − l| 3
|R + l|3

It is seen from Fig. 1.10 that

|l × R| = l R sin θ.

It is now assumed that l R. Then we have to first order in l/R.

1 1 6l
− = 4 cos θ.
|R − l|3 |R + l|3 R

© Springer Nature Switzerland AG 2020 409

Ø. Grøn, Introduction to Einstein’s Theory of Relativity,
Undergraduate Texts in Physics, https://doi.org/10.1007/978-3-030-43862-3
410 Solutions to the Exercises

The equation of motion of the pendulum now takes the form

3G M
2θ̈ + sin 2θ = 0.
R3
This is the equation of motion of a simple pendulum in the variable 2θ instead
of as usual θ as a variable. The equation shows that the pendulum oscillates about
a vertical equilibrium position. The reason for 2θ instead of θ is that the tidal force
pendulum is invariant under a change θ → θ + π while the simple pendulum is
invariant under a change θ → θ + 2π .
Assuming small angular displacements leads to

3G M
θ̈ + θ = 0.
R3
This is the equation of a harmonic oscillator with period
 1/3
R3
T = 2π .
3G M

Note that the period of the tidal force pendulum is independent of its length. This
means that tidal forces can be observed on a system of arbitrarily small size. Also,
from the equation of motion it is seen that in a uniform field, where F1 = F2 , the
pendulum does not oscillate.
The acceleration of gravity at the position of the pendulum is g = G M/R 2 , so
that the period of the tidal pendulum may be written
 1/2
R
T = 2π .
3g

The mass of a spherical body with density ρ is M = (4π/3)ρ R 3 , which gives for
the period of the tidal pendulum at its surface
 1/2
πG
T = .
ρ

Hence, the period depends only upon the density of the body. For a pendulum at
the surface of the Earth the period is about 50 min. The region in spacetime needed
in order to measure the tidal force is not arbitrarily small.

1.2 Newtonian potential for a spherically symmetric body

(a) We can find the potential using Gauss’ law

Solutions to the Exercises 411
 
 )d S =
F(r  )dV ,
∇ · F(r
∂V V

where the integration on the left-hand side is over a spherical surface (which may
be inside or outside the mass shell), and the integral on the right-hand side is over
the volume enclosed by this surface. Using Newton’s law of gravity as formulated
locally, we have

 r ) = −m∇ 2 φ(
∇ · F( r ) = −4π mGρ(
r ),

where φ(r ) is the gravitational potential, ρ(r ) is the mass density at a point with
position vector r, and m is the mass of a test particle at this position. Inserting this
into Gauss’ law gives

4πr 2 F(r ) = −4π Gm ρ(
r )dV

The value of the integral at the right-hand side depends upon whether the boundary
surface of the integration volume is inside or outside the mass distribution. We get

 r) = − G rMm
3 rr ≥ R
F( .
0 r<R

Defining zero potential infinitely far from the mass distribution, the potential at a
finite distance r from the centre is found by calculating the (negative) work performed
to move a particle downwards from the zero level to a distance r from the centre,
⎧ r
⎪ 
r ⎪
⎨−
GM
r2
dr = − GrM r ≥ R
1
φ(
r) = − F · ds = ∞
R .
m ⎪
⎪
∞ ⎩− GM
r2
dr = − GRM r < R
∞

(b) To calculate the gravitational potential outside an inside a sphere of constant

density we use the same procedure as in point (a). Outside the sphere the result
becomes the same as in point (a). Inside the sphere we now get

r
4π 3 r3
ρ(
r )dV = ρr = M 3 .
3 R
0

Hence,
412 Solutions to the Exercises

G Mmr 3
4πr 2 F(r ) = −4π .
R3
The gravitational force outside and inside the sphere is then

 r) = − G rMm
3 rr ≥ R
F( .
− R 3 r r < R
G Mm

This leads to the gravitational potential


− GrM r≥R
φ(
r) = G M (3R 2 −r 2 ) .
− 2R 3
r<R

1.3 Frictionless motion in a tunnel through the Earth

In order to find the equation of motion of a particle moving without friction in the
tube under the action of gravity only, we insert the expression for the gravitational
force inside a sphere from Exercise 1.2 into Newton’s 2. law and get

G Mmr
m r̈ = − .
R3
This is the equation for harmonic oscillations with period

R3
T = 2π .
GM
√
This is just 3 times the period of the tidal force pendulum at the surface of the
Earth (see Exercise 1.1). The period is 1 h and 24 min.
(b) The figure illustrates the situation with a tube not passing through the centre of
the Earth, but having a closest distance s from the centre.

Here x = r sin θ and the component of the gravitational force upon the direction
of motion of the particle is
Solutions to the Exercises 413

G Mmr G Mm
F = F(r ) sin θ = − 3
sin θ = − x.
R R3
Hence the equation of motion takes the form

G Mm
m ẍ = − x.
R2
This gives the same period as when the tube passes the centre of the Earth, showing
that the period does not depend upon the direction of the tube. Using this procedure
(and neglecting friction and the rotation of the Earth) it takes about 42 min to travel
from an arbitrary point on the surface on the Earth to whatever other point on the
Earth, and the traveller would be weightless during all of the travel.

(c) The equation of motion of a satellite moving in a circular path around the Earth
at the surface of the Earth takes the form

G Mm v2
= m ,
R2 R
where v is the velocity of the satellite. Hence, the velocity is

GM
v= .
R

Since the magnitude of the velocity is constant, the time taken to move around
the Earth is

2π R R3
TC = = 2π ,
v GM
which is equal to the period of the frictionless motion through the tube.

1.4 The Earth-Moon System

(a) The Earthand the Moon have positions rJ , rM and masses M J , M M , respectively,
as shown in the figure.
414 Solutions to the Exercises

The equations of motion for the Earth and the Moon are

G M J MM
M J r¨ J = −M M r¨ M = r M − rJ ).
(
|
r M − rJ |3

The position of the centre of mass is

M J rJ + M M rM
rC M = .
M J + MM

Since there are no external forces on the Earth-Moon system the mass centre is
at rest. We may therefore choose a coordinate system co-moving with an inertial
reference frame, in which the mass centre is at the origin, M J rJ = −M M rJ .
We shall show that there is a solution where the Earth and the Moon are moving
in circular orbits around their common centre of mass. For the Earth such a motion
may be represented by

x J (t) = A sin(ωt), y J (t) = A cos(ωt).

where x J and y J are the components of the position vector rJ , and A is the distance
of the Earth from the mass centre. Hence the motion of the Moon is given by

MJ MJ
x M (t) = − A sin(ωt), y M (t) = − A cos(ωt).
MM MM

The x-component of the equation of motion of the Earth can now be written
 
G M J MM MJ
M J ω2 A sin(ωt) = A 1 + sin(ωt).
|
r M − rJ |3 MM

Giving
Solutions to the Exercises 415
 
G MM MM
ω2 = 1 + .
|
r M − rJ |3 MJ

Furthermore
  
MJ
|
r M − rJ | = (x M − x J ) + (y M − y J ) = A 1 +
2 2
.
MM

Hence the equation of motion is fulfilled for the circular path if

G MM
ω2 A3 = .
(1 + M J /M M )2

Inserting the values of the gravitational constant and the masses of the Earth and
the Moon together with the period of the Moon’s motion, T = 27.3 days, we get
ω = 2π/T = 2.66 × 10−6 s −1 . Hence the radius of the circular motion of the Earth
is A = 4670 km, and that of the Moon’s motion (M J /M M )A = 3.8 × 105 km.

(b) The gravitational potential of the Earth–Moon system is the sum of the potential
of each of the bodies. Along the line connecting the Earth and the Moon we get

G MJ G MM
φ(r ) = − − ,
r R −r

where r is the distance from the centre of the Earth, and R is the distance between
the centres of the Earth and the Moon. The gravitational force upon a test particle
with mass m is
G MJ m G MM m
F(r ) = − + .
r2 (R − r )2

This force vanishes at two points given by

√
M J ± M J MM
r= R.
M J − MM

Inserting the Masses of the Earth and the Moon and the distance R = 3.8×105 km
gives r1 = 3.46 × 105 km and r2 = 4.32 × 105 km. The first one is between the
Earth and the Moon, and the second one is beyond the Moon.

(c) The difference between the Moon’s attraction upon a particle with mass m at
a point on the surface of the Earth closest to the Moon and at the most remote
point is

G MM m G MM m RE
F = − ≈ 4G M M m 3 ,
(R − R E )2 (R + R E )2 R
416 Solutions to the Exercises

where R E is the radius of the Earth. This gives F = 2.2 × 10−6 N .

1.5 The Roche limit

(a) A Moon with mass m moves around a planet with mass M. A stone with mass μ
on the surface of the Moon is acted upon by the gravitational field of the planet
and the Moon. Calculating with absolute values, the gravitational force acting
upon the stone towards the planet is equal to the gravitational force due to the
planet minus the gravitational force due to the Moon,

G Mμ Gmμ
F= − ,
(r − R)2 R2

where R is the radius of the Moon and r the distance between the centres of mass of
the Moon and the planet. The centripetal acceleration of the stone is
 
F GM Gm GM R −2 Gm
a= = − 2 = 2 1− − 2.
μ (r − R)2 R r r R

Since R r we can use the series expansion

 
R −2 2R
1− ≈1+ ,
r r

giving

GM 2G M R Gm
a≈ + − 2.
r2 r3 R
The Moon has a centripetal acceleration

GM
a0 = .
r2
In order that a stone at the point of the surface of the Moon closest to the Earth shall
not be lifted up from the surface of the Moon, the centripetal acceleration of the stone
cannot be larger than the centripetal acceleration of the Moon. So the acceleration of
the stone must obey the condition a ≤ a0 . Inserting the expressions above this gives

2M R m
3
≤ 2,
r R
or
 1/3
2M
r≥ R.
m
Solutions to the Exercises 417

The right-hand expression is called the Roche limit. If a Moon is inside the Roche
limit of the Moon-planet system, matter will be drawn from the surface of the Moon
towards the planet. Similarly, if a planet is inside the Roche limit of the planet-star
system matter will be drawn from the planet onto the star. Or if a star is inside the
Roche limit of a system consisting of the star and for example a super massive black
hole at the centre of a galaxy, matter will be drawn from the star into the black hole.
This is an essential part of the mechanism behind quasars.
(b) From the observations of what happened to the comet Shoemaker–Levy 9 when
it passed Jupiter in 1992, it is reasonable to assume that the smallest distance of
the comet from Jupiter was equal to the Roche limit of the comet-planet system.
Then the radius of the comet nucleus is given by

 m 1/3
R= s.
2M

Inserting m = 2.0 × 1012 kg, M = 1.9 × 1027 kg and s = 96, 000 km gives
R = 775 m.

Solution for Chapter 2

2.1 Robb’s Lorentz invariant spacetime interval formul

Consider Fig. 2.2. We shall find a formula for the invariant spacetime interval s AB
between the emission event A and the reflection event B of a light signal reflected
by a mirror, in terms of the emission point of time t A and the point of time tC when
the reflected light signal arrives back at the emission position.
Our point of departure is the usual Formula (2.41) for the invariant spacetime
interval, the Einstein synchronization Formula (2.1) and Eq. (2.2). Hence we have

1 c
t AB = (t A + tC ), x AB = (tC − t A ),
2 2
and

c2 c2
s 2AB = c2 t AB
2
− x 2AB = (t A + tC )2 − (tC − t A )2 = c2 t A tC .
4 4
Hence we arrive at Robb’s formula
√
s AB = c t A tC .

2.2 The twin paradox

(a) In the reference frame of the Earth A travels with a velocity v = 0.8c. B remains
at the Earth. Hence according to B the travel takes 8 light years/0.8 c = 10 years.
Hence B sends 10 greetings.
418 Solutions to the Exercises

Let us now describe the situation as seen from the reference frame of A. The
twin A thinks of himself as at rest while the Earth and Alpha Centauri travels with
the velocity v = 0.8c. Hence the distance  between the Earth and Alpha Centauri
becomes length contracted to L A = L 1 − v 2 /c2 = 2.4l.y. The time taken for
Alpha Centauri to arrive at A is therefore 2.4 l y/0.8 c = 3 years. The return trip takes
the same time, so A sends 6 greetings.
(b) The figure shows the world lines of the twins and the signals.

The figure shows that the twin A receives his first greeting at the moment he
arrives at Alpha Centauri. He receives the 9 other signals from B at the return travel.
(c) The figure shows that B receives the first greeting from A after 3 years, the next
after 6 years and the last year he receives the 3 signals A send during the return
travel.

(d) Due to the Doppler effect a light signal sent by A with a frequency ν A is measured
by B to have a frequency


1 − v/c
νB = νA.
1 + v/c

Hence, if the emitted frequency is 1/year, the received frequency is 1/3 years, and
during the return travel the received frequency is 3 signals per year.

2.3 Faster than the speed of light?

(a) The situation is schematically illustrated on the Figure. The radiation source
is moving from A towards B with a velocity v0 . The arrows at the bottom of
the figure point towards an observer at the Earth. Here θ is the angle between
Solutions to the Exercises 419

the direction of motion of the light source and the direction of sight. We now
consider light emit at A and B. The point A has the same distance from the
Earth as B. We see from the figure that
420 Solutions to the Exercises

a a
AB = , A A = .
sin θ tan θ
The difference in arrival time on the Earth for light emitted from A and B is

AB A A a a
t = − = − ,
v0 c v0 sin θ c tan θ

giving

1 − (v0 /c) cos θ

t = a.
v0 sin θ

(b) The observed transverse velocity normal to the line of sight is

a v0 sin θ
v= =
t 1 − (v0 /c) cos θ

Hence, the velocity of the light source is

v
v0 = .
sin θ + (v/c) cos θ

Inserting v = 10c and θ = 100 gives v0 = 0.998c.

The observed transverse velocity is larger than c if
c
v0 > .
sin θ + cos θ

with θ = 100 this requires v0 > 0.86c.

2.4 Time dilation and Lorentz contraction


t v2 v2
(a) Theformula for the time dilation is t0
= 1
γ
= 1− . Hence 1
γ2
= 1−
 c2 c2

and v = c 1 − γ12 .

In the present case 1

γ
= 0.6 giving v = 0.8c.

(b) Let L x and L y be the components of the rod in the x- and y-directions, respec-
tively, as measured by an observer following the rod. As measured by this
observer the angle θ0 which the rod makes with the x-axis, is given by

Ly
tan θ0 =
Lx
Solutions to the Exercises 421

As measured by an observer at rest on the x-axis the component in the y-direction

is unchanged, and the component in the x-direction is L x /γ . Hence this observer
measures that the rod makes an angle θ given by

Ly
tan θ = γ .
Lx

Thus

1 v2 tan θ0
= 1− = ,
γ c 2 tan θ

giving
  2
tan θ0
v =c 1− .
tan θ
√
With θ0 = π/4 and θ = π/3 we get v = 2/3c ≈ 0.82c.

2.5 Atmospheric mesons reaching the surface of the Earth.

(a) The travel time of a muon is T = Lv = 34 × 10−6 s, or 21.8 half lives. Thus the
survival rate is I /I0 = 2−21.8 ≈ 0.3 × 10−6 . Hence only 3 of ten million created
muons in the upper atmosphere reaches the surface of the Earth.
(b) Due to the relativistic time dilation the actual half life of the travelling muons is
t = γ t0 =  t0 v2 ≈ 5t0 = 7.8 × 10−6 s. Then the survival rate is 0.049 meaning
1− c2
that 4900 of ten million muons reach the surface of the Earth.
(c) As observed by a co-moving observer with a muon its half life is 1.56 · 10−6 s.
But as measured by this observer the distance is Lorenz contracted to L =
(1/γ )L 0 = 2 km. Hence the survival rate is in agreement with the result of the
Earth-bound observer.

2.6 Relativistic Doppler effect

Solving the equation for the redshift with respect to the velocity we get

(1 + z)2 − 1
v= .
(1 + z)2 + 1

Since |z| 1 in the present case, we can use the approximation (1 + z)2 ≈
1 + 2z. This gives to 1 order in z, v ≈ zc. For the centre of the Andromeda galaxy
z = −0.0004, giving v = −120 km/s. This means that the Andromeda galaxy moves
towards the Milky Way galaxy with this velocity.
422 Solutions to the Exercises

2.7 The velocity of light in a moving medium

The velocity of light in a medium with index of refraction, n, moving with velocity
v is

u0 + v c
+v c + nv
u= u0 v = n
v
= c.
1 + c2 1+ n c
1 nc + v

This formula is exact. In most applications v c and one uses an approximate

formula,
   
c 1+ nv
c v 1v c 1
u= c
v ≈ 1+n 1− ≈ +v 1− 2 .
n 1+ vc
n c nc n n

2.8 Cherenkov radiation

The geometry of the light cone formed when a source of light passes through a
medium having index of refraction n with a velocity v greater than that of light in
the medium is shown in the Figure (From Wikipedia: https://en.wikipedia.org/wiki/
Cherenkov_radiation#/media/File:Cherenkov.svg

We see from the figure that the angle θ is given by cos θ = c

nv
.

(a) In order that the electron shall emit Cherenkov radiation it must move through
the glass with a velocity v > c/n. Hence the threshold kinetic energy in order
that it shall emit Cherenkov radiation is

   
1 n
K =  − 1 m e c2 = √ − 1 m e c2 .
1 − v 2 /c2 n2 − 1

Inserting the given data gives K = 0.29 MeV.

Solutions to the Exercises 423

(b) The limiting velocity for increasing energy is v → c. In this limit the angle θ
approaches θmax given by cos θmax = n1 . From the figure is seen that the corre-
sponding maximal half angle is given by sin φmax = n1 , Inserting the refractive
index for glass gives φmax ≈ 50◦ .
2.9 Relativistic form of Newton’s 2. law
 −1/2
Using the chain rule of differentiation we first differentiate γ = 1 − v 2 /c2 ,

dγ 1 −1/2  v dv v dv
= − 1 − v 2 /c2 −2 2 = γ3 2 .
dt 2 c dt c dt
The momentum of the particle is p = γ m 0 v. Differentiating we get
 
1 dp dγ dv v 2 dv dv v2 dv dv
=v +γ = γ3 2 +γ = γ2 2 + 1 γ = γ3 .
m 0 dt dt dt c dt dt c dt dt

Hence Newton’s 2. law has the relativistic form

dv
F = γ 3m0 .
dt

2.10 Lorentz transformation of electric and magnetic fields

We write out Eqs. (5.281) and (5.282) in component form

∂ Bx ∂ By ∂ Bz
+ + = 0,
∂x ∂y ∂z
∂ Ez ∂ Ey ∂ Bx ∂ E x ∂ Ez ∂ By ∂ E y ∂ Ex ∂ Bz
− =− , − =− , − =− .
∂y ∂z ∂t ∂z ∂x ∂t ∂x ∂y ∂t

Inserting the transformations of the partial derivatives into the y-component of

the second equation, we get
   
∂ Ex ∂ Ez v ∂ Ez ∂ By ∂ By
−γ + 2  =− +v  .
∂z  ∂x c ∂t ∂t  ∂x

This equation shall be compared with the similar Maxwell equation in the marked
coordinate system,

∂ E x ∂ E z ∂ B y
− = − .
∂z  ∂x ∂t 

Hence, we rearrange the above equation as follows

∂ E x ∂   ∂  v
− γ E z − cB y = − γ B y − Ez .
∂z  ∂x ∂t  c2
424 Solutions to the Exercises

Requiring equivalence of these equations gives the transformation equations

   v
E x = E x , E z = γ E z − v B y , B y = γ B y − 2 E z .
c
Using all the Maxwell-equations in this way we end up with the transformation
equations of the electric and magnetic field

E x = Ex , 
 Bx = Bx , 
E y = γ E y + v Bz , B y = γ B y − cv2 E z

   
E z = γ E z − v B y , Bz = γ Bz + cv2 E y

Solutions for Chapter 3

3.1 Four-vectors

(a) The scalar products of the vectors with themselves are

   
A · A = 4et + 3ex + 2e y + ez · 4et + 3ex + 2e y + ez = −2,
B · B = 5et + 4ex + 3e y · 5et + 4ex + 3e y = 0, 
.
C · C = et + 2ex + 3e y + 4ez · et + 2ex + 3e y + 4ez = 28

(b) We now assume that A · B = 0. If A is time-like, i.e. A · A < 0, we may chose a

basis where A points in the time direction, A = At et  . Then A · B = −At B t .
  

Hence, the condition A · B = 0 requires that B t = 0, which means that B is



space-like.

If A is light-like, then A · A = 0. Assume now that B is time-like. Hence, we can

chose a basis where B = B t et  . This gives A · B = −At B t . The vector component
  

At cannot vanish since A is light-like. This means that A · B = 0 in contradiction

to the initial assumption. Thus B cannot be time-like.
If both A and B are light-like we may choose a basis where A = A(et  + ex  ).

Then A · B = A −B t + B x . In this case A · B = 0 implies that B x = B t , giving
  

 y  2  z  2  y  2  z  2
B · B = B + B . Since B is light-like B · B = B + B = 0 which
y z 
requires B = B . Hence B = B (et  + ex  ) showing that the vectors A and B have
t 
the same direction; i.e. they are proportional to each other.
If A is space-like we may chose a basis in which only one of the spatial components
of the vectors are is different from zero, say A = A x ex  . This gives A · B =


  
A x B x , and hence B x = 0. There are, however, no further requirements on the other
components of B.  This means that the vector B can be either time-like, light-like or
space-like.
Solutions to the Exercises 425

(c) A Lorentz transformation along the x-axis has the form, measuring distance in
light seconds so that c = 1,

t  + vx  x  + vt 
t=√ , x=√ , y = y, z = z
1 − v2 1 − v2

where v is the relative velocity of the reference frames. We now define the velocity
parameter θ by tanh θ = v. Using the relationships

1 tanh θ
cosh θ =  , sinh θ = 
1 − tanh θ 2
1 − tanh2 θ

The Lorentz transformation takes the form

t = t  cosh θ + x  sinh θ, x = x  cosh θ + t  sinh θ, y = y, z = z.

Hence the transformation matrix becomes

⎛ ⎞
cosh θ sinh θ 0 0
∂xμ ⎜ sinh θ cosh θ 0 0 ⎟
μμ = =⎜
⎝ 0
⎟.
∂ x μ 0 1 0⎠
0 0 01

(d) The 4-velocity u is a vector in 4-dimensional spacetime. This vector may be

described by expressing its components in a chosen basis. But the vector itself
exist independently of any basis. This is expressed mathematically by


u = u μ eμ = u μ eμ ,

where the components u μ and u μ depend upon the basis, but u is independent of it.
The component of the 4-velocity in an arbitrary basis are given by

u = γ (1, v),
   −1/2
where v = dx i /dt ei is the ordinary velocity in 3-space and γ = 1 − v 2 . The
3-velocity is not a vector in 4-dimensional spacetime. It depends upon the reference
frame of the observer.
It follows from the 4-velocity identity, u · u = −1 that u is a time-like vector.
Also the 4-momentum p = m u is time-like. In the limit m → 0 the 4-momentum
becomes light-like.
426 Solutions to the Exercises

The energy of the particle with rest mass m and 3-velocity v is E = γ m. The
velocity of the observer is uobs = (1, 0). Hence

u obs · p.
E = −

This formula is valid independent of the basis.

3.2 Tensor product

(a) We get

α− ⊗ β (e0 , e1 ) = α0 β1 = 0, β ⊗ α− (e0 , e1 ) = β0 α1 = −1.

− −

Hence α− ⊗ β = β ⊗ α− .
− −

(b) Calculation of all the components of α− ⊗ β give

−

⎛ ⎞
−1 0 1 0
⎜ −1 0 1 0⎟
αμ βν = ⎜
⎝ 0
⎟.
0 0 0⎠
0 0 0 0

The symmetric part of α− ⊗ β has components

−

⎛ ⎞
−2 −1 1 0
1  1 ⎜ −1 0 1 0⎟
α(μ βν) = αμ βν + αν βμ = ⎜ ⎟,
2 2⎝ 1 1 0 0⎠
0 0 0 0

and the antisymmetric part of α− ⊗ β has components

−

⎛ ⎞
0 1 1 0
1  1 ⎜ −1 0 1 0⎟
α(μ βν) = αμ βν − αν βμ = ⎜ ⎟.
2 2 ⎝ −1 −1 0 0⎠
0 0 0 0

3.5 Symmetric and antisymmetric tensors

We have
Solutions to the Exercises 427

⎛ ⎞
0 1 0 0
⎜1 −1 0 2⎟
T αβ =⎜
⎝2
⎟.
0 0 1⎠
1 0 −2 0

1. The symmetric part of this tensor is

⎛ ⎞
0 2 2 1
1  αβ  1⎜2 −2 0 1 ⎟
T (αβ) = T +T βα
= ⎜ ⎟.
2 2⎝2 0 0 −1 ⎠
1 2 −1 0

The antisymmetric part of the tensor is

⎛ ⎞
0 0 −2 −1
1  αβ  1 ⎜ 0 0 0 1 ⎟
T (αβ) = T − T βα = ⎜ ⎟.
2 2⎝2 0 0 3 ⎠
1 −1 −3 0

2. The mixed components of the tensor are

⎛ ⎞
0 1 0 0
⎜ −1 −1 0 2⎟
Tβα = ηβμ T αμ =⎜
⎝ −2
⎟.
0 0 1⎠
−1 0 −2 0

3. The covariant components are

⎛ ⎞
0 −1 0 0
⎜ −1 −1 0 2⎟
Tαβ = ηαμ ηβν T μν =⎜
⎝ −2
⎟.
0 0 1⎠
−1 0 −2 0

(b) It is not possible to define a symmetric or antisymmetric part of a mixed tensor

because it is not allowed to exchange a vector by a form in the argument.
3.4 Contractions of tensors with different symmetries
SinceA is antisymmetric we may write

1  αβ 
Aαβ = A − Aβα ,
2
and since B is symmetric we may write
428 Solutions to the Exercises

1 
Bαβ = Bαβ + Bβα .
2
Hence, we get

1  αβ  
Aαβ Bαβ = A − Aβα Bαβ + Bβα
4 ,
1  αβ 
A Bαβ + Aαβ Bβα − Aβα Bαβ − Aβα Bβα
=
4
1   1  αβ 
Aαβ Cαβ = Aαβ − Aβα Cαβ = A Cαβ − Aβα Cαβ
2 2 ,
1  αβ  1 
= A Cαβ − A Cβα = Aαβ Cαβ − Cβα = Aαβ C[αβ]
αβ
2 2
1   1  
Bαβ D αβ = Bαβ D αβ + Bβα D αβ = Bαβ D αβ + D βα = Bαβ D (αβ) .
2 2
3.5 Coordinate transformation in an Euclidean plane

(a) The given transformation can be written

    
x 2 −1 x
= .
y 1 1 y

  
This means that we have a transformation matrix M ij given by x i = Mii x i which
is
 
 2 −1
Mii = .
1 1
 
The inverse transformation matrix is given by Mii M ij  = δ ij  and is
 
1 1 −1
M ij  = .
3 1 2

(b) The transformation of the basis vectors is ei  = Mii ei , giving
   
ex  = (1/3) ex − ey , ey  = (1/3) ex + 2e y .
!
(c) The scalar products of the basis vectors in the x  , y  -system are

1    2 1   
ex  · ex  = ex − ey · ex − ey = , ey  · ey  = ex + 2e y · ex + 2e y
9 9 9 .
5 1    1
= ex  · ey  = ex − ey · ex + 2e y = − = ey  · ex 
9 9 9
Solutions to the Exercises 429

Hence, in this system the covariant components of the metric tensor are
 
1 2 −1
gi  j  = .
9 −1 5
!
(d) The line-element in the x  , y  -system is
   
ds 2 = gi  j  dx i dx j = (1/9) 2dx  2 + 5dy  2 − 2dx  dy  .

(e) The vectors ω
 i can be expressed in terms of the Cartesian basis vectors ex and
ey by
  
i  2−1 ex
 =
ω Mii ei = ,
1 1 ey

giving
 
 x = 2ex − ey ω
ω  y = ex + ey .

The scalar products of these vectors are

           
ω  x = 2ex − ey · 2ex − ey =
x · ω  5, ω
 y · ω
 y = ex + ey · ex + ey = 2
    .
y · ω
ω x = ω  y = 2ex − ey · ex + ey = 1
x · ω
!
Hence, the contravariant components of the metric tensor in the x  , y  -system
are
 
i j 51
g = .
12

Solutions for Chapter 4

4.1 Relativistic rotating disk

We use units so that the velocity of light is c = 1.

(a) Let dl be the physical distance measured by standard measuring rods at

rest in the reference frame between two points with coordinates (r, φ) and
(r + dr, φ + dφ), where (r, φ) are co-moving coordinates with a frame RF rotat-
ing with angular velocity ω. A radial distance dlr = dr is equal to that of the
non-rotating inertial system IF of the axis. A physical distance, In the tangential
direction the standard measuring rods will be Lorentz contracted, and therefore
 −1/2
the measured distance will be larger in RF than in IF, dlφ = 1 − r 2 ω2 r dφ.
430 Solutions to the Exercises

Using the Pythagorean rule the square of the distance of an interval with both a
radial and tangential component is

r2
dl 2 = dlr2 + dlφ2 = dr 2 + dφ 2 .
1 − r 2 ω2

Hence

r2
f 1 (r, φ) = 1, f 2 (r, φ) = .
1 − r 2 ω2

The distance in the radial direction between the axis and a point with coordinates
R is lr = R. The distance around a circle with this radius is

2π R
lφ = √ .
1 − R 2 ω2

Since the observer finds that lφ = 2π R he will conclude that the geometry is not
Euclidean. On a positively curved surface (for example a part of a spherical surface)
the periphery is less than 2π R and on a part of a negatively curves surface (saddle
like) it is larger. Hence the observer concludes that the simultaneity surface with
constant z is negatively curved in the rotating frame.

(b) In order to express the line-element

ds 2 = −dt 2 + dx 2 + dy 2

in terms of t˜, x̃ and ỹ we must first use Eqs. (4.54) to express t, x and y in terms of
t˜, x̃ and ỹ. This gives t = t˜ and

x = r cos(ωt + φ) = r cos φ cos(ωt) − r sin φ sin(ωt) = x̃ cos(ωt) − ỹ sin(ωt),

y = r sin(ωt + φ) = r sin φ cos(ωt) + r cos φ sin(ωt) = ỹ cos(ωt) + x̃ sin(ωt).

The differentials are dt = dt˜ and

        
dx = cos ωt˜ dx̃ − sin ωt˜ d ỹ + −ω x̃ sin ωt˜ − ω ỹ cos ωt˜ dt˜,
        
dy = sin ωt˜ dx̃ − cos ωt˜ d ỹ + −ω ỹ sin ωt˜ − ω x̃ cos ωt˜ dt˜.

Squaring and inserting the result into the line-element above gives the form of the
line-element in the rotating frame
  
ds 2 = − 1 − x̃ 2 + ỹ 2 ω2 dt˜2 + dx̃ 2 + d ỹ 2 − 2 ỹωdt˜dx̃ + 2ω x̃dt˜d ỹ.
Solutions to the Exercises 431

Fig. 4.12 In a rotating frame

the path of a light signal
emitted from the axis is an
Archimedian spiral

(c) A light signal is emitted from the axis of RF. As observed from IF the light moves
along a straight radial curve in 3-space, say along the x-axis, x = t, y = 0.
Inverting the coordinate transformation above we get

x̃ = x cos(ωt) + y sin(ωt), ỹ = −x sin(ωt) + y cos(ωt).

Inserting x = t, y = 0 gives

x̃ = t cos(ωt), ỹ = t sin(ωt).

This is the equation of an Archimedian spiral in parametric for, as shown in

Fig. 4.12. 
Using that r = x̃ 2 + ỹ 2 and the physical interpretation of the line element for
a time-like interval, ds 2 = −dτ 2 , we see from the line element that the proper time
interval for an observer at rest in the rotating frame, i.e. with constant values of x̃
and ỹ is related to a corresponding coordinate time interval by

dτ = 1 − r 2 ω2 dt.

The frequency this observer measures may be written ν = dN /dτ , where dN

is the number of wave fronts passing the observer during the time interval dτ . The
frequency of the emitted light is ν0 = dN /dt. Hence, the observer measures the
frequency
432 Solutions to the Exercises

dN dN dt ν0
ν= = =√ .
dτ dt dτ 1 − r 2 ω2

This frequency is larger than the frequency at the axis. In the rotating frame the
observer experiences a gravitational field away from the axis. Hence he concludes
that light moving downwards in a gravitational field gets an increased frequency.

(d) We shall describe the synchronization process from the point of view of an
observer at rest in the laboratory frame. A Lorentz transformation from the
rotating frame to the laboratory frame gives a time difference for two neigh-
bouring clocks with the same distance from the axis and angular coordinates φ
and φ + φ that are synchronized in the rotating frame,
 
t = γ τ + Rωlφ = γ 2 R 2 ωφ

where we have used that τ = 0 for the synchronized clocks and that their distance in
the rotating frame is lφ = γ Rφ. Integrated around the disk the angular difference
is 2π . It follows that the clocks at (R, 0) and (R, 2π ) will have a time difference
2π γ 2 R 2 ω. But these clocks are at the same position on the disk. So it is not possible
to Einstein-synchronize clocks globally on a rotating disk.

(e) In the cylindrical coordinate system co-moving with the rotating reference frame
the spacetime line-element has the form
 
ds 2 = − 1 − r 2 ω2 dt 2 + dr 2 + r 2 dφ 2 + 2r 2 ωdφdt

The non-vanishing dφdt−term means that the coordinate basis vectors are not
orthogonal. We shall find the vectors in an orthonormal basis field co-moving with
RF.
Let us start by the time-like unit basis vector. It shall have the same direction as
e. Hence, according to Eq. (4.57),

et et et et

etˆ = =√ =√ =√ ,
|et | et · et |gtt | 1 − r 2 ω2

1 − r 2 ω2 /c2 r ω/c2
er̂ = er , eφ̂ = eφ +  et .
r 1 − r 2 ω2 /c2
Solutions to the Exercises 433

Fig. 4.13 The lines

represent simultaneity planes
in the accelerated frame AF.
They are not simultaneous in
the inertial frame IF. The
curves are world lines of
reference points in AF
plotted with reference to IF.
Three sets of basis vectors in
the orthonormal basis field
co-moving with AF are also
shown

4.2 Uniformly accelerated system of reference

(a) It follows from the given coordinate transformation (4.91) that

T = tanh(at)X, X2 − T 2 = x2

For t = constant the first of these equations describe a set of straight lines in the
(t, X )-system that represent simultaneity planes for the reference particles of AF
at different points of time. For x = constant the second equation represents a set
of hyperbolae. They are the world lines of fixed reference points in the accelerated
reference frame as referred to the instantaneous Inertial rest frame IF of the particles
at the point of time T = 0. These sets of lines and curves are shown in the Minkowski
diagram4.12.

(b) In IF there is Minkowski metric

 
−1 1
ημ̂ν̂ = .
1 1

We can find the line element ds 2 by applying the transformation matrix

   
∂T ∂T
∂t ∂x ax cosh(at) sinh(at)
Mμμ̂ = ∂X ∂X
=
∂t ∂x
ax sinh(at) cosh(at)
434 Solutions to the Exercises

to the components of the metric tensor. This gives

μ̂
gtt = Mt Mtν̂ ημ̂ν̂ = −[ax cosh(at)]2 + [ax sinh(at)]2 = −a 2 x 2 ,
gt x = Mtμ̂ Mxν̂ ημ̂ν̂ = −ax cosh(at) sinh(at) + ax sinh(at) cosh(at) = 0,
μ̂
gx x = Mx Mxν̂ ημ̂ν̂ = [sinh(at)]2 + [cosh(at)]2 = 1.

Hence, the line-element takes the form

ds 2 = −a 2 x 2 dt 2 + dx 2 .

(c) We now consider a fixed particle in AF moving along one of the hyperbolae in
Fig. 4.13. Using the transformation 4.91 we find that Its velocity in IF is

dX dX/dt xa sinh(at)
v= = = = tanh(at).
dT dT /dt xa cosh(at)

and its acceleration

dv dt dv 1
= = .
dT dT dt x cosh3 (at)

Lorentz transforming the acceleration to the instantaneous rest frame of the

particle gives
 3/2
 −3/2 dv 1 1 1
g = 1 − v2 = = .
dT 1 − tanh2 (at) x cosh (at)
3 x

Since the particle has a constant value of x in AF it follows that the proper
acceleration of the particle is constant.

(d) The time axis of an orthonormal basis co-moving with an instantaneous inertial
rest frame of AF is equal to the four-velocity of the reference particles in AF,

dt 1
etˆ = et = et .
dτ ax
Since the metric is diagonal the coordinate basis in AF is orthogonal, and since
the values of the spatial components of the metric tensor are equal to 1, the spatial
basis vector in the orthonormal basis is ex̂ = ex . The basis vectors etˆ and ex measure
unit distance in the time- and spatial directions, respectively. Hence ex has unit
length in the spatial direction, but et = ax etˆ does not measure unit length in the
time direction. Along the world line of a reference particle in AF x is constant. The
Solutions to the Exercises 435

time-like coordinate basis vector has unit length along the world line of a reference
particle with x = 1/a.
We see from Figure 4.13 that AF does not cover the region |t| > |x|. From the
line element it is seen that there is a coordinate singularity at x = 0. The relationship
between the proper time in AF and the coordinate time is dτ = axdt, where the rate
of coordinate time is position independent. Hence at x = 0 time does not proceed in
AF.

(e) The accelerated reference frame AR is required to move in a Born rigid way.
This means that all the points on the x-axis move in such a way that the rest length
of all elements of the x-axis remains unchanged. As observed in IF the elements
then get an increasing Lorentz contraction due to the increasing velocity.

We now think of a part of the x-axis in AF as a rod with rest length L. All the points
of the rod move with constant proper acceleration. In AF the front end of the rod is at
x = x0 , and the rear end of the rod has a position x = x0 − L. The position in IF of
a point with coordinate x in √AF is given above by the by an equation of a hyperbola
which may be written X = T 2 + x 2 . Differentiation gives the coordinate velocity
as measured in IF
dX T
V = =√ .
dT T + x2
2

Hence, the velocity of the front end of the rod and its rear end is

T T
VF =  VR =  .
T 2 + x02 T 2 + (x0 − L)2

We see that the rear end of the rod moves faster than the front end. This is due to
the increasing Lorentz contraction. As observed in IF the rod becomes shorter due
to the increasing velocity.
All points of the rod have to move slower than the velocity of light. The longer
the rod is the faster its rear end moves. We have the limit lim = 1. Hence if the rest
L→x0
length of the rod approaches x0 the velocity of its rear end approaches that of light.
In this limit the rear end of the rod falls together with the origin of the x-axis which
also represents the limit in the backwards x-direction of the accelerated reference
frame.
The motion of this limiting point can be inferred from Fig. 4.13. For negative
time the reference frame moves in the negative x-direction with decreasing velocity.
Then the rod has a decreasing Lorenz contraction. It gets longer, and the left end of
the rod moves faster than the right one. As observed in IF the origin of the x-axis
approaches the origin in IF with the velocity of light. At the point of time T = 0 the
motion is reversed, and the limiting point moves in the positive X-direction with the
velocity of light.
436 Solutions to the Exercises

Fig. 4.14 World lines of

Earth, light signals from the
Earth and the spaceship

4.3 Uniformly accelerated spaceship

(a) A spaceship moves with constant proper acceleration g = 10m/s2 . The position
of the spaceship as a function of its proper time is given in Eq. (4.62),

c2  gτ
X = cosh + k,
g c

where k is a constant of integration. The spaceship starts at X = τ = 0 giving

k = −c2 /g. Hence,

c2 "  gτ #
X = cosh −1 .
g c

We shall find the distance which the spaceship has moved during a proper time
τ = 10years. There are 3.15 × 107 s in a year, and a light year is 9.46 × 1015 m.
Hence the acceleration of gravity at the surface of the Earth may be expressed as
g = 9.8 m/s2 = 1.03l.y./y2 . This gives c2 /g = 0.97l.y. and for τ = 10y. we get
gτ/c = 10.3. Inserting this in the formula above gives X = 1.44 × 104 l.y..

(b) In Fig. 4.14 we have shown the world lines of the spaceship and of light signals
emitted from the earth.

This figure shows that light signals emitted from points on the X-axis with negative
values of X will never arrive at the spaceship even if the spaceship always travel slower
than light. If the Earth is at a position X E and the spaceship departs at a point of time
T = 0, a light signal emitted from the Earth later than T = X E /c = c/g will never
arrive at the spaceship. Hence there is a horizon far down in the gravitational field
experienced in the spaceship, at a position x = X E = c2 /g lower than the initial
position of the spaceship.
Solutions to the Exercises 437

A light signal emitted from the Earth at a point of time TE is received at the
spaceship at a point of time τ E . We shall find the emitter time TE corresponding to a
receiver time τ E = 10 years. The position in IF at a point of time T of a light signal
emitted at a point of time TE from the Earth at X E = 0 is

X = c(T − TE ).

The position of the spaceship at this point of time is

⎛ ⎞
 
c2 ⎝ gT 2
X= 1+ − 1⎠.
g c

Putting the two expressions for X equal to each other and solving the resulting
equation with respect to TE gives
⎛  ⎞
 
c⎝ gT gT 2 ⎠
TE = 1+ − 1+ .
g c c

The transformation to the proper time τ of the spaceship is

c gτ
T = sinh .
g c

Using this in the expression for TE gives

c gτ E gτ E
TE = 1 + sinh − cosh .
g c c

Inserting the acceleration of gravity at the surface of the earth, g = 9.8m/s 2 =

1.03l.y/y2 gives c2 /g = 0.97l.y. and for τ = 10y we get gτ E /c = 10.3, which gives
TE = 0.97years.
The expression for the point of time of emission of the signal can be written
c gτ E

TE = 1 − e− c .
g

It follows that lim TE = c/g. Hence, signals emitted later than at TE = c/g are
τ E →∞
not able to reach the spaceship.

(c) Due to the Doppler effect the frequency of the signals is less than the frequency
observed at the emitter. The frequency ν measured in the spaceship is given by
438 Solutions to the Exercises
 
1 − v/c 1 − tanh(gτ E /c)
ν= ν0 = ν0 ,
1 + v/c 1 + tanh(gτ E /c)

where ν0 is emitted frequency measured on the Earth, and τ E is the point of time
measured on the spaceship when the signal is received. The formula in point b) for
the emission point of time may be written

1 − tanh(gτ E /c) gTE
=1− .
1 + tanh(gτ E /c) c

Hence the frequency of the received signal measured on the spaceship is

 
gTE
ν = 1− ν0 .
c

The velocity of the spaceship increases, and the observed frequency decreases due
to an increasing Doppler effect. Since signals emitted later than at TE = c/g will
not arrive at the spaceship, the expression for the frequency has physical meaning
only for TE ≤ c/g. The observed frequency at the spaceship decreases towards zero
at this limiting point of time.
4.4 Light emitted from a point source in a gravitational field
Consider a point like light source at the point x = x1 , y = 0 in a uniformly
accelerated reference frame AF. Let a photon be emitted from the source at a point
of time t = 0. It is emitted in the (x, y)−plane in a direction making an angle θ0
with the x-axis. In the inertial laboratory frame IF the coordinates of the emission
event is T = 0, X = x1 . The photon follows a null-geodesic curve which in this
frame is a straight line given by

X = x1 + cT cos θ0 , Y = cT sin θ0 .

Using the transformation

T = x sinh(gt/c), X = x cosh(gt/c) Y = y

between the coordinate systems co-moving with IF and AF gives the equations for
the world lines of the photon in AF

x1 x1 sinh(gt/c) sin θ0
x= , y= .
cosh(gt/c) − sinh(gt/c) cos θ0 cosh(gt/c) − sinh(gt/c) cos θ0

Dividing the expressions for x and y by each other, we get

y
sinh(gt/c) = .
x sin θ0
Solutions to the Exercises 439

Fig. 4.15 Light rays and wave fronts emitted by a point particle in a uniformly accelerated reference
frame

Inserting this into the expressions for x and y gives an equation which may be
written as
 2
x1
x 2 + (y − x1 cot θ0 )2 = .
sin θ0

This is the equation of a circle with radius r = x1 / sin θ0 . A set of these circular
trajectories and the corresponding wave front is shown in Fig. 4.15.

4.5 Geometrical optics in a gravitational field

(a) Asphere photographed from above. The situation described in the text is
illustrated in Fig. 4.16.
440 Solutions to the Exercises

Fig. 4.16 There is a camera at x = x1 on the vertical x-axis in a uniformly accelerated reference
frame AF. The centre of a sphere with radius r is on the x-axis a distance b below the camera. Due
to the gravitational field experienced in AF a light signal does not move along a straight line, but
along a circle with radius R. Light emitted tangentially from the sphere to the camera makes an
angle θ with the x-axis at the camera.

The radius of the circular light path is c2 /g = 0.97 light years in the gravitational
field of the Earth, so the optical effects of the gravitational field are very small at the
surface of the Earth.
The camera P is at a position x = x1 . The acceleration of gravity at the position
of the camera is g = c2 /x1 . The centre of a sphere with radius r is a distance b below
P. A ray of light is emitted tangentially from a point on the surface of the sphere
such that it arrives at P. The angle θ is half the apex angle.
We see from the figure that sin θa = x1 /R and R 2 − x12 = (R − r )2 − (x1 − b)2
which gives

x1 x 2 + r 2 − (x1 − b)2 r 2 + 2x1 b − b2

R= = 1 = .
sin θa 2r 2r

Hence
2r x1
sin θa = .
r2 + 2x1 b − b2

The half of the apex angle without the gravitational field is sin θ0 = r/b. Inserting
this into the above expression and using that x1 = c2 /g gives

sin θ0 r/b
sin θa =   =   .
1 − gb/2c cos θ0
2 2 1 − gb/2c2 1 − r 2 /b2

This shows that as photographed from above the sphere appears enlarged.

(b) A sphere photographed from below. This situation is illustrated in Fig. 4.17.

In the present case the figure shows that R 2 − x12 = (R + r )2 − (x1 + b)2 . Then
a similar deduction to that in point a) gives
Solutions to the Exercises 441

Fig. 4.17 Similar to the

situation in a, but with the
camera below the sphere

sin θ0 r/b
sin θb =   =   .
1 + gb/2c2 cos2 θ0 1 + gb/2c2 1 − r 2 /b2

This shows that as photographed from below the sphere appears diminished.
(c) The difference in θ as photographed from above and below at the same distance
is given by

 
gb/c2 cos2 θ0 sin θ0
sin θa − sin θb =  2 .
1 − gb/c2 cos4 θ0

With b = 1m we get gb/c2 = 1m/0.97l.y. = 1.06 × 10−16 . Hence the dif-

ference, θ , between the angels θa and θb is very small. We can therefore use the
approximations

sin θa − sin θb ≈ sin(θ + θ ) − sin θ = sin θ cos θ + cos θ sin θ − sin θ

≈ θ cos θ,

and

gb gr r2
θ ≈ 2 sin θ0 cos θ0 = 2 1− .
c c b2
With r = 0.1m and b = 1m we can further use the approximation
gr
θ ≈ .
c2

Letting g = 9.8m/s2 and r = 0.1m we get θ ≈ 1, 0 × 10−17 . This corresponds

to θ ≈ 2.2 × 10−12 arc seconds, which is beyond what we are able to observe.
442 Solutions to the Exercises

Solutions for Chapter 5

5.1 Dual forms

Let{eî } be a Cartesian basis in the 3-dimensional Euclidean space. Using a vector
A = Aî eî there are two ways of constructing a form:
(i) By constructing a 1-form from its covariant components A j = g ji Ai :

A = Aî dx î .

(ii) By constructing a 2-form from its dual components, defined by

1 1
a= a dx î ∧ dx ĵ = εî ĵ k̂ Ak̂ dx î ∧ dx ĵ
2 î ĵ 2
We write this form as a = A, where  means to take the dual form.

(a) Given the vectors a = ex + 2e y − ez and b = 2ex − 3e y + ez .

We shall find the corresponding 1-forms A and B, and the dual 2-forms a = A
and b = B, and also the dual form θ to the 1-form σ = dx − 2dy .
Since the space has Euclidean geometry with an orthonormal basis gî ĵ = δî ĵ , the
component of the 1-forms A and B are equal to the components of the corresponding
vectors Aî = δî ĵ A ĵ and Bî = δî ĵ B ĵ .
Hence,

A = dx + 2dy − dz B = 2dx − 3dy + dz.

The dual 1-forms α = ∗A and β = ∗B have components αî ĵ = εî ĵ k̂ Ak̂ and
βî ĵ = εî ĵ k̂ B k̂ . This gives

∗A = −dx ∧ dy − 2dx ∧ dz + dy ∧ dz
.
∗B = dx ∧ dy + 3dx ∧ dz + 2dy ∧ dz

Given a 1-form σ = σî dx î = dx − 2dy. The 2-form θ = ∗σ has components

ˆ
αî ĵ = εî ĵ k̂ σ k̂ = αî ĵ = εî ĵ k̂ g k̂l σlˆ which gives

θ = 2dx ∧ dz + dy ∧ dz.

(b) Inserting the expressions for A and B found in point a) in θ = A ∧ B we find

 
θ− = dx + 2dy − dz ∧ 2dx − 3dy + dz = − 7dx ∧ dy + 3dx ∧ dz
Solutions to the Exercises 443

− dy ∧ dz.

Calculating the vector product of A and B in the usual way we find

C = A × B = −ex − 3e y − 7ez .

We see that C and θ have the same components, where C z corresponds to θx y and
so forth.
Let us show this in general. The components of the vector product are C k̂ =
εm̂ n̂ k̂ Am̂ B n̂ . Hence,

εî ĵ k̂ C k̂ = εî ĵ k̂ εm̂ n̂ k̂ Am̂ B n̂ = δî m̂ δ ĵ n̂ − δî n̂ δ ĵ m̂ Am̂ B n̂ = Aî B ĵ − A ĵ B î .

Since the exterior product of two 1-forms gives

 
θî ĵ = A ∧ B î ĵ = Aî B ĵ − A ĵ Bî ,

we see that with a metric gî ĵ = δî ĵ there is agreement between the two expressions.
Note that if we replace δî m̂ δ ĵ n̂ − δî n̂ δ ĵ m̂ by gim g jn − gin g jm the deduction will
bevalid in an
arbitrary basis.

(c) Direct calculation gives

 
A ∧ ∗B = Alˆωî ∧ 21 εî ĵ k̂ B k̂ ωî ∧ ω ĵ
   
= Ax ωx + A y ω y + Az ωz ∧ B z ωx ∧ ω y −  B y ω x ∧ ω z + B x ω y ∧ ω z .
 
= A x B x + A y B y + A z B Z ω x ∧ ω y ∧ ω z = A · B ω x ∧ ω y ∧ ω z

(d) The exterior derivative of A is

 
dA = Ai, j ω j ∧ ωi = A y,x − A x,y ω x ∧ ω y
   
+ A z,x − A x,z ω x ∧ ω z A z,y − A y,z ω y ∧ ω z .

The curl of a vector is

     
∇ × A = A z , y −A y ,z ex − A z ,x −A x ,z ey + A y ,x −A x , y ez .

Hence
   k
dA i j = εi jk ∇ × A

(e) The exterior derivative of the dual ∗A is

444 Solutions to the Exercises

1  1 
d∗ A = ∗A i j ,k ωk ∧ ωi ∧ ω j = εi jl Al ,k ωk ∧ ωi ∧ ω j
2 2
= Ak ,k ω x ∧ ω y ∧ ω z .

The exterior derivative of a scalar field f is

d f = f,i ωi .

Which has the same components as ∇ f . Inserting A = d f in the expression for

d ∗ A we get

d ∗ d f = ∇ 2 f ωx ∧ ω y ∧ ωz

5.2 Differential operators in spherical coordinates

(a) Wefind the metric tensor by transforming from the metric in the Cartesian
coordinate system,

∂ x î ∂ x ĵ
gi j = δ .
∂ x i ∂ x j î ĵ
The coordinate transformation is

x = r sin θ cos φ, y = r sin θ sin φ, z = r cos θ

Hence, the transformation matrix has the elements

⎛ ∂x ∂x ∂x
⎞ ⎛ ⎞
∂r ∂θ ∂φ sin θ cos φ r cos θ cos φ −r sin θ sin φ
⎜ ⎟ ⎝
⎠ = sin θ sin φ r cos θ sin φ r sin θ cos φ ⎠.
∂y ∂y ∂y
⎝ ∂r ∂θ ∂φ
∂z ∂z ∂z cos θ −r sin θ 0
∂r ∂θ ∂φ

Inserting this into the transformation formula for the components of the metric
tensor, we find the following non-vanishing components

grr = 1, gθθ = r 2 , gφφ = r 2 sin2 θ.

This shows that the line-element has the form

dl 2 = dr 2 + r 2 dθ 2 + r 2 sin2 θ dφ 2

In the spherical coordinate system.

(b) The gradient of the scalar field f in the spherical coordinate system is gives by
Solutions to the Exercises 445

∂f ∂f ∂f
∇f = er̂ + eθ̂ + e
∂r ∂θ ∂φ φ̂

The orthonormal basis vectors are given in terms of the coordinate basis vectors
as,
 √
eî = ei /|ei | = ei / ei · ei = ei / gii .

This gives

1 1
er̂ = er , eθ̂ = eθ , eφ̂ = eφ .
r r sin θ
Inserting these in the expression for the gradient gives

∂f 1 ∂f 1 ∂f
∇f = er + eθ + eφ .
∂r r ∂θ r sin θ ∂φ

(c) The curl of a vector is given as

   
1 ∂ A 3̂
∂ A 2
1 ∂ A 1̂
∂ A 3̂
∇ × A = √ − e1 + √ − e2
g22 g33 ∂ x 2 ∂x3 g11 g33 ∂ x 3 ∂x1
 
1 ∂ A2̂ ∂ A1̂
+√ − e3 .
g11 g22 ∂ x 1 ∂x2

The components of the vector field in orthonormal basis are given in terms of its
coordinate component by

Ar̂ = Â, Aθ̂ = r Aθ , Aφ̂ = r sin θ Aφ .

Using this together with the components of the metric tensor from point a) in the
above expression for the curl gives in spherical coordinates
      
1 ∂ sin θ Aφ ∂ A θ
1 ∂ A r ∂ r Aφ
∇ × A = − er + − sin θ eθ
r sin θ ∂θ ∂φ r sin θ ∂φ ∂r
   
1 ∂ r Aθ ∂ Ar
+ − eφ
r ∂r ∂θ

(d) The divergence of a vector field is given by

446 Solutions to the Exercises

∇ · A ε = d ∗ A,

where

ε= |g|ω1 ∧ ω2 ∧ ω3 ,

is the volume form and |g| the determinant of the matrix formed by the components
of the metric tensor.
From Exercise 5.1 we have that the dual 2-form of a vector is given by
  
∗A i j = |g|εi jk Ak ,

where εi jk is the Levi-Civita symbol with εr θφ = +1. Hence, we get

 
d ∗ A r θφ
∇ · A = √ .
|g|

In spherical coordinates we have


|g| = r 2 sin θ,

giving
     
∗A r θ = r 2 sin θ Aφ , ∗A r φ = −r 2 sin θ Aθ , ∗A θφ = r 2 sin θ Ar .

The exterior derivative has only one component

       
d ∗ A r θ φ = ∗A θ φ ,r − ∗A r φ ,θ + ∗A r θ ,φ
    .
= sin θ r 2 Ar ,r +r 2 sin θ Aθ ,θ +r 2 sin θ Aφ ,φ

where we have used Einstein’s comma-notation for partial derivatives. This gives
 
d ∗ A r θφ 1  2 r 1  
∇ · A = = r A ,r + sin θ Aθ ,θ +Aφ ,φ .
r2 sin θ r 2 sin θ
We can also express the divergence of a vector in term of it components in an
orthonormal basis. Using that

1 θ̂ 1
Ar = Ar̂ , Aθ = A , Aφ = Aφ̂ ,
r r sin θ
we get
Solutions to the Exercises 447

Fig 5.9 Geodesic curves on

a non-rotating (dashed line)
and rotating (solid line) disc

1 1  1
∇ · A = 2 r 2 Ar̂ ,r + sin θ Aθ̂ ,θ + Aφ ,φ .
r r sin θ r sin θ

Inserting A = ∇ f , where ∇ f is given in point b), we get the expression for the
Laplacian in spherical coordinates

1 2  1 1
∇2 f = r f,r ,r + 2 (sin θ f,θ ),θ + 2 2 f,φφ .
r2 r sin θ r sin θ
5.3 Spatial geodesics in a rotating frame of reference
In Fig. 5.9, we see a rotating disc. We can see two geodesic curves between P1 and P2 .
The dashed line is the geodesic for the non-rotating disc. The other curve is a geodesic
for the 3-space of a rotating reference frame. We can see that the geodesic is curved
inward when the disc is rotating. The curve has to curve inward since the standard
measuring rods along the curve are longer there (because of less Lorentz contraction
close to the axis where the velocity in the length direction of the measuring rods is
less). Thus, the minimum distance, i.e. the minimum number of standard measuring
rods along the curve between P1 and P2 , will be achieved by an inwardly bent curve.
We will show this mathematically, using the Lagrangian equations.
(a) We first deduce the form of the tangent vector identity for the present case. The
line element for the space dtˆ = dz = 0 of the rotating reference frame is

r 2 dθ 2
dl 2 = dr 2 + r 2 ω2
.
1− c2
448 Solutions to the Exercises

The Lagrangian function for the spatial geodesics in the rotating frame is

1 2 1 r 2 θ̇ 2
L= ṙ + .
2 2 1 − r 2cω2 2

where the dot means differentiation with respect to an invariant parameter represent-
ing the arc length along the curve. In the present case the 3-vector identity for the
tangent vectors of the spatial geodesic curves take the form

r 2 θ̇ 2
ṙ 2 + r 2 ω2
= c2 .
1− c2

(b) It is seen from the Lagrangian function that θ is cyclic (∂ L/∂θ = 0), implying
that

∂L r 2 θ̇
pθ = = 2 2 = constant.
∂ θ̇ 1 − r cω2

(c) This gives

 
r 2 ω 2 pθ pθ ω 2 pθ
θ̇ = 1 − 2 = − .
c r2 r2 c2

Inserting into tangent vector identity we get

ω2 pθ2 pθ2
ṙ 2 = 1 + − .
c2 r2
This gives the equation of the geodesic curve between P1 and P2 .

ω2 p 2 p2
ṙ dr r 2 1 + c2 θ − r 2θ
=± =  .
θ̇ dθ pθ 1 − r cω2
2 2

(d) The boundary conditions are ṙ = 0, r = r0 , for θ = 0, where r0 is the distance

from the rotational axis to the point of the curve with minimal distance to the
axis (Fig. 5.10). Inserting this into the expression for pθ gives


pθ pθ2 ω2
= 1+ .
r0 c2

(e) Solving this equation with respect to pθ gives

Solutions to the Exercises 449

Fig. 5.10 Geodesic curves

on a rotating disc,
coordinates

r0
pθ =  .
1 − r02 ω2 /c2

Inserting this into the equation of the geodesic curve, it takes the form

dr r r 2 − r02
=  ,
dθ r0 1 − r 2 ω2 /c2

which may be written

dr ω2 r dr dθ
 − 2 = .
r r 2 − r02 c r 2 − r02 r0

Integrating this yields


r0 ω2 2 r0
θ =± 2
r − r02 ∓ arccos .
c r
The curve represented by this equation is shown in Fig. 5.10.

5.4 Christoffel symbols in a uniformly accelerated reference frame

(a) Theonly non-vanishing Christoffel symbols in the coordinate system co-moving

with a uniformly accelerated reference frame are
450 Solutions to the Exercises

∂ x ∂2T ∂ x ∂2 X ∂t ∂ 2 T ∂t ∂ 2 X
ttx = + , xt
t
= tt x = + .
∂ T ∂t 2 ∂ X ∂t 2 ∂ T ∂t∂ x ∂ X ∂t∂ x
The coordinate transformation formulae are
gT    
= 1 + gx sinh gtc , 1 + gcX2 = 1 + gx cosh gt ,
c c2  2  c2 2  c
gT gx 2
c
= (1 + gcX2 ) tanh gtc , 1 + gcX2 − gT
c
= 1+ c2
.

Differentiation gives

∂x gT gT ∂x 1 + g X/c2
= −   =− , = ,
∂T 2 1 + gx/c2 ∂X 1 + gx/c2
1 + g X/c2 − (gT /c)2
∂t 1 + g X/c2 cosh(gt/c) ∂ct gT /c sinh(gt/c)
= 2 = , = 2 = ,
∂T 1 + gx/c2 1 + gx/c 2 ∂X 1 + gx/c2 1 + gx/c2
∂2T g gx gt ∂2 X  gx gt
= 1 + sinh , = g 1 + cosh ,
∂t 2 c c 2 c ∂t 2 c 2 c
∂2T g gt ∂2 X g gt
= 2 cosh , = sinh
∂t∂ x c c ∂t∂ x c c
Inserting these expressions into the formulae of the Christoffel symbols leads to

  g/c2
ttx = g 1 + gx/c2 , xt
t
= tt x = .
1 + gx/c2

(b) We shall use the formula for the Christoffel symbols in terms of the derivatives
of the components of the metric tensor,

α 1 αβ
μν ≡ g (gβμ,ν + gβν,μ − gμν,β )
2
To calculate the same Chritoffel symbols as above.
The components of the metric tensor in the coordinate system co-moving with a
uniformly accelerated reference frame are
 gx 2
gtt = − 1 + 2 c2 , gx x = g yy = gzz = 1
c

only the term ∂g

∂x
tt α
contributes to μν . Thus the only non-vanishing Christoffel symbols
are
Solutions to the Exercises 451
   
1 tt ∂gtt 1 ∂gtt 2 1 + gx g 1 g
xt
t
= tt x
= g = =  c2
2 =  gx  2 ,
2 ∂x 2gtt ∂ x gx
2 1 + c2 c 2 1 + c2 c
 
1 ∂gtt 1$  gx g %  gx
ttx = − g x x = − −2 1 + 2 2 c2 = 1 + g.
2 ∂x 2 c c c2

5.5 Relativistic vertical projectile motion

A particle is thrown vertically upwards with velocity v from the origin of the coor-
dinate system in the gravitational field of a uniformly accelerated reference frame.
We shall calculate the maximal height of the particle.
The line element, including only the vertical x-direction, has the form
 gx 2
ds 2 = − 1 + 2 c2 dt 2 + dx 2
c
the Lagrange function is

1 1 gx 2 1
L= gμν ẋ μ ẋ ν = − 1 + 2 c2 t˙2 + ẋ 2 ,
2 2 c 2
where the dots imply differentiation with respect to the particle’s proper time
τ . The initial conditions are x(0) = 0, ẋ(0) = γ (c, v, 0, 0), where γ =
 −1/2
1 − v 2 /c2 .
v2
Newtonian description: 21 mv 2 = mgh ⇒ h = 2g .
Relativistic description: t is a cyclic coordinate ⇒ x 0 = ct is cyclic and p0 =
constant.
∂L 1 ∂L  gx 2
p0 = = = −c 1 + t˙.
∂ ẋ 0 c ∂ t˙ c2
In the present case the 4-velocity identity has the form

1 gx 2 1 1
− 1 + 2 c2 t˙2 + ẋ 2 = − c2 .
2 c 2 2
Since the maximum height h is reached when ẋ = 0we have
 
gh
1 + 2 t˙x=h = 1.
c

Since p0 is a constant of the motion we can determine ist value from the initial
condition, p0 = −ct˙(0) = −γ c, and at x = h:
 2
gh
p0 = −γ c = −c 1 + 2 t˙x=h
c
452 Solutions to the Exercises

It follows from the two last equations that the maximal height of the projectile is

c2
h= (γ − 1).
g

In the Newtonian limit this becomes

  2 
c2 1 c 1 v2 v2
h= − 1  1 + − 1 ⇒ h .
g (1 − v 2 /c2 )1/2 g 2 c2 2g

5.6 The geodesic equation and constants of motion

(a) Thegeodesic equation can be written

du μ α
− μβ u α u β = 0.
ds
Inserting the expression for the Christoffel symbols give

α 1 
μβ u α u β = αμβ u α u β = gαμ,β + gαβ,μ − gμβ,α u α u β .
2
Since the metric tensor is symmetrical this can be written

α 1 
μβ uα uβ = gαμ,β + gαβ,μ − gβμ,α u α u β .
2

Since gαμ,β −
 gβμ,α is antisymmetric
 α β in α and β, and u α u β is symmetric in α and
β we have that gαμ,β − gβμ,α u u = 0. Hence

α 1
μβ uα uβ = gαβ,μ u α u β .
2
Hence the geodesic equation in the form above reduces to

du μ 1
− gαβ,μ u α u β = 0.
ds 2
(b) From the form of the geodesic equation in b) it follows that since the metric is
independent

of t, θ and z in the static and cylindrically symmetric case, so that gμν,t = gμν,θ =
gμν,z = 0, then u t ,u θ and u z are constants of motion in this case.
Solutions to the Exercises 453

Solutions for Chapter 6

6.1 Parallel transport

Parallel transport of a vector A along a curve means that the covariant directional
derivative of the vector along the curve vanishes,

∇u A = 0. (S6.1.1)

where u is the tangent vector field of the curve. Let the curve parameter be λ. Then
the component equation of parallel transport along the curve takes the form

μ μ dx ν
A;ν u ν = A;ν = 0. (S6.1.2)
dλ
μ μ
Inserting the expression A;ν = Aμ,ν + λν Aλ of the covariant derivative, we get

dx ν dAμ μ dx ν
Aμ,ν = = −λν Aλ . (S6.1.3)
dλ dλ dλ
For an infinitesimal displacement along the curve the change of the vector
component Aμ thereby becomes
μ
dAμ = −λν Aλ dx ν . (S6.1.4)

(b) The result from point a) may be written as a Taylor expansion of Aμ to first
order at the point x,

μ
Aμ (x + dx) = Aμ (x) − λν Aλ dx ν . (S6.1.5)

The Taylor expansion is generally given as

∂ Aμ (x) ν 1 ∂ 2 Aμ (x) ν γ
Aμ (x + dx) = Aμ (x) + dx + dx dx + · · · (S6.1.6)
∂xν 2 ∂xν∂xγ
We see that
∂ Aμ μ λ
= −λν A . (S6.1.7)
∂xν
We also need the next term in the Taylor expansion,
 μ 
∂ 2 Aμ ∂ ∂ Aμ ∂ −λν Aλ μ μ λ
= γ = = −λν,γ Aλ − λν A,γ
∂xν∂xγ ∂x ∂xν ∂xγ
454 Solutions to the Exercises

Fig. 6.8 Infinitesimalt

parallelogram

μ μ
= −λν,γ Aλ + λν αγ
λ
Aα . (S6.1.8)

Hence, we get to 2.order that by parallel from x to x + dx the component Aμ of

the vector is
μ λ μ μ λ
Aμ (x + d x) = Aμ (x) − λν A dx ν − λν,γ Aλ dx ν dx γ + λν αγ Aα dx ν dx γ .
(S6.1.9)

Let A AB be the vector A parallel transported from point A to B in the parallelogram

shown in Fig. 6.8.
Using Eq. (S6.1.9) we get
μ μ μ μ
A AB = Aμ − λν (A)Aλ da ν − λν,γ (A)Aλ da ν da γ + λν (A)αγ
λ
(A)Aα da ν da γ ,
(S6.1.10)

where (A) means that the quantity is evaluated at the point A. Further parallel
transport to C gives
μ μ μ μ μ
A ABC = A AB − λν (B)Aλ dbν − λν,γ (B)Aλ dbν dbγ + λν (B)αγ
λ
(B)Aα dbν dbγ .
(S6.1.11)

Note that since the last parallel transport starts at the point B, we must evaluate the
connection coefficients at point B. These connection coefficients deviate infinitesi-
mally from their values in A. We want to express all the functions by their values in
A. This is performed by means of a Taylor expansion. Since the connection coeffi-
cients in (S6.2.11) are all multiplied by differentials, it is sufficient to make a Taylor
expansion to first order in the differentials. This gives
μ μ μ μ
λν (B) = λν (A + d a ) = λν (A) + λν,γ (A)da γ + · · · (S6.1.12)

Inserting Eqs. (S6.1.10) and (S6.1.12) into Eq. (S6.1.11) we get

μ μ μ μ
A ABC = A AB − Γ λν Aλ da ν − Γλν Aλ dbν
 
μ 1 1
− Γλν,γ Aλ dbν da γ + da ν da γ + dbν dbγ
2 2
 
μ 1 1
+ Γ λν Γ λαγ Aα dbν da γ + da ν da γ + dbν dbγ (S6.1.13)
2 2
Solutions to the Exercises 455

Parallel transport to C via D gives a similar expression, the only difference being
that da and db are exchanged,
 
AμABC = AμAB − λν
μ λ
A dbν − λν μ λ μ
A da ν − λν,γ Aλ da ν dbγ + 21 dbν dbγ + 21 da ν da γ
μ λ   .
+λν αγ Aα da ν dbγ + 21 dbν dbγ + 21 da ν da γ
(S6.1.14)

The difference is

μ μ μ μ μ α μ α
A ABC − A ADC = λν,γ − λγ ,ν + αγ λν − αν λγ Aλ da ν dbγ , (S6.1.15)

where we have changed some of the summation indices in order to have the same
indices on A, da and db in all the terms.
μ μ
The difference A ABC − A ADC corresponds to adding parallel transport of A from
a to C via B and then adding the negative of the transport from A via D to C. The
latter is the same as adding parallel transport from C via D to A. Hence the difference
corresponds to parallel transporting A around the parallelogram. So the right-hand
side is the change of the vector A by this round trip. Comparing with Eq. (6.17) we
see that the quantities inside the parenthesis in equation (S6.1.15) are the components
of the Riemann curvature tensor in coordinate basis. Hence,
μ
δ Aμ = −Rλνγ Aλ da ν dbγ . (S6.1.16)

The curvature of a space makes a vector change when it is parallel transported

around a closed curve in the space.

6.2 Curvature of the spatial simultaneity space of a rotating reference frame

Considerthe 2. dimensional simultaneity space, tˆ = constant, z = constant, with

line element (4.20),

r 2 dθ 2
dl 2 = dr 2 + . (S6.2.1)
1 − r 2 ω2 /c2

We shall calculate the Riemann curvature scalar of the 2-space described by this
line-element. The 1-foms of an orthonormal basis are
r
ωr̂ = dr, ωφ̂ =  dφ. (S6.2.2)
1 − r 2 ω2 /c2

Taking the exterior derivative gives

dωr̂ = 0 
.
dωφ̂ = d √ r
∧ dφ = 1
dr ∧ dφ = 1
ωr̂ ∧ ωφ̂
1−r 2 ω2 /c2 ( 1−r 2 ω2 /c2)3/2
r( )
1−r 2 ω2 /c2
3/2

(S6.2.3)
456 Solutions to the Exercises

From Cartan’s 1. structure equation

μ̂
dωμ̂ = −ν̂ ∧ ων̂ . (S6.2.4)

We then have

φ̂ 1
r̂φ̂ = f (r )ωφ̂ r̂ =  φ̂
3/2 ω + g(r )ω .
r̂
(S6.2.5)
r 1 − r ω /c
2 2 2

The antisymmetry νμ = −μν now gives

1
f (r ) =  3/2 , g(r ) = 0. (S6.2.6)
r 1 − r 2 ω2 /c2

Hence the only non-vanishing connection forms are

φ̂ 1
r̂ = r̂φ̂ =   ωφ̂ . (S6.2.7)
r 1 − r 2 ω2 /c2

We then calculate the curvature forms from Cartan’s 2. structure equation

R μ̂ν̂ = dμ̂ν̂ + μ̂λ̂ ∧ λ̂ν̂ . (S6.2.8)

This gives only one non-vanishing curvature form

φ̂ 3ω2 /c2 φ̂
R r̂ = −  2 ω ∧ ω .
r̂
(S6.2.9)
1 − r 2 ω2 /c2

From

R μ̂ν̂ = Rν̂μ̂α̂β̂ ωα̂ ∧ ωβ̂ (S6.2.10)

we then get the non-vanishing components of the Riemann curvature tensor

φ̂ 3ω2 /c2
Rφ̂r̂ r̂ φ̂ = Rr̂ φ̂r̂ = −  2 . (S6.2.11)
1 − r 2 ω2 /c2

Contraction gives the components of the Ricci curvature tensor

φ̂ 3ω2 /c2 3ω2 /c2

Rr̂ r̂ = Rr̂ φ̂r̂ = −  2 , Rφ̂ φ̂ = Rφ̂r̂ φ̂ = − 
r̂
2 . (S6.2.12)
1 − r 2 ω2 /c2 1 − r 2 ω2 /c2

Finally the curvature scalar is the trace of the Ricci tensor

Solutions to the Exercises 457

μ̂ φ̂ 6ω2 /c2
R = Rμ̂ = Rr̂r̂ + Rφ̂ = −  2 . (S6.2.13)
1 − r 2 ω2 /c2

According to Eq. (6.55) this is equal to the Gaussian curvature of the surface, which
is again equal to the product of the curvatures of two geodesic curves orthogonal to
each other on the surface. When the Gaussian curvature is negative, these so-called
principal curvatures have opposite signs, meaning that the geodesics curve in opposite
ways, like they do on a saddle surface. Hence, the surface has negative curvature.

6.3 The tidal force pendulum and the curvature of space

(a) The particles of the pendulum are not moving geodesically. However, the forces
on the particles from the rod connecting them, have no component in the direc-
tion normal to the rods, and are thus of no consequence for the period of the
pendulum. We may therefore calculate the period of the tidal force pendulum
using the equation of geodesic deviation,

D 2 μ μ
+ Rανβ u α ν u β = 0. (S6.3.1)
dτ 2
where τ is the proper time of the pendulum.

(b) We introduce co-moving geodesic normal coordinates with u = (1, 0, 0, 0),

vanishing Christoffel symbols and origin at the mass centre of the pendulum.
The coordinate basis of this coordinate system is orthonormal. The pendulum
is permanently at rest at the surface of the Earth. Then it moves with non-
relativistic velocity in a weak gravitational field. Hence, the proper time of the
pendulum can be approximated by the coordinate time t of the Earth, and Eq.
(S6.3.1) reduces to

d2 î
+ R0̂î ĵ 0̂  ĵ = 0. (S6.3.2)
dt 2
(c) From Fig. 6.6 is seen that the θ -component of the equation is

d2 θ̂ φ̂
+ Rtˆφ̂ tˆφ̂ cos θ − Rtr̂ˆr̂ tˆr̂ sin θ = 0, (S6.3.3)
dt 2
where

θ̂ = θ, φ̂ =  sin θ, r̂ =  cos θ. (S6.3.4)

458 Solutions to the Exercises

Hence, for small angular amplitudes the equation of motion of the pendulum can
be approximated by

d2 θ 
+ R ˆ ˆ
φ̂ t φ̂ t − Rr̂ ˆ
t r̂ ˆ
t θ = 0. (S6.3.5)
dt 2
This equation describes harmonic oscillations with period

2π
T = , (S6.3.6)
Rφ̂ tˆφ̂ tˆ − Rr̂ tˆr̂ tˆ

showing that the period of the tidal force pendulum depends only upon the curvature
of spacetime at its position.
At a distance R from the centre of a spherical body of mass M [6.1],

Rφ̂ tˆφ̂ tˆ = G M/R 2 , Rr̂ tˆr̂ tˆ = −2G M/R 2 , (S6.3.7)

giving

R3
T = 2π , (S6.3.8)
3G M
in agreement with the Newtonian calculation in Exercise 1.1 of the period of the tidal
force pendulum.
[1] Misner, C.W., Thorne, K.S., Wheeler, J.A.: Gravitation, p. 821. Freeman, San
Franscisco (1973)

Solutions for Chapter 7

7.1 Newtonian approximation of perfect fluid

(a) Thecovariant formulation of energy–momentum is that the covariant divergence

of the energy–momentum tensor vanishes,
μν
T;ν = 0.

Writing out the covariant derivative we have

T,νμν + αν
ν
T μα + αν
μ
T αν = 0

In the Newtonian approximation we neglect the curvature of spacetime and intro-

duce a Cartesian coordinate system with Minkowski metric and where all the com-
ponents of the metric tensor are constant. Hence all the Christoffel symbols vanish,
and the conservation laws reduce to
Solutions to the Exercises 459

T,νμν = 0.

(b) In the Newtonian limit the components of the energy–momentum tensor are

T 00 = ρc2 , T 0i = ρcv i , T i j = pδ i j + ρv i v j .

Hence the time component of the conservation laws is

 
∂ρ ∂ ρv i ∂ρ
T,ν0ν = T,000 + T,i0i = + = 0 or + ∇ · (ρ v) = 0.
∂t ∂x i ∂t
This is the equation of continuity of the fluid.

(c) The spatial components of the conservation laws are

   
∂ ρv i
ij ∂ pδ i j + ρv i v j
T,νiν = +
T,0i0 = T, j +
 i  ∂t  ∂x j
∂ ρv ∂ ρv v
i j
∂p
= + + =0
∂t ∂x j ∂ xi

This equation can be simplified by using the equation of continuity in the

form (∂ x j ) = − ∂ρ
∂ ρv j
∂t
. Hence we need to rewrite the second last term in the conservation
equation as follows
   j
∂ ρv i v j j ∂v
i
i ∂ ρv j ∂v
i
∂ρ
= ρv + v = ρv − vi .
∂x j ∂x j ∂x j ∂x j ∂t
Inserting this into the conservation equation, and using the product rule in the first
term gives

∂ρ ∂v i ∂v i ∂ρ ∂p
vi +ρ + ρv j j − v i + = 0,
∂t ∂t ∂x ∂t ∂ xi
or

∂v i ∂v i ∂p
ρ + ρv j j + = 0.
∂t ∂x ∂ xi

This is usually written as

   
∂v i ∂v i ∂p ∂ v
ρ + vj j =− or ρ + (
v · ∇)
v = −∇ p.
∂t ∂x ∂ xi ∂t

which is the Euler equation of motion applied to a fluid in a region without a

gravitational field, i.e. the Minkowski spacetime.
460 Solutions to the Exercises

7.2 The energy–momentum tensor of LIVE

(a) Weshall deduce the form of an energy–momentum tensor with Lorentz invariant
components as decomposed in orthonormal basis fields, i.e. with

β̂
Tμ̂ν̂ = Tμ̂ ν̂  = α̂μ̂ ν̂  Tα̂β̂

where μ̂μ̂ are the elements of a Lorentz transformation matrix. Let us first consider
a Lorentz transformation in the x 1̂ - direction,
⎡ ⎤
γ γv 0 0
⎢γv γ 0 0⎥
μ̂ = ⎢ ⎥.
μ̂
⎣0 0 1 0⎦
0 0 0 1
 −1/2
where γ = 1 − v 2 , and we have used units so that c = 1. Inserting this in the
above equation leads to
 
v T0̂0̂ + T1̂1̂ + T0̂1̂ + T1̂0̂ = 0.

In a similar way transformation of T0̂1̂ and T1̂0̂ leads to

 
v T0̂1̂ + T1̂0̂ + T  + T1̂1̂ = 0.
0̂ 0

It follows from these equations that

T0̂0̂ = −T1̂1̂ , T0̂1̂ = −T1̂0̂ .

Transformation of T0̂2̂ and T1̂2̂ give, respectively

   
T0̂2̂ = γ T0̂2̂ + vT1̂2̂ , T1̂2̂ = γ T1̂2̂ + vT0̂2̂ ,

which demands that

T0̂2̂ = T1̂2̂ = 0.

In the same way one finds

T2̂0̂ = T2̂1̂ = T0̂3̂ = T1̂3̂ = T3̂0̂ = T3̂1̂ = 0.

Thus, as a result of Lorentz invariance of the components Tμ̂ν̂ under a Lorentz

transformation in the x 1̂ -direction the energy–momentum tensor must have the form
Solutions to the Exercises 461

⎡ ⎤
T0̂0̂ T0̂1̂ 0 0
⎢ −T −T 0 0 ⎥
Tμ̂ν̂ =⎢
⎣ 0
0̂1̂ 0̂0̂ ⎥.
0 T2̂2̂ T2̂3̂ ⎦
0 0 T3̂2̂ T3̂3̂

Demanding Lorentz invariance under a Lorentz transformation in the x 2̂ -direction

gives the additional conditions

T0̂1̂ = T1̂0̂ = T2̂3̂ = T3̂2̂ = 0, T2̂2̂ = T0̂0̂ .

Lastly Lorentz invariance under a Lorentz transformation in the x 3̂ -direction gives

the additional condition

T3̂3̂ = T0̂0̂ .

It follows that the energy–momentum tensor for LIVE has the form

Tμ̂ν̂ = T0̂0̂ diag(−1, 1, 1, 1) = T0̂0̂ ημ̂ν̂ ,

where ημ̂ν̂ are the components of the Minkowski metric. This is valid in orthonormal
basis. Transforming to an arbitrary basis with metric gμν , the energy–momentum
tensor takes the form

Tμν = T0̂0̂ gμν .

From the physical interpretation of the components of the energy–momentum

tensor (including here the velocity of light) we have that T0̂0̂ = −ρLIVE c2 , where
ρLIVE is the mass density of LIVE. Thus

(TLIVE )μν = −ρLIVE gμν .

(c) We now assume that LIVE can be described as a perfect fluid with energy–
momentum tensor of the form (7.14),

 pLIVE
(TLIVE )μν = ρLIVE + u μ u ν + pLIVE gμν .
c2
Since the energy–momentum tensor of LIVE is proportional to the metric tensor,
the first term must vanish. Hence theequation of stateof LIVE is

pLIVE = −ρLIVE c2 .
462 Solutions to the Exercises

(d) Einstein’ field equations with a cosmological constant for empty space, (7.41),
can be written

1
Rμν − gμν R = −gμν .
2
We comparing this with Einstein’s field equations without a cosmological
constant, (7.29),

1
Rμν − gμν R = κ Tμν .
2
Hence, a spacetime with LIVE, and nothing else, is mathematically equivalent to
an Einstein space if LIVE is represented by an energy–momentum tensor

(TL I V E )μν = −(/κ)gμν .

We see that the cosmological constant represents the density of LIVE

 = κρ L I V E .

Solutions for Chapter 8

8.1 Non-relativistic Kepler motion
(a) According to Newtonian Lagrange dynamics the Lagrange function of a planet
with mass m (considered as a particle) moving in the gravitational field of the
Sun is it kinetic energy minus its potential energy. In spherical coordinates,
(r, θ, φ), the kinetic energy is

1 2 m 2 
T = mv = ṙ + r 2 θ̇ 2 + r 2 sin2 θ φ̇ 2 , (S8.1)
2 2
where ṙ = dr/dt and so forth. The potential energy is

G Mm
V =− , (S8.2)
r
where M is the mass of the Sun. Hence, the Lagrange function is

m 2 m 2 2 m 2 2 2 G Mm
L= ṙ + r θ̇ + r sin θ φ̇ L + . (S8.3)
2 2 2 r
Since L is does not depend upon φ, the momentum
Solutions to the Exercises 463

∂L
pφ = = mr 2 sin2 θ φ̇ (S8.4)
∂ φ̇

is a constant of motion. It is the angular momentum of the planet.

The Lagrange equation for θ ,
 
∂L d ∂L
− =0 (S8.5)
∂θ dt ∂ θ̇

takes the form

mr 2 sin θ cos θ φ̇ 2 − 2mr ṙ θ̇ − mr 2 θ̈ = 0 (S8.6)

Substituting for φ̇ from Eq. (S8.4) we get

2 pφ2 cos θ θ̇
− 4mr 3ṙ θ̇ 2 − 2mr 4 θ̇ θ̈ = 0 (S8.7)
mr 2 sin3 θ
which may be written
 
d pφ2
mr θ̇ +
4 2
= 0. (S8.8)
dt m sin2 θ

Integrating gives

pφ2
mr 4 θ̇ 2 + = K, (S8.9)
m sin2 θ

where K is a constant determined from the initial condition θ(0) = π/2, θ̇ (0) = 0.
This gives K = pφ2 /m. Hence Eq. (S8.9) reduces to

pφ2 pφ2
mr 4 θ̇ 2 + = . (S8.10)
m sin2 θ m

This may be written

 
m 2 r 4 sin2 θ θ̇ 2 = pφ2 sin2 θ − 1 = − pφ2 cos2 θ. (S8.11)

In general the left-hand side of this equation is positive, and the right-hand side
is negative. Hence, the only possibility of having this equation fulfilled is that both
sides vanish, which happens for θ = π/2, θ̇ = 0. This shows that the planet moves
in a plane.

(b) We shall now consider motion in this plane. The Lagrange equation for r,
464 Solutions to the Exercises

 
∂L d ∂L
− =0 (S8.12)
∂r dt ∂ ṙ

takes the form

GM
r̈ − r θ̇ 2 − r sin2 θ φ̇ 2 = − . (S8.13)
r2
which is the same as the r-component of Newton’s 2. law applied to the planet. With
θ = π/2 the Lagrange Eq. (S8.4) for φ becomes
pφ
φ̇ = . (S8.14)
mr 2
and the Lagrange Eq. (S8.13) for r reduces to

GM
r̈ + r φ̇ 2 = − . (S8.15)
r2
In order to find the orbit equation we express the derivatives with respect to t in
terms of derivatives with the respect to φ,

dr dr dφ pφ 
=
ṙ = = φ̇r  = r, (S8.16)
dt dφ dt mr 2
   
d dr pφ d pφ dr pφ2  2 pφ2   2
r̈ = = = r − r . (S8.17)
dt dt mr 2 dφ mr 2 dφ m 2r 4 m 2r 5

where r  = dr/dφ. Introducing a new radial coordinate, u = 1/r , we get

 2
 u  u  2 u
r = − 2, r = − 2 + . (S8.18)
u u u3
Inserting this into the expressions (S8.15) and (S8.16) for the radial velocity and
acceleration, we get

pφ  pφ2
ṙ = − u , r̈ = − 2 u 2 u  . (S8.19)
m m
Substituting this into Eq. (S8.15) gives the orbit equation

G Mm 2
u  + u = . (S8.20)
pφ2
Solutions to the Exercises 465

This is an inhomogeneous equation for harmonic oscillations with general solution

G Mm 2
u= [1 + ε sin(φ − φ0 )], (S8.21)
pφ2

where ε and φ0 are integration constants. With the boundary condition u(0) =
G Mm 2 / pφ2 we have φ0 = 0.
The equation of an ellipse with major half axis a and eccentricity ε in polar
coordinates with origin at one of the focal points has the form

1
u=   (1 + ε sin φ). (S8.22)
a 1 − ε2

Hence with non-relativistic Kepler motion the planet moves along an elliptical
path.
Equation (S8.20) has a particular solution representing circular motion with radius

pφ2
R= . (S8.23)
G Mm 2
Using Eq. (S8.14) the period of the motion is

2π m R2
T0 = = 2π . (S8.24)
φ̇ pφ

From Eq. (S8.23) we have

pφ2 = G Mm 2 R. (S8.25)

Inserting this into Eq. (S8.24) gives

R 3/2
T0 = 2π √ . (S8.26)
GM

(c) We shall now consider the effect upon the motion of the planets of a flattening
of the Sun. The gravitational potential of the Sun is then

GM S
V (r ) = − − 3. (S8.27)
r r
In this case the orbit Eq. (S8.20) is modified to

G Mm 2 3Sm 2 2
u  + u = 2
+ u . (S8.28)
pφ pφ2
466 Solutions to the Exercises

Again there is a particular solution representing circular motion. This time the
radius R = 1/u 0 is given by

pφ2 GM
u 20 − u0 + = 0. (S8.29)
3Sm 2 3S
Hence

3Sm 2
pφ2 = G Mm 2 R + . (S8.30)
R
The period is still given by Eq. (S8.24). Inserting Eq. (S8.30) leads to

T0
T = . (S8.31)
1 + 3S/G M R 2

The flattening of the Sun makes the orbital period of the planets a little longer
than in the spherical case.
(d) We shall now assume that the planet moves along an elliptical path which devi-
ates slightly from a circle, and calculate the precession of the perihelion due to
the flattening of the Earth. Hence, u = u 0 +u 1 where u 0 = 1/R and |u 1 | |u 0 |.
Inserting this into Eq. (S8.28) we get

m2  
u 1 + u 0 + u 1 = 2
G M + 3Su 20 + 6Su 0 u 1 + 3Su 21 . (S8.32)
pφ

Calculating to 1. order in u 1 and utilizing that u 0 fulfils Eq. (S8.29) we get

6Sm 2
u 1 + u 1 = u1, (S8.33)
pφ2 R

or

6Sm 2
u 1 + ω2 u 1 = 0, ω2 = 1 − . (S8.34)
pφ2 R

This is an equation for harmonic oscillations for 6Sm 2 < pφ2 R. The solution with
u 1 (0) = 0 is

u 1 = εu 0 sin(ωφ), (S8.35)

representing an ellipse with eccentricity ε. The period of the orbit is 2π/ω. For
ω = 1 the path is a closed ellipse with fixed orientation. For ω = 1 the ellipse will
Solutions to the Exercises 467

rotate. There will be a precession of the ellipse, where the ellipse rotates an angle


1
φ = 2π −1 (S8.36)
ω

for each travel around the sun. Using the expression (S8.34) for ω and inserting the
expression (S8.30) for pφ gives
√ 
G M R 2 + 3S 6π S
φ = 2π √ −1 ≈ . (S8.37)
G M R 2 − 3S G M R2

The quantity S is usually written

S = (1/2)J2 G M R Sun
2
, (S8.39)

where J2 is a numerical factor determined by measurements, J2 = 3 × 10−5 and

R Sun = 7.0 × 108 m. For the planet Mercury the distance from the Sun is R =
5.8 × 1010 m. This gives φ = 4.1 × 10−8 radians per round trip, which corresponds
to 3.5 per century, too small to explain the disagreement of 43 per century between
the Newtonian prediction and measurements.

8.2 The Schwarzschild solution in isotropic coordinates

(a) Incurvature coordinates the line-element of the Schwarzschild space time has
the form

 
RS 2 2 dr 2
ds 2 = − 1 − c dt + + r 2 d2 . (S8.40)
r 1 − RrS

We introduce a so-called radial isotropic coordinate, ρ(r ), where the line-element

of the Schwarzschild spacetime takes the form
 
RS
ds 2 = − 1 − dt 2 + f 2 (ρ)(dρ 2 + ρ 2 d2 ), (S8.41)
r (ρ)

This requires that

dr 2
= f 2 (ρ)dρ 2 , (S8.42)
1− RS
r

and

r 2 = f 2 (ρ)ρ 2 . (S8.43)
468 Solutions to the Exercises

Dividing these equations by each other and taking the positive square root gives

dρ dr
= . (S8.44)
ρ r − RS r
2

Integrating gives

ρ RS  2
=r− + r − RS r . (S8.45)
C 2
It is usual with isotropic coordinates, to choose C = 1/2. This gives

r RS 1 2
ρ= − + r − RS r . (S8.46)
2 4 2
Solving this equation with respect to r gives

(ρ + R S /4)2
r= . (S8.47)
4ρ

From (S8.43) and (S8.47) we have

 
r RS 2
f (ρ) = = 1+ . (S8.48)
ρ 4ρ

Inserting the expressions (S8.47) and (S8.48) into Eq. (S8.41) gives the line-
element of the Schwarzschild spacetime in isotropic coordinates
   
1 − R S /4ρ 2 2 RS 4 2 
ds = −
2
dt + 1 + dρ + ρ 2 d2 . (S8.49)
1 + R S /4ρ 4ρ

A series expansion of this line element to 1. order in R S /ρ gives

 
ds 2 = −(1 − R S /ρ)dt 2 + (1 + R S /ρ) dρ 2 + ρ 2 d2 . (S8.50)

(b) This coordinate system does only exist outside the horizon of a black hole. From
Eq. (S8.46) we find that the value of the isotropic radius at the horizon is

ρ(R S ) = R S /4. (S8.51)

8.3 Proper radial distance in the external Schwarzschild space

The proper radial distance from the horizon to a point in the Schwarzschild spacetime
with radial curvature coordinate r is
Solutions to the Exercises 469

r  ,   -
dr RS r RS
r̂ = r̂ = √ =r 1 − + R S ln 1− 1− . (S8.52)
1 − r/R S r RS r
RS

8.4 The Schwarzschild–de Sitter metric

(a) We shall here generalize the Schwarzschild solution (8.35) to a space-time

with a non-vanishing cosmological constant. We consider a static space-time
with a spherically symmetric 3-space outside a spherical body (the Sun) with
Schwarzschild radiusR S as described in orthonormal basis associated with cur-
vature coordinates. Then the line-element has the form (8.1) and the non-
vanishing components of the Einstein tensor are given in Eq. 8.15). Einstein’s
field equations are

E tˆtˆ = , Er̂ r̂ = −. (S8.53)

Adding the equations and inserting the expressions (8.15) for the components of
the Einstein tensor, leads to β = −α and

2 −2β  1
e β + 2 (1 − e−2β ) = , (S8.54)
r r
which may be written as

d  
r 1 − e−2β = r 2 . (S8.55)
dr
The general solution of this equation is

  
r 1 − e−2β = r 3 + K , (S8.56)
3
where K is an integration constant. Going to the Newtonian limit we find in the same
manner as in a spacetime without a cosmological constant K = R S . Thus, we get

RS 
e2α = e−2β = 1 − − r 2, (S8.57)
r 3
and the line-element takes the form
 
RS  dr 2
ds 2 = − 1 − − r 2 c2 dt 2 + + r 2 d2 . (S8.58)
r 3 1 − RrS − 3 r 2
√
Introducing a De Sitter radius R = 3/, the line element takes the form
470 Solutions to the Exercises
 
RS r2 dr 2
ds = − 1 −
2
− 2 c2 dt 2 + r2
+ r 2 d2 . (S8.59)
r R 1 − RrS − 2
R

(b) In globally empty space, but with LIVE, R = 0, and the line element reduces
to

 
r2 dr 2
ds 2 = − 1 − 2 c2 dt 2 + 2 + r d .
2 2
(S8.60)
R 1 − Rr 2


The spacetime described by this line element is called the DeSitter spacetime after
the Duch astronomer Willem De Sitter who found it as a solution of Einstein’s field
equations with a cosmological constant for empty space already in 1917.
Standard clocks at rest in this coordinate system show a time τ related to the
coordinate time t by

dτ = 1 − r 2 /R
2
dt. (S8.61)

The coordinate clocks of the line-element (S8.60) are adjusted to go at the same
rate as a standard clock at the origin independent of their position. Hence the standard
clocks go at a slower rate the farther they are from the origin. This means that they are
in a gravitational field pointing outwards from the origin. The time does not proceed
at r = R .
Hence, the line-element has a singularity at r = R . In order to find its physical
significance, we consider light moving radially. Along the world line of light, ds 2 = 0,
and hence, for light emitted in the negative r-direction, i.e. towards an observer at
the origin, we have
 
dr r2
= − 1 − 2 c. (S8.62)
dt R

Since the coordinate clocks go at the same rate as a standard clock at the origin,
Eq. (S8.61) represents the velocity of light as measured by an observer at the origin.
As measured by this observer light emitted towards him from r = R does not
proceed at all. This means the observer cannot receive information from the region
outside the spherical surface with radius r = R . Hence this surface is a horizon for
observers inside it.
In our universe observations indicate that the cosmological constant has a value
which corresponds approximately to a density of LIVE equal to the critical density
of the universe (see Chap. 12),  ≈ 10−52 m−2 , giving R ≈ 1026 m. This is
approximately equal to ten billion light years.

8.5 The perihelion precession of Mercury and the cosmological constant

Solutions to the Exercises 471

(a) Weshallnow find the orbit equation of a body moving in the gravitational field
of the central mass distribution. It follows from Eq. (S8.45) that the Lagrange
function of the body is

  1 2
1 Rs  ṙ 1 1
L=− 1− − r 2 t˙2 + 2
+ r 2 θ̇ 2 + r 2 sin2 θ φ̇ 2 (S8.63)
2 r 3 1− Rs
r
− 3 r 2 2 2

with constants of motion

 
∂L RS 
pt = =− 1− − r 2 t˙, pφ = r 2 sin2 θ φ̇. (S8.64)
∂ t˙ r 3

In the same way as in the case  = 0 the motion is planar, so we can put
θ = π/2, θ̇ = 0. Then the 4-velocity identity takes the form

pt2 ṙ 2 pφ2
− + + = −1, (S8.65)
1− RS
r
− 3 r 2 1− RS
r
− 3 r 2 r2

giving
  
RS  pφ2
ṙ =
2
pt2 − 1− − r2 1+ 2 . (S8.66)
r 3 r

The orbit equation is found in the same way as with  = 0, and it turns out that
Eq. (8.117) is generalized to

d2 u RS 3 
2
+u = 2
+ RS u2 − 2 3 . (S8.67)
dφ 2 pφ 2 3 pφ u

(c) As with  = 0 the solution of this equation represents a slowly rotating ellipse.
We shall calculate the perihelion precession of the ellipse. A circular motion
has a constant radius with u = u 0 , where u 0 fulfils

RS 3 
u0 = 2
+ R S u 20 − 2 3 , (S8.68)
2 pφ 2 3 pφ u 0

giving

1 (3/2)R S u 30 −  3R S /r03 − 
pφ2 = 4 1 − (3/2)R u
≈ r04 . (S8.69)
3u 0 S 0 1 − 3R S /2r0
472 Solutions to the Exercises

The path of a planet is a perturbation of the circle, u = u 0 +u 1 , where |u 1 | << |u 0 |.

Inserting this into Eq. (S8.66) and calculating to 1. Order in u 1 gives

d2 u 1 3 r05
+ u 1 = 2ku 0 u 1 , k = R S + . (S8.70)
dφ 2 2 2 pφ2

This is an equation of harmonic oscillations with general solution


u 1 = εu 0 cos[ f (φ − φ0 )], f = 1 − 2ku 0 ≈ 1 − ku 0 . (S8.71)

The period of the oscillations is 2π/ f . Equation (S8.71) describes a rotating

ellipse. The precession angle per round trip is
  
1 3R S r04
φ = 2π − 1 ≈ 2π ku 0 = π + 2 . (S8.72)
f r0 pφ

(c) Inserting the approximate value of pφ2 from Eq. (S8.69) gives

 
3R S2 + r0 − R2S r03
φ ≈ π   . (S8.73)
r0 R S − 3 r03

Since R S r0 the last term in the parenthesis in the numerator can be neglected.
With the unit we use here the cosmological constant has dimension m−2 . Hence
we can introduce
√ a length characterizing the value of the cosmological constant,
R = 1/ . Inserting this into Eq. (S8.73) we get

3π 3R S2 R
2
+ r04
φ ≈ . (S8.74)
r0 3R S R2
− r03

The cosmological constant has a value which corresponds approximately to a

density of LIVE equal to the critical density of the universe (see Chap. 12),  ≈ 10−52
m−2 , giving R ≈ 1026 m. The Schwarzschild radius of the Sun is R S ≈ 3 × 103 m,
and Mercury’s distance from the Sun r0 = 5.8 × 1010 m. This gives r03 R S R 2
.
Hence the expression for the perihelion precession of Mercury can be approximated
by

RS r3
φ ≈ 3π +π 0 2 . (S8.75)
r0 R S R
Solutions to the Exercises 473

Here the first term is the general relativistic precession given in equation (8.145),
and the last term is the contribution from the cosmological constant. Inserting numer-
ical values givesφ ≈ 5.1 × 10−7 + 2 × 10−23 . Hence, the contribution from the
cosmological constant is negligible.

8.6 Relativistic time effects and GPS

According to the general physical interpretation of a time-like spacetime interval the

proper time interval measured on a standard clock moving in spacetime with a metric
gμν is
 
gμν dx μ dx ν 1/2
dτ = − . (S8.76)
c2

In order to find the magnitude of the relativistic effects it is sufficient to consider

a satellite moving along a circular path with radius rs in the equatorial plane. For a
clock moving with velocity v φ = r dφdt
in the Schwarzschild metric this gives
  2 1/2
RS vφ
dτ = 1 − − dt, (S8.77)
rs c

where t is the coordinate time as shown by coordinate clocks adjusted to go with a

position independent rate of time. Here R S = 0.01 m is the Schwarzschild radius of
the Earth. The radius of the Earth is approximately r E = 6.4 × 106 m. Hence at the
surface of the Earth R S /r E = 1.6 × 10−9 .
A typical height and velocity of the GPS satellite are rs = 2 × 104 km and
φ
v = 4 × 103 m/s. Hence the magnitudes of the terms inside the parenthesis are
 2
RS vφ
= 5 × 10−10 , = 1.7 × 10−10 .
rE c

These are the relative magnitudes of the relativistic effects. Hence we can with
 2
good accuracy make a Taylor expansion to first order in R S /r and v φ /c . To this
order the difference between the proper time interval as measured on the satellite
clocks and coordinate time interval corresponding to a coordinate interval t is
 2
1 RS h 1 vφ
τ = τG + τV = t − t, (S8.78)
2 r E (r E + h) 2 c

where τG is the gravitational effect and τV the kinematical effect. Here h is the
height of the satellite above the surface of the Earth. Inserting numerical values gives
for a coordinate time interval corresponding to a day, t = 24h = 8.6 × 104 s that
τG = 5.2 × 10−5 s and τV = −7.0 × 10−6 s. The minus sign means that the
kinematical effect makes the satellite clocks go slower due to the velocity-dependent
474 Solutions to the Exercises

time dilation. The gravitational effect acts in the opposite way, making the satellite
clocks go faster that the clocks on the Earth. It should be noted that the gravitational
effect is larger than the kinematical effect. The gravitational effect makes the satellite
clock go faster that the Earth clock by 52 microseconds in 24 h, while the kinematical
effect slows down the satellite clock with 7 microseconds.
To compute the position of an object by means of the GPS-system with a precision
of 1m the GPS satellite clocks must measure time with a precision of one part in
1013 . Our calculation shows that to obtain such a precision both the special relativistic
kinematical time dilation and the general relativistic gravitational time effect must
be taken into account.

8.7 The photon sphere

We consider a photon travelling at constant radius. Due to the spherical symmetry

there is no loss of generality if we choose to consider a photon which travels in the
equatorial plane. We the put dr = dθ = 0 and have θ = π/2 in the Schwarzschild
line-element. Furthermore since photons follow null-geodesic curves we have ds =
0. Hence the equation of motion of the considered photon is
 
RS 2 2
1− c dt = r 2 dφ 2 , (S8.79)
r

Giving
 2  
dφ c2 RS
= 2 1− . (S8.80)
dt r r

We now use the radial geodesic equation

d2 r
+ μν
r
u μ u ν = 0. (S8.81)
dτ 2
The only non-vanishing Christoffel symbols with an upper index r are
   
c2 RS RS RS RS
ttr = 1− , rr = − 
r
 , θθ = φφ = −r 1 −
r r
.
2 r r2 2 1 − RrS r 2 r
(S8.82)

d2 r
Furthermore dr
dτ
= dθ
dτ
= dτ 2
= 0. Hence the geodesic equation reduces to
 2  2
dφ dt
φφ
r
= −ttr , (S8.83)
dτ dτ
or
Solutions to the Exercises 475
 2
dφ ttr
=− r . (S8.84)
dt φφ

Inserting the expressions (S8.82) for the Christoffel symbols leads to

 2
dφ c2 R S
= . (S8.85)
dt 2r 3

Equations (S8.80) and (S8.85) give

 
c2 RS c2 R S
2
1 − = . (S8.86)
r r 2r 3

Solving this equation with respect to r gives the radius of the photon sphere in the
Schwarzschild spacetime

3
r= RS . (S8.87)
2

Solutions for Chapter 10

10.1 A spaceship falling into a black hole

(a) We shall consider a spaceship falling radially into a Schwarzschild black hole
with the mass of the Sun, M = 2, 0 × 1030 kg. The Schwarzschild radius of
the black hole is

2G M 2 · 6.67 × 10−11 m3 /kgs2 · 2.0 × 1030 kg

RS = =  2 = 3.0 km. (S10.1)
c2 3.0 × 108 m/s

The line-element has the form

 
RS 2 2 dr 2
ds = − 1 −
2
c dt + + r 2 d2 . (S10.2)
r 1 − RrS

Hence, the Lagrange function of the spaceship is

 
1 R S 2 2 1 ṙ 2
L=− 1− c t˙ + . (S10.3)
2 r 2 1 − RrS
476 Solutions to the Exercises

Since t is a cyclic coordinate,

 
RS 2
pt = 1 − c t˙ (S10.4)
r

is a constant of motion. The 4 velocity identity takes the form

 
RS 2 2 ṙ 2
− 1− c t˙ + = −c2 . (S10.5)
r 1 − RrS

Substituting for t˙ from the previous equation we get

 
pt2 RS 2
ṙ 2 = − 1 − c . (S10.6)
c2 r

We consider a spaceship which falls from r = r0 at τ = 0, i.e. with ṙ (r0 ) = 0.

This gives
 
RS 4
pt2 = 1− c . (S10.7)
r0

Hence,

dr  1 1
= −c R S − . (S10.8)
dτ r r0

Integration gives
    
r0 r r r0 r0
τ (r ) = 1− + arctan −1 . (S10.9)
c r0 RS RS r

The proper time taken to fall from the Earth to the Sun is
   
r0 RS r0 r0
τ (R S ) = 1− + arctan −1 . (S10.10)
c r0 RS RS

Since R S r0 a good approximation is


π r0 r0
τ (R S ) ≈ , (S10.11)
2 c RS

where we have used that lim arctan x = π/2 Here r0 is the distance from the Earth
x→∞
to the Sun, i.e. r0 = 1.5 × 108 km. Inserting numerical values gives τ (R S ) ≈ 64
Solutions to the Exercises 477

days. This is the same as the Newtonian result (1.11) for the case considered here,
where we neglect the gravitational field of the Earth.
(b) Let A be an observer on the spaceship and B a stationary observer at a distance
r B from the Sun. Both A and B emit signals with angular frequency ω. A receives
the signals from B with an angular frequency ω A , and B receives the signals
from A with angular frequency ω B .

We shall find ω A and ω B by utilizing an inertial observer C instantaneously at rest

in the coordinate system at r A . This observer is at rest relatively to B. The observer
C receives signals from A with an angular frequency ωC A and from B with ωC B .
Let us first consider the signals from A to B via C. Since A moves away from C
the signal from A to C will be redshifted due to the kinematical Doppler effect, but
there is no gravitational frequency shift since A and C is at the same position during
their exchange of signals. Hence

1 − |v|/c
ωC A = ω. (S10.12)
1 + |v|/c

The velocity as measured by standard clocks in a local inertial frame is given in

Eq. (S10.8).
We then consider the signal from C to B. Then there is no kinematical Doppler
effect since B and C are at rest relatively to each other. But there is a gravitational
frequency shift,
 
(gtt )C 1 − R S /r A
ωB = ωC A = ωC A . (S10.13)
(gtt ) B 1 − R S /r B

Inserting the expression (S10.8) for the velocity into Eq. (S10.12) and combining
with Eq. (S10.13) gives
.
/ 
 / 1 − √R
S rA −
1 1
1 − R S /r A /
/ r0
ωB =  ω.
1 − R S /r B 0 1 + √ R
(S10.14)
S rA −
1 1
r0

We then consider the signals from B to the spaceship A in a similar manner. The
signal is first sent from B to C. Then there is no kinematical Doppler effect, only
a gravitational blue shift, since these signals move downwards in the gravitational
field of the sun,
 
(gtt ) B 1 − R S /r B
ωC B = ω= ω. (S10.15)
(gtt ) A 1 − R S /r A
478 Solutions to the Exercises

Then the signals are emitted further from C to A. Since A moves away from C
there is now a redshift,
.
/ 
/ 1 − √R
S rA −
1 1
/
/ r0
ωA = 0 √  ωC B . (S10.16)
1 + R S r1A − 1
r0

Combining the last two expressions we get

.
/ 
/ 1 − √R 
S rA −
1 1
/ 1 − R S /r B
ωA = /
r0
0 √  ω. (S10.17)
1 + R S r1A − 1 1 − R S /r A
r0

In the limit that the spaceship falls from an infinitely far away position we get

1 − rRBS
lim ω A =  ω. (S10.18)
r0 →∞
1 + Rr AS

In this case there is a redshift, which means that the kinematical redshift then
dominates over the gravitational blue shift.

10.2 Kinematics in the Kerr spacetime

(a) We shall consider light moving in negative and positive direction of φ in the
Kerr spacetime in the equatorial plane θ = π/2, as described in Boyer–Lindquist
coordinates. Using that ds 2 = 0 along the world line of light, we then get for
the angular velocity of light, ω L = dφ/dt, from the line element (10.26),

ω2L − 2ωω L + ω2 − e2ν−2ψ = 0. (S10.19)

where

RS a ν−ψ r r (r − R S ) + a 2
ω= 3 , e = 3 . (S10.20)
r + a 2 (r − 2R S ) r + a 2 (r − 2R S )

The solution of Eq. (S10.19) is


R S a ± r r (r − R S ) + a 2
ω± =ω±e ν−ψ
= . (S10.21)
L
r 3 + a 2 (r − 2R S )
Solutions to the Exercises 479

The velocity of light in the equatorial plane of the Kerr spacetime is c K φ = r ω L

which approaches 1, or c, when r → ∞. In the special case without rotation, a =
0, which represents the Schwarzschild spacetime, the coordinate velocity of light
reduces to

RS
c Sφ = 1 − . (S10.22)
r

which is isotropic. But in the Kerr spacetime the velocity is different in the positive
and the negative φ−direction. The time difference for light sent around the equator
in the positive and negative φ−direction is
 
1 1 4π R S a r 3 + (r − 2R S )a 2
t = 2π − − + = . (S10.23)
ωL ωL r − R S r 3 + (r + R S )a 2

This time difference vanishes for the non-rotating case, a = 0, and in general
when r → ∞.
(b) We are still describing particles and observers in the equatorial plane of the
Kerr space. In the following a ZAMO in the Kerr spacetime will be describing
particles with fixed r- and θ -coordinates. We introduce an orthonormal basis
(etˆ , er̂  , eθ̂  , eφ̂ ), where etˆ is the 4-velocity of the ZAMO. It follows from the
form of the line element

ds 2 = −e2ν dt 2 + e2μ dr 2 + e2λ dθ 2 + e2ψ (dφ − ωdt)2 , (S10.24)

that the dual basis 1-forms are

tˆ
ω
−
= eν ω
−
t
, ω
−
r̂ 
= eμ ω
−
r
,
θ̂
. (S10.25)
ω
−
= eλ ω
−
θ
, ω
−
φ̂
= eψ (ω
−
φ
− ωω
−
t
).

Using the fundamental contraction

ωμ (eν ) = δνμ (S10.26)

It follows that the basis-vectors of an orthonormal basis associated with the ZAMO
are

etˆ = e−ν (et + ωeφ ), er̂  = e−μ er ,

eθ̂  = e−λ eθ , eφ̂ = e−ψ eφ , (S10.27)

We shall now find the physical velocity as measured by a ZAMO of a particle

with 4-velocity components
480 Solutions to the Exercises

  dφ
U μ = t˙, φ̇ = t˙(1, ),  = (S10.28)
dt
in the Boyer–Lindquist coordinate system. Here the dot denotes differentiation with
respect to the proper time of the particle. We shall find the components of its velocity
with reference to the orthonormal basis above, carried by the ZAMO. Using the
tensor transformation of the vector components we get
 
U tˆ = eν U t = eν t˙, U φ̂ = eψ U φ − ωU t = eψ t˙( − ω).
 
(S10.29)

The physical velocity of the particle measured by the ZAMO is


 U φ̂
v φ̂ = = eψ−ν ( − ω). (S10.30)
U tˆ
A particle at rest in the coordinate system has  = 0 and hence the velocity

φ̂  RS a
v0 = −eψ−ν ω = −  . (S10.31)
r r (r − R S ) + a 2

We shall later need the factor

.
/  
1 / 1 + 1 − R S −1 a22
γ̂  =  0
2 =   r .
r
 RS a2
(S10.32)
φ̂  1 + 1 +
1 − v0 r r2

In the non-rotating case, with a = 0, the particle is at rest relative to the ZAMO;
i.e. observers at rest in the coordinate system are ZAMOs in the Schwarzschild
spacetime. But in the Kerr spacetime observers at rest in the coordinate system are
not ZAMOs. At r = R S the velocity of a particle at rest in the coordinate system is
φ̂
v0 = −1, i. the particle moves with the velocity of light relative to the ZAMO.
(c) We shall introduce an orthonormal basis co-moving with observers at rest in the
Boyer–Lindquist coordinate system. The vectors of this orthonormal basis are

etˆ = (−gtt )−1/2 et , eî = (γii )−1/2 [ei − (git /gtt )et ], (S10.33)

where

γii = gii − git2 /gtt . (S10.34)

are the components of the spatial metric tensor. Using that

Solutions to the Exercises 481
   2 
φ̂
gtt = e2ψ ω2 − e2ν = −e2ν 1 − v0 = −e2ν /γ̂  2 ,

grr = e2μ , gφφ = e2ψ , gθθ = e2λ , gtφ = −e2ψ ω (S10.35)

we get

γrr = grr = e2μ γθθ = gθθ = e2λ ,

2    2 
gφt ω2 e2ψ φ̂
γφφ = gφφ − =e 2ψ
+ 2ν −2 = e 2ψ
1 + γ̂ v0
2
= γ̂ 2 e2ψ . (S10.36)
gtt e γ̂

Inserting the expressions (S10.35) and (S10.36) into Eq. (S10.33) gives

etˆ = γ̂ e−ν et , er̂ = e−μ er ,

φ̂
eθ̂ = e−λ er , eφ̂ = γ̂ −1 e−ψ eφ + γ̂ e−ν v0 et . (S10.37)

The vector etˆ is the 4-velocity of an observer at rest in the BL-coordinate system,
i.e. of a static observer. Using the fundamental contraction (S10.26) we get the
corresponding basis 1-forms,

φ̂ 
ωtˆ = γ̂ −1 eν ωt − γ̂ eψ v0 ωφ , ωr̂ = eμ ωr , ωθ̂ = eλ ωθ , ωφ̂ = γ̂ eψ ωφ . (S10.38)

(d) The transformation matric leading from the coordinate basis vectors in the
Boyer–Lindquist system to the ZAMO can be read off from equation (S10.37),

⎛ ⎞
φ̂ 
γ̂ e−ν 0 0 γ̂ e−ν v0
 ⎜ ⎟
⎜
Mμ̂μ = ⎜ 0 e−μ 0 0 ⎟
⎟. (S10.39)
⎝0 0 e−λ 0 ⎠
0 0 0 γ̂ eψ

The inverse transformation matrix is

⎛ ⎞
φ̂ 
γ̂ −1 eν 0 0 −γ̂ eψ v0
 ⎜ ⎟
⎜
Mμμ̂ = ⎜ 0 eμ 0 0 ⎟
⎟. (S10.40)
⎝0 0 e−λ 0 ⎠
0 0 0 γ̂ eψ

The components in the orthonormal basis (S10.37) of the Boyer–Linquist system

of the 4-velocity of aa particle with the 4-velocity (S10.28) are found by applying
the transformation (S10.40), U μ̂ = Mμμ̂ U μ , which gives,
482 Solutions to the Exercises

φ̂  φ̂ 
U tˆ = γ̂ −1 eν U t − γ̂ eψ v0 U φ = γ̂ −1 eν t˙ − γ̂ eψ v0 t˙, U φ̂ = γ̂ eψ U φ = γ̂ eψ t˙.
(S10.41)

The physical velocity of the particle as observed by an observer at rest in the

Boyer–Lindquist coordinate system is

U φ̂ eψ−ν 
v φ̂ = = . (S10.42)
U tˆ γ̂ −2 + e2ψ−2ν ω

Using Eqs. (S10.30) and (S10.31) this can be written as

 φ̂ 
φ̂ v φ̂ − v0
v = , (S10.43)
φ̂ 
1 − v φ̂  v0

which is the relativistic formula for velocity addition.

10.3 A gravitomagnetic clock effect
Following [10.1] and [10.2] we consider two clocks moving freely in opposite direc-
tions in the equatorial plane of the Kerr spacetime outside a rotating body. The clocks
move along a path with r = constant and θ = π/2.
(a) In this case the radial geodesic equation

α β
d2 r r dx dx
+  αβ =0 (S10.44)
dτ 2 dτ dτ
reduces to

ttr dt 2 + 2φt
r
dφdt + φφ
r
dφ 2 = 0.. (S10.45)

Inserting the expression (5.65) for the Christoffel symbols we obtain

 2
dt dt
gtt,r + 2gφt,r + gφφ,r = 0. (S10.46)
dφ dφ

From the line element (10.28) we see that in the equatorial plane of the Kerr
spacetime the non-vanishing components of the metric are
 
RS RS a RS a2
gtt = − 1 − , gφt = , gφφ = r 2 + a 2 + . (S10.47)
r r r

Performing the differentiations and inserting the result into equation (S10.46)
gives
Solutions to the Exercises 483
 2
dt dt 2r 3
− 2a + a2 − = 0. (S10.48)
dφ dφ RS

The solution of this equation is

dt 1
=a± , (S10.49)
dφ ωK

where ω K is the Keplerian angular frequency,

2R S GM
ω2K = = 3 , (S10.50)
r3 r
and equation (S10.50) is an expression of Kepler’s 3. law.

(b) To find the gravitomagnetic clock effect one integrates equation (S10.49) over
2π for co-rotating and counter-rotating clocks. This gives

t± = TK ± 2πa, (S10.51)

where
2π
TK = (S10.52)
ωK

is the Keplerian period. The time difference in the travel time around the central
rotating body is

t = 4πa = 4π J/M, (S10.53)

where J is the angular momentum of the central body. Inserting the velocity of light
we have,

t = 4π J/Mc2 . (S10.54)

The mass of the Earth is m = 6 × 1026 kg and its angular momentum J =

10 kg m2 /s. Hence outside the Earth the travel time difference for clocks travelling
34

freely around the Equatior in opposite directions is 2 × 10−7 s.

References
10.1 Cohen, J.M., Mashoon, B.: Standard clocks, interferometry, and gravitomag-
netism. Phys. Lett. A181, 353–358 (1993).
10.2. Faruque, S.B.: Gravitomagnetic clock effect in the orbit of a spinning particle
orbiting the Kerr black hole. Phy. Lett. A327, 95–97 (2004).
484 Solutions to the Exercises

Solutions for Chapter 11

11.1 The Schwarzschild-de Sitter metric

In the static spherically symmetric case the line element can be written as in
Eq. (8.1). Then there are only 2 independent field equations, which can be taken
to be the tˆtˆ−and r̂ r̂ −equations. The components of the Einstein curvature tensor
are given in equation (8.15). Then the tˆtˆ−and r̂ r̂ −field equations (7.43) with a
cosmological constant for empty space take the form

2 −2β  1
e β + 2 (1 − e−2β ) = , (S11.1)
r r
2 −2β  1
e α − 2 (1 − e−2β ) = −. (S11.2)
r r
Adding the equations we get with a suitable coordinate condition β = −α.
Equation (S11.1) can be written as

d  
r 1 − e−2β = . (S11.3)
dr
The general solution of this equation is

K 
e−2β = 1 − − r 2. (S11.4)
r 3
Demanding that the solution reduces to the exterior Schwarzschild solution with
vanishing
√ cosmological constant, we get K = R S . Introducing the length R H =
3/ the line element can then be written as
 
RS r2 dr 2
ds = − 1 −
2
+ 2 c2 dt 2 + r2
+ r 2 d2 , (S11.5)
r RH 1 − RrS + R 2H

which describes the Schwarzschild-De Sitter spacetime.

11.2 A spherical domain wall in empty space described by the Israel formalism

Theenergy–momentum tensor of a domain wall is

Si j = −σ (3) gi j . (S11.6)

Hence, the only non-vanishing mixed components of the energy–momentum

tensor are
φ
Sθθ = Sφ = −σ. (S11.7)
Solutions to the Exercises 485

In empty space the equation of continuity of the domain wall reduces to (11.58)
which then gives

dσ
= 0 σ = constant (S11.8)
dτ
Hence, the mass density of the domain wall remains constant during radial motion
even if the area of the domain wall changes.
Inserting Eq. (S11.6) into Eq. (11.55) gives
 
3 κ
[K θθ ] = κ −σ (3) gθθ + σ (3) gθθ = σ R 2 . (S11.9)
2 2

This equation has the same form as Eq. (11.74) for dust, but there σ R 2 =constant,
while here σ =constant. The equation of motion (11.89) for dust is still valid, but
with M S replaced by 4π σ R 2 ,

M= 1 + Ṙ 2 − 2π σ R 4π σ R 2 . (S11.10)

A static spherical domain wall has

M = (1 − 2π σ R)4π σ R 2 , (S11.11)

which may be written

R S = R SS (1 − R SS /4R). (S11.12)

Here R S is the Schwarzschild radius of the domain wall as measured by an observer

far from it, R SS is Schwarzschild radius as measured by an observer at the wall, and
R is its radius.

Solutions for Chapter 12

12.1 Gravitational collapse

(a) We shall here study particles falling freely from rest at a position infinitely far
from a black hole with Schwarzschild radiusRS . In Exercise 10.1 we considered
a spaceship falling from a position r0 outside a black hole and into the hole.
Hence we shall here consider the limit r0 → ∞.

We found the following expression for the radial component of the 4-velocity
 
pt2 RS 2
ṙ 2 = − 1 − c . (S12.1)
c2 r
486 Solutions to the Exercises

The boundary condition ṙ (∞) = 0 gives pt = c2 . Hence


RS
ṙ = −c . (S12.2)
r

New integration with r (0) = ρ gives

2  3/2 
τ= √ ρ − r 3/2 . (S12.3)
3c RS

Inserting this expression into Eq. (S12.2) gives

√
 r
dt = −c RS dr. (S12.4)
r − RS

Integration with the boundary condition t(0) = τ gives

√
r/RS + 1 
1
ct = cτ + RS ln √ 1 − 2 RS r . (S12.5)
1 r/RS − 11

Solving the expression (S12.3) with respect to r gives

 √ 2/3  √ 
3c RS 3c RS −3/2
r = ρ 3/2 − τ =ρ 1− τρ . (S12.6)
2 2

Inserting this into Eq. (S12.5) gives the relationship between the proper time of a
freely falling particle and the coordinate time, which is the same as the proper time
of an observer at rest far away from the black hole. Note that ρ has a constant value
for a freely falling particle.

(b) We shall now assume that there is Schwarzschild spacetime outside the black
hole, and find the form of the spacetime line-element, using τ as time coordinate
and ρ as radial coordinate. The radial coordinate ρ is commoving with the freely
falling particles. Solving Eq. (S12.6) with respect to ρ gives

3 
ρ 3/2 = r 3/2 + c RS τ. (S12.7)
2
We see that ρ increases with τ . As seen from the freely falling particles points
with fixed Schwarzschild coordinates move outwards.
Taking the differential of t, using Eqs. (S12.5) and (S12.4) we get
√
∂t ∂t  r
dt = dτ + dr = dτ − c RS dr. (S12.8)
∂τ ∂r r − RS
Solutions to the Exercises 487

Inserting this into the Schwarzschild line-element (8.35) gives

  
RS 2 2 RS
ds = − 1 −
2
c dτ + 2 cdτ dr + dr 2 + r 2 d2 . (S12.9)
r r

Taking the differential of Eq. (S12.6) gives

 
∂r ∂r ρ RS
dr = dρ + dτ = dρ − dτ. (S12.10)
∂ρ ∂τ r r

Inserting this into Eq. (S12.8) and using Eq. (S12.6) gives

− 23
 43
3 − 23 3 − 23
ds = −c dτ + 1 −
2 2 2
RS cτρ dρ + 1 −
2
RS cτρ ρ 2 d2 .
2 2
(S12.11)

This metric is regular at the Schwarzschild radiusr = RS which has an increasing

co-moving radial coordinate,
 2/3
3c
ρS = RS 1 + τ , (S12.12)
2RS

which increases with time. But the metric is singular at

 √ 2/3
3 RS
ρ= cτ , (S12.13)
2

which corresponds to the centre of the black hole, r = 0.

(c) We shall now consider a collapsing star which has a position-dependent energy
density ρ(τ ), assuming that the pressure is zero. Assume further that the inte-
rior spacetime can be described by a Friedmann metric with Euclidean spatial
geometry (k = 0). We shall find the metric inside the star.

We assume that the star is homogeneous so that the density of the star is constant
in space, but increases with time since the star collapses. The total mass of the star
is assumed to be constant. The surface of the star has a radius given by equation
(S12.6). Let the radius of the star at an arbitrary point of time be R and at τ = 0 be
R0 . Hence the density of the star is

M 6M
ρ(τ ) = =  √ 2 . (S12.14)
(4/3)π R 3 −3/2
4π R03 2 − 3 RS cτ R0
488 Solutions to the Exercises

Inserting this into the Friedmann equation (12.46) for a flat universe with vanishing
cosmological constant gives,

ȧ 2 A2 RS
=  2 , A2 = . (S12.15)
a2 1 − 23 Acτ R03

Since the star collapses we take the negative root, getting

da Acdτ
=− . (S12.16)
a 1 − 23 Acτ

Integration with the initial condition a(0) = 1 gives

 2/3
3
a(τ ) = 1 − Acτ . (S12.17)
2

Hence the metric in the collapsing star is

√  4/3
3 R S  2 
ds 2 = −c2 dτ 2 + 1 − cτ dρ + ρ 2 d2 . (S12.18)
2 R03/2

12.2 The volume of a closed Lemaître-Friedmann-Robertson-Walker (LFRW)

universe

The volume of the region contained inside a radius r = a R0 χ in a closed LFRW

universe is
χ  
1
V = 4πa 3
R03 sin χ dχ = 2πa
2 3
R03 χ − sin 2χ . (S12.19)
2
0

12.3 Conformal time

The conformal time η of the Isotropic and homogeneous universe models is defined
by

dt = a(η)dη. (S12.20)

Inserting this into the line-element (12.1) gives

 
ds 2 = a(η)2 −c2 dη2 + R02 dχ 2 + r (χ )2 d2 . (S12.21)

Light moving radially have ds 2 = d2 = 0, and hence a coordinate velocity

Solutions to the Exercises 489

dχ c
= . (S12.22)
dη R0

The physical velocity of the light is

R0 dχ
= c. (S12.23)
dη

12.4 Lookback time and the age of the universe

(a) Accordingto Eq. (12.111) the Hubble parameter at the emission point of time is
given as a function of the redshift of the source by

 1/2
H (z) = H0 rad0 (1 + z)4 + m0 (1 + z)3 + k (1 + z)2 + L0 . (S12.24)

Using Eq. (12.113) the lookback time of a source with redshift z is

∞ ∞ z
dz dz dz
tLB (z) = t0 − tE (z) = − = ,
(1 + z)H (z) (1 + z)H (z) (1 + z)H (z)
0 z 0
(S12.25)

giving

z
dz
tLB (z) = tH 
(1 + z) rad0 (1 + z) + m0 (1 + z)3 + k (1 + z)2 + L0
4
0
(S12.26)

where tH = 1/H0 is the Hubble age of the universe, and L0 is the present value of
the density parameter of LIVE.
The age of the universe is

∞
dz
t0 = tLB (∞) = tH  .
(1 + z) rad0 (1 + z)4 + m0 (1 + z)3 + k (1 + z)2 + L0
0
(S12.27)

(b) An empty universe (the Milne universe) has rad0 = m0 = L0 = 0, k = 1
giving
490 Solutions to the Exercises

z
dz z
tLB (z) = tH =tH . (S12.28)
(1 + z) 2 1 + z
0

Hence the age of the Milne universe is equal to its Hubble age, t0 = tLB (∞) = tH .

(c) The Einstein–de Sitter universe is a flat, mass-dominated universe with rad0 =
L0 = k = 0, m0 = 1. Then the integral (S12.26) reduces to

z

dz 2 1
tLB (z) = tH = tH 1 − . (S12.29)
(1 + z)5/2 3 (1 + z)3/2
0

The age of this universe is

2
t0 = tLB (∞) = tH . (S12.30)
3
(d) A matter-dominated universe with curved 3-space has

z
dz
tLB (z) = tH √ . (S12.31)
(1 + z) m0 z + 1
2
0

For positive spatial curvature, m0 > 1, this gives

⎡ √ √ ⎤
m0 m0 − 1 1 + m0 z − 1
√ √
tH ⎢ ⎥
arcsin
⎢ m0 − 1 m0 1 + z ⎥
tLB (z) = ⎢ √ ⎥, (S12.32)
m0 − 1 ⎣ 1 + m0 z ⎦
+ −1
1+z

The age of the universe is

  
tH m0 1
t0 = tLB (∞) = √ arccos −1 . (S12.33)
m0 − 1 m0 − 1 m0

For negative curvature, m0 < 1, the integral (S12.30) gives

⎡ √ ⎤
1 + m0 z m0
1− −√ arcsinh⎥
tH ⎢ ⎢ 1+z 1 − m0 ⎥
tLB (z) = ⎢√ √  ⎥. (S12.34)
1 − m0 ⎣ 1 − m0 1 + m0 z − 1 ⎦
√
m0 1 + z
Solutions to the Exercises 491

The present age of this universe is

  
tH m0 1
t0 = tLB (∞) = 1− √ arccosh . (S12.35)
1 − m0 1 − m0 m0

(e) We shall now determine the age of a matter-dominated universe from the para-
metric solution (S12.140)–(S12.144). We shall first consider the universe model
with negatively curved 3-space. The present value of the conformal time is fund
from the requirement a(η0 ) = 1 which gives

 
cη0 2 − m0
= arccosh . (S12.36)
R0 m0

Inserting this into Eq. (S12.141) gives

 
tH 1 m0 2 − m0
t0 = t(η0 ) = 1− √ arccosh . (S12.37)
1 − m0 2 1 − m0 m0

This seems to be in conflict with equation (S12.35). However applying the identity

1  
arccoshx = arccosh 2x 2 − 1 (S12.38)
2
√
with x = 1/ m0 , shows that the expressions are identical.
Using Eqs. (S12.143) and (S12.144) we find the age of a universe with a positively
curved 3-space

  
tH 1 m0 2 − m0
t0 = t(η0 ) = √ arccos −1 . (S12.39)
m0 − 1 2 m0 − 1 m0

(f) For a flat universe with cold matter and LIVErad0 = k = 0, and the integral
(S12.26) reduces to

z
dz
tLB (z) = tH  , (S12.40)
(1 + z) m0 (1 + z)3 + L0
0

which gives
√ 
2tH L0 L0 +  m0 (1 + z) 3
− 1
tLB = √ arctanh  . (S12.41)
3 L0 L0 + m0 (1 + z)3 − L0
492 Solutions to the Exercises

The age of this universe is equal to the look back time of an object with infinitely
great redshift,
√
2 arctanh L0
t0 = tLB (∞) = tH √ (S12.42)
3 L0

in agreement with Eqs. (12.164) and (12.168). Since lim arctan x

x
= 1 we get
x→0

2
tLB = tH (S12.43)
3
for a flat matter-dominated universe with L0 = 0, in agreement with Eq. (S12.30).
(g) In the case of a flat LIVE-dominated universe with m0 = 0, L0 = 1 the
integral (S12.140) gives

tLB = tH ln(1 + z). (S12.44)

12.5 The LFRW universe models with a perfect fluid

Inthis problem we will investigate FRW models with a perfect fluid. We will assume
that the perfect fluid obeys the equation of state

p = wρc2 , (S12.45)

where −1 ≤ w ≤ 1.
(a) The Friedmann equations (12.46) and (12.57) with vanishing cosmological
constant are

ȧ 2 + kc2 /R02
3 = κρ (S12.46)
a2
and
κ p
ä = − a(ρ + 3 2 ). (S12.47)
6 c
Inserting equation (S12.45) the last equation takes the form
κ
ä = − a(1 + 3w)ρ. (S12.48)
6
The integral (12.106) of the energy–momentum conservation equation with the
normalization a(t0 ) = 1 takes the form
Solutions to the Exercises 493

ρ = ρ0 a −3(1+w) . (S12.49)

(b) For w > −1/3 Eq. (S12.48) gives ä < 0, i.e. decelerated expansion. Hence
at a point of time the expansion will stop, ȧ = 0 and the universe will start to
collapse. It follows from equation (S12.46) that this is only possible for k > 0,
i.e. for a closed universe.
(c) We now assume that the spatial curvature is Euclidean, k = 0. Inserting (S12.49)
into (S12.46) then gives

κ
a 1+3w ȧ 2 = ρ0 . (S12.50)
3
Integration for w = −1 with the conditions a(0) = 0, a(t0 ) = 1 gives
  3(1+w)
2
t
a(t) = . (S12.51)
t0

The Hubble parameter is

ȧ 2 1
H= = , (S12.52)
a 3(1 + w) t

and the deceleration parameter is

a ä 1
q=− 2
= (1 + 3w). (S12.53)
ȧ 2
It follows from Eqs. (S12.49) and (S12.51) that the time evolution of the mass
density is
 2
t0
ρ(t) = ρ0 . (S12.54)
t

(d) For the present universe model the time evolution of the radius of the particle
horizon is given in Eq. (12.194),

3(1 + w)
lPH = t. (S12.55)
1 + 3w

This shall be expressed in terms of the redshift and the Hubble age of the universe.
Using Eq. (S12.51) the relation between the redshift and the cosmic time is
494 Solutions to the Exercises

  3(1+w)
2
1 t0
1+z = = . (S12.56)
a t

Hence

t = t0 (1 + z)− 2 (1+w) .
3
(S12.57)

Inserting this into Eq. (S12.55) gives

3(1 + w)
t0 (1 + z)− 2 (1+w) .
3
lPH = (S12.58)
1 + 3w

The relationship between the age of the universe and its Hubble age is given in
equation (12.193),

2
t0 = tH . (S12.59)
3(1 + w)

Inserting this into equation (S12.54) gives

2
tH (1 + z)− 2 (1+w) .
3
lPH = (S12.60)
1 + 3w

(e) For a dust-dominated universe, w = 0, the above results give ρa 3 = constant,

which means that the mass of dust in a co-moving volume is constant. The
time evolution of the scale factor, the Hubble parameter and the deceleration
parameter is a(t) = (t/t0 )2/3 , H = 23 1t and q = 1/2, the first two in agreement
with Eqs. (12.142) and (12.145). For this universe model the radius of the particle
horizon evolve as lPH = 3t = 2tH (1 + z)−(3/2) .

For a radiation-dominated universe model, w = 1/3, we get ρ = ρ0 a −4 . Hence

the energy of radiation in a co-moving volume decreases with time. This is due to
the work performed by the radiation upon the region outside a co-moving volume.
The time evolution of the scale factor, the Hubble parameter and the deceleration
parameter is a(t) = (t/t0 )1/2 , H = 21 1t and q = 1. The time evolution of the radius
of the particle horizon is lPH = 2t = 2tH (1 + z)−(3/2) = t H (1 + z)−2 .
For a LIVE-dominated universe, w = −1, we get ρ = ρ0 ; i.e. the density of
the vacuum energy is constant in the expanding universe. Hence the vacuum energy
in a co-moving volume increases. This is due to the negative work performed at
the boundary of the co-moving volume due to the negative pressure of the vacuum
energy. Hence vacuum energy is extracted from an infinitely remote region into the
regions at a finite distance from the origin.
Integration of equation (S12.46) for w = −1 gives a(t) = e H0 t . The Hubble
parameter is constant, H = H0 , and the deceleration parameter is q = −1. This
universe model has no particle horizon.
Solutions to the Exercises 495

(f) We shall find a general formula for the deceleration parameter of a LFRW
universe model,

ä
q=− . (S12.61)
aH2

From ȧ = a H we have ä = ȧ H + a Ḣ . Hence

Ḣ
q = −1 − . (S12.62)
H2
Differentiating Eq. (12.129) we get

Ḣ 3 m0 2rad0 k0

− 2
= 3
+ 4
+ 2 . (S12.63)
H0 2 a a a

Using that a(t0 ) = 1 and that k0 = 1 − m0 − rad0 −  L0 , we find that the
present value of the deceleration parameter is

m0
q0 = + rad0 − L0 . (S12.64)
2
12.6 Age—density relation for a radiation-dominated universe

Theage of a radiation-dominated universe is given by equation (S12.27) with m0 =

L0 = 0, and rad0 + k = 0, i.e.

∞
dz
t0 = t H  . (S12.65)
(1 + z) 2
rad0 (1 + z)2 + 1 − rad0
0

This gives

tH
t0 = √ . (S12.66)
1+ rad0

12.7 Redshift–luminosity relation for matter-dominated universe: Mattig’s fomula

Itfollows from 1 + z = a1 that dz = − aȧ2 dt = − Ha dt. The equation of motion

for light moving radially towards the origin is R0 adχ = −dt. Hence we obtain a
relation between the difference in redshift of to objects and the difference of their
radial coordinate,

dz
R0 dχ = . (S12.67)
H (z)
496 Solutions to the Exercises

Inserting the expression (12.111) of the Hubble parameter gives the redshift-
distance relation
   −1/2
R0 H0 dχ = rad0 1 + z 4 + m0 (1 + z)3 + k0 (1 + z)2 +  L0 dz.
(S12.68)

This formula is valid in curved as well as flat universe models. In a matter-

dominated universe it gives

z
1 dz
R0 χ = √ . (S12.69)
H0 (1 + z) 1 + m0 z
0

The so-called distance function is

D(χ ) = R0 Sk (χ ). (S12.70)

Performing the integral (S12.69) and inserting the result into equation (S12.70)
gives the distance function of a matter-dominated universe is
√ 
2 m0 z − (2 − m0 ) 1 + m0 z − 1
D(z) = . (S12.71)
H0 2m0 (1 + z)

This is called Mattig’s formula.

The luminosity distance to an object with redshift z is
√ 
2 m0 z − (2 − m0 ) 1 + m0 z − 1
d L (z) = (1 + z)D(z) = . (S12.72)
H0 2m0

This is the redshift–luminosity relation for matter-dominated universe. For the

Einstein–de Sitter universe, with m0 = 1, this relation reduces to

2c √
dL = (1 + z − 1 + z). (S12.73)
H0

12.8 Newtonian approximation with vacuum energy

(a) The time-time component of Einstein’s field equations is

G tt + gtt = 8π Gρ. (S12.74)

In orthonormal basis gtˆtˆ = −1. Hence the field equations without a cosmologi-
cal constant corresponds to those with a cosmological constant if the cosmological
constant is intepreted to represent the constant density ρL of LIVE according to
Solutions to the Exercises 497

 = −8π GρL . (S12.75)

When equations (9.15) and (9.16) of the linear field approximation are generalized
to include a cosmological constant we obtain

∇ 2 φ +  = 4π Gρ, (S12.76)

which may be written

∇ 2 φ = 4π G(ρ + 2ρL ). (S12.77)

(b) In the spherically symmetric case this equation takes the form

 
1 d 2 dφ
r = 4π G(ρ + 2ρL ). (S12.78)
r 2 dr dr

Assume there is a particle with mass m at the origin. In the space outside the
particle Eq. (S12.78) then reduces to
 
1 d 2 dφ
r = 8π Gρ L . (S12.79)
r 2 dr dr

The solution of this equation is

Gm 4
φ=− − π Gρ L r 2 . (S12.80)
r 3
The acceleration of gravity is

Gm 8
g=− 2
er + π GρL r er . (S12.81)
r 3
Hence, the acceleration of gravity vanishes at a radius
 1/3
3m
r0 = . (S12.82)
8πρL

The mass of LIVE inside this surface is

4π
mL = ρLr03 . (S12.83)
3
It follows that the mass of the LIVE inside the surface with vanishing acceleration
of gravity is only half as large as the mass at the centre, m L (r0 ) = m/2.
498 Solutions to the Exercises

As an illustration we calculate the transition radius for the Solar system in the
present Universe. Then the density of the LIVE is ρL = L ρcr , where the mass
parameter of LIVE is L = 0.7 and the critical density is ρcr ≈ 10−26 kg/m3 . The
mass of the Sun is m = 2 × 1030 kg. Inserting these quantities into Eq. (13) gives
a transition radius of approximately 300 l.y. Since the distances in the solar system
are much smaller than this, the cosmological constant is of negligible significance
for solar system gravitational effects.

12.9 Universe models with constant deceleration parameter

The deceleration parameter is

a ä
q=− . (S12.84)
ȧ 2
Integration with constant q and the boundary conditions a(0) = 0, a(t0 ) = 1
gives
 1/(1+q)
t
a= . (S12.85)
t0

In a flat universe filled by a perfect fluid with equation of state p = wρenergy–

momentum conservation implied Eq. (12.106),

ρ = ρ0 a −3(1+w) . (S12.86)

For a flat universe with vanishing cosmological constant equation (12.47) then
takes the form’

ȧ 2
3 = κρ = κρ0 a −3(1+w) . (S12.87)
a2
Integration with a(0) = 0 gives
  3(1+w)
2
t
a(t) = . (S12.88)
t0

Comparison with Eq. (S12.85) shows that the solution with constant deceleration
parameter is identical to the time evolution of a plat universe filled by a perfect fluid
with equation of state

1
p= (2q − 1)ρc2 . (S12.89)
3
12.10 Density parameters as functions of the redshift
Solutions to the Exercises 499

The density parameter in a universe with radiation matter in the form of dust, and
LIVE is
ρ κ ρ
= = = rad + m + L . (S12.90)
ρcr 3 H2

It follows from Eq. (12.106) that

ρrad = ρrad0 (1 + z)4 , ρm = ρm0 (1 + z)3 , ρL = ρL0 . (S12.91)

Combining this with Eqs. (12.111) and (S12.90) can be written

rad0 (1 + z)4 + m0 (1 + z)3 + L0

(z) = . (S12.92)
rad0 (1 + z)4 + m0 (1 + z)3 + k (1 + z)2 + L0

where

rad0 + m0 + k + L0 = 1. (S12.93)

The density parameters of radiation, matter and LIVE as functions of the redshift
are

rad0 (1 + z)4
(z) = , (S12.94)
rad0 (1 + z) + m0 (1 + z)3 + k (1 + z)2 + L0
4

m0 (1 + z)3
(z) = , (S12.95)
rad0 (1 + z) + m0 (1 + z)3 + k (1 + z)2 + L0
4

L0
(z) = . (S12.96)
rad0 (1 + z) + m0 (1 + z)3 + k (1 + z)2 + L0
4

12.11 LFRW universe with radiation and matter

We shall deduce the solutions of Einstein’s field equations for LFRW-universes with
dust and radiation, following [1]. We introduce conformal time in the same way as
in Eq. (12.126). Then Friedmann’s 1. equation takes the form
 2
da κ
= (ρm0 a + ρr 0 ) − ka 2 . (S12.97)
dη 3

This equation can be written as

 2
da
= 2αa + β 2 − ka 2 , (S12.98)
dη

where
500 Solutions to the Exercises

κ 1 κ 
α = ρm0 = m0 H02 , β = ρrad0 = H0 rad0 . (S12.99)
6 2 3

The solutions of Eq. (S12.98) with the initial condition a(0) = 0 are
For k = 1:

a = α(1 − cos η) + β sin η, t = α(η − sin η) + β(1 − cos η). (S12.100)

For k = 0:
α 2 α β
a= η + βη, t = η3 + η2 . (S12.101)
2 6 2
For k = −1:

a = α(cosh η − 1) + β sinh η, t = α(sinh η − η) + β(cosh η − 1). (S12.102)

12.12 Event horizons in de Sitter universe models

Theeventhorizon represents the barrier between the future events that can be
observed, and those that cannot. It is a spherical surface around an observer. The
event horizon sets up a limit in the future observable universe, since in the future the
observer will be able to obtain information only from events which happen inside
their event horizon. According to its definition the coordinate radius, rEH , of the event
horizon is given by

r E H ∞
dr cdt
 = . (S12.103)
1 − k(r/R0 )2 a(t)
0 t

The proper distance to the horizon is

rEH ∞
dr cdt
lEH = a(t)  =a(t) . (S12.104)
1 − k(r/R0 )2 a(t)
0 t

The scale factors of the de Sitter universe models are given in Eq. (12.160). We
have 3 cases
k = 1: In this case the scale factor is
  
R0 3 
a(t) = cosh t . (S12.105)
c  3

Inserting this into Eq. (S12.103) gives

Solutions to the Exercises 501

r E H  ∞
dr c2 3 dt
 = √ , (S12.106)
1 − (r/R0 )2 R0  cosh /3t
0 t

which leads to
   √
rEH 6c2 − /3t
arcsin = arctan e . (S12.107)
R0 R02

Hence the coordinate radius of the event horizon is

2  √ 
6c − /3t
rEH = R0 sin arctan e . (S12.108)
R02

k = 0: The flat de Sitter universe has scale factor

a(t) = e(/3)t . (S12.109)

In this case Eq. (S12.104) takes the form

r E H ∞
cdt
dr = √
/3t
, (S12.110)
e
0 t

giving
 √
rEH = c /3e− /3t . (S12.111)

k = −1: For this model the scale factor is

  
R0 3 
a(t) = sinh t . (S12.112)
c  3

Equation (S12.104) then takes the form

r E H  ∞
dr c2 3 dt
 = √ , (S12.113)
1 + (r/R0 )2 R0  sinh /3t
0 t

which leads to
  ,   -
rEH 3c2 1 
arcsinh =− ln tanh t . (S12.114)
R0 R02 2 3

Hence
502 Solutions to the Exercises
 ,   -2
3c2 1 
rEH = R0 sinh − ln tanh t . (S12.115)
R02 2 3

12.13 Flat universe model with radiation and LIVE

(a) We shall find the scale factor as function of cosmic time following the procedure
in [1]. It follows from Eq. (12.106) that in a universe containing non-interacting
radiation and LIVE the density of LIVE is constant, and the density of the
radiation decreases as the fourth power of the expansion factor. Also in a flat
universe the sum of the densities of radiation and LIVE is equal to the critical
density. Hence the total density can be written

ρ = ρL0 + ρrad0 a −4 = 1 − ρrad0 + ρrad0 a 4 (S12.116)

with the usual normalization of the scale factor, a(t0 ) = 1. Then the first
Friedmann equation takes the form
κ 
a 2 ȧ 2 = (ρcr0 − ρrad0 )a 4 + ρrad0 . (S12.117)
3
which can be written as
 
a 2 ȧ 2 = (ω/2)2 a 4 + A4 , (S12.118)

where

κ 
ω=2 (ρcr 0 − ρrad0 ) = 2H0 1 − rad0 ,
3
 1/4  1/4
ρrad0 rad0
A= = . (S12.119)
ρcr0 − ρrad0 1 − rad0

Introducing the function y = (a/A)2 Eq. (S12.118) takes the form

 
ẏ 2 = ω2 y 2 + 1 . (S12.120)

The solution with the initial condition y(0) = 0 is

y = sinh(ωt). (S12.121)

Hence the scale factor is


a(t) = A sinh(ωt). (S12.122)
Solutions to the Exercises 503

(b) The Hubble parameter and the critical density are

ω
H= coth(ωt) (S12.123)
2
and

ρcr = (3/κ)H 2 = ρL0 coth2 (ωt). (S12.124)

Hence the density parameter of LIVE and radiation is

1
L = tanh2 (ωt), rad0 = . (S12.125)
cosh(ωt)

A graph essentially of the Hubble parameter as a function of time is shown in

Fig. 12.22.
The graph shows that the Hubble parameter approaches the constant value ω/2
for ωt > 2. Also the scale factor then approaches an exponential function. Hence this
model approaches the flat de Sitter universe for large times. This is a consequence of
the fact that the density of LIVE is constant while the density of radiation decreases
during the expansion, so that the model ends up by being LIVE dominated.
(c) The age of the universe is determined from the condition a(t0 ) = 1. Equation
(S12.122) then gives


  t H 1 − rad0
t0 = (1/ω)arcsinh 1/A2 = √ arcsinh . (S12.126)
2 1 − rad0 rad0

An alternative expression follows from Eq. (S12.125),

√
tH arctanh 1 − rad0
t0 = √ . (S12.127)
2 1 − rad0

The acceleration of the scale factor is


Aω2 1
ä = sinh1/2 (ωt) 1 − coth2 (ωt) . (S12.128)
2 2

There is a transition from decelerated expansion due to the attractive gravity of

the radiation to accelerated expansion due to the repulsive gravity of LIVE at a point
of time t1 given by ä(t1 ) = 0. This leads to
√
coth(ωt1 ) = 2. (S12.129)
504 Solutions to the Exercises

Fig. 12.22 The graph of cothx

Combining this with Eq. (S12.127) gives

 √
artanh 1/ 2
t 1 = t0 √ . (S12.130)
artanh 1 − rad0

The present density parameter of the cosmic background radiation is rad0 ≈

10−4 . Inserting this in the above equation gives t1 ≈ 0.17t0 = 2.3 billion years.
Hence neglecting dust, a flat universe with LIVE and radiation equal to the CMB of
our universe, will experience a transition from decelerated expansion to accelerated
expansion 2.3 billion years after the Big bang.
It follows from Eqs. (S12.121) and (S12.129) that the observed redshift of an
object emitting the observed radiation at the point of time t1 is
Solutions to the Exercises 505

1 1 1 1 1 − rad0
z1 = −1= √ −1= √ −1= √ − 1.
a(t1 ) A sinh(ωt1 ) 2A 2 rad0
(S12.131)

Inserting rad0 ≈ 10−4 gives z 1 = 6.1.

12.14 The De Sitter Universe

(a) The formula for the cosmic redshift of light emitted at a point of time te and
received at a point of time t0 is: z 1 = a(t1e ) − 1. The scale factor of the “flat” de
Sitter universe is a(t) = e H t . This gives z 1 = e H (t0 −te ) − 1. Light moves along
a null geodetic curve, ds 2 = 0.

For light moving radially this gives dr = −ce−H t dt. Integration leads to
c  −H te  c  
r= e − e−H t0 = e−H t0 e H (te −t0 ) − 1 , (S12.132)
H H
Giving

z 1 = (H/c)e H t0 re . (S12.133)

(b) The 4-acceleration of an arbitrary particle is

 
d
u du μ μ
a = = + αβ u α u β eμ . (S12.134)
dτ dτ
μ
With the present line-element the 4-velocity of a reference particle is u μ = dxdτ =
μ
(c, 0, 0, 0) all the time. Hence dudτ
= 0. Thus the expression for the 4-acceleration
μ 2
reduces to a = tt c eμ . These Christoffel symbols are
 
μ 1 μν ∂gνt ∂gtν ∂gtt
tt = g + − ν . (S12.135)
2 ∂t ∂t ∂x

Since the metric is diagonal the first two terms vanish, and since gtt = −c2 the
last term vanish. Hence the four-acceleration of a reference particle in this reference
frame is a = 0. This means that the reference particles are freely falling. Hence an
observer with constant radial coordinate r does not experience any acceleration of
gravity.

(c) For a static metric the frequency change of light measured locally at the position
of the emitter, e, and the receiver, r, comes from the position dependence of the
rate of time in a gravitational field. Let τe be the period of light emitted at R
as measured locally by a standard clock at rest, and te the as measured by a
506 Solutions to the Exercises

coordinate clock. The corresponding quantities as measured at the receives are

τr and tr . Then

 
τe = −(gtt )e te , τr = −(gtt )r tr . (S12.136)

A static metric means that the coordinate clocks tick with a position independent
rate, te = tr . Hence the proper periods of the light at the emitter and the receiver
are related by

(gtt )r
τr = τe . (S12.137)
(gtt )e

Locally the wavelength and the period are related by λ = cτ . Hence the
gravitational redshift (for light moving upwards, blueshift downwards) is

λr (gtt )r
z= −1= − 1. (S12.138)
λe (gtt )e

With the given static form of the de Sitter metric we get

1
z2 =  − 1. (S12.139)
1 − H 2 Re2

The reason that the redshifts in (a) and (c) are different, is that the experiments are
not identical, since the emitter in (a) moves relative to the receiver, but the emitter in
(c) is at rest relative to the receiver.
(d) From point (b) we have that the 4-acceleration of a reference particle perma-
nently at rest in the coordinate system is

1 ∂gT T 1 ∂gT T 1 ∂gT T

a = TμT c2 eμ = − g μν ν
eμ = − g R R eR = − eR .
2 ∂x 2 ∂R 2g R R ∂ R
(S12.140)
   −1
Inserting gT T = − c2 − H 2 R 2 , g R R = 1 − H 2 R 2 /c2 gives
 
a = − 1 − H 2 R 2 /c2 H 2 Re R . (S12.141)

The particle accelerates in the negative R-direction. Hence the observers at rest in
this reference frame experiences an acceleration of gravity in the positive R-direction.
Solutions to the Exercises 507

(e) It follows from the given coordinate transformation that a particle with r = r0
will have a motion with

 r0 e H T
R = r0 e H T 1 − H 2 R 2 /c2 or R =  . (S12.142)
H 2r 2
1 + c2 0 e2H T

Relatively to the (T, R)-system the reference particles of the (t, r )-system have
an accelerated outwards motion.

(f) In the (t, r )-system the redshift is explained as an expansion effect, but in the
(T, R)-system the redshift is explained as a gravitational effect. The light which
moves inwards towards the observer at the origin, moves upwards in the outwards
directed gravitational field that is experienced in this reference frame, because
the reference particles are not falling freely.

12.15 The Milne Universe

(a) This line element describes an expanding, isotropic and homogeneous universe
model, a FLRW-universe. It has negative spatial curvature, and thus infinitely
large spatial extension, and the scale factor is a(t) = t/t0 .
(b) The Hubble parameter is H (t) = 1/t with a present value H (t0 ) = 1/t0 . Hence
the age of this universe is equal to the inverse of the present value of its Hubble
parameter.
(c) The deceleration parameter is q ≡ −a ä/ȧ 2 = 0. Hence this universe model
expands with a constant velocity.   2
(d) Given the coordinate transformation T = t 1 + ctr0 , R = rt0t . It follows that

 2  2
rt rt
c T −R =c t +
2 2 2 2 2
− = c2 t 2 . (S12.143)
t0 t0

Hence

t= T 2 − R 2 /c2 . (S12.144)

Furthermore it follows from the given coordinate transformation that

R r
= . (S12.145)
cT c2 t02 + r 2

Solving this equation with respect to r gives

508 Solutions to the Exercises

ct0 R
r=√ . (S12.146)
c T 2 − R2
2

Equations (d2) and (d4) are the inverse transformation.

(e) The world lines of the reference frame E in which the coordinates t, r are co-
moving are given by r = r0 =constant. It follows from Eq. (ds)√that these world
lines are given by the curves described by cT = K 1 R, K 1 = 1 + (ct0 /r0 ) in
a Minkowski diagram referring to the T, R-system. These curves are straight
lines.
According to Eq. (d1) the simultaneity curves of the t, r -system, t = a = constant,
are hyperbolae.
The world lines and simultaneity curves of the t, r -system relative to the T, R-
system are shown in the Minkowski diagram below.
12.16 Natural Inflation
(a) We here consider a natural inflation model with potential

V (φ) = V0 1 + cos φ̃ , (S12.147)

where φ̃ = φ/M, and M is the spontaneous symmetry breaking scale. Writing

the Einstein gravitational constant as κ = 1/M P2 , where M P is the Planck mass, and
inserting the potential (S12.147) into Eq. (12.304) we find the slow-roll parameters
 2
b 1 − cos φ̃ cos φ̃ MP
ε= , η = −b , b= . (S12.148)
2 1 + cos φ̃ 1 + cos φ̃ M

Here φ is the initial value of the field giving rise to N e-folds. The parameter b
represents the symmetry breaking scale and b < 1 for M > M P .
The spectral parameters are

3 − cos φ̃ 1 − cos φ̃ 1 − cos φ̃

δns = b , n T = −b , r = 8b . (S12.149)
1 + cos φ̃ 1 + cos φ̃ 1 + cos φ̃

(b) It follows from these equations that

r
r = 4(δns − b), n T = − , (S12.150)
8
The symmetry breaking parameter b can be determined from observations from
the first of the relationships (S12.150),
r
b = δns − . (S12.151)
4
Solutions to the Exercises 509

The observed value δns = 0.032 and r ≥ 0 gives b < 0.02 and hence M ≥ 5.6M P .
This shows that the symmetry breaking energy is higher than the Planck energy which
is a weakness of the original natural inflation model.
(c) Inserting the potential (S12.147) into equation (12.322) and performing the
integration gives the number of e-folds


1 1 − cos φ̃ f
N = ln  (S12.152)
b 1 − cos φ̃

It is usual to specify that the inflationary era ends when ε = 1. Inserting φ = φ f

in the first of the Eq. (S12.148) with ε = 1 we get
 2−b
cos φ̃ f = − . (S12.153)
2+b

Inserting this into Eq. (S12.152) gives

 4 −bN
cos φ̃ = 1 − e . (S12.154)
2+b

It follows from Eqs. (6.5.22) and (6.5.29) that

(2 + b)ebN + 2 2b 16b
δns = b , nT = − ,r = .
(2 + b)e − 2
bN (2 + b)e − 2
bN (2 + b)ebN − 2
(S12.155)

Inserting δns = 0.032 and N = 50 into equation the first of the expressions
(S12.155) we get b = 0.02 and bN = 1, and the last of the equations gives r ≈ 0.09
which is a little larger than the observational requirement r < 0.04.
Index

A 1st identity, 185, 186

Absolute space, 18 component form, 185
Accelerated expansion, 321 2nd identity, 186, 201
Accelerated frame, 108, 109, 112 Bianchi type-I, 351
Accelerated frame hyperbolically, 108 Big Bang, 275
Accelerated reference frame, 89, 112 Big Bang cosmology, 361
4-acceleration, 272 Big Rip, 351
Acceleration in a rotating reference frame, Binary system, 265, 266
142 Black body radiation, 44
Acceleration of gravity, 14, 16, 135, 192, Black hole, 221, 222, 263, 271, 273, 476
194, 271, 283 horizon, 271
Acceleration of particle in non-rotating radiation, 273
frame, 142 temperature, 273
Acceleration of particle in rotating frame, Blueshift, 56
142 Boltzmann’s constant, 273, 363
Acceleration scalar, 272 Boost, 35
Action, 127 Born-rigid motion, 108
Adiabatic, 327 Born-stiff motion, 108
Age-density relation, 384, 496 Boyer-Lindquist coordinates, 276, 280, 299
Age of the universe, 311, 315, 383, 489 Boyer-Lindquist coordinate system, 277,
Age-redshift relationship, 329, 341 281, 480, 482
Ampere’s circuit law, 54
Brill, 19, 377
Angular acceleration, 95
Angular momentum, 228
Anisotropic universe models, 351
Anisotropy parameter, 352 C
Antisymmetric tensor, 78, 86, 119, 426 Canonical momentum, 128
Antisymmetrization, 80 Cartan
Archimedean spiral, 99 surface curvature, 183
Atmospheric mesons, 421 Cartan connection, 147, 150
Cartan formalism, 183, 211
Cartan’s structure equations, 178, 395
B Cartan structure equations 1st , 148, 149, 178,
Basis forms, 178 185, 212, 395, 456
Basis vector, 71, 147 Cartan structure equations 2nd , 178, 179,
Bergmann, 393 185, 186, 212, 397, 457
Bianchi Cartesian basis, 169
© Springer Nature Switzerland AG 2020 511
Ø. Grøn, Introduction to Einstein’s Theory of Relativity,
Undergraduate Texts in Physics, https://doi.org/10.1007/978-3-030-43862-3
512 Index

Cartesian coordinates, 47, 73, 75, 155, 164, Coordinate basis connection coefficients,
169 123
Cartesian coordinate system, 23, 34, 49, 61, Coordinate basis forms, 161
66, 67, 115, 116, 123, 140, 152, 177, Coordinate basis vector, 64, 66, 67, 141, 161,
198, 444 179
Causal, 76 Coordinate clocks, 24, 94, 109, 141, 216
Causal mass, 378, 380 Coordinate cyclic, 128, 129
CDM universe, 380 Coordinate Gaussian, 123
Centrifugal acceleration, 143 Coordinate singularity, 239
Centrifugal barrier, 230 Coordinate system, 23, 64
Centrifugal force, 19, 95 Coordinate time, 24, 98, 109, 112
Centripetal acceleration, 99, 105 Coordinate transformation, 65, 87, 91, 93
Chaotic inflation, 375 Coordinate vector, 62
Cherenkov radiation, 57, 422 Coriolis acceleration, 143, 193
Chirp, 260, 267 Cosmic fluids, 317
Chirp mass, 263, 264 Cosmic Microwave Background (CMB),
Christoffel, 122 372
Christoffel symbols, 122, 130, 135, 141, 142, Cosmic microwave background radiation,
146, 155, 171, 177, 189, 190, 192, 372
195, 215, 244, 272, 282, 402, 449, Cosmic red-shift, 323, 326
459, 475 Cosmic time, 311
Christoffel symbols in coordinate basis, 123 Cosmological constant, 204, 242, 318, 367,
Christoffel symbols physical interpretation, 462, 471, 484
215 Cosmological horizon, 359
Christoffel symbols rotating reference Cosmological red-shift, 315–317
frame, 141 Cosmology, 311
Classical electron radius, 300 Covariance principle, 17, 122
Clock, 78, 98 Covariant, 18
Codifferential, 157, 166 Covariant components, 86
Cohen, 19, 377 Covariant derivative, 119, 122, 144–146,
Coincidence problem, 348 153, 189
Commutator, 68, 143, 149, 155 Covariant differentiation, 119
Co-moving coordinate system, 64, 311 Covariant directional derivative, 124, 140,
Conformally flat, 295, 296 144, 145, 153, 173, 188, 453
Conformal time, 331, 383, 489 Covariant equation, 72
Connection coefficients, 123, 140, 142, 144, Covariant Euler-Lagrange equations, 127,
145, 148, 176, 188 128
Connection coefficients in coordinate basis, Covariant tensor, 70
123 Critical density, 345
Connection coefficients rotating reference Curl, 157, 160–162, 445
frame, 140 Current form, 166
Connection forms, 147, 177, 183, 212, 457 Curvature, 173, 195, 456
Connections coefficients, 177 forms
Conservation energy-momentum, 197 of surface, 183
Conservation of mass-energy, 198 Gaussian, 183, 184
Constants of motion, 128, 156, 171, 453 geodesic, 181
Continuity equation, 197, 198, 366 normal, 181
Contraction, 69, 86, 149, 168, 177 principal, 182
Contractions of tensors, 427 Ricci, 177
Contravariant components, 74 Riemann tensor, 175
Contravariant tensor, 70 surface (Cartan formalism), 183
Convective derivative, 151, 152 Curvature coordinates, 211, 216, 220, 222,
Coordinate basis, 120, 143, 150 226
Index 513

Curvature forms, 177, 212, 457 Eddington–Finkelstein coordinates, 216,

Curvature of space, 196 219
Curvature scalar, 457 Eddington-Finkelstein coordinate system,
Cyclic coordinate, 128, 228, 250 220
Cylinder conditions, 394 Effective potential, 230
Cylinder coordinates, 93, 225 Ehrenfest’s paradox, 89
Cylinder dimension, 404 Einstein, 15, 17, 19, 23, 31, 44, 45, 52, 99,
197, 201, 393
Einstein–de Sitter, 333
D Einstein–de Sitter model, 342
D’Alembertian wave operator, 162, 164 Einstein–de Sitter universe, 378, 379, 490
D’Alembert’s wave operator, 157, 244 Einstein field equations, 202, 241
Deceleration parameter, 324, 342, 385, 494, Einstein’s constant of gravity, 202
498 Einstein’s curvature tensor, 201, 211, 285,
Deflection of light, 211, 236 484
Density parameter, 322, 345, 386, 499 Einstein’s field equations, 197, 201, 202,
de Sitter spacetime, 241, 321, 387, 471 206–208, 238, 243, 244–247, 255,
de Sitter universe, 505 275, 284, 285, 295, 326, 462, 497
de Sitter universe models, 386, 501 Einstein spaces, 204
Determinant, 81 Einstein’s theory, 211
Differential forms, 211 Einstein synchronization, 92, 101
curvature Einstein synchronized, 25
of surface, 183 Einstein tensor, 202, 211, 213, 284, 469
Differential forms connection, 212 Electrical field, 53
Differential forms curvature, 212
Electric field, 209
Differential forms, directional derivatives of,
Electric field strength, 163
145
Electricity, 51
Differential geometry, 179
Electric scalar potential, 163
Differential operator, 66, 140, 169, 444
Electromagnetic current form, 165
Differentiation covariant, 124
Electromagnetic field, 51, 167, 168, 201, 207
Differentiation of forms, 119
Electromagnetic field form, 163, 166
Dimension, 59
Electromagnetic field tensor, 207
Dirac delta function, 6
Electromagnetic potential form, 162
Directional derivative, 66, 140, 153, 156, 179
Distance coordinate, 211 Electromagnetic vector potential, 163, 207
Distance physical, 211 Electromagnetic waves, 33, 167
Divergence, 157, 162, 167, 170, 198, 445 Electromagnetism, 162
D’Lambertian wave operator, 159 Ellipse, 266
Domain wall, 309, 485 Embedding, 225
Doppler effect, 26, 311, 326, 418, 477 Embedding of the Schwarzschild metric, 225
Dragging angular velocity, 251, 278 Emission point of time, 329
Dual, 82, 83, 160, 161, 167 Empty space, 204
Dual forms, 84, 169, 442 Energy, 44, 47
Dust, 200, 205, 317, 321, 328, 348 Energy conservation, 168
Dust dominated universe, 331 Energy-momentum conservation, 197, 208,
Dust-dominated universe model, 331, 333 326
Energy-momentum tensor, 197–200, 205,
207, 209, 210, 238, 246, 247, 460
E Eötvös, 15
Earth, 12, 20, 33, 34, 117, 226, 227, 232, Equation of continuity, 167, 210, 460
253, 264, 412, 413 Equation of geodesic deviation, 189, 196
Earth-Moon system, 21, 413, 414 Equation of state, 462
Eccentricity, 266 Equatorial plane, 229
Eddington, 255 Euclidean geometry, 94
514 Index

Euclidean plane, 87, 155, 428 G

Euclidean space, 169, 442 Galilean kinematics, 23
Euclidean 3-space, 82, 161 Galilei, 14
Euclidean spatial geometry, 323, 361 Galilei transformation, 18, 35
Euler equation, 240 Galileo, 1
Euler equation of motion, 198, 199, 210, 460 Gamma ray burst, 269
Euler–Lagrange equation, 119, 128, 131, Gauge transformation, 164, 394
135 Gauss
Event, 23, 77 curvature, 183
Event horizons, 386, 501 theorema egregium, 184
Expansion, 153 Gauss’ equation, 180
Exterior derivative, 119–121, 147, 150, 157– Gaussian coordinates, 123
161, 163, 165, 166, 178 Gaussian curvature, 183, 184, 457
Exterior differentiation, 150 Gauss integral theorem, 6
Exterior Schwarzschild metric, 215 Gauss’ law, 410
Extrinsic curvature tensor, 298 Gauss’s integral theorem, 246
General principle of relativity, 17
General theory of relativity, 1, 16, 19, 62,
F 95, 119, 122, 123, 215, 232, 233, 243,
Fabry-Pérot cavities, 260 266, 290, 305
False vacuum, 362 Geodesic
Fermi gamma ray space telescope, 269 curvature, 181
Field equations Einstein, 202 Geodesic curve, 78, 125, 130, 156, 181, 206,
Field equations vacuum, 204 448
First order phase transition, 363 Geodesic deviation, 188, 257
Flamm paraboloid, 226 Geodesic equation, 125, 171, 312, 453
Flatness problem, 360, 365 Geodesic normal coordinates, 189
Flat spacetime, 76 Geodesic postulate, 204
Flat universe model, 319 Geodesic world lines, 133
Force, 47–49 Geometrical optics, 118, 439
Form-components, 80 GPS, 242, 473
Form-invariant, 18 GPS-satellite, 474
Forms, 59, 78, 119 Gradient, 157, 159, 161, 170, 444
Four-acceleration, 62 Gravitation, 18
Four momentum, 61 Gravitational acceleration, 214
Four-vectors, 85, 424 Tolman-Whittaker expression, 284
Four-velocity, 60, 65, 76 Gravitational antenna, 258
Four-velocity identity, 76 Gravitational collapse, 382, 486
Free fall, 15, 217 Gravitational field, 4, 16, 95, 112, 118, 138,
Freely falling reference frame, 15 192, 243, 438
Free particle, 130, 135 Gravitational force, 412
Frequency, 116 Gravitational frequency shift, 211
Friedman-Lemaître Eq., 328 Gravitational mass, 283
Friedmann–Lemaître, 338 Gravitational mass–density, 321
Friedmann–Lemaître model, 342, 348 Gravitational potential, 193, 246, 411
Friedmann–Lemaître universe model, 347 Gravitational radiation emission, 266
Friedmann equation, 366, 488 Gravitational redshift, 140
Friedmann-Lemaître equations, 319, 331, Gravitational self energy, 295
332 Gravitational shift of wavelength, 326
Friedmann metric, 488 Gravitational time dilation, 89, 98, 99, 112,
Friedmann models, 360, 362, 365 134
Friedmann’s 1st equation, 500 Gravitational wave detector, 264
Index 515

Gravitational waves, 243, 245, 253–257, Inertial reference frame, 16, 40, 41, 64, 123,
260, 263, 266 133, 140
Gravito-electric’ field, 252 Inertial rest frame, 116
Gravitoelectromagnetic fields, 253 Infalling coordinates, 382
Gravitoelectromagnetism, 251 Inflation, 363
Gravitomagnetic clock effect, 281, 483 Inflationary era, 364, 366, 372, 377
Gravitomagnetic field, 252 Inflationary model, 363
Gravitomagnetic vector potential, 251 Inflationary universe models, 361
Gravitomagnetism, 243 Inflaton field, 366, 367, 375
Gravity, 119 Inflaton potential, 368
Gravity probe B, 253 Inhomogeneous universe models, 355
Guth, Alan, 363 Instantaneous rest frame, 92
INTEGRAL telescope, 269
Interference, 99
H Interference fringes, 99
Hafele-Keating experiment, 232 Interferometer, 260, 264
Hamiltons principle, 127 Internal coordinate transformation, 91
Harrison-Zel’dovich spectrum, 373 Internal Schwarzschild solution, 283
Hawking, 273 Internal Schwarzschild space-time, 289
Hawking radiation, 271, 273 Interval, 37, 40, 77, 78
Heat capacity of black hole, 274 Invariant, 72, 73
Heisenberg uncertainty relationships, 200 Invariant basis, 155, 156
Higgs field, 361, 362 Invariant equation, 72
Higgs mechanism, 361 Invariant interval, 116
Hilbert, 204, 206 Isotropic coordinates, 241, 468
Hilbert’s variational principle, 206 Israel formalism, 283, 290, 291, 309, 485
Hodge Laplacian, 158
Horizon, 113, 220, 222, 271, 277, 380, 471
Horizon problem, 358, 364 J
Horizon radius, 379 Jupiter, 21
Hubble age, 315, 334, 340, 342, 359, 360,
379, 490, 494
Hubble constant, 314, 320, 322, 339 K
Hubble flow, 305, 308, 315 Kaluza-Klein theory, 206, 393, 400, 404
Hubble-Lemaître expansion law, 311 Kasner universe, 353
Hubble-Lemaître law, 314, 315 Kepler motion, 240, 463, 465
Hubble length, 315, 323 Kepler’s 2nd law, 235
Hubble parameter, 314, 320, 322, 342, 348, Kepler’s 3rd law, 236, 263, 484
350, 367, 368, 383, 489, 494 Kerr, 275
Hubble slow roll parameters, 368 Kerr black hole, 278, 280
Hydrostatic equilibrium, 8 Kerr metric, 275, 278
Hyperbolically accelerated frame, 107 Kerr-Newman spacetime, 299, 302, 304
Hyperbolic motion, 41–43 Kerr solution, 276
Kerr spacetime, 271, 275, 277, 280, 281,
283, 479, 481, 483
I Killing equation, 154–156
Incompressible star, 7, 288 Killing vectors, 154–157
Inertial dragging, 243, 250, 253, 271, 277, Kinetic energy, 46
377, 378, 405 Klein, 404
Inertial dragging angular velocity, 377 Koszul connection coefficient, 140
Inertial effects, 193 Kretschmann curvature scalar, 216, 297
Inertial forces, 19 Kruskal–Szekeres, 216, 222, 224
Inertial frame, 16, 47, 93, 108, 123, 198, 250 Kruskal–Szekeres line element, 224
516 Index

L Line element free particle, 129

LAGEOS satellites, 253 LIVE, 200, 210, 242, 285, 306, 311, 318,
Lagrange density, 204 321, 328, 336, 338, 339, 344, 345,
Lagrange dynamics, 463 348, 353, 354, 357, 358, 366, 374,
Lagrange equation, 128, 129, 228, 464 386, 460, 462, 492, 497, 502
Lagrange equations free particle, 129 LIVE-Cold-Dark-Matter, 338
Lagrange function, 127–129, 131, 171, 206, Local derivative, 151
250, 276, 403, 463, 476 Local inertial frame, 215
Lagrange function free particles, 129 Local inertial reference frame, 135
Lagrange multiplicator, 181 Locally Cartesian coordinates, 192
Lagrangian, 206, 306 Locally Cartesian coordinate system, 186
Lagrangian function, 217, 447 Lookback time, 329, 335, 383, 489
Lambda-Cold-Dark-Matter, 338 Lopez, 300
Laplace–Beltrami operator, 158, 167 Lopez’s source of the Kerr-Newman metric,
Laplace equation, 5 303
Laplacian, 157, 158, 162, 170 Lorentz contracted, 95
LCDM-universe, 357, 358 Lorentz contraction, 33, 35, 56, 95, 96, 108,
Leads to an Einstein–de Sitter universe, 333 420
Leibowitz, 401 Lorentz factor, 30
Leibowitz-Rosen tensor, 401 Lorentz force, 252
Lemaître, 200, 321 Lorentz invariance, 200
Lemaître-Friedmann-Robertson-Walker Lorentz invariant, 37, 47, 75, 460
universe models, 311 Lorentz invariant vacuum energy, 200, 311,
Lemaître-Tolman-Bondi, 355 318, 338
Length, 35 Lorentz’s force law, 52
Length contraction, 33 Lorentz transformation, 34–37, 47, 58, 92,
Lense, 19 95, 109, 423, 425, 460
Lense-Thirring effect, 19, 250 Lorenz condition, 164
Lever paradox, 49 Lorenz gauge, 164
Levi-Civita, 124, 295 Lorenz gauge condition, 252, 254
Levi-Civita Bertotti-Robinson metric, 283 Lowering an index, 74
Levi-Civita Bertotti Robinson solution, 295 Luminosity distance, 385
Levi-Civita Bertotti Robinson spacetime, Lyth bound, 371
297
Levi-Civita connection, 123
Levi-Civita symbol, 80, 251 M
LFRW cosmological models, 330 Mach, 19
LFRW universe, 500 Mach’s principle, 18
LFRW universe models, 325, 348, 384, 493 Magnetic flux density, 163
Lie derivative, 155, 156 Magnetism, 51
Light, 33 Manifestly covariant, 18
Light cone, 24, 218 Mass, 44, 45
Light cone coordinates, 223 Material derivative, 151
Light deceleration of, 226 Matrix, 81
Light deflection of, 236 Matter-dominated universe, 385
Light-like, 37, 39, 78 Mattig’s formula, 385, 496
LIGO, 243, 260, 263, 269 Maximally linearly independent, 59
Linear combination, 59 Maximally symmetric, 155
Linear field approximation, 243, 252 Maxwell, 33
Linearized field equations, 246 Maxwell’s equations, 18, 51, 162, 252, 424
Linearly independent, 59 Maxwell source equations, 166
Line element, 40, 75, 83, 93, 94, 98, 101, Maxwell’s source free equations, 165, 208
104, 161 Measuring rods, 97
Index 517

Mechanical energy, 266 Newtonian gravitational potential, 190, 251

Mercury, 227, 241, 471 Newtonian gravity, 230, 289
Mercury’s orbit, 236 Newtonian hydrodynamics, 151, 198
Mercury’s perihelion precession, 233 Newtonian law of gravitation, 203
Meson, 56 Newtonian limit, 135, 191, 203, 214, 325,
Metric, 72 459, 470
Kerr (rotating), 275 Newtonian physics, 190
Metric covariant differentiation, 146 Newtonian potential, 20, 21, 203, 230, 410
Metric perturbations, 257 Newtonian theory, 265
Metric static, stationary, 129 Newtonian tidal tensor, 191
Metric tensor, 72, 74, 81, 82, 94, 146, 200, Newton’s 1st law, 16, 64, 135, 192
216, 232 Newton’s 2nd law, 14, 57, 412, 423
Michelson and Morley, 33 Newton’s law of gravitation, 1, 4, 201, 409
Michelson interferometers, 260 Newton’s law of gravity, 411
Michelson–Morley experiment, 33 Newton’s theory, 16, 95, 119, 193, 230
Milky Way, 107, 108 Newton’s theory of gravitation, 1, 192
Milne universe, 353, 383, 388, 490, 507 Non-Euclidean, 94
Milne universe model, 319 Non-tidal gravitational field, 195
Mini-black hole, 275 Normal curvature, 181
Minkowski diagram, 23, 24, 30, 42, 102, Null-geodesic curve, 17, 359
106, 133, 388, 433 Number of e-folds, 370
Minkowski-force, 62
Minkowski force form, 168
Minkowski line-element, 76 O
Minkowski metric, 76, 86, 148, 210, 243, One-form basis, 69
317, 433, 459 Orbit equation, 231
Minkowski space, 116 Orthogonal basis, 82
Minkowski spacetime, 75, 78, 133, 166, 214, Orthonormal basis, 64, 67, 76, 102, 104, 109,
223, 244, 277, 295, 296, 300, 319, 142, 150, 160, 211, 212, 239
320, 377, 460 Orthonormal basis field, 148, 150
Outer product, 120
Minkowski space-time from Schwarzschild
solution, 214
Minkowski universe, 353
P
Mixed components, 86
Parallel transport, 124, 174, 195, 453
Mixed metric tensor, 74 Partial derivative, 122, 144
Mixed tensor, 70, 71 Particle horizon, 359, 379, 380, 494, 495
Møller coordinates, 108 Particle trajectories, 228
Momentum, 47, 48 Particle trajectory Schwarzschild 3-space,
Momentum-energy tensor, 198 228
Moon, 12, 21, 413 Past light cone, 378
Multilinear function, 70 Peculiar velocity, 315
Multi messenger astronomy., 269 Perfect dragging, 378
Perfect fluid, 199, 210, 317, 459, 493
Perfect inertial dragging, 341, 379
N Perihelion precession, 471
Natural Inflation, 389, 508 Perihelion precession of Mercury, 242
Newton, 1, 18 Perihelion shift of Mercury, 211
Newton fluid, 198 Permeability of empty space, 166, 207
Newtonian 3-velocity, 151 Permeability of vacuum, 54
Newtonian approximation, 190, 210, 215, Permittivity of empty space, 166
459, 497 P-form, 78
Newtonian fluid, 197, 198 Phantom energy, 348, 350
Newtonian force, 16 Photon, 99
518 Index

Photon clock, 28 Kerr spacetime, 276

Photon sphere, 242, 474 Redshift cosmological, 315
Physical components, 160 Redshift–luminosity relation, 385, 496
Physical laws, 18 Reduced mass, 263
Physical singularity, 216 Reference frame, 16, 17, 64
Planck length, 240 Reference frame inertial, 123
Planck mass, 274, 375 Reference particle, 94
Planck spectrum, 273 Reinterpretation principle, 50
Planck time, 275 Reissner-Nordström solution, 211, 239, 406
Plane polar coordinates, 66, 69, 73, 75, 123, Reissner-Nordström spacetime, 238
140, 150 Relativistic Doppler effect, 67, 421
Poincaré’s lemma, 121, 158, 160, 162, 164, Relativistic Doppler shift, 56
165, 167, 178, 185, 395 Relativistic gravitational potential, 325
Poisson equation, 7, 385 Relativistic gravity, 230
Polarization, 256, 257, 372 Relativistic rotating disc, 115, 429
Polynomial inflation, 375 Relativity of simultaneity, 30, 32, 50, 92, 93,
Pound-Rebka experiment, 140 96
Power spectra, 373 Repulsive gravitation, 285, 321
Precession, 235 Rest acceleration, 42, 43
Precession of Mercury’s perihelion, 233 Rest frame, 61
Pressure, 44 Rest length, 33
Pressure forces, 8 Ricci
Primordial black hole, 275 curvature, 177
Principal curvatures, 182 identity, 184
Principle of equivalence, 1, 15, 16, 95, 192, component form, 184
215 Ricci curvature scalar, 177, 179, 207, 213,
Principle of general relativity, 136 243
Principle of relativity, 1, 17, 18, 41, 95, 380
Ricci curvature tensor, 177, 179, 213, 243,
Projectile motion, 171
457
Projection, 89, 153
Ricci curvature tensor, divergence, 201
Projection tensor, 113, 114, 153
Ricci identity, 184
Proper acceleration, 105, 107, 114
Ricci scalar, 244
Proper distance, 92
Ricci tensor, 201
Proper time, 30, 38, 78, 98, 112, 116, 127,
Riemann curvature forms, 179
134, 216
Riemann curvature tensor, 173, 175–178,
Proper time-interval, 40, 106, 232
184, 186, 190, 195, 213, 243, 457
Proxima Centauri, 137
Riemannian space, 188
Riemann tensor, 175–177, 179, 184, 189,
Q 255
Quasars, 417 Rigid rotation, 104
Rindler coordinates, 108
River model of space, 283
R River of space, 306
Radar method, 25 Robb, 54
Radially moving photons, 219 Robb’s formula, 417
Radiation, 45, 200, 317, 321, 328 Robertson–Walker line element, 314
Radiation dominated universe, 330, 361, Robertson–Walker metric, 312, 359
384, 496 Robertson–Walker universe, 383
Radiation-dominated universe model, 330 Roche limit, 21, 416, 417
Radiation pressure, 44 Rosen, 401
Rank, 70, 71 Rosser, 47
Rapidity, 36, 109, 110 Rotating black hole, 275
Redshift, 56 Rotating frame, 94–96, 101, 102, 141
Index 519

Rotating frame of reference, 170, 447 Simultaneity space, 91, 102, 104, 115, 153
Rotating motion, 95 Simultaneous events, 93
Rotating reference frame, 92, 99, 101, 195, Singularity coordinate, 216
447, 456 Singularity physical, 216
Rotating reference frame connection coeffi- Singular mass shells, 290
cients, 140 Slow roll approximation, 367
Rotating spherical shell, 247 Slow roll era, 374
Rotation, 152, 153 Slow roll parameters, 368, 375
Rotational motion, 380 Space, 42
Running of the spectral indices, 373 Space-like, 37, 78
Space ship, 117
Spacetime, 16, 61, 64, 65, 247
S Spacetime curvature, 17, 190
Sagnac effect, 99, 280 Spacetime interval, 37
Sagnac experiment, 89 Spatial geodesics, 447
Scalar field, 318, 444 Spatial geometry, 89, 95
Scalar perturbation, 372 Spatial line-element, 90, 94, 114
Scalar product, 72 Spatial metric, 102, 114
Scalar quantity, 70 Spatial metric tensor, 89, 91, 94, 114
Scale factor, 311, 317, 379 Spatial simultaneity space, 456
Scale invariant, 373 Special principle of relativity, 17
Schwarzchild spacetime, 223, 272 Special theory of relativity, 23, 25, 41, 98,
Schwarzschild 3-space, 228 200
Schwarzschild 3-space energy, 228 Spectral indices, 373
Schwarzschild 3-space momentum, 228 Spectral parameters, 509
Schwarzschild black hole, 273, 279, 476 Spectral tilt, 375
Schwarzschild-de Sitter metric, 241, 309, Spherical coordinates, 161, 169, 214, 247,
469, 484 444
Schwarzschild-de Sitter spacetime, 306, 485 Spherical coordinate system, 82
Schwarzschild horizon, 241, 308 Spherically symmetric spacetime, 209
Schwarzschild light cones, 218 Spontaneous symmetry breaking, 361, 362
Schwarzschild line-element, 223, 475, 487 Standard clock, 98, 112, 116, 134, 217
Schwarzschild metric, 215, 216, 225, 239, Standard measuring rod, 95, 96
241, 293 Static border, 278
Schwarzschild metric embedding, 225 Static metric, 129, 275
Schwarzschild radius, 3, 19, 216, 218, 222, Stationary metric, 129, 275
226, 227, 236, 238, 243, 264, 268, Stefan–Boltzmann law, 274, 330, 364
271, 278, 279, 293, 299, 378–382, Structure coefficients, 68, 143, 148, 176, 188
469, 474, 486, 488 Sun, 226, 227, 236, 238, 264, 269, 274
Schwarzschild’s exterior solution, 211 Surface curvature, 183
Schwarzschild’s interior solution, 287 Surface gravity, 271, 273
Schwarzschild solution, 211, 241, 468, 469 Synchronization, 25
Schwarzschild space, 241
Schwarzschild spacetime, 211, 216–218,
225, 238, 247, 250, 272, 468, 479, T
481, 487 Tachyons, 50
Semi-latus rectum, 266 Tachyon telephone paradox, 51
Shapiro, 226 Tangent vector, 66
Shapiro experiment, 226 Tangent vector field, 62, 64, 125
Shear, 152, 153 Taylor expansion, 193
Shoemaker-Levy, 21 Tensor, 59, 69, 78, 146
Shoemaker-Levy 9, 417 contraction of, 177
Simultaneity, 101 Ricci curvature, 177
520 Index

Riemann curvature, 175 Uniformly accelerated space ship, 436

Tensor covariant differentiation of, 146 Uniformly accelerated system of reference,
Tensor Einstein, 202 433
Tensor equation, 18, 72 Unit vector, 67
Tensor momentum energy, 198 Universe, 19
Tensor perturbation, 373 Universe model dust dominated, 331
Tensor product, 70, 86, 426 Universe model inflationary, 363
Tensor Riemann curvature, 213 Universe model radiation dominated, 330
Tensor-to-scalar ratio, 373
Tetrad, 102
The de Sitter universe models, 336 V
The Friedmann–Lemaître model, 337 Vacuum energy, 200
The Hubble parameter, 503 Vacuum field equations, 204
Theorema egregium, 184 Variational principle, 130, 131, 204, 206
Theory of relativity, 95 Vector, 18, 59
The river of space, 305 Vector field, 64
The static border, 278 Vectorial form, 147
The twin paradox, 54, 417 Vector perturbation, 372
Thin dust shell, 291 Velocity, 47
Thirring, 19, 377 Velocity addition, 36
Tidal acceleration, 190 Velocity field, 151
Tidal field, 14 4-velocity identity, 129
Tidal force, 1, 10, 16, 173, 193 Velocity of light, 24
pendulum Venus, 226
relativistic, 196 Vertical free fall, 131
Tidal force pendulum, 19, 196, 409, 410, 458 Vertical projectile motion, 451
Tidal gravitational field, 194 Virgo, 263, 269
Tidal potential, 14 Volume-element, 82
Tilt, 373 Volume expansion, 152
Time dilation, 28, 30, 38, 39, 56, 98, 134, Volume form, 80, 82, 83
420 Vorticity, 372
Time-like, 37, 76
Tolman–Oppenheimer–Volkoff (TOV)
equation, 285, 287, 288
W
Tolman–Whittaker expression, 284
Wave number, 373
Tolman-Whittaker formula, 299
Weak gravitational field, 216
Torque, 49
Wedge product, 79, 84
Torsion, 187
Weingarten’s equation, 181
Torsion form, 187
Wheeler, 204
Total derivative, 151, 153
White hole, 221
Transformation matrix, 81
Wien’s displacement law, 273
Transverse traceless gauge condition, 256
World line, 24, 38, 42, 76
Twin paradox, 40, 41, 136, 138

U Z
Uniformly accelerated motion, 42, 43, 114 Zel’dovich fluid, 366
Uniformly accelerated reference frame, 43, Zero-Angular-Momentum-Observers
89, 105, 107, 112, 116, 131, 139, 171, (ZAMO), 250, 276, 277, 280, 377,
449 480