EXACT SOLUTIONS IN THREE-DIMENSIONAL

GRAVITY

This self-contained text systematically presents the determination and classifica-

tion of exact solutions in three-dimensional Einstein gravity. The book explores
the theoretical framework and general physical and geometrical characteristics
of each class of solutions, and includes information on the researchers responsible
for their discovery. Beginning with the physical character of the solutions, they
are identified and ordered on the basis of their geometrically invariant proper-
ties, their symmetries, and their algebraic classifications, or from the standpoint
of their physical nature; for example, electrodynamic fields, fluid, scalar field,
or dilaton. Consequently, this text serves as a thorough catalogue of (2 + 1)-
exact solutions to the Einstein equations coupled to matter and fields, and to
the vacuum solutions of topologically massive gravity with a cosmological con-
stant. The solutions are also examined from different perspectives, building a
conceptual bridge between exact solutions of three- and four-dimensional gravi-
ties and thus providing graduates and researchers with an invaluable resource in
this important topic in gravitational physics.

A l b e rt o A . G a rc ı́ a - D ı́ a z is Emeritus Professor at the Center for

Research and Advanced Studies of the National Polytechnic Institute, Mexico
(CINVESTAV-IPN). His research throughout his career has focused on algebraic
classification in four-dimensional gravity, nonlinear electrodynamics and dilaton
fields.
 CAMBRIDGE MONOGRAPHS ON MATHEMATICAL PHYSICS

General Editors: P. V. Landshoff, D. R. Nelson, S. Weinberg

S. J. Aarseth Gravitational N-Body Simulations: Tools and Algorithms †

J. Ambjørn, B. Durhuus and T. Jonsson Quantum Geometry: A Statistical Field Theory Approach †
A. M. Anile Relativistic Fluids and Magneto-fluids: With Applications in Astrophysics and Plasma
Physics
J. A. de Azcárraga and J. M. Izquierdo Lie Groups, Lie Algebras, Cohomology and Some
Applications in Physics †
O. Babelon, D. Bernard and M. Talon Introduction to Classical Integrable Systems †
F. Bastianelli and P. van Nieuwenhuizen Path Integrals and Anomalies in Curved Space †
D. Baumann and L. McAllister Inflation and String Theory
V. Belinski and E. Verdaguer Gravitational Solitons †
J. Bernstein Kinetic Theory in the Expanding Universe †
G. F. Bertsch and R. A. Broglia Oscillations in Finite Quantum Systems †
N. D. Birrell and P. C. W. Davies Quantum Fields in Curved Space †
K. Bolejko, A. Krasiński, C. Hellaby and M-N. Célérier Structures in the Universe by Exact Methods:
Formation, Evolution, Interactions
D. M. Brink Semi-Classical Methods for Nucleus-Nucleus Scattering †
M. Burgess Classical Covariant Fields †
E. A. Calzetta and B.-L. B. Hu Nonequilibrium Quantum Field Theory
S. Carlip Quantum Gravity in 2+1 Dimensions †
P. Cartier and C. DeWitt-Morette Functional Integration: Action and Symmetries †
J. C. Collins Renormalization: An Introduction to Renormalization, the Renormalization Group
and the Operator-Product Expansion †
P. D. B. Collins An Introduction to Regge Theory and High Energy Physics †
M. Creutz Quarks, Gluons and Lattices †
P. D. D’Eath Supersymmetric Quantum Cosmology †
J. Dereziński and C. Gérard Mathematics of Quantization and Quantum Fields
F. de Felice and D. Bini Classical Measurements in Curved Space-Times
F. de Felice and C. J. S Clarke Relativity on Curved Manifolds †
B. DeWitt Supermanifolds, 2nd edition †
P. G. O. Freund Introduction to Supersymmetry †
F. G. Friedlander The Wave Equation on a Curved Space-Time †
J. L. Friedman and N. Stergioulas Rotating Relativistic Stars
Y. Frishman and J. Sonnenschein Non-Perturbative Field Theory: From Two Dimensional
Conformal Field Theory to QCD in Four Dimensions
J. A. Fuchs Affine Lie Algebras and Quantum Groups: An Introduction, with Applications in
Conformal Field Theory †
J. Fuchs and C. Schweigert Symmetries, Lie Algebras and Representations: A Graduate Course
for Physicists †
Y. Fujii and K. Maeda The Scalar-Tensor Theory of Gravitation †
J. A. H. Futterman, F. A. Handler, R. A. Matzner Scattering from Black Holes †
A. S. Galperin, E. A. Ivanov, V. I. Ogievetsky and E. S. Sokatchev Harmonic Superspace †
R. Gambini and J. Pullin Loops, Knots, Gauge Theories and Quantum Gravity †
T. Gannon Moonshine beyond the Monster: The Bridge Connecting Algebra, Modular Forms and
Physics †
A. Garcı́a-Dı́az Exact Solutions in Three-Dimensional Gravity
M. Göckeler and T. Schücker Differential Geometry, Gauge Theories, and Gravity †
C. Gómez, M. Ruiz-Altaba and G. Sierra Quantum Groups in Two-Dimensional Physics †
M. B. Green, J. H. Schwarz and E. Witten Superstring Theory Volume 1: Introduction
M. B. Green, J. H. Schwarz and E. Witten Superstring Theory Volume 2: Loop Amplitudes,
Anomalies and Phenomenology
V. N. Gribov The Theory of Complex Angular Momenta: Gribov Lectures on Theoretical Physics †
J. B. Griffiths and J. Podolský Exact Space-Times in Einstein’s General Relativity †
S. W. Hawking and G. F. R. Ellis The Large Scale Structure of Space-Time †
F. Iachello and A. Arima The Interacting Boson Model †
F. Iachello and P. van Isacker The Interacting Boson-Fermion Model †
C. Itzykson and J. M. Drouffe Statistical Field Theory Volume 1: From Brownian Motion to
Renormalization and Lattice Gauge Theory †
C. Itzykson and J. M. Drouffe Statistical Field Theory Volume 2: Strong Coupling, Monte Carlo
Methods, Conformal Field Theory and Random Systems †
G. Jaroszkiewicz Principles of Discrete Time Mechanics
C. V. Johnson D-Branes †
P. S. Joshi Gravitational Collapse and Spacetime Singularities †
J. I. Kapusta and C. Gale Finite-Temperature Field Theory: Principles and Applications, 2nd
edition †
V. E. Korepin, N. M. Bogoliubov and A. G. Izergin Quantum Inverse Scattering Method and
Correlation Functions †
J. Kroon Conformal Methods in General Relativity
M. Le Bellac Thermal Field Theory †
Y. Makeenko Methods of Contemporary Gauge Theory †
S. Mallik and S. Sarkar Hadrons at Finite Temperature
N. Manton and P. Sutcliffe Topological Solitons †
N. H. March Liquid Metals: Concepts and Theory †
I. Montvay and G. Münster Quantum Fields on a Lattice †
P. Nath Supersymmetry, Supergravity, and Unification
L. O’Raifeartaigh Group Structure of Gauge Theories †
T. Ortı́n Gravity and Strings, 2nd edition
A. M. Ozorio de Almeida Hamiltonian Systems: Chaos and Quantization †
L. Parker and D. Toms Quantum Field Theory in Curved Spacetime: Quantized Fields and
Gravity
R. Penrose and W. Rindler Spinors and Space-Time Volume 1: Two-Spinor Calculus and
Relativistic Fields †
R. Penrose and W. Rindler Spinors and Space-Time Volume 2: Spinor and Twistor Methods in
Space-Time Geometry †
S. Pokorski Gauge Field Theories, 2 nd edition †
J. Polchinski String Theory Volume 1: An Introduction to the Bosonic String †
J. Polchinski String Theory Volume 2: Superstring Theory and Beyond †
J. C. Polkinghorne Models of High Energy Processes †
V. N. Popov Functional Integrals and Collective Excitations †
L. V. Prokhorov and S. V. Shabanov Hamiltonian Mechanics of Gauge Systems
S. Raychaudhuri and K. Sridhar Particle Physics of Brane Worlds and Extra Dimensions
A. Recknagel and V. Schiomerus Boundary Conformal Field Theory and the Worldsheet
Approach to D-Branes
R. J. Rivers Path Integral Methods in Quantum Field Theory †
R. G. Roberts The Structure of the Proton: Deep Inelastic Scattering †
C. Rovelli Quantum Gravity †
W. C. Saslaw Gravitational Physics of Stellar and Galactic Systems †
R. N. Sen Causality, Measurement Theory and the Differentiable Structure of Space-Time
M. Shifman and A. Yung Supersymmetric Solitons
H. Stephani, D. Kramer, M. MacCallum, C. Hoenselaers and E. Herlt Exact Solutions of
Einstein’s Field Equations, 2nd edition †
J. Stewart Advanced General Relativity †
J. C. Taylor Gauge Theories of Weak Interactions †
T. Thiemann Modern Canonical Quantum General Relativity †
D. J. Toms The Schwinger Action Principle and Effective Action †
A. Vilenkin and E. P. S. Shellard Cosmic Strings and Other Topological Defects †
R. S. Ward and R. O. Wells, Jr Twistor Geometry and Field Theory †
E. J. Weinberg Classical Solutions in Quantum Field Theory: Solitons and Instantons in High
Energy Physics
J. R. Wilson and G. J. Mathews Relativistic Numerical Hydrodynamics †
†
Available in paperback
Exact Solutions in Three-Dimensional
Gravity

A L B E RTO A . G A RC ÍA-D Í A Z
Center for Research and Advanced Studies of the
National Polytechnic Institute, Mexico
(CINVESTAV)
 University Printing House, Cambridge CB2 8BS, United Kingdom
One Liberty Plaza, 20th Floor, New York, NY 10006, USA
477 Williamstown Road, Port Melbourne, VIC 3207, Australia
4843/24, 2nd Floor, Ansari Road, Daryaganj, Delhi – 110002, India
79 Anson Road, #06–04/06, Singapore 079906

Cambridge University Press is part of the University of Cambridge.

It furthers the University’s mission by disseminating knowledge in the pursuit of
education, learning, and research at the highest international levels of excellence.

www.cambridge.org
Information on this title: www.cambridge.org/9781107147898
DOI: 10.1017/9781316556566

c Alberto A. Garcı́a-Dı́az 2017
This publication is in copyright. Subject to statutory exception
and to the provisions of relevant collective licensing agreements,
no reproduction of any part may take place without the written
permission of Cambridge University Press.
First published 2017
Printed in the United Kingdom by Clays, St Ives plc
A catalogue record for this publication is available from the British Library.
Library of Congress Cataloging-in-Publication Data
Names: Garcı́a-Dı́az, Alberto A., 1942- author.
Title: Exact solutions in three-dimensional gravity / Alberto A.
Garcı́a-Dı́az (Center for Research and Advanced Studies of the National
Polytechnic Institute, Mexico (CINVESTAV).
Other titles: Cambridge monographs on mathematical physics.
Description: Cambridge, United Kingdom ; New York, NY : Cambridge University
Press, 2017. | Series: Cambridge monographs on mathematical physics
Identifiers: LCCN 2017012130 | ISBN 9781107147898 (hardback ; alk. paper) |
ISBN 1107147891 (hardback ; alk. paper)
Subjects: LCSH: Gravitation–Mathematics. | Quantum gravity–Mathematics. |
Einstein field equations. | Space and time.
Classification: LCC QC178 .G365 2017 | DDC 539.7/54–dc23
LC record available at https://lccn.loc.gov/2017012130
ISBN 978-1-107-14789-8 Hardback
Cambridge University Press has no responsibility for the persistence or accuracy of
URLs for external or third-party internet websites referred to in this publication,
and does not guarantee that any content on such websites is, or will remain,
accurate or appropriate.
 Contents

Preface page xvii

1 Introduction 1
1.1 Main Features of (2 + 1) Gravity 1
1.1.1 Field Equations and Curvature Tensors 2
1.1.2 Matter Distribution Locally Curves the Spacetime 2
1.1.3 Point Particles Produce Global Effects on the Spacetime 3
1.1.4 Newtonian Limits 3
1.1.5 No Geodesic Deviation for Dust 5
1.1.6 No Dynamic Degrees of Freedom 6
1.1.7 Black Holes in (2 + 1) Gravity 7
1.1.8 Gravity in the Presence of Other Fields and Matter 7
1.2 Algebraic Classification 7
1.2.1 Classification of the Cotton–York Tensor 7
1.2.2 Classification of the Energy–Momentum Tensor 9
1.2.3 Classification of the Traceless Ricci Tensor 11
1.3 Brown–York Energy, Mass, and Momentum for Stationary
Metrics 11
1.3.1 Summary of Quasilocal Mass, Energy, and Angular
Momentum 14
1.4 Decomposition with Respect to a Frame of Reference 15
1.4.1 Kinematics of the Frame 15
1.4.2 Perfect Fluid Referred to a Frame of Reference 16

2 Point Particle Solutions 19

2.1 Staruszkiewicz Point Source Solutions 19
2.1.1 Relationship Between the Deficit Angle and Mass 20
2.2 Staruszkiewicz Single Point Source Solution 21
2.2.1 No Parallelism With the (3 + 1) Schwarzschild Solution 22
2.3 Staruszkiewicz Two Point Sources Solution 22
2.4 Deser–Jakiw–’t Hooft Static N Point Sources Solution 23
2.4.1 Energy and Euler Invariant 23
2.4.2 Energy–Momentum Tensor for N Point Particles 24
2.5 Clément Rotating Point–Particles Solution 24
viii Contents

3 Dust Solutions 27
3.1 Cornish–Frankel Dust Heaviside Function Solution 27
3.2 Giddings–Abott–Kuchař Dust Solutions 28
3.2.1 Time-Dependent Class of Dust Solutions
Ω = ln (tf (x, y)) 29
3.2.2 Static Class of Dust Solutions Ω = ln g(x, y) 30
3.3 Barrow–Shaw–Tsagas Anisotropic Dust Solution; Λ = 0 30
3.4 BST Diagonal Anisotropic Dust Solutions with Λ 33
3.5 BST (t, x, y)-Dependent Cosmological Solutions with
Comoving Dust 34
3.5.1 BST Class 2 of Solutions 37
3.5.2 BST Class 1 Spacetime 38
3.5.3 BST Class 3 of Dust Solutions 39
3.6 Rooman–Spindel Dust Gödel Non-Diagonal Model 40

4 A Shortcut to (2+1) Cyclic Symmetric Stationary Solutions 44

4.1 Cyclic Symmetric Stationary Solutions in Canonical Coordinates 44
4.1.1 Bañados–Teitelboim–Zanelli Solution in Canonical
Polar ρ Coordinate 45
4.1.2 BTZ Solution Counterpart 46
4.1.3 Coussaert–Henneaux Metrics 47
4.2 Static AdS Black Hole 48
4.2.1 Static BTZ Solution 49
4.2.2 Static AdS Solution Counterpart 51
4.3 Symmetries of the Stationary and Static Cyclic Symmetric BTZ
Metrics 52
4.3.1 Symmetries of the AdS Metric for Negative M , M = −α2 56

5 Perfect Fluid Static Stars; Cosmological Solutions 58

5.1 Static Circularly Symmetric Fluid Solutions 58
5.1.1 Cotton Tensor Types 59
5.2 Incompressible Static Star 59
5.2.1 Collas Static Star with Constant Density μ0 60
5.2.2 Giddings–Abott–Kuchař Static Star with μ0 60
5.2.3 Cornish–Frankel Static Star with μ0 61
5.3 Cornish–Frankel Static Polytropic Solutions 62
5.3.1 Static Star with a Stiff Matter p(r) = μ(r) 64
5.3.2 Static Star with Pure Radiation p = μ(r)/2 65

6 Static Perfect Fluid Stars with Λ 66

6.1 Equations for a (2+1) Static Perfect Fluid Metric 67
6.1.1 General Perfect Fluid Solution with Variable ρ(r) 68
6.2 Canonical Coordinate System {t, N, θ} 69
 Contents ix

6.3 Perfect Fluid Solutions for a Barotropic Law p = γ ρ 70

6.4 Perfect Fluid Solutions for a Polytropic Law p = Cργ 71
6.5 Oppenheimer–Volkoff Equation 73
6.6 Perfect Fluid Solution with Constant Density 74
6.6.1 (3+1) Static Spherically Symmetric Perfect Fluid Solution 76
6.6.2 Comparison Table 78

7 Hydrodynamic Equilibrium 80
7.1 Generalized Buchdahl’s Theorem 80
7.2 Stellar Equilibrium in (2 + 1) Dimensions with Λ 81
7.2.1 Cruz–Zanelli Existence of Hydrostatic Equilibrium for
Λ≤0 83
7.2.2 No Buchdahl’s Inequality in (2 + 1) Hydrostatics 83
7.2.3 Static Star with Constant Density μ0 and Λ = −1/l2 ≤ 0 84
7.3 Buchdahl Theorem in d Dimensions 85
7.3.1 Buchdahl’s Inequalities 86
7.3.2 Constant Density Solution 89

8 Stationary Circularly Symmetric Perfect Fluids with Λ 92

8.1 Stationary Differentially Rotating Perfect Fluids 93
8.2 Garcia Stationary Rigidly Rotating Perfect Fluids 94
8.2.1 Rigidly Rotating Perfect Fluid Solution with
W (r) = J/(2r2 ) 96
8.2.2 Garcia Interior Solution with Constant Energy Density 97
8.2.3 Interior Perfect Fluid Solution to the BTZ Black Hole 100
8.2.4 Alternative Parametrization 100
8.2.5 Barotropic Rotating Perfect Fluids Without Λ 102
8.3 Lubo–Rooman–Spindel Rotating Perfect Fluids 102
8.3.1 Equations for Rigidly Rotating Fluids 104
8.3.2 Garcia Representation of Stationary Perfect Fluid
Solutions 105
8.3.3 Barotropic Class of Solutions p = γ μ 105
8.3.4 Constant Density Stationary Solution; p = p(r), μ = μ0 105
8.3.5 Lubo–Rooman–Spindel Perfect Fluids u = θ0 and grr = 1 106
8.3.6 LBR Rotating Perfect Fluid with μ0 106
8.3.7 Rooman–Spindel Rotating Fluid Model; gtt = −1 = −grr 107

9 Friedmann–Robertson–Walker Cosmologies 108

9.1 Einstein Equations for FRW Cosmologies 108
9.1.1 Einstein Equations for (3+1) FRW Cosmology 108
9.1.2 Einstein Equations for (2+1) FRW Cosmology 109
9.2 Barotropic Perfect Fluid FRW Solutions 110
9.2.1 Barotropic Perfect Fluid (3 + 1) Solutions 110
x Contents

9.2.2 Barotropic Perfect Fluid (2 + 1) Solutions 111

9.2.3 Comparison Between (3+1) and (2+1) Barotropic
Solutions 112
9.3 Polytropic Perfect Fluid FRW Solutions 113
9.3.1 Polytropic Perfect Fluid (3 + 1) Solutions 113
9.3.2 Polytropic Perfect Fluid (2 + 1) Solutions 114
9.3.3 Comparison Between (3+1) and (2+1) Polytropic
Solutions 114
9.4 Mann–Ross Collapsing Dust FRW Solutions with Λ 114
9.4.1 Cosmological dS–FRW Solution 115
9.4.2 Asymptotically AdS–FRW Dust Solution 115
9.4.3 Matching the AdS–FRW Dust to the Static BTZ 116
9.4.4 Determination of Kij 117
9.4.5 Gidding–Abbott–Kuchař Dust FRW Solution 119

10 Dilaton–Inflaton Friedmann–Robertson–Walker
Cosmologies 121
10.1 Equations for a FRW Cosmology with a Perfect Fluid and a
Scalar Field 122
10.1.1 Einstein Equations for (3+1) FRW Dilaton Cosmology 122
10.1.2 Einstein Equations for (2+1) FRW Cosmology 123
10.1.3 Correspondence Between (3+1) and (2+1) Solutions 125
10.2 Single Scalar Field to Linear State Equations; Λ = 0 127
10.2.1 (2+1) Solutions for a Scalar Field 127
10.2.2 (3+1) Solutions for a Scalar Field 128
10.2.3 Slow Roll Spatially Flat FRW Solutions 129
10.3 Spatially Flat FRW Solutions for Barotropic Perfect Fluid and
Scalar Field 131
10.3.1 Spatially Flat FRW (3+1) Solutions ; γ4 = 2Γ4 131
10.3.2 Spatially Flat FRW (2+1) Solutions; γ3 = 2Γ3 133
10.3.3 Barrow–Saich Solution; γ = 2 Γ 134
10.4 Single Scalar Field Spatially Flat FRW Solutions to
pφ + ρφ = Γ ρφ β 135
10.4.1 Spatially Flat (3+1) Solutions with
V (φ) = A(αφ2/(1−β) − φ2β/(1−β) ) 135
10.4.2 Spatially Flat (2+1) Solutions with
V (φ) = A(αφ2/(1−β) − φ2β/(1−β) ) 136
10.4.3 Barrow–Burd–Lancaster (2+1) and Madsen (3+1)
Solutions 137
10.5 Scalar Field Solutions for a Given Scale Factor 139
10.5.1 Second (2+1) BBL Solution 139
10.5.2 (3+1) Generalization of the Second (2+1) BBL
Solution 141
 Contents xi

11 Einstein–Maxwell Solutions 142

11.1 Stationary Cyclic Symmetric Einstein–Maxwell Fields 143
11.1.1 Stationary Cyclic Symmetric Maxwell Fields 143
11.1.2 General Stationary Metric and Einstein Equations 145
11.1.3 Complex Extension and Real Cuts 148
11.1.4 Positive Λ Solutions 149
11.1.5 Characterizations of Einstein–Maxwell Solutions 149
11.1.6 Static Cyclic Symmetric Equations for Maxwell Fields 152
11.2 Electrostatic Solutions; b = 0, a = 0 153
11.2.1 General Electrostatic Solutions 154
11.2.2 Gott–Simon–Alpern, Deser–Mazur, and
Melvin Electrostatic Solution 155
11.2.3 Charged Static Peldan Solution with Λ 156
11.3 Magnetostatic Solutions; a = 0, b = 0 159
11.3.1 General Magnetostatic Solutions 160
11.3.2 Melvin, and Barrow–Burd–Lancaster Magnetostatic
Solution 161
11.3.3 Peldan Magnetostatic Solution with Λ 162
11.3.4 Hirschmann–Welch Solution with Λ 165
11.4 Cataldo Static Hybrid Solution 168
11.4.1 Mass and Energy 170
11.4.2 Field, Energy–Momentum, and Cotton Tensors 171
11.5 Uniform Electromagnetic Solutions Fμν ;σ = 0 173
11.5.1 General Uniform Electromagnetic Solution for
a = 0, = b 173
11.5.2 Uniform “Stationary” Electromagnetic
A = r/(b l2 )(dt − ω0 dφ) Solutions 175
11.5.3 Matyjasek–Zaslavskii Uniform Electrostatic
A = r/(b l2 ) dt Solution 176
11.5.4 Uniform “Stationary” Electromagnetic
A = r/(a l2 )(dφ + W0 dt) Solutions 179
11.5.5 No Uniform Stationary Magnetostatic Solution
for Λ = −1/l2 180
11.6 Constant Electromagnetic Invariants’ Solutions 180
11.6.1 General Constant Invariant Fμν F μν = 2γ for a = 0 = b 181
11.6.2 Constant Electromagnetic Invariant F F = ∓2/l2
Solution 182
11.6.3 Constant Electromagnetic Invariant F F = −2/l2
Solution for b = 0 182
11.6.4 Constant Electromagnetic Invariant F F = 2/l2
Stationary Solution for a = 0 183
11.6.5 Vanishing Electromagnetic Invariant F F = 0
Solution 184
xii Contents

11.6.6 Kamata–Koikawa Solution √ 185

11.6.7 Proper Kamata–Koikawa Solution, ρ0 = ±Q/ Λ 189
11.7 Ayón–Cataldo–Garcia Stationary Hybrid Solution 191
11.7.1 ACG Hybrid Solution Allowing for BTZ Limit 192
11.7.2 Mass, Energy, and Momentum 194
11.7.3 Constant Electromagnetic Invariants’ Hybrid Solution
for Λ = 0 198
11.8 Stationary Solutions for a = 0 or b = 0 199
11.8.1 Stationary Magneto-Electric Solution for a = 0 = b 199
11.8.2 Stationary Electromagnetic Solution for b = 0 = a 202
11.9 Garcı́a Stationary Solutions for a = 0 and b = 0 204
11.9.1 Alternative Representation of the Einstein Equations 205
11.9.2 Garcı́a Stationary Electromagnetic Solution with BTZ
limit 206
11.9.3 Garcı́a Stationary Solution with BTZ-Counterpart
Limit 213
11.10 Generating Solutions via SL(2, R)–Transformations 215
11.11 Transformed Electrostatic b = 0 Solutions 217
11.11.1 Stationary Electromagnetic Solution 218
11.11.2 Clément Spinning Solution 219
11.11.3 Martı́nez–Teitelboim–Zanelli Solution 224
11.12 Transformed Magnetostatic a = 0 Solutions 228
11.12.1 Stationary Magneto-Electric Solution 229
11.12.2 Dias–Lemos Magnetic BTZ–Solution Counterpart 229
11.13 Transformed Cataldo Hybrid Static Solution 235
11.13.1 Mass, Energy and Momentum 237
11.14 Summary on Electromagnetic Maxwell Solutions 238

12 Black Holes Coupled To Nonlinear Electrodynamics 240

12.1 Nonlinear Electrodynamics in (2 + 1) Dimensions 241
12.2 General Nonlinear Electrostatic Solution 242
12.2.1 Static Charged Peldan Solution 244
12.3 Cataldo–Garcı́a Nonlinear EBI Charged Black Hole 244
12.3.1 Static Cyclic Symmetric EBI Solution 246
12.3.2 Cataldo–Garcı́a Black Hole to EBI 247
12.4 Regular Black Hole Solution 249
12.4.1 Regularity 250
12.4.2 Horizons 251
12.4.3 Thermodynamics 252
12.5 Coulomb-Like Black Hole Solution 252
12.5.1 Horizons for the Coulomb-Like Solution 254
12.6 Stationary Nonlinear Electrodynamics Black Holes 256
 Contents xiii

13 Dilaton Field Minimally Coupled to (2 + 1) Gravity 257

13.1 Scalar Field Minimally Coupled to Einstein Gravity 257
13.2 Static Black Hole Coupled to a Scalar Ψ(r) = k ln(r) 258
13.2.1 Quasi Local Momentum, Energy, and Mass 261
13.2.2 Classification of the Energy–Momentum and Cotton
Tensors 262
13.3 General Static Chan–Mann Solution 263
13.3.1 Regular F (r)+ Function for the Metric g+ 264
13.3.2 Chan–Mann Solution 265
13.4 Stationary Solution Coupled to Ψ(r) = k ln(r) 266
13.4.1 Momentum, Energy, and Mass for a Rotating Dilaton 268
13.4.2 Classification of the Energy–Momentum and Cotton
Tensors 269
13.5 Stationary Dilaton Solutions Generated via SL(2, R)
Transformations 270
13.5.1 Sub-Class of Rotating Dilaton Black Holes 272
13.5.2 Rotating Chan–Mann Dilaton Black Hole 273
13.6 Dilaton Coupled to Einstein–Maxwell Fields 274
13.6.1 Einstein–Maxwell-Scalar Field Equations 274
13.7 Static Charged Solution Coupled to Ψ(r) = k ln(r) 275
13.7.1 Quasi-Local Mass, Momentum, Energy for Charged
Dilaton 277
13.7.2 Algebraic Classification of the Field,
Energy–Momentum, and Cotton Tensors 277
13.8 Stationary Charged Dilaton Generated via SL(2, R) 279
13.8.1 Quasi-Local Mass and Momentum 280
13.8.2 Algebraic Classification of the Field,
Energy–Momentum, and Cotton Tensors 282
13.8.3 Particular Stationary Charged Dilaton via SL(2, R)
Transformation 284
13.9 Summary of Dilaton Minimally Coupled to Gravity 284

14 Scalar Field Non-Minimally Coupled to (2+1) Gravity 286

14.1 Einstein Equations for Non-Minimally Coupled Scalar Field 286
14.1.1 Martinez–Zanelli Black Hole Solution with Tμ μ = 0 287
14.2 Stationary Black Hole for a Non-Minimally Coupled Scalar Field 287
14.2.1 Quasi-Local Momentum, Energy, and Mass 288
14.2.2 Algebraic Classification of the Ricci,
Energy–Momentum, and Cotton Tensors 289

15 Low-Energy (2+1) String Gravity 292

15.1 n-Dimensional Heterotic String Dynamical Equations 292
xiv Contents

15.1.1 String Frame 292

15.1.2 Einstein Frame 293
15.2 Dynamical Equations in (2+1) String Gravity 295
15.3 Horne–Horowitz Black String 296
15.4 Horowitz–Welch Black String 298
15.5 Chan–Mann String Solution 300
15.5.1 Einstein–Maxwell-Scalar Field Equations 300
15.5.2 Static and Stationary Black String Solutions 301

16 Topologically Massive Gravity 303

16.1 Chern–Simons Action and Field Equations of TMG 304
16.2 Exact Vacuum Solutions of TMG with Λ 305

17 Bianchi-Type (BT) Spacetimes in TMG; Petrov Type D 307

17.1 Generalities on Bianchi-Type (BT) 3D Spaces 307
17.2 Nutku–Baekler–Ortiz “Timelike” BT VIII Spacetime 309
17.2.1 Nutku Timelike Biaxially Squashed Metric 310
17.3 Nutku–Baekler–Ortiz “Spacelike” Squashed BT VIII Spacetime 311
17.3.1 Spacelike Biaxially Squashed Metric; Nutku Solution
Counterpart 313
17.4 Nutku–Baekler–Ortiz Solutions of Bianchi Type III 314
17.4.1 Nutku–Baekler–Ortiz BT III Timelike Solution with
Λ=0 315
17.4.2 Nutku–Baekler–Ortiz BT III Spacelike Solution with
Λ=0 316
17.5 Timelike Biaxially Squashed Metrics 317
17.5.1 Representation of the Vacuum Biaxially Squashed
Solutions 317
17.5.2 Eigenvectors of the Cotton Tensor; Triad Formulation 319
17.5.3 Complex Extension Toward the Spacelike Squashed
Metric 320
17.5.4 Alternative Metric Representation of dstl 2 321
17.6 Spacelike Biaxially Squashed Metrics 327
17.6.1 Eigenvectors of the Cotton Tensor; Triad Formulation 328
17.6.2 Alternative Metric Representation of ds2sl 329

18 Petrov Type N Wave Metrics 333

18.1 Brinkmann-Like 3D Metric 333
18.2 AdS3 Non-Covariantly Constant TN-Waves 334
18.2.1 AdS3 TN-Waves with Λ = 0 336
18.2.2 Nutku TN-Wave Solution 336
18.2.3 Clément TN-Wave Solution 337
18.2.4 Ayón–Hassaı̈ne TN-Wave Solution 337
 Contents xv

18.2.5 Ölmez–Sarioğlu–Tekin TN-Wave Solution 338

18.2.6 Dereli-Sarioğlu TN-Wave Solution 338
18.2.7 Carlip–Deser–Waldron–Wise TN-Wave Solution 338
18.2.8 Gibbons–Pope–Sezgin TN-Wave Solution 338
18.2.9 Anninos–Li–Padi–Song–Strominger TN-Wave Solution 338
18.2.10 Garbarz–Giribet–Vásquez TN-Wave Solution 339
18.3 pp-Wave Solutions; Λ = 0 340
18.3.1 Martinez–Shepley pp-Wave Solution; Λ = 0 340
18.3.2 Aragone pp-Wave Solution; Λ = 0 341
18.3.3 Percacci–Sodano–Vuorio pp-Wave Solution; Λ = 0 341
18.3.4 Hall–Morgan–Perjés pp-Wave Solution; Λ = 0 341
18.3.5 Dereli–Tucker pp-Wave Solution; Λ = 0 342
18.3.6 Deser–Steif pp-Wave Solution; Λ = 0 342
18.3.7 Clément pp-Wave Solution; Λ = 0 342
18.3.8 Cavaglia pp-Wave Solution; Λ = 0 343
18.3.9 Dereli–Sarioğlu pp-Wave Solution; Λ = 0 343
18.3.10 Garcı́a–Hehl–Heinicke–Macı́as pp-Wave Solution; Λ = 0 343
18.3.11 Macı́as–Camacho pp-Wave Solution; Λ = 0 344

19 Kundt Spacetimes in TMG 345

19.1 Null Geodesic Vector Field 345
19.2 General Kundt Metrics 347
19.2.1 3D Kundt Metric 348
19.3 3D Canonical Kundt Metric 350
19.3.1 Petrov Classification of the Cotton and Traceless Ricci
Tensors 352
19.3.2 Sub-Branch W1 (r) of the General Kundt Metric in TMG 355
19.3.3 Kundt Metric Structure for W1 (r) 356
19.4 Type II CSI Kundt Metric; W1 = 2μ/3 357
19.4.1 Negative Cosmological Constant; Λ = −m2 358
19.4.2 Positive Cosmological Constant; Λ = m2 359
19.4.3 Zero Cosmological Constant; Λ = 0 359
19.5 Type D CSI Kundt Solutions; W1 = 2μ/3, F0 = 0 359
19.6 Petrov Type III Kundt Metrics 360
19.7 Type III Kundt Solution; Λ = 0, W1 = 0 361
19.7.1 Type N pp-Wave Limit 362
19.8 Type III Kundt Solution; Λ = 0, W1 = −2/r 362
19.8.1 Type N Limit 365
19.9 Type III Kundt Solution; Λ = −m2 , W1 = −2m 366
19.9.1 Type III Kundt Solution; Λ = −m2 , W1 = −2m, μ = ±m 366
19.9.2 Type III Kundt Solution; Λ = −m2 , W1 = −2m, μ = m 367
19.9.3 Type III Kundt Solution; Λ = −m2 , W1 = −2m,
μ = −m 368
xvi Contents

19.10 Type III Kundt Metric; Λ = −m2 , W1 = −2 m coth(m r) 369

19.10.1 Type III Kundt Solution; Λ = −m2 ,
W1 = −2m coth(m r), μ = ±m 370
19.10.2 Multi Exponent–Integral Representation of φ 375
19.10.3 Type III Kundt Solution; Λ = −m2 ,
W1 = −2m coth(m r), μ = −m 377
19.10.4 Type III Kundt Solution; Λ = −m2 , W1 =
−2m coth(mr), μ = m 381
19.11 Type III Kundt Metric; Λ = −m2 ; W1 = −2 m tanh(m r) 384
19.11.1 Type III Kundt Solution; Λ = −m2 ,
W1 = −2m tanh(mr), μ = ±m 386
19.11.2 Type III Kundt Solution; Λ = −m2 , W1 =
−2m tanh(mr), μ = −m 391
19.11.3 Type III Kundt Solution; Λ = −m2 , W1 =
−2m tanh(mr), μ = m 394
19.12 Type III Kundt Metric; Λ = m2 , W1 = −2m cot(m r), μ 398
19.12.1 Solution Φ Through VP 399
19.12.2 Multi Exponent–Integral Representation of Φ 401

20 Cotton Tensor in Riemannian Spacetimes 404

20.1 Bianchi Identities and the Irreducible Decomposition
of the Curvature 405
20.2 Cotton 2-Form 409
20.3 Conformal Correspondence 415
20.4 Criteria for Conformal Flatness 416
20.5 Classification of the Cotton 2-Form in 3D 417
20.5.1 Euclidean Signature 418
20.5.2 Lorentzian Signature 419

References 421
Index 430
 Preface

I have been working on this book project over a period of many years in order
to create a concise but comprehensive account of exact solutions in the three-
dimensional Einstein theory of gravity. This theory, although tangentially related
to the real gravity world, is a good model from which to extract some relevant
conclusions about that world, taking advantage of simplifications due to the
reduction in dimensions. Of course, after fifty years of existence, (2 + 1) gravity
has been approached from many perspectives, but, to my mind, there is still
no satisfactory account of solutions of physical interest. I therefore wrote the
present book on exact solutions in (2 + 1) gravity following closely the pattern
of the classic book, Exact solutions in Einstein’s field equations with the aim of
bringing to each class of solutions presented its theoretical framework, its general
physical and geometrical characteristics, and references to the researchers who
discovered or studied them.
I am greatly indebted to my late friends and collaborators J.F. Plebański, N.
R. Sibgatullin and S. del Campo for stimulating talks. Many thanks are due to
my collaborators E. Ayón-Beato, C. Campuzano, M. Cataldo, G. Gutierrez, F.
Hehl, A. Macı́as, V. Manko, N. Mitskievich, and C. Terrero for many enlightening
talks, and advice on various problems I faced when dealing with this task. This
work could not have come about without a fruitful sabbatical stay at the Physics
Department of the California University at Davis under the sponsorship of Pro-
fessor S. Carlip, to whom I am indebted, and to UC MEXUS-CONACYT for
financial support. I cordially thank my friends, particularly Y. Gurievich and M.
Lopez, and colleagues from the Departamento de Fı́sica, Centro de Investigación
y de Estudios Avanzados del I.P.N., for their encouragement. Many thanks to
my assistant E.Vargas-Dı́az for her daily support and typographical help in the
final preparation of the manuscript of this book. The support of the Consejo
Nacional de Ciencia y Tecnologı́a (CONACYT), México, through various grants
(at present: Grant CONACyT 178346) is acknowledged.
I really do appreciate the patience and support of my daughters, Ana Alicia
and Alexandra Sofia, and sons, Sergey, Albert, Alberto-Tiko, and Alex, over all
these years of writing this book.
Please do not hesitate to let me know about inconsistencies, unevenness, omis-
sions or errors to be found in this text. Constructive comments are welcome:
aagarcia@fis.cinvestav.mx.
 1
Introduction

Given the large number of exact solutions that exist today in (2 + 1) Einstein
gravity the purpose of the present book is to present a complete and concise list
of exact solutions with emphasis on their physical and geometrical properties
from the beginnings of the field in 1963 to the present, to be useful for the
audience of experts and young researchers. Emphasis is given to solutions to
the Einstein equations in the presence of matter and fields, for instance, point
particle solutions, perfect fluids, cosmological spacetimes, dilatons, inflatons, and
stringy solutions. The second part of this book deals with solutions to vacuum
topologically massive gravity with a cosmological constant, as there exist three
big families of spacetimes: the inhomogeneous Bianchi class of solutions, the
Kundt spacetimes and the Cotton type N wave fields.
To avoid unnecessary typing, the cosmological constant is denoted by Λ, AdS
spacetime stands for an asymptotically anti-de Sitter spacetime with Λ < 0,
dS spacetime stands for an asymptotically de Sitter spacetime with Λ > 0, 3D
stands for three dimensions, while (2+1)D spacetime denotes (2+1)-dimensional
spacetime, PF stands for perfect fluid, and ρ or μ denotes the fluid energy den-
sity. Occasionally we use SL for spacelike, TL for timelike, and ST to denote
spacetime. On the other hand, when publications by various authors are cited,
an abbreviation of their family names, including their first capital initials, are
given; for instance, EEqs. and EM mean, respectively, Einstein equations and
Einstein–Maxwell, MTW stands for Misner, Thorne, and Wheeler; FRW reads
Friedmann–Robertson–Walker, and BTZ denotes Bañados–Teitelboim–Zanelli.

1.1 Main Features of (2 + 1) Gravity

In the early work by Giddings, Abbott, and Kuchař (GAK) (Giddings et al.,
1984) it is stated that “the lowest dimension in which the Einstein Theory makes
2 Introduction

sense is n = 3.” Consequently, bearing in mind the self-contained nature of this

book, the main features of (2 + 1)-dimensional gravity are presented following
the GAK pattern and essentially maintaining their wording.

1.1.1 Field Equations and Curvature Tensors

Einstein’s theory of relativity, as a theory of a gravitational spacetime, can be
based on two postulates which are independent of the spacetime dimensions;
these postulates demand that the field equations take the form of the Einstein
equations:
1
Gμν := Rμν − R gμν = κ Tμν − Λ gμν , (1.1)
2
where Gμν is the Einstein tensor, Tμν is the energy–momentum tensor (which, by
virtue of the Bianchi identity, fulfills the energy conservation condition T μ ν;μ =
0), Λ is a cosmological constant, and κ is a coupling constant. And second, the
spacetime geometry is determined by the Riemann curvature tensor Rα βγδ .
The Riemann tensor in three dimensions possesses six algebraically indepen-
dent components: as many as the number of independent components of the
Ricci tensor; therefore, the Riemann tensor is completely determined by the
Ricci tensor and the scalar curvature, namely:
Rαβγδ = gαγ Rβδ − gαδ Rβγ − gβγ Rαδ + gβδ Rαγ
1
− (gαγ gβδ − gαδ gβγ )R. (1.2)
2
There is no room for the Weyl conformal tensor, which is thus zero.
Due to the Einstein equations, the Riemann tensor can be expressed in terms
of the Einstein tensor or, in turn, through the energy momentum tensor Tαγ as
Rαβγδ = κ[gαγ Tβδ − gαδ Tβγ ] + κ[gβδ Tαγ − gβγ Tαδ ]
−κ(gαγ gβδ − gαδ gβγ )T. (1.3)
In three dimensions, the coupling constant κ is measured in units of 1/mass, and
therefore defines a natural mass unit.

1.1.2 Matter Distribution Locally Curves the Spacetime

Since, in a 3D spacetime, the Riemann curvature tensor is expressible solely
through the energy–momentum tensor (1.3), thus, in an empty spacetime, where
Tαγ = 0, the spacetime is locally flat:
Rαβγδ = 0.
Therefore, the flat spacetime is the field solution to the vacuum Einstein equa-
tions Gμν = 0 → Rμν = 0 = R, (Tμν = 0 and Λ = 0). Staruszkiewicz (1963), in
 1.1 Main Features of (2 + 1) Gravity 3

his pioneering article, stressed this fact by writing: “three-dimensional gravita-

tion theory is a theory without a field of gravitation; where no matter is present,
space is flat. Curvature can arise only if matter or energy are present.”

1.1.3 Point Particles Produce Global Effects on the Spacetime

The first publication on gravity in (2 + 1) dimensions by Staruszkiewicz (1963)
was devoted to the description of static solutions determined by point sources.
Point particles move along geodesics. In the points where the particles are located
there arise conical defects (conical singularities) that can be felt at infinity; the
total mass in the spacetime is proportional to the deficit angle at infinity. Because
the angle deficit cannot increase by 2π, the mass is bounded from above. See
Chapter 2 for details.

1.1.4 Newtonian Limits

This section has to be subdivided into three subsections: first, to recall the con-
tent of the Newtonian theory; second, to reveal the existence of the Newtonian
limit in the standard (3 + 1), or (1 + (n − 1)), gravity via the weak gravitational
field treatment; and finally, to show that the slow motion limit of the (2 + 1)
gravity occurs without acceleration.

Newtonian Theory of Gravity

The Newtonian theory of gravity is based on the Newtonian potential φ fulfilling
the Poisson equation with matter density ρ,
∇2 φ = 4π G ρ, (1.4)
which generates the Newtonian field causing the accelerated motion of test
particles in it:
d2 xi
= −δ i j ∂j φ. (1.5)
dt2
G stands for the Newton constant of gravitation.

Weak Gravitational Theory in n Dimensions

In this paragraph, starting from the n-dimensional Einstein equations for weak
gravitational fields (expansion of the metric components), the equation obeyed
by the weak fields, and related to the energy density, is derived. The limit of the
geodesic equation of motion for test particles in the case of slow velocities and
weak gravitational fields is derived and its consequences analyzed.

GAK Linearized Approach

Following GAK, the linearized Einstein equations in the Hilbert–de Donder gauge
reduce to the inhomogeneous wave equation
4 Introduction
 
τ
 hαβ = −2κ ταβ − ηαβ . (1.6)
n−2
For small perturbations,
hαβ = gαβ − ηαβ , (ηαβ ) = diag (−1, 1, 1, 1). (1.7)
In weak fields, the linearized stresses τij (spatial components) are negligible in
comparison to the mass density τ00 = ρ, and assuming additionally the quasi-
staticity of the field and sources, one arrives from (1.6) at
n−3
∇2 h00 = −2κ τ00 , (1.8)
n−2
or, identifying h00 = −2 φ, one gets
n−3
∇2 φ = κ ρ. (1.9)
n−2
The geodesic equation
dxα α dxμ dxν
+ Γ μν = 0, (1.10)
ds2 ds ds
for slow motion in the linearized limit, becomes
dxi 1 dxi
2
− δ ij ∂j h00 = 0 → 2 = − δ ij ∂j φ. (1.11)
dt 2 dt

Carlip Linearized Approach

In Chapter 1 of Carlip (1998), it is stated that “general relativity in 2+1
dimensions has a Newtonian limit in which there is no force between static
point masses.” The starting point to establish this assessment is the approxima-
tion (1.7), and it continues with the n-dimensional field equations given in the
harmonic gauge (which can always be chosen) such that:
1
− η μν ∂μ ∂ν h̄αβ + O(h2 ) = κTαβ ,
2
η μν ∂μ h̄νβ = 0, (1.12)
where
1 1
h̄αβ = hαβ − ηαβ η μν hμν → hαβ = h̄αβ − ηαβ η μν h̄μν . (1.13)
2 n−2
The Newtonian limit is obtained:
by setting T00 = ρ, where ρ is the mass density;
by equating to zero all other components of the stress–energy tensor;
by ignoring time derivatives;
then, identifying
h̄00 = −4φ, (1.14)
the linearized equations (1.12) reduce to
 1.1 Main Features of (2 + 1) Gravity 5

κ
∇2 φ = ρ. (1.15)
2
In this limit, the geodesic equation (1.10) reduces to
d2 xi 1
− ∂i h00 = 0 (1.16)
dt2 2
and, taking into account (1.13) and (1.14), becomes
d2 xi n − 3 ij
= −2 δ ∂j φ. (1.17)
dt2 n−2

Discrepancies
Comparing these results reported by GAK, Giddings et al. (1984), and Carlip
(1998), one notices that the numerical coefficients in the GAK equations for the
limits of Newton and particle motion equations correspond to those in the limit
of the geodesic motion and to the Newton equation respectively of Carlip (1998).

The Outcome of the Dilemma

A detailed derivation of the linearized Einstein theory is done in 3+1 dimensions
by Tonnelat (1959), Chapter 12, §1, 2, 3, which can be extended to (1 + (n − 1))
dimensions, practically without any changes. In this manner, using the de Don-
√
der conditions, σ β = √1−g ∂α ( −g g α β ) = 0 – which allow for the existence of
(isothermal) harmonic coordinates and the use of the subclass of quasi-Lorentzian
coordinates in the linearization problem of the Einstein equations – one gets the
Newton limit in the form (1.15). Moreover, the limit of the geodesic equation
for slow motion is also established by Tonnelat (1959), which in terms of the
“bar” quantities, in the terminology of MTW, Misner et al. (1973, Chapter 18),
gives (1.16), and consequently, in term of the Newtonian potential, it reduces to
(1.17).

Weak Gravitational Theory in (2 + 1) Dimensions

Having at hand the linearized expressions of the Einstein equations and of the
geodesic equations, in any dimension n, by means of the equations (1.15) and
(1.17), correspondingly, one easily recognizes that in (2 + 1) gravity the Newto-
nian limit holds in the two spatial dimensions, but the Newtonian acceleration
equation fails to be true; the geodesic slow motion occurs without acceleration,
d2 xi
dt2 = 0.

1.1.5 No Geodesic Deviation for Dust

It is apparent that the geodesic deviation for neighboring moving particles has
to vanish. Consider a congruence of geodesics with tangent vectors uα , and let
the separation vectors from one geodesic to another be V μ ; then the geodesic
deviation equation is
6 Introduction

∇u ∇u V α = Rα βγδ uβ uγ V δ . (1.18)

Assume now that this congruence is modeled by a tube of dust with energy–
momentum tensor
T αβ = ρ uα uβ . (1.19)

Substituting this tensor into the expression of the Riemann tensor (1.3) and
contracting it with u, one gets

Rα βγδ uβ uγ = 0, (1.20)

which in turn implies

∇u ∇u V α = 0. (1.21)

Therefore, in a (2 + 1) spacetime the world lines of dust do not deviate. In par-

ticular, the trajectories of point particles do not approximate to one another;
in other words, there is no acceleration between them. This final observation is
another effect which is present in (2 + 1) gravity. Summarizing, in this theory
there is no action at distance: in a 3D spacetime, gravitational effects do not
propagate outside the matter content; test particles outside the matter region
move along geodesics without experiencing acceleration and geodesic deviation.

1.1.6 No Dynamic Degrees of Freedom

In more than three dimensions, the Weyl tensor encodes the information about
the Riemann curvature not caused by matter. Since in 3D spacetime the Weyl
tensor vanishes – that is, there is no room for it – then, because curvature
is only produced by matter, the gravitational field has no dynamic degrees of
freedom. Another way to arrive at this absence of degrees of freedom is through
a counting argument appealing to the canonical geometrodynamics, as was done
in 3D by Giddings et al. (1984) and by Carlip (1998) in nD. Roughly speaking,
in the canonical geometrodynamics, the spacetime is foliated by means of a one-
parameter family of spacelike hypersurfaces, xα = xα (xa , t); thus one can define
an intrinsic metric
∂xα
gab = gαβ Xaα Xbβ , Xaα := , a = 1, . . . , n − 1, α = 1, . . . , n − 1, t,
∂xa
and a field of unit normals U α , to these hypersurfaces, together with the extrinsic
curvatures, Kab = −Uα;β Xaα Xbβ . The Einstein equations are then decomposed
with respect to the normal and tangential directions to the hypersurfaces; then
we introduce the lapse function N , the shift vector N a , the intrinsic metric
gab and the momentum conjugate to the metric pab , defined by means of the
extrinsic metric, which are elevated to the category of canonical variables. Ein-
stein equations are then expressed in terms of all of them, with constraints:
the projections Gαβ U α U β , (1), Gαβ U α Xaβ , (n − 1), the evolution eq. for
p: ṗab ∼ Gab Xaα Xbβ , n(n − 1)/2, and the evolution eq. for g: ġab , n(n − 1)/2.
 1.2 Algebraic Classification 7

In n dimensions, the intrinsic metric possesses n(n − 1)/2 components, and the
conjugate momentum also has n(n − 1)/2 components, and together the number
of their components is n(n − 1). On the other hand, one can fix n of them by
choosing n coordinates, and additionally n by the constraints; consequently, the
number of degree of freedom in the canonical data is: n(n − 1) − 2n = n(n − 3).
Hence, in three dimensions there is no freedom in the prescription of the initial
data in the initial hypersurface.
As a consequence of this lack of degrees of freedom, there are no gravitational
waves in 3D flat spacetime; in the terminology of Gott and Alpert (1984), there
are no gravity waves in flatland; no gravitons.

1.1.7 Black Holes in (2 + 1) Gravity

So there are no black holes in asymptotically flat spacetime; the asymptotically
flat space is flat everywhere, and as such it does not allow for any solution
different to the one corresponding to the Minkowskian 3D metric. However, let
us consider a different situation from the flat asymptotic: for asymptotically anti-
de Sitter (2 + 1) spacetime there exists a black hole solution found by Bañados,
Teitelboim, and Zanelli: the BTZ black hole, Bañados et al. (1992). Since that
original discovery, various classes of black hole solutions, in the presence of fields
and matter, have been reported in the literature.

1.1.8 Gravity in the Presence of Other Fields and Matter

3D gravity in the presence of other fields is worthy of deep study; many
achievements have been made since studies began. The present text is sim-
ply designed to show the existence of big families of exact solutions in three
dimensions and their parallelism, if any, with those classes of the standard
4D gravity. For instance, one may attempt a bridge between 3D conformally
flat metrics and nD conformally flat metrics for incompressible perfect flu-
ids, Friedmann–Robertson–Walker cosmology, and FRW dilaton–inflaton theory,
among others.

1.2 Algebraic Classification

In 4D gravity, to characterize adequately the gravitational field, there exist two
main classifications: the Petrov classification of the Weyl conformal tensor –
Petrov types of gravitational fields – and the Plebański–Pirani classification of
the tensor of matter or of the traceless Ricci tensor. In (2+1) gravity the situation
simplifies considerably: one is dealing with symmetric tensors for the matter from
one side, and, to characterize the conformal properties of the metric, the Cotton
tensor from the other.

1.2.1 Classification of the Cotton–York Tensor

The role of the conformal tensor in (2 + 1) gravity is played by the Cotton
tensor, Cotton (1899); see Stephani et al. (2003) for a more recent reference. In
8 Introduction

3D geometry, the conformal property of the space is guaranteed by the vanishing

of the conformal Cotton tensor. This tensor is defined by means of the covariant
derivatives of the Ricci tensor and of the scalar curvature according to
 
1
C αβ = C βα = η μν(α Rβ) μ − R δ β) μ , (1.22)
4 ;ν

where the symmetry has been introduced explicitly. Notice that the Cotton
tensor is traceless:
C α α = 0. (1.23)
To classify the Cotton tensor with respect to its eigenvalues, one has to solve a
generalized eigenvalue problem:
 αβ 
C − λ g αβ Vβ = 0 , C [αβ] = 0 , C αβ gαβ = 0 . (1.24)
By lowering one index, one can reformulate this task as an ordinary eigenvalue
problem for the matrix Cα β . However, in that case, the symmetry C αβ = C βα
is no longer present:
 β 
Cα − λ δαβ Vβ = 0 , Cα α = 0 . (1.25)
Accordingly, the matrix Cα β is no longer symmetric and the roots of the
characteristic polynomial
 
det Cα β − λ δαβ = 0 (1.26)
may be complex. This point seems to have been overlooked by Barrow et al.
(1986).
I will present a classification of Cα β along the lines of Garcia–Hehl–Heineke–
Macı́as (GHHM), Garcı́a et al. (2004), where the components are referred to an
orthonormal basis:
g = gαβ dxα dxβ = ηab Θa Θb , (ηab ) = diag(−1, 1, 1). (1.27)
The trace-free condition (1.25)2 reads explicitly
C1 1 + C2 2 + C3 3 = 0 . (1.28)
Accordingly, we can eliminate C3 3 , e.g., from (1.25)1 . Then the secular determi-
nant reads
 
 C1 1 − λ C1 2 C1 3 
 
det  −C1 2 C2 2 − λ C2 3  = 0,
 (1.29)
 −C1 3
C2 3
−C1 − C2 2 − λ
2 

with the five matrix elements C1 2 , C1 2 , C1 3 , C2 2 , C2 3 . The equation to determine

the eigenvalues λ amounts explicitly to
λ3 + b λ + c = 0 , (1.30)
where
b : = −(C1 1 )2 − C1 1 C2 2 − (C2 2 )2 + (C1 2 )2 + (C1 3 )2 − (C2 3 )2 , (1.31)
 1.2 Algebraic Classification 9

c : = (C1 1 )2 C2 2 + C1 1 (C2 2 )2 + C1 1 (C1 2 )2 + C1 1 (C2 3 )2 (1.32)

+(C1 ) C2 + 2 C1 C1 C2 − (C1 ) C2
2 2 2 2 3 3 3 2 2
.
The roots of (1.29) are given by
√ √
A 3 A 3
λ1 = A , λ2 = − + i B, λ3 = − − i B, (1.33)
2 2 2 2
with
D2 − 12b D2 + 12b 1/3
A := , B := , D := −108c + 12 12b3 + 81c2 .
6D 6D
A cubic polynomial with real coefficients has at least one real root and the
complex roots have to be complex conjugates. The Petrov types, Jordan normal
forms and Segré notations of the Cotton tensor read:
Table 1.2.1 Algebraic classification of the Cotton tensor

“Petrov” type Jordan form Segré notation eigenvalues relation

⎛ ⎞
λ1 0 0
I ⎝ 0 λ2 0 ⎠ [111] λ1 = λ2 , λ3 = −λ1 − λ2
0 0 −λ1 − λ2
⎛ ⎞
λ1 0 0
D ⎝ 0 λ1 0 ⎠ [(11)1] λ1 = λ2 = 0, λ3 = −2λ1
0 0 −2λ1
⎛ ⎞
λ1 1 0
II ⎝ 0 λ1 0 ⎠ [21] λ1 = λ2 = 0, λ3 = −2λ1
0 0 −2λ1
⎛ ⎞
0 1 0
N ⎝ 0 0 0 ⎠ [(21)] λ1 = λ2 = λ3 = 0
0 0 0
⎛ ⎞
0 1 0
III ⎝ 0 0 1 ⎠ [3] λ1 = λ2 = λ3 = 0
0 0 0
⎛ ⎞
0 0 0
O ⎝ 0 0 0 ⎠
0 0 0

This parallels exactly the Petrov classification of the Weyl tensor in four dimen-
sions in Stephani et al. (2003). This comes about since the Weyl tensor in 4D
is equivalent to a (complex) 3 × 3 trace-free matrix, as Cα β in 3D; for a similar
classification of Cαβ , see Hall and Capocci (1999). A detailed derivation of the
Cotton tensor in any dimension together with an account of its properties is
presented here in Chapter 20; see also Garcı́a et al. (2004).

1.2.2 Classification of the Energy–Momentum Tensor

The standard classification of the energy–momentum tensor Tab takes advantage
of its symmetry property, Tab = Tba , where the Latin letters denote the indices
10 Introduction

with respect to the orthonormal basis (1.27). For that reason the eigenvectors
are found by solving the matrix equation
Tab V b = λ ηab V b , → (Tab − λ ηab ) V b = 0. (1.34)
Searching the values of λ that cancel the determinant of the matrix
⎡ ⎤
T11 + λ T12 T13
⎢ ⎥
⎢ T12 T22 − λ T23 ⎥, (1.35)
⎣ ⎦
T13 T23 T33 − λ
namely the roots of the eigenvalue polynomial
λ3 + c2 λ2 + c1 λ + c0 = 0,
c0 := T11 , T22 T33 − T13 2 T22 + 2 T12 T23 T13 − T12 2 T33 − T11 T23 2 ,
c1 := −T11 T22 − T11 T33 + T12 2 + T22 T33 + T13 2 − T23 2 ,
c2 := T11 − T22 − T33 , (1.36)
which allows for three roots, with its possible degenerations,
c2 1√ √
−3
λ1 = − Rd − 6 F Rd,
3
+
3 6
c2 1 √ √
−3 1 √ √ √
−3
λ2,3 = − − Rd + 3 F Rd ±
3 3
i 3 Rd + 36 F Rd ,
3 12 12
Rd := 36 c1 c2 − 108 c0 − 8 c2 3 + 12 D,
D := 12 c1 3 − 3 c1 2 c2 2 − 54 c1 c2 c0 + 81 c0 2 + 12 c0 c2 3 ,
1 1
F = c1 − c2 2 , (1.37)
3 9
one would be able to determine the eigenvectors corresponding to each root.
The nomenclature used for eigenvectors and algebraic types of tensors is bor-
rowed from Plebański (1964): timelike, spacelike, null, and complex vectors are
denoted respectively by T, S, N, and Z. For algebraic types are used the sym-
bols: {λ1 T, λ2 S2 , λ3 S3 } ≡ {T, S, S}, meaning that the first real eigenvalue λ1
gives rise to a timelike eigenvector T, the second real eigenvalue λ2 is associ-
ated with a spacelike eigenvector S2 , and finally the third real eigenvalue λ3 is
related to a spacelike eigenvector S3 ; for the sake of simplicity I use the sym-
bols {T, S, S}. It is clear that {N, N, S} stands for the algebraic type allowing
for two different real eigenvalues giving rise to two null eigenvectors while the
third real root is associated with a spacelike eigenvector. When there are single
and double real eigenvalues giving rise correspondingly to timelike and spacelike
eigenvectors, the algebraic type is denoted by {T, 2 S}; consequently, for a triple
real eigenvalue, if that were the case, the types could be {3T }, {3N }, or {3S}.
For a complex eigenvalue λZ , in general, the related eigenvectors are complex
and are denoted by Z and Z̄ for its complex conjugate; the possible types are
{T, Z, Z̄}, {N, Z, Z̄}, or {S, Z, Z̄}.
 1.3 Brown–York Energy, Mass, and Momentum Stationary Metrics 11

In standard (3+1) gravity, the classification of the tensor of matter is replaced

by the classification of the traceless Ricci tensor S α β = Rα β − 14 Rδ α β , S α α = 0;
this is because the number of algebraically independent components of the Weyl
tensor is 10, of the traceless Ricci tensor is 9, and of the scalar curvature is
1 – which total 20: the number of independent components of the Riemann
tensor.

1.2.3 Classification of the Traceless Ricci Tensor

In the same spirit, one may classify the traceless symmetric 3D Ricci tensor,
leaving aside the scalar curvature:
1
S α β = Rα β − Rδ α β , S α α = 0. (1.38)
3
Since this tensor S α β shears the same properties owed by the Cotton tensor in
3D, then established algebraic classification for the Cotton tensor extends and
holds for the S α β tensor too. Using the above-proposed notation, the algebraic
types for the Cotton tensor can also be given as: {T, S, S}, {T, N, N }, {S, Z, Z̄},
and so on.

1.3 Brown–York Energy, Mass, and Momentum for Stationary

Metrics
In this section we establish the general forms of the energy and momentum
functions for metrics with non-flat anti-de Sitter asymptotics following the
Brown–York approach (Brown and York Jr., 1993); see also Brown, Creighton,
and Mann (Brown et al., 1994). The (2 + 1)-dimensional stationary cyclic
symmetric metric to be used is given by

ds2 = −N 2 dt2 + L−2 dr2 + K 2 [dφ + W dt]2 , (1.39)

where the structural functions N , L, K, and W depend on the variable r. The

timelike vector uμ normal to the hypersurface Σ : tΣ = const. and the spacelike
2
vector nμ normal to the surface B : r 2 = R = const. are given by
B

1 μ W μ
uμ = −N δμt , uμ = δ − δ ,
N t N φ
1 r μ
nμ = δ , n = Lδrμ . (1.40)
L μ
Therefore the projection metrics are:

ds2 |Σ:t=const. = L−2 dr2 + K 2 dφ2 = hij dxi dxj ,

ds2 | 2 = −N 2 dt2 + K 2 (dφ + W dt)2 = γAB dxA dxB ,
B :r=R=const.
ds2 |B:t=const.,r=R=const. = K 2 dφ2 = σıj dxı dxj . (1.41)
12 Introduction

The components of the projection tensor h are explicitly given by

hμν = gμν + uμ uν ,
hμν = K 2 W 2 δμt δνt + W K 2 (δμt δνφ + δμφ δνt ) + L−2 δμr δνr + K 2 δμφ δνφ ,
hμ ν = W δμt δφν + δμr δrν + δμφ δφν , hi j = δij , det(hij ) = K 2 /L2 . (1.42)
The components of the projection tensor γ amount to
γμν = gμν − nμ nν ,
γμν = −(N 2 − K 2 W 2 )δμt δνt + W K 2 (δμt δνφ + δμφ δνt ) + K 2 δμφ δνφ ,
γμ ν = δμt δtν + δμφ δφν , det(γAB ) = −K 2 N 2 . (1.43)
Notice that the indices μ, ν associated with the 3D spacetime can be replaced
by indices A, B running 0 ∼ t, 3 ∼ φ. The components of the projection tensor
σ amount to
σμν = gμν + uμ uν − nμ nν ,
σμν = K 2 W 2 δμt δνt + W K 2 (δμt δνφ + δμφ δνt ) + K 2 δμφ δνφ , σıj = K 2 δıφ δjφ ,
σ μν = δφμ δφν /K 2 , σ ıj = δφı δφj /K 2 , det(σıj ) = K 2 , (1.44)
where ı, j run only 3 ∼ φ.
To evaluate extrinsic curvatures, one needs the expressions of the symmetric
Christoffel symbols, which amount to
1   1
Γt tr = 2
2N N,r − K 2 W W, r , Γt rφ = − 2
K 2 W, r ,
2N  2N
Γr tt = L2 N N,r − KW 2 K, r − K 2 W W, r ,
1
Γr tφ = − L2 K (2W K,r + KW, r ) ,
2
1
Γ rr = − L,r , Γr φφ = −L2 KK,r ,
r
L
1  
Γφ tr = 2
−2W KN N,r + W 2 K 3 W, r + 2N 2 W K, r + N 2 KW, r ,
2KN
φ 1  
Γ rφ = 2
W K 3 W,r + 2N 2 K, r , (1.45)
2KN
while all other components vanish. The extrinsic curvature to the hypersurface
Σ : t = const. is given by the spatial tensor Kμν , namely
Kμν = −hαμ ∇α uν = −hμβ g
βα
uν;α
1 2 1 2
Kμν = K W W, r (δμt δνr + δμr δνt ) + K W, r (δμφ δνr + δμr δνφ ),
2N 2N
1 2
K μν = L W, r (δφμ δrν + δrμ δφν ),
2N
1 2 2 1 2 2 1
Kμ ν = L K W W, r δμt δrν + L K W, r δμφ δrν + W, r δμr δφν ,
2N 2N 2N
(1.46)
 1.3 Brown–York Energy, Mass, and Momentum Stationary Metrics 13

thus the trace of Kμν is zero, Kμμ = 0.

The momentum tensor P μν = 21κ det(hij ) [Kαα hμν − K μν ] for the hypersur-
face Σ becomes
1
P μν = − LKW, r (δrμ δφν + δφμ δrν ), (1.47)
4κ N
while the surface momentum density vector jμ = −2σμν P να nα / det hij
amounts to
1 L 2 1 L 2
jμ = K W W, r δμt + K W, r δμφ . (1.48)
2κ N 2κ N
Consequently, the surface momentum density ja reduces in the studied case to
1 1 L 2
jφ = −2 σφφ P φ r = K W, r (1.49)
K 2κ N
modulo the additive constant related to the reference spacetime.
The energy density is evaluated by using the tensor

kμν = −σμα hβα nλ;β hλν = hμα Θαβ hβν , Θμν = −γμβ nν; β ,

which amounts to
φ t
kμν = −LKW K, r [W δμt δνt + 2δ(μ δν) ] − LKK, r δμφ δνφ , (1.50)

while
1 1
kμ ν = − LW K, r δφμ δνt − L K, r δφμ δνφ . (1.51)
K K
Rising with σφφ = K 2 = σ φφ one of the indexes of the component kφφ =
2
−LKK, r of the extrinsic curvature k associated with the metric of B , one arrives
at
1
k := kıı = σ φφ kφφ = − LK, r . (1.52)
K
Therefore the energy density  becomes
1 cl 1
= k|0 = − LK, r |R − 0 . (1.53)
κ κK
As far
 as the integral characteristics are concerned, the total quasilocal energy
√
E = B dx σ = 2π K  is given by
π
E = −2 LK, r |R − 2π K(R)0 , (1.54)
κ
while the total
 mass
∂ μ
related to the timelike Killing vector ξ μ = ( ∂t ) = δtμ ,
√
M ( ∂ t ) = − B dx σ( uμ + jμ )ξ amounts to
∂ μ

π πL 3
M (∂/∂t) = −2 N LK, r |R − K W W, r |R − 2π N K|R 0 . (1.55)
κ κN
14 Introduction
 √
Finally, the total momentum J( ∂∂φ ) = B dx σjμ ζ μ associated with the Killing
vector ζ μ = ( ∂∂φ )μ = δφμ , is given by
 2π
πL 3 1 J(R)
J(∂/∂φ) = dφ jφ K = K W, r |R , jφ = . (1.56)
0 κ N 2π K(R)
Incidentally, other representations of the mass and momentum density are:
M (∂/∂t) = N (R)E(R) − W (R)J(R). (1.57)
The extrinsic curvature Θμν = −γμα ∇α nν = −nν;α γ α μ of the surface boundary
2
B reduces to
Θμν = −L(KW 2 K, r + K 2 W W, r − N N, r )δμt δνt
t φ
−LK(2W K, r + KW, r )δ(μ δν) − LKK, r δμφ δνφ , (1.58)
with trace Θ equal to
L
Θ=− (KN, r + N K, r ), (1.59)
NK
which, used in the definition of the boundary momentum
1
π μν = − − det γAB (Θ γ μν − Θμν ),
2κ
taking into account that det γAB = −K 2 N 2 , gives
L L (μ ν)
π μν = − K, r δtμ δtν + (2W K, r + KW, r )δt δφ
2κ N 2κ N
L
+ (N N, r − W 2 KK, r − K 2 W W, r )δφμ δφν . (1.60)
2κ N K
This tensor is used in the construction of the stress tensor
2
sαβ = √ σ α π μν σνβ − s0 αβ . (1.61)
σN μ

1.3.1 Summary of Quasilocal Mass, Energy, and Angular

Momentum
Summarizing, to evaluate quasilocal mass, energy, and angular momentum of
spacetimes with asymptotics different from the flat case, in particular the anti-
de Sitter, one uses the quasilocal formalism developed in Brown and York Jr.
(1993); Brown et al. (1994). For a stationary cyclic symmetric metric of the form
g = −N (ρ)2 d T 2 + L(ρ)−2 dρ2 + K(ρ)2 [dΦ + W (ρ)d T ] ,
2

the surface energy density is given by

L(ρ) d
(ρ) = − K(ρ) − 0 , (1.62)
π K(ρ) dρ
 1.4 Decomposition with Respect to a Frame of Reference 15

where 0 is the reference energy density, which in the case of solutions with a
negative cosmological constant Λ = −1/l2corresponds to the density of the
2
anti-de Sitter spacetime, namely 0 = − πρ
1
1 + ρl2 . The momentum density is
determined from
K(ρ)2 L(ρ) d
j(ρ) = W (ρ). (1.63)
2 π N (ρ) dρ
The integral momentum J(ρ), global energy E(ρ), and integral mass M (ρ) are
correspondingly given by
J(ρ) = 2π K(ρ) j(ρ),
E(ρ) = 2π K(ρ) (ρ),
M (ρ) = E(ρ) K(ρ) − W (ρ) J(ρ). (1.64)
Evaluation of these physical quantities for the studied classes of solutions is
presented in the corresponding sections.

1.4 Decomposition with Respect to a Frame of Reference

Barrow, Shaw, and Tsagas (BST), Barrow et al. (2006), developed the covariant
decomposition of the spacetime with respect to a family of timelike observers
with worldlines tangential to the 3-velocity field ua in analogy to the standard
(3 + 1) covariant approach to general relativity introduced by Ehlers (1961); see
also Ellis and Elst (1999). This topic is included here for the sake of reference;
I retain the original presentation and wording with certain minor changes; for
instance, the energy density is denoted here by μ instead of ρ. Throughout this
section, Latin indices run the spacetime index values.

1.4.1 Kinematics of the Frame

The 3-velocity field determines the time direction and is normalized so that
ua ua = −1. The 2D space orthogonal to ua is defined by means of the projection
tensor
hab = gab + ua ub , (1.65)
with properties
hab = h(ab) , hab ub = 0, hab hb c = hac , ha a = 2. (1.66)
Using hab one also defines the covariant derivative operator of the 2D space as
Da = ha b ∇b . (1.67)
The irreducible kinematic variables, which describe the motion of the observers,
are obtained by decomposing the covariant derivative of the 3-velocity. This
splitting gives
1
∇a ub = σab + ω[ab] + Θ hab − u̇a ub , (1.68)
2
16 Introduction

where
the shear : σab = D(a ub) , σab ub = 0, (1.69a)

the vorticity : ωab = D[a ub] , ωab ub = 0, (1.69b)

the area expansion : Θ = Da ua = ∇a ua , (1.69c)

the acceleration : u̇a = ub ∇b ua , u̇a ua . (1.69d)

The meaning of these quantities is explained very well in the BST paper: “The
tensor
1
Da ub = hca hdb ∇c ud = σab + ωab + Θ hab (1.70)
2
describes changes in the relative position of the worldlines of two neighboring
observers. When the latter follow the motion of a fluid, the effect of Θ is to change
the area of a given fluid element, without causing rotation or shape distortion.
This scalar also defines the average scale factor a by
ȧ 1
= Θ. (1.71)
a 2
The shear monitors distortions in the element’s shape that leave the area unaf-
fected, while vorticity describes changes in its orientation under constant area
and shape. The symmetric and trace-free nature of σab ensures that it has only
two independent components, while the antisymmetry of ωab guarantees that the
vorticity tensor is determined by a single component. In other words, the shear
and the vorticity correspond to a vector and a scalar respectively. The latter
reflects the fact that the rotational axis has been reduced to a point. Defining
the 2D permutation tensor ab , with ab ub = 0, the vorticity scalar is
1
ω= ab ω ab , (1.72)
2
with ωab = ωab . Note that ab = ηabc uc , by definition, where ηabc is the
3D totally antisymmetric alternating tensor. The latter satisfies the condition
p q r
ηabc η pqr = 3!δ[a δb δc] , which ensures that ab cd = 2hb[a hdc] .”

1.4.2 Perfect Fluid Referred to a Frame of Reference

Suppose that the matter source of the 3D gravitational field governed by the
Einstein equations
1
Eab := Rab − gab R + Λ gab − κ Tab = 0 (1.73)
2
is a perfect fluid. Then, relative to an observer moving with 3-velocity ua , the
energy–momentum tensor of the material component takes the form
Tab = (μ + p) ua ub + p gab = μ ua ub + p hab , T = Taa = 2p − μ, (1.74)
 1.4 Decomposition with Respect to a Frame of Reference 17

where μ is the energy density, and p is pressure. Substituting the above into the
Einstein field equations (1.73), the latter read

Rab = 2(κp − Λ)ua ub + [κ(μ − p) + 2Λ]hab , R = 2κ(μ − 2p) + 6Λ.

(1.75)

The above also provides the following auxiliary relations:

Rab ua ub = 2(κp − Λ), hca hdb Rcd = [κ(μ − p) + 2Λ]hab , hba Rbc uc = 0. (1.76)

The Bianchi identities imply that

∇a Gab = 0 → ∇a Tab = 0. (1.77)

Explicitly, it reads

ub ua ∇a μ + (μ + p)ub ∇a ua + (μ + p)ua ∇a ub + hab ∇a p = 0

→ ub ua ∇a μ + (μ + p)Θ ub + (μ + p)u̇b + Db p = 0. (1.78)

The timelike and spacelike parts of this relation lead to the 3D fluid conservation
laws:
for the energy density : ua ∇a μ := μ̇ = −Θ(μ + p), (1.79)

for the momentum density : (μ + p)u̇a = −Da p. (1.80)

Notice that these conservation laws of a perfect fluid have the same functional
form as their 4D counterparts.
The intrinsic curvature of the 2D space orthogonal to ua is determined by
projecting the Riemann tensor, (1.2), onto the 2D space by means of hba :

Rabcd = hqa hsb hfc hpd Rqsf p − vac vbd + vad vbc ,
1
vab = Db ua = σab + ωab + Θhab , (1.81)
2
where vab is the relative position tensor. Note that vab characterizes the extrinsic
curvature (i.e., the second fundamental form of the space.) Assuming perfect-
fluid matter, i.e., equations (1.75), the projected Riemann tensor of the 2D
(spatial) sections reads
1
Rabcd = (κ μ − Θ2 + Λ)(hac hbd − had hbc )
4
1
− Θ [(σac + ωac )hbd + (σbd + ωbd )hac − (σad + ωad )hbc − (σbc + ωbc )had ]
2
−(σac + ωac )(σbd + ωbd ) + (σad + ωad )(σbc + ωbc ), (1.82)

with Rabcd = R[ab][cd] . Similarly, as in the standard (3 + 1) gravity, the isotropic

pressure does not contribute to the curvature of the space orthogonal to ua .
Also, in the absence of anisotropy (i.e., when σab and ωab vanish), the above
reduces to
18 Introduction

1
Rabcd = (κ μ − Θ2 + Λ)(hac hbd − had hbc ). (1.83)
4
Defining
Rab = Rc acb (1.84)
as the local 2D Ricci tensor, by contracting expression (1.82) one obtains the
following 3D analogue of the Gauss–Codacci formula
1
Rab = (κ μ − Θ2 + σ 2 − ω 2 ) hab , (1.85)
4
where the shear and rotation scalars are defined correspondingly as
2σ 2 = σab σ ab , 2ω 2 = ωab ω ab . (1.86)
In deriving the above result (1.85), the relations
σc[a ω c b] = 0, σc[a σ c b] = 0, ωc[a ω c b] = 0 (1.87)
1 2
have been used. The former holds because σ12 (ω 1 + ω 2 ) = 0 (i.e., the sin-
gle independent component vanishes due to the trace-free nature of the shear).
Similarly, the two independent components of
σc(a σ c b) = 0
are also identically zero. Last, the result ωc(a ω c b) = 0 is guaranteed by the
relation ωab = ωηab and the properties of ηab . The absence of a skew and also of
a symmetric and trace-free part from Eq. (1.84) agrees with the symmetries of
the Riemann tensor in 2D spaces.
Finally, the trace of (1.84) leads to the curvature scalar of the spatial sections,
which may also be seen as the generalized Friedmann equation for 3D spacetimes:
1
R = Raa = 2(κ μ − Θ2 + σ 2 − ω 2 + Λ). (1.88)
4
Besides these theoretical aspects of fluids, the paper by Barrow et al. (2006)
deals with the kinematics of fluid matter, too. However, that material is outside
the scope of the present work.
 2
Point Particle Solutions

The first publication (Staruszkiewicz, 1963) on (2 + 1) gravity dated back to

1963, and it was devoted to the description of static solutions determined by
point sources.
In (2 + 1) gravity, it is assumed that the spacetime is 3D and obeys the 3D
Einstein equations. The flat spacetime, Rαβγδ = 0, is the field solution to the
vacuum Einstein equations
Gμν = 0 → Rμν = 0 = R, (Tμν = 0 and Λ = 0.)
In Staruszkiewicz’s wording: 3D gravitation theory is a theory without a field of
gravitation; where no matter is present, space is flat. Curvature can arise only
if matter or energy are present. Point particles are the most simple material
objects to be considered in 3D; they move along geodesics, and do not interact
with each other; their world-lines are lines of conical singularities.

2.1 Staruszkiewicz Point Source Solutions

This section answers the question: What does the gravitational field of a point
mass look like? The source of the Einstein equations is located along a single time-
like world line (a geodesic) on which curvature may be present. The spacetime
is certainly static, thus for a foliation t = const. the spatial hypersurface is flat
except for a single point. Hence, one is dealing with a cone; or, more precisely,
with a family of cones with vertexes on the world line of each point particle, if
there are many. Choosing the static, circularly symmetric Schwarzschild metric
g = −A(r)2 dt2 + B(r)2 dr2 + r2 dφ2 , 0 ≤ r ≤ ∞, 0 ≤ φ < 2π, (2.1)
the Einstein equations Rμν = 0 requires A(r) = A0 = const, and B(r) = B0 =
const. Adopting a new time coordinate τ = A0 t one gets
g = −dτ 2 + B02 dr2 + r2 dφ2 , 0 ≤ r ≤ ∞, 0 ≤ φ < 2π, (2.2)
20 Point Particle Solutions

Next, accomplishing the coordinate transformation

r = ρ/B0 , φ = Φ B0 (2.3)
one arrives at the (2 + 1)–Minkowski metric in polar coordinates
g = −dτ 2 + dρ2 + ρ2 dΦ2 , 0 ≤ ρ ≤ ∞, 0 ≤ Φ < 2π/B0 . (2.4)
The surface τ = const describes, because of 0 ≤ Φ < 2π/B0 , a cone with vertex
at ρ = 0 and deficit angle defΦ := 2π(1 − 1/B0 ).
In Gott and Alpert (1984), the metric (2.2) is embedded as a cone z(r) =
(B02 − 1)1/2 r in the Minkowski (3 + 1) spacetime:
g = −dt2 + dz 2 + dr2 + r2 dφ2 , cone: z(r) = (B02 − 1)1/2 r. (2.5)

2.1.1 Relationship Between the Deficit Angle and Mass

The deficit angle is related to the mass of the point particle. To establish this
fact, let us consider a perfect fluid source with energy–momentum tensor Tμν =
(ρ + p)uμ uν + p δμν , uμ = A(r)δμt . The Einstein equations for the metric (2.1) are
 r
dB −2
− + κ B r ρ (r) = 0 → B = B0 − 2κ
3
s ρ(s) d s,
dr 0
dA d2 A dA dB
− κ p (r) B 2 A r = 0, − 2 B + + κ p (r) A B 3 = 0. (2.6)
dr dr dr dr
For dust p = 0, one has
 r
B −2 = B0 − 2κ M (r), M (r) = s ρ(s) ds,
0
A (r) = A0 , A (r) → 1. (2.7)
In Cornish and Frankel (1991) one finds that for a point source ρ = M δ(r), and
p = 0,
κ
(e−2η ) = B −2 = 1 − M, κ3 = 2π G, (2.8)
π
which results from the integration of
 r  2π  r
1
r M δ(r)d r = r M δ(r)d r d φ
2π 0
0
 x y0
1 M
= M δ(x)δ(y)d x d y = .
2π 2π
Thus, one has a conical geometry
 
dr2 κ κ
g = −dt +
2
+ r2 dφ2 , r = r̃ 1 − M , φ = φ̃/ 1 − M ,
1 − πκ M π π

κ
g = −dt2 + dr̃2 + r̃2 dφ̃2 , 0 ≤ r̃ ≤ ∞, 0 ≤ φ̃ < 2π 1 − M , (2.9)
π
 2.2 Staruszkiewicz Single Point Source Solution 21

with a deficit angle proportional to the mass M

  
κ
defφ̃ := 2π 1 − 1 − M ≈ κ M + O((κM )2 ), (2.10)
π

which geometrically corresponds to remove a wedge of angle defφ̃ ≈ κ M with

the identification of the cut edges.
This Dirac δ point source has been analyzed in detail in Cornish and Frankel
(1991), where the lack of a Newtonian limit is linked to the absence of acceleration
in the geodesic equations for particles initially at rest. Also one finds some lines
on this single point particle solution in the appendix of Clément (1976); this
article is devoted to extended particles in 2D space.

2.2 Staruszkiewicz Single Point Source Solution

A static cyclic symmetric (2 + 1)-dimensional metric can be given always in
isotropic coordinates as
 
g = −A(x, y)2 dt2 + e2 η(x,y) dx2 + dy 2 . (2.11)

The Einstein equations Gνμ = 0 yield

A(x, y) = A0 → 1,
 2 
−2η ∂ η ∂2η
0
G0 = e + = 0 → η = F (x + iy) + F ∗ (x + iy), (2.12)
∂y 2 ∂x2
because of the flat Laplace character of the partial equation for η(x, y). In
particular, for a single point particle located at the point (x1 , y1 ), one may choose
−α
η(x, y) = ln (x − x1 )2 + (y − y1 )2 . (2.13)

Thus, the metric (2.11) becomes

−2α  2 
g = −dt2 + (x − x1 )2 + (y − y1 )2 dx + dy 2 . (2.14)

Accomplishing the coordinate transformation x = x1 + r cos θ, y = y1 + r sin θ,

one gets
 
g = −dt2 + r−2α dr2 + r2 dθ2 , (2.15)

carrying out a second transformation r = ρa one arrives at

g = −dt2 + a2 ρ2a(1−α)−2 dρ2 + ρ2a(1−α) dθ2

→ g = −dt2 + a2 dρ2 + ρ2 dθ2 , for a(1 − α) = 1. (2.16)

Comparing with the conical metric (2.9) one establishes that

 
κ κ κ
a = 1/ 1 − M , α = 1 − 1 − M ≈ M,
π π 2π
22 Point Particle Solutions

κ
g = −dt + dρ / 1 − M + ρ2 dθ2 ,
2 2
(2.17)
π

which explicitly gives the relation between the exponent α for the single point
particle solution in isotropic coordinates with a mass M :
−2α  2 
g = −dt2 + (x − x1 )2 + (y − y1 )2 dx + dy 2 ,

κ defθ
α=1− 1− M = . (2.18)
π 2π

2.2.1 No Parallelism With the (3 + 1) Schwarzschild Solution

The Schwarzschild solution is thought of as an external gravitational field
produced by a static mass. Nevertheless, it is well known that the origin
of the Schwarzschild coordinates, r = 0, is a singularity, which, all the
worse, is spacelike, and as such cannot be associated with a Dirac δ point
source.
Assume for a while that the Schwarzschild mass is located at the ori-
gin r = 0, as it is in the book by Ortı́n (2004), (§7.2), p. 202, and that
it follows a world-line with timelike 4-velocity uμ . Hence, one can repre-
sent it by means of an energy–momentum δ-tensor: Tμν = m uμ uν δ(x).
The Einstein equations in Schwarzschild coordinates evaluated in the limit
as r → 0 yield not only the Gtt different from zero proportional to
Ttt = mδ (3) (r), but the remaining diagonal components are proportional
to a δ-function; this is related to the spacelike nature of the Schwarzschild
singularity.

2.3 Staruszkiewicz Two Point Sources Solution

The static cyclic (2 + 1)-dimensional metric for two point masses can be given
in isotropic coordinates as
   
g = −dt2 + e2 η(x,y) dx2 + dy 2 = −dt2 + B(x, y)2 dx2 + dy 2 ,
 −α1 /2  −α2 /2
2 2 2 2
B (x, y) = B0 2 (x − x1 ) + (y − y1 ) (x − x2 ) + (y − y2 ) ,
(2.19)

where (x1 , y1 ) and (x2 , y2 ) stand for the spatial location of the first and second
point masses m1 and m2 , respectively, α1 = m m2
κπ and α2 = κπ . The proper
1

Staruszkiewicz two point sources solution corresponds to the arrangement of the

masses along the real axis of z, i.e., with y1 = 0 = y2 .
 2.4 Deser–Jakiw–’t Hooft Static N Point Sources Solution 23

2.4 Deser–Jakiw–’t Hooft Static N Point Sources Solution

The static N point sources solution derived by Deser, Jackiw and ’t Hooft (Deser
et al., 1984) is described, in isotropic coordinates, by the metric
   
g = −dt2 + e2 η(x,y) dx2 + dy 2 = −dt2 + Φ(x, y) dx2 + dy 2 . (2.20)
The Einstein tensor corresponding to this metric is given by:
 2 
−2η ∂ η ∂2η 1
μ μ 0 0 0
Gν = δ0 δν G0 , G0 = e 2
+ 2
= e−2η Δη = Δ ln Φ,
∂y ∂x 2Φ
1
G = Gμμ = − R = G00 . (2.21)
2
For vacuum Gμν = 0, the Cartesian Laplace equation allows for real solutions of
the form
η = F (x + iy) + F (x − iy) = F (z) + F (z ∗ ), Φ = f (z) f (z ∗ ), z := x + iy,
which made evident the flatness of the metric in this case
 z
∗ ∗ ∗
g = −dt + f (z)f (z ) dz dz = −dt + dZ d Z , Z :=
2 2
f (z) dz.

For n static point particles, the DJ’tH solution is given by

  −αn /2
|r − rn |−αn = C0 2
2 2
Φ(x, y) = C0 2 (x − xn ) + (y − yn ) ,
n n
κ κ
2η = ln Φ = − mn ln |r − rn | + ln C02 , αn := mn , (2.22)
π n π

where κ = 8π GE . This choice is dictated by the δ-function character of the point

sources; from the properties of the Dirac δ-function; see for instance Infeld and
Plebański (1960), one has
n = 3 : Δln r = 2 π δ 2 (r),
1
n ≥ 3 : Δ n−2 = −(n − 2)Ωn δ n (r), (2.23)
r
where Ωn is equal to the area of the sphere Sn of unitary radius. Hence, for the
DJ’tH solution one gets
κ 
Δ ln Φ = − mn Δ ln |r − rn | = −2κ mn δ 2 (r − rn ). (2.24)
π n n

2.4.1 Energy and Euler Invariant

The parameters mk stand for the masses of the point particles. Notice that
√ 1√ 1
−g G00 = − −g R = Δ ln φ, (2.25)
2 2
24 Point Particle Solutions

hence, the energy and Euler invariant, see Eqs. (3.1–3.2) of Deser et al. (1984)
and comments therein, read
  
2 √ 2 √
2κ E = −2 d x −g G0 = 0
d x −gR = − d 2 xΔ ln Φ
  
= 2κ d 2x mn δ 2 (r − rn ) = 2κ mn . (2.26)
n n

2.4.2 Energy–Momentum Tensor for N Point Particles

For point particles with masses mk located at rest at rk the energy–momentum
tensor may be given as
Tμν = m(x, y) δμ0 δν0 , Tνμ = − m(x, y) δ0μ δν0 , T = Tμμ = − m(x, y),
The Einstein equation G00 = κ T00 yields
1 1 
Δ ln Φ = −κ m (x, y) → − 2κ mn δ 2 (r − rn ) = −κ m (x, y).
2Φ 2Φ n

Therefore

Tμν = Φ−1 mn δ 2 (r − rn ) δμ0 δν0 = Φ−1 Tμν DJtH (2.27)
n

which is a weighted version of the T μν of Deser et al. (1984); compare with

DJ’tH(2.5).

2.5 Clément Rotating Point–Particles Solution

Clément (1985) reported two families of stationary solutions to 3D vacuum
Einstein equations: the point particles class and a second class which suffers
unsatisfactory physical interpretation. Following Clément, the stationary metric
(changing the signature) is given as
ds2 = −h2 (dt + ωi dxi )2 + ḡij dxi dxj , (2.28)
referring to Landau and Lifshitz (1970), where the structural functions depend
on space coordinates xi . The arbitrariness of the vector potential ωi , via the
transformation t → t + F (r), can be restricted to
ω i ;i = 0, (2.29)
and the spatial indices are raised and lowered by means of ḡij . The vacuum
Einstein equations can be written as
h4
R00 = 0 : −h h;i ;i + fij f ij = 0, (2.30a)
4
1  
R0i = 0 : − h−1 h3 f ij ;j = 0, (2.30b)
2
 2.5 Clément Rotating Point–Particles Solution 25

h2 ik j
Rij = 0 : − f f k − h−1 h;i;j + R̄ij = 0, (2.30c)
2
where
fij = ωj;i − ωi;j , (2.31)

and R̄ij is constructed with the spatial metric ḡ ij . In 3D spacetime, for the
spatial metric one can choose the isotropic coordinates such that

ḡij = e2 u δij . (2.32)

The equation (2.29) yields

ωi,j = 0 → ωi = ij ∂j φ → fij = −ij Δφ, (2.33)

where Δ is the 2D flat Laplace operator.

The equations (2.30b) leads to
 
ij e−2 u h3 Δφ ,j = 0 → Δφ = λ0 e2 u h−3 , (2.34)

where λ0 is a constant. Substituting this result in (2.30a), one gets

λ20 2 u
h3 Δ h = − e . (2.35)
2
The trace and the traceless part of (2.30c) give
3 2 −4
e−2 u Δ u = λ0 h , (2.36)
4

1
h;i;j − h;k ;k ḡij = 0, (2.37)
2
Introducing the complex variable z = x + iy, the equation (2.37) becomes
∂2 ∂u ∂h ∂h
2
h=2 → (z, z̄) = μ(z̄) e2 u(z,z̄) , (2.38)
∂z ∂z ∂z ∂z
where μ(z̄) is an arbitrary function. The most interesting class of solutions arises
for μ(z̄) = 0; then h is a constant, which without loss of generality can be chosen
equal to 1, h = 1. Then, the equation (2.35) implies λ0 = 0. Consequently, the
Einstein equations, (2.34) and (2.36), become

Δφ = 0 = Δ u. (2.39)

The multicenter solutions of these equations

κ 
φ=− Jα ln |z − aα |
2π α
κ 
u=− mα ln |z − aα | (2.40)
2π α
26 Point Particle Solutions

(where κ = 8πG) lead to the metric

 2
κ  ij (xi − ai α ) dxj 
ds = − dt +
2
Jα + |r − aα |−κmα /π dr · dr,
2π α |r − aα |2
α
(2.41)
representing a system of point particles of “masses” mα , and “spins” Jα .
In the case of a single massless spinning point particle at the origin, the above
metric gives rise to
κ 2
ds2 = − dt + J dθ + dr2 + B 2 r2 dθ2 , (2.42)
2π
therefore one recovers the rotating Deser–Jakiw–t’Hooft solution (Deser et al.,
1984). An interpretation of this spacetime is given in Carlip (1998) Eqs. (3.20)–
(3.25), where the coordinates {t, θ} are subjected to the transformations
κJ
t → t = t + θ, θ → θ  = B θ,
2π
therefore, the spacetime experiences a conical singularity together with a
timeshift.
Of course, each one of these point particle solutions possesses a vanishing
Riemann tensor as a consequence of the vacuum character of the spacetimes,
hence all these multiple point particle spaces are Minkowski flat; although it
may sound redundant, the Cotton tensor has zero value.
 3
Dust Solutions

This chapter is devoted to the derivation of dust solutions in (2 + 1) gravity,

i.e., solutions to Einstein’s equations for an energy momentum tensor Tαβ =
μ(xν )uα uβ , where uα , uα uα = −1, is the dust particles’ velocity.

3.1 Cornish–Frankel Dust Heaviside Function Solution

In Cornish and Frankel (1991) there were reported extended static Heaviside
function dust distrinbutions of the form
M
κ μ (r) = 2π G H(r − R0 ), κ := 2π G, (3.1)
πR0 2
with embedding line element

ds2 = −dt2 + dz 2 + dr2 + r2 dθ2 ,

for the embedding function z

 
R0 M r2
z(r) = 1− 1−κ , 0 ≤ r ≤ R0 , (3.2)
κ M/π π R02

which gives a (2 + 1) metric

 −1
M r2
ds2 = −dt2 + 1 − κ dr2 + r2 dθ2 , 0 ≤ r ≤ R0 ,
π R02
with value at the border
R0
z(r = R0 ) = (1 − 1 − κ M/π),
κ M/π
and energy momentum tensor for dust

Tαβ = μ(r)δαt δβt .

28 Dust Solutions

On the other hand, the external region has an embedding function Z given by
 
κ M/π 1 M
Z(r) = r − R0 1− 1−κ ,
1 − κ M/π κ M/π 1 − κ M/π π
r ≥ R0 , (3.3)

such that

ds2 = −dt2 + dZ 2 + dr2 + r2 dθ2

 −1
M
= −dt2 + 1 − κ dr2 + r2 dθ2 , r ≥ R0 , Z(r = R0 ) = z(r = R0 ).
π

with zero energy momentum tensor, although a conical singularity is present.

3.2 Giddings–Abott–Kuchař Dust Solutions

This solution has been sought for irrotational dust, which geometrically means
that the world lines of the dust are hyperface forming, i.e., the field of dust veloc-
ity is tangential to the generatrices of the hypersurface. Consequently, following
the Giddings–Abott–Kuchař work in Giddings et al. (1984), one can introduce
a Gaussian coordinate system attached to the world lines of the dust such that
the time coordinate is comoving with the dust. Since any 2D space is always
conformally flat space, the isotropic coordinates can be used, hence the metric
to deal with is
 
g = −dt2 + e2 Ω(t,x,y) dx2 + dy 2 , (3.4)

thus the corresponding Einstein equations for dust Tλν = κ m(t, x, y) δλt δνt , are
   2
∂2 ∂2 ∂
Ett :→ e−2 Ω Ω + Ω − Ω = −κ m (t, x, y) , (3.5a)
∂y 2 ∂x2 ∂t
∂2
Ext = 0 = Etx :→ Ω = 0 → Ω = F2 (t, y) + F1 (x, y) , (3.5b)
∂x∂t
∂2
Eyt = 0 = Ety :→ Ω = 0 → Ω = F3 (t, x) + F1 (x, y) , (3.5c)
∂y∂t
 2
∂2 ∂ ∂ ∂
Exx = 0 = Eyy , → 2 Ω + Ω = 0, → ( Ω)−1 = 1
∂t ∂t ∂t ∂t
→Ω = ln (t f (x, y) + g(x, y)) . (3.5d)

One may distinguishes two branches of solutions:

a) the time-dependent class f (x, y) = 0,
b) and the static case f (x, y) = 0.
 3.2 Giddings–Abott–Kuchař Dust Solutions 29

3.2.1 Time-Dependent Class of Dust Solutions Ω = ln (tf (x, y))

The equations (3.5b) and (3.5c), after substitution of Ω from (3.5d), yield
∂ g ∂ g
=0= → g = c0 f → Ω = ln ((t + c0 ) f (x, y)) ;
∂x f ∂y f
hence, shifting the time coordinate, one has without lost of generality

Ω(t, x, y) = ln (t f (x, y)) . (3.6)

Substituting this Ω(t, x, y) into (3.5a) one arrives at

∂2 ∂2
f −2 ( ln f + ln f ) = 1 − κ t2 m (t, x, y)
∂y 2 ∂x2
1
→ κ m (t, x, y) = 2 (1 − f −2 ∇ 2 ln f ). (3.7)
t
Moreover, the conservation law T μν ;ν = 0 gives rise to a single equation for
m(t, x, y), namely
∂m ∂ ∂m m 1
+ 2m Ω = 0 → +2 = 0 → m(t, x, y) = 2 μ0 (x, y). (3.8)
∂t ∂t ∂t t t
Therefore, the structural function f (x, y) is the only remaining to be determined
function subjected to the partial nonlinear differential equation

f −2 ∇ 2 ln f = 1 − κ μ0 (x, y) . (3.9)

The Riemann tensor is equal to

3 2 3 2
Rαβμν = R3232 δ[α δβ] δ[μ δν] = κ μ0 (x, y) f 4 t2 δ[α
3 2 3 2
δβ] δ[μ δν] , (3.10)

    2
−2 Ω ∂2 ∂2 ∂ 4
R3232 = − e + Ω− Ω f (x, y) t4 = κ μ0 f 4 t2 .
∂y 2 ∂x2 ∂t
(3.11)

The evaluation of the Cotton yields

⎡ ⎤
0 1 ∂m
4 ∂y − 14 ∂m
∂x
⎢ ⎥
(C α β ) = 2κ ⎢ 1 −2 Ω ∂m
⎣ −4 e ∂y 0 m ∂Ω
2 ∂t + 1 ∂m
4 ∂t
⎥,
⎦ (3.12)
1
4 e−2 Ω ∂m
∂x −m
2
∂Ω
∂t − 1 ∂m
4 ∂t 0

thus, taking into account that Ω = tf (x, y), and μ = μ0 (x, y)/t2 , one gets
⎡ ⎤
∂y μ0 (x, y) − ∂x μ0 (x, y)
∂ ∂
0
κ ⎢ ⎥
(C α β ) = 2 ⎢ − e−2 Ω ∂y
∂
μ0 (x, y) 0 0 ⎥. (3.13)
2t ⎣ ⎦
e−2 Ω ∂x
∂
μ0 (x, y) 0 0
30 Dust Solutions

The search for its eigenvectors yields

λ1 = 0; V1 = [1, 0, 0], V 1μ = −δμt , V 1μ V 1μ = −1, V1 = T1,

−Ω ∂μ0 2 ∂μ0 2 ∂μ0 1 −Ω ∂μ0 1
λ2 = i e ( ) +( ) ; V2 = [1, −e−Ω ,e ] = Z,
∂x ∂y ∂y λ2 ∂x λ2
λ3 = λ̄2 ; V3 = Z̄, (3.14)
therefore the corresponding tensor type is Type I: {T, Z, Z̄}.
The eigenvectors V2 and V3 are complex conjugated, while the zero eigenvalue
λ1 gives raise to a timelike eigenvector V1, which points in the t-coordinate
direction.

3.2.2 Static Class of Dust Solutions Ω = ln g(x, y)

The only remaining equation to be solved is the resulting from equation (3.5a)
from the substitution of Ω = ln g(x, y), namely
g −2 ∇ 2 ln g = −κ m (x, y) . (3.15)
Thus, for a given structural function g(x, y) one determines the density m(x, y),
which is required to fulfill the energy conditions, among them, the positiveness
of the mass. This backward way of looking at this problem is an alternative too.
For instance, for a set of point sources – see the previous section 2.4 and Deser
et al. (1984) – one arrives at
 1
|r − rn |− π mn ; 2 Δ ln g = −κ m (x, y) ,
κ
g(x, y)2 = C0 2
n
g
2 
m (x, y) = 2 mn δ 2 (r − rn ). (3.16)
g n

3.3 Barrow–Shaw–Tsagas Anisotropic Dust Solution; Λ = 0

The simplest (2 + 1) homogeneous and anisotropic Bianchi I–type universe, in
the terminology of Barrow et al. (2006), is given by the spacetime metric
ds2 = −dt2 + A2 (t)dx2 + B 2 (t)dy 2 . (3.17)
For this metric, the Einstein equations for dust, with energy momentum tensor
Tαβ = ρ(t) uα uβ , uα = δαt ,
yield
ȦḂ
Ett = 0 → κ ρ(t) = , (3.18)
AB
2
d B
Exx = 0 → = 0 → B(t) = B1 t + B0 , (3.19)
dt2
 3.3 Barrow–Shaw–Tsagas Anisotropic Dust Solution; Λ = 0 31

d2 A
Eyy = 0 → = 0 → A(t) = A1 t + A0 , (3.20)
dt2
where Ai , Bi , i = 0, 1 are integration constants.
The area expansion Θ and the shear σ scalars are
Ȧ Ḃ
2σ(t) = − , (3.21a)
A B
Ȧ Ḃ
−Θ =
+ . (3.21b)
A B
The evaluation of the dust energy density gives
B1 A1
κ ρ (t) = , (3.22)
(B1 t + B0 ) (A1 t + A0 )
while the shear and the area expansion scalars are
A1 B0 − B1 A0
2σ = , (3.23)
(A1 t + A0 ) (B1 t + B0 )
2 A1 B1 t + A1 B0 + B1 A0
−Θ = . (3.24)
(A1 t + A0 ) (B1 t + B0 )
Moreover, from the energy conservation law T αβ ;β = 0 one establishes
 
dρ Ȧ Ḃ
=− + ρ = Θ ρ, (3.25)
dt A B

and for σ, using (3.21a) and (3.21b) and the fact that solutions arise for the field
equations Ä = 0 and B̈ = 0, one gets
 
dσ Ȧ Ḃ
=− + σ = Θ σ. (3.26)
dt A B

For Θ = 2ȧ/a, where a(t) is the “geometric–mean scale factor”, the solution for
a(t) obeys
ȧ Ȧ Ḃ α0
2 = −( + ) → a(t)2 = . (3.27)
a A B A(t)B(t)
For the A(t), and B(t) above, (3.20) and (3.19), one has
α0
a(t)2 = , (3.28)
(A1 t + A0 ) (B1 t + B0 )
and for the choice
t κρ0 t + 2σ0
A(t) = A0 , B(t) = B0 . (3.29)
t0 κρ0 t0 + 2σ0
one has
α0 t0 (κρ0 t0 + 2σ0 ) 1 1
a(t)2 = =: a20 , (3.30)
A0 B0 t (κρ0 t + 2σ0 ) t (κρ0 t + 2σ0 )
32 Dust Solutions

which is the inverse of the “geometric–mean scale factor” a(t) reported in Barrow
et al. (2006).
In the cited work is given an alternative representation of the solutions
derived above following the standard integration of the Einstein equations. Using
the energy conservation law and the 3D analogue of Raychaudhuri’s equation,
the dynamical equations in the case of a perfect fluid obeying a barotropic
state equation p = wρ, where w is the barotropic index, are given in the
form of
ρ̇ = −(1 + w)Θ ρ, (3.31a)
1
Θ̇ + Θ2 = −2κ w ρ − 2σ 2 (3.31b)
2
σ̇ + Θσ = 0, (3.31c)

ω = 0, (3.31d)
with the constraint
1 2
Θ = κρ + σ 2 . (3.32)
4
For definitions of the area expansion, shear and vorticity, correspondingly Θ, σ,
and ω; see Sec. 1.4.
To my mind, there is a discrepancy in the sign at the right-hand side of
the equation (3.31a). To establish this fact, let us review the energy density
conservation law:
uα ∇α ρ = −(ρ + p)uα ;α = −(ρ + p)Θ, (3.33)
since, for the uα aligned along ∂t
d
uα ∇α = − ,
dt
therefore
d dρ dρ
− ρ = −(ρ + p) Θ → = (ρ + p) Θ, = (1 + w)ρ Θ. (3.34)
dt dt dt
The solutions of this last equation for Θ = 2ȧ/a, and of (3.31c), are
 2(1+w)  2
a(t) a(t)
ρ(t) = ρ0 , σ = σ0 (3.35)
a0 a0
which certainly are inverse to the ones of Barrow et al. (2006): there, integrating
(3.31a) and (3.31c) for Θ = 2ȧ/a, one arrives at
a0 2(1+w) a0 2
energy density : ρ = ρ0 , and shear: σ = σ0 .
a a
For the specific problem under consideration, the sign of Θ in Eq. (38) of Barrow
et al. (2006) is opposite to the correct one calculated here for (3.21b).
 3.4 BST Diagonal Anisotropic Dust Solutions with Λ 33

Cotton Tensor
The Cotton tensor is
 
α 1 B1 A1 (B1 A0 − A1 B0 ) δαx δy β δαy δxβ
C β = + . (3.36)
2 (B1 t + B0 ) (A1 t + A0 ) A(t)2 B(t)2
Searching for its eigenvectors, one establishes that they are

λ1 = 0, T α = δ α t , T α Tα = −1, (3.37a)
1 B1 A1 (B1 A0 − A1 B0 ) δ α
x δ α
y
λ2 = , S1α = + , S1α S1α = 2, (3.37b)
2 A(t)2 B(t)2 A(t) B(t)
1 B1 A1 (B1 A0 − A1 B0 ) δα x δαy
λ3 = − , S2α
= − + , S2α S2α = 2, (3.37c)
2 A(t)2 B(t)2 A(t) B(t)
where T and S are used to denote respectively timelike and spacelike vectors,
therefore, this Cotton tensor is of Type I: {T, S, S}.

3.4 BST Diagonal Anisotropic Dust Solutions with Λ

Homogeneous and anisotropic (2 + 1) universes in the presence of a cosmological
constant are given by the metric

ds2 = −dt2 + A2 (t)dx2 + B 2 (t)dy 2 . (3.38)

Einstein equations for dust with Λ are

Eμν := Gμν + Λ gμν − κTμν = 0, Tμν = ρ(t) uμ uν , uμ = δ t μ , (3.39)

which for the metric (3.38) explicitly read

d  d
t B dt A
Et : κ ρ (t) + Λ = dt , (3.40a)
BA
d2 √ √
Exx : 2 B (t) − Λ B (t) = 0 → B (t) = B1 e Λt + B2 e− Λt , (3.40b)
dt
d2 √ √
Eyy : 2 A (t) − Λ A (t) = 0 → A (t) = A1 e Λt + A2 e− Λt . (3.40c)
dt
The evaluation of the characteristic functions ρ, σ and Θ yields
(B1 A2 + B2 A1 )
κ ρ (t) = −2 Λ √ √ √ √ , (3.41a)
B1 e Λt + B2 e− Λt A1 e Λt + A2 e− Λt

√
Λ (B2 A1 − B1 A2 )
σ(t) = √ √ √ √ , (3.41b)
B1 e Λt + B2 e− Λt A1 e Λt + A2 e− Λt

√ √ √
Λ −A1 B1 e2 Λt + A2 B2 e−2 Λt
Θ=2 √ √ √ √ . (3.41c)
B1 e Λt + B2 e− Λt A1 e Λt + A2 e− Λt
34 Dust Solutions

3D Gödel Universes
One can bring the derived solutions a Gödel-like form by introducing the
trigonometric functions, namely
A(t, Λ < 0) = a1 sin |Λ|t + a2 cos |Λ|t,
B(t, Λ < 0) = b1 sin |Λ|t + b2 cos |Λ|t, (3.42a)
√ √
A(t, Λ > 0) = a1 sinh Λt + a2 cosh Λt,
√ √
B(t, Λ > 0) = b1 sinh Λt + b2 cosh Λt. (3.42b)

3.5 BST (t, x, y)-Dependent Cosmological Solutions with

Comoving Dust
In this section, anisotropic dust solutions with structural functions depending
on the (t, x, y) coordinates are derived. Thus, the class of metrics of interest is
of the form
2 2
ds2 = −dt2 + A(t, x, y) dx2 + B(t, x, y) dy 2 . (3.43)
The Einstein equations for dust with velocity uμ = δμt are
 2   
∂ ∂ ∂
Etx : − B A+ A B = 0, (3.44a)
∂x∂t ∂t ∂x
 2   
∂ ∂ ∂
Ety : A B− B A = 0, (3.44b)
∂y∂t ∂t ∂y
∂2
Exx : 2 B − Λ B = 0, (3.44c)
∂t
∂2
Eyy : 2 A − Λ A = 0, (3.44d)
∂t
with the energy density resulting from equation Ett = 0, namely
∂2A ∂2B ∂B ∂A
[κ ρ(t, x, y) + Λ]A3 B 3 = −A2 B 2
− B 2 A 2 + A2 B 2
∂y ∂x ∂t ∂t
∂A ∂B ∂A ∂B
+B 2 + A2 , (3.45)
∂x ∂x ∂y ∂y
with structural functions depending on the three variables t, x, and y.
The equations (3.44c) and (3.44d) integrate as
√ √
A(t, x, y) = A1 (x, y) e Λt
+ A2 (x, y) e− Λt
,
√ √
Λt − Λt
B(t, x, y) = B1 (x, y) e + B2 (x, y) e , (3.46)
which determine the structure of the metric functions of all the classes of solu-
tions√with time dependence
√ t via a cosmological constant Λ in terms of hyperbolic
sinh Λt and cosh Λt functions for positive Λ = |Λ|, or through trigonometric
 3.5 BST (t, x, y)-Dependent Cosmological Solutions 35

sin |Λ|t and cos |Λ|t functions for negative Λ = −|Λ| = −1/l2 . When
replacing the exponential functions in terms of the hyperbolic, or trigonometric
functions depending on the sign of the Λ, the coefficient functions in front of the
sines and cosines are linear combinations of the previous coefficients multiplying
the exponentials (in the negative Λ these coefficients are complex conjugated);
for them, Greek symbols will be used, and one has these possibilities: for positive
Λ: Λ > 0, one gets
√ √
A(t, x, y) = α1 (x, y) sinh Λt + α2 (x, y) cosh Λt,
√ √
B(t, x, y) = β1 (x, y) sinh Λt + β2 (x, y) cosh Λt, (3.47a)

while for negative Λ: Λ = −|Λ|, one obtains

A(t, x, y) = α1 (x, y) sin |Λ|t + α2 (x, y) cos |Λ|t,

B(t, x, y) = β1 (x, y) sin |Λ|t + β2 (x, y) cos |Λ|t. (3.47b)

Therefore, the metric structure becomes

 √ √ 2
ds2 = −dt2 + A1(x, y) e Λt + A2(x, y) e− Λt dx2
 √ √ 2
+ B1(x, y) e Λt + B2(x, y) e− Λt dy 2 ,
(3.48)

where Λ may be considered positive or negative. The remaining Einstein

equations for the above functions read

∂ ∂
Ext : −A2 (x, y) B1 (x, y) + A1 (x, y) B2 (x, y) = 0, (3.49)
∂x ∂x
∂ ∂
Eyt : −B2 (x, y) A1 (x, y) + B1 (x, y) A2 (x, y) = 0. (3.50)
∂y ∂y

Since there are only two constraints on four unknown structural functions, then
one may consider two of them as arbitrary; let them be A1 and A2. The most
general solutions of these equations, B1 (x, y) and B2 (x, y), depending on both
spatial variables x and y, are
∂A2
∂y
B2 (x, y) = B1 (x, y) ∂A1
, (3.51)
∂y
 
 A1 ∂ 2 A2 ∂A1
− ∂A2 ∂ 2 A1
∂y∂x ∂y ∂y ∂y∂x
B1 (x, y) = b (y) exp   dx. (3.52)
∂A1
∂y A2 ∂A1
∂y − A1 ∂A2
∂y

Of course, certain conditions on the behavior of the derivatives of A1 and A2

have to be imposed.
36 Dust Solutions

The area expansion Θ and shear constant σ are given by

√ √ √
Λ B1e2 Λt A1 − B2e−2 Λt A2
Θ = −2 √ √ √ √ , (3.53)
A1e Λt + A2e− Λt B1e Λt + B2e− Λt
√
(−B1A2 + A1B2) Λ
σ= √ √ √ √ . (3.54)
A1e Λt + A2e− Λt B1e Λt + B2e− Λt

Assume we are working with the hyperbolic functions; then

√ √
A (t, x, y) = F1 (x, y) sinh Λt + F2 (x, y) cosh Λt ,
√ √
B (t, x, y) = G1 (x, y) sinh Λt + G2 (x, y) cosh Λt , (3.55)

constrained to the pair of equations

∂ ∂
G2 (x, y) F1 (x, y) − G1 (x, y) F2 (x, y) = 0, (3.56a)
∂y ∂y

∂ ∂
F1 (x, y) G2 (x, y) − F2 (x, y) G1 (x, y) = 0, (3.56b)
∂x ∂x
thus two functions should stay arbitrary; let them be F1 (x, y) and F2 (x, y).
From the (3.56a) equation one gets
∂
∂y F1 (x, y)
G1 (x, y) = ∂
G2 (x, y) (3.57)
∂y F2 (x, y)

which, substituted into the (3.56b) and integrating for G2, yields
 
 F2 ∂2
F2 ∂
F1 − ∂2
F1 ∂
F2
∂y∂x ∂y ∂y∂x ∂y
G2 (x, y) = g2 (y) exp   dx
∂
∂y F2 F2 ∂
∂y F1 − F1 ∂
∂y F2
(3.58)

with arbitrary functions F1 and F2 . Notice that the integration function g2 (y)
can be equated to 1 by accomplishing a coordinate transformation in y.

Algebraic Classification of the Cotton Tensor

For this class of metric the Cotton tensor is given by
⎡ ⎤
0 A2 ∂y ∂
m − B 2 ∂x
∂
m
κ ⎢ ⎢   ⎥
(C α β ) = − ∂y
∂
m 0 ∂
m B2 ⎥. (3.59)
2AB ⎣ ∂t ⎦
 
∂
∂x m − ∂t
∂
m A2 0
 3.5 BST (t, x, y)-Dependent Cosmological Solutions 37

The secular equation for the eigenvalue S is given by

 
S 3 − C23 C32 + C21 C12 + C31 C31 S − C31 C12 C23 − C21 C32 C13 = 0,
where the correspondence with the coordinates is {1, 2, 3} → {t, x, y}: for the
structural functions determining this metric, none of the factors of this equation
is zero, hence, in general there exist three solutions for S, such that S1 +S2 +S3 =
0; consequently the Cotton tensor is of Type I.

3.5.1 BST Class 2 of Solutions

Another representation of this class of solutions has been given in the same
article by Barrow et al. (2006) as the Class 2. First, the structural functions F1
and F2 are chosen as
F2 (x, y) = eν(x,y) , F1 (x, y) = eν(x,y) F (x, y) , (3.60)
which substituted into (3.56a) determines
 
∂ν ∂F
∂y F + ∂y
G1 (x, y) = G2 (x, y) . (3.61)
∂ν
∂y

Entering G1 into (3.56b) and integrating for G2 one arrives at

∂ν  x  ∂ν
1 ∂y −α(x,y) ∂F ∂y
G2 = e , α (x, y) := dx. (3.62)
S(y) ∂F ∂x ∂F
∂y ∂y

Finally, the sought solution can be given as

2 2
ds2 = −dt2 + A (t, x, y) dx2 + B (t, x, y) dy 2 ,
 √ √ 
A (t, x, y) = eν(x,y) F (x, y) sinh Λt + cosh Λt ,
 
∂ν
e−α(x,y) ∂y
∂F
∂y
√ √
B (t, x, y) = ∂F
F + ∂ν sinh Λt + cosh Λt ,
S(y) ∂y ∂y
 x  ∂ν
∂F ∂y
α (x, y) = dx, (3.63)
∂x ∂F ∂y

where F (x, y) and ν(x, y) are arbitrary functions ∂F∂y = 0, and S(y) is an
integration function, which can be set equal to unity by a redefinition of the
y-coordinate. The evaluation of the energy density can done by means of the
expression (3.45). This general dust solution corresponds to the one given in
paragraph 6.2 Class 2 in Barrow et al. (2006), through
[∂y (R(x, y, t)eν(x,y) )]2 −2α(x,y) 2
ds2 = −dt2 + e−2ν(x,y) e dy
S 2 (y)(∂y F (x, y))2
+R2 (x, y, t)e2ν(x,y) dx2 ,
38 Dust Solutions

R(x, y, t) = chΛ t + F (x, y)shΛ t,

1 √ √
shΛ t := √ sinh Λt, chΛ t := cosh Λt. (3.64)
Λ

Szekeres Solutions
Szekeres solutions arise for ∂x F = 0, and

e−ν(x,y) = A(y)x2 + 2B(y)x + C(y).

3.5.2 BST Class 1 Spacetime

Apart from the general solutions above, there is a subclass of solutions arising for
A1 (x, y) = a1(x) and A2 (x, y) = a2(x) depending on the single spatial variable
x. In such case, the equation (3.50) is identically fulfilled, while equation (3.49)
allows the solutions

B1 (x, y) = a1 (x) L (x, y) dx + b1 (y) ,

B2 (x, y) = a2 (x) L (x, y) dx + b2 (y) , (3.65)

where L (x, y) is an arbitrary function. In this representation, the energy density

amounts to
 ∂
 √
2 Λ a2 (x) B1 + 2 Λ B2a1 (x) + ∂x L e2 Λt
κρ = − √ √ , (3.66)
e2 Λt B1 + B2 a2 (x) + e2 Λt a1 (x)
or
√ √ √ √
− ∂x
∂
L+Λ e Λt
B1 − B2e− Λt
e Λt
a1 − a2 e− Λt

κρ + Λ = √ √ √ √ , (3.67)
e Λt B1 + B2e− Λt e2 Λt a1 + a2e− Λt

to maintain the parallelism with the results of Barrow et al. (2006).

In Barrow et al. (2006) this solution was reported in the alternative form

ds2 = −dt2 + [C(x)shΛ (t) + D(x)chΛ (t)]2 dx2

+[W (x, y)shΛ (t) + V (x, y)chΛ (t)]2 dy 2 ,
 x
∂L(ξ, y)
W (x, y) = C(ξ) dξ + f (y),
x ∂ξ
 x0
∂L(ξ, y)
V (x, y) = D(ξ) dξ + g(y), (3.68)
x0 ∂ξ
with dust energy density
−∂x2 L(x, y) + [W chΛ (t) + Λ V shΛ (t)][CchΛ (t) + Λ DshΛ (t)]
κρ + Λ = ,
[W shΛ (t) + V chΛ (t)][CshΛ (t) + DchΛ (t)]
 3.5 BST (t, x, y)-Dependent Cosmological Solutions 39

where the dependence on x, and y has been omitted. Notice that the definitions
of the functions L(x, y) differ; L(3.65) = ∂x L(3.68) .

3.5.3 BST Class 3 of Dust Solutions

This class of solutions arises for the choice A1 (x, y) = α0 A2 (x, y), which sub-
stituted into equation (3.50) yields B1 (x, y) = α0 B2 (x, y); for these relations
the equation (3.49) is immediately satisfied. The metric amounts to
√ √
+ e−
2 2
ds2 = −dt2 + (α0 e Λt Λt 2
) [A2 (x, y) dx2 + B2 (x, y) dy 2 ]. (3.69)

Since any 2D space is conformally flat, one may write the spatial sector of this
metric as
√ √
e2 φ(x,y) (α0 e Λt
+ e− Λt 2
) (dx2 + dy 2 ). (3.70)

Thus the spacetime for positive Λ > 0 is

√ √
ds2 = −dt2 + e2 φ(x,y) (b0 sinh Λt + a0 cosh Λt)2 (dx2 + dy 2 ), (3.71a)

where α0 and β0 are constants, with energy density ρ(t, x, y) given by

2 2
−2 ν(x,y)
Λ (a0 2 − b0 2 ) + [ ∂y
∂ ∂
2 ν (x, y) + ∂x2 ν (x, y)]e
κρ = − √ √ (3.72)
[b0 sinh Λt + a0 cosh Λt]2
and scalar area expansion
√ √
√ a0 sinh Λt + b0 cosh Λt
Θ = −2 Λ √ √ . (3.73)
a0 cosh Λt + b0 sinh Λt

The proper BST formulation is given by

ds2 = −dt2 + e2φ(x,y) (μ shΛ t + chΛ t)2 (dx2 + dy 2 ), (3.74)

with dust energy density

(μchΛ t + ΛshΛ t)2 − e−2φ(x,y) ∇2 φ(x, y)
κ ρ(x, y) + Λ = . (3.75)
(μ shΛ t + chΛ t)2
The shear tensor and rotation tensor are zero. The FRW solution emerges when
solution arises for e−2φ(x,y) ∇2 φ(x, y) = constant.
The second solution, for negative cosmological constant Λ = −L2 , is given by

ds2 = −dt2 + e2 ν(x,y) [b0 sin (L t) + a0 cos (L t)]2 (dx2 + dy 2 ), (3.76)

with energy density

2
∂2
L2 (a0 2 + b0 2 ) − e−2 ν(x,y) [ ∂x
∂
2 ν (x, y) + ∂y 2 ν (x, y)]
κρ = 2 , (3.77)
[a0 cos (Lt) + b0 sin (Lt)]
40 Dust Solutions

and scalar area expansion

 2 
a0 − b0 2 sin (2 Lt) − 2b0 a0 cos (2 Lt)
Θ=L 2 . (3.78)
[a0 cos (Lt) + b0 sin (Lt)]
The shear tensor vanishes.

3.6 Rooman–Spindel Dust Gödel Non-Diagonal Model

Rooman and Spindel (1998) reported various families of dust models depending
on one spatial coordinate, exhibiting certain properties belonging to the 4D
Gödel solution.
The metric in comoving coordinates is given by Eq.(80) in Rooman and Spindel
(1998) as
ds2 = −[dt + Z(r)dφ]2 + dr2 + Y (r)2 dφ2 . (3.79)

The conservation of the energy momentum tensor Tμν = (μ + p)uμ uν + p gμν for
a perfect fluid (dust) moving along the time direction ∂t ,

uμ = −δ μ t , uμ = δμ t + Zδμ φ , (3.80)
dp
requires the constancy of the pressure, dr = 0, p(r) = p0 .
The Einstein equation
d2 Z dZ dY
Gφ t = 0 : Y − = 0, (3.81)
dr2 dr dr
integrates as
d
Y (r) = C1 Z (r) , (3.82)
dr
while Gr r = 0 = (Gφ φ ) gives the relation
1
κ p0 = Λ + = constant, (3.83)
4C12
and the equation Gt t = 0 = (Gt φ ) becomes
 
d3 Z 3 dZ
3
+ κ μ (r) + Λ − 2 = 0. (3.84)
dr 4C1 dr
Although the obtained solution possesses a constant pressure throughout the
entire spacetime, it is endowed with matter modeled through the energy density
subject to the equation (3.84). As pointed out in Rooman and Spindel (1998),
this equation can be thought of as defining the matter content for a given function
Z(r), or, conversely, for a given plausible matter distribution one determines the
structural function Z(r).
The evaluation of the fluid covariant derivative properties yields zero, i.e., the
area expansion Θ, shear σ, rotation ω and acceleration vector all vanish.
 3.6 Rooman–Spindel Dust Gödel Non-Diagonal Model 41

As far as the algebraic type of the Cotton tensor is concerned, it is clearer and
more straightforward to establish it from the alternative representation of this
metric in term of the the structural function Y (r). Instead of (3.82), one now
uses
d
Z (r) = F0 Y (r) . (3.85)
dr
Thus, there remains a single constraint
 
d2 3 2
Y (r) + κ m (r) + Λ − F0 Y (r) = 0. (3.86)
dr2 4
The Cotton tensor in terms of Y (r) and M (r), 4 κ m (r) + F0 2 + 4 Λ = 8 M (r),
becomes
⎡   ⎤
−2 F0 Y M − Z dM
dr 0 −3 ZF0 Y M − Y 2 + Z 2 dM dr
⎢ ⎥
Y (C α β ) = ⎢
⎣ 0 F0 Y M 0 ⎥,
⎦
dM
dr 0 Z dM
dr + F0 Y M
(3.87)

from which it is evident that the roots for the eigenvalues could be real or complex
conjugated, fulfilling the traceless condition λ1 + λ2 + λ3 = 0, therefore the
algebraic types could be Type I, and Type IZ .
In the next paragraph, for the case of incompressible dust m(r) = m0 , the
eigenvectors of the Cotton tensor are shown explicitly.

Rooman–Spindel Dust
If the pressure is set equal to zero, p = 0, thus C21 = − 4Λ
1
, then one is dealing
with an AdS, Λ < 0, dust star, with metric
 2
dZ
ds = −[dt + Z(r)dφ] + dr + C1
2 2 2
dφ2 , (3.88)
dr
and energy density subjected to
d3 Z dZ
3
+ (κ μ (r) + 4Λ) = 0. (3.89)
dr dr

Incompressible Dust Gödel Models

A quite particular class of solution emerges under the constancy of the energy
density, μ = μ0 . The energy equation becomes
d3 d
Z (r) + M0 κ Z (r) = 0, (3.90)
dr3 dr
where
3
κM0 = κμ0 − + Λ = κμ0 + 4Λ, Λ < 0. (3.91)
4 C1 2
42 Dust Solutions

allowing the general solutions

Z (r) = C0 + C2 sin κM0 r + C3 cos κM0 r ,

M0 > 0, a−2 = κM0 , (3.92a)

or

Z (r) = C0 + C2 sinh κ M0 r + C3 cosh κ M0 r ,

M0 < 0, a−2 = −κM0 . (3.92b)

Following Lubo et al. (1999), see Eq. (95), these solutions can be given as
r r
ds2 = −dt2 + dr2 + γ 2 sinh2 ( ) − a2 c2 (κ + cosh( ))2 dφ2
a a
r
−2acγ(κ + cosh( ))dt dφ, (3.93a)
a
2r r
ds2 = −dt2 + dr2 + γ 2 e( a ) − a2 c2 (κ + e( a ) )2 dφ2
r
−2acγ(κ + e( a ) )dt dφ, (3.93b)
r r
ds2 = −dt2 + dr2 + γ 2 cosh2 ( ) − a2 c2 (κ + sinh( ))2 dφ2
a a
r
−2acγ(κ + sinh( ))dt dφ, (3.93c)
a
Locally, these metrics are mutually equivalent; for coordinate transformations
relating to them, see Rooman and Spindel (1998); for low density stars, κμ <
−4Λ, they are determined by restricting the domain of r to r < rboundary , hence
they correspond to different non-diffeomorphic subsets of a larger space on which
r is unrestricted.
On the other hand, the metric of the high density star κμ > −4Λ is given by
r r
ds2 = −dt2 + dr2 + γ 2 sin2 ( ) − a2 c2 (κ + cos( ))2 dφ2
a a
r
−2acγ(κ + cosh( ))dt dφ. (3.94)
a
For the physical interpretation of the above solutions one has to return to the
original Gödel works.
The Cotton tensor (3.87), for this incompressible case

m(r) = m0 → M (r) = M0 ; 2F0 M0 = C0 ,

amounts to
⎡ ⎤
−C0 0 − 32 Z C0
⎢ ⎥
(C α β ) = ⎢
⎣ 0 1
2 C0 0 ⎥,
⎦
1
0 0 2 C0
 3.6 Rooman–Spindel Dust Gödel Non-Diagonal Model 43

thus, the eigenvalues are: a single λ1 = −C0 and a double one λ2 = C0 /2 with
eigenvectors
λ1 = −C0 , V 1α = δ α t , V 1α V 1α = −1,
C0
λ2 = , V 2α = −ZV3 δ α t + V2 δ α x + V3 δ α y , V α Vα = V2 2 + V3 2 Y 2 ,
2
C0
λ3 = , V 3α = −Zv3 δ α t + v2 δ α x + v3 δ α y , V α Vα = v2 2 + v3 2 Y 2 ,
2
(3.95)
in particular, for the components
V3 2 Y 2
v1 = V1 , v2 = − , v3 = V3 ,
V2
the vectors V 2 and V 3 are orthogonal. Consequently, the Cotton tensor for these
Gödel kinds of solutions is of Type I: {T, S, S}.
 4
A Shortcut to (2+1) Cyclic Symmetric
Stationary Solutions

From a general cyclic symmetric stationary metric of the (2 + 1)-gravity with a

negative Λ = −1/l2 a shortcut in the derivation of the three families of cyclic
solutions is given. There arise: two branches for the BTZ black hole solution, the
stationary Coussaert–Henneaux (CH) solution, the cyclic CH SO(2) × SO(2)
cosmological metric, and additionally the static AdS class with its sub-branches:
the one-parametric AdS solution, called the AdS static BTZ black hole, and the
AdS naked singularity metric.

4.1 Cyclic Symmetric Stationary Solutions in Canonical Coordinates

Within (2+1)-dimensional Einstein theory, the general form of a stationary cyclic
symmetric line element – allowing for a timelike Killing vector ∂t and a spacelike
Killing vector field ∂φ – can be given as
F (r) 2 dr2
g=− dt + + H(r)(dφ + W (r)dt)2 . (4.1)
H(r) F (r)
For this metric, the Einstein equations Eμ ν := Gμ ν − δμ ν /l2 = 8π Tμ ν , with
negative cosmological constant, in the vacuum case Tμ ν = 0, yield:
1 d dW dW J
Eφ t = (H 2 )=0→ = 2, (4.2)
2 dr dr dr H
 
1 d2 F 8 4
Et t + E φ φ + 2 E r r = − = 0 → F (r) = 2 r2 + F1 r + F0 , (4.3)
2 dr2 l2 l

F d2 H
Et t − Er r − W Eφ t = = 0 → H(r) = H1 r + H0 . (4.4)
2H dr2
The integration of (4.2), assuming H1 = 0, gives
J 1 J 1
W (r) = − + W0 = − + W0 . (4.5)
H1 H(r) H1 H1 r + H 0
 4.1 Stationary Solutions in Canonical Coordinates 45

Without loss of generality one can set W0 = 0. Moreover, the substitution of the
above structural functions into the remaining Einstein equations gives rise to a
condition on the integration constants, which can be solved, for instance, for F0

H0 J2 4 H2
F0 = F1 + 2 − 2 02 (4.6)
H1 H1 l H1
Therefore the metric tensor components become
1 4 H0
gtt = − ( 2 r + F1 − 4 2 ),
H1 l l H1
4 J2 H0 H0
1/grr = 2 r 2 + F1 r + 2 + (F1 − 4 2 ),
l H1 H1 l H1
J
gt φ =− , gφ φ = H1 r + H0 . (4.7)
H1

4.1.1 Bañados–Teitelboim–Zanelli Solution in Canonical

Polar ρ Coordinates
Accomplishing in (4.1), with structural functions determined by (4.3)–(4.5) and
constraint (4.6), the linear transformations and re-parameterizations

t→t H1 , φ → 2φ/ H1 , r → ρ2 /4 − H0 /H1 , 4F (ρ)/ρ2 → f (ρ),

8 H0
J → J0 H1 /4, F1 → 2 − M, (4.8)
l H1
one arrives at the standard BTZ form presented in Bañados et al. (1992)
 2
dρ2 J0
g = −f (ρ)dt2 + + ρ2 dφ − 2 2 dt ,
f (ρ) ρ
ρ2 J02
f (ρ) = 2 − M + 4 2 . (4.9)
l ρ
For a detailed analysis see Bañado et al. (1993). This representation of the
BTZ metric in polar circumference coordinate ds = ρ dφ is useful when one
considers, for instance, the vacuum limit of the metrics describing electric fields.

Brown–York Quasilocal Characteristics of BTZ

The Brown–York quasilocal characterization of stationary (2+1) metrics is given
in detail in Section 1.3 and in the summary 1.3.1. Using these quantities, the
energy–momentum characterization for the metric (4.9) amount to: momentum
J0 J0
jφ = , jφ ≈ ,
2πρ 2πρ
Jφ = J0 , Jφ ≈ J0 , (4.10a)
46 A Shortcut to (2+1) Cyclic Symmetric Stationary Solutions

energy density

f (ρ) 1 l M
(ρ, M, 0 ) = − − 0 , (ρ, M, 0) ≈ − + ,
πρ πl 2π ρ2
l M − M0
(ρ, M, 0 (M0 )) ≈ ,
2π ρ2
ρ2 / l2 − M0 1 l M0
0 (M0 ) := − , 0 (M0 ) ≈ − + , (4.10b)
πρ π l 2π ρ2
total quasi-local energy

E(ρ, M, 0 ) = −2 f (ρ) − 2πρ0 ,

ρ l l
E(ρ, M, 0) ≈ −2 + M, E(ρ, M, 0 (M0 )) ≈ (M − M0 ), (4.10c)
l ρ ρ
total quasi-local mass
ρ2 ρ2
M (ρ, M, 0 ) = −2 + 2M − 2πρ f (ρ)0 , M (ρ, M, 0) = −2 + 2M,
l2 l2
M (ρ, M, 0 (M0 )) ≈ M − M0 , (4.10d)

where ≈ denotes “approximate value of the considered quantity in the limit as

ρ → ∞.”
As we see, these quantities depend on the choice of the referential energy
density 0 : the referential 0 = 0 can be thought of as√the energy corresponding
ρ2 / l2 −M0
to the flat Minkowski spacetime, while 0 (M0 ) = − πρ corresponds to
energy density of the static BTZ metric – a common choice.

4.1.2 BTZ Solution Counterpart

On the other hand, accomplishing in (4.1) the following linear transformations
and re-parametrization
2 l2 H0
t→ t H1 l, φ → √ Φ, r → ρ2 /4 − F1 + , 4F (ρ)/ρ2 → f (ρ),
l H1 4 H1
8 H0
J → J0 H1 , F1 → 2 − M, (4.11)
l H1
one arrives at the BTZ solution in the form
 2
J0 dρ2
g = −ρ2 dt + 2 dφ + + f (ρ)dφ2
2ρ f (ρ)
f (ρ) dρ2 J0 /2
= −ρ2 dt2 + + (ρ2 /l2 + M )[dφ − 2 2 dt]2 ,
ρ2 /l2
+M f (ρ) ρ /l + M
ρ2 J2
f (ρ) = 2 + M + 02 . (4.12)
l 4ρ
 4.1 Stationary Solutions in Canonical Coordinates 47

It becomes apparent that this metric form is just another real cut of the met-
ric (4.9) when subjecting it to the complex transformations t → i φ and φ → i t.
This representation of the BTZ metric counterpart is useful when one considers,
for example, the vacuum limit of the metrics describing magnetic fields.
Moreover, by accomplishing in the above metric the transformation of the
radial coordinate

t → t/l, ρ → l ρ2 /l2 − M0 , φ → l φ, (4.13)

one gets the standard BTZ metric (4.9).

Brown–York Quasilocal Characteristics of BTZ Counterpart

As far as to the energy–momentum characteristics are concerned, for the second
form of the metric (4.12) one has
J0 J0
jφ = − , jφ ≈ − ,
2 l2
π ρ2 / l2
+M 2 l πρ
J0 J0
Jφ = − 2 , Jφ ≈ − 2 ,
l l
1 f (ρ) 1 l M
(ρ, M, 0 ) = − 2 2 2 − 0 , (ρ, M, 0) ≈ − + ,
l π ρ /l + M πl 2π ρ2
2 f (ρ)
E(ρ, M, 0 ) = − 2
ρ − 2π 0 ρ2 / l2 + M ,
l ρ / l2 + M
2

ρ 1
E(ρ, M, 0) ≈ −2 2 , E(ρ, M, 0c (M0 )) ≈ (M − M0 ),
l ρ
2 2 ρ2
M (ρ, M, 0 ) = − 2 ρ − 2π ρ 0 f (ρ), M (ρ, M, 0) = −2 2 ,
l l
M (ρ, M, 0c (M0 )) ≈ M − M0 ,
ρ
0c (M0 ) := − ,
πl 2 ρ / l 2 + M0
2

1 l M0
0c (M0 ) ≈ − + ≈ 0 (M0 ). (4.14)
π l 2π ρ2

4.1.3 Coussaert–Henneaux Metrics

There exists the branch of solutions with constant H(r) = H0 = J l/2, thus the
integration of (4.2) yields W (r) = 4r/(Jl2 ), and therefore the resulting metric
can be given as
 2
F (r) 2 dr2 Jl 4r
g = −2 dt + + dφ + 2 dt ,
Jl F (r) 2 Jl
4 2
F (r) = 2 r + F1 r + F0 , (4.15)
l
48 A Shortcut to (2+1) Cyclic Symmetric Stationary Solutions

which, by changing t/ Jl/2 → t, φ Jl/2 → φ, can be brought to the form

 2
dr2 2r
g = −F (r)dt +
2
+ dφ + dt ,
F (r) l
4 4 l2
F (r) = 2 r2 + F1 r + F0 = 2 [(r + F1 )2 ± A2± ],
l l 8
2 2
l l
A2± := ± (F0 − F 2 ). (4.16)
4 16 1

Coussaert–Henneaux Stationary Metric

Subjecting (4.16) to the transformations
l2 l2 l 3 F1 l
t = t̃, r = − F1 + A+ sinh r̃, φ = − t̃ + φ̃,
4 A+ 2 4 A+ 2

l l2 2 4
A+ := F0 − F1 , F (r) → 2 A2+ cosh2 r̃ (4.17)
2 16 l
one gets the standard Coussaert and Henneaux (1994b) metric; see also Ayón–
Beato et al. (2004),
l2  2 2

g= −dt̃ + 2 sinh r̃dt̃ dφ̃ + dφ̃ + dr̃2 . (4.18)
4

Time-Dependent SO(2) × SO(2) Cyclic Metric

Moreover, if one were allowing the structural function F (r) to range over negative
2
values, which takes place in the case of A2− = − l4 (F0 −l2 F12 ), with lower negative
sign in F (r) from (4.16), subjecting (4.16) to the transformations
l2 l2 l 3 F1 l
t = θ, r = − F1 + A− sin τ, φ = θ + Φ,
4 A− 8 16 A− 2

l l 2 4
A− := F 2 − F0 , F (r) → − 2 A2 cos2 τ , (4.19)
2 16 1 l
one arrives at a time-dependent (τ -coordinate) spacetime
l2
g= dθ2 + dΦ2 + 2 sin τ dθ dΦ − dτ 2 . (4.20)
4
This spacetime is the self-dual spacetime with isometry SO(2) × SO(2); see
equation (34) of Ayón–Beato et al. (2004) and details therein.

4.2 Static AdS Black Hole

Working with the canonical metric (4.1) the Einstein equations in the static case,
W (r) = 0, simplify considerably. In principle, one obtains the anti-de Sitter static
black hole solution from the BTZ metric simply by setting J = 0, or following
step by step the procedure exhibited above in the stationary case.
 4.2 Static AdS Black Hole 49

4.2.1 Static BTZ Solution

In polar coordinates the static BTZ solution assumes the form
dρ2
g = −f (ρ, M0 )dt2 + + ρ2 dφ2 ,
f (ρ, M0 )
ρ2
f (ρ, M0 ) := − M0 , (4.21)
l2
where the mass parameter is now denoted by M0 .
In these coordinates the energy characteristics acquire the form
1 1 l M0
(ρ, M0 , 0 ) = − f (ρ, M0 ) − 0 , (ρ, M0 , 0) ≈ −
+ ,
πρ πl 2π ρ2
ρ l
E(ρ, M0 , 0 ) = −2 f (ρ, M0 ) − 2π ρ 0 , E(ρ, M0 , 0) ≈ −2 + M0 ,
l ρ
M (ρ, M0 , 0 ) = −2f (ρ, M0 ) − 2πρ f (ρ, M0 ) 0 ,
ρ2
M (ρ, M0 , 0) ≈ −2 2 + 2M0 . (4.22)
l
Introducing the energy density function (ρ, M0 ),
1
(ρ, M0 ) := − f (ρ, M0 ) (4.23)
πρ
one brings the above functions to the forms
(ρ, M0 , 0 ) = (ρ, M0 ) − 0 ,
E(ρ, M0 , 0 ) = 2πρ [(ρ, M0 ) − 0 ] ,
M (ρ, M0 , 0 ) = f (ρ, M0 )E(ρ, M0 , 0 ). (4.24)

Choosing the Base Energy Density; AdS Spacetime

For M0 = −1 one obtains the standard anti-de Sitter solution in polar
coordinates
dρ2
g = −f (ρ)dt2 + + ρ2 dφ2 ,
f (ρ)
ρ2
f (ρ) = + 1, (4.25)
l2
with energy density

1 ρ2 1 l
(ρ, −1, 0) = AdS (ρ) := − + 1, AdS (ρ) ≈ − − ,
πρ l2 πl 2π ρ2
(ρ, −1, 0 ) = AdS (ρ) − 0 , (4.26)
therefore, for the referential energy density choice 0 = AdS (ρ), the mass–energy
functions of the anti-de Sitter become zero:
(ρ, −1, AdS ) = 0, E(ρ, −1, AdS (ρ)) = 0, M (ρ, −1, AdS (ρ)) = 0.
50 A Shortcut to (2+1) Cyclic Symmetric Stationary Solutions

This is one of the plausible choices for the base energy density of the
anti-de Sitter spacetime; see Brown et al. (1994) in the paragraph below
Eq. (4.12). Another possibility takes place for the naked singularity solution with
M0 = 0,
dρ2 ρ2
g = −f (ρ, 0)dt2 + + ρ2 dφ2 , f (ρ, 0) = 2 , (4.27)
f (ρ, 0) l
with energy characteristics
1 1
(ρ, 0) := − , (ρ, 0, 0 ) = − − 0 ,
πl πl
ρ
E(ρ, 0, 0 ) = 2π ρ (ρ, 0, 0 ), M (ρ, 0, 0 ) = E(ρ, 0, 0 ). (4.28)
l
For the particular referential energy density 0 = − π1l , all energy characteristics
of this naked anti-de Sitter solution vanish,
1 1 1
(ρ, 0, 0 = − ) = 0, E(ρ, 0, 0 = − ) = 0, M (ρ, 0, 0 = − ) = 0,
πl πl πl
in agreement with Brown et al. (1994), Eq. (4.12).

Limits of the Mass–Energy Functions of the Static BTZ Black Hole

For the static BTZ metric (4.21), with M0 parameter, the energy characteristics
at infinity, ρ → ∞, become:
for the “naked” referential 0 = − π1l
 
1 M0 l
 ρ, M0 , 0 = − ≈ ,
πl 2πρ2
 
1 M0 l
E ρ, M0 , 0 = − ≈ ,
πl ρ
 
1
M ρ, M0 , 0 = − ≈ M0 , (4.29)
πl
while for the AdS referential 0 = AdS
l
(ρ, M0 , AdS ) ≈ (1 + M0 ),
2πρ2
l
E(ρ, M0 , AdS ) ≈ (1 + M0 ),
ρ
M (ρ, M0 , AdS ) ≈ 1 + M0 . (4.30)

These results can be gathered in a more compact form by introducing the base
energy density equipped with the discrete parameter m = −1, and 0,

1 ρ2 1 lm
0 (ρ, m) =:= − − m, 0 (ρ, m) ≈ − + , (4.31)
πρ l2 πl 2π ρ2
 4.2 Static AdS Black Hole 51

as
l
(ρ, M0 , 0 (ρ, m)) ≈ (M0 − m),
2πρ2
l
E(ρ, M0 , 0 (ρ, m)) ≈ (M0 − m),
ρ
M (ρ, M0 , 0 (ρ, m)) ≈ M0 − m. (4.32)

4.2.2 Static AdS Solution Counterpart

Setting J0 = 0 in the metric (4.12), one arrives at the static BTZ metric
counterpart
d ρ2
g = −ρ2 d t2 + + (ρ2 /l2 + M0 )dφ2 . (4.33)
(ρ2 /l2 + M0 )
In these coordinates the energy characteristics are given by
ρ 1 1 l M0
(ρ, M0 , 0 ) = − − 0 , (ρ, M0 , 0) ≈ − + ,
πl2 ρ2 /l2 + M0 πl 2π ρ2
ρ ρ
E(ρ, M0 , 0 ) = −2 2 − 2π 0 ρ2 /l2 + M0 , E(ρ, M0 , 0) ≈ −2 2 ,
l l
ρ2 ρ2
M (ρ, M0 , 0 ) = −2 2 − 2πρ 0 ρ2 /l2 + M0 , M (ρ, M0 , 0) ≈ −2 2 .
l l
(4.34)
One could define the base energy density as the one corresponding to the black
hole limit of the anti-de Sitter spacetime or that associated with the naked
singularity; both cases can be handled by introducing the discrete parameter
m = 0, −1, and defining the base energy density function (ρ, m),
ρ 1 1 l m
0 (ρ, m) := − , 0 (ρ, m) ≈ − + . (4.35)
π l2 ρ2 /l2 +m πl 2π ρ2
For such a choice of the base energy density, the mass–energy characteristics of
the proper anti-de Sitter solution counterpart, as it should be, vanish:
(ρ, m, 0 (ρ, m)) = 0, E(ρ, m, 0 (ρ, m)) = 0, M (ρ, m, 0 (ρ, m)) = 0.
The limits at spatial infinity ρ → ∞ of the energy quantities are
l
(ρ, M0 , 0 (ρ, m)) ≈ (M0 − m)
2πρ2
1
E(ρ, M0 , 0 (ρ, m)) ≈ (M0 − m),
ρ
M (ρ, M0 , 0 (ρ, m)) ≈ M0 − m. (4.36)
By accomplishing in metric (4.33) the transformations (4.13) one gets the anti-de
Sitter black hole metric (4.21). The presence of powers of l in these transfor-
mations explains the appearance of different powers in the limits at spatial
52 A Shortcut to (2+1) Cyclic Symmetric Stationary Solutions

infinity ρ → ∞ of the energy–momentum characteristics for the BTZ and AdS

black hole solutions compared with the corresponding quantities of their solution
counterparts.

4.3 Symmetries of the Stationary and Static Cyclic Symmetric

BTZ Metrics
Although it is known that the BTZ solution possesses two Killing vectors – the
timelike symmetry along the time coordinate and the spacelike symmetry along
the orbits of the periodic angular variable – in my opinion, some comments on this
respect can be added to clarify how the number of six Killing vectors solutions
for the BTZ metric structure reduces to the quoted two. In this framework,
the six symmetries of the anti-de Sitter space with parameter M0 , AdS(M0 ),
are derived; the static BTZ allowing for time + polar coordinates possesses a
timelike and one 2π-periodic circular symmetries. For the anti-de Sitter space
with parameter M0 = −1, denoted simply by AdS, there are six symmetries:
time, circular, and four boots symmetries.
The study of the symmetries of the stationary and static cyclic symmetric
BTZ families and AdS classes of solutions starts with the stationary metric for
the standard BTZ solution
dr2 2
g = −F (r)2 d t2 + + r2 [dφ + W (r)d t] ,
F (r)2
2 r2 J2 J
F (r) = 2 − M + 2 , W (r) = − 2 . (4.37)
l 4r 2r
The covariant components of the Killing vectors Vα are denoted by vα, namely

V1 (t, r, φ) = v1, V2 (t, r, φ) = v2, V3 (t, r, φ) = v3. (4.38)

The Killing equations EQμν := Vμ;ν + Vν;μ = 0 amount explicitly to

2
∂ rF (r)
EQ11 = v1 − v2 = 0, (4.39a)
∂t l2
∂ ∂ J r
2EQ12 = v1 + v2 − 2 v3 − 2 2 2 v1 = 0, (4.39b)
∂r ∂t 2
l rF (r) l F (r)

∂ ∂
2EQ13 = v1 + v3 = 0, (4.39c)
∂φ ∂t

∂ d F1 (t, φ)
EQ22 = F (r) v2 + v2 F (r) → v2 = = 0, (4.39d)
∂r dr 2lF (r)

∂ ∂ J (M l2 − r2 )
EQ23 = v2 + v3 + 2 v1 + 2 2 v3 = 0, (4.39e)
∂φ ∂r rF (r) rl2 F (r)
 4.3 Symmetries of the BTZ Metrics 53

∂ 2
EQ33 = v3 + r F (r) v2 = 0, (4.39f)
∂φ
where Fi (t, φ) , i = 1, 2, 3, are integration functions.
Isolating v1 from (4.39e) in terms of v2 and v3 and their derivatives, one gets
 
  ∂ ∂
rl2 Jv1 = 2 r r2 − M l2 v3 − r2 l2 F (r)2 v2 + v3 . (4.40)
∂φ ∂r

Next, substituting v1 from above and the first integral of v2 from (4.39d) into
equation (4.39b), one arrives at a linear second-order equation for v3 with
integrals
 
r2 1 r l F (r) ∂ ∂
v3 = F2 (t, φ) + F3 (t, φ) − J +M F1 (t, φ)
2 2 M 2 l2 − J 2 ∂t ∂φ
(4.41)

which, substituted together with v2 from (4.39d) into Eq. (4.40) for v1, gives
 2 
r − M l2 1
v1 = 2 F3 (t, φ) − J F2 (t, φ)
Jl2  4 
1 r l F (r) J ∂ ∂
+ +M F1 (t, φ) . (4.42)
2 (l2 M 2 − J 2 ) l2 ∂φ ∂t

The dependence of the Killing vector components on the r variable has been
established; it remains still to determine their dependence on the t and φ vari-
ables hiding in the F1 (t, φ) , F2 (t, φ) and F3 (t, φ) functions. Substituting the
expressions of v1 from (4.42), v3 from (4.41), and v2 from (4.39d), one arrives
at the independent equations
 2 
∂2 ∂
F1 J 2 + l2 J F1 + l 2 1 M − F1 M l = 0,
2 2
F (4.43a)
∂t∂φ ∂φ2

 
∂2 ∂2
2
F1 J + l J 2
F1 + M F1 l4 − F1 M 2 l2 = 0, (4.43b)
∂t∂φ ∂t2

with integral
√ √ √ √
M l−J(lφ+t) M l+J(lφ−t) M l+J(lφ−t) M l−J(lφ+t)
− −
F1 = C1 e l3/2 + C2 e l3/2 + C3 e l3/2 + C4 e l3/2 .
(4.44)

Furthermore, there have to be solved constraints on F2 (t, φ) and F3 (t, φ), namely
∂ ∂
F3 = 0, F3 = 0, F3 (t, φ) = J C6 = const.,
∂t ∂φ
∂ ∂
F2 = 0, F2 = 0, F2 (t, φ) = C5 = const. (4.45)
∂t ∂φ
54 A Shortcut to (2+1) Cyclic Symmetric Stationary Solutions

where the integration constants are denoted through Ci , i = 1, . . . , 6. Finally,

the covariant Killing vector components of the BTZ solution are
 √
1 r F (r) √ Ml − J
V1 = √ C1 M l + J exp( (lφ + t))
2 l 3/2 M l −J
2 2 2 l3/2
√
√ J + Ml
−C2 M l − J exp( 3/2
(lφ − t))
√l
√ J + Ml
+C3 M l − J exp(− 3/2
(lφ − t))
√ l 
√ Ml − J
−C4 M l + J exp(− (lφ + t))
l3/2
 2 
J M l − r2
− C5 − 2 C6 , (4.46a)
4 l2
 √ √
1 Ml − J J + Ml
V2 = C1 exp( (lφ + t)) + C2 exp( (lφ − t))
2lF (r) l3/2 l3/2
√ √ 
J + Ml Ml − J
+C3 exp(− (lφ − t)) + C4 exp(− (lφ + t)) , (4.46b)
l3/2 l3/2
 √
1 r F (r) √ Ml − J
V3 = − √ C 1 M l + Jexp( (lφ + t))
2l1/2 M 2 l2 − J 2 l3/2
√
√ Ml + J
+C2 M l − J exp( 3/2
(lφ − t))
√l
√ Ml + J
−C3 M l − Jexp(− 3/2
(lφ − t))
√ l 
√ Ml − J r2
−C4 M l + Jexp(− (lφ + t)) + C5 + J C6 . (4.46c)
l3/2 2
These expressions allow one to determine the Killing vector ki associated with
its corresponding integration constant Ci for each of the possible i, by means of
!6
Vμ = i=1 kiμ , where kiμ = Ci Viμ , for each fixed value of i. The contravariant
vectors’ components ki μ , ∂ki = ki μ ∂x∂ μ , are derived from the relationship V μ =
!6 !6 !6
Vν g νμ = i=1 kiν g νμ = i=1 ki μ = i=1 Ci Viμ . Explicitly, these contravariant
Killing vectors’ components ki μ , ∂ki = ki μ ∂x∂ μ = Ci Viμ , i = 1, . . . , 6 are given
below. The reason to include the integration constants in the definitions of the
Killing vectors ki μ is related to the domain of definition of the spatial coordinates;
in the case of the existence of a periodic coordinate some Killing vectors vanish,
which can be easily achieved by setting certain structural constant equal to zero.
Explicitly, these Killing vectors are:
√ 
Ml − J 1 Jl − 2 r2 1 1
C1 , ∂k1 ; k1 μ = C1 exp (lφ + t) √ ,
l 3/2 4 M l − J r F (r) l3/2

1 F (r) 1 Jl + 2 r2 − 2 M l2 1 1
,− √ , (4.47a)
2 l 4 Ml − J l5/2 r F (r)
 4.3 Symmetries of the BTZ Metrics 55
√ 
J + M l (lφ − t) 1 l J + 2 r2 1 1
C2 , ∂k2 ; k2 μ
= C2 exp 3/2
√ 3/2
,
l 4 J + Ml l r F (r)
  
1 F (r) 1 l J + 2 M l2 − 2 r2 1 1
, √ , (4.47b)
2 l 4 J + Ml l5/2 r F (r)
 √ 
J + Ml 1 l J + 2 r2 1 1
C3 , ∂k3 ; k3 μ
= C3 exp − 3/2
(lφ − t) − √ 3/2
,
l 4 J + Ml l r F (r)
  
1 F (r) 1 l J + 2 M l2 − 2 r2 1 1
,− √ , (4.47c)
2 l 4 J + Ml l5/2 r F (r)
 √ 
M l − J (lφ + t) 1 J l − 2 r2 1 1
C4 , ∂k4 ; k4 μ
= C4 exp − − √ ,
l 3/2 4 Ml−J l 3/2 r F (r)

1 F (r) 1 J l + 2 r2 − 2 M l2 1 1
, √ , (4.47d)
2 l 4 Ml−J l 5/2 r F (r)

C5 , ∂k5 ; k5 μ = C5 [0, 0, 1/2], (4.47e)

C6 , ∂k6 ; k6 μ = C6 [−2, 0, 0]. (4.47f)

For completeness, the list of the independent Killingian commutators is given:
√ √
Ml − J Ml + J
∂[k6 ∂k1 ] = −2C6 3/2
∂k1 , ∂[k6 ∂k3 ] = −2C6 ∂k 3 ,
√ l √ l3/2
Ml − J Ml + J
∂[k6 ∂k4 ] = 2C6 3/2
∂k4 , ∂[k6 ∂k2 ] = 2C6 ∂k 2 , (4.48a)
√l l3/2
√
1 Ml − J 1 Ml + J
∂[k5 ∂k1 ] = C5 ∂k1 , ∂[k5 ∂k3 ] = −C5 ∂k 3 ,
2 √l1/2 2 √ l1/2
1 Ml − J 1 Ml + J
∂[k5 ∂k4 ] = −C5 1/2
∂k4 , ∂[k5 ∂k2 ] = C5 ∂k 2 , (4.48b)
2 l 2 l1/2
C1 C4 1 C1 C4 2
∂[k1 ∂k4 ] = − √ ∂k + √ ∂k , (4.48c)
2C6 l5/2 M l − J 6 C5 l7/2 M l − J 5
C3 C2 1 C3 C2 2
∂[k3 ∂k2 ] = − √ ∂k 6 − √ ∂k5 . (4.48d)
2C6 l 5/2 Ml + J C5 l 7/2 Ml + J
All anti-de Sitter metrics for coordinates {t, ρ, φ} – merely names – ranging
−∞ ≤ t ≤ ∞, −∞ ≤ ρ ≤ ∞, −∞ ≤ φ ≤ ∞
allows for six symmetries, i.e., six Killing vectors. All these spaces in these
coordinates are maximally symmetric spaces.
Another is the situation if the spatial coordinates are constrained to range
0 ≤ ρ ≤ ∞, 0 ≤ φ ≤ 2π,
in such case ρ and φ are polar coordinates with φ being the angular coordinate
with period 2π. Since the expressions of four of the Killing vector fields depend-
ing on φ do not exhibit the angular symmetry in 2π, invariance under the change
56 A Shortcut to (2+1) Cyclic Symmetric Stationary Solutions

φ → φ + 2π, therefore there is no room for the corresponding symmetries and

the integration constants associated with those vectors ought to be zero. Conse-
quently the metric with positive M allowing for polar angular coordinate, and
only that, possesses only two Killing vectors, ∂t and ∂φ (two symmetries: the time
translation and the 2π-periodic angular rotation). This spacetime is known as
the stationary BTZ black hole. For more about this, see also Ayón–Beato et al.
(2004).
A similar situation takes place in the case of the static anti-de Sitter metric. By
setting the rotation parameter equal to zero, J = 0, the above expressions (4.48)
give the Killing vectors for the static anti–de Sitter spacetime. Again, in the
case of the coordinates restricted to ranges 0 ≤ ρ ≤ ∞, 0 ≤ φ ≤ 2π, the static
anti-de Sitter metric allows only for two Killing vectors: ∂t , and ∂φ , i.e., the time
translation and the 2π-periodic angular rotation; otherwise, when there are six
constants, the space is maximally symmetric.

4.3.1 Symmetries of the AdS Metric for Negative M , M = −α2

If M is negative, one can equate it to −α2 . Moreover, instead of complex expo-
nential function, it will be better to use trigonometric sine and cosine functions.
Thus, one can give the Killing vector components as V = Vμ dxμ = Ca Vaμ dxμ
√     
r α2 l 2 + r 2 αt αt
V1 = 3
−C1 sin (α φ) cos + C2 sin (α φ) sin
αl l l
   
αt αt
−C3 cos (α φ) cos + C4 cos (α φ) sin
l l
+C5 (α2 l2 + r2 ), (4.49)
    
α t α t
V2 = α2 l2 + r2 C1 sin (α φ) sin + C2 sin (α φ) cos
l l
   
αt αt
+C3 cos (α φ) sin + C4 cos (α φ) cos , (4.50)
l l
√     
r α2 l 2 + r 2 αt αt
V3 = 2
C 1 cos (α φ) sin + C 2 cos (α φ) cos
αl l l
   
αt αt
−C3 sin (α φ) sin − C4 sin (α φ) cos + C6 r 2 . (4.51)
l l
This anti-de Sitter metric (cosmological constant negative, Λ = −1/l2 ) for
the coordinates {t, ρ, φ} – merely names – ranging −∞ ≤ t ≤ ∞, −∞ ≤
ρ ≤ ∞; −∞ ≤ φ ≤ ∞ allows for six symmetries, i.e., six Killing vectors.
For these ranges of determination of the coordinates, the space is maximally
symmetric.
If the spatial coordinates are restricted to range

0 ≤ ρ ≤ ∞, 0 ≤ φ ≤ 2π,
 4.3 Symmetries of the BTZ Metrics 57

and α is set equal to unity, α = 1 = −M , then in such case ρ and φ become polar
coordinates with φ being the angular coordinate with period 2π. This spacetime
– the (proper) anti-de Sitter space (with M = −1) – allows for six symmetries,
and as such it is maximally symmetric.
In this chapter, in the framework of the (2+1)-dimensional Einstein theory
with cosmological constant different families of exact solutions for cyclic sym-
metric stationary (static) metrics in the presence of a negative cosmological
constant have been derived. Specific branches of solutions in the general case are
determined via a straightforward integration. In this systematic approach, all
known cyclic symmetric solutions of the considered class are properly identified.
 5
Perfect Fluid Static Stars; Cosmological Solutions

The purpose of this chapter is to determine the static circularly (cyclic) symmet-
ric spacetimes coupled to perfect fluids via a straightforward integration of the
Einstein equations. The structural functions of the metric depend on the energy
density, which remains in general arbitrary. Spacetimes for fluids fulfilling linear
(barotropic) and polytropic state equations are explicitly derived. By the way,
we demonstrate here that the incompressible perfect fluid solution is the only
conformally flat (with vanishing of the Cotton tensor) circularly symmetric solu-
tion. Since there is no boundary to determine a zero-pressure surface, all these
solutions fall in the cosmological category. The perfect fluid is characterized by
a fluid velocity uμ and energy–momentum tensor
Tμν = (p + μ)uμ uν + pgμν , uμ uμ = −1, (5.1)
where μ and p are the energy density and the pressure; throughout this text,
the energy density will be denoted with μ or ρ, as are commonly used in
hydrodynamics.

5.1 Static Circularly Symmetric Fluid Solutions

The static circularly symmetric metric commonly is chosen in curvature coordi-
nates
g = −e2 ν(r) dt2 + e2 λ(r) dr2 + r2 dθ2 , (5.2)
therefore, the Einstein equations are
d d −2 λ(r)
Ett : λ (r) = κ r μ (r) e2 λ(r) , → e = −2κ rμ (r) →
dr  r dr
κ
e−2 λ(r) = C0 − 2 κ rμ (r) dr = C0 − M (r) ,
π
 r
M (r) := 2π r μ (r) dr, (5.3a)
 5.2 Incompressible Static Star 59
 r
d rp (r) dr
Err : ν (r) − κ re2 λ(r) p (r) = 0 → ν (r) = κ , (5.3b)
dr C0 − πκ M (r)
 2
d2 ν dν dν dλ
Eθθ : + − − κ pe2 λ = 0. (5.3c)
dr2 dr dr dr
By the way, the substitution of the derivatives of λ from (5.3a) and ν from
(5.3b) into equation (5.3c) yields the energy momentum conservation equation
(5.4) given below. The conservation law T μν ;ν = 0 gives rise to a single equation
equation after Oppenheimer and Volkoff (1939), namely
dp dν dp rp
= − (p + μ) → = −κ (p + μ) . (5.4)
dr dr dr C0 − κ M (r) /π
In hydrodynamics, usually a state equation – a dependence between the pressure
and the energy – has to be provided.
Moreover, an integral quantity arise from the combination of (5.3b) and (5.3c),
(Eθθ − Err /r)eν(r) , namely
 
d e−λ deν
= 0. (5.5)
dr r dr

5.1.1 Cotton Tensor Types

The Cotton tensor for this perfect fluid is
κ dμ(r) eν
C μ ν = − e−λ (r e−ν δtμ δνθ − δθμ δνt ). (5.6)
4 dr r
Since the Cotton vanishes for constant density μ(r) = μ0 then this kind of
solution is conformally flat. Therefore, the incompressible perfect fluid (μ(r) =
μ0 ) is conformally flat.
The search for its eigenvectors yields
λ1 = 0; V1 = [0, V 2 , 0], V 1μ = V 2 grr δμr , V 1μ V 1μ = (V 2 )2 grr , V1 = S1,
κ dμ(r) ieν V 1
λ2 = i e−λ ; V2 = [V 1 , 0, V 3 = − ], V2 = Z,
4 dr r
κ dμ(r) ieν V 1
λ3 = −i e−λ ; V3 = [V 1 , 0, V 3 = ], V3 = Z̄, (5.7)
4 dr r
consequently the corresponding tensor type is Type I: {S, Z, Z̄}.

5.2 Incompressible Static Star

For this class of solutions, the corresponding spacetimes are conformally flat; the
Cotton becomes Type O. The easiest case to handle is the fluid with constant
density μ(r) = μ0 , which integrates completely:
60 Perfect Fluid Static Stars; Cosmological Solutions

e−2 λ(r) = C0 − r2 κ μ0 ,
d κ p (r) r (μ0 + p (r)) B0 μ0 C0 − r2 κ μ0
p (r) = − → p(r) = ,
dr C0 − r κ μ0
2
1 − B0 C0 − r2 κ μ0
ν (r) = ν0 + ln 1 − B0 C0 − r2 κ μ0 , (5.8)

where C0 and B0 are integration constants which one fixes via boundary
conditions.

5.2.1 Collas Static Star with Constant Density μ0

In Collas (1977), the search for a perfect fluid solution was posed in the para-
graph: B Interior solutions. The condition that the component grr , at the
boundary of the mass distribution r0 , has to be 1, as for the exterior Minkowski
metric, gives rise to the Eq. (37) in Collas (1977), namely
 r0
e−2 λ(r) = C0 − 2 κ rμ (r) dr = 1 (5.9)

from which one determines C0 , which in the particular case of constant density
μ0 results in C0 = κμ0 r02 = 8πμ0 r02 . The pressure is expressible as
[1 + 8πμ0 (r02 − r2 )]1/2
p(r) = Bμ0 , (5.10)
[1 + 8πμ0 r02 ]1/2 − B[1 + 8πμ0 (r02 − r2 )]1/2
comparable with Eq. (40) of Collas (1977). Thus the requirement of vanishing
pressure at the boundary of the fluid ball fails to be fitted, hence the author
concludes that it is not possible to have a fluid with constant density μ and
nonzero pressure p = 0.
This dilemma found a solution in the forthcoming formulation for the constant
density μ0 and variable bounded pressure p(r) branch of metrics, but not allowing
for a match to the exterior Minkowski spacetime.

5.2.2 Giddings–Abott–Kuchař Static Star with μ0

In the work by Giddings et al. (1984) a nontrivial perfect fluid solution with
constant density and variable pressure is found. The first requirement to fulfill
was the vanishing of the pressure at some border R, which yields to C0 = R2 κ μ0 ,
therefore
√ √
B0 μ0 κ μ0 R2 − r2
p(r) = √ √ . (5.11)
1 − B0 κ μ0 R2 − r2
Next, the finite value of the pressure pc = p(r = 0) at the center r = 0 of the
ball fixes the constant B0 to be
√
B0 κ μ0 R = pc /(μ0 + pc ),
 5.2 Incompressible Static Star 61

consequently
√
R2 − r 2
p(r) = pc μ0 √
R(pc + μ0 ) − pc R2 − r2

2
1 − (r/R)
= pc  , (5.12a)
2
1 + μpc0 − μpc0 1 − (r/R)

e−2 λ(r) = κ μ0 (1 − r2 ) = κ μ0 R2 (1 − (r/R) ) = g rr ,

2
(5.12b)
    
μ0 pc pc 2
ν (r) = ν0 + ln + ln 1 + − 1 − (r/R) . (5.12c)
μ0 + pc μ0 μ0
Notice that at this boundary also g rr (rb ) → 0, grr (rb ) → ∞, which means that
one is facing a coordinate singularity there.
Accomplishing the coordinate transformations
r √ μ0
r = √ , θ  = R κ μ0 θ, ν0 = − ln , (5.13)
R κ μ0 μ0 + pc
one arrives, dropping primes, at the GAK metric, Eq. (118) of Giddings et al.
(1984):
 2
pc pc 1
g =− 1+ − 1 − r2 κ μ0 dt2 + dr2 + r2 dθ2
μ0 μ0 1 − r2 κ μ0
pc μ0 1 − r2 κ μ0
p (r) = , μ(r) = μ0 , M (r) = π μ0 r2 , (5.14)
pc + μ0 − pc 1 − r2 κ μ0
where p (r), μ (r), m (r) denote correspondingly the pressure, the mass density,
and the integral mass. The range of the circular coordinate θ has been restored
to be 0 ≤ θ ≤ 2π. In this formulation of the fluid solution with constant μ, as
before, the center is located at r = 0, p(r = 0) = pc , and the vanishing of the
√
pressure requires rb = 1/ κ μ0 , at which the metric component grr (rb ) → ∞,
therefore the matching to an exterior metric, what ever it may be, is not possible;
the continuity of the metric components is one of the main requirements to be
satisfied. The solution with μ0 is simply a cosmological one.
As was stated before, the solutions with constant energy density μ0 are Cotton
zero, i.e., conformally flat as in the (3 + 1) Schwarzschild–Kottler case.

5.2.3 Cornish–Frankel Static Star with μ0

The incompressible perfect fluid, μ(r) = μ0 , has been treated by various authors:
Collas (1977), Giddings et al. (1984), and again by Cornish and Frankel (1991)
in their Example C with the solution in the form
62 Perfect Fluid Static Stars; Cosmological Solutions
  2 
pc r 1/2 r −1 2
g =− 1+ 1 − 1 − ( )2 dt2 + 1 − ( )2 dr + r2 dθ2 ,
μ0 R R
   −1
r 2 1/2 pc r 2 1/2
p(r) = pc 1 − ( ) 1+ 1− 1−( ) . (5.15)
R μ0 R
The pressure vanishes on the boundary R of the star, but on this circumference
the metric exhibits a coordinate singularity, therefore the fluid ball blows up in
these confines.

5.3 Cornish–Frankel Static Polytropic Solutions

Cornish and Frankel (1991) successfully treated static polytropic perfect fluid
cases for the metric (5.2). The starting point in deriving this class of fluids is the
Oppenheimer–Volkoff pressure equation (5.4) together with the mass–energy–
density definition (5.3a). Isolating C0 π − κ M (r) from (5.4), one has

π rκ p (r) (μ (r) + p (r))

C0 π − κ M (r) = − d
, (5.16)
dr p (r)

which, differentiated with respect to r, yields

2
p (μ + p) r p ddrμ r p (μ + p) ddrp2
r (2 p − μ) + dp
+ dp
− 2 = 0. (5.17)
dp
dr dr
dr

For the family of polytropic fluids, the pressure and energy density can be
given as
p(r) = p0 Θ(r)n+1 , μ(r) = μ0 Θ(r)n , (5.18)

where n is the polytropic index, not necessarily an integer. Moreover, one may
impose one condition on the pressure and energy at the center of the fluid ball,
namely
p(r = 0) = p0 Θ(0)n+1 , μ(r = 0) = μ0 , Θ(r = 0) = 1. (5.19)

A second condition, such as the value of Θ(r = R) = 0, where R denotes

the boundary value of r, is not needed; on the contrary, if this were the case,
then a relation between the boundary value R and the center values of the
pressure and the energy density would arise. In any case, at the boundary R
a relation (a transcendent one) between Θ(R), μ0 , p0 , and R will arise; see
below.
The above equation (5.17), in terms of Θ, can be rearranged into the form
   
d
d ln (r) − ln Θ − (n + 1) ln (Θ) + (2 n + 3) ln (μ0 + p0 Θ) = 0.
dr
 5.3 Cornish–Frankel Static Polytropic Solutions 63

Therefore, one arrives at

2 n+3
dΘ (μ0 + p0 Θ)
= B0 r
dr Θn+1
2 
B0 r Θn+1
→ = 2 n+3 dΘ + A0 . (5.20)
2 (μ0 + p0 Θ)
The total integral is given in terms of polynomials of Θ, namely
 
n+1
Θn+1 1
− 2 n+3 dΘ = n+1 n+2
Skn , (5.21)
(μ0 + p0 Θ) μ0 p0
k=0

k (n + 1)! μ0 k 1
Skn (Θ(r)) = (−1) .
k! (n + 1 − k)! (n + 1 + k) (p0 Θ + μ0 )n+1+k
(5.22)
Hence, evaluating A0 at r = 0, Θ(r = 0) = 1, one arrives at

B0 2 1 
n+1
− r = n+1 n+2 [Skn (Θ(r)) − Skn (Θ = 1)] . (5.23)
2 μ0 p0
k=0

At this stage, let us see if one can fix the value of the integration constant B0 .
The evaluation of
κ π rκ p (μ + p) κr dr
e−2λ(r) = C0 − M (r) = − dp
=− (μ0 + p0 Θ) Θn+1
π dr
n + 1 dΘ
1 κ Θ2n+2 1
=− = ,
B0 n + 1 (μ0 + p0 Θ)2n+2 grr
1 n+1 1
gΘΘ =− , (5.24)
B0 κr2 (μ0 + p0 Θ)2n+2
therefore, the positiveness of the spatial metric components requires B0 to be
negative, B0 < 0, say B0 = −2. The integration of ν(r), (5.3b) is straightforward,
−2 n−2
exp (2ν) = N02 (μ0 + p0 Θ) .
Changing Θ to a more conventional radial coordinate ρ ≡ Θ, the CF solution
in the (t, ρ, φ) coordinates can be given as
−2 n−2 n+1 dρ2
ds2 = − (μ0 + p0 ρ) dt2 + + r2 dφ2 , (5.25)
2κ r2 (μ0 + p0 ρ)2n+2
1 
n+1
r(ρ)2 := [Skn (ρ) − Skn (ρ = 1)] , (5.26)
μ0 n+1 p0 n+2
k=0

k (n + 1)! μ0 k 1
Skn (ρ) = (−1) , (5.27)
k! (n + 1 − k)! (n + 1 + k) (μ0 + p0 ρ)n+1+k
p(ρ) = p0 ρn+1 , μ(ρ) = μ0 ρn . (5.28)
Since this solution does not allow for a zero-pressure circle p(ρ) = 0 except at
ρ = 0, this kind of solution is a cosmology that extends over the whole space.
64 Perfect Fluid Static Stars; Cosmological Solutions

As far as to the mass is concerned, let us return to the equations (5.24), with
B0 = −2,
κ 1 κ Θ2n+2 1
C0 − M (r) = = . (5.29)
π 2 n + 1 (μ0 + p0 Θ)2n+2 grr

From this equation one concludes that at the origin r = 0, and at the boundary
r = R one correspondingly has
κ 1 κ 1
C0 = M (r = 0) + , (5.30)
π 2 n + 1 (μ0 + p0 )2n+2
2n+2
κ 1 κ Θ(R) 1
C0 − M (r = R) = = . (5.31)
π 2 n + 1 (μ0 + p0 Θ(R))2n+2 grr (R)

From the second condition, it becomes clear that the choice Θ(R) = 0, made in
Cornish and Frankel (1991), is not an adequate one since in that case g rr (R) = 0,
grr (R) → ∞–a black fluid ball-plate. Moreover, as was been mentioned above,
the condition Θ(R) = 0 impose a dependence of the form R(μ0 , p0 ), which is also
unwelcome. Consequently, it is better to adopt as the mass within the radius r
the quantity
2 n+2
π Θ(r) π
M (Θ) = M (0) + 2 n+2 − 2 n+2 .
2 (μ0 + p0 ) (n + 1) 2 (μ0 + p0 Θ(r)) (n + 1)
In particular, the mass inside the ball with r = R, is given by
2n+2
π 1 1 π 1 Θ(R)
M (r = R) = M (0) + 2n+2 − .
2 n + 1 (μ0 + p0 ) 2 n + 1 (μ0 + p0 Θ(R))2n+2

On the other hand, at this boundary r = R, the metric components are finite if
Θ(R) = 0, and μ0 + p0 Θ(R) = 0; see (5.31). Recall that Θ(R) is a root of the
algebraic polynomial equation of degree n + k + 1 for Θ(R) arising from

1 
n+1
R2 = n+1 n+2
[Skn (Θ(R)) − Skn (Θ(0) = 1)] .
μ0 p0
k=0

This is a complete analytic solution, for any polytropic index n and has no
parallel comparing with (3 + 1)-dimensional case, where only few polytropic
solutions are known.

5.3.1 Static Star with a Stiff Matter p(r) = μ(r)

The integration of the Einstein equations yields
1 C1 −C1 r2 /2
e−C1 r /2 , e2 λ(r) = −
2 2
e2 ν(r) = e , p(r) = C2 e C1 r /2 ,
2κ C2 2κ C2
p(0) = p0 = C2 , C1 = −2κ C2 = −2κ p0 ,
 5.3 Cornish–Frankel Static Polytropic Solutions 65

hence, this simple conformally flat solution can be given as

2 2
d s2 = −eκ p0 r dt2 + eκ p0 r dr2 + r2 dθ2 ,
p(r) = p0 e−κ p0 r ,
2
(5.32)
from which its cosmological character becomes apparent.

5.3.2 Static Star with Pure Radiation p = μ(r)/2

The static radiation or ultra-relativistic particles solution p = μ(r)/2 is given by
the metric (corrected) and pressure of the form
 r −2 2  r −4 2
g = − 1 − ( )2 dt + 1 − ( )2 dr + r2 dθ2 ,
R R
 r 3 2
p(r) = γ 1 − ( )2 , γ = . (5.33)
R κ R2
In Cornish and Frankel (1991), (Example B ), there is a misprint in the component
2
gtt = − 1 − ( Rr )2 , instead of the correct one of (5.33) with power −2.
 6
Static Perfect Fluid Stars with Λ

The purpose of this chapter is to determine all static circularly (cyclic) symmet-
ric spacetimes with a cosmological constant coupled to perfect fluids with and
without zero-pressure surfaces via a straightforward integration of the Einstein
equations. The structural functions of the metrics depend on the energy den-
sity, which remains in general arbitrary. Spacetimes for fluids fulfilling linear
(barotropic) and polytropic state equations are explicitly derived; they con-
tain, among other families, stiff matter, incoherent radiation, and non-relativistic
degenerate fermions. By the way, we demonstrate here that the incompressible
perfect fluid solution is the only conformally flat – in the sense of the vanishing
of the Cotton tensor – circularly symmetric solution.
In Section 6.1, the Einstein equations for the static circularly symmetric (2+1)
metric with a cosmological constant coupled to a perfect fluid solution with vari-
able density ρ and pressure p are exhibited and integrated; in this chapter we
are using ρ to denote the energy density instead of the μ used in the previous
chapters. Section 6.2 is devoted to representing this whole class of spacetimes in
a canonical coordinate system. For a given equation of state of the form p = p(ρ),
certain particular families of perfect fluid solutions are derived; as concrete exam-
ples, the subcases of fluids obeying the barotropic law p = γ ρ in Section 6.3, and
those fluids subjected to a polytropic law p = C ργ in Section 6.4, are derived.
Moreover, the incompressible fluid, ρ = const., is the only conformally flat static
circularly symmetric solution coupled to perfect fluids. In Section 6.5, from the
Oppenheimer–Volkoff equation certain properties of the studied solutions are
established: for positive pressure p and positive density ρ, a microscopically sta-
ble fluid possesses a monotonically decreasing energy density; and conversely,
these results are in close relation with the ones reported by Cruz and Zanelli
(1995), which unfortunately contains various misprints.
In Section 6.6, to facilitate the comparison of the interior Schwarzschild (3+1)
solution with cosmological constant, the incompressible perfect fluid (2 + 1) solu-
tion is derived. With this aim in mind, we search for an adequate representation
 6.1 Equations for a (2+1) Static Perfect Fluid Metric 67

of the corresponding structural functions and related quantities of these (3 + 1)

and (2 + 1) spacetimes. A comparison table is presented. Therefore, by dimen-
sional reduction from the interior Schwarzschild with Λ solution, one can obtain
the incompressible perfect fluid (2+1) solution.

6.1 Equations for a (2+1) Static Perfect Fluid Metric

As far as we know, in most of the publications dealing with the search for perfect
fluid solutions in (2+1) gravity – see for instance Cornish and Frankel (1991);
Cruz and Zanelli (1995) – in the integration process the energy–momentum con-
servation, i.e., the Oppenheimer–Volkoff equation, has been used as a clue to
obtain the desired results. By contrast, we prefer to solve directly the corre-
sponding Einstein equations; in such case the energy–momentum conservation
equations trivially hold.
The line element of static circularly symmetric (2+1) spacetimes, in coordi-
nates {t, r, θ}, is given by
dr2
ds2 = −N (r)2 dt2 + + r2 dθ2 . (6.1)
G(r)2
The Einstein equations with cosmological constant for a perfect fluid energy–
momentum tensor Tab :
1
Gab = Rab − gab R = κTab − Λgab ,
2
Tab = (p + ρ)ua ub + pgab , ua = −N δat , (6.2)

for the metric (6.1) explicitly amount to

N 2 dG2
Gtt = = −N 2 (κρ + Λ),
2r dr
1 dN 1
Grr =− = − 2 (κp − Λ),
rN dr G 
r2 d2 N 1 dN dG2
Gθθ =− G2 2 + = −r2 (κp − Λ). (6.3)
N dr 2 dr dr
Notice that the combination of the Einstein equations r2 G(r)2 Grr − Gθθ = 0,
for N (r) = const., gives rise to an important equation, namely
   
d2 N dN r dG2 d G dN
G2 r2 2 + r − G2 = 0, −→ = 0, (6.4)
dr dr 2 dr dr r dr
which will be extensively used throughout this paper.
The energy conservation yields the Oppenheimer–Volkoff equation (Oppen-
heimer and Volkoff, 1939; see subsection 6.5):
dp dN
N + [p(r) + ρ(r)] = 0, (6.5)
dr dr
68 Static Perfect Fluid Stars with Λ

which used in the Cotton tensor yields to

 
κ dρ r α φ N α φ
α
C β= F − δ tδ β + δ φδ β . (6.6)
2 dr N r
Thus the eigenvectors are
κ dρ
λ1 = 0, S; λ2 = iF , Z, λ3 = λ̄2 , Z̄. (6.7)
2 dr
Therefore the Cotton tensor is of Type I: {S, Z, Z̄}. The incompressible fluid,
ρ(r) = ρ0 = constant, determines a conformally flat spacetime; Type O. There-
fore, we conclude that the incompressible perfect fluid ρ(N ) = const. solution
for static circularly symmetric spacetimes is unique. This result can be stated as
a theorem: the incompressible perfect fluid solution is the only conformally flat
static circularly symmetric spacetime for a perfect fluid source with or without
cosmological constant. The similarity of this theorem with the corresponding one
formulated for Schwarzschild metric, see Garcı́a (1988), is noteworthy. We shall
return to this spacetime in Section 6.6.

6.1.1 General Perfect Fluid Solution with Variable ρ(r)

In this section, we derive the most general static circularly symmetric solution
via a straightforward integration of the Einstein equation with Λ for a perfect
fluid following the approach presented in Garcı́a and Campuzano (2003). It is
easy to establish that the structural functions G(r) and N (r) can be integrated
in quadratures.
Integrating the Gtt –(6.3), one arrives at
 r  r
G(r)2 = −Λr2 − 2κ rρ(r)dr ≡ C − Λr2 − 2κ rρ(r)dr, (6.8)
0

where C is an integration constant in which we have incorporated the constant

value of the integral at the lower integration limit r = 0, thus the remaining
integral depends on the upper integration limit r; we use the r–notation for
the upper integration limit as well as to denote the integration variable in the
integrand because this notation is unambiguous; this convention will be used
hereafter. From the second relation of (6.4), one obtains
dN r
= n1 (6.9)
dr G(r)
therefore  
r r
r r
N (r) = n1 dr ≡ n0 + n1 dr (6.10)
0 G(r) G(r)
and the evaluation of the pressure p(r) yields
1
κp(r) = (n1 G(r) + Λ N (r)) . (6.11)
N (r)
 6.2 Canonical Coordinate System {t, N, θ} 69

The metric (6.1), with G(r) from (6.8), and N (r) from (6.10), determines the
general static circularly symmetric (2+1) solution of the Einstein equations (6.3)
with Λ for a perfect fluid, characterized by a pressure given by (6.11), and an
arbitrary density ρ(r). The fluid-velocity is aligned along the timelike Killing
direction ∂t . To deal with realistic matter distributions one has to impose pos-
itivity conditions on the density, ρ > 0, and the pressure, p > 0, requiring
additionally ρ > p.
For a finite distributed fluid, the pressure p becomes zero at the boundary,
say r = a; this value of the radial coordinate r is determined as solution of the
equation p(r) = 0.
For non-vanishing cosmological constant, assuming that the values of the struc-
tural functions at the boundary r = a are N (a) and G(a), the vanishing at r = a
of the pressure p(r), given by (6.11), requires n1 = −ΛN (a)/G(a), hence
Λ
κ p(r) = (N (r) G(a) − N (a) G(r)) . (6.12)
N (r)G(a)
If one is interested in matching the obtained perfect fluid metric with a vacuum
metric with cosmological constant Λ, the plausible choice at hand is the anti-de
Sitter metric, with Λ = −1/l2 , for which G(a) = N (a) = −M∞ + a2 /l2 at the
boundary r = a. Incidentally, for a cosmological constant different from zero,
there is no room for dust. The zero character of the pressure would yield the
vanishing of the density, and consequently the metric reduces to the (anti-)de
Sitter spacetime.
For vanishing cosmological constant, the expression of the pressure (6.11) is
G(r)
κp(r) = n1 , (6.13)
N (r)
from which it becomes apparent that the corresponding solution represents a
cosmological spacetime; there is no a surface of vanishing pressure.
For vanishing Λ and zero pressure, the situation slightly changes: the function
N becomes a constant, and the corresponding metric can be written as
dr2
ds2 = −dt2 + r + r2 dθ2 , (6.14)
C − 2κ rρ(r)dr
for any density function ρ. Of course, the choice of ρ is restricted by physically
reasonable matter distributions.

6.2 Canonical Coordinate System {t, N, θ}

In this section we show that an alternative formulation of our general solution
can be achieved in coordinates {t, N, θ}. Indeed, from (6.9) for the derivative of
the function N , in which we are including (without any loss of generality) the
constant n1 , N/n1 −→ N, n1 t −→ t, one obtains
70 Static Perfect Fluid Stars with Λ

dN dr
= , (6.15)
r G(r)
hence
 N
2
r = C0 + 2 G dN . (6.16)

To derive G as a function of the new variable N , one uses the Gtt –(6.3) in the
form of
G dG = −(κ ρ + Λ)r dr = −(κ ρ + Λ)GdN, (6.17)

therefore, integrating one gets

 N
G(N ) = C1 − Λ N − κ ρ(N ) dN . (6.18)

Substituting this function G into the expression of r one obtains

 N N
H(N ) := r2 = C0 + 2 C1 N − Λ N 2 − 2 κ ρ(S) dSdN . (6.19)

Finally, our metric in the new coordinates {t, N, θ} amounts to

dN 2
ds2 = −N 2 dt2 + + H(N ) dθ2 , (6.20)
H(N )
which is characterized by pressure
 N
C1 1 1
p(N ) = − ρ(N ) dN (6.21)
κ N N
and an arbitrary energy density ρ(N ) depending on the variable N , both
functions p and ρ have to be positive.
The metric (6.20) together with the function H from (6.19) give an alternative
representation of our general solution. This representation will be used to derive
particular solutions for a given state equation of the form p = p(ρ). In this
approach the expression of the pressure (6.21) play a central role.

6.3 Perfect Fluid Solutions for a Barotropic Law p = γ ρ

Although in the above section we provide the general solution to the posed
question of finding all solutions for circularly symmetric static metrics in (2+1)-
gravity coupled to a perfect fluid in the presence of the cosmological constant,
from the physical point of view, even in this lower-dimensional spacetime, it is of
interest to analyze certain specific cases: for instance, the solution corresponding
to a fluid obeying the barotropic law p = γ ρ, or the more complicated case of a
polytropic law p = ργ .
 6.4 Perfect Fluid Solutions for a Polytropic Law p = Cργ 71

The barotropic law establishes a linear relation between pressure and energy
density of the form
p(N ) = γ ρ(N ). (6.22)

Substituting p(N ) from Eq. (6.23) into this relation, one gets
 N
C1
− ρ(N ) dN = γ N ρ(N ). (6.23)
κ
Differentiating this equation with respect to the variable N , one obtains
d 1
(N ρ) + (N ρ) = 0, (6.24)
dN γN
which has as general integral
γ − 1 − γ+1
ρ(N ) = C2 N γ , (6.25)
γ2
where C2 is an integration constant. Since we arrived at the above simple linear
equation (6.24) through differentiation, then one has to replace the obtained
result into the relation (6.23), or equivalently into (6.22), to see if there arises
any constraint from it:
C1 1
p(N ) = + γ ρ(N ) = γ ρ(N ) −→ C1 = 0, (6.26)
κ N
in such manner, we establish that the constant C1 vanishes. Replacing the func-
tion ρ(N ) from (6.25) into the expression of H(N ), (6.19), and accomplishing
the integration one arrives at

H(N ) = C0 − ΛN 2 + C2 N (γ−1)/γ = r2 . (6.27)

Thus, the metric for a perfect fluid fulfilling a barotropic state equation in
coordinates {t, N, θ} is given by
dN 2 γ−1
ds2 = −N 2 dt2 + γ−1 +(C0 −ΛN 2 +2C2 κN γ )dθ2 . (6.28)
C0 − ΛN 2 + 2κC2 N γ

To give this solution in terms of the radial variable r, one has to be able to solve
the algebraic equation, in general a transcendent one, for N = N (r). Within this
class of solutions merits mention the pure radiation γ = 1/2, and the stiff matter
γ = 1 where the speed of sound equals the speed of light.

6.4 Perfect Fluid Solutions for a Polytropic Law p = Cργ

This section is devoted to the derivation of all solutions obeying the poly-
tropic law
p = C ργ . (6.29)
72 Static Perfect Fluid Stars with Λ

Using again the expression of p(N ) from (6.21), the above polytropic relation
can be written as
 N
C1
− ρ(N ) dN = C N ργ (N ). (6.30)
κ
Differentiating with respect to N , one obtains
d
− ρ=C (N ργ ) (6.31)
dN
which, by introducing the auxiliary function Z := N 1/γ ρ, can be written as
 
1 1
d(Z γ−1 ) + d(N (γ−1)/γ ) = 0 → d (ργ−1 + ) N (γ−1)/γ = 0, (6.32)
C C
therefore, integrating
 
1
ργ−1 + N (γ−1)/γ = B C −(γ−1)/γ , (6.33)
C
where B is an integration constant. Consequently, the general integral of this
equation can be given as
−1 −1
 −1 γ−1
 γ−1
1

ρ=C γ N γ B−C γ N γ . (6.34)

Entering this ρ into the equation (6.30), taking into account that the integral of
the density ρ amounts to
 N  N   γ−1
γ
−1 γ−1
ρ(N ) dN = − d B−C γ N γ , (6.35)

one arrives at
n1 C1
p(N ) = + C ργ = C ργ −→ C1 = 0. (6.36)
κ N
Considering that the first integral of ρ is given by (6.35), the expression of the
structural function H(N ) becomes
 N  γ−1
γ
−1 γ−1
H(N ) = r2 = C0 − Λ N 2 + 2 κ B−C γ N γ dN. (6.37)

Notice that the integral can be expressed in terms of hypergeometric functions,

hence

γ γ γ
H(N ) = C0 − Λ N + 2 κ B
2 γ/(γ−1)
NF [ ,− ], [ + 1],
γ−1 γ−1 γ−1
N (γ−1)/γ C −1/γ C1 −1 . (6.38)

Summarizing, in the case of a polytropic equation of state p = C ργ , the metric

is given by
 6.5 Oppenheimer–Volkoff Equation 73

dN 2
ds2 = −N 2 dt2 +
N  −1 γ−1
 γ−1
γ

C0 − Λ N 2 + 2κ B−C γ N γ dN
  N 
−1 γ−1
 γ−1
γ

+ C0 − Λ N 2 + 2κ B−C γ N γ dN dφ2 . (6.39)

while the pressure and density are:

1   γ−1   γ−1
γ 1
−1 γ−1 −1 −1 −1 γ−1
p= B−C γ N γ , ρ=C γ N γ B−C γ N γ . (6.40)
N
Within this class of solutions, the nonrelativistic degenerate fermions γ = 2,
and the nonrelativistic matter and radiation γ = 3/2, are worthy of mention.
Incidentally, the study of static circularly symmetric cosmological (absence of
the cosmological constant) spacetimes, coupled to perfect fluids fulfilling the
polytropic law was accomplished in Cornish and Frankel (1991).

6.5 Oppenheimer–Volkoff Equation

Although when Einstein equations have been fulfilled the Bianchi identities
(energy–momentum conservation) trivially hold, it is of interest to establish cer-
tain properties arising from the Oppenheimer–Volkoff equation in Oppenheimer
and Volkoff (1939): see for instance Cruz and Zanelli (1995) in (2+1) gravity. An
alternative derivation of this equation consists in differentiating with respect to
r the Einstein Grr –equation (6.3); this yields
   
dp 1 dN dG2 G2 G2 d2 N 1 dN 2
κ = − + − ( ) , (6.41)
dr rN dr dr r rN dr2 N dr
2
substituting the second derivative ddrN2 from (6.4), and the first derivative
dN
dr from the Grr -equation into(6.41) one arrives at the Oppenheimer–Volkoff
equation:
dp r
= − 2 (κ p − Λ)(ρ + p). (6.42)
dr G
At the circle of vanishing pressure p(a) = 0 at r = a one obtains that the pressure
gradient amounts to
dp Λa
|r=a = ρ(a). (6.43)
dr G(a)2
Since inside the circle the pressure is positive, p(r < a) > 0, hence at the circle
r = a the pressure gradient has to be non-positive; consequently the cosmological
constant ought to be negative, Λ = −1/l2 < 0. We shall continue to use Λ instead
of −1/l2 , keeping in mind that Λ is a negative constant.
The definition of the mass contained in the circle of radius a is given by
 a
M := 2π ρ(r) r dr, (6.44)
0
74 Static Perfect Fluid Stars with Λ

and since the metric components grr = 1/G(r)2 has to be positive in the domain
of definition of the solution, then there exits an upper limit for the mass, namely
π
M ≤ (C − Λ a2 ). (6.45)
κ
Matter is said to be microscopically stable if dp/dρ ≥ 0, which is equivalent
to the statement that the speed of sound is less than the velocity of light. Since
(6.41) can be written as
dp r dρ
= − 2 (κ p − Λ)(ρ + p)/ , (6.46)
dρ G dr
one concludes that for a microscopically stable fluid with positive pressure p and
positive density ρ, this density is monotonically decreasing dρ/dr < 0.
For our general solution in coordinates {t, N, θ}, metric (6.20), from the
expression (6.21) for the pressure, one establishes
dp 1 dρ
= − (ρ + p)/ , (6.47)
dρ N dN
therefore the density is monotonically decreasing dρ/dN < 0 if the matter is
microscopically stable dp/dρ ≥ 0, and conversely.
Moreover, our fluids, fulfilling the barotropic state equation p = γρ, γ > 0,
as well as those ones obeying the polytropic law p = Cργ , C > 0, γ > 0, are
microscopically stable fluids.

6.6 Perfect Fluid Solution with Constant Density

As has been established above, for ρ = const. the Cotton tensor vanishes, and
consequently the corresponding conformally flat space is unique; we shall refer
to it as the incompressible perfect fluid.
In this section, it is shown that one can achieve a full correspondence of the
metrics and structural functions for incompressible perfect fluids in (2+1) and
(3+1) gravities. By an appropriate choice of the constant densities and cos-
mological constants, via a dimensional reduction (freezing of one of the spatial
coordinates of the (3+1) spacetime), one obtains the (2+1) metric structure from
the (3+1) solution. To achieve the intended purpose, the conformally flat static
spherically symmetric perfect fluid (3+1) solution with cosmological constant
is presented in a form which allows comparison with the incompressible static
circularly symmetric perfect fluid with Λ–term of the (2+1) gravity.
In the canonical coordinate system {t, N, θ}, for ρ = ρ0 , the metric, the expres-
sion of the function H, which in its turn establishes the relation to the radial
coordinate r, and the pressure are given by:
dN 2
ds2 = −N 2 dt2 + + Hdθ2 , (6.48)
H
H = C0 + 2C1 N − (Λ + κρ0 )N 2 =: r2 , (6.49)
C1 1
p = −ρ0 + . (6.50)
κ N
 6.6 Perfect Fluid Solution with Constant Density 75

This unfamiliar-looking solution can be given in terms of the radial variable r

by expressing N as function of r, N = N (r).
Having in mind the comparison of the (2+1) incompressible perfect fluid with
its (3+1) relative – the Schwarzschild interior solution – we shall derive it from
the very beginning by integrating the Einstein equations (6.3) in coordinates
{t, r, θ}.
For ρ = const., the integral of (6.8) gives

G(r) = C − (κρ + Λ)r2 . (6.51)

Substituting G(r) from (6.51) into (6.10), one obtains

n1
N (r) = n0 − G(r), (6.52)
Λ + κρ

which can be written as N (r) = C1 + C2 G(r).

The evaluation of pressure p(r) from (6.3) yields

1
κp(r) = [n1 κρ G(r) + n0 Λ(κρ + Λ)] . (6.53)
(κρ + Λ)N (r)

This pressure has to vanish at the boundary r = a, which imposes a relation

on the constants: n0 = −n1 κρG(a)/[Λ(κρ + Λ)], where G(a) is the value of the
function G(r) at the boundary, i.e., G(a) is equal to the external value for the
G(r) corresponding with the vacuum solution plus Λ. A similar comment applies
to N (a). Replacing n0 in (6.52), the function N (r) becomes
n1
N (r) = − [κρG(a) + ΛG(r)] . (6.54)
Λ(κρ + Λ)

Evaluating N (r) at r = a, establishes that n1 = −Λ N (a)/G(a). Consequently,

N (r) amounts to

N (a)
N (r) = [κρ G(a) + Λ G(r)] . (6.55)
G(a)(κρ + Λ)

Substituting n0 , n1 , and N (r) into (6.53), one gets

G(a) − G(r)
p(r) = ρ Λ . (6.56)
κρG(a) + ΛG(r)

Summarizing, the (2+1) metric for an incompressible perfect fluid is given by

dr2
ds2 = −N (r)2 dt2 + + r2 dθ2 (6.57)
G(r)2

with structural functions G(r) from (6.51) and N (r) from (6.55) and character-
ized by a density ρ = const. and pressure p given by (6.56).
76 Static Perfect Fluid Stars with Λ

6.6.1 (3+1) Static Spherically Symmetric Perfect Fluid Solution

In this subsection we review the main structure of the interior perfect fluid solu-
tion in the presence of the cosmological constant Λ – the interior Schwarzschild
metric with Λ – for the (3 + 1) static spherically symmetric metric of the form

2 dr2  
ds2 = −N (r) dt2 + 2
+ r2 dθ2 + sin2 θdφ2 . (6.58)
G(r)

The Einstein equations with cosmological constant for perfect fluids for the
metric (6.58) explicitly amount to
 
N2 dG2
Gtt = 2 r + G2 − 1 = −N 2 (κρ + Λ),
r dr
 
1 dN 1
Grr = − 2 2 2rG2 − N + N G2 = − 2 (κp − Λ),
G Nr dr G
 2 2

r dN 1 dG d N r dN dG2
Gθθ = − G2 + N + rG2 2 + = −r2 (κp − Λ),
N dr 2 dr dr 2 dr dr
Gφφ = sin2 θ Gθθ . (6.59)

In what follows we shall omit the dependence on r of the functions we are dealing
with, except when needed for clarity. The Einstein equations for a perfect fluid
energy–momentum tensor in four dimensions have the same form as the ones
in three dimensions, except for the modifications due to the change of dimen-
sionality; for instance, the expressions of the scalar curvatures R are different.
Because of the corresponding equations can be found in textbooks, we do not
exhibit them here explicitly.
Since we are interested in conformally flat solutions, we require the vanishing of
the conformal Weyl tensor, which for static spherically symmetric perfect fluids
fulfills the following equation
 
d G2 − 1
= 0 −→ G(r) = 1 + c0 r2 . (6.60)
dr r2

On the other hand, from the equation Gtt , one arrives at


1
G(r) = 1 − (κρ + Λ)r2 , (6.61)
3
therefore, comparing with (6.60), one has c0 = −(κρ + Λ)/3, −→ ρ = const.
Hence, the resulting solution corresponds to an incompressible perfect fluid;
see Adler et al. (1965).
Moreover, from (r2 G2 Grr − Gθθ ) = 0, taking into account the form of the
function G from (6.61), the general expression of N (r) is

N (r) = C1 + C2 G(r). (6.62)

 6.6 Perfect Fluid Solution with Constant Density 77

The evaluation of the pressure p, from Grr Einstein equation, yields

1
κp(r) = [C1 (2Λ − κρ) − 3C2 κ ρ G(r))] . (6.63)
3N (r)
where G(r) and N (r) are determined in (6.61) and (6.62), respectively.
This result can be stated in the form of a generalization of the Gürses and
Gürsey (1975) theorem to the case of Λ: the only conformally flat spherically
symmetric static solution to the Einstein equations with cosmological constant
for a perfect fluid is given by the metric (6.58) with structural functions G(r)
and N (r) defined respectively by (6.61) and (6.62). Moreover, replacing in the
metric (6.58) sin2 θ by sinh2 θ, and θ2 , one obtains correspondingly the pseudo-
spherical and flat branches of the solutions.
The constants C1 and C2 are determined through the values of structural
functions at the boundary r = a, where the pressure vanishes, p(r = a) = 0;
they are:
G(a) N (a) 2Λ − κρ
C1 = 3 C2 κρ , C2 = , (6.64)
2Λ − κρ 2G(a) Λ + κρ
where, G(a) is the value of the function G(r) at the boundary r = a, i.e., G(a) is
equal to the external value of G(r) corresponding to the vacuum plus Λ solution.
A similar comment applies to N (a). We shall return to this point at the end of
this section.
Substituting the expressions of C1 and C2 into Eq. (6.62), one has
N (a)
N (r) = [3κρG(a) + (2Λ − κρ)G(r)] . (6.65)
2G(a)(κρ + Λ)
Replacing C1 , C2 and the above expression of N (r) into (6.63), one gets
G(a) − G(r)
p(r) = ρ(2Λ − κρ) . (6.66)
3κρG(a) + (2Λ − κρ)G(r)
For the external Schwarzschild with Λ solution, known also the Kottler solu-
tion, Stephani et al. (2003); Kottler (1918), the functions N (r) and G(r) are
equal one to another, N (r) = G(r), namely

2m Λ 2
N (r) = G(r) = 1 − − r , for r ≥ a. (6.67)
r 3
Evaluating the mass contained in the sphere of radius a for a constant density
κρ, one obtains 2m = κρa3 /3, therefore

κρ a3 Λ
N (r) = G(r) = 1 − − r2 , for r ≥ a, (6.68)
3 r 3
consequently at r = a, one has

κρ + Λ 2
N (a) = G(a) = 1− a . (6.69)
3
78 Static Perfect Fluid Stars with Λ

In the limit of vanishing cosmological constant, Λ = 0, one arrives at the interior

Schwarzschild solution.

6.6.2 Comparison Table

A comparison table of perfect fluid solutions with constant ρ is given:

Table 6.1 Incompressible perfect fluid solutions

(3 + 1) solution (2 + 1) solution
dr 2 dr 2
ds2 = −N 2 dt2 + G2
+ r 2 dΩ2 ds2 = −N 2 dt2 + G2
+ r 2 dφ2
G2 = 1 − 13 (κρ + Λ)r 2 G2 = C − (κρ + Λ)r 2
1 N (a) 1 N (a)
N = 2(κρ+Λ) G(a)
[3κρG(a) N = (κρ+Λ) G(a)
[κρG(a) + ΛG(r)]
+(2Λ − κρ)G(r)] +ΛG(r)]
(2Λ−κρ)(G(a)−G(r)) G(a)−G(r)
p = ρ 3κρG(a)+(2Λ−κρ)G(r) p = ρΛ κρG(a)+ΛG(r)

Kottler: (anti-)de Sitter:

2m
G(a)2 = 1 − r
− 13 Λr 2 ; G(a)2 = −M∞ − Λa2 ;
2m = κρa3 /3 C = κρa2 − M∞ > 0
N (a) = G(a) N (a) = G(a)

2Λ4 − κ4 ρ4 → 6Λ3 , κ4 ρ4 → 2κ3 ρ3

G4 (r) → G3 (r), N4 (r) → N3 (r), κ4 p4 (r) → 2κ3 p3 (r)

Comparing the structure corresponding to perfect fluid solutions with con-

stant ρ in (3+1) gravity with the structure of the (2+1) incompressible perfect
fluid solution one arrives at the following correspondence: 2Λ4 − κ4 ρ4 → 6Λ3 ,
3κ4 ρ4 → 6κ3 ρ3 , which yields G4 (r) → G3 (r), N4 (r) → N3 (r), κ4 p4 (r) →
2κ3 p3 (r). Remembering that in (2+1) gravity there is no Newtonian limit; the
choice of κ3 is free, thus by selecting κ3 appropriately one can achieve that
p4 (r) → p3 (r) and ρ4 → ρ3 .
From this comparison table one can easily conclude that the (2+1) incom-
pressible perfect fluid can be derived from the Schwarzschild interior metric
by a simple dimensional reduction: freezing one of the spatial coordinates, say
θ = π/2, in the (3+1) solution, one obtains the corresponding (2+1) spacetime.
Since we accomplished a scaling transformation of the r-coordinate, accompa-
nied with the inverse scaling of the angular coordinate φ, one may argue that a
conical singularity could arise; one may overcome this problem by saying that
the angular coordinate should be fixed once one brings the canonical form to the
(2+1) metric with G3 (r) = 1 − (κρ + Λ)r2 .
In this chapter all perfect fluid solutions for the static circularly symmetric
spacetime have been derived. The general solution is presented in the standard
 6.6 Perfect Fluid Solution with Constant Density 79

coordinate system {t, r, θ}, and alternatively, in a system – the canonical one –
with coordinates {t, N, θ}. From the physical point of view, particularly interest-
ing are those fluids fulfilling the linear (barotropic) equation of state, p = γρ, as
well as those subjected to the polytropic law p = ργ ; both families are derived
in details from our general metric referred to the coordinate system {t, N, θ}.
Therefore, the derived solutions describe, among other things, stiff matter, pure
radiation, incoherent radiation, nonrelativistic degenerate fermions, etc. The
incompressible perfect fluid solution with cosmological constant of the (2+1)
gravity is singled out among all static circularly spacetimes as the only confor-
mally flat space – its Cotton tensor vanishes – sharing this conformally flatness
property with its (3+1) relative, the Schwarzschild interior perfect (incompress-
ible) fluid solution with Λ; a comparison table for these incompressible fluids is
included.
 7
Hydrodynamic Equilibrium

Under Buchdahl’s conditions on the behavior of the density and the pressure for
regular fluid static stars in the presence of a cosmological constant, the bounds of
the mass are determined in any dimension. Here we work out the n-dimensional
case because of its parallelism with the 3D case.
For (2 + 1)-dimensional perfect fluid stars in hydrodynamic equilibrium there
are no bounds on the mass, except for their positiveness; the metric for a con-
stant density distribution is derived, and its matching with the external static
solution with a negative cosmological constant is accomplished. In the (d ≥ 4)-
dimensional case the existence of bounds for the mass is established. The metric
for a constant density is derived and its matching with the external static solu-
tion is carried out. Some mistakes in previous works on the topic are pointed
out.

7.1 Generalized Buchdahl’s Theorem

The main objective of this section is to establish that a generalized Buchdahl’s
theorem, Buchdahl (1959), holds in d ≥ 4-dimensions. For lower dimensions,
d < 4, there is no room for bounds of the mass distribution.

Theorem 7.1 If a perfect fluid distribution fulfills the conditions:

● it is described by a one-parameter state equation p = p(μ),

● dμ
the density is positive, μ > 0, and monotonically decreasing, dr < 0,
● dp
it is microscopically stable, dμ ≥ 0 → dp
dr ≤ 0,

in d dimensions, there is a bound on the density given by


4
M (R) (d − 2)
2 (d − 2) − 2 Λ R 2 (d − 1) (d − 2) 2mskt
≤ 2 + 2 ≥ d−3 , (7.1)
R d−3
(d − 1) (d − 1) R
 7.2 Stellar Equilibrium in (2 + 1) Dimensions with Λ 81

where the subscript skt stands for Schwarzschild–Kottler–Tangherlini; see Kottler

(1918), Tangherlini (1963). For vanishing cosmological constant one gets
2
1 (d − 2) 2
M (R) ≤ 2 2 ≥ mskt . (7.2)
R d−3
(d − 1) R d−3
Moreover, for four dimensions, d = 4, this inequality reduces just to the well-
known expression
1 8 2
M (R) ≤ ≥ mskt . (7.3)
R 9 R
In (2+1) dimensions there is no bound on the mass.

7.2 Stellar Equilibrium in (2 + 1) Dimensions with Λ

The work Cruz and Zanelli (1995) is devoted to the study of the equilibrium
for static perfect fluid solutions with cosmological constant, generalizing in this
manner the GAK analysis. Unfortunately, their work suffers from various mis-
prints; see Garcı́a (2014). Hence, we shall address the topic from the very
beginning.
The Einstein’s equations for a static (2 + 1) metric in curvature coordinates

g = −e2 ν(r) dt2 + e2 λ(r) dr2 + r2 dθ2 (7.4)

for a perfect fluid in the presence of a cosmological constant, Tμν = (p(r) +

μ(r))uμ uν + p(r) gμν , uα = eν(r) δαt , can be given as
1
E β α := Rβ α − δ β α R − κT β α + Λδ β α ,
2
T β α = −ρ(r)δ β t δ t α + p(r) δ β r δ r α + p(r) δ β θ δ θ α , (7.5)

explicitly
d κ
E11 = 0 : λ = r (κ μ + Λ)e2 λ → e−2 λ(r) = C0 − m (r) ,
dr
 r  r π
π π
m (r) := 2π r (μ + Λ/κ) dr = 2π r μ dr + Λ r2 =: M (r) + Λ r2 ,
κ κ
(7.6a)
 r
dν (κp − Λ) r dr
E22 = 0 : − re2 λ(r) (κ p − Λ) = 0 → ν = , (7.6b)
dr C0 − πκ m (r)
 2  
d2 d d d
3
E3 = 0 : ν+ ν − ν λ + (Λ − κ p) e2 λ = 0. (7.6c)
dr2 dr dr dr
The substitution of the derivative dν
dr from (7.6b) into (7.6c) yields the same
equation arising from the energy–momentum conservation law T αβ ;β = 0,
namely
82 Hydrodynamic Equilibrium

dp dν
= − (μ + p) = −e2 λ r (κ p − Λ) (μ + p)
dr dr
dp r (κ p − Λ) (μ + p)
→ =− . (7.7)
dr C0 − κ M (r) /π − Λ r2
On the other hand, the substitution of p(r) from (7.6b) into (7.6c) gives rise to
 2  
d2 ν dν dν dλ 1 dν
+ − − = 0, (7.8)
dr2 dr dr dr r dr
dν
which is a first-order equation for N (r) := dr , which can be written in a very
simple form by introducing the functions

ξ(r) := r eλ(r) , Z := N/ξ, (7.9)

namely
   r
d N N 2
+( ) ξ = 0 → dZ −1 = ξ dr → Z −1 = C1 + ξ(r)dr. (7.10)
dr ξ ξ 0

The equation for ν becomes

 r
ξ dr
dν = Zξ dr = r → ν(r) = ln [C1 + ξ(r)dr] + ln C2 /2,
C1 + 0 ξ(r)dr 0
 r 
λ(r)
ν (r) = ln e rdr + C1 + ln C2 /2, (7.11)
0

the const C2 /2 → 1 by scaling the time coordinate.

Substituting this integral in (7.6b) one obtains the pressure
1 Λ
p (r) =  r + . (7.12)
κ eλ(r) 0
eλ(r) rdr + C1 κ

The integration of the equation for λ (r) gives

 
κ r
e λ(r)
= 1/ C0 − (2π μ(r̃) r̃ dr̃) − Λ r2 ,
π 0
κ d
λ (r) = −1/2 ln C0 − M (r) − Λ r2 , M (r) = 2 π μ (r) r, (7.13)
π dr
consequently

 r  r
  κ 2
e ν(r)
= C1 + [ r dr / C0 − (2π μ(r̃) r̃ dr̃) − Λ r ]. (7.14)
0 π 0

Finally, the pressure results in

 r
C0 − πκ 0 (2π μ(r̃) r̃ dr̃) − Λ r2
κ p (r) =  + Λ. (7.15)
r    r 2
C1 + 0 [ r dr / C0 − π 0 (2π μ(r̃) r̃ dr̃) − Λ r ]
κ
 7.2 Stellar Equilibrium in (2 + 1) Dimensions with Λ 83

This relation determines the pressure p through the energy density μ in a func-
tional manner: if p were expressed by a state equation of the form p = p(μ), the
equation (7.15) gives rise to an integral differential equation for the energy as
function of the variable r.
The pressure has to vanish at boundary rb (perimeter) of the circle, p(rb ) = 0,
where the mass function M (r) determines the total mass of the fluid M (rb ) on
the circle. Because the metric signature has to be preserved throughout the whole
spacetime, the positiveness of grr imposes an upper bound on the value of the
total mass, namely
π
M (rb ) ≤ (C0 − Λ rb2 ). (7.16)
κ

7.2.1 Cruz–Zanelli Existence of Hydrostatic Equilibrium for Λ ≤ 0

Cruz and Zanelli (1995) established that: a perfect fluid in hydrostatic equilib-
rium (pressure monotonically decreasing),
dp
μ(r ≤ rb ) > 0, p(r ≤ rb ) > 0, |r≤rb < 0,
dr
is only possible for Λ ≤ 0.
The condition on Λ follows from the energy–momentum conservation equation
(7.7)
d r (κ p − Λ) (μ + p)
p (r) = − ,
dr C0 − κ M (r) /π − Λ r2
which evaluated at the boundary yields
dp rb Λ
|r=rb = μ(rb ),
dr C0 − Λ rb − κ M (rb ) /π
2

which for μ(r) ≥ 0, and M (rb ) fulfilling (7.16) is non-positive only if Λ ≤ 0.

Moreover, since for Λ ≤ 0 the right-hand side of
d
p (r) = − r(κ p − Λ) (p + μ) grr
dr
is always negative, then p (r) is a decreasing function such that
p (r = 0) = pc > p (rb ) = 0.

7.2.2 No Buchdahl’s Inequality in (2 + 1) Hydrostatics

The existence of a Buchdahl-like bound on the mass density is based, in four
and more dimensions, on an inequality arising from the energy conservation law,
the Oppenheimer–Volkoff equation, expressed in terms of the metric structural
functions, see (7.27). In three dimensions the corresponding equation (7.8), to
which (7.27) reduces for d = 3, does not allow for an inequality; it integrates
84 Hydrodynamic Equilibrium

in quadratures, (7.11). Therefore, in three dimensions there is no Buchdahl–like

bound on the mass density.

7.2.3 Static Star with Constant Density μ0 and Λ = −1/l2 ≤ 0

The static star with uniform density μ0 is characterized by mass and pressure
given respectively as:
π
M (r) = π μ0 r2 , m(r) = π μ0 r2 + Λ r2 ,
κ
C0 − π m(r) − C0 − πκ m(rb )
κ
p(r) = μ0 . (7.17)
− C0 − πκ m(r) − Λ κ
μ0 C0 − πκ m(rb )
A star of uniform density in hydrostatic equilibrium (Λ = −1/l2 ) possesses
central mass and pressure of the form

M (0) = 0, m(0) = 0,
√
C0 − C0 − πκ m(rb )
pc = μ0 √ , (7.18)
− C0 + κ l2 μ0 C0 − πκ m(rb )
at the boundary r = rb

M (rb ) = π rb2 μ0 = μ0 S , p(rb ) = 0,

 
while for rb = κCμ00 1 + κ l21 μ2 the pressure becomes infinity, p → ∞, and the
0
mass equates to
κ C0 1 κ C0 1
M= (1 + 2 ), (m = (1 − 2 4 2 )).
π κ l μ0 π κ l μ0
The evaluation of e2ν or the metric component gtt = −e2ν yields
 2
2ν κ μ0 l2 C0 − πκ m(rb ) − C0 − πκ m(r)
e = . (7.19)
κ μ0 l2 − 1

which is different compared with the expression determined by Eq. (23) of Cruz
and Zanelli (1995).
The external solution to which the uniform fluid solution can be matched is
the static anti-de Sitter metric with parameter M0 , known also as the static BTZ
solution
r2 r2
g = −(−M0 + 2 )dt2 + dr2 /(−M0 + 2 ) + r2 dφ2 (7.20)
l l
the continuity at the boundary rb of the metric for the fluid is achieved
rb2 κ r2
e2ν(rb ) = e−2λ(rb ) = C0 +
2
− M = −M0 + 2b
l π l
→ κ = π, M (rb ) = M = M0 − C0 . (7.21)
 7.3 Buchdahl Theorem in d Dimensions 85

The static perfect fluid solution with Λ = −1/l2 exhibits an event horizon at
2
rh = κ μC0 ll2 −1 .

7.3 Buchdahl Theorem in d Dimensions

The static spherical symmetric metric in curvature Schwarzschild coordinates for
a d-dimensional spacetime is given by

g := −e2 ν(r) dt2 + e2 λ(r) dr2 + r2 dΩ2d−2 ,

dΩ2d−2 = dχ22 + sin2 (χ2 )dχ23 + . . . + sin2 (χ2 ) sin2 (χ3 ) . . . sin2 (χd−2 ) dχ2d−1 .
(7.22)

In the presence of a perfect fluid and a cosmological constant Λ, the Einstein

equation are:
1
E β α := Rβ α − δ β α R − κT β α + Λδ β α = 0,
2
T β α = −ρ(r)δ β 0 δ 0 α + p(r) δ β 1 δ 1 α + · · · + p(r) δ β d−1 δ d−1 α . (7.23)

The time component Et t of the Einstein equations yields

 
−2 λ(r) 1 dλ (r) (d − 3) (d − 3) 2 λ(r)
κ ρ (r) + Λ = (d − 2) e − + e , (7.24)
r dr 2r2 2r2
while the radial Einstein equation component Er r gives rise to
 
1 dν (r) (d − 3) (d − 3) 2 λ(r)
κ p (r) − Λ = (d − 2) e−2 λ(r) + − e . (7.25)
r dr 2r2 2r2
The remaining Einstein equations Ed−2 d−2 related to angular components yield
2
 2
2d ν dν dν dλ dν
e (κ p (r) − Λ) = r
2λ
+ r2 + r (d − 3) − r2
dr2 dr dr dr dr
 
d−3 dλ
− e2λ (d − 4) − (d − 4) + 2 r . (7.26)
2 dr
Replacing the expression of p(r) from (7.25) into (7.26) one gets a second-order
equation for ν(r), namely
 2    
d2 ν dν 1 dν dν dλ 1 dλ e2λ 1
+ − − = (d − 3) − + . (7.27)
dr2 dr r dr dr dr r dr r2 r2
This equation can be rewritten as
     
(λ−ν) d 1 −λ deν 2λ r d 1 −2λ 1
re e = − (d − 3) e e + 2 . (7.28)
dr r dr 2 dr r2 r
The right-hand side of (7.28), containing λ and its derivative, can be signi-
ficatively simplified by using the solution of the density equation (7.24) for λ,
namely
86 Hydrodynamic Equilibrium

r2 Ω (r) r−d+3 Λ r2
e−2λ = 1 − 2 + 2C1 −2 , (7.29)
d−2 d−2 (d − 1) (d − 2)
where the function Ω(r) is defined by

1−d
Ω = κr rd−2 ρ (r) dr. (7.30)

Therefore, one may write (7.28) as

   
−ν d 1 −λ deν d−3 λ d (d − 1)
re e = re Ω (r) + C1 . (7.31)
dr r dr d−2 dr rd
To bring (7.31) to the Buchdahl’s form one introduces the variables
r
ζ := eν , y := e−λ , dξ = 2 dr; x = r2 ; (7.32)
y
consequently one gets
   
d2 ζ 1d−3 1 d C1 d − 1 1d−3 d C1 d − 1
= ζ Ω + = ζ Ω + .
dξ 2 2d−2 y dξ 2 r1+d 2d−2 dx 2 x(1+d)/2
(7.33)
To have solutions with regular centers one imposes certain conditions on the
density and on the metric components; therefore, one requires, for d > 4, C1
to be zero: C1 = 0. Thus for any dimension equal to or greater than four, the
leading term in g rr is 1.
The three-dimensional case, d = 3, is singled out, since in this case the leading
term is a constant 1 + C1 = M related to the mass for negative cosmolog-
2
ical constant. Moreover, the second-order equation equates to zero, ddξζ2 = 0.
Consequently, its solution becomes
 r
d2 ζ
= 0 → ζ = A1 ξ + A0 e ν
= A0 + A 1 2reλ dr, (7.34)
dξ 2 0

where it has been taking into account the definitions given in (7.32). Hence, there
is no room for a restriction on the mass via a Buchdahl procedure.

7.3.1 Buchdahl’s Inequalities

Under the conditions of the Lagrange mean value theorem – see, for instance,
Korn and Korn (1961)1 – the integral of dζdξ yields

dζ ¯ m.v. ξ dζ
(ξ − 0) (ξ) = dξ = [ζ(ξ) − ζ(0)]
dξ 0 dξ
dζ ζ(ξ) − ζ(0)
→ (ξm.v. ) = , 0 < ξ¯ < ξ, (7.35)
dξ ξ
1 Lagrange mean value theorem. If f (x) is continuous in [a, b]:
i) and differentiable on (a, b), then in (a, b) there exists a number X such that
f (b) − f (a) = f  (X)(b − a),

ii) then on (a, b) there exists a number X such that ab f (x)dx = f (X)(b − a).
 7.3 Buchdahl Theorem in d Dimensions 87

where ξm.v. is the value of ξ where the mean value Lagrange theorem holds; let
2
us call it the “mean value” ξ coordinate. Moreover, since ddξζ2 ≤ 0 → dζ
dξ (ξ0 ) ≥
dζ
dξ (ξ), ξ0 < ξ, and considering that ζ(ξ) is a positive (increasing) function, then

dζ ¯ ζ(ξ) − ζ(0)
ζ(ξ0 ) < ζ(ξ), ξ0 < ξ → (ξ > ξm.v. ) ≤ , 0 < ξm.v. < ξ¯ < ξ.
dξ ξ
(7.36)

Because of the positiveness of ξ and ζ(ξ) and its increasing property implying
dξ ≥ 0, one can write
dζ

dζ ¯ ζ(ξ)
(ξ) ≤ , 0 < ξm.v. < ξ¯ < ξ. (7.37)
dξ ξ
In various publications, the inequality referred to as the one due to the mean
value theorem is
dζ ζ(ξ)
(ξ) ≤ ; (7.38)
dξ ξ
see, for instance, Straumann (1984), paragraph 6.6.3, Ponce de Leon and Cruz
(2000), Mak and Harko (2000), and Zarro (2009). This last inequality is treated
as a function of ξ in the whole range of its variability, without any comment on
the restrictions imposed by the mean value theorem. The writing of (7.38) in that
way is an abuse of typing, although it would be considered correct if one should
have in mind that ξ stands for the ξ¯ at the left-hand side of this inequality, which
should be very doubtful. Whatever the case, it is essentially misleading.
In terms of the original variables, the above inequality reads
1 −λ(r̄) ν(r̄) dν 1
e e (r̄) ≤  r λ eν(r) . (7.39)
2r̄ dr 2 0 re dr

Isolating from (7.25) the derivative dν

dr and substituting in the middle of the
resulting equation the expression for e−2λ from (7.29), one gets
 
1 d κ d−3 2
ν=e 2λ
p (r) + Ω (r) − Λ , (7.40)
r dr d−2 d−2 (d − 1) (d − 2)
which substituted into (7.39) yields the inequality
  r
1 λ(r̄)+ν(r̄) κ p (r̄) d−3 2Λ
e + Ω (r̄) − 2 reλ dr ≤ eν(r) .
2 d−2 d−2 (d − 1) (d − 2) 0
(7.41)

The actual problem of handling the inequality (7.41) is hidden in the evaluation
r
of the integral ξ(r) = 0 2 reλ dr. To avoid dealing with specific models, i.e., to
give particular densities ρ(r), one can follow the Buchdahl procedure consisting
in replacing Ω(r) by Ω(R) and use the resulting inequality and the arising inte-
gral expressed in terms of elementary functions, namely, radicals. Since ρ(r) is
88 Hydrodynamic Equilibrium

assumed to be a monotonically decreasing function, then dρ

dr ≤ 0, therefore Ω is
a decreasing function too; consequently
dΩ
≤ 0 → Ω(r) > Ω(R), eλ(Ω(R),r) ≤ eλ(Ω(r),r) ,
dr
 r  r  r̄
e dr ≥
λ(r) 2
e dr ≥
λ(Ω(R),r) 2
eλ(Ω(R),r) dr2 , 0 < r̄ ≤ r ≤ R.
0 0 0
(7.42)

Thus, the inequality (7.41) can be rewritten as

   r̄
1 λ(r̄) κ p(r̄) d − 3 2
e + Ω(r̄) − Λ 2 reλ(Ω(R),r) dr ≤ 1.
2 d−2 d−2 (d − 1) (d − 2) 0
(7.43)

The integration of ξ(r) for Ω(R) is straightforward:

 r
dr2 d − 2 1/2
ξ(Ω(R), r) =  =− Δ (r)|r0 ,
0 1 − 2 d−2 ω(R)
r 2 ω(R)

r2 Λ
Δ := 1 − 2 ω(R), ω(R) := Ω (R) + ,
d−2 (d − 1)
 R
1 Λ M (R) Λ
ω(R) = d−1 κ rd−2 ρ (r) dr + =: d−1 + . (7.44)
R 0 d−1 R d−1
At the boundary r = R the fluid metric has to be matched to the external vacuum
with Λ solution known as the d-dimensional Schwarzschild–Kottler–Tangherlini
(SKT) solution, which is given by the metric (7.22) with structural functions
 
r2 2m Λ r2
e−2λ(r) = e2ν(r) = 1 − 2 + =1−2 ωskt (r),
d−2 r d−1 d−1 d−2
2m Λ
ωskt (r) := d−1 + . (7.45)
r d−1
At the hypersphere boundary r = R one has

−λ(R) ν(R) −λ(R) ν(R) R2
eskt = eskt = epf = epf = 1−2 ω(R) = Δ1/2 (R),
d−2
ωskt (R) = ωpf (R) = ω(R). (7.46)

Thus, at r = R, taking into account that p(R) = 0 and that

d − 2 1/2 d−2
ξ(Ω(R), R) = − Δ (r)|R
0 = (1 − Δ1/2 (R))
ω(R) ω(R)
the inequality (7.43) becomes
 
1 d−2 d−3 1
ω (R) − Λ 1 − Δ1/2 (R) ≤ Δ1/2 (R). (7.47)
2 ω(R) d − 2 d−2
 7.3 Buchdahl Theorem in d Dimensions 89

Isolating Δ1/2 (R) one arrives at

(d − 3)ω (R) − Λ
Δ1/2 (R) ≥
(d − 1)ω (R) − Λ
r2 [(d − 3)ω (R) − Λ]2
→ Δ(R) = 1 − 2 ω(R) ≥ . (7.48)
d−2 [(d − 1)ω (R) − Λ]2
Hence, for vanishing Λ one obtains the well-known inequality Δ(R, Λ = 0) ≥
(d−3)2
(d−1)2 , which, for dimension four, d = 4, yields Buchdahl’s inequality, Δ ≥ 9 .
1

The algebraic equation for ω is a cubic equation with one of the roots being
zero, ω = 0; the remaining quadratic equation amounts to
 
2
(d − 2) + Λ R 2 (d − 1) Λ 2 (d − 2) + Λ R 2
ω2 − 2 2 ω + 2 = 0, (7.49)
R 2 (d − 1) R 2 (d − 1)
with roots ω± where ω+ ≥ ω(R) is given by

4
2
(d − 2) + Λ R (d − 1)
2 (d − 2) − 2 Λ R 2 (d − 1) (d − 2)
ω(R) ≤ 2 + 2 .
R 2 (d − 1) R 2 (d − 1)
(7.50)
Taking into account the definition of ω(R), (7.44), through Ω and M , one finally
obtains the main inequality:

4
M (R) (d − 2)
2 (d − 2) − 2 Λ R 2 (d − 1) (d − 2) 2mskt
≤ 2 + 2 ≥ d−3 . (7.51)
R d−3
(d − 1) (d − 1) R
For vanishing cosmological constant one gets
2
M (R) (d − 2) 2
≤2 2 ≥ R d−3 mskt . (7.52)
R d−3 (d − 1)
Moreover, for the fourth dimension, d = 4, this inequality reduces just to the
well-known expression R1 M (R) ≤ 89 ≥ R2 mskt .

7.3.2 Constant Density Solution

In this subsection the static perfect fluid regular solution with constant den-
sity and a cosmological constant is derived and its matching to the external
d-dimensional SKT solution is accomplished.
The family of static perfect fluid with constant density and Λ is unique: if the
integral mass density Ω (7.30) is constant Ω0 , then ρ = constant too;

d
rd−1 Ω0 = κ rd−2 ρ (r) dr → Ω0 rd−1 = κrd−2 ρ (r)
dr
1
→ ρ (r) = Ω0 = ρ0 . (7.53)
(d − 1)κ
90 Hydrodynamic Equilibrium

The integral of e−2λ(r) , for metrics which are regular at the origin of the
coordinates, yields
κ ρ0 r2 Λ r2
e−2λ(r) = 1 − 2 −2 . (7.54)
(d − 1) (d − 2) (d − 1) (d − 2)
As far as the integral for ν(r) from (7.27) is concerned, that equation becomes
   r
d 1 −λ deν
e = 0 → eν = C0 + C1 reλ dr (7.55)
dr r dr 0

which, substituting eλ and integrating, can be given as

C1 (d − 1) (d − 2) −λ(r)
eν(r) = C0 − e . (7.56)
2 κ ρ0 + Λ
By substituting eλ(r) and eν(r) and its derivatives into (7.25) one obtains the
expression of the pressure p(r),
2 C1 (d − 2) (κ ρ0 + Λ) e−λ(r)
p(r) =
κ 2 C0 (κ ρ0 + Λ) − C1 (d − 1) (d − 2) e−λ(r)
2Λ d−3
+ − ρ0 . (7.57)
κ (d − 1) d−1
At the boundary rb = R of the fluid distribution, the pressure has to vanish,
p(R) = 0; this condition determines one of the constants, namely
C0 (κ ρ0 + Λ) [κρ0 ( d − 3) − 2 Λ] λ(R)
C1 = 2 2 e , (7.58)
κ ρ0 (d − 2) (d − 1)
therefore the pressure vanishing at the boundary can be written as
 
(κ dρ0 − 3 κ ρ0 − 2 Λ) eλ(R) − eλ(r) e−λ(r)
p = ρ0 . (7.59)
(d − 1) κ ρ0 − (κ dρ0 − 3 κ ρ0 − 2 Λ) eλ(R)−λ(r)

Matching to the External SKT Solution

The external solution is the d–dimensional Schwarzschild–Kottler–Tangherlini
solution, which is given by the metric (7.22) with structural functions
Λ r2 m r−d+3
e−2λ(r) = e2ν(r) = 1 − 2 −4 . (7.60)
(d − 1) (d − 2) d−2
From the continuity of the metric components across the fluid edge one gets:
κ ρ0 Rd−1
e−2λf (R) = e−2λskt (R) → m = , (7.61)
2 d−1
and
C1 (d − 1) (d − 2) −λ(R)
e2νf (R) = e2νskt (R) → eν(R) = C0 − e = e−λ(R)
2 κ ρ0 + Λ
 
C1 (d − 1) (d − 2)
→ C0 = 1 + e−λ(R) . (7.62)
2 κ ρ0 + Λ
 7.3 Buchdahl Theorem in d Dimensions 91

Taking into account the expression (7.58) for C1 arising from the vanishing of
the pressure at the border rb = R, one arrives at
κ ρ0 (d − 1) −λ(R)
C0 = e ,
2(κ ρ0 + Λ)
κ ρ0 (d − 3) − 2Λ −λ(R)
C1 = e . (7.63)
(d − 2)(d − 1)
Finally, the metric components of a regular static solution in d-dimensions for a
perfect fluid with constant density ρ0 in the presence of a cosmological constant
Λ allowing for a matching on the hypersphere R with the external Schwarzschild–
Kottler–Tangherlini solution can be given as
κ ρ0 r2 Λ r2
1/grr = e−2λ(r) = 1 − 2 −2 ,
(d − 1) (d − 2) (d − 1) (d − 2)
 2
−2λ(R) 1 κ ρ0 (d − 3) − 2Λ −λ(R) −λ(r)
−gtt = e 2ν(r)
=e 1+ e −e ,
2 κ ρ0 + Λ
 λ(R) 
[κ ρ0 (d − 3) − 2 Λ] e − eλ(r)
p = ρ0 e−λ(r) , (7.64)
(d − 1) κ ρ0 − [κ ρ0 (d − 3) − 2 Λ] eλ(R)−λ(r)
where the behavior at the frontier
 R becomes apparent. The event horizon, if
there is any, is located at rh = (d−1)(d−2)
2(κ ρ0 +Λ) .
 8
Stationary Circularly Symmetric Perfect Fluids
with Λ

In the theory of black holes, an important place is assigned to the problem of

their interior: understanding it as the matter that generates the external grav-
itational field. Commonly, the matter is modeled through a perfect fluid. It is
worth recalling that in (3 + 1) gravity the interior solution to the Schwarzschild
black hole is modeled by an interior Schwarzschild perfect fluid solution with
constant energy density, but the question still remains open as to what is the
interior solution for the rotating Kerr black hole? In (2 + 1) gravity one may ask
the same question with respect to the rotating BTZ black hole. As we shall see
in this chapter, there is an interior solution to the BTZ black hole, modeled by
a perfect fluid with constant energy density. Moreover, we succeeded in deriving
perfect fluid solutions, such that their fluid velocity possesses differential rota-
tion. On the other hand, rigidly rotating exact perfect fluid solutions derived
by Rooman and Spindel (1998) and Lubo et al. (1999) are also reported. Those
papers deal with stationary circularly symmetric gravitational sources of the per-
fect fluid type, with a cosmological constant with focus on restrictions on the
physical parameters of the solutions due to the matching conditions between the
interior and exterior geometries. In particular, it is established there that finite
sources and absence of closed timelike curves privilege negative values of the
cosmological constant. Moreover, for stationary configurations, the field equa-
tions for constant energy densities have been explicitly solved; if, additionally,
the pressure vanishes, interior Gödel-like stars arise.
It is worth pointing out that the literature on stationary perfect fluid solutions
of (2+1) gravity is rather scarce; among it one may cite Rooman and Spin-
del (1998); Lubo et al. (1999); Garcı́a (2004); Cataldo (2004); Gürses (1994);
Obukhov (2003). From Section 8.1 to Section 8.2.5 I follow the presentation
given in Garcı́a (2004), while for the remaining two sections I follow Rooman
and Spindel (1998), and Lubo et al. (1999).
 8.1 Stationary Differentially Rotating Perfect Fluids 93

8.1 Stationary Differentially Rotating Perfect Fluids

The main goal of this section is the derivation of interior solutions modeled
through a differentially rotating perfect fluid in the presence of a cosmological
constant for the stationary circularly symmetric (2+1) metric
dr2 2
ds2 = −N (r)2 dt2 + + r2 (dφ + W (r)dt) . (8.1)
F (r)2
It will be demonstrated that this class of stationary circularly symmetric (2+1)
spacetimes coupled to a perfect fluid are determined completely by an arbitrary
function N (r); for a given function N (r) one determines the remaining function
F (r), the energy density, and the pressure. Various families of solutions will be
explicitly given.
The Einstein equations to be considered are
R
Rαβ − g + gαβ Λ = κ Tαβ ,
2 αβ
where the perfect fluid energy–momentum tensor is given by

Tαβ = (p + ρ) uα uβ + p gαβ ,
1
uα = (δt α + Ω δφ α ) . (8.2)
N − r (W + Ω)2
2 2

The function Ω(r) describes the differential rotation property of the fluid; if
dΩ(r) = 0, one is dealing with a differentially rotating perfect fluid, while if
Ω(r) = const., the fluid is rigidly rotating. κ stands for the (2+1) gravitational
constant, ρ(r) and p(r) denote respectively the perfect fluid energy density and
isotropic pressure.
The independent Einstein equations for the metric (8.1), coupled to a perfect
fluid and a cosmological constant Λ, are:
the pressure
F2
κp (r) − Λ = r3 (Ẇ )2 + 4 N Ṅ , (8.3a)
4r N 2
the energy density
    2
2
W 1 W Ṅ
ρ (r) + Λ = r − F Ḟ − rF 2
N r N N
    
2 2
r2 F F Ḟ Ṅ
− (Ẇ )2 + r − (Ω 2 + 2 W Ω), (8.3b)
4 N N F N

the equation for the metric function W

  
3 Ḟ Ṅ 2 2
EQW d := Ẅ + Ẇ + − Ẇ + W + Ω = 0, (8.3c)
r F N r r
94 Stationary Circularly Symmetric Perfect Fluids with Λ

and the equation for the metric function N

  2 
Ḟ W 1 Ḟ W 2 (Ẇ )2
EQN d = N̈ + −r − Ṅ + r − r2
F N r F N N

r Ḟ Ṅ
+ − (Ω2 + 2 W Ω) = 0. (8.3d)
N F N

Incidentally, replacing (Ω2 + 2 W Ω) from (8.3d) in (8.3b), one gets ρ in terms of

the metric functions only,
   
N̈ Ṅ 3 r2 2 Ṅ 1
κρ (r) + Λ = −F 2
− − Ẇ − F Ḟ + . (8.3e)
N rN 4 N2 N r

Therefore, we have established the following result:

The class of stationary circularly symmetric (2+1) solutions coupled to a differ-
entially rotating perfect fluid, dΩ/dr = 0, and a cosmological constant is defined
by: the metric (8.1), the isotropic pressure (8.3a), the fluid density (8.3b), or
equivalently (8.3e), and the structural functions: W (r), and N (r) fulfilling the
differential equations (8.3c), and (8.3d), respectively, for the four unknown func-
tions N , F , W , and Ω. A constraint on these functions is provided by a state
equation – a relation between the pressure p and the energy density ρ, thus still
there is a freedom in one function; one may think of equation (8.3c) as defining
the differential rotation function Ω and replace it into the equation (8.3d) for
N ; in this manner the constraints occur in terms of the metric functions. Some
mistakes contained in the expression for Ω reported in Garcı́a (2004) are here
corrected.

8.2 Garcia Stationary Rigidly Rotating Perfect Fluids

The main goal of this section is the derivation of interior solutions modeled
through a rigidly rotating perfect fluid in the presence of a cosmological con-
stant for the stationary circularly symmetric (2+1) metric (8.1). It will be
demonstrated that this class of stationary circularly symmetric (2+1) spacetimes
coupled to a perfect fluid is determined completely by an arbitrary function N (r);
for a given function N (r) one determines the remaining function F (r), the energy
density, and the pressure. Various families of solutions will be explicitly given.
The perfect fluid energy–momentum tensor associated to the Einstein equations
is given by
1
Tαβ = (p + ρ) uα uβ + p gαβ , uα = √ (δt α ) . (8.4)
N 2 − r2 W 2
Therefore, the explicit expressions of the Einstein equations are derived from
the previous ones (8.3) by setting the differential rotation equal to zero, Ω = 0,
hence the pressure
 8.2 Garcia Stationary Rigidly Rotating Perfect Fluids 95

F2
κp (r) − Λ = r3 (Ẇ )2 + 4 N Ṅ , (8.5a)
4r N 2
the energy density
  2  2  2
W 1 W Ṅ r2 F
ρ (r) + Λ = r − F Ḟ − rF 2 − Ẇ 2 , (8.5b)
N r N N 4 N

the equation for the metric function W

  
3 Ḟ Ṅ 2
EQW := Ẅ + Ẇ + − Ẇ + W = 0, (8.5c)
r F N r

and the equation for the metric function N

  2 
Ḟ W 1 Ḟ W 2 Ẇ 2
EQN := N̈ + −r − Ṅ + r − r2 . (8.5d)
F N r F N N

Another expression for ρ is given by (8.3e). Certainly, the energy–momentum con-

servation equations are fulfilled modulo the equations arising from the Einstein
equations. Therefore, a state equation ρ = ρ(p) gives rise to a third equation
for the three unknown structural functions F (r), W (r), and N (r). A second
possibility of looking at the Einstein–perfect fluid set of equations consists in a
particular choice of one of the structural functions, say for instance W , solving
for a second structural function, N or F , and substituting this last relation into
the energy and the pressure; one is then left with two functions expressed in
terms of one unknown function and its derivatives. An explicit example of this
approach is given in the next subsection.
The Cotton tensor for this class of metrics is of the general form
⎡ ⎤
c11 0 c13
⎢ ⎥
(C α β ) = ⎢
⎣ 0 c22 0 ⎥,
⎦ (8.6)
c31 0 −(c11 + c22 )

in the standard correspondence {t, r, φ} → {1, 2, 3}. The eigenvalues of this

tensor are

λ1 = c22 → S1,

1 1
λ2,3 = − c22 ± (c22 + 2 c11 )2 + 4 c31 c13 , (8.7)
2 2
Depending on the radical the roots λ2,3 could be complex, hence the type of
the Cotton tensor would be Type I: {S, Z, Z̄}. This kind of metric allows for
conformally flat solutions too.
96 Stationary Circularly Symmetric Perfect Fluids with Λ

8.2.1 Rigidly Rotating Perfect Fluid Solution with W (r) = J/(2r 2 )

Incidentally, the W (r) = W1 /r2 singles out a unique class of rigidly rotating
perfect fluid solutions. In fact, the equation EQW (8.5c) allows for a unique
solution for W independent of the functions F and N ; requiring

d2 W 3 dW W1
2
+ = 0, → W = 2 , W0 = 0,
dr r dr r
dW 2 W1
+ W = 0, → W = 2 , W1 = −J/2, BTZ-like rotation. (8.8)
dr r r
Hence, the stationary circularly symmetric (2+1) metric to be studied in this sub-
section for the derivation of interior solutions modeled through a rigidly rotating
perfect fluid with a unit 4 velocity
r
uα = 2 √ δt α ,
4 r2 N2 − J2
in the presence of a cosmological constant Λ amounts to
 2
dr2 J
ds = −N (r) dt +
2 2 2
+ r dφ − 2 dt .
2
(8.9)
F (r)2 2r

This class of stationary (2+1) spacetimes coupled to a perfect fluid are deter-
mined completely by an arbitrary function N (r); for a given function N (r) one
determines the remaining function F (r), the energy density, and the pressure.
The equation EQN (8.5d), once substituted W = −J/(2 r2 ), can be written as
 
d F
EQN : 4 r3 Ṅ N + J 2 = 0, (8.10)
dr N r4

therefore
N r4 1 rF0 1 J2
F (r) = F0 → Ṅ = − , (8.11)
4 r3 N Ṅ + J 2 4 F 4 r3 N
where dots denote derivatives with respect to the radial coordinate r. Substitut-
ing W = −J/(2r2 ) and using Ṅ from (8.11) into the expression (8.5a) for the
pressure, one obtains
F0 F
κ p(r) − Λ = . (8.12)
4 N
The expression of the energy density for W = −J/(2r2 ) from (8.5b) yields

2  
1 J r Ṅ + N F2 1 J 2 − 4 r2 N 2
ρ (r) κ + Λ = − + F Ḟ , (8.13)
4 r4 N 3 4 N 2 r3
which, modulo the function F (r) from (8.11), can be written entirely in terms
of N and its derivatives as
 8.2 Garcia Stationary Rigidly Rotating Perfect Fluids 97

F02 r4 
κρ (r) + Λ = − 3 4r4 N N̈ (J 2 − 4r2 N 2 )
4 4r3 N Ṅ + J2

+ 4r3 Ṅ J 2 r Ṅ + 4r2 N 3 + J 2 N + 16 J 2 r2 N 2 − 3J 4 . (8.14)

Consequently, the class of rigidly rotating fluid stationary circularly symmetric

solutions for the metric (8.9) are determined through an arbitrary function N (r);
for any given function N (r) one determines the metric function F (r) from (8.11),
the isotropic pressure (8.12), and the fluid density (8.14).
Various families of solutions will be explicitly given in the forthcoming subsec-
tions. The particular interior Garcı́a (2004) solution with constant energy density
matchable to the exterior BTZ black hole will also be derived.

8.2.2 Garcia Interior Solution with Constant Energy Density

In this section the incompressible branch of solutions is studied in detail. The
equation (8.14) for ρ = ρ0 to determine N (r) amounts to
3 F2 
(κ ρ0 + Λ) 4 r3 N Ṅ + J 2 = 0 r4 4r4 N N̈ (4r2 N 2 − J 2 )
4 
−4J 2 r3 Ṅ (rṄ + N ) − 16r2 N 2 (J 2 + r3 N Ṅ ) + 3J 4 . (8.15)

We search N (r) by means of an auxiliary function Q(r) as

J2
N (r)2 = + Q(r), (8.16)
4r2
which, substituted in (8.15), yields a simple nonlinear equation
 3  2
d2 Q dQ dQ dQ
2F02 rQ 2 − 2F02 Q − 8(κ ρ0 + Λ) − F02 r = 0.
dr dr dr dr
(8.17)
Introducing a new independent variable x = r2 , the above equation can be
rewritten as
 3  2
d2 Q κ ρ0 + Λ dQ 1 dQ
Q 2 −8 − = 0, (8.18)
dx F02 dx 2 dx
which can be brought to the form

d( dQ d( 12 + 8 κ ρF0 2+Λ dQ
dx )
dx ) dQ
dQ
− 0
= , (8.19)
dx
1
2 + 8 κ ρF0 2+Λ dQ
dx
2Q
0

and allows the first integral


| dQ
dx | |Q|
= , (8.20)
| 12 + 8 κ ρF0 2+Λ dQ
dx |
|Q0 |
0
98 Stationary Circularly Symmetric Perfect Fluids with Λ

where vertical bars stand for absolute values. The above relation can be put into
the differential form
    2
κ ρ0 + Λ κ ρ0 + Λ |Q|
d B0 − 4 x =d 1−8 ; (8.21)
F02 Q0 F02 |Q0 |

hence, integrating one obtains

 2
F02 2
Q(r) = Q Q0 (1 − K(r)) ,
8(κ ρ0 + Λ)

κ ρ0 + Λ 2
K(r) := B0 − 4Q r , (8.22)
F02 Q0

where Q = ± and  = ± assume their signs ± independently; of course one could

choice Q Q0 as a single constant Q̃; nevertheless, as we shall see in the forth-
coming treatment, it is more convenient to consider Q0 as a positive constant,
2
F2
proportional to q02 := Q0 8(κ ρ00+Λ) , and the signs switch to .
Summarizing, with all generality, the structural functions can be given as

J2
W (r) = − , (8.23)
2r2
1 J2 2
N (r)2 = + N q02 [1 − K(r)] , (8.24)
4 r2
κ ρ0 + Λ K(r)
F (r) = 4 N (r), (8.25)
F0 1 − K(r)

1 F02
K(r) := B0 − r2 , (8.26)
16 (κ ρ0 + Λ)q02

2
F02
where q02 := Q0 8(κ ρ0 +Λ) ,  = ±, and N = ±.

The pressure amounts to

 κ ρ0 K(r) + Λ
κp(r) = . (8.27)
1 − K(r)

From this expression one establishes that at certain value r = rzp , where the
subscript zp denotes zero pressure, the pressure may vanish for  = 1, Λ = −1/l2 ,
and  = −1, Λ = 1/l2 ; one has to √ keep in mind these sign conditions when
extracting square roots; for instance, Λ2 = (−Λ)2 = −Λ. The vanishing of
the numerator of p(r), K(rzp ) = − Λ/(κ ρ0 ), yields

F02 2 Λ2
B0 = N 2 rzp + 2 2. (8.28)
16(κ ρ0 + Λ)q0 κ ρ0
 8.2 Garcia Stationary Rigidly Rotating Perfect Fluids 99

Therefore, for vanishing pressure at the circle rzp , the structural functions N (r)
and F (r) amount to
J2 2
N (r)2 = + N q02 [1 − K(r, rzp )] , (8.29)
4r2

κ ρ0 + Λ K(r, rzp )
F (r) = 4 N (r), (8.30)
F0 1 − K(r, rzp )

Λ2 F02
K(r, rzp ) = +  (r2 − r2 ). (8.31)
κ2 ρ20 N
16(κ ρ0 + Λ)q02 zp

Moreover, at the radius rzp , the functions N (r) and F (r) become
1 J2 2 (κ ρ0 + Λ)
2
N (rzp )2 = +  q 0 , (8.32)
2
4 rzp N
κ2 ρ20

4Λ
F (rzp ) = − N (rzp ). (8.33)
F0
On the other hand, the vacuum solution for the metric (8.9) is determined by
the structural functions
1 J2
N (r)2 = F (r)2 = − Λ r2 − M, (8.34)
4 r2
which in the case of negative cosmological constant Λ = −1/l2 becomes the
well-known BTZ black hole solution given by the metric (8.9) with
1 J2 r2
N (r)2 = F (r)2 = 2
+ 2 − M. (8.35)
4r l
Comparing (8.34) with (8.32) at rzp , one arrives at
κ2 ρ20
q02 = N 2
(−Λ rzp − M ), (8.36)
(κ ρ0 + Λ)2
and consequently we have the following sub-branches of solutions:
A)N = 1,
2
κ2 ρ20 rzp
A1: Λ = −1/l2 , q02 = (κ ρ0 −1/l2 )2 ( l2 − M ), rzp
2
/l2 − M > 0,
2
κ2 ρ20 rzp
A2: Λ = 1/l2 , q02 = − (κ ρ0 +1/l 2 )2 ( l 2
2
+ M ), rzp /l2 + M < 0,
B) N = −1,
2
κ2 ρ2 rzp
B1:Λ = −1/l2 ,q02 = − (κ ρ0 −1/l
0
2 )2 ( l2 − M ), rzp
2
/l2 − M < 0,
2
κ2 ρ20 rzp
B2: Λ = 1/l2 , q02 = (κ ρ0 +1/l2 )2 ( l2
2
+ M ), rzp /l2 + M > 0.
Substituting F0 = −4 Λ, which in turn yields F (rzp ) = N (rzp ), and q0 from
(8.36) into (8.29–8.31), one obtains the expressions of the structural functions
100 Stationary Circularly Symmetric Perfect Fluids with Λ

satisfying matching conditions with external gravitational fields in the presence

of a cosmological constant of both signs:
2
1 J2 Λ rzp +M 2 2 2
N (r)2 = − κ ρ0 [1 − K(r, rzp )] , (8.37)
4 r2 (κ ρ0 + Λ)2

κ ρ0 + Λ K(r, rzp )
F (r) = − N (r), (8.38)
Λ 1 − K(r, rzp )

Λ κ ρ0 + Λ 2
K(r, rzp ) = − 1− 2 +M
(rzp − r2 ), (8.39)
κ ρ0 Λ rzp

accompanied by the pressure


1 − Λκrρ20 +Λ
1− 2 − r2 )
(rzp
zp +M
p(r) = ρ0 Λ  . (8.40)
κ ρ0 + Λ 1 − Λκrρ20 +Λ
+M (rzp − r )
2 2
zp

8.2.3 Interior Perfect Fluid Solution to the BTZ Black Hole

One of the most interesting interior solutions of the studied class, which matches
with the exterior BTZ black hole, arises for Λ = −1/l2 , and is determined by
the metric (8.9) with structural functions
 2
1 J2 M − rzp2
/l2 1
N (r)2 = − κ ρ0 − K(r, rzp ) , (8.41)
4 r2 (κ ρ0 − 1/l2 )2 l2

K(r, rzp )
F (r) = (κ ρ0 − 1/l2 ) N (r), (8.42)
κ ρ0 − l12 K(r, rzp )

κ ρ0 − 1/l2 2
K(r, rzp ) := 1− (r − r2 ), (8.43)
M − rzp
2 /l2 zp

This solution is characterized by the pressure


κ ρ0 −1/l2
ρ0 1 − 1− M −rzp2 /l2 (rzp − r )
2 2
p(r) = − 2  . (8.44)
l κ ρ − 1 1 − κ ρ0 −1/l2 (r2 − r2 )
0 l2 M −rzp
2 /l2 zp

and a constant energy density ρ0 .

8.2.4 Alternative Parametrization

By introducing new constants c1 , and c2 through
c1 c2
q0 = , B0 = 1 − 16 2 , F0 = − 4Λ (8.45)
4 c1
 8.2 Garcia Stationary Rigidly Rotating Perfect Fluids 101

one brings the structural functions N (r) and F (r), (8.24) and (8.25), to the form
 2
2 1 J2 N Λ
N (r) = + c1 + 4 K(r) ,
4 r2 16 κ2 ρ20 − Λ2

K(r)
F (r) = 4(κ ρ0 + Λ) N (r),
c1 κ2 ρ20 − Λ2 + 4ΛK(r)

κ2 ρ20 − Λ2 16Λ2 2
K(r) := − c21 − 16c2 − N r , (8.46)
4Λ κ ρ0 + Λ
accompanied by the isotropic pressure,
c1 κ2 ρ20 − Λ2 − 4κρ0 K(r)
κ p(r) = Λ ,
c1 κ2 ρ20 − Λ2 + 4ΛK(r)
(8.47)
where  = ±, and N = ± independently. As far as the behavior of the pressure
is concerned, one may establish that for r running from the value rpress↑∞ =
−N c2 (κ ρ0 + Λ)/Λ2 , at which point the denominator of the pressure p (8.47)
becomes equal to zero, up to the value rzp at which the numerator of the pressure
vanishes, K(rzp ) = c1 κ2 ρ20 − Λ2 /(4κρ0 ), the pressure is a positive function of
the radial coordinate, which decreases from a very large value at rpress↑∞ to zero
at the boundary circle rzp , at which
1 J2 c21 (κ ρ0 + Λ)2
N (rzp )2 = F (rzp )2 = +  ,
2
4 rzp N
16 κ2 ρ20

κ ρ0 Λ2
c1 = 4 c2 + N r2 .
κ ρ0 − Λ
2 2 2 κ ρ0 + Λ zp
Taking into account the existence of a zero-pressure circle r = rzp , the structural
functions from (8.46) can be given as
1 J2  2
N (r)2 = + 2 2N 2 [κ ρ0 K(rzp ) + Λ K(r)] ,
4 r2 κ ρ0 − Λ
K(r)
F (r) =  (κ ρ0 + Λ) N (r),
κ ρ0 K(rzp ) + ΛK(r)

N Λ2 rzp
2
K(r) := c2 + + (κ ρ0 − Λ) (rzp
2 − r 2 ), (8.48)
κ ρ0 + Λ
together with
K(r) − K(rzp )
p(r) = −ρ0 Λ , (8.49)
κ ρ0 K(rzp ) + Λ K(r)
where K(rzp ) = K(r = rzp ), c1 = 4 κ ρ0 K(rzp )/(κ2 ρ20 − Λ2 ), meaningful for
positive and negative values of the cosmological constant Λ and the parameter M .
102 Stationary Circularly Symmetric Perfect Fluids with Λ

Notice that the vacuum solution arises as a limiting case of the above interior
metric functions, (8.49), for ρ0 = 0 accompanied with the identification c2 →
−N M .
Comparing (8.34) with the expression (8.49), one establishes that the ful-
fillment of the matching conditions, Next (rzp ) = Nfluid (rzp ), together with
Fext (rzp ) = Ffluid (rzp ), is guaranteed by
κ ρ0 − Λ
K(rzp )2 = −N 2
(Λ rzp + M ).
κ ρ0 + Λ
Substituting K(rzp ) into K(r) from (8.48), one has

κ ρ0 − Λ 2
K(r) = K(rzp ) 1 − 2 +M
(rzp − r2 ), (8.50)
Λ rzp

and hence the structural functions assume the form given in (8.37)–(8.39).

8.2.5 Barotropic Rotating Perfect Fluids Without Λ

The equations (8.13) and (8.14), for the linear state equation p = (γ−1)ρ without
Λ, gives rise to a single differential equation
 
4r4 N 4r2 N 2 − J 2 N̈ + 4r4 4 (1 − γ) r2 N 2 − J 2 (Ṅ )2
−4r3 N 4r2 N 2 − (1 − 2γ)J 2 Ṅ + J 2 (4 − γ)J 2 − 16 r2 N 2 = 0
(8.51)

with nontrivial solution

J2 1  2
N (r)2 = 2
+ c2 r2 − c1 2−γ . (8.52)
4r 4
Substituting the above N (r) into (8.11), one determines of the function F (r)
2−γ  − γ
F (r) = F0 N (r) c2 r2 − c1 2−γ , (8.53)
2 c2
and finally the explicit expressions of the pressure and the energy density are
achieved:
2−γ 2  − γ 1
κ p = F02 r c2 r2 − c1 2−γ , ρ = p. (8.54)
8 c2 γ−1
It is clear from the energy conditions that 0 < γ < 2. These perfect fluid solutions
can be thought of as cosmological solutions which extend to infinity. At rzp =
c1 /c2 both the pressure and F (r) vanish.

8.3 Lubo–Rooman–Spindel Rotating Perfect Fluids

In this section the field equations for a stationary circularly symmetric metric in
the presence of a perfect fluid as gravitational source are established; see Lubo
 8.3 Lubo–Rooman–Spindel Rotating Perfect Fluids 103

et al. (1999), assuming (anti-)de Sitter exterior geometries. The metric in coor-
dinates {t, r, φ} with structural functions ρ(r), Z(r), B(r), and Y (r) is given as
2 2 2
ds2 = −(ρ(r)dt + Z(r)dφ)2 + B(r)2 dr2 + Y (r)2 dφ2 = −θ0 + θ1 + θ2 ,
θ0 = ρ(r)dt + Z(r)dφ, θ1 = B(r)dr, θ2 = Y (r)dφ, (8.55)
where θa , a = 0, 1, 2 is the orthonormal triad basis. Comparing with the corre-
sponding metric of Lubo et al. (1999), the function T (r) has been replaced by
ρ(r) here; the reason for proceeding in this way is due to the convenience of a
variable change to the spatial radial coordinate ρ(r). For the fluid energy density
here is reserved μ(r).
The Einstein tensor triad components are Ga b = Ra b − Rδ a b /2 = κT a b − δ a b Λ
are
      2
1 Y 1 ρ Z
0
G 0= − , (8.56a)
BY B 4 BY Y
   2
1 ρ Y  1 ρ Z
G 1 = 2 + , (8.56b)
B ρY 4 BY Y
     2
2 1 ρ 1 ρ Z
G 2 = + , (8.56c)
Bρ B 4 BY Y
   
1 1 ρ3 Z
G 2
0 = −G 0
2 = . (8.56d)
2 Bρ2 BY ρ
∂
The fluid one-form is chosen as linear combination os the Killing vectors ∂t and
∂
∂φ , namely

u = cosh ν(r)θ0 + sinh ν(r)θ2 ,

uα = ρ cosh ν δαt + (Z cosh ν + Y sinh ν) δαφ , uα uα = −1. (8.57)
The energy–momentum perfect fluid tensor
Tμν = (p + μ)uμ uν + pgμν , (8.58)
where μ and p are the energy density and the pressure, referred to the
orthonormal basis possesses the components
T 0 0 = −μ (r) cosh2 ν − p (r) sinh2 ν,
T 2 0 = −T 0 2 = cosh ν sinh ν (μ (r) + p (r)) ,
T 1 1 = p(r), T 2 2 = μ (r) sinh2 ν + p (r) cosh2 ν, (8.59)
throughout this section, the energy density will be denoted with μ instead of σ
used in Lubo et al. (1999). The energy–momentum conservation T αβ ;β = 0 yields
  
 μ+p   Z
p =− cosh ν ρ Y − sinh ν ρ Y − cosh ν sinh ν ρ
2 2 2
. (8.60)
ρY ρ
104 Stationary Circularly Symmetric Perfect Fluids with Λ

8.3.1 Equations for Rigidly Rotating Fluids

Following Lubo et al. (1999), the stellar models to be analyzed are those with
u aligned along θ0 , i.e., for ν = 0, leaving out of consideration the differen-
tially rotating case. Thus the components of the energy momentum tensor (8.59)
become

T 0 0 = −μ (r) , T 2 0 = −T 0 2 = 0,
T 1 1 = p(r) = T 2 2 , (8.61)

together with
ρ
p = − (μ + p). (8.62)
ρ

The equation G2 0 = 0 yields

     
ρ3 Z ρ3 Z
=0→ = −2β = constant, (8.63)
BY ρ BY ρ

and from G2 2 = G1 1 one gets

  
ρ Y  ρ ρ ρ
= d ln(Y ) = d ln( ) → Y (r) = y0 , y0 → 1, (8.64)
BY B B B

which substituted into (8.63) gives

 
ρ3 Z Z β
= −2β d ( ) = β d ρ−2 → Z(r) = αρ(r) + . (8.65)
ρ ρ ρ ρ(r)

Therefore, the metric for a rigidly rotating fluid in all generality can be given as
 2
ρ
ds2 = −(ρ dt + Z dφ)2 + B 2 dr2 + dφ2 ,
B
β ρ
Z(r) = α ρ(r) + , Y (r) = , (8.66)
ρ(r) B

where α, β, and B are integration constants, with the energy and pressure
fulfilling
  
  2
1 Y
1 ρ Z
−(κμ (r) + Λ) = − ,
BY B
4 BY Y
   2
ρ Y  1 ρ Z
κp(r) − Λ = 2 + . (8.67)
B ρY 4 BY Y

The energy conservation p = − ρρ (μ + p) does not provide new conditions; for
the given μ and p, it becomes an identity.
 8.3 Lubo–Rooman–Spindel Rotating Perfect Fluids 105

8.3.2 Garcia Representation of Stationary Perfect Fluid Solutions

1
A further transformation based on (8.64), B(r)dr = y0 dρ/Y =: ρ A(ρ) dρ, brings
the metric to the form
1
ds2 = −(ρ dt + Z(ρ) dφ)2 + 2 dρ2 + ρ2 A(ρ)2 dφ2 ,
ρ A(ρ)2
Z(ρ) = α ρ + β/ρ, (8.68)
with energy density
β2 dA2 ρ2 d2 A2
κ μ (ρ) + Λ = 3 − A2
− 2 ρ − , (8.69)
ρ4 dρ 2 dρ2
and pressure
β2 ρ dA2
κ p (ρ) − Λ = + A2
+ . (8.70)
ρ4 2 dρ
Providing specific state equations μ = μ(p) one gets different families of
stationary fluid solutions, evidently in a non-familiar spatial polar coordinate.

8.3.3 Barotropic Class of Solutions p = γ μ

For a linear dependence between pressure and density – the barotropic state
equation – of the form
p=γμ (8.71)
one arrives at linear equation allowing for the general solution
γ+1
−1 β2 C2 γ
A(ρ)2 = −2 ρ− γ γ C1 (γ − 1) −Λ+ +2 2 . (8.72)
ρ4 ρ (γ − 1)

Radiation solution p = μ/2

Radiation is determined for γ = 1/2, thus
C2 C1 β2
A(ρ)2 = −Λ − 2 + 2 + . (8.73)
ρ2 ρ3 ρ4

8.3.4 Constant Density Stationary Solution; p = p(r), μ = μ0

For constant mass energy density, one can obtain analytic solutions in (2 + 1)D.
Integrating the equation (8.69) for μ = μ0 = constant with respect to A2 , one
gets
2 β2 C1 C2
A (ρ) = −Λ − M0 κ + 4 + 2 2 − 2 , (8.74)
ρ ρ ρ
which, substituted into (8.70), gives the pressure
C1
κp = −κμ0 − . (8.75)
ρ
106 Stationary Circularly Symmetric Perfect Fluids with Λ

The transformation to the radial coordinate is achieved by solving

gφφ = ρ2 A2 − Z 2 = r2 , (8.76)
which explicitly reads
r2 = −aρ2 − 2 C2 ρ − 2 α β + 2 C1 , a := Λ + α2 + κ μ0 ,
C2 2 r2 2 (C1 − α β) a + C2 2
(ρ + ) =− + . (8.77)
a a a2

8.3.5 Lubo–Rooman–Spindel Perfect Fluids u = θ 0 and grr = 1

In this paragraph interior solutions containing, among others, one parameter
family of Gödel-like geometries, are derived. All the obtained solutions are
causally well behaved; some are nevertheless physically unacceptable; their angu-
lar momentum is too large compared to their mass, and such solutions would
lead to naked causal singularities in case of collapse unless centrifugal forces were
to prohibit them from evolving into black holes.
The choice of the radial coordinate in the stationary metric studied in Rooman
and Spindel (1998) – see Eqs. (80) and (81) – is such that grr = B(r) = 1,
therefore the metric becomes
ds2 = −[ρ dt + (α ρ + β/ρ)dφ]2 + dr2 + ρ y02 dφ2 .
2
(8.78)
When regularity conditions at the origin r = 0 are imposed, then the following
relations arises
β = −α, ρ (0) = 0, C/2 = α ρ (0) = ω, ρ(0) → 1, (8.79)
equivalent to Eq. (1.8) of RS. Hence
 
ρ
2
1 2
ds = −ρ dt + dr + α
2 2 2 2 2
− (ρ − ) dφ2 − 2α(ρ2 − 1)dt dφ. (8.80)
ω2 ρ
Moreover, the matching to the external AdS metric requires additionally
L∞ = −ω, ω 2 < −Λ.

8.3.6 LBR Rotating Perfect Fluid with μ0

Assuming in the metric (8.55) the metric components gtt = −1 = −grr , one
integrates its equations to arrive at a first-order equation for ρ(r), namely
 2
dρ 1
= ω 2 ( 2 − 1) + κ(p(0) + μ0 )(ρ − 1) − (κμ0 /2 + Λ)(ρ2 − 1), (8.81)
dr ρ
which integrates in terms of elliptic functions. In Rooman and Spindel (1998)
additional considerations dealing with the angular momentum and quasi local
mass values are stressed.
 8.3 Lubo–Rooman–Spindel Rotating Perfect Fluids 107

8.3.7 Rooman–Spindel Rotating Fluid Model; gtt = −1 = −grr

There is a sub-branch of stationary fluid solutions for the metric (8.55) with
metric component gtt = −1. The metric in the gauge grr = B(r) = 1 becomes
ds2 = −[dt + Z(r)dφ]2 + dr2 + Y (r)2 dφ2 . (8.82)
The equation G2 0 = 0, (8.56d), yields
d
Y (r) = C1 Z (r) , (8.83)
dr
while (8.67) gives the pressure in the form of a relation
1
κ p (r) = Λ + = constant. (8.84)
4C1 2
The energy fulfills
 
d3 Z 3 dZ
+ κ μ (r) + Λ − 2 = 0. (8.85)
dr3 4C1 dr
As pointed out in Rooman and Spindel (1998), this equation can be thought of
as defining the matter content for a given function Z(r), or, conversely, for a
given plausible matter distribution one determines the structural function Z(r).
The star possesses a boundary if the pressure is set equal to zero, p = 0, thus
Λ = − 4C1 2 < 0, hence for this AdS dust star
1

d3 Z dZ
+ (κ μ (r) + 4Λ) = 0.
dr3 dr
The incompressible dust was treated in detail in Section 3.6.
 9
Friedmann–Robertson–Walker Cosmologies

This chapter is devoted to the derivation of exact solutions to the FRW model
in (3+1) and (2+1) dimensions and to establish their correspondence following
Garcı́a, Cataldo and del Campo, Garcı́a et al. (2003). It is widely known that
any (3+1) FRW perfect fluid fulfilling a barotropic equation of state of the form
p(ρ) = (γ − 1) ρ, under algebraic transformation rules of the parameters, can be
transformed into its (2+1) FRW counterpart. It is noteworthy that the physical
content of the solutions depends on the dimensionality in which they are viewed.
Using four-dimensional terminology, for a vanishing cosmological constant, the
(2+1) analog of the (3+1) dust (γ4 = 1), p4 = 0 is a radiation-dominated universe
(γ3 = 3/2), p3 = ρ3 /2. Conversely, for a (3+1) radiation-dominated universe
(γ4 = 4/3), p4 = ρ4 /3, one finds that the (2+1) counterpart is the stiff matter
(γ3 = 2), p3 = ρ3 . Moreover, the analog of the (3+1) de Sitter spacetime is the
(2+1) de Sitter spacetime, γ4 = 0 = γ3 , with equation of state p = −ρ = const.
in both cases. The family of FRW barotropic solutions are derived in details.
Moreover, the full description of collapsing dust is given for: the Mann–Ross
dust FRW solution with Λ, Section 9.4, and the Gidding–Abbott–Kuchař dust
FRW solution, Section 9.4.5.

9.1 Einstein Equations for FRW Cosmologies

The starting point in the study of FRW cosmologies begins with the homogeneous
and isotropic metric equipped with a scale factor a(t) (function depending on
time), and a constant k, the curvature index, related to the curvature character
of the three-space (closed, flat, open).

9.1.1 Einstein Equations for (3+1) FRW Cosmology

In (3+1)-dimensional gravity, the metric of the FRW model is given by
 
dr2
ds2 = −dt2 + a(t)2 + r 2
dθ 2
+ r 2
sin2
θdφ 2
, (9.1)
1 − kr2
 9.1 Einstein Equations for FRW Cosmologies 109

where, as usual, a(t) is the scale factor, and k = −1, 0, 1. The FRW metric
is conformally flat, i.e., its Weyl tensor vanishes everywhere in the domain of
definition of the spacetime. The scale factor a(t) of the metric (9.1) is governed
by equations modeled in terms of the perfect fluid energy density ρ4 , the matter
isotropic pressure p4 , and a cosmological constant Λ, if present: the energy–
momentum tensor is given by
Tμν = (ρ4 (t) + p4 (t)) uμ uν + p4 (t) gμν , uμ = −δμt ,
hence, the Einstein equations
R
Rμν − gμν + Λ gμν = κ Tμν ,
2
amounts explicitly to
3  2 
Ett : ȧ + k = Λ4 + κ4 ρ, (9.2a)
a2

E i i : 2 a ä + ȧ2 + k + a2 (κ p4 − Λ4 ) = 0
a
→ a ä + (κ ρ4 + 3 κ p4 − 2 Λ4 ) = 0, (9.2b)
6
d   da3
ρ4 a3 + p4 = 0 ≡ T μ ν ;μ = 0. (9.2c)
dt dt
The equation (9.2c) represents the conservation of the matter content: T μ ν ;μ = 0.
It results also from the substitution of ȧ from (9.2a) into ä of (9.2b); thus only
two of the equations of this system are independent.

9.1.2 Einstein Equations for (2+1) FRW Cosmology

In (2+1)-dimensional gravity, the analogous to the FRW metric (9.1) is given by
 
dr2
ds = −dt + a(t)
2 2 2 2 2
+ r dθ . (9.3)
1 − kr2
This metric is characterized by the vanishing of the Cotton tensor, and hence,
from the 3D point of view, this conditions leads to the conformal flatness of
the (2+1) FRW metric. Moreover, the metric (9.3) can be considered as a
dimensional reduction of the metric (9.1) for any fixed value of the azimuthal
angle φ.
The corresponding (2+1) Einstein equations for the fluid, Tμν = (ρ3 (t) +
p3 (t))uμ uν + p3 (t) gμν , uμ = −δμt , are
ȧ2 + k
= κ3 ρ3 + Λ3 , (9.4a)
a2
ä + a (κ p3 − Λ3 ) = 0, (9.4b)

d  2 da2
ρa + p = 0. (9.4c)
dt dt
110 Friedmann–Robertson–Walker Cosmologies

Substituting ȧ from (9.4a) into (9.4b) one arrives at (9.4c), which is the equation
arising from the energy conservation T μ ν ;μ = 0.
For dust, the pressure vanishes, p = 0, and the conservation of the energy–
momentum tensor implies that

3 + 1 : ρ a3 = ρ0 a30 = const.,

2 + 1 : ρ a2 = ρ0 a20 = const.,

where ρ0 is the initial density and a0 the value of the scale factor at the beginning
of the time counting.

9.2 Barotropic Perfect Fluid FRW Solutions

In this section, by a straightforward integration of the field equations in the case
of a perfect fluid, fulfilling linear state equations of the form p + ρ = γ ρ, the
general solutions in (3+1) and (2+1) spacetimes are derived.

9.2.1 Barotropic Perfect Fluid (3 + 1) Solutions

In (3+1) FRW spaces, for linear equations of state

p4 = (γ4 − 1) ρ4 , (9.5)

the equation (9.2c) amounts to

d ln ρ4 + 3γ4 d ln a = 0, (9.6)

and hence its general integral is

ρ4 = ρ40 a−3γ4 , (9.7)

where ρ40 is an integration constant. The integral of (9.2a) is given by


a 2 γ4 −1 da
3

t − t0 = a  , (9.8)
Λ4 κ4
a3γ4 − ka2( 2 γ4 −1) +
3
3 3 ρ40

where a = ±1.
From (9.8), it becomes apparent that one can not, in general, express in terms
of elementary functions t as function of a.
Nevertheless, for Λ4 = 0 and arbitrary γ4 , the above integral is given in terms
of hypergeometric functions, namely
  9  
2 γ4 − 2 a 2 γ4 −1
3 3 3
a a 2 γ4 1 2 γ4
t(a) = t0 + F , , ,k . (9.9)
ρ40 κ4 /3 32 γ4 2 3γ4 − 2 3γ4 − 2 ρ40 κ4 /3
 9.2 Barotropic Perfect Fluid FRW Solutions 111

(3 + 1) FRW Solution a(t) ∼ t

For γ4 = 2/3, p4 = −ρ4 /3, the scale factor amounts to

κ4
a(t) = a ρ − k t + a0 , (9.10)
3 40
where a0 is a constant of integration. This solution is quite unphysical but it is
included for reason of comparison with the linear dependent a(t) of the (2 + 1)
case.

(3 + 1) de Sitter Cosmological Solution

On the other hand, for γ4 = 0, p4 = −ρ4 = −ρ40 = const., i.e., de Sitter
spacetime, the scale factor is

e−a C0 (t−t0 )  2a C0 (t−t0 )
 κ4
a(t) = k+e , C0 = ρ . (9.11)
2C0 3 40

9.2.2 Barotropic Perfect Fluid (2 + 1) Solutions

For (2+1) FRW cosmology with linear state equations of the form p3 = (γ3 −1)ρ3 ,
the dynamical field equation (9.4c) possesses as general integral
ρ3 = ρ30 a−2γ3 , (9.12)
where ρ30 is a constant of integration. The integral of t, from (9.4a), amounts to

aγ3 −1 da
t − t0 = a  . (9.13)
Λ3 a2γ3 − ka2(γ3 −1) + κ3 ρ30

For Λ3 = 0, the integral (9.13) can be written in terms of hypergeometric

functions
    
a aγ3 1 γ3 3γ3 − 2 a2γ3 −2
t(a) = t0 + √ F , , ,k . (9.14)
κ3 ρ30 γ3 2 2γ3 − 2 2γ3 − 2 κ3 ρ30

Saslaw, Collas, Gidding–Abbott–Kuchař, and Cornish–Frankel Dust Solution,

a(t) ∼ t
Moreover, for zero cosmological constant Λ = 0, γ3 = 1, and p3 = 0, the scale
factor can be given as
a(t) = a κ3 ρ30 − k t + a0 , (9.15)
where a0 is an integration constant.
In Collas (1977), Eq.(58)–Eq.(64), the dust-filled open universe linearly
expands in time, therefore there is no deceleration, and consequently it satis-
fies the Hubble law, which has been derived in detail for the first time to our
knowledge; although the work by Saslaw (1977), published in September 1977
(the same month of the Collas publication), also contains the same results.
112 Friedmann–Robertson–Walker Cosmologies

This kind of solution was also reported in Giddings et al. (1984), Eq.(77). In
the paragraph prior to this equation, it is pointed out “that in structure the
equation for a radiation-dominated Universe in n dimensions is identical to that
for a dust-filled Universe in n + 1 dimensions.”
Later, in 1991, this kind of cosmology was reported in Cornish and Frankel
(1994), Eq.(4.5–4.6).
(2 + 1) de Sitter Cosmological Solution
On the other hand, for γ3 = 0, consequently p3 = −ρ3 = −ρ30 = const., i.e., one
is dealing with the de Sitter metric, the scale factor amounts to
e−a C0 (t−t0 )   √
a(t) = k + e2a C0 (t−t0 ) , C0 = κ3 ρ30 . (9.16)
2C0
Saslaw (1977), in the paragraph following equation (6), wrote the sentence: “If
Λ = 0, one can similarly compare models for 2+1 and 3+1 dimensional universes.
For p = 0 – dust – one obtains a = a0 exp t/τ , which is the analogous of the de
Sitter model.”

Saslaw, Cornish–Frankel Radiation-Dominated Universes; ρ3 = 2p3

In Saslaw (1977), after Eq. (6), it is stated that, in (3 + 1) dimensions, the
dust-filled universe k = 0, a(t) ∼ t2/3 corresponds, in (2 + 1) dimensions, to the
radiation-dominated universe with ρ ∼ a−3 .
In Cornish and Frankel (1994), Eq.(4.4) and Eq.(5.20), the solutions for
expanding universes in the case of radiation are given explicitly as:
2 1
k=0: t= √ a3/2 , (9.17a)
3 2GM0 a0

k = 1 : t = 2GM0 a0 arcsin A1/2 − A1/2 (1 − A)1/2 , (9.17b)

k = −1 : t = 2GM0 a0 A1/2 (1 + A)1/2 − sinh−1 A1/2 , (9.17c)

√ a
where A = 2GM0 a0
.

9.2.3 Comparison Between (3+1) and (2+1) Barotropic Solutions

Assuming that the time t and the scale factor function a(t) are structurally
invariant functions for the studied (3+1) and (2+1) metrics, let us compare the
integral solution t given by (9.8) with the one defined by (9.13); by accomplishing
parameter scaling transformations
κ4 Λ
→ κ3 , 4 → Λ3 , 3γ4 → 2γ3 (9.18)
3 3
in t from (9.8), one arrives at the integral t of (9.13); the equivalence in the
opposite direction holds too. Thus the time t as function of the variable a is a
 9.3 Polytropic Perfect Fluid FRW Solutions 113

structurally invariant function. One reaches the same conclusion when dealing
with the hypergeometric function representation of t determined by (9.9) and
(9.14). On the other hand, if one were able to express the scale factor a as a
function of the variable t, then, via parameter scaling, one would arrive at the
structurally invariant character of the function a(t).
We have established in this way that any FRW cosmology, filled with a perfect
fluid fulfilling a linear state equation, determined in (3+1) dimensions, can be
reduced to its (2+1) counterpart by using the correspondence (9.18); the converse
statement holds, too.
Moreover, considering (3+1) and (2+1) FRW cosmologies as independent enti-
ties, dominant energy conditions for fluids: ρ ≥ 0 and −ρ ≤ p ≤ ρ, have to
hold on their own account in (2+1) and (3+1) dimensions. Therefore, the (3+1)
dimensional state parameter γ4 has to fulfill the condition 0 ≤ γ4 ≤ 2, while
independently the (2+1) dimensional state parameter γ3 has to range the values
0 ≤ γ3 ≤ 2.
On the other hand, assuming that the considered spacetimes are in the corre-
spondence (9.18), one arrives at restrictions for the values one can assign to the
state parameters γ, namely:
4
0 ≤ γ3 ≤ 2, and 0 ≤ γ4 ≤ .
3
Thus, the class of (3+1) perfect fluid cosmologies which participates in the cor-
respondence with the whole family of (2+1) perfect fluid cosmologies is more
narrow compared with the whole (3+1) perfect fluid cosmology; (3+1) cosmol-
ogy with 43 < γ4 ≤ 2 are out of the comparison scheme. Hence, thinking in
terms of dimensionally reduced spaces, a perfect fluid FRW solution given in
(3+1) dimensions, which can be reduced to its (2+1) cosmological counterpart,
possesses state parameters given in the above-specified ranges. From this point
of view, using the four-dimensional terminology, for vanishing cosmological con-
stants Λ4 = Λ3 = 0, the (2+1) analog of the (3+1) dust (γ4 = 1), p4 = 0, is
a radiation-dominated universe (γ3 = 3/2), p3 = ρ3 /2. Conversely, for a (3+1)
radiation-dominated universe (γ4 = 4/3), p4 = ρ4 /3, one finds that the (2+1)
counterpart is the stiff matter (γ3 = 2), p3 = ρ3 . Moreover, the (3+1) de Sit-
ter spacetime coincides with the (2+1) de Sitter spacetime, γ4 = 0 = γ3 , with
equation of state p = −ρ = const.

9.3 Polytropic Perfect Fluid FRW Solutions

In what follows, solutions for perfect fluids subjected to polytropic state
equations p = α ργ are derived.

9.3.1 Polytropic Perfect Fluid (3 + 1) Solutions

Under the polytropic state equation

p4 = α4 ρ4 γ4 , (9.19)
114 Friedmann–Robertson–Walker Cosmologies

the equation (9.2c) becomes

dρ4
+ 3d ln a = 0. (9.20)
ρ4 + α4 ρ4 γ4
hence

d ln [(α4 + ρ(−γ
4
4
+1) 3(−γ4 +1)
)a ] = 0 → ρν4 4 = A40 a−3ν4 − α4 , (9.21)

where ν4 = 1 − γ4 .
Moreover, the integral of t, (9.2a), amounts to

da
t − t0 =  . (9.22)
Λ4 2 1/ν4
3 a −k+ 3 a
κ4 2
A40 a−3ν4 − α4

9.3.2 Polytropic Perfect Fluid (2 + 1) Solutions

In (2+1) cosmology one encounters, as expected, a similar treatment for the
polytropic case, p3 = α3 ρ3 γ3 , which is determined by:
the fluid energy density
ρν3 3 = A30 a−2ν3 − α3 , (9.23)

and the time variable t


da
t − t0 =  . (9.24)
1/ν3
Λ3 a2 − k + κ3 a2 A30 a−2ν3 − α3

9.3.3 Comparison Between (3+1) and (2+1) Polytropic Solutions

Comparing the expressions of coordinate time t, which is assumed to have the
same meaning in both dimensions, as well as the scale factor a(t), one arrives at:
Λ4 κ4 
3  Λ3 , 3  κ3 , but one can not establish a relation between γ s, or equiva-

lently for ν s, to reproduce the energy densities functions from one another. At
most one conclude, due to the structural invariance of a(t), that
  3ν1   2ν1
1 1 ν 4 1 ν 3
= (ρ4 4 + α4 ) = (ρ3 3 + α3 ) . (9.25)
a(t) A4 A3

Hence, for perfect fluids fulfilling the polytropic state equation p = αρ(1−ν) , we
have no relations between exponential factors νd alone.

9.4 Mann–Ross Collapsing Dust FRW Solutions with Λ

In this section, following Mann and Ross (1993), the standard cosmological con-
stant is subjected to the change: Λ → −Λmr , thus for asymptotically AdS spaces
Λmr = 1/l2 , while for asymptotically dS spaces Λmr = −1/l2
 9.4 Mann–Ross Collapsing Dust FRW Solutions with Λ 115

The Einstein equations for a perfect fluid with a cosmological constant for the
(2 + 1) Friedmann–Robertson–Walker metric (9.3),
 
dr2
ds = −dt + a(t)
2 2 2 2 2
+ r dφ ,
1 − kr2

are given by (9.4).

For dust, which is the case to be treated from now on, the pressure vanishes,
p = 0, and the conservation of the energy–momentum tensor implies that

ρ a(t)2 = ρ0 a20 = constant, (9.26)

where ρ0 is the initial density and a20 is the value of the scale factor at the
beginning of the time. The remaining equations are:
 2
d2 a da
+ Λmr a = 0, + k − κ ρ0 a20 + Λmr a2 = 0. (9.27)
dt2 dt

9.4.1 Cosmological dS–FRW Solution

In the case of negative Λmr = −1/l2 , i.e., positive standard cosmological constant
Λ, the Einstein equations (9.27) are solved by
ȧ0 da
a(t) = a0 cosh( −Λmr t) + sinh( −Λmr t), ȧ0 = |t=t0 ,
−Λmr dt
ȧ20 = κ ρ0 a20 − k − Λmr a20 ; κ ρ0 a20 − k − Λmr a20 ≥ 0. (9.28)

For a monotonically increasing scale factor, ȧ(t) < 0, in particular, ȧ(t = t0 ) =

− κ ρ0 a20 − k − Λmr a20 .
If the collapse occurs, it takes place at time t = tc when the scale factor
vanishes, a(tc ) = 0, i.e., for

1 a0 −Λmr
tc = − arctanh
−Λmr ȧ0

1 a0 −Λmr
=− arctanh . (9.29)
−Λmr κ ρ0 a20 − k − Λmr a20

The exterior metric has to be the de Sitter one, with negative parameter M < 0
to have the correct metric signature; see (9.32) below.

9.4.2 Asymptotically AdS–FRW Dust Solution

In the case of positive Λmr = 1/l2 , i.e., negative standard cosmological constant
Λ, the Einstein equations (9.27) are satisfied by
116 Friedmann–Robertson–Walker Cosmologies

ȧ0 da
a(t) = a0 cos( Λmr t) + sin( Λmr t), ȧ0 = |t=t0 ,
Λmr dt

ȧ0 = κ ρ0 a20 − k − Λmr a20 ; κ ρ0 a20 − k − Λmr a20 ≥ 0, real a(t).
(9.30)
At time of the collapse, t = tc , this solution has to vanish, a(tc ) = 0, hence

1 a0 Λmr
tc = arctan . (9.31)
Λmr κ ρ0 a20 − k − Λmr a20

9.4.3 Matching the AdS–FRW Dust to the Static BTZ

To match the dust solution of the FRW metric to the exterior static BTZ solution
endowed with mass and a negative cosmological constant Λ = −1/l2 = −Λmr ,
one has to fit the continuity conditions of the metrics and the extrinsic curvature
tensors. The static BTZ can be given in the exterior coordinates {T, R, φ} as
dR2
g = −(Λmr R2 − M )dT 2 + + R2 dφ2 . (9.32)
Λmr R2 − M
The boundary of the dust distribution is taken to be at rb , and correspondingly
at R(t) in the exterior coordinates. On the boundary edge, the solutions ought
to fulfill the continuity conditions, i.e., the vanishing of the jump of the metric
MR (−) BT Z (+)
gμν , following the Israel (1966) notation: g μν ≡ g μν , g μν ≡ g μν , and of the
extrinsic curvature Kμν :
(−) (+)
[gμν ] := g μν (rb )− g μν (R(t)) = 0, (9.33a)

[Kμν ] = 0. (9.33b)
(−) (+) (−) (+)
From g (rb )− g (R(t)) = 0, and g φφ (rb ) = g φφ (R(t)) one gets
correspondingly
 2  2
dT 1 dR
− 1 = −(Λmr R(t) − M ) 2
+ ,
dt Λmr R(t)2 − M dt
rb a(t) = R(t). (9.34)
The latter condition R(t) = rb a(t) means that in the exterior coordinates the
position R of the boundary is equal to the proper distance rb a(t) from the origin
to the dust edge. The first condition yields

dT Λmr R(t)2 − M + Ṙ(t)2
= , (9.35)
dt Λmr R(t)2 − M
d
on the dust edge; the over-dot denotes dt .
 9.4 Mann–Ross Collapsing Dust FRW Solutions with Λ 117

The initial conditions a0 = 1, and ȧ0 = 0 represent a ball of dust with initial
radius R(t0 ) = rb a(t0 ) = rb initially at rest Ṙ(t)|t0 = rb ȧ(t)|t0 = 0 in the
exterior coordinates.

9.4.4 Determination of Kij

(±)
The full determination of the Kij is accomplished here in detail. The standard
definition of the extrinsic curvature tensor, see Eisenhart (1966), is used
(±)
Kij = Nα ei ν ∇ν ej α = −ei ν ej α ∇ν Nα , Nα ei α = 0, (9.36)

where ei ν are the components of the tangent vector ∂i to the coordinate curve
ξ i defined on the surface, Nα is the normal vector to the surface, and as such
is orthogonal to ei ν . Operationally, it is recommendable to use the directional
description ei ν ∂x∂ ν = ∂ξ∂ i , thus
(±)
   
∂ α ∂
Kij = Nα ej + ei ν ej σ Γα νσ = −ej α N α − e i
ν
N σ Γ σ
να , (9.37)
∂ξ i ∂ξ i
which is close to the Israel formulation.

Extrinsic Curvature for the Interior AdS–FRW Dust Metric

The boundary surface is located at {r = rb , t = t, φ = φ}, consequently the
tangent and normal vectors can be chosen as
1 a(t)
et ν = δt μ , eφ ν = δφ μ , Nα = δr α. (9.38)
rb a(t) 1 − k rb2
The components of the extrinsic curvature are
(−) a(t) (−)
Ktt = Nα Γα tt = Γt tt = 0, Ktφ = 0, (9.39)
1 − k rb
2

(−) 1 1 1 − k rb2
Kφφ = N α Γ α
φφ = Γ r
φφ = − . (9.40)
rb2 a(t)2 rb2 a(t) 1 − k rb2 rb a(t)

Extrinsic Curvature for Exterior Static BTZ Metric

For the exterior metric, the tangent and normal vectors to the surface T (λ),
R(λ), φ = φ are chosen as

eλ μ = T,λ δTμ + R,λ δR

μ
, eφ μ = 1/Rδφμ , (9.41)
Nμ = −R,λ δμT + T,λ δμR , μ
eλ Nμ = 0 = eφ Nμ , μ
(9.42)

Thus the extrinsic curvature tensor components are

(+) (+)
Kλλ = Nμ eλ ν ∇ν eλ μ = −eλ ν eλ μ ∇ν Nμ , Kφφ = Nμ eφ ν ∇ν eφ μ . (9.43)
118 Friedmann–Robertson–Walker Cosmologies

(+)
For Kφφ one has
(+) Λmr R2 − M
Kφφ = Nμ Γμ φφ /R2 = −T,λ , (9.44)
R
where the non-vanishing Christoffel symbols for the external metric are
Λmr R
ΓT T R = = −ΓR RR , ΓR T T = Λmr R(Λmr R2 − M ),
Λmr R2 − M
1
ΓR φφ = − R(Λmr R2 − M ), Γφ Rφ = . (9.45)
R
(+)
For Kλλ one has
(+)
Kλλ = Nμ ∂λ eλ μ + Nμ eλ ν eλ σ Γμ νσ . (9.46)
Consider first
Nμ ∂λ eλ μ = Nμ ∂λ (T,λ δTμ + R,λ δR
μ
) = (−R,λ δμT + T,λ δμR ) (T,λλ δTμ + R,λλ δR
μ
)
= −T,λλ R,λ + R,λλ T,λ . (9.47)
On the other hand,
 
Nμ eλ ν eλ σ Γμ νσ = −R,λ T,λ T,λ ΓT T T + 2 T,λ R,λ ΓT T R + R,λ R,λ ΓT RR
 
+T,λ T,λ T,λ ΓR T T + 2 T,λ R,λ ΓR T R + R,λ R,λ ΓR RR . (9.48)
Substituting the Christoffel symbols one arrives at
Λmr R
Nμ eλ ν (eλ σ Γμ νσ ) = Λmr R(Λmr R2 − M )(T,λ )3 − 3 T,λ (R,λ )2 .
Λmr R2 − M
(9.49)
Setting λ = t, and using T,t from (9.35), together with
ṘR̈
T,tt = 
(Λmr R2 − M ) Λmr R2 − M + Ṙ2
 
Λmr R Ṙ Λmr R2 − M + 2Ṙ2
−  (9.50)
(Λmr R2 − M )2 Λmr R2 − M + Ṙ2
one arrives at
(+) 
∂
Ktt = Λmr R2 − M + Ṙ 2 . (9.51)
∂R
Therefore, since on the boundary r = rb the smoothness of Kφφ from (9.40) and
(9.44) yields

1 − k rb2
Λ R −M
2 Λmr R2 − M + Ṙ 2
− = −T,λ mr =− , (9.52)
rb a(t) R R
 9.4 Mann–Ross Collapsing Dust FRW Solutions with Λ 119

where it has been used T,λ = T,t from (9.35). Since on the boundary

R(t) = rb a(t), Ṙ(t) = rb ȧ(t), (9.53)

hence, one arrives at the condition

2
Λmr rb 2 a(t)2 − M + rb 2 ȧ(t) = 1 − k rb 2 , (9.54)

isolating M one gets

M = (k + Λmr a(t)2 + ȧ(t)2 ) rb2 − 1 = κ rb2 a20 ρ0 − 1 (9.55)

where (9.27) was used in agreement with the results reported in Mann–Ross.
The existence of an event horizon in the static BTZ black hole around the
collapsing dust, Rh = M Λmr , M > 0, requires ρ0 > 1/(κ rb2 a20 ).

9.4.5 Gidding–Abbott–Kuchař Dust FRW Solution

In Giddings et al. (1984), §7, one finds the (2+1) dynamical dust solution without
a cosmological constant, analogous to the Oppenheimer–Snyder collapsing dust
of (3 + 1) dimensions; see Oppenheimer and Snyder (1939).
This case is contained in the above-derived solution. Setting the cosmological
constant equal to zero in (9.27), one gets
d2 a
= 0 → a(t) = ȧ0 t + a0 ,
dt2
 2
da
+ k − κ ρ0 a20 = 0 → ȧ20 + k − κ ρ0 a20 = 0
dt

→ ȧ0 = ± κ ρ0 a20 − k. (9.56)

This scalar factor a(t) can also be obtained from (9.30) as a limit as Λ goes to
zero. The collapse, if any, occurs at a(tc ) = 0, i.e.,

tc = −a0 /ȧ0 = a0 / κ ρ0 a20 − k,

where ȧ0 = − κ ρ0 a20 − k has been chosen to have tc > 0, assuming a0 > 0; the
collapse has to take place after the initial time t = 0.
The exterior metric (9.32) for vanishing Λ becomes a flat metric with conical
singularity
dR2
g = −C dT 2 + + R2 dφ2 , C := −M > 0. (9.57)
C
Choosing the dust edge at r = rb , with equation R(t), the matching conditions
(9.33a) and (9.33b) of the FRW metric (9.3) to the metric (9.57) are those of
(9.34), (9.35), (9.52) and (9.51) for Λ = 0, namely

R(t) = rb a(t) (9.58a)

120 Friedmann–Robertson–Walker Cosmologies

2
dT C + Ṙ(t)
= , (9.58b)
dt C

(−) 1 − k rb2 C (+)

− Kφφ = = Ṫ = − Kφφ , (9.58c)
rb a(t) R

(−)
 (+)
d 2
Ktt = 0 = C + Ṙ(t) =Ktt , (9.58d)
dR
Using Ṫ from (9.58b) in (9.58c), taking into account that R(t) = rb a(t), one
arrives at
1 − k rb2 = C + Ṙ(t)2 , (9.59)
which, isolating C, considering that R(t) = rb a(t) and the derivative ȧ(t) from
(9.56), yields
2
C = 1 − (k + ȧ(t) ) rb2 = 1 − κ rb2 a20 ρ0 > 0, (9.60)
in agreement with Mann and Ross (1993), Eq. (32).
 10
Dilaton–Inflaton Friedmann–Robertson–Walker
Cosmologies

The purpose of this chapter is to provide a new insight on (2+1) and (3+1)
Friedmann–Robertson–Walker (FRW) cosmologies by establishing a bridge
between them. In order to achieve this goal, I shall begin with a comparison
of the dynamical equations corresponding to (2+1) and (3+1) FRW spacetimes
coupled to matter perfect fluid sources, scalar field (inflaton, dilaton) fields, and
cosmological constants. A (2+1) FRW spacetime may be considered as a dimen-
sional reduction of the associated (3+1) FRW spacetime, arising as result of
the freezing (constant value assignation) of the azimuthal angle (in spherical
coordinates) of this last (3+1) space. A similar approach has been applied suc-
cessfully by Cataldo, del Campo and Garcia, Cataldo et al. (2001), to the (3+1)
Plebański–Carter[A] metric – see Plebański (1975) and Carter (1968) – to derive
the (2+1) BTZ black hole solution.
It is shown that FRW cosmological models coupled to a single scalar field
and to a perfect fluid fitting a wide class of matter perfect fluid state equations,
determined in (3+1) dimensional gravity, can be related to their (2+1) cosmo-
logical counterparts, and vice versa, by using simple algebraic transformations
relating gravitational constants, state parameters, perfect fluid and scalar field
characteristics. It should be pointed out that the demonstration of these relations
for the scalar fields and potentials does not require the fulfillment of any state
equation for the scalar field energy density and pressure. As far as the perfect
fluid is concerned, one has to demand the fulfillment of state equations of the
form p + ρ = γ f (ρ). If the considered cosmologies contain the inflaton field alone
φ, then any (3+1) scalar field cosmology possesses a (2+1) counterpart, and vice
versa.
Notice that one is tacitly assuming that coordinates remain the same ones
for both (3+1) and (2+1) FRW metrics. It is notable that these spacetimes are
both conformally flat, i.e., correspondingly their Weyl and Cotton ten-
sors vanish. By associated (corresponding) spacetimes we mean spaces that
belong to a specific family: for instance, spaces fulfilling a (linear) barotropic
state equation, or those fitting a polytropic law. Moreover, (2+1) FRW solu-
tions to a barotropic perfect fluid state equation are in correspondence with
122 Dilaton–Inflaton Friedmann–Robertson–Walker Cosmologies

(3+1) FRW cosmological solutions to a barotropic perfect fluid state equation,

but with different values of the state parameters. Among other considerations,
maintaining the (3+1) terminology, the (2+1) counterpart of the (3+1) radiation
is the stiff matter. By a straightforward integration of the corresponding field
equations, wide families of solutions are derived, which are used to check indepen-
dently the fulfillment of the theorem. Special attention is addressed to solutions
associated to inflaton scalar field cosmologies, for instance, to Barrow–Saich, and
Barrow–Burd–Lancaster–Madsen solutions.
The outline of the present chapter is as follows:
In Section 10.1 we briefly review the Einstein field equations for (3+1) and (2+1)
FRW models and demonstrate a theorem. In Section 10.2 we derive single scalar
field solutions to barotropic state equations. Section 10.3 deals with spatially flat
FRW solutions to barotropic state equations for both matter and scalar fields
simultaneously present. In Section 10.4 we derived single scalar field spatially
flat FRW solutions to state equations of the form pφ + ρφ = Γ ρφ β . In Section
10.5, single scalar field spatially flat FRW solutions for a given scale factor a(t)
are determined.
Some conventions are followed: Latin and Greek letters with the subscript 0
denote constants, for instance, A0 , t0 or ρφ , although constants of common use,
30
such as gravitational constant κ, cosmological constant Λ, and those appearing in
state equations, γ, Γ, and β, will be typed without any subscript. When extract-
ing the square root of a quantity, say m, the ± sign will be denoted by m . In the
derivation of solutions practically the same pattern will be followed. Moreover,
the conventional perfect fluid description of the field equations is widely used to
derive most of the solutions presented here. Abbreviations for perfect fluid (PF)
and scalar field (dilaton) are used.

10.1 Equations for a FRW Cosmology with a Perfect Fluid and a

Scalar Field
In this section, the field equations for isotropic homogeneous FRW models filled
with a perfect fluid and a single scalar field φ minimally coupled to gravity with
a self-interacting potential V (φ) are explicitly given. A theorem relating (3+1)
and (2+1) solutions is demonstrated.
Throughout this chapter, the energy density will be denoted by ρ(t, r) instead
of μ to correspond with earlier publications.

10.1.1 Einstein Equations for (3+1) FRW Dilaton Cosmology

In (3+1) dimensional gravity, the metric of the FRW model is given by
 
dr2
ds2 = dt2 − a(t)2 + r 2
dΩ2
, (10.1)
1 − kr2
 10.1 Equations for a FRW Cosmology with a Perfect Fluid 123

where, as usual, dΩ2 := dθ2 + sin2 θdφ2 , a(t) is the scale factor, and k = −1, 0, 1
denotes the curvature index. This metric is conformally flat, i.e., its Weyl tensor
vanishes everywhere in the domain of definition of the spacetime. The scale factor
a(t) of the metric (10.1) is governed by equations modeled in terms of the perfect
fluid energy density ρ4 , the matter isotropic pressure p4 , the scalar field φ, the
self-interacting potential V (φ), and a cosmological constant Λ, if present.

Standard Formulation of the (3+1) Field Equations

These dynamical equations are:
ȧ2 + k 1 2
3 = κ4 (ρ4 + φ˙4 + V4 ) + Λ4 , (10.2)
a2 2
ȧ
ρ˙4 + 3 (ρ4 + p4 ) = 0, (10.3)
a
ȧ d
φ¨4 + 3 φ̇4 + V (φ4 ) = 0. (10.4)
a dφ4

Perfect Fluid Formulation of the (3+1) Field Equations

In the conventional perfect fluid notation to describe the scalar field, one
defines the energy density and the pressure associated with the scalar field
correspondingly as:
1 2
ρφ = φ˙4 + V4 , (10.5)
4 2
1 ˙2
φ − V4 .
pφ = (10.6)
2 4
4

In terms of these quantities, the (3+1) dynamical equations are:

ȧ2 + k
3 = κ4 (ρ4 + ρφ ) + Λ4 , (10.7)
a2 4

ȧ
ρ˙4 + 3 (ρ4 + p4 ) = 0, (10.8)
a
ȧ
ρ̇φ + 3 (ρφ + pφ ) = 0, (10.9)
4 a 4 4

The (10.3) represents the conservation of the matter content, while (10.4)
corresponds to the energy conservation of the scalar field.

10.1.2 Einstein Equations for (2+1) FRW Cosmology

In (2+1)-dimensional gravity, the analogous to the FRW metric (10.1) is given by
 
dr2
ds2 = dt2 − a(t)2 + r 2
dθ 2
. (10.10)
1 − kr2
124 Dilaton–Inflaton Friedmann–Robertson–Walker Cosmologies

This metric is characterized by the vanishing of the Cotton tensor, and hence,
from the 3D point of view, it is conformally flat. Moreover, the metric (10.10)
can be considered as a dimensional reduction of the metric (10.1) for any fixed
value of the azimuthal angle φ.

Standard Formulation of the (2+1) Field Equations

The corresponding (2+1) Einstein equations are
ȧ2 + k 1 2
2
= κ3 (ρ3 + φ˙3 + V3 ) + Λ3 , (10.11)
a 2

ȧ
ρ˙3 + 2 (ρ3 + p3 ) = 0, (10.12)
a

ȧ d
φ¨3 + 2 φ̇3 + V (φ3 ) = 0. (10.13)
a dφ3

Perfect Fluid Formulation of the (2+1) Field Equations

Introducing the scalar field density and pressure defined through
1 ˙2
ρφ = φ + V3 , (10.14)
3 2 3

1 ˙2
pφ = φ − V3 , (10.15)
3 2 3
the (2+1) dynamical equations become
ȧ2 + k
= κ3 (ρ3 + ρφ ) + Λ3 , (10.16)
a2 3

ȧ
ρ˙3 + 2 (ρ3 + p3 ) = 0, (10.17)
a

ȧ
ρ̇φ + 2 (ρφ + pφ ) = 0. (10.18)
3 a 3 3

It is apparent that the field equations for metrics (10.1) and (10.10) are dif-
ferent because of the difference in dimensions. Nevertheless, one may assume
that the time coordinate t remains the same in both (2+1) and (3+1) dimen-
sions. Moreover, one also may assume that the scale factor a(t) is a structurally
invariant function depending on t and certain constants; by structural invari-
ance we mean that under dimensional reduction the function a(t) maintains its
form with respect to the t variable as well as its dependence on the constants
involved. The extension of this concept to functions depending on other variables
is straightforward.
The main result of this section can be formulated as a theorem.
 10.1 Equations for a FRW Cosmology with a Perfect Fluid 125

10.1.3 Correspondence Between (3+1) and (2+1) Solutions

Theorem 10.1 Assuming that the time coordinate t and the scale factor
a(t) are structurally invariant functions in both (2+1) and (3+1) dimensional
FRW cosmologies coupled, in each dimension, to a single scalar field and to a
perfect fluid subjected to state equations p + ρ = γf (ρ), where f (ρ) is a struc-
turally invariant functions, then the constants and structural functions of these
cosmologies are related according to the following rules:
κ4 Λ
parameter scaling:  κ3 , 4  Λ3 , 3γ4  2γ3 ,
3 3
3
function scaling: ρ4  ρ3 , φ  φ3 ,
2 4
1 2 1 2
V4 − φ˙4 → V3 , V3 + φ˙3 → V4 . (10.19)
4 6

Proof Considering that the time coordinate t as well the scale factor a(t) remain
unchanged, comparing (10.2) and (10.11) one has
ȧ2 + k κ4 1 2 Λ 1 2
= (ρ4 + φ˙4 + V4 ) + 4 = κ3 (ρ4 + φ˙3 + V3 ) + Λ3 ⇒
a2 3 2 3 2
κ4 Λ4
 κ3 ,  Λ3 , ρ4  ρ3 , (10.20)
3 3
together with
1 ˙2 1 2
φ4 + V4  φ˙3 + V3 , ∼ ρφ  ρφ . (10.21)
2 2 4 3

Next, assuming that in each space the state equation for matter is of the form
p + ρ = γf (ρ), where f (ρ) is a structurally invariant function, i.e., it is a form-
invariant function as viewed from the spaces under consideration, the matter
conservation equations yield
da 1 dρ4 1 dρ3
=− =− , (10.22)
a 3γ4 f (ρ4 ) 2γ3 f (ρ3 )
hence, because of by assumption f (ρ4 )  f (ρ3 ), one has
 ρ4  ρ3
a 1 dρ 1 dρ
ln =− =− , (10.23)
a0 3γ4 f (ρ) 2γ3 f (ρ)
therefore
3γ4  2γ3 . (10.24)
To establish the remaining relationships on scalar fields φ and potentials V (φ)
we rewrite (10.4) and (10.13) correspondingly as:
d 1 ˙2 d ȧ 2
(3 + 1) : φ4 + V (φ4 ) + 3 φ˙4 = 0,
dt 2 dt a
d 1 ˙2 d ȧ ˙ 2
(2 + 1) : φ + V (φ3 ) + 2 φ3 = 0. (10.25)
dt 2 3 dt a
126 Dilaton–Inflaton Friedmann–Robertson–Walker Cosmologies

Starting from the (3+1) equation, assuming again that t and a(t) remain
unchanged under dimensional reduction, we shall establish the transformations
of φ4 and V4 to recover the (2+1) equation. First, one has
 2  2
1 d 1 3 d ȧ 3 ˙
(1 − ) φ̇ + V (φ4 ) + 2 φ = 0, (10.26)
3 dt 2 2 4 dt a 2 4

which rewrites as
 2  2
d 1 3 d 1 ˙2 ȧ 3 ˙
φ̇ + [V (φ4 ) − φ4 ] + 2 φ = 0. (10.27)
dt 2 2 4 dt 4 a 2 4

Comparing with the corresponding (2+1) equation, one establishes


3 1 2
φ → φ3 , V4 − φ˙4 → V3 . (10.28)
2 4 4
The inverse transformations read:

2 1 2
φ3 → φ4 , V3 + φ˙3 → V4 . (10.29)
3 6
Finally, we demonstrate that (10.21) does not contribute with an independent
2
relation; in fact, since by definition ρφ = 12 φ˙4 +V4 , replacing the correspondence
4
2
above, (10.28), one obtains ρ = 1 φ˙ + V =: ρ .
φ 2 3 3 φ
4 3

Corollary In (2+1) and (3+1) dimensional FRW cosmologies, with cosmolog-

ical constants, coupled to a single scalar field, under the assumption of invariance
of the time coordinate as well as the scale factors in both (3+1) and (2+1) spaces,
the constants and structural functions are related according to the following rules:
κ4 Λ
 κ3 , 4  Λ 3 ,
3
 3
3 1 1
φ  φ3 , V4 − φ̇24 → V3 , V3 + φ̇23 → V4 . (10.30)
2 4 4 6
The proof follows immediately from the theorem above. Therefore, any (3+1)
inflationary cosmology possesses a (2+1) partner and conversely.

Nevertheless, scalar fields, modeled conventionally in terms of fluid quantities,

by definition fulfill pφ + ρφ = φ̇2 = 0. Thus, one can think of this last equation as
a state equation for the scalar field and assume the existence of relations of the
form pφ + ρφ = ΓF (ρφ ). If one were assuming the structural invariant property
of functions F (ρφ ), because of the equations φ̇24 = Γ4 F (ρφ ) and φ̇23 = Γ3 F (ρφ ),
4 3
one could straightforwardly determine the relationship between Γs, namely

3Γ4  2Γ3 . (10.31)

 10.2 Single Scalar Field to Linear State Equations; Λ = 0 127

In the next sections, cosmologies with scalar fields subjected to state equa-
tions pφ + ρφ = ΓF (ρφ ), for which (10.31) holds, are derived. The advantage
of using these equations resides in the uniqueness of the derived solutions. For
such branches of solutions one tacitly assumes that the above condition (10.31)
is fulfilled.
For matter perfect fluids, dominant energy conditions require that ρ ≥ 0, and
−ρ < p < ρ, therefore determining this kind of solution, one has additionally
to take care of the fulfillment of this inequality in each spacetime, no matter its
dimension. The validity of this physical requirement is assumed to hold beyond
four dimensions.

10.2 Single Scalar Field to Linear State Equations; Λ = 0

By integrating the dynamical equations for a vanishing cosmological constant in
the case of a single scalar field, described in the conventional fluid formulation,
fulfilling linear state equations pφ +ρφ = γ ρφ , the general solutions for (3+1) and
(2+1) FRW spacetimes are derived. The following two subsections are devoted
to general non-flat (k = 0) cosmologies, while the third one deals with spatially
flat (k = 0) FRW spacetimes. By simple comparison one establishes that the
conditions stated in the theorem hold.

10.2.1 (2+1) Solutions for a Scalar Field

The derivation of solutions for a scalar field, modeled through perfect fluid
quantities, obeying a linear state equation pφ = (Γ3 − 1)ρφ , is given in some
3 3
detail.
The equation for ρφ can be written as
3

d 2
ρ + (ρφ + pφ ) = 0, (10.32)
da φ3 a 3 3

thus, for the considered linear state equation, one obtains

ρφ = ρφ a−2Γ3 , ρφ = const. (10.33)
3 30 30

Moreover, (10.16) yields

ȧ2 = κ3 ρφ a2 − k = κ3 ρφ a−2(Γ3 −1) − k, (10.34)
3 30

hence

aΓ3 −1
t − t0 = a  da, (10.35)
κ3 ρφ − k a2(Γ3 −1)
30

which can be given in terms of hypergeometric functions as

    
a aΓ3 1 Γ3 3Γ3 − 2 a2Γ3 −2
t(a) = t0 + F , , ,k . (10.36)
κ3 ρφ Γ3 2 2Γ3 − 2 2Γ3 − 2 κ3 ρφ
30 30
128 Dilaton–Inflaton Friedmann–Robertson–Walker Cosmologies

Adding (10.14) and (10.15) one gets (dφ3 /dt)2 = ρφ + pφ , which can be
3 3
written as
 2
dφ3
ȧ2 = Γ3 ρφ . (10.37)
da 3

Substituting in the above equation ȧ2 from (10.34), and the expression of ρφ
3
from (10.33), one obtains
Γ3 da
dφ3 = a φ ρφ  , (10.38)
− ka2Γ3 −2 a
30
κ3 ρφ
30

Γ3 −1
which, by introducing the variable z = a , amounts to
a φ Γ3   
dφ3 = − √ d ln κ3 ρφ + κ3 ρφ − kz 2 − ln z . (10.39)
(Γ3 − 1) κ3 30 30

Integrating this last expression one arrives at

Γ 1  
φ3 − φ30 = −a φ √ 3 ln κ3 ρφ + κ3 ρφ − ka2(Γ3 −1)
κ 3 Γ3 − 1 30 30

− ln a(Γ3 −1) , (10.40)

where φ30 is an integration constant. Moreover, introducing

√
C3 = a φ κ3 (Γ3 − 1)/ Γ3 ,
the expression of a in terms of φ3 amounts to
exp (−C3 (φ3 − φ30 ))
a(Γ3 −1) = 2 κ3 ρφ . (10.41)
30 k + exp (−2C3 (φ3 − φ30 ))
On the other hand, subtracting the (10.15) from (10.14), one has 2V3 = 2ρφ −
3

φ̇3 = (2 − Γ3 )ρφ , which explicitly becomes

2
3

 −2 ΓΓ−1
3
2 − Γ3 exp (−C3 (φ3 − φ30 )) 3
V3 = ρφ 2 κ3 ρφ . (10.42)
2 30 30 k + exp (−2C3 (φ3 − φ30 ))

10.2.2 (3+1) Solutions for a Scalar Field

The general solution for a scalar field, fulfilling a linear equation of state pφ =
4
(Γ4 − 1)ρφ , is explicitly given by:
4
the scalar field density
ρφ = ρφ a−3Γ4 , ρφ = const., (10.43)
4 40 40

the time variable t as function of the scale factor a


a 2 Γ4 −1
3

t − t0 = a  da, (10.44)
κ4 2( 32 Γ4 −1)
ρ
3 φ − k a
40
 10.2 Single Scalar Field to Linear State Equations; Λ = 0 129

which, in terms of hypergeometric functions, can be expressed as

  9  
3 3
− 3Γ4 −2
a a 2 Γ4 1 Γ Γ 2 a
t = t0 +  κ F , 2 4
, 2 4
, k κ4 , (10.45)
4
ρ
3
2 Γ4 2 3Γ4 − 2 3Γ4 − 2 3 ρφ40
3 φ 40

the scalar field

   
a φ Γ4 1 κ4 κ4 2( 32 Γ4 −1)
φ4 = − ln ρ + ρ − ka
κ4 /3 23 Γ4 − 1 3 φ40 3 φ40

− ln a( 2 Γ4 −1) + φ40 ,
3
(10.46)

κ
where φ40 is an integration constant. Introducing C4 = a φ 34 ( 32 Γ4 − 1)/ Γ4 ,
the expression of the function a(φ4 ) amounts to

( 32 Γ4 −1) κ4 exp (−C4 (φ4 − φ40 ))
a =2 ρφ . (10.47)
3 40 k + exp (−2C4 (φ4 − φ40 ))
Finally, the potential V4 can be expressed as
  −6 3ΓΓ4−2
2 − Γ4 κ4 exp (−C4 (φ4 − φ40 )) 4
V4 = ρφ 2 ρφ . (10.48)
2 40 3 40 k + exp (−2C4 (φ4 − φ40 ))
It is clear that these (2+1) and (3+1) solutions for a single scalar field fulfill the
requirements of the theorem (corollary) and the condition (10.31).

10.2.3 Slow Roll Spatially Flat FRW Solutions

This subsection is devoted to the derivation of (3+1) and (2+1) inflationary
solutions. A comparison with the existing solutions in the literature is carried
out.

Lucchin–Matarrese Power Law (3+1) Solution

In (3+1) inflationary theory, one of the most important solutions for a single
scalar field was found by Lucchin and Matarrese (1985); see also Liddle and
Lyth (2000), which gives rise to power law inflation. In the conventional scalar
field representation, this unique solution arises for a linear state equation
2
pφ + ρφ = ρ =: Γ4 ρφ . (10.49)
4 4 3α φ4 4

Therefore, the equation for ρφ , (10.9), reads

4

2
dln ρφ + dln a = 0, (10.50)
4 α
with solution
ρφ = ρφ (a/a0 )−2/α , (10.51)
4 40
130 Dilaton–Inflaton Friedmann–Robertson–Walker Cosmologies

where the constant a0 has been introduced for further convenience; notice that
the dynamical equations, for k = 0 and Λ4 = 0, are invariant under the change
a → a/a0 , thus without lost of generality one may set a0 = 1. Substituting this
expression of ρφ into (10.7), which yields
4

d(a/a0 )1/α = 1/α κ4 ρφ /3 dt, (10.52)
40

thus its integral becomes

a = a0 (κ4 /α2 ρφ /3)α/2 tα . (10.53)

40

Equating ρφ to 3α2 /κ4 , one arrives at the power law

40

a(t) = a0 tα . (10.54)

Next, the equation for φ4 becomes

φ˙4 = φ 2α/κ4 t−1 , (10.55)

with general solution


α
φ4 − φ40 = φ 2 ln t. (10.56)
κ4

Finally, the evaluation of V4 = ρφ − φ˙4 /2 yields the slow roll the self-interacting
4
potential

α −2 α κ
Vφ = (3α − 1) t = (3α − 1) e−φ 2 α4 (φ4 −φ40 )
. (10.57)
4 κ4 κ4

Power Law (2+1) Solution

In (2+1) cosmology one may construct the analog to the inflaton solution via
the correspondence (10.19) and the relation (10.31). In this way one obtains the
(2+1) power law slow roll inflaton solution:
power law scale factor
a(t) = a0 tα , (10.58)

the homogeneous scalar field


α
φ3 − φ30 = φ ln t, (10.59)
κ3
slow roll the self-interacting potential

α −2 α −2φ κ
3 (φ3 −φ30 )
Vφ = (2α − 1) t = (2α − 1) e α . (10.60)
3 2κ3 2κ3
In the conventional scalar field representation, this solution corresponds to a
2
−2
linear state equation of the form pφ + ρφ = α1 ρφ =: Γ3 ρφ , with ρφ = ακ t .
3 3 3 3 3 3
 10.3 Spatially Flat FRW Solutions for Barotropic Perfect Fluid 131

Cruz–Martı́nez (2+1) Flat FRW Solution

Cruz and Martı́nez (2000) have obtained a solution which describes a (2+1) flat
FRW cosmology determined through:
 √ 1/Γ3
a(t) = t0 + a Γ3 κ3 t , (10.61)
1  √ 
φ3 (t) − φ30 = ln t0 + a Γ3 κ3 t , (10.62)
κ 3 Γ3

2 − Γ3 −2√κ3 Γ3 (φ3 −φ30 )

V (φ3 ) = e . (10.63)
2
It is clear from the comparison with the (2+1) inflaton solution exhibited
above that the Cruz–Martı́nez solution is a slightly different parametrization of
the inflaton solution.

10.3 Spatially Flat FRW Solutions for Barotropic Perfect Fluid and
Scalar Field
The derivation of the general solutions for spatially flat (k = 0) FRW spacetimes
filled simultaneously with matter and scalar field, modeled by two perfect fluids –
one related to matter and the second one related to the scalar field – is presented
in some details.

10.3.1 Spatially Flat FRW (3+1) Solutions ; γ4 = 2Γ4

For a (3+1) spatially flat FRW spacetime, the equations for ρ4 and ρφ can be
4
written as
d 3
ρ4 + (ρ4 + p4 ) = 0,
da a
d 3
ρ + (ρφ + pφ ) = 0, (10.64)
da φ4 a 4 4

thus, for linear state equations

ρ4 + p4 = γ4 p4
ρφ + pφ = Γ4 pφ (10.65)
4 4 4

their integrals are

ρ4 = ρ40 a−3γ4 , ρ40 = const., ρφ = ρφ a−3Γ4 , ρφ = const. (10.66)

4 40 40

On the other hand, (10.7) for k = 0 yields

 2
ȧ κ
= 4 ρ40 a−3γ4 + ρφ a−3Γ4 , (10.67)
a 3 40
132 Dilaton–Inflaton Friedmann–Robertson–Walker Cosmologies

which, by introducing the new variable z = a3γ4 /2 , can be written as


(γ −Γ )
3 2 4γ 4
ż = a γ4 κ4 /3 ρ40 + ρφ z 4 . (10.68)
2 40

Incidentally, the simplest solutions to (10.68) arise for γ4 = Γ4 , and γ4 = 2Γ4 ,

and amount to
1 
a3γ4 /2 = a γ4 3κ4 (ρ40 + ρφ )(t − t0 ), (10.69)
2 40

 
1 3 2 2
a3γ4 /2 = z = γ ρ κ (t − t0 )2 − ρ40 . (10.70)
ρφ 16 4 φ40 4
40

the first scale factor, (10.69), gives rise to the slow roll power law inflationary
solution, while the last one, (10.70), yields the proper (3+1) Barrow and Saich
(1993) solution; see below.
For the general case γ4 = 2Γ4 , the integral of (10.68), is given in terms of
hypergeometric functions by
   
2 a 1 γ4 3γ4 − 2Γ4
t − t0 = a3γ4 /2
F , , ,
3 γ4 κ4 /3ρ40 2 2γ4 − 2Γ4 2γ4 − 2Γ4

ρφ
− 40 a3(γ4 −Γ4 ) . (10.71)
ρ40
Adding (10.5) and (10.6) one gets (dφ4 /dt)2 , which can be written as
 2
dφ4
ȧ2 = Γ4 ρφ . (10.72)
da 4

Substituting above ȧ2 from (10.67), and the expression of ρφ from (10.66), one
4
obtains
Γ4 2
a 2 (γ4 −Γ4 )
3
φ4 − φ40 = a φ ln ρφ
κ4 /3 3(γ4 − Γ4 ) 40


+ ρ40 + ρφ a3(γ4 −Γ4 ) ,
40

(10.73)
√
κ /3
hence, introducing C4 = a φ √ 4 32 (γ4 − Γ4 ), the expression of a in terms of φ
Γ4
amounts to
1
a 2 (γ4 −Γ4 ) = eC4 (φ4 −φ40 ) − ρ40 e−C4 (φ4 −φ40 ) .
3
(10.74)
2 ρφ
40

Since 2V4 = 2ρφ − φ̇24 = (2 − Γ4 )ρφ , then V4 as function of φ is given by

4 4

 −2 γ
Γ4
−Γ
2 − Γ4 1 4 4

V4 = ρφ eC4 (φ4 −φ40 ) − ρ40 e−C4 (φ4 −φ40 ) (10.75)

.
2 40 2 ρφ
40
 10.3 Spatially Flat FRW Solutions for Barotropic Perfect Fluid 133

10.3.2 Spatially Flat FRW (2+1) Solutions; γ3 = 2Γ3

In the (2+1) case, for linear state equations ρ3 + p3 = γ3 p3 , ρφ + pφ = Γ3 pφ ,
3 3 3
the integrals of the dynamical equations are:
the energy densities

ρ3 = ρ30 a−2γ3 , ρ30 = const., ρφ = ρφ a−2Γ3 , ρφ = const., (10.76)

3 30 30

the time variable t, given in terms of hypergeometric functions depending on the

argument a, is given by
   
a 1 γ3 3γ3 − 2Γ3
t − t0 = √ aγ3 F , , ,
γ3 κ3 ρ30 2 2γ3 − 2Γ3 2γ3 − 2Γ3

ρφ
− 30 a2(γ3 −Γ3 ) , (10.77)
ρ30

the scalar field

1 Γ3
φ3 − φ30 = a φ √ ln ρφ aγ3 −Γ3
κ3 γ3 − Γ3 30


+ ρ0 + ρφ a2(γ3 −Γ3 ) , (10.78)
30

and finally the potential

 −2 γ
Γ
3
3 −Γ3
2 − Γ3 1
V3 = ρφ eC3 (φ3 −φ30 ) − ρ0 e−C3 (φ3 −φ30 ) ,
2 30 2 ρφ
30

(10.79)
√
κ3
where it as been introduced the constantC3 = a φ √ (γ3 − Γ3 ). Incidentally,
Γ3
the scale factor a in terms of φ3 amounts to

1
a(γ3 −Γ3 ) = eC3 (φ3 −φ30 ) − ρ0 e−C3 (φ3 −φ30 ) . (10.80)
2 ρφ
30

By the way, the simplest solutions, expressible in terms of elementary

functions, arise for γ3 = Γ3 , and γ3 = 2Γ3 , and amount respectively to

aγ3 = a γ3 κ3 ρ30 + ρφ (t − t0 ), (10.81)
30

 
1 γ32 2
aγ3
=z= ρ κ (t − t0 )2 − ρ30 , (10.82)
ρφ 4 φ30 3
30

the first scale factor corresponds to the (2+1) power law solution, and the last
scale factor gives rise to the (2+1) Barrow–Saich solution.
134 Dilaton–Inflaton Friedmann–Robertson–Walker Cosmologies

10.3.3 Barrow–Saich Solution; γ = 2 Γ

From the general solutions given above, one can reproduce the Barrow and Saich
(1993) solution arising for the particular branch γ4 = 2Γ4 . Alternatively, one may
use for the scale factor the expression (10.70). In the Barrow–Saich solution the
matter and scalar field perfect fluids fulfill the equation of state: p4 + ρ4 =
γ
γ4 ρ4 , pφ + ρφ = 24 ρφ . The energy density conservation equations give ρ4 =
4 4 4
A4 a−2γ4 and ρφ = Aφ a−γ4 , where A4 , and Aφ are constants of integration.
4 4 4
The expressions for the scale factor, scalar field and its potential are
 3κ 2/(3γ4 )
4
16 γ42 A2φ4 (t − t0 )2 − A4
a(t) = , (10.83)
Aφ4

  
2 2 3γ4 κ4
φ4 (t) − φ40 = √ ln A (t − t0 )
3 γ4 κ4 4 φ4 3

3γ42
+ κ A2 (t − t0 )2 − A4 , (10.84)
16 4 φ4

√
e 3κ4 γ4 /2(φ4 −φ40 )
V (φ4 ) = (4 − γ4 )Aφ4  √ 2
2 . (10.85)
e 3κ4 γ4 /2(φ4 −φ40 ) − A4

Using now the correspondence (10.19) and the relation (10.31), accompanied by
the changes ρ4 → ρ3 , and ρφ → ρφ , one obtains the following (2+1) spatially
40 30
flat FRW Barrow–Saich counterpart:
κ 1/γ3
3
4 γ32 A2φ3 (t − t0 )2 − A3
a(t) = , (10.86)
Aφ3


2 γ √
φ3 (t) − φ30 = ln 3 Aφ3 κ3 (t − t0 )
γ3 κ3 2

γ32
+ κ A2 (t − t0 )2 − A3 , (10.87)
4 3 φ3

√
e 2κ3 γ3 (φ3 −φ30 )
V (φ3 ) = (4 − γ3 )Aφ3  √ 2
2 . (10.88)
e 2κ3 γ3 (φ3 −φ30 ) − A3

As far as the interpretation of these solutions is concerned, we direct readers to

the original Barrow–Saich publication, where the inflationary behavior is clearly
exhibited.
 10.4 Single Scalar Field Spatially Flat FRW Solutions 135

10.4 Single Scalar Field Spatially Flat FRW Solutions to

pφ + ρφ = Γ ρφ β
In this section we derive scalar field solutions to nonlinear state equations of
the form pφ + ρφ = Γ ρφ β , for any value of the parameter β except for β = 1
– the linear case. In particular, for β = 1/2 one recognizes the (2+1) Barrow–
Burd–Lancaster, Barrow et al. (1986), and the (3+1) Madsen (1986) solutions.
Cosmological constants are set equal to zero.

10.4.1 Spatially Flat (3+1) Solutions with

V (φ) = A(αφ2/(1−β) − φ2β/(1−β) )
For the derivation of solutions of this kind we consider the following state
equation
pφ = Γ4 ρβφ − ρφ . (10.89)
4 4 4

Consequently, (10.8) becomes

d 3
ρφ + Γ4 ρβφ = 0, (10.90)
da 4 a 4

which has, β = 1, the general integral

1
a(t) = a0 exp (− ρ(1−β) ). (10.91)
3Γ4 (1 − β) φ4
The equation for a, from (10.7) for k = 0, reads
ȧ
= a κ4 /3ρ1/2 , (10.92)
a φ
4

therefore, substituting a from (10.91), one arrives at

1 d
− ρφ = a κ4 /3ρ1/2 , (10.93)
3Γ4 ρβφ4 dt 4
φ
4

which possesses, for β = 1/2, the general solution

1/(1/2−β)
ρφ (t) = ρφ − 3a Γ4 κ4 /3(1/2 − β) t . (10.94)
4 40

On the other hand, substituting the above expression into (10.91), one obtains
a(t), namely
 (1−β)
1 2 (1−2β)
a(t) = a0 exp − ρφ − 3a Γ4 κ4 /3(1/2 − β) t . (10.95)
3Γ4 (1 − β) 40

√
The equation to determine φ4 (t), φ̇ = φ pφ + ρφ , amounts to
dφ4 β/2
= φ Γ4 ρφ , (10.96)
dt 4
136 Dilaton–Inflaton Friedmann–Robertson–Walker Cosmologies

which can be written as

1
dφ4 = −φ a ρφ −(1+β)/2 dρφ , (10.97)
3Γ4 κ4 4 4

consequently
1
φ4 − φ40 = −2φ a ρφ (1−β)/2 . (10.98)
3Γ4 κ4 (1 − β) 4
Finally, the evaluation of V (φ4 ) gives
 2/(1−β)
1 2 1
V (φ4 ) = ρφ − φ˙4 = −a φ 3Γ4 κ4 (1 − β)(φ4 − φ40 )
4 2 2
 2β/(1−β)
Γ4 1
− −a φ 3Γ4 κ4 (1 − β)(φ4 − φ40 ) . (10.99)
2 2
Recall that these families of solutions arise for β = 1/2. The case β = 1/2 gives
rise to the (3+1) Madsen cosmology.

10.4.2 Spatially Flat (2+1) Solutions with

V (φ) = A(αφ2/(1−β) − φ2β/(1−β) )
For the nonlinear state equation pφ = Γ3 ρβφ − ρφ , in the case of β = 1 and
3 3 3
β = 1/2, the integrals of the dynamical equations are:
the scale factor can be given in the form
 
1
a(t) = a0 exp − ρ(1−β)
, (10.100)
2Γ3 (1 − β) φ3
the scalar field energy density is
 √ 1/(1/2−β)
ρφ (t) = ρφ − 2a Γ3 κ3 (1/2 − β) t , (10.101)
3 30

thus, substituting this last expression into the scale factor a one has alternatively
 (1−β)
1 √ 2 (1−2β)
a(t) = a0 exp − ρ − 2a Γ3 κ3 (1/2 − β) t . (10.102)
2Γ3 (1 − β) φ30

The scalar field amounts to

1
φ3 − φ30 = −φ a ρφ (1−β)/2 , (10.103)
Γ3 κ3 (1 − β) 3
finally, the evaluation of V (φ3 ) gives
1 2  2/(1−β)
V (φ3 ) = ρφ − φ˙3 = −a φ Γ3 κ3 (1 − β)(φ3 − φ30 )
3 2
Γ3  2β/(1−β)
− −a φ Γ3 κ3 (1 − β)(φ3 − φ30 ) . (10.104)
2
 10.4 Single Scalar Field Spatially Flat FRW Solutions 137

It should be mentioned that all these solutions are determined under the
condition β = 1/2 and β = 1. The case β = 1/2 yields the (2+1)
Barrow–Burd–Lancaster solution, which is treated in detail below.

10.4.3 Barrow–Burd–Lancaster (2+1) and Madsen (3+1) Solutions

The derivation of the solutions in the case of β = 1/2 for flat FRW spacetimes
subjected to the nonlinear state equations under consideration is described below.

Barrow–Burd–Lancaster (2+1) Solution

Barrow et al. (1986), (BBL), reported two exact solutions exhibiting the
evolution of cosmological models containing self-interacting scalar fields with
physically interesting potentials, in the zero-curvature FRW model. One of them
is derived in what follows, starting from the nonlinear state equation

pφ = Γ3 ρ1/2
φ
− ρφ , (10.105)
3 3 3

the (10.17) for ρφ amounts to

3

d 2
ρ + Γ3 ρ1/2 = 0. (10.106)
da φ3 a φ
3

Its general integral is given by

1 1/2
a = a0 exp (− ρ ). (10.107)
Γ3 φ3
From (10.16) for a, in general one has
ȧ √
= a κ3 ρ1/2 . (10.108)
a φ
3

Substituting this derivative of a into (10.106), one gets

1 d √
ρ 1/2 + a κ3 ρφ 1/2 = 0, (10.109)
Γ3 dt φ3 3

therefore, integrating one obtains

√
ρφ (t) = ρφ e−2a Γ3 κ3 t
. (10.110)
3 30

Using this expression in (10.107) one arrives at a(t), namely

1 1/2 −a Γ √κ t
a = a0 exp(− ρ e 3 3 ). (10.111)
Γ3 φ30
√
The equation for φ3 , φ̇ = φ pφ + ρφ , yields

φ̇3 = φ Γ3 ρ1/4
φ
, (10.112)
3
138 Dilaton–Inflaton Friedmann–Robertson–Walker Cosmologies

therefore, using (10.110), one obtains

1 1 √
φ3 − φ30 = −2φ a ρ1/4 exp (− a Γ3 κ3 t). (10.113)
κ 3 Γ3 φ
30 2

On the other hand, replacing

1
ρ1/4 = − (φ3 − φ30 )a φ Γ3 κ 3 , (10.114)
φ
3 2
into V3 = ρφ − φ̇23 /2, one gets
3

 
Γ2 κ 1
V3 = 3 3 κ (φ − φ30 ) − (φ3 − φ30 ) .
4 2
(10.115)
8 2 3 3
To identify the derived solution with the original Barrow–Burd–Lancaster ex-
pressions – see Barrow et al. (1986) – one has to accomplish the following changes:
1
Γ3 −→ 8μ/κ3 , ρφ −→ μκ A4 , −a −→ a = ±1, (10.116)
30 2 3
in this manner one obtains:
κ3 2 a √8μt
a = a0 exp (− A e ),
4
1 √
ρφ (t) = μκ3 A4 e2a 8μ t ,
3 2 √
φ3 − φ30 = φ a Aea 8μ t/2 ,
 
1
V3 = μ κ3 (φ3 − φ30 )4 − (φ3 − φ30 )2 , (10.117)
2
where a = ±1, and φ30 is an integration constant.

Madsen (3+1) Solution

In this subsection we shall give a four-dimensional cosmological model ful-
filling nonlinear state equation, pφ = Γ4 ρ1/2
φ
− ρφ , which has been found
4 4 4
by Madsen (1986). This inflationary solution admits symmetry breaking and
is determined by:
the scale factor
2 1/2 −3/2a Γ4 √κ4 /3 t
a = a0 exp(− ρ e ). (10.118)
3Γ4 φ40
the scalar field energy density
√
ρφ (t) = ρφ e−3a Γ4 κ4 /3t
, (10.119)
4 40

thus, a allows for the alternative representation

2 1/2
a = a0 exp (− ρ ). (10.120)
3Γ4 φ4
 10.5 Scalar Field Solutions for a Given Scale Factor 139

The scalar potential can be expressed as

4 1 3
φ4 − φ40 = − φ a ρ1/4 exp (− a Γ4 κ4 /3 t), (10.121)
3 Γ4 κ4 /3 φ
40 4

while the potential can be brought to the form

 
9 2 κ4 9 κ4
V4 = Γ (φ − φ40 ) − (φ4 − φ40 ) .
4 2
(10.122)
32 4 3 8 3 4
A more suitable representation of the Madsen’s solution can be achieved by
accomplishing the following changes
√
2 8μ 1
Γ4 −→ , ρφ −→ μκ4 A4 , −a −→ a = ±1; (10.123)
3 κ4 /3 40 6

in this way one obtains:

1 κ4 2 a √8μt
a = a0 exp (− A e ),
4 3
μ κ4 4 2a √8μ t
ρφ (t) = A e ,
4 2 3

2 √
φ4 − φ40 = φ a Aea 8μ t/2 ,
3
 
9 κ4
V4 = μ (φ − φ40 ) − (φ4 − φ40 ) ,
4 2
(10.124)
8 3 4
where a = ±1, and φ40 is an integration constant.
It is really easy to establish the relationship existing between the BBL and
Madsen solutions by means of the correspondence (10.19).

10.5 Scalar Field Solutions for a Given Scale Factor

In this section we present solutions for a scalar field alone, namely the second
Barrow–Burd–Lancaster cosmology – see Barrow et al. (1986) – and its (3+1)
generalization, such that a simple state equation of the form pφ + ρφ = Γ F (ρφ )
is difficult to establish. In order to reach our goal, we consider that the scale
factor a(t) is a known function of the argument t, and proceed to evaluate ρφ ,
next to integrate φ, and finally evaluate V (φ).

10.5.1 Second (2+1) BBL Solution

The approach we shall apply to derive the second BBL solution assumes the
function a(t) as a given one, namely

2 A
a(t) = t 1 + 3 . (10.125)
t
140 Dilaton–Inflaton Friedmann–Robertson–Walker Cosmologies

From (10.16), one evaluates ρφ ,

3

1 ȧ2 1 (A + 4t3 )2
ρφ = = . (10.126)
3 κ3 a2 4κ3 t2 (A + t3 )2
2
Further, since (10.18) reads ρ̇φ + 2(pφ + ρφ )ȧ/a = 0, and φ˙3 = pφ + ρφ ,
3 3 3 3 3
hence the equation for φ3 to be integrated is
ȧ 2
ρφ˙ + 2 φ˙3 = 0. (10.127)
3 a
Substituting the derivative of ρφ from (10.126), one obtains
3

(2t3 − A)
φ˙3 = √ , (10.128)
2κ3 t(A + t3 )
which has the following integral
 
1 A + t3
φ3 = √ ln C0 . (10.129)
2κ3 t
2
The evaluation of V (φ) = ρφ − φ˙3 /2 yields
3

3 t 3 √
V (φ3 ) = 3
= C0 e− 2κ3 φ3 . (10.130)
κ3 A + t κ3
Summarizing, the second (2+1) inflationary BBL solution is determined by

2 A
a(t) = t 1 + 3 , (10.131)
t
 
1 A + t3
φ3 = √ ln C0 , (10.132)
2κ3 t

3 t 3 √
− φ3 2κ3
V (φ3 ) = = C0 e , (10.133)
κ3 A + t3 κ3
where A and C0 are constants.
To get an insight into the form of the conventional state equation, i.e., on the
dependence of F (ρ) on ρ, one expresses t in terms of ρ := 4κ3 ρφ by solving
3
(10.126) with respect to t, which yields
  12
√ 1  16 − 2 ρ3/2
A
t ρ=1+ 4 + ρ1/2 A1/3 Δ + t
8 −ρ A Δ+
1/2 1/3
,
2 2 4 + ρ1/2 A1/3 Δ
(10.134)
 1/3
where Δ := Aρ3/2 − 16 . On the other hand, since
2 1
φ˙3 = pφ + ρφ = Γ F (ρφ ) = (2t3 − A)2 /(t2 (A + t3 )2 ),
3 3 3 2κ3
substituting t from (10.134), one obtains a very involved function F on ρ.
 10.5 Scalar Field Solutions for a Given Scale Factor 141

10.5.2 (3+1) Generalization of the Second (2+1) BBL Solution

Using the relations (10.19) together with (10.31) we obtain the following (3+1)
flat FRW cosmology:

2 A
a(t) = t 1 + 3 , (10.135)
t
 
1 A + t3
φ4 = √ ln C0 , (10.136)
κ4 t

1 40t6 + 32At3 + A2
V (φ4 ) = , (10.137)
4κ4 t2 (t3 + A)2
The expression of V (φ4 ) in terms of φ4 , which is very involved, can be achieved
by substituting the roots of t in terms of φ4 from (10.136) into (10.137).
 11
Einstein–Maxwell Solutions

The purpose of this chapter is to provide a new approach on the search of electro-
magnetic–gravitational solutions to the Einstein–Maxwell fields of the (2 + 1)
gravity in the presence of a cosmological constant, allowing for stationary and
cyclic symmetries, establishing their relationship with known current solutions,
and to point out the families allowing for black hole interpretation. The search
and interpretation of this kind of solution has been the goal and realm of several
authors’ investigations, starting from quite different perspectives and using a
variety of approaches, which have sometimes brought about duplication of results
and efforts. Consequently, the completeness of the electromagnetic classes of
stationary cyclic symmetric solutions under consideration will be demonstrated
via straightforward integration of the field equations. A full characterization
of the physical content of these solutions would require considerable work; for
this reason, some short related comments are made close to those contained in
the relevant references, if there are any, and also about newly discovered families
with special emphasis on their black hole feature. Nevertheless, a full geometrical
characterization based on the algebraic classification of the physical tensors has
been produced.
From a general metric for stationary cyclic symmetric gravitational fields cou-
pled to Maxwell electromagnetic fields within the (2 + 1)-dimensional gravity
the uniqueness of wide families of exact solutions is established, including all
uniform electromagnetic solutions possessing electromagnetic fields with vanish-
ing covariant derivatives, all fields having constant electromagnetic invariants
Fμν F μν and Tμν T μν , the whole classes of hybrid electromagnetic solutions, and
also wide classes of stationary solutions, derived for third-order nonlinear key
equations. Certain of these families can be thought of as black hole solutions.
For the most general set of Einstein–Maxwell equations, reducible to three non-
linear equations for the three unknown functions, two new classes of solutions
– having anti-de Sitter spinning metric limits – are derived. The relationship
of various families with those reported by different authors’ solutions has been
 11.1 Stationary Cyclic Symmetric Einstein–Maxwell Fields 143

established. Among the classes of solutions with cosmological constant are: the
electrostatic Peldan solution, the magnetostatic Peldan metric, the stationary
uniform and spinning Clément classes, the constant electromagnetic invariant
branches with the particular Kamata–Koikawa solution, the Ayón–Cataldo–
Garcı́a hybrid cyclic symmetric stationary black hole fields, and the no less
important solutions generated via SL(2, R) transformations where the Clément
spinning charged solution, the Martı́nez–Teitelboim–Zanelli black hole solution,
and Dias–Lemos metric merit mention.

11.1 Stationary Cyclic Symmetric Einstein–Maxwell Fields

11.1.1 Stationary Cyclic Symmetric Maxwell Fields
To begin with, we consider a stationary cyclic symmetric spacetime with sig-
nature (−, +, +), i.e., a space endowed with stationary symmetry k = ∂t ,
k · k < 0, such that £k g = 0, and cyclic symmetry m = ∂φ , m · m > 0,
such that £m g = 0, with closed integral curves from 0 to 2π, which in turn
commute [k, m] = 0. Hence the Killing vector fields k and m generate the group
SO(2) × R. The electromagnetic field, described by the antisymmetric tensor
field F = 12 Fμν dxμ ∧ dxν , is assumed to be stationary cyclic symmetric, i.e.,
£k F = 0 = £m F . It should be pointed out that, in contrast to the general (3+1)
stationary cyclic symmetric spacetime, any (2 + 1) stationary cyclic symmetric
spacetime is necessarily circular, i.e., the circularity conditions

k ∧ m ∧ dk = 0 = k ∧ m ∧ dm (11.1)

are identically fulfilled because of their 4-form character and hence there exists
the discrete symmetry when simultaneously t → −t and φ → −φ. One may find
a coordinate system such that the metric tensor components g(k dr) = 0 and
g(m dr) = 0, where the coordinate direction dr is orthogonal to the surface
spanned by k ∧ m. Commonly one introduces the coordinate system {t, φ, r} in
(2+1)-dimensional gravity .
The main goal of this section is to demonstrate of the following theorem:

Theorem 11.1 The general form of stationary cyclic symmetric electromag-

netic fields in (2+1) dimensions is given by
grr
∗ F = adt + bdφ + c √ dr, (11.2)
−g

where the constants a, b and c are subjected, by virtue of the Ricci circularity
conditions, to the equations

a c = 0 = b c, (11.3)
144 Einstein–Maxwell Solutions

which gives rise to two disjoint branches

grr
c = 0, ∗F = c √ dr, (11.4)
−g
and
c = 0, ∗F = adt + bdφ, (11.5)

with its own sub-classes a = 0 or b = 0.

Proof To establish that the field ∗F possesses the form given by (11.2) one uses
the source-free Maxwell equations

dF = 0 = d ∗ F , (11.6)

where ∗ denotes the Hodge star operation. Let us evaluate the exterior derivative
of the t-component ∗F (k) of ∗F ,

d∗F (k) = d ik ∗ F = £k ∗ F − ik d ∗ F = 0 − 0 → ∗F (k) =: a = const.,

(11.7)

the first zero arises from the stationary character of the field F , while the second
one corresponds to the Maxwell equation. Similarly, for the φ–component ∗F (m)
one has

d∗F (m) = d im ∗ F = £m ∗ F − im d ∗ F = 0 → ∗F (m) =: b = const.

(11.8)

In this manner we have established that the t and φ components of the dual field
∗F are constants given correspondingly by a and b. The component of ∗F along
the vector direction ∂r remains to be determined. For this purpose, consider the
tφ-component F (k, m) of the field F , which can be expressed as F (k, m) =
im ik F = (−im ik ∗ ∗F = im ∗ (k ∧ ∗F ) = ∗(m ∧ k ∧ ∗F )) = − ∗ F (∗(k ∧ m)),
thus its derivative yields

dF (k, m) = d(im ik F ) = d im (ik F )

= (£m − im d)(ik F ) = ik £m F + i[k,m] F − ik (£m − ik d)F
= 0 → F (k, m) =: c = const.. (11.9)

Since the constant c can be written as c = − ∗ F (∗(k ∧ m)), to determine it, one
evaluates ∗(k ∧ m). Identifying the Killing vectors accordingly with k = ∂t and
m = ∂φ , then
√ √
∗ (k ∧ m) = − −gdr = − −g g rr ∂r , (11.10)

thus
√ √
c = − ∗ F (− −g g rr ∂r ) = −g g rr ∗ F (∂r ). (11.11)
 11.1 Stationary Cyclic Symmetric Einstein–Maxwell Fields 145

Conversely, from the above mentioned relation one determines the r component
of the field ∗F , namely ∗F (∂r ) = √−g
c
grr . In this manner, the structure of F ,
explicitly given by (11.2), has been established.
The vanishing conditions (11.3) straightforwardly arise from the Ricci circu-
larity conditions m ∧ k ∧ R(k) = 0 and k ∧ m ∧ R(m) = 0. Correspondingly,
the vanishing conditions a c = 0 = b c can be established immediately, as we
shall see in the next section, from the Einstein equations Rμν − 12 R gμν =
κTμν − Λ gμν , where the electromagnetic energy–momentum tensor compo-
nents are defined through the electromagnetic field Fμν = −Fνμ as 4π Tμν =
Fμσ Fν σ − 14 gμν Fασ F ασ .

A first formulation of this theorem with an outline of its demonstration has

been reported by Ayón, Cataldo and Garcı́a in Ayón–Beato et al. (2005).

11.1.2 General Stationary Metric and Einstein Equations

In general, in (2 + 1)-dimensional gravity any stationary cyclic symmetric metric
can be given as

g = gtt dt2 + 2gtφ dt dφ + gφφ dφ2 + grr dr2 . (11.12)

When a Maxwell electromagnetic field is present, the field tensor, as we

established previously, possesses the structure
⎡ ⎤
0 b −√
cgrr
−g
 αβ  1 ⎢ ⎢
⎥
⎥
F =√ ⎢ −b 0 a ⎥, (11.13)
−g ⎣ ⎦
cg
√ rr
−g
−a 0

where g := det(gμν ), which makes apparent the fulfillment of the divergence

equation

( − det(gμν )F αβ );β = 0

for constants a, b, and c. The Maxwell electromagnetic energy–momentum tensor

is given as usual as
1 1
Tμ ν = (Fμσ F νσ − δμν Fτ σ F τ σ ). (11.14)
4π 4
Without loss of generality, one can choose the coordinates for a stationary cyclic
symmetric (2+1) metric, developed with respect to the cyclic symmetry m = ∂φ ,
in such a way that it becomes

F (r) 2 dr2 2
g=− dt + + H(r) [dφ + W (r)dt] . (11.15)
H(r) F (r)
146 Einstein–Maxwell Solutions

On the other hand, if one chooses the stationary symmetry k = ∂t as the

fundamental Killing field, the stationary cyclic symmetric (2 + 1) metric can be
written as

F (r) 2 dr2
g=− [dt − ω(r) dφ] + h(r) dφ2 + ,
h(r) F (r)
F H ω HF
F = F, H = h − ω 2 , W = ,h = . (11.16)
h F h F − W 2 H2

Mostly we will use the metric (11.15) in the forthcoming developments, but
occasionally the metric representation (11.16) will be used . When doing so,
the derived expressions will be given in terms of the set {F (r), h(r), ω(r)} of
structural functions. Omitting the dependence of the structural functions on the
variable r, the Maxwell electromagnetic field contravariant tensor is given by
⎡ ⎤
0 b − Fc
⎢ ⎥
(F μν ) = ⎢
⎣ −b 0 a ⎥,
⎦ (11.17)
c
F −a 0

where a, b, and c are constants related with the character of the field. For
instance, if only b is different from zero, while a and c vanish, the field is called
(pure) electric field. When a = 0, b = 0 = c, one deals with a pure magnetic field;
other possibilities do not receive a particular name. The covariant components
Fμν of the field tensor are given by:

Ftr = −b/H − W H(a − b W )/F, Ftφ = c, Frφ = H(a − b W )/F. (11.18)

The electromagnetic field quadratic invariant F F := Fμν F μν is given by

2
c2 H (a − W b) b2
F F = −2 +2 −2 . (11.19)
F F H
Notice that if one uses the vector-potential description of the electromagnetic
field
1
F := Fμν dxμ ∧ dxν = d (Aμ dxμ ) =: d A
2
(11.20)

one would have

 r  r
1 H 2 H H H
F =d [( − W )b + W a]dr × dt + d [− W b + a]dr × dφ
H F F F F
 
1
+d c(tdφ − φdt) = d A. (11.21)
2
 11.1 Stationary Cyclic Symmetric Einstein–Maxwell Fields 147

The energy–momentum tensor matrix (Tνμ ) associated with the metric (11.15) is
⎡ 2 2 2 ⎤
b (H W −F )−a2 H 2 −c2 H a[b(H 2 W 2 −F )−aH 2 W ]
ac
−
⎢ 8π F H 4π 4π F H ⎥
⎢ ⎥
⎢ cH(W b−a) −b F +H (W b−a) +c H
2 2 2 2 c[b(H W 2 −F )−aH 2 W ]
2
⎥
⎢ − ⎥
⎣ 4 πF 2 8π F H 4π F 2 H ⎦
bH(W b−a) b (H W −F )−a H +c H
2 2 2 2 2 2

4 πF
bc
4π − 8π F H
(11.22)
and possesses the trace T := Tμμ given by
2
1 c2 1 H (a − W b) 1 b2 1
T =− + − = F F, (11.23)
8π F 8π F 8π H 16π
and the electromagnetic energy momentum quadratic invariant
2
μν3 H 2 (a − b W )2 − b2 F − c2 H 3
T T = Tμν T = = F F 2 . (11.24)
64π 2 F 2H 2 256π 2
The Einstein–Maxwell equations
R
Eμν := Rμν − gμν + Λgμν − 8π Tμν = 0 (11.25)
2
for a negative cosmological constant Λ = −1/l2 explicitly read:
1 H,r,r 1 H,r F,r 1 F 1
Et t = F + − 2
H,r 2 + H 2 W W,r,r + HW W,r H,r
2 H 4 H 4H 2
1 2 F − H 2W 2 c2 a2 H 1
+ H W,r 2 + b2 + + − 2, (11.26a)
4 FH F F l
Et r = −2 ca, (11.26b)
1 H,r,r H,r F,r HW
Et Φ = W F,r,r − F W −W − 2 a2
2 H H F
 
F −H W 2 2
1  H,r
− 2ab − 2
F +H W 2
W,r,r + 2W,r
FH 2 H
H,r 2
+F W − H 2 W W,r 2 , (11.26c)
H2
H
Er t = −2c (W b − a) , (11.26d)
F2
1 H,r F,r 1 H,r 2 1 b2
Er r = − F 2 + H 2 W,r 2 +
4 H 4 H 4 H
c2 H 1
− − (bW − a)2 − 2 , (11.26e)
F F l
F − H 2W 2 HW
Er Φ = −2 c b 2
− 2ca 2 , (11.26f)
F H F
1 2 H
EΦ t = H W,r,r + HW,r H,r − 2 b (W b − a) , (11.26g)
2 F
148 Einstein–Maxwell Solutions

EΦ r = −2 bc, (11.26h)

1 1 1 H,r,r 3 H,r F,r 3 H,r 2 1

EΦ Φ = − + F ,r,r − F − + F − H 2 W W,r,r
l2 2 2 H 4 H 4 H2 2
3 2 F − H 2W 2 c2 a2 H
− H W W,r H,r − H W,r 2 − b2 + − . (11.26i)
4 FH F F
The vanishing of Et r and EΦ r yields respectively ac = 0 = bc. Therefore one can
distinguish the branches:
c = 0 with a and b, not vanishing simultaneously; and
c = 0 with a and b vanishing simultaneously.
In the forthcoming sections we shall deal with the integration and characteriza-
tion of each branch starting from the simplest static solutions.

11.1.3 Complex Extension and Real Cuts

It would be of some interest to add some lines about the complex extension of
the metric under consideration. Accomplishing in the metric (11.15) the complex
transformations
t → i Φ, φ → −i T , (11.27)

one arrives at
 
F dr2
gc = − HW 2 dΦ2 + − Hd T 2 + 2H W d T dΦ, (11.28)
H F
which can be brought to the form
F 2 dr2 2
gc = − dT + + H (dΦ + W dT ) , (11.29)
H F
accompanied by the identification
F HW
F = F, H = − HW 2 , W = . (11.30)
H H
At the level of the field tensor F μν one has
⎡ ⎤ ⎡ ⎤
0 B − FC 0 −i a c
F
⎢ ⎥ ⎢ ⎥
(F μν ) = ⎢
⎣ −B 0 A ⎥ ⎢
⎦ = ⎣ ia 0 ib ⎥
⎦, (11.31)
C
F −A 0 − Fc −ib 0
thus the following correspondence for the field constants arises

−i a → B, i b → A, −c → C. (11.32)

Summarizing, one may say that the role of the Killingian coordinates has been
interchanged: the timelike coordinate t becomes the spacelike Φ–coordinate,
 11.1 Stationary Cyclic Symmetric Einstein–Maxwell Fields 149

while the cyclic φ-coordinate becomes the new time-coordinate T . Correspond-

ingly, one has to think of the tensor components of the participating quantities
from this perspective. This procedure can be used to determine new classes of
solutions from known ones. For instance, one can generate magnetic solutions
from electric ones. The relations arising from this kind of complex transfor-
mations have been called “duality mapping” by Cataldo and Salgado (1996),
although strictly there is no electric–magnetic duality in (2+1) dimensions.

11.1.4 Positive Λ Solutions

For completeness and to avoid duplication of works, it is worth noticing that
solutions for positive cosmological constant are easily obtainable from the anti-
de Sitter (Λ = −1/l2 ) ones; first, notice that the Einstein equations for any
sign of a cosmological constant Λ – positive or negative – are recovered from
equations (11.26) simply by replacing −1/l2 → Λ. Next, having at disposal a
concrete solution of the Einstein equations mentioned above (11.26), by replacing
there l2 by − l2 , one determines the corresponding metric structure for positive
cosmological constant Λ = 1/l2 . This replacement is equivalent to accomplishing
the complex change l → i l in the Λ < 0 solution, nevertheless one ought to take
care of possible additional arrangements of constants, if any, and also of possible
changes in the signature.

11.1.5 Characterizations of Einstein–Maxwell Solutions

The physical characterization is based on the determination of the local and
global energy–momentum–mass quantities using the Brown–York approach
(Brown and York Jr., 1993). As far as the algebraic–geometrical character-
ization is concerned, the eigenvalue problem for the electromagnetic field,
energy–momentum and Cotton tensors is solved and their types are established.
The general form of electromagnetic fields for stationary cyclic symmetric
√
(2+1) spacetimes is given by: ∗F = adt + bdφ + cgrr / −gdr, which splits
into various sub-families: the electric b = 0 fields, the magnetic a = 0 fields,
the uniform fields characterized by the vanishing of the covariant derivatives
Fαβ;γ = 0, the class of stationary fields with constant invariant Fμν F μν , and
consequently, due to the structure of the electromagnetic fields with constant
energy–momentum tensor invariants, the gravitational stationary cyclic solutions
√
for the hybrid electromagnetic field ∗F = cgrr / −gdr; the explicit derivation
of the solutions belonging to the quoted branches can be found in Garcı́a (2009),
and the evaluation of quasi-local energy–momentum quantities has been done in
Garcia–Diaz (2013).
The families of Einstein–Maxwell solutions to be considered are: all uni-
form electromagnetic solutions possessing electromagnetic fields with vanishing
150 Einstein–Maxwell Solutions

covariant derivatives (stationary uniform and spinning Clément classes), all fields
having constant electromagnetic field and energy–momentum tensors’ invariants
(Kamata–Koikawa solution), the whole classes of hybrid electromagnetic Ayon–
Cataldo–Garcı́a solutions, a new family of stationary electromagnetic solutions,
the electrostatic and magnetostatic solutions with Peldan limit, the Clément
spinning charged metric, the Martı́nez–Teitelboim–Zanelli black hole solution,
and Dias–Lemos electromagnetic solution.
The application of the Hayward black hole dynamics formulation, Hayward
(2008), and the Ashtekar isolated horizon approach, Ashtekar and Krishnan
(2004), to the static and stationary black hole solutions reported here is
straightforward.

Cotton Tensor Algebraic Classification

In (n + 1)-dimensional spacetimes, for n ≥ 3, the invariant decomposition of
the Riemannian curvature tensor gives rise to the conformal Weyl tensor, the
traceless Ricci tensor, and the scalar curvature; for the classification of gravity
one classifies the Weyl tensor, and the classification of matter is achieved through
the classification of the traceless Ricci tensor. For details, in (3 + 1)-dimensional
spacetimes; see, for instance, the book by Stephani et al. (2003), and in the
present book, Chapter 20.
In (2 + 1)-dimensional spacetimes there is no room for the conformal Weyl
tensor, the Riemannian curvature tensor decomposes into the Ricci tensor, and
the scalar curvature. The role of the conformal tensor in (2+1) gravity is played
by the Cotton tensor – see Stephani et al. (2003) and Chapter 20 – which is
defined by means of the Ricci tensor and the scalar curvature through their
covariant derivatives
1
C αβ = αγδ (Rβ γ − R δ β γ );δ , C α α = 0. (11.33)
4
For the standard stationary (static) cyclic symmetric metric

ds2 = −N 2 dt2 + L−2 dr2 + K 2 [dφ + W dt]2 ,

the traceless Cotton tensor, in the form C α β , is

⎡ 1 ⎤
C 1 0 C 13
⎢ ⎥
C := (C α β ) = ⎢
⎣ 0 C 22 0 ⎥ 1 2 3
⎦ ; C 1 + C 2 + C 3 = 0. (11.34)
C 31 0 C 33

Determining the eigenvalues and eigenvectors of the Cotton matrix (11.34) one
establishes the algebraic Cotton type of the spacetime one is dealing with.
Accordingly, the characteristic equation for the eigenvalue λ amounts to
 
(C 2 2 − λ) (C 1 1 − λ)(C 3 3 − λ) − C 3 1 C 1 3 = 0, (11.35)
 11.1 Stationary Cyclic Symmetric Einstein–Maxwell Fields 151

or, in terms of its solutions, as

 
1 2 1
(C 2 −λ) λ + C 2 +
2 2 1 2 3 1
(C 2 + 2 C 1 ) + 4 C 1 C 3
2 2
 
1 2 1
× λ+ C 2− 2 1 2 3 1
(C 2 + 2 C 1 ) + 4 C 1 C 3 = 0, (11.36)
2 2
while the eigenvector equations are
(C 1 1 − λ)V 1 + C 1 3 V 3 = 0,
(C 2 2 − λ)V 2 = 0,
C 3 1 V 1 + (C 3 3 − λ)V 3 = 0. (11.37)
For each eigenvalue the corresponding solution is:
λ1 = C 2 2 , V1 = (0, V 2 , 0),
1 1
λ2 = − C 2 2 + (C 2 2 + 2 C 1 1 )2 + 4 C 3 1 C 1 3 ,
2 2
C 13
V2 = (− 1 V 3 , 0, V 3 ),
C 1 − λ2
1 1
λ3 = − C 2 2 − (C 2 2 + 2 C 1 1 )2 + 4 C 3 1 C 1 3 ,
2 2
C 13
V3 = (− 1 V 3 , 0, V 3 ). (11.38)
C 1 − λ3
Thus, the eigenvector V1 – a real one – is oriented in the ρ-direction, the remain-
ing two vectors V2 and V3 might be real vectors lying on the surface spanned
by the t and φ coordinate directions or complex eigenvectors depending, corre-
spondingly, upon whether the value of the radical (C 2 2 + 2 C 1 1 )2 + 4 C 3 1 C 1 3 is
positive or negative.
The nomenclature to be used for eigenvectors and algebraic types of tensors is
borrowed from Plebański (1964); see also Plebański (1967), Chapter VI: timelike,
spacelike, null, and complex vectors are denoted respectively by T, S, N, and Z.
For algebraic types are used the symbols:
{λ1 T, λ2 S2 , λ3 S3 } ≡ {T, S, S,}
meaning that the first real eigenvalue λ1 gives raise to a timelike eigenvector
T, the second real eigenvalue λ2 is associated with a spacelike eigenvector S2 ,
finally the third real eigenvalue λ3 is related to a spacelike eigenvector S3 ; for
the sake of simplicity I use the typing {T, S, S}. It is clear that {N, N, S} stands
for the algebraic type allowing for two different real eigenvalues giving rise to
two null eigenvectors while the third real root is associated with a spacelike
eigenvector. When there is a single and a double real eigenvalues giving rise
correspondingly to a timelike and spacelike eigenvector, the algebraic type is
denoted by {T, 2 S}; consequently, for a triple real eigenvalue, if that were the
case, the types could be {3T }, {3N }, or {3S}. For a complex eigenvalue λZ ,
152 Einstein–Maxwell Solutions

in general, the related eigenvectors occur to be complex and are denoted by Z

and Z̄ its complex conjugated, the possible types are {T, Z, Z̄}, {N, Z, Z̄}, or
{S, Z, Z̄}.
In general, the spaces described by the stationary (static) cyclic symmetric
metric above belong to the Cotton type I; if the three eigenvectors are real
the type is IR , otherwise the type is IZ with eigenvectors S, N, T, Z, and Z̄.
Following the notation above-proposed, the algebraic type for the Cotton tensor
could be Type I: {S, S, S}, {S, Z, Z̄}, and so on.
As far as the Jordan form is concerned, one achieves it by subjecting the
matrix under consideration to (elementary) similarity transformations: adding
to a row another row multiplied by a suitable constant or function in order to
simplify the resulting row, or to add to a column another one multiplied by a
constant or function. To bring C, (11.34), the corresponding Jordan form, one
has to eliminate the components C 1 3 and C 3 1 , which is achieved by multiplying
C from the right by the matrix A, and from the left by the matrix B, namely
⎡ ⎤ ⎡ ⎤
1 0 −C 1 3 /C 1 1 1 0 1
⎢ ⎥ ⎢ ⎥
A=⎢ ⎣ 0 1 0 ⎥, B = ⎢
⎦ ⎣ 0 β 0 ⎥ ⎦,
0 0 1 −C 3 1 /C 1 1 0 1
⎡ 1
⎤
C 1 0 0
⎢ ⎥
BCA = ⎢
⎣ 0 βC 2 2 0 ⎥,
⎦
0 0 γC 3 3
C 13C 31 C 13C 31
γ =1− 1 3
, β := 1 + 1 2 , (11.39)
C 1C 3 C 1C 2
and finally by multiplying the resulting matrix by E = diag(1, 1/β, 1/γ), one
arrives at
⎡ ⎤
λ1 0 0
⎢ ⎥
(B C A)E ∼ JI = ⎢ ⎣ 0 λ2 0 ⎥.
⎦ (11.40)
0 0 −λ1 − λ2

An alternative treatment of the Cotton tensor and conformal symmetries for

(2+1)-dimensional spaces is given in Hall and Capocci (1999), and also in Garcı́a
et al. (2004), where the analysis on Cotton tensors in (n+1)-dimensions is also
developed.

11.1.6 Static Cyclic Symmetric Equations for Maxwell Fields

In the forthcoming three sections, we derive all the static solutions of the
Einstein–Maxwell equations (11.26); there are only three families within this
 11.2 Electrostatic Solutions; b = 0, a = 0 153

class. For the static metric W (r) = 0, consequently the metric (11.15) becomes

F (r) 2 dr2
g=− dt + + H(r)dφ2 , (11.41)
H(r) F (r)
and the Einstein–Maxwell equations simplify drastically:
1 H,r,r 1 H,r F,r 1 H,r 2 b2 c2 a2 H 1
Et t = F + − F 2 + + + − 2,
2 H 4 H 4 H H F F l
1 H,r F,r 1 H,r 2 b2 c2 a2 H 1
Er r = − F 2 + − − − 2,
4 H 4 H H F F l
1 1 H,r,r 3 H,r F,r 3 H,r 2 b2 c2 a2 H 1
EΦ Φ = F,r,r − F − + F − + − − 2,
2 2 H 4 H 4 H2 H F F l
1
Et r = −2 ca, Et Φ = −2a b ,
H
H 1
Er t = 2a c 2 , Er Φ = −2b c ,
F FH
H
EΦ t = 2a b , EΦ r = −2 bc, (11.42)
F
Each of these Eμ ν equations has to be equated to zero, therefore one can distin-
guish the following three families of static solutions:
the electric class: b = 0, a = 0, c = 0,
the magnetic class: a = 0, b = 0, c = 0,
the hybrid class: c = 0, a = 0, b = 0.
In the next sections we proceed to integrate each class separately.

11.2 Electrostatic Solutions; b = 0, a = 0

In this section, the electrostatic cyclic symmetric Einstein–Maxwell solutions
are derived in the presence of a negative cosmological constant Λ = −1/l2 . It
seems that static Einstein–Maxwell solutions with cosmological constant were
first derived in Peldan (1993); in that publication Peldan mentioned his failure
in finding any work done on explicit solutions to Einstein–Maxwell solutions with
cosmological constant, although static and rotationally symmetric solutions with
vanishing cosmological constant did exist, namely the Deser and Mazur (1985)
and Melvin (1986) solutions. Moreover, in the BTZ publication, Bañados et al.
(1992) also presented a charged rotating metric, which was assumed to be cor-
rect and to satisfy the Einstein–Maxwell equations. Nevertheless, according to
the comments by Kamata and Koikawa (1995), and later by Garcı́a (1999), it
happened to be wrong; the azimuthal Maxwell equation yields JQ/r3 = 0, hence
one could distinguish the rotating BTZ solution and the charged static solution
as two possible independent branches but not a single expression equipped with
two non-vanishing simultaneously parameters as a “solution” as it was erro-
neously assumed in the BTZ publication. To my mind it is reasonable to name
154 Einstein–Maxwell Solutions

the electrostatic cyclic symmetric Einstein–Maxwell solution with cosmological

constant “the charged static–Peldan solution.” The subclass of solutions without
Λ are explicitly given and identified, while the electrostatic Peldan solution with
Λ is here analyzed in detail.

11.2.1 General Electrostatic Solutions

The substraction Et t (a = 0 = c) − Er r (a = 0 = c) from (11.42) yields

d2
H (r) = 0 ⇒ H(r) = C0 + C1 r, (11.43)
dr2
where C0 , and C1 are constants of integration; C1 at this stage is assumed to be
different from zero; the zero case deserves special attention and will be treated
separately. Substituting this structural function H into the equation Et t (a =
0 = c) one arrives at a first-order differential equation for F
 2
H,r H,r b2 4
F,r − F + 4 − 2 = 0, (11.44)
H H H l
which by introducing an auxiliary function f (r) through

F (r) = H(r) f (r) = (C0 + C1 r) f (r),

reduces to the simple equation

df (r) 4 C1 r + C0 − b2 l2
= (11.45)
dr C1 l2 C0 + C1 r
with general integral
4  
f= K0 + C1 r − b2 l2 ln (C0 + C1 r) , (11.46)
C12 l2
where K0 is a new integration constant, into which of course one has incorpo-
rated C0 .
Summarizing, one arrives at the metric

F (r) 2 dr2
g=− dt + + h(r)dφ2 ,
h(r) F (r)
4
F (r) = 2 2 K0 + h(r) − b2 l2 ln h(r) h(r),
C1 l
h(r) = C1 r + C0 . (11.47)

This solution is characterized by:

the vector field
b
A = At dt = ln h dt, (11.48)
C1
 11.2 Electrostatic Solutions; b = 0, a = 0 155

the electromagnetic field tensors

b
F μν = 2bδ[t μ δr] ν , Fμν = −2 δ[μ t δν] r (11.49)
h(r)
with field invariant
b2
Fμν F μν = −2 , (11.50)
h
the energy momentum tensor
b2 1
Tμ ν = − δμ t δt ν + δμ r δr ν − δμ φ δφ ν , (11.51)
8π h
with quadratic energy momentum invariant and trace
3 b4 1 b2
Tμν T μν = 2 2
, Tμμ = − . (11.52)
64π h 8π h
A familiar representation of the above mentioned solution is achieved for the
choice C0 = 0, C1 = 2, K0 = b2 l2 ln 2r0 , which yields
 
2r r dr2
g = − 2 − b ln 2
dt2 + 2 rdφ2 +  ,
l r0 2 r 2l2r − b2 ln rr0
b r
A= ln dt. (11.53)
2 r0
This solution, endowed with mass, electric charge, a cosmological constant −1/l2 ,
and radial parameter, allows for a charged black hole interpretation. The mass
may assume positive as well as negative values, whereas the charge is not upper-
bounded.

11.2.2 Gott–Simon–Alpern, Deser–Mazur, and

Melvin Electrostatic Solution
According to the existing references, Gott, Simon, and Alpern were the first
authors to derive solutions within Maxwell theory in (2+1) gravity: Gott and
Alpert (1984); Gott et al. (1986); they found, among other things, the electro-
static solution without cosmological constant given by Eq. (42) in Gott et al.
(1986) in “Schwarzschild” coordinates.
Introducing in the above expressions, (11.47) and (11.48), the radial coordinate
ρ through C0 + C1 r → ρ2 together with t → C1 t/2, K0 → l2 k0 , and by letting
1/l2 → 0 one arrives at the electrostatic solution in the form
dρ2 κ 2 ρc
g = −F dt2 + + ρ2 dφ2 , F (ρ) = k0 − 2b2 ln ρ = Q ln ,
F 2π ρ
A = At dt = b ln ρ dt. (11.54)
Some authors refer to thecoordinate system {t, ρ, φ}, in which the perimeter of
2π
the circle equates 2π ρ = 0 ρ dφ, as to the “Schwarzschild” coordinates.
156 Einstein–Maxwell Solutions

Two years later, Deser and Mazur (1985) published their version of the elec-
trostatic solution for Λ = 0. Moreover, by then, the work by Melvin (1986) was
published with the derivation of the electrostatic as well as the magnetostatic
solutions for vanishing Λ.
Melvin (1986) introduced three kinds of coordinate systems, and defined K as
the electric charge, while κ = 8π G , M as the total mass, and the scale parameter
a = (1 − 4G M )2 /(2 K), among other quantities:
physical radial coordinates, in which the electrostatic solution amounts to
dS
ds2 = −N (ρ)2 dt2 + dρ2 + S(ρ)2 dθ2 , dρ = ,
(1 − 4G M )2 − 2 K ln S
N= (1 − 4G M )2 − 2 K ln S, (11.55a)

Schwarzschild radial coordinates, in which the electrostatic solution is given by

ds2 = −N (r)2 dt2 + ψ(r)2 dr2 + r2 dθ2 ,

1
N (r) = (1 − 4G M )2 − 2 K ln r, ψ(r) = , (11.55b)
N (r)
conformal radial coordinates, in which the electrostatic solution becomes
 
ds2 = −N (R)2 dt2 + Ψ(R)2 dR2 + R2 dθ2 ,
N (R) = (1 − 4G M ) − K ln R,
 
 K
Ψ(R) = exp −4G M ln R − (ln R) .
2
(11.55c)
2
Deser and Mazur (1985) reported the electrostatic solution, Eq. (12), in the
conformal radial coordinate system (11.55c), with the following identifications
√
{G , M, (1 − 4G M ), K, N, t}M → {G, m, α, G e2 , α N, t/ α}DM .

There is a sign missprint in the function grr of Deser and Mazur (1985) , Eq. (12);
in the factor in front of (ln r)2 should be −G e2 (ln r)2 .
Kogan (1992) reported and analyzed the (electro- and magneto-) static solu-
tions of the (2+1)-dimensional Einstein–Maxwell equations for both positive and
negative signs of the gravitational constant κ; recall that in the three dimensions
there is no restriction on its sign. The r-coordinate used there was such that
grr = 1 for the signature used in the present report, consequently, the solutions
are given in the (11.55a) representation.

11.2.3 Charged Static Peldan Solution with Λ

The Peldan electrostatic solution with cosmological constant in polar coordinates
arises from the general expressions above, (11.47) and (11.48), by means of the
coordinate and parameter changes C0 + C1 r → ρ2 , t → C1 t/2, K0 → −l2 m. In
this way one obtains
 11.2 Electrostatic Solutions; b = 0, a = 0 157

dρ2 ρ2
g = −F dt2 + + ρ2 dφ2 , F (ρ) = 2 − m − 2b2 ln ρ,
F l
A = b ln ρ dt. (11.56)

The corresponding field tensors are given by

b2 1 b
Tμ ν = − 2
δμ t δt ν + δμ ρ δρ ν − δμ φ δφ ν , Fμν = −2 δ[μ t δν] r , (11.57)
8π ρ ρ
To achieve the specific Peldan’s writing, one has to accomplish the additional
identifications m → −C1 , 1/l2 → −λ/2, b2 → q 2 /4, t → C2 t, ρ → r.
The electrostatic Peldan (1993) solution, Eq. (71) – see Eq. (4.15) in Garcı́a
(2009) – in canonical representation is given by the metric
1
ds2 = −N (ρ)2 dt2 + dρ2 + K(ρ)2 [dφ + W (ρ)dt]2 ,
L(ρ)2

ρ2
L(ρ) = N (ρ) = − 2b2 ln ρ − M , K(ρ) = ρ, W (ρ) = 0. (11.58)
l2

Mass and Energy

The surface energy density  is
1
(ρ, 0 ) = − N (ρ) − 0 . (11.59)
πρ
Consequently the global energy and mass are given by

E(ρ, 0 ) = −2N (ρ) − 2π ρ 0 ,

ρ2
M (ρ, 0 ) = −2 2 + 2m + 4b2 ln ρ − 2πρ N (ρ)0 . (11.60)
l
Thus, for the natural choice of a vanishing reference energy density 0 = 0, one
has at the spatial infinity ρ → ∞ that
1 lM l b2
(ρ → ∞, 0 = 0) ≈ − + + ln ρ,
π l 2π ρ2 π ρ2
ρ Ml l b2
E(ρ → ∞, 0 = 0) ≈ −2 + +2 ln ρ,
l ρ ρ
ρ2
M (ρ → ∞, 0 = 0) ≈ −2 2 + 2M + 4b2 ln ρ, (11.61)
l
while if the reference energy corresponds to the anti-de Sitter spacetime
2
AdS(M0 ), 0 = − π ρ ρl2 − M0 , 0|∞ (M0 ) ≈ − π1l + 2π
1 lM0
ρ2 , then the energies
and mass at spatial infinity are expressed as
M − M0 l b2
(ρ → ∞, 0|∞ (M0 )) ≈ l 2
+ ln ρ,
2π ρ π ρ2
M − M0 lb2
E(ρ → ∞, 0|∞ (M0 )) ≈ l +2 ln ρ,
ρ ρ
M (ρ → ∞, 0|∞ (M0 )) ≈ M − M0 + 2 b2 ln ρ. (11.62)
158 Einstein–Maxwell Solutions

Comparing with the static BTZ one recognizes M as the BTZ mass. Notice that
the energy and mass include an amount of energy due to the electric field through
the logarithmical term; because of this dependence, these quantities diverge at
infinity logarithmically.

Field, Energy–Momentum, and Cotton Tensors

The electromagnetic tensor field associated with the charged Peldan solution is
given by
⎡ b
⎤
0 L2 ρ 0
⎢ ⎥
⎢ 2 ⎥
(F α β ) = ⎢ b L 0 0 ⎥, (11.63)
⎣ ρ ⎦
0 0 0
Searching for its eigenvectors, one arrives at
2
λ1 = 0; V1 = [0, 0, V 3 ], Vμ V μ = ρ2 V 3 , V1 = S1,
b V2
λ2 = ; V2 = [ 2 , V 2 , 0], V μ Vμ = 0, V2 = N2,
ρ L
b V2
λ3 = − ; V3 = [− 2 , V 2 , 0], V μ Vμ = 0, V3 = N3,
ρ L
Type:{S, N, N }. (11.64)
As far as the electromagnetic energy momentum tensor is concerned, its matrix
amounts to
⎡ 1 b2 ⎤
− 8 π ρ2 0 0
⎢ ⎥
⎢ 2 ⎥
(T α β ) = ⎢ 0 − 18 πbρ2 0 ⎥, (11.65)
⎣ ⎦
1 b2
0 0 8 π ρ2

with the following eigenvalues and their corresponding eigenvector

1 b2
λ1 = − ; V1 = [V 1 , V 2 , 0],
8π ρ2
V 1μ V 1μ = (V 1 )2 gt t + (V 2 )2 gρρ , V1 = {T1, S1, N1},
1 b2
λ2 = − ; V2 = [Ṽ 1 , Ṽ 2 , 0],
8π ρ2
V 2μ V 2μ = (Ṽ 1 )2 gt t + (Ṽ 2 )2 gρρ , V2 = {T2, S2, N2},
1 b2
λ3 = ; V3 = [0, 0, V 3 ],
8π ρ2
V μ Vμ = (V 3 )2 gφ φ , Ṽ3 = S3. (11.66)
For V1 and V2, the character of these vectors depends on the sign of their
magnitudes; for instance, choosing
√
V 1 = s gρ ρ / |gt t | V 2 , s = constant, V1μ V1μ = (1 − s2 )gρ ρ (V 2 )2 ;
s > 1 → V1 = T, s = ±1 → V1 = N, s < 1 → V1 = S.
 11.3 Magnetostatic Solutions; a = 0, b = 0 159

The spacelike vector V3 is aligned along the circular Killing direction ∂φ . Thus
one may have the spacetime arrangement {T1, S2, S3}, or {N1, N2, S3}, and so
on.
The Cotton tensor for electrostatic cyclic symmetric gravitational field is
given by
⎡ ⎤
b2
0 0 2ρ 2
⎢ ⎥
⎢ ⎥
(C α β ) = ⎢ 0 0 0 ⎥. (11.67)
⎣ ⎦
b2
− 2ρ 4L
2
0 0

The search for its eigenvectors yields

λ1 = 0; V1 = [0, V 2 , 0], V 1μ V 1μ = (V 2 )2 gρρ , V1 = S1,

i L b2 iL 1
λ2 = 3
; V2 = [V 1 , 0, V ], V2 = Z,
2 ρ ρ
i L b2 iL
λ3 = − 3
; V3 = [V 1 , 0, − V 1 ], V3 = Z̄, (11.68)
2 ρ ρ
therefore the corresponding tensor type is

Type I: {S, Z, Z̄}.

The eigenvectors V2 and V3 are complex conjugated, or, if one wishes, one
may consider the component V 1 differently for each of the complex vectors. For
the zero eigenvalue λ1 , the vector V1 is a spacelike vector that it points along
the ρ–coordinate direction.
It is worth pointing out that the field and Cotton tensors of the solutions
generated via coordinate transformations, in particular SL(2, R) transforma-
tions, applied onto this electrostatic cyclic symmetric Peldan solution will shear
the eigenvalues λi of the corresponding field and Cotton tensors of the charged
Peldan solution; recall that eigenvalues are invariant characteristics of tensors,
although the components of the eigenvectors, in general, look different in differ-
ent coordinate systems; this remark also applies to the (eigenvalues) eigenvectors
of the seed and resulting solutions.

11.3 Magnetostatic Solutions; a = 0, b = 0

In this section, the magnetostatic cyclic symmetric Einstein–Maxwell solutions
are derived in the presence of a negative cosmological constant Λ = −1/l2 . As
was noted in the previous section, it seems that static Einstein–Maxwell solutions
with a cosmological constant were first derived in Peldan (1993). For this class
of solutions, Peldan left the Schwarzschild frame of coordinates and used those
ones in which gφφ = H(ρ).
160 Einstein–Maxwell Solutions

The subclass of solutions without Λ are explicitly given and identified, while
the magnetostatic Peldan and Hirschmann–Welch solution representation with
Λ are analyzed in detail.

11.3.1 General Magnetostatic Solutions

To derive the magnetostatic solution, one starts from the addition Et t + Er r
from (11.42), which yields
 
d F d
H − 4 r /l2 = 0, (11.69)
dr H dr
with integral
F d r
H = 4 2 + C1 . (11.70)
H dr l
The substraction Et t − Er r gives
d2
F2 H + 4a2 H 2 = 0. (11.71)
dr2
Substituting F (r) from (11.70) into (11.71) one arrives at a first-order equation
d
for dr H
 2
  2 d2 H dH
4 r + C1 l2 + 4 a2 4
l = 0, (11.72)
dr2 dr
which is rewritten as
d  −1
d( H)−1 = −a2 l4 d 4 r + C1 l2 (11.73)
dr
with first integral
d C2 (4 r + C1 l2 ) − a2 l4
( H)−1 = . (11.74)
dr 4r + C1 l2
A subsequent integration gives
 
R(r)l2 + a2 l4 ln R(r)l2 + C3
H(r) = ,
4C22
F (r) = R(r)H(r),
R(r) l2 : = C2 (4 r + C1 l2 ) − a2 l4 , (11.75)

where (11.70) it has been used to evaluate F (r). These structural functions
completely determine the magnetostatic solution; without any loss of generality,
by letting

C2 → C1 l2 /4, C1 → 4 (a2 l2 + C0 )/(C1 l2 ), C3 → K0 l2 − a2 l4 ln l2 ,

the magnetostatic metric can be given as

 11.3 Magnetostatic Solutions; a = 0, b = 0 161

dr2
g = −h(r)dt2 + + H(r)dφ2 ,
H(r)h(r)
4
H(r) = K0 + h(r) + a2 l2 ln h(r) ,
C1 2 l2
F (r) = H(r) h(r), h(r) := C1 r + C0 . (11.76)

This solution is characterized by:

the electromagnetic field vector
a
A= ln h dφ, (11.77)
C1
the electromagnetic field tensors
a
F μν = −2aδ[φ μ δr] ν , Fμν = −2 δ[μ φ δν] r (11.78)
h(r)
with field invariant
a2
Fμν F μν = 2 , (11.79)
h(r)
the energy–momentum tensor
a2 1
Tμ ν = −δμ t δt ν + δμ r δr ν + δμ φ δφ ν , (11.80)
8 π h(r)
with energy field invariants
3 a4 1 1 1
Tμν T μν = , T μ = a2 . (11.81)
64 π 2 h(r)2 μ 8π h(r)
This class of solutions allows for a hydrodynamics interpretation in terms of a
perfect fluid energy–momentum tensor for a stiff fluid, μ = p, where μ and p are,
respectively, the fluid energy density and the fluid pressure. In fact, the energy
momentum tensor for a perfect fluid is given by

Tμν = (μ + p)uμ uν + p gμν .

√
Therefore choosing the fluid 4-velocity along the time direction uμ = δtμ / −gtt
2
a H
one establishes that μ = 8π F 2 = p.

11.3.2 Melvin, and Barrow–Burd–Lancaster Magnetostatic Solution

Melvin (1986) derived the electric and the magnetic static solutions for vanishing
cosmological constant Λ = 0. The corresponding solution can be obtained from
the metric (11.76) introducing new coordinates according to h = C1 r + C0 →
ρ2 , t → t, φ → φ C1 /2, and setting K0 = 0 arriving at

dρ2
g = −ρ2 dt2 + + F (ρ)dφ2 , F (ρ) = k0 + 2a2 ln ρ, (11.82)
F (ρ)
162 Einstein–Maxwell Solutions
2
or by introducing a4 r2 ek0 /a = k0 + a2 ln ρ2 , and scaling the variables t and φ
one brings it to the form
2
g = er (−dt2 + dr2 ) + r2 dφ2 . (11.83)

In the paragraph devoted to stiff perfect fluid, Barrow, Burd, and Lancaster
– see Barrow et al. (1986) – pointed out that for a fluid aligned along the
time-coordinate, “in (2 + 1) dimensions the stiff fluid has an energy–momentum
tensor identical to that of a static magnetic field,” and they continued with a
statement very close to the following: if one sets the electric field components
√
F0i = 0 and magnetic components Fi j = i j 2μ in the electromagnetic energy–
momentum tensor Tμν = Fμλ Fν λ − gμν Fαλ F λα /4 reduces to the perfect fluid
energy–momentum tensor Tμν = (μ + p)uμ uν + p gμν , with energy density μ
equalling the pressure p, μ = p.

11.3.3 Peldan Magnetostatic Solution with Λ

Introducing in the metric (11.76) new coordinates according to h = C1 r + C0 →
ρ2 , t → t, φ → φ C1 /2, K0 → k0 l2 one gets

dρ2 ρ2
g = −ρ2 dt2 + + F (ρ)dφ2 , F (ρ) = k0 + 2 + 2a2 ln ρ,
F (ρ) l
A = a ln ρ dφ, (11.84)

characterized by the field tensors

a2 1
Tμ ν = −δμ t δt ν + δμ ρ δρ ν + δμ φ δφ ν ,
8 π ρ2
a
Fμν = −2 δ[μ φ δν] ρ . (11.85)
ρ
This solution has also been derived and analyzed in Peldan (1993), Eq. (86).
The Peldan (1993) magnetostatic solution – see also Garcı́a (2009), Eq.
(4.30) – with a negative cosmological constant is determined, in canonical
representation, by
1
ds2 = −N (ρ)2 dt2 + dρ2 + K(ρ)2 [dφ + W (ρ)dt]2 ,
L(ρ)2

ρ2
L(ρ) = K(ρ) = + 2 a2 ln ρ + m, N (ρ) = ρ, W (ρ) = 0. (11.86)
l2
Notice that the metric functions L2 and K 2 are positive functions for values of
ρ > ρroot , where
 
m 1 1
ρroot = exp − 2 − LambertW ( 2 2 e− a2 ) , LW (x) exp(LW (x)) = x,
m

2a 2 l a
 11.3 Magnetostatic Solutions; a = 0, b = 0 163

where for short LW := LambertW , ρroot is solution of the equation,

g ρ ρ (ρroot ) = L2 (ρroot ) = 0, ρ2root /l2 + 2 a2 ln ρroot + m = 0.
Therefore the coordinate ρ does not cover the expected range 0 ≤ ρ ≤ ∞. For
ρ ≤ ρroot the metric suffers an unacceptable signature change. This fact also
points out on the non-existence of a horizon ρ = const for the Peldan solution.
Consequently, one has to modify the choice of the ρ-coordinate in order to be
able to rich the origin of coordinates; with this purpose in mind a new coordinate
system is chosen in the forthcoming paragraph (11.3.4).

Mass and Energy

The surface energy density  is given by
1 ρ a2
(ρ) = − ( 2 + )0 . (11.87)
πK l ρ
Consequently the global energy and mass are given by
ρ a2
E(ρ, 0 ) = −22
− 2 − 2π K 0 ,
l ρ
2
ρ
M (ρ, 0 ) = −2 2 − 2a2 − 2π ρ K 0 . (11.88)
l
For the natural choice of a vanishing reference energy density 0 = 0, one has at
the spatial infinity ρ → ∞ that
1 m − 2a2 l a2
(ρ → ∞, 0 = 0) ≈ − +l 2
+ ln ρ,
πl 2π ρ π ρ2
ρ a2
E(ρ → ∞, 0 = 0) = −2 2 − 2 ,
l ρ
ρ2
M (ρ → ∞, 0 = 0) = −2 2 − 2a2 , (11.89)
l
while if the reference energy is the one 
corresponding to the anti-de Sitter space-
2
√
time with M0 parameter, 0 = − πρl2 / ρl2 + M0 , 0|∞ (M0 ) ≈ − π
Λ
+ M0
√
2π Λ ρ2
,
then the energies are expressed at spatial infinity as
m − M0 − 2a2 a2
(ρ → ∞, 0|∞ (M0 )) ≈ l + ln ρ,
2π ρ2 π ρ2
m − M0 − 2a2 a2
E(ρ → ∞, 0|∞ (M0 )) ≈ + 2 ln ρ,
ρ ρ
M (ρ → ∞, 0|∞ (M0 )) ≈ m − M0 − 2a + 2 a2 ln ρ.
2
(11.90)
Comparing these quantities with the corresponding ones of the static BTZ
solution counterpart, one sees a complete correspondence for vanishing electro-
magnetic parameter a, thus one recognizes m as the mass parameter. Notice that
the energy and mass include an amount of energy due to the magnetic field, in
164 Einstein–Maxwell Solutions

a way similar to the electric one, through logarithmical terms; because of this
dependence, these quantities logarithmically diverge at infinity.

Field, Energy–Momentum and Cotton Tensors

The electromagnetic field tensor for this solution is given by
⎡ ⎤
0 0 0
⎢ ⎥
⎢ L2 a ⎥
(F α β ) = ⎢ 0 0 ⎥, (11.91)
⎣ ρ
⎦
0 − L2 ρ
a
0

and is algebraically characterized by the following eigenvectors

λ1 = 0; V1 = [V 1 , 0, 0], V μ Vμ = −(V 1 )2 (ρ2 ), V1 = T1,

a 1
λ2 = i ; V2 = [0, V 2 , i 2 V 2 ], V2 = Z,
ρ L
a 1
λ3 = −i ; V3 = [0, V 2 , −i 2 V 2 ], V3 = Z̄,
ρ L
Type:{T, Z, Z̄}. (11.92)

As far as the electromagnetic energy momentum tensor is concerned, its matrix

amounts to
⎡ 1 a2
⎤
− 8π ρ2 0 0
⎢ ⎥
⎢ 1 a2 ⎥
(T α β ) ⎢ 0 0 ⎥ (11.93)
⎣ 8π ρ2 ⎦
1 a2
0 0 8π ρ2

with the following eigenvalues and their corresponding eigenvectors

1 a2
λ1 = − ; T1 = [V 1 , 0, 0], V μ Vμ = −ρ2 (V 1 )2 ,
8π ρ2
1 a2
λ2 = ; S2 = [0, V 2 , V 3 ], V μ Vμ = (V 2 )2 / L2 + (V 3 )2 L2 ,
8π ρ2
1 a2
λ3 = ; S3 = [0, Ṽ 2 , Ṽ 3 ], V μ Vμ = (Ṽ 2 )2 / L2 + (Ṽ 3 )2 L2 ,
8π ρ2
Type: {T, 2S}. (11.94)

This tensor structure corresponds to that describing a perfect fluid energy

momentum tensor, but this time for the state equation: energy = pressure.
Again, the solutions generated from this metric by using coordinate transfor-
mations will possess this perfect fluid feature because the invariance of the
eigenvalues.
 11.3 Magnetostatic Solutions; a = 0, b = 0 165

The Cotton tensor for electrostatic cyclic symmetric gravitational field is

given by
⎡ 2 2 ⎤
0 0 a2ρL4
⎢ ⎥
⎢ ⎥
(C α β ) = ⎢ 0 0 0 ⎥. (11.95)
⎣ ⎦
2
− 2ρ
a
2 0 0
Searching for its eigenvectors, one arrives at
λ1 = 0; V1 = [0, V 2 , 0], V μ Vμ = (V 2 )2 gρρ , V1 = S1,
i L a2 iL
λ2 = ; V2 = [− V 3 , 0, V 3 ], V2 = Z,
2 ρ3 ρ
2
i La iL
λ3 = − ; V3 = [ V 3 , 0, V 3 ], V3 = Z̄,
2 ρ3 ρ
Type I: {T, Z, Z̄}. (11.96)
The eigenvectors V2 and V3 are complex conjugated while the vector V1, associ-
ated to the zero eigenvalue, is spacelike – the only physically tractable ρ-direction
vector in this case. It is worth pointing out that the solutions generated via coor-
dinate transformations, in particular the SL(2, R) transformations, applied onto
this magneto-static cyclic symmetric metric will shear the eigenvalues λi of the
Cotton tensor quoted above; recall that eigenvalues are invariant characteristics
of tensors, although their components in different coordinate systems are dif-
ferent – this last also applies to the eigenvectors of the seed and the resulting
solutions.

11.3.4 Hirschmann–Welch Solution with Λ

Accomplishing in the general magnetic static metric (11.76) the transformations
C1 r + C0 → (ρ2 + r+
2
− ml2 )/l2 =: h(ρ),
2φ/(C1 l2 ) → φ, a2 l4 = χ2 , K0 = m, (11.97)
one ends with the Hirschmann and Welch (1996) representation of the magnetic
solution
1
g = − 2 (ρ2 + r+
2
− ml2 )dt2 + [ρ2 + r+
2
+ χ2 ln(|h(ρ)|)]dφ2
l
l2 ρ2 dρ2
+ 2 2 + χ2 ln(|h(ρ)|)] ,
(ρ + r+ − ml2 )[ρ2 + r+
2

h(ρ) = (ρ2 + r+
2
− ml2 )/l2 , (11.98)
with vector potential
1
A= χ ln |(ρ2 + r+
2
)/l2 − m|dΦ, (11.99)
2
166 Einstein–Maxwell Solutions

For ρ = 0, one determines the constant r+ fulfilling

2
r+ + χ2 ln |r+
2
/l2 − m| = 0. (11.100)

This solution is endowed with mass, magnetic charge, and radial parameters. The
coordinate ρ ranges from zero to infinity. This magnetic solution does not allow
the existence of an event horizon since timelike geodesics can reach the origin
at finite proper time, while null geodesics approach the origin at finite affine
parameter; hence it does not describe a magnetic black hole. Moreover the Ricci
tensor, and consequently the curvature tensor, as well as the electromagnetic
field, are well behaved in this spacetime.
Cataldo et al. (2004) commented on this static circular magnetic solution of
the (2+1) Einstein–Maxwell equations, derived previously by other authors, and
came to the conclusion that this solution, considered up to that moment as a two-
parameter one, is in fact a one-parameter solution, which describes a distribution
of a radial magnetic field in a (2+1) anti-de Sitter background spacetime, and
that the mass parameter is just a pure gauge and can be rescaled to minus one.
Accomplishing in the original Peldan solution (11.86) the coordinate transfor-
mation

t → t, ρ → (ρ2 + r+ 2 − m l2 )/l2 , φ → φ l2 , χ2 := a2 l2 ,

one obtains the Hirschmann and Welch (1996) representation of the magnetos-
tatic solution, which is given by the metric functions
1
ds2 = −N (ρ)2 dt2 + dρ2 + K(ρ)2 [dφ + W (ρ)dt]2 ,
L(ρ)2
ρ2 + r+
2
− m l2 H(ρ)
H(ρ) = , L(ρ) = K(ρ),
l2 ρ

K(ρ) = ρ2 + r+ 2 + χ2 ln H(ρ),

N (ρ) = H(ρ), W (ρ) = 0. (11.101)

In the original Hirschmann and Welch (1996) work there is a condition to be

fulfilled by the parameter r+ , arising from the vanishing of K at ρ = 0, namely
2
r+ r2 2 2
2
r+ + χ2 ln( 2
− m) = 0 → ( + − m)e(r+ /χ ) = 1. (11.102)
l l2
This equation has been used in the quoted publication to determine the conical
angle deficit: as ρ → 0 the behavior of

K 2 /ρ2 = 1 + χ2 /ρ2 ln [(ρ2 + r+

2
− ml2 )/(r+
2
− ml2 )]

is given by
2
r+ − ml2 + χ2 χ2 2 2
(K 2 /ρ2 )|ρ→0 → = 1 + 2 e(r+ /χ ) ,
r+ − ml
2 2 l
 11.3 Magnetostatic Solutions; a = 0, b = 0 167

hence the spatial sector ( L12 dρ2 + K 2 dφ2 )|ρ→0 of the studied metric behaves as
⎡ ⎤2
2 2 2 2  2
⎣d ρ  e(r+ /2χ )
⎦ + ρ2 e(r+ /χ )
2
(−r+ /2χ2 ) χ2 (r+2 /χ2 )
χ2 2 2)
dφe (1 + 2 e )
1+ χ2 2
e(r+ /χ
2)
1+ l2 e(r+ /χ l
l2

=: dρ̃2 + ρ̃2 dφ̃2 ,

hence the angles range


2
(r+ /2χ2 ) l 2 − ml2
r+
e
0 ≤ φ̃ ≤ 2π → 0 ≤ φ ≤ 2π 2 /χ2 )
χ2 (r+
= 2π 2 − ml2 + χ2 ,
1+ r+
l2 e

thus, the conical singularity at ρ = 0, as pointed out in the HW paper, arises

2 2 2 2 2
in φ with the period Tφ = 2πν := 2π e(r+ /2χ ) /(1 + χl2 e(r+ /χ ) ), consequently
the angle deficit is δTφ = 2π(1 − ν) as reported also in Dias and Lemos
(2002).

Mass and Energy

For this electromagnetic field solution the surface energy density is given by

1 ρ2 + r+
2
− ml2 + χ2
(ρ, 0 ) = −   − 0 , (11.103)
π l ρ2 + r2 + χ2 ln H ρ2 + r2 − l2 m
+ +

while the integral energy and mass amount to

2 ρ2 + r+
2
− ml2 + χ2
E(ρ, 0 ) = −  − 2π K 0 ,
l 2 − l2 m
ρ2 + r+
2 2
M (ρ, 0 ) = − 2
(ρ + r+ − ml2 + χ2 ) − 2π N K0 . (11.104)
l2
The evaluation of the above functions independent of 0 behave at infinity
according to
1 ml2 − 2χ2 χ2 ρ
(ρ → ∞, 0 = 0) ≈ − + 2
+ 2
ln ( ),
πl 2π l ρ πlρ l
ρ m l2 − r+2
− 2χ2
E(ρ → ∞, 0 = 0) ≈ −2 + ,
l lρ
2
M (ρ → ∞, 0 = 0) = − 2 (ρ2 + r+
2
− ml2 + χ2 ). (11.105)
l
Using in the expressions (11.104) the energy
 density for the anti-de Sitter solu-
2
tion counterpart, namely 0 = − π l2 ρ/ M0 + ρl2 , which at the spatial infinity
1

behaves as 0|∞ (M0 ) ≈ − π1l + l M0

2π ρ2 , the series expansions of the corresponding
quantities at ρ → ∞ result in
168 Einstein–Maxwell Solutions

Table 11.3.1 Electro(b)–Magneto(a)-Static Solutions

References § Eqns. elec. mag. Λ

√ √ √
Static E–M Eqns 11.1.6 (11.42)
√ √
Electrostatic 11.2.1 (11.47) 0
√
Gott and Alpert (1984) (11.54) 0 0
Gott et al. (1986)
√
Deser and Mazur (1985) (11.55c) 0 0
√
Melvin (1986) (11.55) 0 0
√
Kogan (1992) (11.55a) 0 0
√ √
Peldan (1993) (11.56) 0
√ √
Magnetostatic 11.3.1 11.76 0
√
Kogan (1992) (11.55a) 0 0
√ √
Peldan (1993) (11.86) 0
√ √
Hirschmann and Welch (1996) (11.98) 0

l M0 ml2 − 2χ2 χ2 ρ
(ρ → ∞, 0|∞ (M0 )) ≈ − 2
+ 2
+ ln ( ),
2π ρ 2π lρ π l ρ2 l
l M0 ml − 2χ
2 2
χ 2
ρ
E(ρ → ∞, 0|∞ (M0 )) ≈ − + + 2 ln ( ),
ρ lρ lρ l
χ2 χ2 ρ
M (ρ → ∞, 0|∞ (M0 )) ≈ m − M0 − 2 2 + 2 2 ln ( ). (11.106)
l l l
Therefore, comparing with the energy characteristics of the BTZ solution, one
concludes that the mass logarithmically diverges at spatial infinity, and that the
role of mass is played by m.
Some data about the electro-magneto–static families of solutions can be
represented schematically in the table above.

11.4 Cataldo Static Hybrid Solution

This class of electromagnetic static fields has no analog in (3 + 1) Maxwell elec-
tromagnetism, where one finds fields A = F (r)dt or A = F (r) dφ but not
A = 2c (tdφ − φdt) which is the present case. The integration starts from the
combination Et t (a = 0 = b) + 2Er r (a = 0 = b) + EΦ Φ (a = 0 = b) from (11.42),
which yields
d2 8
2
F − 2 = 0, (11.107)
dr l
with integral
(r − r1 ) (r − r2 )
F (r) = 4 . (11.108)
l2
As the equation for H(r) one may consider the first-order equation Er r (a =
0, b = 0), which can be written as
 2  2
H,r F,r F,r c2 1
− = − 2− 2 . (11.109)
2H 4F 4F F l F
 11.4 Cataldo Static Hybrid Solution 169

Evaluating the right–hand side of this equation, one arrives at

  2  
d H (r2 − r1 )2 − c2 l4
ln =4 .
dr F 1/2 l4 F 2
For definiteness we assume r2 > r1 . Accomplishing the integration one obtains
   
H (r2 − r1 )2 − c2 l4 r − r1
ln √ =∓ × ln .
F 2 (r2 − r1 ) r − r2
Introducing the constant α through
2
l4 c2 (r2 − r1 ) (1 − α)
α=1− 2
2, c = , (11.110)
(r2 − r1 ) l4

one obtains H(r) in the form

 ± √2α
r − r1
H (r) = K0 2 F (r) . (11.111)
r − r2
Summarizing, this class of solutions is given by the metric
F 2 1
g=− dt + dr2 + H dφ2 ,
H F
4
F = (r − r1 )(r − r2 ),
l2
2 K02 √ √
H= (r − r1 )(1± α)/2 (r − r2 )(1∓ α)/2 ,
l
c
A = (tdφ − φdt). (11.112)
2
The field tensor characterization of this solution is given by
c [μ ν]
F μν = −2 δ t δ φ , Fμν = 2cδ[μ t δν] φ ,
F
c2  μ t 
T μν = −δ t δ ν + δ μ r δ r ν − δ μ φ δ φ ν ,
8πF
c2 3 c4
F F = −2 , T T = ,
F 64 π 2 F 2
2
1 c
Tμμ = − . (11.113)
8 πF
By scaling transformations of the Killingian coordinates φ and t, the arbitrary
constant K0 can be equated to 1.
Subjecting the metric (11.112) to a further coordinate transformation
1 √ 1 √ 1
t = √ K0 l∓ α/2 t , φ = √ l± α/2 φ , r = ρ2 + r1 , M := 2 (r2 − r1 ),
2 2K0 l
(11.114)
170 Einstein–Maxwell Solutions

dropping primes, one brings the static hybrid metric to the form

√
 (1±√α)/2 √
 2 (1∓√α)/2
ρ2 ρ
g = −ρ1∓ α
−M 2
dt + ρ1± α
−M dφ2
l2 l2
 2 −1
ρ
+ 2 −M dρ2 . (11.115)
l
The electromagnetic field tensor under the above mentioned transformations
becomes
√
Fμν = M 1 − αδ[μ t δν] φ ,
M2 (1 − α)
Tν μ = × −δν t δt μ + δν r δr μ − δν φ δφ μ . (11.116)
32π ρ (ρ /l2 − M )
2 2

This solution corresponds to the static charged solution reported in Cataldo

(2002), where the name of azimuthal static solution was coined.

11.4.1 Mass and Energy

The structural functions of the Cataldo static solution – see Garcı́a (2009), Eq.
(4.46), in its canonical representation – are given by
1
ds2 = −N (ρ)2 dt2 + dρ2 + K(ρ)2 [dφ + W (ρ)dt]2 ,
L(ρ)2
√ √
N (ρ) = ρ(1/2− α/2)
(ρ2 /l2 − M )(1/4+ α/4)
,
L(ρ) = (ρ /l − M )
2 2 (1/2)
,
√ √
K(ρ) = ρ (1/2+ α/2)
(ρ /l − M )(1/4−
2 2 α/4)
, W (ρ) = 0. (11.117)

The corresponding surface densities are

√
1 1 (1 + α)
(ρ, 0 ) = − [ρ2 − M l2 ] − 0 ,
ρ2 − M l2
πlρ 2
jφ (ρ) = 0 = J(ρ), (11.118)

while the integral quantities amount to

√ √
(− α/4+1/4) (√α/2+1/2)
E(ρ, 0 ) = −2π0 l( α/2−1/2)
(ρ2 − M l2 ) ρ ,
√
√ √
2 (− α/4−1/4) ( α/2−1/2)
√
−l (ρ − M l )
( α/2−3/2) 2
ρ [2ρ − (1 +
2
α)M l2 ]
ρ 2 √ 2π
M (ρ, 0 ) = −2 2 + (1 + α)M − 0 ρ ρ2 − M l2 . (11.119)
l l
The evaluation of energy and mass functions independent of 0 behave at
infinity as
 11.4 Cataldo Static Hybrid Solution 171
√
1 l αM
(ρ → ∞, 0 = 0) ≈ − + ,
πl 2π ρ2
√
√
1 ρ (1 + α)M
E(ρ → ∞, 0 = 0) ≈ l 2 (1+ α)
[−2 2 + ],
l 2ρ
√ ρ2
M (ρ → ∞, 0 = 0) ≈ M (1 + α) − 2 2 . (11.120)
l
Using in the expressions(11.118) and (11.119) as the reference energy density
ρ2
the quantity 0 = − πρ
1
−M0 + l2 , which at the spatial infinity behaves as
0|∞ (M0 ) ≈ − π1l
+ M0
2π ρ2 the series expansions of the corresponding quantities at
ρ → infinity result in
l √
(ρ → ∞, 0|∞ (M0 )) ≈ (−M0 + αM ),
2π ρ2
√
(−M0 + αM ) 1 (1+√α)
E(ρ → ∞, 0|∞ (M0 )) ≈ l2 ,
ρ
√
M (ρ → ∞, 0|∞ (M0 )) ≈ −M0 + αM. (11.121)
Therefore, comparing with the energy characteristics of the BTZ solution,
one concludes that the mass parameter at spatial infinity is determined by
√
the product αM , although the mass function diverges at infinity as fast as
1/ρ2 , a similar behavior is exhibited by the energy density in that spatial
region.

11.4.2 Field, Energy–Momentum, and Cotton Tensors

The electromagnetic field of this solution is given by
l M (1 − α)1/2
(F α β ) =
(ρ2 − M l2 )1/2 ρ
⎡ √  2 1/2 √α ⎤
l √α
0 0 − 12 ρ α
ρ − M l2
⎢ ⎥
⎢ ⎥
×⎢ 0 0 0 ⎥,
⎣ √
⎦
√
ρ√ α
 
2 −1/2 α
− 21 l α
ρ2 − M l 0 0
(11.122)
and is characterized by the following eigenvalues and eigenvectors
λ1 = 0; V1 = [0, V 2 , 0], V μ Vμ = (V 2 )2 gρρ , V1 = S1,
√
Ml 1 − α √  −1/2 √α √α
λ2 = −1/2 ; N2 = [l α ρ2 − M l2 ρ , 0, 1],
ρ2 − M l2 ρ
V μ Vμ = 0, V2 = N2,
√
Ml 1 − α √  −1/2 √α √α
λ3 = 1/2 ; N3 = [−l α ρ2 − M l2 ρ , 0, 1],
ρ2 − M l2 ρ
172 Einstein–Maxwell Solutions

V μ Vμ = 0, V3 = N3,
Type : {S, N, N }. (11.123)

On the other hand, the energy–momentum tensor, having the structure

⎡ ⎤
1 0 0
1 M 2 l2 (1 − α) ⎢ ⎥
⎢ 0 −1 0 ⎥ ,
(T α β ) = − ⎣ ⎦ (11.124)
32 ρ (ρ − M l ) π
2 2 2

0 0 1
2 2
1 M l (1−α)
allows for the eigenvalues λ1 = 32 ρ2 (ρ2 −M l2 )π and the other one, of multiplicity
2 2
1 M l (1−α)
two, λ2 = λ3 = − 32 ρ2 (ρ2 −M l2 )π with the corresponding eigenvectors

1 M 2 l2 (1 − α)
λ1 = ;
32 ρ2 (ρ2 − M l2 ) π
V1 = [0, V 2 , 0], V μ Vμ = (V 2 )2 gρρ , V1 = S1,
1 M 2 l2 (1 − α)
λ2 = − ;
32 ρ2 (ρ2 − M l2 ) π
V2 = [V 1 , 0, V 3 ], V μ Vμ = (V 1 )2 gt t + (V 3 )2 gφ φ ,
V2 = T2, S2, N2,
1 M 2 l2 (1 − α)
λ3 = − ;
32 ρ2 (ρ2 − M l2 ) π
V3 = [Ṽ 1 , 0, Ṽ 3 ], V μ Vμ = (Ṽ 1 )2 gt t + (Ṽ 3 )2 gφ φ ,
V3 = T3, S3, N3. (11.125)

For V2 and V3, the character of the vector depends on the sign of its norm; for
instance, by choosing
√
V 1 = s gφ φ / |gt t | V 3 , s = const., V1μ V1μ = (1 − s2 )gφ φ (V 3 )2 ;
s > 1 → V1 = T, s = ±1 → V1 = N, s < 1 → V1 = S.

Recall that in (3+1) gravity the eigenvectors of the electromagnetic energy–

momentum tensor (and at the same time of the electromagnetic field tensor) are
null in pairs, i.e. they exhibit double coincidence. Hence in the (2+1) case under
study one may think of the alignments {S, 2N} or {N, S, N} as the corresponding
reductions of electromagnetic field eigen-directions of (3+1) gravity.
To complete the characterization of this solution, it is reasonable to add some
comments about the conformal Cotton tensor, which is given by
⎡ √
α  −√α/2 ⎤
0 0 ρl√α ρ2 − M l2
⎢ ⎥
⎢ ⎥
(C α β ) = C0 ⎢ 0 0 0 ⎥ , (11.126)
⎣ √ ⎦
√
α   α/2
− ρl √α ρ2 − M l2 0 0
 11.5 Uniform Electromagnetic Solutions Fμν ;σ = 0 173
√
l3 α(α−1)M 3
where C0 = 8(ρ2 −M l2 )3/2 ρ3
with eigenvectors

λ1 = 0; V1 = [0, V 2 , 0], V μ Vμ = (V 2 )2 /L2 , V1 = S,

√
i αl3 (1 − α) M 3
λ2 = − 3/2
;
8 (ρ2 − M l2 ) ρ3
√  − √α/2 −√α 3
V2 = [V 1 = −iρ α ρ2 − M l2 l V , 0, V 3 ], V2 = Z,
√ 3
i αl (1 − α) M 3
λ3 = 3/2
;
8 (ρ2 − M l2 ) ρ3
√  − √α/2 l−√α 3
V3 = [V 1 = iρ α ρ2 − M l2 V , 0, V 3 ], V3 = Z̄,
Type I : {S, Z, Z̄}. (11.127)

11.5 Uniform Electromagnetic Solutions Fμν ;σ = 0

To determine all uniform electromagnetic solutions, i.e., those possessing van-
ishing covariant derivatives of Fμν , Fμν ;σ = 0, one has to start the integration
process from the differential relations arising from these conditions. The hybrid
class c = 0 does not allow for such kind of solutions. The other families with
a = 0 and (or) b = 0 give rise to nontrivial solutions.

11.5.1 General Uniform Electromagnetic Solution for a = 0, = b

A class of uniform electromagnetic stationary solutions, for a = 0, b = 0 and
c = 0, can be constructed by demanding the vanishing of the covariant derivatives
of the electromagnetic tensor field, Fμν ;σ = 0, which yields two independent
equations: Ftφ;t = 0 and Ftr;r = 0. From the last one, one isolates dW /dr
dW F dH
= −b 3 , (11.128)
dr H (a − b W ) dr
which when used in the first equation Ftφ;t = 0 allows us to write
dF F dH F2 dH
= − b2 3 . (11.129)
dr H dr H (a − b W )2 dr
As the next step, one substitutes recursively these first derivatives into the Ein-
stein equations. One gets, among other relations, a simple expression for the
equation Eφ t
d2 H H2
= −4 (a − b W )2 , (11.130)
dr2 F2
which when substituted back into the Einstein equations reduces them to a single
relation
F (b2 l2 − H) = l2 H 2 (a − b W )2 , (11.131)
174 Einstein–Maxwell Solutions

from which one has

a F
W (r) = ∓ b2 l2 − H. (11.132)
b lbH
Using the relation (11.131) in (11.130) one obtains
d2 H 4
= − 2 (b2 l2 − H). (11.133)
dr2 l F
On the other hand, substituting W (r) from (11.132) into (11.129) one arrives at
the relation
dF 2 2 dH
(b l − H) + F = 0, (11.134)
dr dr
with integral
b2 l2 − H(r)
F (r) = , β = constant. (11.135)
l2 β 2
The substitution of F (r) from (11.135) into (11.133) yields
d2 H
= −4β 2 , (11.136)
dr2
hence
H(r) = −2β 2 r2 + c1 r + c0 . (11.137)
Consequently the function W (r) becomes
a 1 b2 l 2 − H
W (r) = ∓ 2 . (11.138)
b l bβ H
No restriction arises from the remaining (11.128). Thus, we have determined the
general uniform electromagnetic stationary cyclic symmetric solution given by
the metric and the field vector
b2 l2 − H(r) 2 l2 β 2 dr2
g=− dt +
l2 β 2 H(r) b2 l2 − H(r)
   2
a 1 b2 l 2 − H
+H(r) dφ + ∓ 2 dt ,
b l bβ H
H(r) = −2β 2 r2 + C1 r + C0 ,
1 ± al2 β
A = −β r [dφ − 2 dt], (11.139)
l bβ
characterized by the uniform electromagnetic field tensors
1 ± al2 β t r
Fμν = −2 δ[μ δν] ± 2βδ[μ r δν] φ ,
l2 b
8π l2 Tν μ = −(1 ± 2al2 β) δν t δt μ − δν φ δφ μ
2a
−δν r δr μ ∓ 2 β bl2 δν φ δt μ + (1 ± al2 β)δν t δφ μ . (11.140)
b
with F F invariant F F = −2/l2 .
 11.5 Uniform Electromagnetic Solutions Fμν ;σ = 0 175

Although the solution above has been derived for Λ = −1/l2 , the branch
corresponding to Λ = 1/l2 is achieved from the above expressions by changing
l2 → −l2 .

11.5.2 Uniform “Stationary” Electromagnetic

A = r/(b l2 )(dt − ω0 dφ) Solutions
Consider now the case a = 0 = c for the metric (11.16)
F 2 dr2 F dr2 2
g=− (dt − ω dφ) + h dφ2 + = − dt2 + + H (dφ + W dt) ,
h F H F
F ω F
F = F, H = h − ω 2 , W = . (11.141)
h H h
The electromagnetic tensor amounts to F μν = 2 b δ μ [t δ ν r] , and
b HW [φ r]
Fμν = −2 (F − H 2 W 2 ) δμ [t δν r] + 2 b δμ δν
HF F
b ω(r) [φ r]
= −2 δμ [t δν r] + 2 b δμ δ ν .
h(r) h(r)
The covariant derivatives Fφr;r and Ftr;r of the field Fμν are equal to zero if
ω(r) = ω0 , h(r) = h0 .
Therefore, the structural functions ω(r) = ω0 and h(r) = h0 are constants. The
Einstein–Maxwell equations require the fulfillment of the equations
d2 4 2
F (r) = 2 → F (r) = 2 r2 + c1 r + c0 , h0 = b2 l2 . (11.142)
d r2 l l
Consequently the derived solution can be given as
F dr2
g=− (dt − ω0 dφ)2 + h0 dφ2 + ,
h0 F (r)
2r2
F (r) = 2 + c1 r + c0 , h0 = b2 l2 ,
l
r
A = 2 (dt − ω0 dφ), (11.143)
bl
and hence by a shifting transformation of the t-coordinate the derived metric
becomes a static one. The electromagnetic tensors characterizing this uniform
“stationary” cyclic symmetric solution are given by
1 ω0
Fμ ν = −2 2
δ[μ t δν] r + 2 2 δ[μ φ δν] r ,
bl bl
1 ω0
ν
Tμ = 2
(−δμ δt − δμ r δr ν + δμ φ δφ ν ) +
t ν
δμ t δφ ν . (11.144)
8πl 4 π l2
This solution can be generated from the static solution, which is given in subsec-
tion 11.5.3 by the metric (11.147), via the transformation t → t − ω0 φ, φ → φ.
176 Einstein–Maxwell Solutions

Clément Uniform “Stationary” Electromagnetic Solution

Clément (1993) reported the uniform “stationary” generalization of the electro-
static solution in the form of

F 2 dr2
g=− (dt − ω0 d φ) + + H0 dφ2 ,
H0 F (r)
2r2
F (r) = 2 + c1 r + c0 ,
l
1
A= √ r (dt − ω0 d φ) , (11.145)
H0 l
π02 l2
In Clément’s parametrization one adopts H0 = 4m , with m = 1/(2 κ).

No Uniform Generalization of the Electrostatic Solution for Λ = 1/l2

On the other hand for b = 0 and positive cosmological constant Λ = 1/l2 there is
no a uniform electromagnetic stationary cyclic symmetric solution; the reason is
hidden in the resulting erroneous signature. In the considered case, for the met-
ric (11.16) with structural functions F (r), h(r), ω(r) the electromagnetic tensor
amounts, for a = 0, to

F μν = 2 b δ μ [t δ ν r] ,
Fμν = −2 b/h(r) δμ [t δν r] + 2 b ω(r)/h(r) δμ [φ δν r] .

Therefore Fμν;λ = 0 is achieved for h(r) = h0 , ω(r) = ω0 . The Einstein equations

requires Er r = b2 /h0 +1/l2 = 0 → h0 = −b2 l2 . The covariant tensor components
gtt and grr explicitly amount to gtt = F (r)/(b2 l2 ) > 0, grr = 1/F (r) > 0,
which contradicts the adopted signature {−, +, +}, therefore this case does not
represent a compatible solution.

11.5.3 Matyjasek–Zaslavskii Uniform Electrostatic A = r/(b l2 ) dt

Solution
The sub-branch {a = 0, b = 0, W (r) = 0} of uniform electrostatic solutions
arises for constant H(r), H(r) = H0 = constant. The equation Ett implies that
the constant H0 has to be H0 = b2 l2 . The remaining equation Et φ amounts to

d2 F 4 r2
− = 0, → F (r) = 2 + 4 c1 r + c0 ; (11.146)
d r2 l2 l2
consequently the metric and the field vector become

F (r) 2 dr2 r
g=− 2 2
dt + + b2 l2 dφ2 , A = 2 dt. (11.147)
b l F (r) bl
 11.5 Uniform Electromagnetic Solutions Fμν ;σ = 0 177

The electromagnetic field tensors of this solution possess constant eigenvalues

and also exhibit the uniform character; explicitly they are given by
⎡ ⎤ ⎡ ⎤
0 − b1l2 0 − 8π1l2 0 0
⎢ 1 ⎥ ⎢ ⎥
Fμ ν = ⎢
⎣ b l2 0 0 ⎥ ν
⎦ , Tμ = ⎣
⎢ 0 − 8π1l2 0 ⎥ ⎦, (11.148)
1
0 0 0 0 0 8π l2

with constant field invariants given by

2 3 1
FF = − , TT = .
l2 64 π 2 l4
Incorporating the constant H0 = b2 l2 in the new definitions of t and φ, t/b l → t,
b lφ → φ, one can set b l = 1 in the metric (11.147).
For the sake of comparison with previous reports let us introduce hyperbolic
functions:
 
r2 2(r + l2 c1 )2
F (r) = 2 2 + 4 c1 r + c0 = |c0 − 2 l2 c21 | 2 ± 1 ,
l l |c0 − 2 l2 c21 |

+ : r + l2 c1 = l2 |c0 − 2 l2 c21 |/2 sinh(αx), F (r) → |c0 − 2 l2 c21 | cosh2 (αx),

− : r + l2 c1 = l2 |c0 − 2 l2 c21 |/2 cosh(αx), F (r) → |c0 − 2 l2 c21 | sinh2 (αx),
√ 2 √
c0 = 2 l2 c21 : r + l2 c1 = exp 2x/l, F (r) → 2 exp (2 2x/l). (11.149)
l
The above mentioned quantities, i.e., the metric (11.147) and the structural
functions (11.149), determine the solution derived in Matyjasek and Zaslavskii
(2004). Expressions (11.147) and (11.149) are equivalent to the Bertotti (1959)
and Robinson (1959) uniform electromagnetic–gravitational field solution, for a
constant slice of one of the spatial coordinates; the BR (3 + 1) solution allows
for a product of two surfaces of constant curvature as manifold. Moreover, the
2D BR metric sector ds2− reduces to r− 2
dφ2 in the (2 + 1) case, therefore the
(2+1)-dimensional uniform electrostatic field can be considered as a dimensional
reduction of the (3 + 1) Bertotti–Robinson solution.
With the aim of demonstrating the uniqueness of this class of uniform solutions
with H(r) = h0 = constant, even in the framework of stationary fields, let us
consider the general metric with W (r):
In the case {a = 0 = c, b = 0, H(r) = h0 }, the combination of equations Et t −
2
h0 b2
Er r − W Et φ = 2 W (r) F (r) implies W (r) = 0 and consequently the gravitational
field is static; further integration gives rise to the above mentioned uniform
electrostatic fields.
Case {a = 0, b = 0 = c, H(r) = h0 }: from Et t − Er r − W Et φ = 2 a2 h0 /F (r)
therefore there is no solution in this case.
Case {a = 0 = b, c = 0, H(r) = h0 }: the combination Et t − Er r − W Eφ t =
2c2 /F (r), hence there is no solution.
178 Einstein–Maxwell Solutions

Vanishing Mass, Energy and Momentum

The uniform electrostatic solution Matyjasek and Zaslavskii (2004) – see
also Garcı́a (2009), Eq. (5.20) – is given by the metric functions
1
ds2 = −N (ρ)2 dt2 + dρ2 + K 2 [dφ + W dt]2 ,
L(ρ)2

2 2
L(ρ) = N (ρ) = ρ + 4c1 ρ + c0 , K = 1, W = 0. (11.150)
l2
Since the surface energy density  is proportional to 0 ,  = −0 , consequently
all the energy–mass quantities are given through it

 = −0 , M (ρ, 0 ) = −2π N (ρ)0 , E(ρ, 0 ) = −2π0 . (11.151)

Thus, for the natural choice of a vanishing reference energy density 0 = 0 all
the energy quantities vanish:  = 0, M (ρ, 0) = 0 = E(ρ, 0). On the other hand,
if the reference
 energy is the one corresponding to the anti-de Sitter spacetime,
ρ2
0 = − π1ρ l2 − M0 , the energies M (ρ, 0 ) and E(ρ, 0 ) will be again expressed
through 0 .

Field, Energy–Momentum, and Cotton Tensors

As far as the eigenvalue–vector properties of this solution go, one establishes
straightforwardly that the Cotton tensor ought to vanish because of uniform
character of the electromagnetic field, hence the (2+1) Matyjasek–Zaslavski grav-
itational field is conformally flat, C α β = 0, Type O. On the other hand the
electromagnetic field tensor
⎡ 1 ⎤
0 N 2l 0
⎢ 2 ⎥
(F α β ) = ⎢ N
⎣ l 0 0 ⎥
⎦, (11.152)
0 0 0
allows for the eigenvectors
2
λ1 = 0; V1 = 0, 0, V 3 , Vμ V μ = V 3 , V1 = S1,
1
λ2 = ; V2 = V 1 , N 2 V 1 , 0 , V μ Vμ = 0, V2 = N2,
l
1
λ3 = − ; V3 = V 1 , −N 2 V 1 , 0 , V μ Vμ = 0, V3 = N3,
l
Type I : {S, N, N }. (11.153)

For the electromagnetic energy–momentum tensor we have

⎡ ⎤
− 8l12 π 0 0
⎢ ⎥
(T α β ) = ⎢
⎣ 0 − 8l12 π 0 ⎥
⎦, (11.154)
1
0 0 8l2 π
 11.5 Uniform Electromagnetic Solutions Fμν ;σ = 0 179

with eigenvectors
1 2
λ1 = ; V1 = 0, 0, V 3 , Vμ V μ = V 1 , V1 = S1,
8 π l2
1
λ2,3 = − 2
; V2, 3 = V 1 , V 2 , 0 ,
8 π
 2 1l  
N V − V 2 N 2V 1 + V 2
Vμ V = −
μ
N2
V2 = T2, S2, N2, V3 = T3, S3, N3, (11.155)
and therefore it allows for the types {S, 2T }, {S, 2N }, {S, 2S}.

11.5.4 Uniform “Stationary” Electromagnetic

A = r/(a l2 )(dφ + W0 dt) Solutions
In the case of positive cosmological constant Λ = 1/l2 there exists a uniform
“stationary” magnetic solution with constant W (r) = W0 .
The electromagnetic tensor possesses the structure
H H
Fμν = −2 a W δμ [t δν r] − 2 a δμ [φ δν r] ,
F F
and its covariant derivatives are zero if
a H 2 dW
Frt;r = − ,
2F dr
a d F dW
Frφ;r =− (−2H 2 + H3 )
4F H dr H dr
vanish. Hence
F (r) = β H(r), W (r) = W0 = constant, (11.156)
and consequently
a a
Ft r = −W0 , Fφ r = − . (11.157)
β β
The Einstein equations reduce to
a2 1 d2 4
Er r = − + 2 = 0 → β = a2 l2 , 2 Et t = 2 F (r) + 2 = 0
β l dr l
2
→ F (r) = − 2 r2 + c1 r + c0 . (11.158)
l
The metric and fields for the derived solution can be expressed as
dr2 F (r)
g = −a2 l2 dt2 + + 2 2 (dφ + W0 dt)2 ,
F (r) a l
2r2
F (r) = − + c1 r + c0 ,
l2
r
A = 2 (dφ + W0 dt), (11.159)
al
180 Einstein–Maxwell Solutions

with uniform electromagnetic field tensor is

W0 t r 1 φ r
Fμν = −2 2
δ[μ δν] − 2 2 δ[μ δν] ,
al al
and energy–momentum tensor
1
8π Tν μ = [−δνt δtμ + δνr δrμ + δνφ δφμ + 2 W0 δνt δφμ ].
l2
This solution is equivalent to the Clément’s solution given by Eq. (26), Λ = 1/l2
of Clément (1993).

11.5.5 No Uniform Stationary Magnetostatic Solution

for Λ = −1/l2
On the contrary, as far as to the stationary uniform electromagnetic branch
with a = 0 and negative cosmological constant Λ = −1/l2 is concerned, one
establishes that there is no solution at all. Following a similar procedure as the
one used in the previous case, where now b = 0, F (r) = β H(r), W (r) = W0 =
2
constant, the Einstein equation Er r = − aβ − l12 = 0 yields β = −a2 l2 . The
covariant tensor components grr and gφ φ explicitly amount to grr = 1/F (r) > 0,
gφ φ = −a2 l2 F (r) < 0, which yields a contradiction with the adopted signature
{−, +, +}. Hence this case does not represent a solution compatible with the
(2+1) metric signature.

11.6 Constant Electromagnetic Invariants’ Solutions

This section is devoted to the derivation of the electromagnetic fields cou-
pled to stationary (static) cyclic symmetric (2+1) gravitational fields such that
their electromagnetic invariants F F , T and T T are constants; because of the
proportionality of T and T T to F F , it is enough to establish under which
conditions
2
c2 H (a − W b) b2
Fμν F μν = −2 +2 −2
F F H
vanishes.
It should be pointed out that the hybrid c = 0 class of spacetimes does not
allow for constant invariants’ solution when a cosmological constant is present.
Therefore it is sufficient to restrict oneself to the case a = 0 or b = 0.
In what follows we shall search for subclasses of solutions with constant
electromagnetic invariant, namely those with {a = 0, b = 0, W (r) = W (r)},
{a = 0, b = 0, ω(r) = ω0 }, and {a = 0, b = 0 , W (r) = W0 } families of solutions.
At this stage it is worth pointing out that constant invariant electromag-
netic fields contain, as subclasses, the covariantly constant electromagnetic field
solutions, while the inverse statement does not hold.
 11.6 Constant Electromagnetic Invariants’ Solutions 181

11.6.1 General Constant Invariant Fμν F μν = 2γ for a = 0 = b

Restricting the present section to the study of the cases a = 0 and b = 0, the
constancy of F F = 2γ is guaranteed by
F 2
(b + γ H) = (a − b W )2 , (11.160)
H2
which yields
a F (r)
W (r) = ± b2 + γ H(r). (11.161)
b b H(r)
In general, the following combinations of the Einstein equations give
d2 F 8 b2 F − H 2 (a − b W )2
Et t + Eφ φ + 2Er r = − + 4 = 0,
dr2 l2 FH
d2 H
2HF (Et t − Er r − W Eφ t ) = F 2 2 + 4H 2 (a − b W )2 = 0. (11.162)
dr
Using the relation (11.160) one brings (11.162) to the form
d2 F 8 d2 H 4γ 4 b2
= + 4γ, + H = − , (11.163)
dr2 l2 dr2 F F
with solution for F (r)
4 4
F (r) = ( + 2γ)r2 + C1 r + C0 = ( 2 + 2γ)(r − r1 )(r − r2 ), (11.164)
l2 l
and the function H, as solution of its second-order equation (11.163), is expressed
in terms of hypergeometric functions. Nevertheless there exists a shortcut for
the integration of the function H(r) by noticing that Er r contains only first
derivatives of the structural functions: replacing W(r) from (11.161) in the quoted
equation, after extracting square root, one arrives at
dH dF b √
γF − (b2 + γ H) = ±4 1 + l2 γ F b2 + γ H,
dr dr l
which, when introducing the auxiliary function Q(r)2 := b2 + γ H(r), becomes
d Q b
√ ∓2 1 + l2 γ = 0.
dr F lF
Integrating this equation, using F (r) (11.164), one obtains
1 + l2 γ ln (r − r2 ) − ln (r − r1 )
Q = ± bl +β F (r). (11.165)
2 + l2 γ r2 − r1
Fulfilling a single constraint still remains; using the expressions of the function
W (r) and its derivatives from the Eμ ν -equations as well as the derivatives of
F(r), together with H(r) in terms of Q(r) and the derivative of the latter from
(11.6.1), one gets a single equation
 √ dF
2(1 + l2 γ) Q(r)2 + b2 F + l b Q(r) 1 + l2 γ = 0, (11.166)
dr
which is incompatible with Q(r) determined in (11.165) except for γ = −1/l2 .
182 Einstein–Maxwell Solutions

Hence, we conclude that there are no solutions for arbitrary constant

electromagnetic invariant F F = 2 γ.

11.6.2 Constant Electromagnetic Invariant F F = ∓2/l2 Solution

A class of constant electromagnetic invariants’ stationary solutions with a = 0,
b = 0 and γ = −1/l2 , F F = −2/l2 , arises for
a F
W (r) = ∓ b2 l 2 − H . (11.167)
b b l H(r)

As in the previous case, F (r) and H(r) fulfill the (11.163) for γ = −1/l2 and
correspondingly their integrals are given by
2 2
F (r) = r + c1 r + c0 , H(r) = b2 l2 − β 2 l2 F (r).
l2
There are no further constraints from the field equations. Consequently the final
result can be written as
b2 l2 − H(r) 2 dr2
g=− dt + l 2 2
β
l2 β 2 H(r) b2 l2 − H(r)
   2
a 1 b2 l 2 − H
+H(r) dφ + ∓ 2 dt ,
b l bβ H
H(r) = −2β 2 r2 + C1 r + C0 ,
1 ± al2 β
A = −β r [dφ − 2 dt],
l bβ
which coincides with the uniform solution (11.139). Therefore we have deter-
mined a class of uniform constant electromagnetic invariant stationary solutions
for both non-vanishing constants a = 0 = b. Although the solution mentioned
above has been derived for Λ = −1/l2 , the branch with positive Λ = 1/l2 is
achieved from the above-mentioned expressions by changing l2 → −l2 .

11.6.3 Constant Electromagnetic Invariant F F = −2/l2 Solution

for b = 0
The constant electromagnetic invariant solution with b = 0 can be determined as
solution of the Einstein–Maxwell equations by considering the stationary metric
in the form
F 2 dr2 F dr2 2
g=− (dt − ω dφ) + h dφ2 + = − dt2 + + H (dφ + W dt) ,
h F H F
F ω F
F = F, H = h − ω 2 , W = .
h H h
 11.6 Constant Electromagnetic Invariants’ Solutions 183

Demanding the electromagnetic invariant F F in the case b = 0 = a to be

constant, one establishes
2b2
FF = − → h(r) = h0 = constant. (11.168)
h(r)

Substituting h(r) = h0 into the Einstein equations one obtains from Eφ t that
d2 ω 2 dF dω
2
=− , (11.169)
dr F dr dr
which when used in Et t –Er r yields
F dω 2
( ) = 0 → ω(r) = ω0 . (11.170)
h20 dr
Replacing ω = ω0 and h = h0 in the remaining equations one establishes
d2 4 2
2
F (r) = 2 → F (r) = 2 r2 + c1 r + c0 , h0 = b2 l2 . (11.171)
dr l l
Therefore we arrive at a constant electromagnetic invariant solution in the form
F dr2
g=− (dt − ω0 dφ)2 + h0 dφ2 + ,
h0 F (r)
2r2
F (r) = 2 + c1 r + c0 , h0 = b2 l2 ,
l
r
A = 2 (dt − ω0 dφ), (11.172)
bl
which in all respects is identical to the uniform electromagnetic solution (11.143)
derived in the previous section. Notice that this solution exists only for negative
cosmological constant, Λ = −1/l2 ; there is no extension to Λ = 1/l2 . It is
evident that this solution can be generated from the static one, (11.147), via the
transformations t → t − ω0 φ, φ → φ.
For ω0 = 0, the above mentioned metric and field reduce to the Matyjasek–
Zaslavskii solutions – see Section (11.5.3) – thus this class of constant electro-
magnetic invariants’ static solutions is unique with the additional property of
being a uniform static solution.

11.6.4 Constant Electromagnetic Invariant F F = 2/l2 Stationary

Solution for a = 0
In the case of positive cosmological constant Λ = 1/l2 there exists a constant
electromagnetic invariant stationary a = 0 solution. Requiring the constancy of
the electromagnetic invariant F F = 2a2 HF , one gets

H
F F = 2a2 → H(r) = β 2 F (r). (11.173)
F
184 Einstein–Maxwell Solutions

From Eφ t one establishes

d2 2 dF dW
W =− , (11.174)
dr2 F dr dr
which when used in Et t –Er r yields
F dW 2
( ) = 0 → W (r) = W0 (11.175)
β 4 dr
Using W = W0 and H(r) = β 2 F (r) in the remaining Einstein equations, one
gets
d2 4 2
F (r) + 2 = 0 → F (r) = − 2 r2 + c1 r + c0 . (11.176)
d r2 l l
Therefore we have established that there is a unique constant electromagnetic
invariants’ solution given by
dr2 F (r)
g = −a2 l2 dt2 + + 2 2 (dφ + W0 dt)2 ,
F (r) a l
2r2
F (r) = − + c1 r + c0 ,
l2
r
A = 2 (dφ + W0 dt), (11.177)
al
which is identical to the uniform electromagnetic solution (11.159) derived in
the previous Section 11.5 dealing with uniform electromagnetic solutions. Notice
that this solution exists only for positive cosmological constant, Λ = 1/l2 , there
is no Λ = −1/l2 solution within this class.

11.6.5 Vanishing Electromagnetic Invariant F F = 0 Solution

A particular family of stationary cyclic symmetric solutions arises by demanding
the vanishing of the electromagnetic invariant Fμν F μν ,
2 √
H (a − b W ) b2 a F
Fμν F μν = 2 −2 = 0 → W (r) = ± . (11.178)
F H b H
For the above mentioned W (r), the equation Er r gives
 2 2
dF (r − C)
l2 − 16 F = 0 → F (r) = 4 . (11.179)
dr l2
After the substitution of W (r) and F(r) into the Einstein equations, the
remaining equation to be solved amounts to

2 d2
(r − C) H + b2 l2 = 0 → H(r) = C0 + C1 r + b2 l2 ln (r − C) .
dr2
The gravitational and electromagnetic fields of this solution can be given as
 11.6 Constant Electromagnetic Invariants’ Solutions 185

F 2 dr2
g=− dt + + H(dφ + W dt)2 ,
H F
2
(r − C)
H(r) = C0 + C1 r + b2 l2 ln (r − C) , F (r) = 4 ,
√ l2
a F l
W (r) = ± , A = ∓ ln (r − C) (adt + bdφ) . (11.180)
b H 2
This solution is characterized by the field tensor
l
Fμν = ± aδ[μ t δν] r + bδ[μ φ δν] r ,
(r − C)
F μν = 2bδt [μ δr ν] − 2aδφ [μ δr ν] , (11.181)
with energy–momentum tensor
l
Tμ ν = −a bδμ t δt ν + a bδμ φ δφ ν + a2 δμ t δφ ν − b2 δμ φ δt ν . (11.182)
4 π (r − C)
Notice that the three invariants Fμν F μν , Tμμ and Tμν T μν are equal to zero.
Without any loss of generality one can always set C = 0.
This solution corresponds to a possible representation of the Kamata and
Koikawa (1995) solution to be treated in detail in Section 11.6.5. It should be
pointed out that this solution does not belong to the family of uniform solutions,
i.e., the fields possessing vanishing covariant derivatives.

11.6.6 Kamata–Koikawa Solution

Kamata and Koikawa (1995) reported their electrically charged BTZ black hole
with negative cosmological constant such that the Maxwell field is self (anti-
self) dual, a condition which is imposed on the orthonormal basis components
of the electric field and the magnetic field. This solution describes an electrically
charged extreme black hole with mass M , angular momentum J, and electric
charge Q. To achieve their representation one accomplishes in metric (11.180)
the substitutions
r = ρ2 , t → t Q/2, φ → φ/ Q, C1 → Q, l → |Λ|−1/2 C0 → −b2 /Λ ln ρ20 ,
aQ
C = ρ20 , H/Q → K 2 , W Q/2 → + N φ , F → 4 ρ2 L2 , (11.183)
b 2
arriving at the solution
L2 2 dρ2 aQ
g = −ρ2 dt + 2 + K 2 [dφ + ( + N φ )dt]2 ,
K2 L b 2 
b2 ρ2 − ρ20 ρL
L = |Λ|(ρ − ρ0 /ρ) , K = ρ +
2 2 2 2 2
ln 2 , Nφ = ± 2 ,
QΛ ρ K
  0
b ρ2 − ρ20 aQ
A= ln ( ) dφ + dt .
2 Q |Λ| ρ20 b 2
186 Einstein–Maxwell Solutions

The electromagnetic field tensors are


1 1
Tμ ν = a b(δμ t δt ν − δμ φ δφ ν )
8π |Λ|(ρ2 − ρ20 )

Q a2 φ ν 2b2 t ν
− δμ δt − δμ δφ ,
2 Q
 
ρ b
Fμν = a Qδ[μ δν] + 2 √ δ[μ δν] .
t r φ r
(11.184)
|Λ|(ρ2 − ρ20 ) Q

Next, one restores the factor π G in the above mentioned solution through the
identifications of the physical parameters:
 √ √
ρ20 = 4π G Q2 /|Λ| = 1/2
J, b = 2 π GQ3/2 , a = ± 4 π G|Λ|1/2 Q1/2 ,
2|Λ|

arriving at the metric (11.184) with structural functions

 2 
ρL ρ − ρ20
L = |Λ|(ρ − ρ0 /ρ) , N = ± 2 , K = ρ + ρ0 ln
2 2 2 φ 2 2 2
(11.185)
K ρ20
and electromagnetic field tensors
 
ρ2 − ρ20 1
A = Q π |Λ| ln ( )× dφ + dt ,
ρ20 |Λ|
 
√ ρ 1
Fμν = −4Q π G 2 t r
δ[μ δν] + φ r
δ[μ δν] ,
ρ − ρ20 |Λ|
 
2 φ ν
Q G δ μ δ t
Tμ ν = 2 −δμ t δt ν + δμ φ δφ ν − + |Λ|δμ t δφ ν . (11.186)
ρ − ρ20 |Λ|

It should be pointed out that Clément (1993) also reported a metric expression
and electromagnetic vector field describing a solution with vanishing electromag-
netic invariants. Comments concerning the mass content of this solution can be
found in Chan (1996). This solution is horizonless and consequently does not
permit a black hole interpretation.
The Kamata and Koikawa (1995, 1997) solution, see also Garcı́a (2009),
Eq. (7.7), is defined by the metric and the structural functions
1
ds2 = −N (ρ)2 dt2 + dρ2 + K(ρ)2 [dφ + W (ρ)dt]2 ,
L(ρ)2
√
Λ 2 √
L(ρ) = (ρ − ρ20 ), Λ = 1/l,
ρ

Q2 ρ2
K(ρ) = ρ2 + ln ( 2 − 1),
Λ ρ0
 11.6 Constant Electromagnetic Invariants’ Solutions 187

√ Q2 ρ2
N (ρ) = ρ L/ K = Λ(ρ2 − ρ20 )/ ρ2 + ln ( 2 − 1),
Λ ρ0
√
(ρ2 − ρ20 ) Λ √
W (ρ) = Q2 2 − Λ, (11.187)
[ρ2 + Λ ln ( ρρ2 − 1)]
0

which are used in the evaluation of the energy and mass.

Mass, Energy, and Momentum

The surface energy and momentum densities are respectively given by
1 Q2 − Λρ20 + Λρ2
(ρ, 0 ) = − √ 2 2 − 0 ,
π Λ [ρ2 + QΛ ln ( ρρ2 − 1)]
0
2

1 Λρ0 − Q + Q ln ( ρ20 − 1)
2 2 2 ρ

j(ρ) = √  2 2
, (11.188)
π Λ ρ2 + QΛ ln ( ρρ2 − 1)
0

while the integral quantities amount to

Λρ20 − Q2 Q2 ρ2
J(ρ) = 2 √ + 2 √ ln ( 2 − 1),
Λ Λ ρ0
2 Q2 − Λρ20 + Λρ2
E(ρ, 0 ) = − √  − 2π K0 ,
Λ ρ2 + Q2 ln ( ρ22 − 1)
Λ ρ0

ρ2 √
M (ρ, 0 ) = −2Λ ρ2 + 2(2Λρ20 − Q2 ) + 2Q2 ln ( 2 − 1) − 2π Λ (ρ2 − ρ20 )0 .
ρ0
(11.189)
The evaluation of the functions above for vanishing 0 , i.e. 0 = 0, behave at
ρ → ∞ according to
√
Λ Λρ20 − Q2 Q2 ρ
(ρ → ∞, 0 = 0) ≈ − + √ +2 √ ln ( ),
π π Λρ 2 π Λρ 2 ρ0
Λρ20 − Q2 Q2 ρ
j(ρ → ∞) ≈ √ +2 √ ln ( ),
π Λρ π Λρ ρ0
Λρ − Q
2 2
Q 2
ρ
J(ρ → ∞) ≈ 2 0√ + 4 √ ln ( ),
Λ Λ ρ0
√ Λρ20 − Q2 Q2 ρ
E(ρ → ∞, 0 = 0) ≈ −2 Λρ + 2 √ + 2√ ln ( ),
Λρ Λρ ρ0
ρ
M (ρ → ∞, 0 = 0) ≈ −2Λ ρ2 + 2(2Λρ20 − Q2 ) + 4Q2 ln ( ). (11.190)
ρ0
Using in the expressions (11.189) as reference energy density the quantity

1 ρ2
0 = − −M0 + 2 ,
πρ l
188 Einstein–Maxwell Solutions
√
which at the spatial infinity behaves as 0|∞ (M0 ) ≈ − πΛ + 2π M
√0
Λ ρ2
, the series
expansions of the corresponding quantities at ρ → ∞ result in
M Λρ20 − Q2 Q2 ρ
(ρ → ∞, 0|∞ (M0 )) ≈ − √0 + √ +2 √ ln ( ),
2π Λ ρ 2 π Λρ 2 π Λρ 2 ρ0
M0 Λρ20 − Q2 Q2 ρ
E(ρ → ∞, 0|∞ (M0 )) ≈ − √ +2 √ + 4√ ln ( ),
Λρ Λρ Λρ ρ0
ρ
M (ρ → ∞, 0|∞ (M0 )) ≈ −M0 + 2(Λρ0 − Q ) + 4Q ln ( ).
2 2 2
(11.191)
ρ0
Therefore, comparing with the energy characteristics of the BTZ solution, one
arrives at the conclusion that there is no mass parameter of the kind M present in
the BTZ solution. All characteristic functions logarithmically diverge at spatial
infinity.

Field, Energy–Momentum and Cotton Tensors

The electromagnetic field tensor
⎡ ⎤
0 − Λ(ρ2Qρ q
−ρ0 2 )2
0
⎢ ⎥
⎢ qQΛ(ρ2 −ρ0 2 ) √ ⎥
(F α β ) = ⎢ ⎥,
qQ Λ(ρ2 −ρ0 2 )
⎢ ρ 0 − ρ ⎥ (11.192)
⎣ ⎦
0 − √Λ(ρQρ q
2 −ρ 2 )2
0
0

allows for a triple zero eigenvalue and the following set of eigenvectors
√
λ1,2,3 = 0; V = [V 1 , V 2 , Λ V 1 ], V μ Vμ = 0, V = N,
Type N : {3N }. (11.193)

The electromagnetic energy momentum tensor with vanishing invariants is

⎡ √ ⎤
− Λ 0 −Λ
1 Q2 ⎢ ⎥
(T α β ) = √ ⎢ 0 0 ⎥
4 π Λ (ρ2 − ρ0 2 ) ⎣
0 ⎦, (11.194)
√
1 0 Λ

while the Cotton tensor for this electromagnetic–gravitational stationary cyclic

symmetric field is given by
⎡ √ ⎤
− Λ 0 1
Q2 ⎢ ⎥
(C α β ) = 2 ⎢ 0 0 ⎥
(ρ − ρ0 )
2 ⎣ 0 ⎦. (11.195)
√
−Λ 0 Λ

It is clear that both the Cotton and Maxwell tensors possess the same eigenvalues,
namely the triple zero eigenvalue λ = 0. Searching for the eigenvectors of these
tensors, one arrives at
 11.6 Constant Electromagnetic Invariants’ Solutions 189

√ (V 2 )2 ρ2
λ1,2,3 = 0; V = [V 1 , V 2 , Λ V 1 ], V μ Vμ = 2, V = S,
Λ (ρ − ρ0 )
V(V 2 = 0) = N,
Type : {S, 2N }. (11.196)

The eigenvectors are spacelike or null vectors depending on the non-vanishing

or vanishing value of the component V 2 . One may consider them different, one
to another, having different V 1 and V 2 components. The most degenerate cases
are {3S} and {3N }.

√
11.6.7 Proper Kamata–Koikawa Solution, ρ0 = ±Q/ Λ
The proper Kamata–Koikawa solution is defined by the √ metric and structural
functions of (11.187) for Λρ20 − Q2 = 0, i.e., ρ0 = ±Q/ Λ, namely
1
ds2 = −N (ρ)2 dt2 + dρ2 + K(ρ)2 [dφ + W (ρ)dt]2 ,
L(ρ)2
√
Λ 2
L(ρ) = (ρ − ρ20 ),
ρ

Q2 ρ2
K(ρ) = ρ2 + ln ( 2 − 1),
Λ ρ0

√ Q2 ρ2
N (ρ) = ρ L/ K = Λ(ρ − ρ0 )/ ρ2 +
2 2
ln ( 2 − 1),
Λ ρ0
√
(ρ2 − ρ20 ) Λ √
W (ρ) = 2 2 − Λ, (11.197)
[ρ2 + Λ ln ( ρ2 − 1)]
Q ρ
0

The surface energy and momentum densities are respectively given by

√
Λ ρ2
(ρ, 0 ) = − 2 2 ,
π [ρ2 + Q
ln ( ρ2 − 1)]
Λ ρ0
2

1 Q2 ln ( ρρ2 − 1)
j(ρ) = √  0
, (11.198)
π Λ ρ2 + Q2 2
Λ ln ( ρρ2 − 1)
0

while the integral quantities amount to

Q2 ρ2
J(ρ) = 2 √ ln ( 2 − 1),
Λ ρ0
√ ρ2
E(ρ, 0 ) = −2 Λ  2 2
− 2π K0 ,
ρ2 + QΛ ln ( ρρ2 − 1)
0

ρ2 √
M (ρ, 0 ) = −2Λ ρ2 + 2Λρ20 + 2Q2 ln ( 2 − 1) − 2π Λ (ρ20 − ρ20 )0 . (11.199)
ρ0
190 Einstein–Maxwell Solutions

The evaluation of the functions above for vanishing 0 , i.e. 0 = 0, behaves at

spatial infinity according to
√
Λ Q2 ρ
(ρ → ∞, 0 = 0) ≈ − +2 √ ln ( ),
π π Λ ρ2 ρ0
Q2 ρ
j(ρ → ∞) ≈ 2 √ ln ( ),
π Λρ ρ0
Q2 ρ
J(ρ → ∞) ≈ 4 √ ln ( ),
Λ ρ0
√ Q2 ρ
E(ρ → ∞, 0 = 0) ≈ −2 Λρ + 2 √ ln ( ),
Λρ ρ0
ρ
M (ρ → ∞, 0 = 0) ≈ −2Λ ρ2 + 2Λρ20 + 4Q2 ln ( ). (11.200)
ρ0

Using in the expressions (11.199) as reference energy density the quantity

√
Λ M0
0|∞ (M0 ) ≈ − + √ ,
π 2π Λ ρ2

the series expansions of the corresponding quantities at ρ = infinity result in

M Q2 ρ
(ρ → ∞, 0|∞ (M0 )) ≈ − √0 +2 √ ln ( ),
2π Λ ρ2 π Λρ 2 ρ0
M0 Q2 ρ
E(ρ → ∞, 0|∞ (M0 )) ≈ − √ + 4√ ln ( ),
Λρ Λρ ρ0
ρ
M (ρ → ∞, 0|∞ (M0 )) ≈ −M0 + 4Q ln ( ).
2
(11.201)
ρ0

In the work by Chan (1996) there are some comments addressed to the evalua-
tion of the global momentum, energy and mass of the proper Kamata–Koikawa
solution: the exact and the approximated expressions of the momentum J coin-
cide with the corresponding ones given in (Chan, 1996, Eq. 9) and (Chan, 1996,
Eq. 7). Moreover, the mass M at spatial infinity, (11.200), coincides with the
M (Chan, 1996, Eq. 10.) for a zero background energy density with the cor-
rect extra term −2Λ ρ2 . From my point of view, it is to be recommended to
accomplish series expansions of the quantities under consideration to deter-
mine how fast they approach zero or diverge at spatial infinity. From this
perspective, the evaluation of the energy density  and the global energy E
yield to quantities different from zero at spatial infinity, although they both
approach faster to zero as ρ → ∞ than the momentum and mass. Compar-
ing with the energy characteristics of the BTZ solution, one concludes that
the mass, energy and momentum functions logarithmically diverge at spatial
infinity.
 11.7 Ayón–Cataldo–Garcia Stationary Hybrid Solution 191

11.7 Ayón–Cataldo–Garcia Stationary Hybrid Solution

The main goal of this section is to derive the stationary cyclic symmetric
spacetime corresponding to the case c = 0, i.e., for the vector potential
c grr
A = (tdφ − φdt) → ∗F = c √ dr. (11.202)
2 −g
It is worthwhile to point out that this case has no analog in stationary axial
symmetric spacetimes of the standard (3+1) Einstein–Maxwell theory. The set
of field equations is given by: {Et t , Et φ , Er r , Eφ t , Eφ φ }. In the forthcoming sub-
sections two main families of solutions exhibiting the hybrid feature of the vector
potential are derived.
The starting point in the integration process of the system of field equations
is Eφ t (a = 0 = b) = 0, (11.26g), which possesses a first integral of the form
J
W,r = , (11.203)
H2
where J is an integration constant. The combination E t t + 2Er r + Eφ φ , for
a = 0 = b, yields
8
F,r,r − 2 = 0, (11.204)
l
which possesses the general solution
4
F = (r − r1 )(r − r2 ), (11.205)
l2
where r1 and r2 are constant of integration. Next, using W,r from (11.203) in
Er r (a = 0 = b), (11.26e), one arrives at
 2
1 H,r 1 F,r J2 F,r 2 c2 1
− − = − − 2 . (11.206)
4 H 2 F 4H 2 F 16F 2 F2 l F
The evaluation the right-hand side of this equation gives the same result as in
the static case, thus one gets
  2
d H J2 (r2 − r1 )2 α
ln − = 4 , (11.207)
dr F 1/2 H2 F l4 F 2
where α is defined through
(r2 − r1 )2
c2 = (1 − α).
l4
From the above equation it becomes apparent that H can be sought in the form of
H(r) = h (r) F (r). (11.208)
Replacing H(r) from above into (11.207) one obtains an equation for h(r) which
can be given as
dh dr √ 2
√ = ∓ , α0 := 2(r2 − r1 ) α/l , (11.209)
α0 2 h2 + J 2 F
192 Einstein–Maxwell Solutions

with integral
  ± √α/2
r − r1
ln α0 h + α0 2 h2 + J 2 = ln k1 , (11.210)
r − r2

where k1 is an integration constant. Therefore h(r) can be expressed as

 ± √α/2  ∓ √α/2
l2 k1 r − r1 J 2 r − r1
h(r) = √ − 2 . (11.211)
4(r2 − r1 ) α r − r2 k1 r − r2

The integration of the (11.203) for W does not present problem.

Summarizing the derived above results, one has that this family of solutions
can be given by
F 2 dr2
g=− dt + + H(dφ + W dt)2 ,
H F
4
F = (r − r1 )(r − r2 ),
l2 ⎡ ⎤
 ± √2α  ∓ √2α
(r − r1 )(r − r2 ) ⎣ r − r1 r − r1 ⎦,
H(r) = l √ × − K12 J 2
2K1 (r2 − r1 ) α r − r2 r − r2
 ± √α −1
4 √ r − r1
W (r) = W0 ± 2 J K1 α(r2 − r1 ) ×
2
− K12 J 2 ,
l r − r2
c
A = (tdφ − φdt), (11.212)
2
where the constant K1 stands for 1/k1 , K1 = 1/k1 , and W0 is an integra-
tion constant. Recall that the parameter α is related to c, r1 and r2 through
c2 = (r2 −r
2
1)
l4 (1 − α).
Correspondingly, the electromagnetic field tensors are
Fμν = 2cδ[μ t δν] φ ,
c2
Tμ ν = (−δμ t δt ν + δμ r δr ν − δμ φ δφ ν ), (11.213)
8π F
with invariants
c2 3 c4 1 c2
Fμν F μν = −2 , Tμν T μν = , T μ
= − .
F 64 π 2 F 2 μ 8 πF
This solution has been reported, for the first time to our knowledge, in Ayón–
Beato et al. (2005). The static hybrid solution (11.112) arises from the stationary
√
one above by setting J = 0 = W0 and identifying 4(r2 − r1 ) α K02 = l2 /K1 .

11.7.1 ACG Hybrid Solution Allowing for BTZ Limit

To achieve a representation of this hybrid solution in terms of the radial
coordinate ρ, such that at the limit of vanishing electromagnetic parameter
 11.7 Ayón–Cataldo–Garcia Stationary Hybrid Solution 193

c = 0 → α = 1, the stationary BTZ solution would arise, one has to accomplish

the coordinate transformations

l 1 − K12 J 2
t= T, (11.214a)
4K1 r2 − r1
 
l 1 − K12 J 2 K1
φ = Φ − W0 + J T, (11.214b)
4K1 r2 − r1 l
 
1 K1
r= r1 − r2 K1 J − 2
2 2
(r2 − r1 ) ρ ,
2
(11.214c)
1 − K12 J 2 l

where with {T, ρ, Φ} are denoted the corresponding BTZ coordinates, which
ought to be accompanied with the identification

1 + K12 J 2
J 2 K1 = −R(−) := − M l − M 2 l2 − J 2 , M = − . (11.215)
2 l K1

In this way this solution can be given in the standard representation as

ρ2 f (ρ) dρ2 2
g=− dT2 + + H(ρ) [dΦ + W (ρ)d T ] ,
H(ρ) f (ρ)
ρ2 J2
f (ρ) = − M + , R± := M l ± M 2 l2 − J 2 ,
l2 4ρ2
2ρ2 − lR− 2ρ2 − lR+  √ √
H(ρ) = √ √ −(2ρ2 − lR− ) α/2 (2ρ2 − lR+ )− α/2
4 αK1 M 2 l2 − J 2
√ √ 
+J 2 K12 (2ρ2 − lR− )− α/2 (2ρ2 − lR+ ) α/2 ,
R−  2 √ √
W (ρ) = (2ρ − lR+ ) α (2 α M 2 l2 − J 2 + R− )R−
Jl
√   √ √ −1
−(2ρ2 − lR− ) α J 2 × (2ρ2 − lR+ ) α R− 2
− (2ρ2 − lR− ) α J 2 .
(11.216)

When the electromagnetic field is turned off, c = 0 → α = 1, the above metric

components reduce to
 −1
ρ2 J ρ2 J2
gT T = M − , gT Φ = , gΦΦ = ρ2 , gρρ = −M + 2 , (11.217)
l2 2 l 2 4ρ

which correspond to the BTZ ones.

This solution possesses mass M , angular momentum J, electromagnetic
parameter α, and negative cosmological constant, and describes a black hole.
The metric defining this kind of stationary electromagnetic solution, see Ayón–
Beato et al. (2005) and Garcı́a (2009), can be given in the standard form as
194 Einstein–Maxwell Solutions

1 ρ2 f 2 dρ2
g = −N 2 d t2 + dρ 2
+ K 2
[dφ + W d t]2
= − dt +
L2 H f
2 2
ρ J
+H(dφ + W d t)2 , f (ρ) = 2 − M + 2 ,
l 4ρ
2ρ − lR− 2ρ − lR+
2 2  √ √
H(ρ) = √ −(2ρ2 − lR− ) α/2 (2ρ2 − lR+ )− α/2
4K1 αSM
√ √ 
+J K1 (2ρ2 − lR− )− α/2 (2ρ2 − lR+ ) α/2 ,
2 2

R−  2 √ √
W (ρ) = − (2ρ − lR+ ) α (2 αSM + R− )R−
Jl
√   √ √ −1
−(2ρ2 − lR− ) α J 2 × (2ρ2 − lR+ ) α R− 2
− (2ρ2 − lR− ) α J 2 ,
R−
K1 : = − , SM := M 2 l2 − J 2 . (11.218)
J2
The structural functions appearing in the definitions of the energy and momen-
tum quantities are expressed as

ρ2 f (ρ)
N (ρ) = , L(ρ) = f (ρ), K(ρ) = H(ρ), W (ρ) = W (ρ). (11.219)
H(ρ)

The corresponding electromagnetic tensors are given as

√
1−α
Fμ ν = − M 2 l2 − J 2 δ[μ t δν] φ ,
l
α − 1 M 2 l2 − J 2 T ν
8 π Tμ ν = [δμ δT − δμ ρ δρ ν + δμ Φ δΦ ν ]. (11.220)
4 l2 ρ2 f (ρ)
When the electromagnetic field is turned off, α = 1, the above metric components
reduce to
 2 −1
ρ2 J ρ J2
gT T = M − 2 , gT Φ = − , gΦΦ = ρ2 , gρρ = − M + ,
l 2 l2 4ρ2
which correspond to the BTZ ones.

11.7.2 Mass, Energy, and Momentum

In terms of the structural metric functions the momentum quantities they allow
for very simple expressions
J 1
j(ρ) = , J(ρ) = J, (11.221)
2π H(ρ)
while the energy and mass characteristics become

1 f (ρ) d
(ρ, 0 ) = − H(ρ) − 0 , (11.222)
2π H(ρ) dρ
 11.7 Ayón–Cataldo–Garcia Stationary Hybrid Solution 195

f (ρ) d
E(ρ, 0 ) = − H(ρ) − 2π0 H(ρ), (11.223)
H(ρ) dρ

f (ρ) d
M (ρ, 0 ) = −ρ H(ρ) − J W (ρ) − 2ρπ0 f (ρ), (11.224)
H(ρ) dρ
Because of the involved dependence of the metric functions upon the ρ coordi-
nate, the evaluation of the energy quantities will be done in the approximation
of the spatial infinity. The momentum density at infinity becomes
J α1/4
j(ρ → ∞) ≈ , (11.225)
2π ρ
while the global momentum remains constant in the whole space

J(ρ) = J. (11.226)

It becomes apparent then that the role of the momentum parameter is played
and coincides with J.
The approximated at ρ → ∞ surface energy density, global energy and mass,
for zero base energy density 0 are given by
√
1 lM α
(ρ → ∞, 0 = 0) ≈ − + ,
πl 2π ρ2
ρ l √
E(ρ → ∞, 0 = 0) ≈ −2 1/4 + 1/4
(1 + α)M,
lα 2 ρα
√
ρ2 α−1
M (ρ → ∞, 0 = 0) ≈ −2 2 + 2M + M 2 l2 − J 2 . (11.227)
l l
Using in the expressions (11.222)–(11.224) as reference energy density
 the
ρ2
energy corresponding to the anti-de Sitter metric, 0 = − πρ l2 − M0 ,
1

0|∞ (M0 ) ≈ − π1l + 2π

l M0
ρ2 , the series expansions of the global energy and mass
quantities at ρ → ∞ result in
l √
(ρ → ∞, 0|∞ (M0 )) ≈ ( αM − M0 ),
2π ρ2
√
( αM − M0 )
E(ρ → ∞, 0|∞ (M0 )) ≈ l ,
ρ α1/4
√
α−1
M (ρ → ∞, 0|∞ (M0 )) ≈ M − M0 + M 2 l2 − J 2 . (11.228)
l
Comparing with the energy characteristics of the BTZ solution, the mass param-
eter is an involved
√ quantity
√ depending on M , the momentum J, and the charge
α, namely M + α−1 M 2 l2 − J 2 , although the mass function is finite at spatial
l √
infinity. On the other hand, if M l >> J then αM becomes the mass param-
eter. The energy density and global energy are proportional at infinity to 1/ρ2
and 1/ρ correspondingly.
196 Einstein–Maxwell Solutions

Field, Energy–Momentum, and Cotton Tensors

The coordinate system {t, r, φ} is more adequate in the derivation of the
eigenvalue–vector characteristics of the considered solution of Garcı́a
(2009), Eq.(8.11). The metric in {t, r, φ}-coordinates is given by
⎡ ⎤
−F/H + H W 2 0 HW
⎢ ⎥
g=⎢
⎣ 0 1/F 0 ⎥,
⎦ (11.229)
HW 0 H

with structural functions

(r − r1) (r − r2)
F (r) = 4 ,
l2  1/2 √α
l (r − r1) (r − r2) r − r1
H(r) = √
2 K1 α (r2 − r1) r − r2
 √
−1/2 α
r − r1
−J 2 K12 ,
r − r2
 −1/2 √α
r − r1 (r − r1) (r − r2)
W (r) = W0 − 2 K1 J . (11.230)
r − r2 l H(r)

The electromagnetic field tensor is given by

⎡ ⎤
−c HFW 0 −c FH
⎢ ⎥ √
⎢ ⎥ (r2 − r1) 1 − α
(F α β ) = ⎢ 0 0 0 ⎥, c = . (11.231)
⎣ ⎦ l2
(F −H W 2 2
)
−c HF 0 c HW
F

In the search of its eigenvectors, one arrives at

2
V2
λ1 = 0; V1 = [0, V 2 , 0], Vμ V μ = , V1 = S1,
F
c
λ2 = √ ;
F
√
HW + F V1
V2 = [V , 0, −
1
], V μ Vμ = 0, V2 = N2,
H
c
λ3 = − √ ;
F
√
−HW + F V1
1
V3 = [V , 0, ], V μ Vμ = 0, V3 = N3, (11.232)
H

hence this tensor is of the type {S, N, N }.

 11.7 Ayón–Cataldo–Garcia Stationary Hybrid Solution 197

For the energy–momentum tensor

⎡ ⎤
c2
−1 0 0
⎢ 8π F
⎥
⎢ 1 c2 ⎥
(T α β ) = ⎢ 0 0 ⎥, (11.233)
⎣ 8π F ⎦
2
0 0 − 8π
1 c
F

the eigenvectors are

2
1 c2 V2
λ1 = ; V1 = [0, V 2 , 0], Vμ V μ = , V1 = S1,
8π F F
1 c2
λ2,3 = − ; V2, 3 = [V 1 , 0, V 3 ],
8π F
2  
V1 F  1 
2
3 2 V 1 F Z2 − 1
Vμ V = −
μ
+ V W +V H= ,
H H
V2 = {T2, N2, S2}, V3 = {T3, N3, S3}, (11.234)

hence it allows for the types: {S, 2T }, {S, 2N }, {S, 2S}.

The Cotton tensor is determined through
⎡ 1 ⎤
C 1 0 C 13
⎢ ⎥
(C α β ) = ⎢
⎣ 0 0 0 ⎥,
⎦
C 31 0 −C 1 1
 
c2 FJ
C 1
1 = −C 3
3 =− W Q+2 ,
32πF 2 H
c2
C 13 = − Q,
32πF 2
   
c2 FJ
C 31 = − −W H 2
W Q + 4 − F Q ,
32πF 2 H 2 H
d d
Q := 2 F H − H F;
dr dr  1/2 √α
2 r − r1
Q = − (r − r1) (r − r2)
lK1 r − r2
 −1/2 √α
2 2 r − r1
+J K1 , (11.235)
r − r2

possesses the following set of eigenvectors

2
λ1 = 0; V1 = [0, V 2 , 0], Vμ V μ = V 2 /F, V1 = S1,
√ 2
1 αc (r2 − r1)
λ2 = i ;
4 π F 3/2 l2
 −1
c2 J F
Z = [−c2 V 3 Q c2 W Q + 2 + 32 λ2 F 2 π , 0, V 3 ],
H
198 Einstein–Maxwell Solutions
√
1 αc2 (r2 − r1)
λ3 = − i ;
4 π F 3/2 l2
 −1
c2 J F
Z̄ = [−c V Q c W Q + 2
2 3 2 2
+ 32 λ3 F π , 0, V 3 ]. (11.236)
H
therefore its type is Type I: {S, Z, Z̄}.

11.7.3 Constant Electromagnetic Invariants’ Hybrid Solution

for Λ = 0
This section is devoted to the studied of the hybrid electromagnetic stationary
solution with constant electromagnetic invariant F F and by virtue of the field
2
structure, constant T and T T . The constant character of F F = − F2c(r) is achieved
by requiring F (r) = F0 , and consequently all electromagnetic invariants equal
to constants
2c2 3 c4 c2
FF = − , TT = , T μ
μ
= − . (11.237)
F0 64 π 2 F02 8π F0
Again the integration of the Einstein equations start from Eφ t (a = 0 = b) = 0,
which gives the relation
d J
W (r) = (11.238)
dr H(r)2
d
for the integration of the function H(r) the substitution of dr W (r) and
F (r) = F0 into the remaining Einstein equations requires the cosmological
constant to vanish, Λ = 0. Under such conditions, the equation for H(r) becomes
 2
d H(r)
F0 − J 2 + 4 c2 F0 H(r)2 = 0
dr
with solution
J 2c
H(r) = H F0 sin (r − C0 ), H = ±1, (11.239)
2c F0
which, used in (11.238), after integration yields
2c 2c
W (r) = W0 + W cot (r − C0 ), W = ±1, (11.240)
J F0
where H and W assume their signs independently; one has to take care on
the ranges of the variable r to guarantee a correct signature. Moreover, notice
that the integration constant C0 can be always equated to zero. Therefore the
corresponding metric and electromagnetic field vector amount to
F0 2 dr2
g=− dt + + H(dφ + W dt)2 ,
H F0
J 2c 2c 2c
H = H F0 sin r, W = W0 + W cot r,
2c F0 J F0
c
A = (tdφ − φdt). (11.241)
2
 11.8 Stationary Solutions for a = 0 or b = 0 199

The electromagnetic field tensors are

Fμν = 2cδμ [t δν φ] ,
c2
Tμ ν = [−δμ t δt ν + δμ r δr ν − δμ φ δφ ν ]. (11.242)
8π F0
By means of scaling transformations F0 can always be set equal to unit, F0 = 1,
hence this solution is endowed with two effective parameters c and J.

11.8 Stationary Solutions for a = 0 or b = 0

This section deals with the search of stationary solutions for the branches where
one of the electromagnetic constants is zero, a = 0 = b or b = 0 = a. For these
families, the integration problem reduces to finding the solution of a master
fourth-order (reducible to a third-order) nonlinear equation for F (r), and to
fitting a differential constraint on the found structural functions F (r) and H(r)
(H(r)). The integration of W (r) (W(r)) is trivial.

11.8.1 Stationary Magneto-Electric Solution for a = 0 = b

If the structural function W (r) is different from a constant (the constant case
will be treated at the end of this paragraph) then Eφ t reads
 
d d
Et φ = H 2 W = 0, (11.243)
dr dr
which yields
d J
W = 2. (11.244)
dr H
The remaining independent Einstein–Maxwell equations arise respectively from
combinations (4 Er r + 2Et t + 2Eφ φ ), (−2H(Er r − Et t + W Et φ )/F ), and Er r :

d2 F H 1
EQF = − 4 a2 − 8 2 = 0,
d r2 F l
d2 H2
EQH = 2 H + 4a2 2 = 0,
dr F
1 dH dF F dH 2 J2 H 1
Er r = − 2
( ) + − a2 − 2 = 0. (11.245)
4H d r d r 4H d r 4 H2 F l
The equation Er r can be written in the form
 2  2
1 dH 1 dF 1 dF a2 H 1 J2 1
− − 2
+ 2 − + 2 = 0.
2H dr 4F dr 16 F dr F 4 F H2 l F
(11.246)
200 Einstein–Maxwell Solutions

On the other hand, using EQF one expresses H in terms of F and its derivative
 2 
1 d F 8
H(r) = − 2 F. (11.247)
4 a2 dr2 l
Substituting the above H(r) into EQH (11.245) one gets
 2 2
d4 F d3 F dF d F 24 d2 F 64
F + 2 + 2 − + 4 = 0. (11.248)
dr4 dr3 dr dr2 l2 dr2 l
Therefore, integrating, if possible, (11.248) for F (r), substituting the solution
F (r) into (11.247) one determines H(r). The resulting functions F (r) and H(r)
ought to fulfil the (11.246) or Er r equation from (11.245). By integrating the
linear first-order (11.244) one determines W (r).
The contravariant components of electromagnetic tensor are
F μν = −2 a δ μ [φ δ ν r] . (11.249)
The equation (11.248) for F (r) can be reduced to a third-order nonlinear equa-
tion. The problem for deriving solutions in this branch actually resides in this
equation.
Another possibility arises with the introduction of the auxiliary function h(r)
by means of
H(r) = F (r)1/2 h(r), (11.250)
the (11.246) acquires the form
2   2
dF dh
EQh = −l h 2 2 2 2
+ 4l F
dr dr
√
+16 l a h F − 4 l J + 16 h2 F = 0
2 2 3 2 2
(11.251)
and one could try to determine solutions for this variant.

“Stationary” Magneto-Electric A = A(r)(dφ − J0 dt) Solution

A particular solution to (11.248) is given by F (r) from (11.76), namely
4 h(r)
F (r) = K0 + h(r) + a2 l2 ln h(r) , h(r) := C1 r + C0 (11.252)
C1 2 l2
which, being substituted into (11.247), leads to
4
H(r) = K0 + h(r) + a2 l2 ln h(r) . (11.253)
C1 2 l2
Entering with these particular solutions F (r) and H(r) in the constraint (11.246)
one arrives at
W (r) = −J0 = constant. (11.254)
 11.8 Stationary Solutions for a = 0 or b = 0 201

Summarizing, this solution is given by the same structural functions (11.76) of

the magnetostatic solution except that in the present case the function W (r)
is a constant. The corresponding metric line element and field vector can be
written as
dr2
g = −h(r)dt2 + + H(r)(dφ − J0 d t)2 ,
H(r) h(r)
a
A= ln h(r)(dφ − J0 d t). (11.255)
C1
The electromagnetic field tensors and their invariants are given by
a2
F μν = 2 a δ μ [r δ ν φ] , F F = 2 , (11.256)
h
and
a2  t μ  3 a4
Tν μ = −δν δt + δνr δrμ + δνφ δφμ − 2 J0 δνt δφμ , T = . (11.257)
8π h 64π 2 h2
Because of the structure of the energy–momentum tensor above, this solution
can be interpreted as a rigidly rotating perfect fluid
1
Tμ ν = (ρ + p)uμ uν + p gμν , uμ = (δ μ t + J0 δ μ φ ), (11.258)
F/H
with energy density ρ and pressure p given by
1 a2
ρ= = p.
8π h
This solution can be generated via transformations t → t, φ → φ − J0 t from
the magnetostatic solution (11.76).

Clément “Rotating” Electromagnetic A = A(r)(dφ + ω0 dt) Solution

Clément (1993) published the dual family of electromagnetic “stationary” cyclic
symmetric solutions, Eq. (24), changing signature and VCl → −V , given by
1 dρ2
g = V (dφ + ω0 dt)2 + − 2ρdt2 ,
ξ02 2ρV
π12 ρ
V = −2Λρ + ln( ),
4m ρ0
π1 ρ
A = − ln( )(dφ + ω0 dt), (11.259)
2 ρ0
where m, π1 , ξ0 and ρ0 are constants, Λ = ±1/l2 stands for the cosmological
constant of both signs; for anti-de Sitter Λ = −1/l2 . The parameter ω0 is related
to the angular momentum constant.
It is worth noticing that the Clément expressions (11.259) satisfy the (2+1)
Einstein–Maxwell equations if ξ02 = 1 and for 2 m = 1/κ; for the adopted in
the Clément’s convention, κ = 1, Gμν + Λ gμν = 4πκ Tμν , the evaluation of the
202 Einstein–Maxwell Solutions

right-hand side of the Einstein equations for the structural functions (11.259),
for ξ02 = 1, yields

π12
Gν μ = [−δνt δtμ + δνr δrμ + δνφ δφμ + 2 ω0 δνt δφμ ],
8mρ
while the right-hand side amounts to

κ π12
4π κ Tν μ = [−δνt δtμ + δνr δrμ + δνφ δφμ + 2 ω0 δνt δφμ ]
4ρ
hence 2m = 1/κ.
If one were adopting κ = 1, then modifying the electromagnetic vector A to
be Amod = − 2√π21 m ln( ρρ0 )(dφ + ω0 dt), one would arrive at the solution in our
convention.
It is apparent that these Clément’s solutions correspond to a variant of
the solution derived in the previous Section (11.8.1), with the identification
r → ρ followed by minor scaling transformations of t and φ. Notice also that
the above generalization with W (r) = ω0 = 0 of the magnetostatic solu-
tion (11.76) can be determined applying to it SL(2, R) transformations of the
form φ → φ + ω0 t, t → t.

11.8.2 Stationary Electromagnetic Solution for b = 0 = a

A straightforward way to derive the equations and solutions of this class of
fields is just by using the complex extension of the stationary magnetic field
we derived in the previous subsection taking into account the specific structure
of the functions (11.30) of the extended metric (11.29) and the metric compo-
nents from (11.255) of the magnetic solution together with the formal change
a2 → −b2 .
Another close possibility is to accomplish the substitution
F HW
F = F, H = − HW 2 , W = , (11.260)
H H
in the corresponding Einstein equations for this case b = 0 = a, arriving at the
following set of independent field equations

d2 F H 1
EQF = + 4 b2 − 8 2 = 0,
d r2 F l
d2 H4
EQH = 2 H2 − 4b2 2 = 0,
dr F
1 d H d F F dH 2 1 H 1
Err = − ( ) + J 2 + b2 − 2 = 0,
4H d r d r 2
4H d r 4H 2 F l
d J
W = 2. (11.261)
dr H
 11.8 Stationary Solutions for a = 0 or b = 0 203

Continuing with the parallelism, isolating H from EQF and replacing it into
EQH one obtains
 2 2
d4 F d3 F dF d F 24 d2 F 64
F + 2 + 2 − + 4 = 0. (11.262)
dr4 dr3 dr dr2 l2 dr2 l
Thus, as before, the first step in the integration of the problem depends upon
the (11.262) for F(r), structurally identical to (11.248). Substituting the solution
F(r) into EQH from (11.261) one determines H. The resulting functions F(r)
and H have to fulfill Er r from (11.261). By integrating the linear first-order
equation for W one determines W(r).

“Stationary” Electromagnetic A = A(r)(dt + J0 dφ) Solution

Hitherto, the only known particular solution for F(r) of (11.262) and its
corresponding solutions for H and W is
4
F= K0 + h(r) − b2 l2 ln h(r) h(r), h(r) := C1 r + C0 ,
C1 2 l2
F
H = , W = −J0 = constant. (11.263)
h
Substituting these expressions into (11.260), one gets
H
F = F = Hh, H = h − HJ0 2 , W = −J0 ,
h − HJ0 2
4
H := K0 + h(r) − b2 l2 ln h(r) , (11.264)
C1 2 l2
therefore, the corresponding metric and field vector can be written as
dr2
g = −H(dt + J0 dφ)2 + + h(r)dφ2 ,
H h(r)
b
A= ln h(r)(dt + J0 dφ). (11.265)
C1
The electromagnetic field tensors and their invariants are given by
b J0 b2
F μν = 2 b δ μ [t δ ν r] , Fμν = −2 δμ [t δν r] − 2 b δμ [φ δν r] , F F = −2 ,
h(r) h(r) h
2 4
b 3 b
Tν μ = [−δνt δtμ − δνr δrμ + δνφ δφμ − 2 J0 δνt δφμ ], T T = .
8π h 64π 2 h2
As we shall see in the forthcoming section, this stationary electromagnetic
solution can be generated via transformations t → t + J0 φ, φ → φ from the
electrostatic solution (11.47).

Clément “Rotating” Electromagnetic A = A(r)(dt − ω0 dφ) Solution

Clément (1993) also published a class of electromagnetic “stationary” cyclic
symmetric metrics, Eq. (23), changing signature, given by
204 Einstein–Maxwell Solutions

1 dρ2
g = −U (dt − ω0 dφ)2 + + 2ρdφ2 ,
ξ02 2ρU
π02 ρ
U = −2Λρ − ln( ),
4m ρ0
π0 ρ
A= ln( )(dt − ω0 dφ), (11.266)
2 ρ0
where m, π0 , ξ0 and ρ0 are constant parameters, Λ = ±1/l2 stands for the cosmo-
logical constant of both signs; for anti-de Sitter Λ = −1/l2 . The parameter ω0 is
a constant related to the angular momentum. The evaluation of the right-hand
side of the Einstein equations for the structural functions (11.266), for ξ02 = 1,
yields
π2
Gν μ = − 0 [δνt δtμ + δνr δrμ − δνφ δφμ − 2 ω0 δνφ δtμ ],
8mr
while the energy–momentum tensor in the left-hand side, for the vector A,
amounts to
π2
4π Tν μ = − 0 [δνt δtμ + δνr δrμ − δνφ δφμ − 2 ω0 δνφ δtμ ],
4r
therefore Einstein–Maxwell equations are fulfilled if 2m = 1/κ or, for κ = 1,
modifying the electromagnetic vector A to be Amod = 2√π20 m ln( ρρ0 )(dt − ω0 dφ).
Recall that additionally one has to set ξ02 = 1.
It is clear that this solution is equivalent to the one treated in Section 11.8.2 for
the identification C1 r +C0 → ρ accompanied with minor scaling transformations
of t and φ.
Notice that this branch of rotating solutions with W (r) = ω0 can be deter-
mined from the static electric field solution, i.e., metric (11.47) and vector A
(11.48), via SL(2, R) transformations: t → t − ω0 φ, φ → φ.

Constant W Electric Solution

In the case W (r) = −J = constant the equation Eφ t , from (11.26), reduces to
b2 H J/F = 0, then J = 0 → W = 0. Hence, the set of equations reduces to the
corresponding one of the static case.

11.9 Garcı́a Stationary Solutions for a = 0 and b = 0

It is clear that the derivation of a general solution to the whole system of
Einstein–Maxwell equations (11.26) is far from being an easy task. Nevertheless,
some simplifications of the system of equations can be achieved by a useful change
of the structural functions and combinations of the Einstein equations; the inte-
gration problem on the whole for the three structural functions is constrained to
three differential equations without any further restrictions. Although we could
not find sufficiently general classes of solutions, we were able to determine new
families of solutions within particular combinations of elementary functions.
 11.9 Garcı́a Stationary Solutions for a = 0 and b = 0 205

11.9.1 Alternative Representation of the Einstein Equations

Having in mind the derivation of other possible families of Einstein–Maxwell
solutions with a and b different from zero, it is desirable to have at hand the
most simple set of equations. For this purpose, introducing W (r) = Ω(r)/H(r),
the independent Einstein equations can be written as;
d2
EQH2 = F 2 H + 4(a H − bΩ)2 = 0, (11.267)
dr2
d2 Ω
EQΩ2 = H F 2 2 + 4 Ω(a H − bΩ)2 + 4b F (a H − bΩ) = 0, (11.268)
dr
 2
dF dH dH 2
EQF 1 =F H − F2 − 4 H (a H − bΩ)
dr dr dr
 2
dΩ dH
+F H −Ω + 4F H(b2 − H/l2 ) = 0, (11.269)
dr dr
 2
d2 F dH dF dH
EQF 2 = Ω H2 2 − 2 Ω H + 2Ω F
dr dr dr dr
 2
dΩ dH
−2 Ω H −Ω = 0. (11.270)
dr dr
It is worth pointing out that the equation EQF 2 can be considered as an integra-
bility condition of the system of equations; differentiating the EQF 1 one obtains
2
the second derivative ddrF2 together with second derivatives of H and Ω, which
d2 F
can be replaced through EQH2 and EQ ; next substituting
 Ω2dF  dr 2 into EQF 2
one arrives at an equation of the form H dr + F dH dr × EQ F 1 , equal to zero
by virtue of the same EQF 1 . Although one can adopt a different point of view;
the EQF 2 equation arises as the differentiation of EQF 1 together with the use
of EQH2 and EQΩ2 , and therefore it is not an independent equation.
Using the experience gathered until now, we shall search for particular
solutions of the form

F (r) = P (r) + Q(r) ln(r),

H(r) = A(r) + B(r) ln(r),
W (r) = Ω(r)/H(r),
Ω(r) = V (r) + Z(r) ln(r), (11.271)

where it is assumed the explicit dependence on ln(r). Substituting these guessed

functions into the quoted system of equations and equating to zero the coeffi-
cients of different powers of ln(r), one arrives at a very large nonlinear system
of equations; since there are five nontrivial Einstein equations, then one may
expect 40 secondary equations. For instance, from equations arising from the
coefficients of ln(r) to the seventh power, one has
206 Einstein–Maxwell Solutions

dB dZ 2 d2 B d2 Z
Ett ln7 = B 2 Q l2 (Z −B ) − 2 B 3 Z Q l2 (Z 2 − B 2 ),
dr dr dr dr
d2 B d2 Z
Eφφ ln7 + 3Err ln7 = 2 l2 B 3 Q(Z − B ),
dr2 dr2
hence
d2 B d2 Z dB dZ
Z 2
− B 2
= 0, Z −B = 0,
dr dr dr dr
therefore
Z(r) = c1 B(r).
After a very lengthy and time-consuming integration process we succeeded in
getting two branches of stationary electromagnetic solutions of the Einstein–
Maxwell equations. The structural functions H and W possess a multiplicative
factor a/b which can be absorbed by re–scaling of the Killingian coordinates
according to: |a/b|t → t and |b/a|φ → φ, |a| |b| = ±α.

11.9.2 Garcı́a Stationary Electromagnetic Solution with BTZ limit

This class of solution, depending on three parameters, is given by
F (r) 2 dr2 2
g = − dt + + H(r) [dφ + W (r)dt] ,
H(r) F (r)
  
r2 r
F (r) = 4 2 + 2 l w1 + l w1 − 4 [w0 + W0 ln(r)] ,
2 2
l l

r 2 2
H(r) = l w1 − 4 − [w0 + W0 ln(r)] ,
l
W (r) = Ω(r)/H(r), Ω(r) := w0 + w1 r + W0 ln(r),
  
1
W0 := − l2 α2 l2 w12 − 2 − l w1 l2 w12 − 4 ,
2
  
1
A = α l w1 − l2 w12 − 4 (dt − dφ) ln(r). (11.272)
4
and electromagnetic tensors
 
F μν = 2 α δ μ [t δ ν r] − δ μ [φ δ ν r] , (11.273)
⎡ 2
(1−W 2 )] 2[F +H 2 W (1−W )]
⎤
− [F +H C 0 C
⎢ ⎥
⎢ ⎥
− [F −H
2
(1−W )2 ]
(T μ ν ) = ⎢ 0 0 ⎥, (11.274)
⎣ C ⎦
−2H 2 (1−W ) [F +H 2 (1−W 2 )]
C 0 C
8π F H
where C = α2 . The electromagnetic invariants are
α2 α2
F F = −2 + 2 H (1 − W )2 ,
H F
 11.9 Garcı́a Stationary Solutions for a = 0 and b = 0 207

and
3 α4 1
TT = [−F + H 2 (1 − W )2 ]2 .
64 π 2 F 2 H 2
Since this solution uses the BTZ solution as a limit for α = 0, it is natural
to search for new coordinates in which the BTZ standard structure will become
apparent. First one determines the radial transformation r = β0 (ρ2 + γ0 ); since
grr → gρρ , then
  
r2 r ρ4 J2
F (r) = 4 2 + 2 l w1 + l w1 − 4 w0 → F (ρ) = 2 − M ρ2 +
2 2 ,
l l l 4
hence
Ml 1
γ02 /l2 + γ0 M + J 2 /4 = 0 → γ0 /l = − ∓ l2 M 2 − J 2 ,
  2 2

w0 l w1 + l2 w12 − 4 = ± 2 β l2 M 2 − J 2 .

Next, the structure of the Killingian transformations is of the form

t = αt T + βt Φ, φ = δt Φ.
Substituting these relations into the metric and comparing the metric compo-
nents with the corresponding ones of the BTZ-metric one establishes that
1 M
w0 = l2 M 2 − J 2 −l M + l2 M 2 − J 2 , β = −1, w1 = 2 . (11.275)
J J
Therefore, the coordinate transformations to be used in the electromagnetic solu-
tion in order to get the proper BTZ limit when the electromagnetic α–parameter
is switched off is given by
M l2 l
r = −ρ2 + + l2 M 2 − J 2 ,
2 2  

1 2 2 
2 −1/4 J l
t= √ l M −J T −lM Φ ,
2 l J

1 l 2 2 1/4
φ= √ l M − J2 Φ,
2 J
√
α = Q J. (11.276)
Under these transformations the metric, in its canonical representation, becomes
1
ds2 = −N (ρ)2 dT 2 + dρ2 + K(ρ)2 [dΦ + W (ρ)dT ]2 ,
L(ρ)2
Hn
H(ρ) := ;
Hd
Hn = 4 ρ2 (ρ2 − M l2 )(M 2 l2 − J 2 ) − J 2 Q4 l6 R−
2
(ln | Z(ρ) |)2
−2Q2 l3 M 2 l2 − J 2 [M J 2 l2 − 2 ρ2 R− M 2 l2 − J 2 ] ln | Z(ρ) |,
Hd = 4 (M 2 l2 − J 2 )(ρ2 − M l2 ) − 2 l3 Q2 R−
2
M 2 l2 − J 2 ln | Z(ρ) |,
208 Einstein–Maxwell Solutions

ρ2 J2 l Q2
L(ρ)2 = − M + + (2ρ2 R− − l J 2 ) ln | Z(ρ) |,
l2 4 ρ2 2 ρ2
K(ρ)2 = H(ρ),
L(ρ)2
N (ρ)2 = ρ2 ,
H(ρ)
W (ρ) Hn = J Q4 l5 R−
3
(ln | Z(ρ) |)2 − 2 J (M 2 l2 − J 2 )(ρ2 − M l2 ),
+Q2 l2 J M 2 l2 − J 2 [J 2 l + 2 l R−
2
− 2ρ2 R− ] ln | Z(ρ) |
l l R−
Z(ρ) := ρ2 − (M l − M 2 l2 − J 2 ) = ρ2 − ,
2 2
R± := M l ± M 2 l2 − J 2 , (11.277)

with electromagnetic vector potential

√  
lQ ln | ρ | J
A := − √ √ R− dt − dφ , (11.278)
2 4 l2 M 2 − J 2 l
therefore the non-vanishing covariant components of the electromagnetic field
tensor Fμν are
√
lQ R− 1 Q J 1 1
Ftρ = √ √ , Fφρ = √ √ √ . (11.279)
2 l M −J
4 2 2 2 ρ 2 l l M −J
4 2 2 2 ρ
Notice that the above gravitational–electromagnetic field, as was pointed out
previously, when the electromagnetic field is switched off, Q = 0, becomes the
rotating BTZ solution, while for vanishing rotation, J = 0, the correspond-
ing solution is represented by the static BTZ metric, i.e., the AdS metric with
M -parameter.
Notice that (gtt gφφ − gtφ
2
) = −ρ2 F (ρ). The correspondence of this function
representation of this electromagnetic solution with the BTZ solution in the limit
of vanishing electromagnetic parameter α becomes apparent:
ρ2 J2 J
F (ρ) = − M + , H(ρ) = ρ2 , W (ρ) = − 2 .
l2 4ρ2 2ρ
Thus, this anti-de Sitter solution has three parameters: mass M , angular momen-
tum J, and electromagnetic parameter α. Because of its close similarity to the
BTZ solution, it could represent a black hole.

Mass, Energy and Momentum

The evaluation of the surface energy density at spatial infinity ρ → ∞ for 0 = 0
yields
1 lM l2 J 2 Q2
(ρ → ∞, 0 = 0) ≈ − + − √
πl 2πρ2 2πρ2 M 2 l2 − J 2
l3 Q2 M
+ √ R− ln (ρ), (11.280)
π ρ2 M 2 l2 − J 2
 11.9 Garcı́a Stationary Solutions for a = 0 and b = 0 209

while the momentum quantities amount to

J l2 JQ2 l2 JQ2 R−
j(ρ → ∞) ≈ − √ R− + √ ln (ρ)
2π ρ 2πρ M 2 l2 − J 2 π ρ M 2 l2 − J 2
l2 JQ2 2 l2 JQ2 R−
J(ρ → ∞) ≈ J − √ R− + √ ln (ρ). (11.281)
M 2 l2 − J 2 M 2 l2 − J 2
The integral energy and mass characteristic at spatial infinity can be evaluated
from the generic expressions E(ρ) = 2 π K (ρ), M (ρ) = N E(ρ) − W J(ρ).
The behavior of the corresponding functions with 0 = 0 as ρ → ∞ is

2ρ l M l2 J 2 Q2
E(ρ → ∞, 0 = 0) ≈ − + − √
l ρ ρ M 2 l2 − J 2
2 2
l Q
+ √ R− 2
ln (ρ),
ρ M 2 l2 − J 2
ρ2 l J 2 Q2
M (ρ → ∞, 0 = 0) ≈ −2 2 + 2 M − √
l M 2 l2 − J 2
2 l Q2
+√ R−2
ln (ρ). (11.282)
M 2 l2 − J 2
The series expansions of the expressions
 of the energy and mass evaluated for
2
the reference energy density 0 = − πρ −M0 + ρl2 , which at the spatial infinity
1

behaves as 0|∞ ≈ − π1l + l M0

2π ρ2 , at ρ → ∞ occur to be

l l2 J 2 Q2
(ρ → ∞, 0|∞ ) ≈ (M − M 0 ) − √
2πρ2 2πρ2 M 2 l2 − J 2
l3 Q2 M R−
+ √ ln (ρ),
π ρ2 M 2 l2 − J 2
l(M − M0 ) l2 J 2 Q2
E(ρ → ∞, 0|∞ ) ≈ − √
ρ ρ M 2 l2 − J 2
3 2
2l Q M
+ √ R− ln (ρ),
ρ M 2 l2 − J 2
l J 2 Q2
M (ρ → ∞, 0|∞ ) ≈ M − M0 − √
M 2 l2 − J 2
2 2
2l M Q
+√ R− ln (ρ). (11.283)
M 2 l2 − J 2
For vanishing electromagnetic field charge Q, which gives rise to the rotat-
ing BTZ black hole, the mass, and the energy–momentum quantities become
just the mass–energy–momentum characteristics of the BTZ solution; hence one
concludes that the parameters M and J are related to the mass and momentum
respectively. Moreover, in the electromagnetic case, the momentum, mass, energy
functions logarithmically diverges at spatial infinity.
210 Einstein–Maxwell Solutions

Field, Energy–Momentum, and Cotton Tensors

To determine the eigenvector structure of the Garcı́a solution it is more
convenient to use another of its representation in the coordinates {τ, r, σ}, namely
⎡ ⎤
−F/H + H 2 0 HW
⎢ ⎥
g=⎢ ⎣ 0 1/F 0 ⎥ ⎦, (11.284)
HW 0 H
where the metric functions are
r2 r
F (r) = 4 +2 lw1 + l2 w1 2 − 4 (w0 + W0 ln (r)),
l2 l
r 2 2
H(r) = −w0 − W0 ln (r) + l w1 − 4,
l
(w0 + W0 ln (r) + w1 r)
W (r) =  √ ,
−w0 − W0 ln (r) + rl l2 w1 2 − 4
l 2 α2 2
W0 = − lw1 − l2 w1 2 − 4 . (11.285)
4
To achieve the metric structure studied in the previous paragraph, one subjects
the above metric to the coordinate transformation
 
1 J t − l2 M φ l2 M l
τ=√ √ , r = −ρ2
+ + l2 M 2 − J 2 ,
2 Jl l M − J
4 2 2 2 2 2
√
l 4 2 2
σ=√ √ l M − J 2 φ,
2 J
together with
√
l2 M 2 − J 2 M l 2 α2 2
w0 = − R − , w 1 = 2 , W0 = − 2 R − ,
J J J
R− = lM − l2 M 2 − J 2 .

followed by the change of the charge α → J 1/2 Q.

In these coordinates, the Maxwell electromagnetic field tensor is given by
⎡ ⎤
0 α/F 0
⎢ ⎥
(F α β ) = ⎢
⎣ α F − H W (W − 1) /H
2
0 −Hα (W − 1) ⎥ ⎦ , (11.286)
0 −α/F 0
and is characterized by the following set of eigenvectors
H 2 (W − 1)
λ1 = 0; V1 = [ V 3 , 0, V 3 ],
F − H 2 W (W − 1)
2
HF F − H 2 (W − 1) 2
μ
Vμ V = 2 V 3 , V1 = T1, S1,
[F − H 2W (W − 1)]
F− H 2 (W − 1)2
λ2 = √ α;
HF
 11.9 Garcı́a Stationary Solutions for a = 0 and b = 0 211

αV 2 αV2
V2 = [ , V 2, − ], V μ Vμ = 0, V2 = N2, Z,
λ2 F (r) λ2 F
F − H 2 (W − 1)2
λ3 = − √ α;
HF
αV 2 2 αV2
V3 = [ ,V ,− ], V μ Vμ = 0, V3 = N3, Z̄. (11.287)
λ3 F λ3 F

This kind of tensor exhibits an enormous variety of types.

For the Maxwell energy–momentum tensor matrix (T α β ) is given as
⎡ α2 [F −H 2 (W 2 −1)]
⎤
1 Hα (W −1)
2
− 8π
1
0
⎢ FH 4π F ⎥
⎢ ⎥
⎢ 1 α [F −H (W −1) ]
− 8π
2 2 2
⎥,
⎢ 0 0 ⎥
⎣ FH
⎦
1 α [F −H W (W −1)] 1 α [F −H (W −1)(W +1)]
2 2 2 2

4π FH 0 8π FH
(11.288)
one has the following eigenvalues and eigenvectors
 
2
1 α F − H (W − 1)
2 2

λ1,2 = − ; V1, 2 = [−V 3 , V 2 , V 3 ],

8π FH
2 2 2
V2 V 3 [F − H 2 (W − 1) ]
Vμ V μ = − ,
F H
V1 = T1, S1, V2 = T2, S2,
 
2
1 α 2
F − H 2
(W − 1)
λ3 = ;
8π FH
H 2 V 3 (W − 1)
V3 = [ , 0, V 3 ],
F − H 2 W (W − 1)
 
2 2
V 3 HF F − H 2 (W − 1)
Vμ V μ = − 2 ,
[F − H 2 W (W − 1)]
V3 = T3, S3. (11.289)

Therefore one has a quite big choice of algebraic types of this tensor.
Finally, the Cotton tensor can be given as
⎡ ⎤
C 11 0 C 13
⎢ ⎥
(C α β ) = ⎢
⎣ 0 C 22 0 ⎥,
⎦ (11.290)
C 31 0 −C 1 1 − C 2 2
2
1 HW α2 (W − 1) (HF,r − F H,r ) 1 α2 (W − 1) F,r
C 11 = − +
32 π F2 32 π F
1 α2 (3 W − 2) H,r 1
− − α2 W,r , (11.291)
32 π H 16 π
212 Einstein–Maxwell Solutions

with components
2
1 α2 H (W − 1) (H F,r − F H,r ) 1 α2 H,r
C 13 = − 2
− , (11.292)
32 π F 32 π H
1 α2 (W − 1) (F,r H − 2 F H,r )
C 22 =−
16 π FH
2
1 α H (W − 1) + F W,r
2 2

+ , (11.293)
16 π  F 
1 α2 (W − 1) −F (1 + W ) + W 2 H 2 (W − 1) F,r
C 31 =
32 π F2
  2
1 F + H 2 W 2 α2 −F + H 2 (W − 1) H,r
− ,
32 π H 3F
2
1 W α −F + H (W − 1) W,r
2 2

− , (11.294)
16 π F
2 2
1 H 2 W α2 (W − 1) F,r 1 α2 H 2 (W − 1) W,r
C 33 = 2
− ,
32 π F 16 π F
2
1 α H W (W − 1) + F (W − 2) H,r
2 2

− (11.295)
32 π FH
possesses, in general, three different eigenvalues, with the possibility of complex
conjugated roots, namely
λ1 = C 2 2 ,
λ2 = −1/2 C 2 2 + 1/2 (C 1 1 + C 2 2 )2 + 4 C 1 3 C 3 1 ,
λ3 = −1/2 C 2 2 − 1/2 (C 1 1 + C 2 2 )2 + 4 C 1 3 C 3 1 ]. (11.296)
The set of eigenvector equations reduces to
V 1 (C 1 1 − λ) + C 1 3 V 3 = 0,
 2 
C 2 − λ V 2 = 0,
C 3 1 V 1 − V 3 (C 1 1 + C 2 2 + λ) = 0 (11.297)
with solutions
2
λ1 = C 2 2 ; V1 = [0, V 2 , 0], Vμ V μ = V 2 /F, V1 = S1,
λ2 = −1/2 C 2 2 + 1/2 (C 1 1 + C 2 2 )2 + 4 C 1 3 C 3 1 ;
C 13 V 3
V2 = [− 1 , 0, V 3 ], V μ Vμ = {0, > 0, < 0},
C 1 − λ2
V2 = S2, N2, Z,
λ3 = −1/2 C 2 2 − 1/2 (C 1 1 + C 2 2 )2 + 4 C 1 3 C 3 1 ;
C 13 V 3
V3 = [− 1 , 0, V 3 ], V μ Vμ = {0, > 0, < 0},
C 1 − λ3
V3 = S3, N3, Z̄. (11.298)
 11.9 Garcı́a Stationary Solutions for a = 0 and b = 0 213

In general, one is dealing with a Type I Cotton tensor; there is a big choice of
eigenvectors.

11.9.3 Garcı́a Stationary Solution with BTZ-Counterpart Limit

The second possible solution in the studied class is given by
F 2 dr2 2
g = − dT + + H (dΦ + W dT ) ,
H F

r2 r
F(r) = 4 2 + 2 (l w1 + l2 w12 − 4) (w0 + W0 ln(r)) ,
l l
Hn
H(r) = ,
Hd
Hn = F(r) − Ω(r)2 ,

r 2 2
Hd = l w1 − 4 − (w0 + W0 ln(r)) ,
l
Ω(r)
W(r) = , Ω(r) = w0 + W0 ln(r) + w1 r,
H(r)
  
1
W0 := l2 α2 l2 w12 − 2 − l w1 l2 w12 − 4 , (11.299)
2
with electromagnetic tensors
 
F μν = 2 α δ μ [t δ ν r] + δ μ [φ δ ν r] , (11.300)
⎡ ⎤
− 2[F −H W(1+W)]
2
(1−W 2 )] 2
− [F +H C 0 C
⎢ ⎥
⎢ ⎥
− [F −H
2
(1+W)2 ]
(T μ ν ) = ⎢ 0 0 ⎥ , (11.301)
⎣ C ⎦
2H2 (1+W) [F +H2 (1−W 2 )]
C 0 C

where C = 8παF2 H , and w0 and w1 are parameters related to mass and angu-
lar momentum, while α is an electromagnetic parameter; the electromagnetic
invariants are
α2
FF = 2 −F + H2 (1 + W)2 ,
FH
3 α4 1
TT = [−F + H2 (1 + W)2 ]2 . (11.302)
64 π 2 F 2 H2
The calligraphic capital letters have been used above to make their relationship
evident to those structural functions arising as real cuts of the complex extensions
of the studied class of metric, see (11.29). This solution of the Einstein–Maxwell
equations can be considered also as a real cut of the complex version of the
stationary electromagnetic solution with BTZ-limit given in the previous para-
graph; the structural functions F, H, and W can be constructed according to
(11.30) with F , H, and W from (11.271) accompanied by the replacement of the
214 Einstein–Maxwell Solutions

sign in front of α2 , α2 el → −α2 mg . If one searches for the anti-de Sitter limit
of this solution, one would arrive at an alternative real cut of the BTZ solution,
namely to the “BTZ solution counterpart,” or “BTZ counterpart” for short.
F dρ2 2
gc = −ρ2dT2 + + H (dΦ + W dT ) ,
H F
ρ2 J2 ρ2 J
F = 2 −M + 2
, H = 2 − M, W = . (11.303)
l 4ρ l 2H
Recall that in the above mentioned metric one can again introduce the radial
coordinate by changing
J
ρ2 → R2 + M l2 , H → R2 , W → ,
2R2
R2 J2
F(ρ) → F (R) = + M + . (11.304)
l2 4 R2
Since this solution possesses the BTZ counterpart as a limit for α = 0, it is
pertinent to search for new coordinates in which the BTZ solution counterpart
structure (11.303) will become apparent.

Transformation to BTZ-Counterpart Coordinates

The constants and the coordinate transformations to be used in this case are
given by
1 M
w0 = − l2 M 2 − J 2 l M − l2 M 2 − J 2 , w1 = 2 ,
J J
M l2 l
r = −ρ2 + + l2 M 2 − J 2 ,
2 2  
 
1 2 2 
2 −1/4 l J
φ= √ l M −J Ml T+ Φ ,
2 J l

1 l 2 2 1/4
t= √ l M − J2 T. (11.305)
2 J
Under these transformations the solution amounts to
F (ρ) 2 dρ2
g = −ρ2 dT + + H(ρ)(dΦ + W (ρ)d T )2 ,
H(ρ) F (ρ)

ρ2 J2 l α2  2 
F = − M + + J l − 2 R(−) ρ2 ln |r|,
l2 4ρ2 2 J ρ2

Hn
H(ρ) = ,
Hd
 
Hn := 4 J 3 l3 l2 M − ρ2 l2 M 2 − J 2 R(+)
2
α2 ln |r|
 2 2
−4 l2 M − ρ2 (l2 M 2 − J 2 )R(+) 4
− l6 J 6 α4 (ln |r|)
 11.10 Generating Solutions via SL(2, R)–Transformations 215

Hd := −2l5 J 3 l2 M 2 − J 2 R(+)
2
α2 ln |r|
 
+4 l2 l2 M − ρ2 (l2 M 2 − J 2 )R(+)4
,

l2 Ω(ρ)
W (ρ) = ,
J Hn
2
Ω(ρ) = −l5 J 6 R(+) α4 (ln |r|)
 
+l2 J 3 l R(+)2
+ 2J 2 l2 M 2 − J 2 + J 2 − R(+)
2
ρ2 R(+)
2
α2 ln |r|
   
+2 J 2 l2 M 2 − J 2 ρ2 − l2 M R(+)4
,
M l2 l
r := −ρ2 + + l2 M 2 − J 2 , R(±) := M l ± l2 M 2 − J 2 . (11.306)
2 2
The correspondence of this representation with the BTZ solution counterpart in
the limit of vanishing electromagnetic parameter α is evident.
Because of the complexity of the system of equations, we have found very hard
to determine other branches, if any, of exact solutions in the general case.

11.10 Generating Solutions via SL(2, R)–Transformations

This section deals with SL(2, R) transformations applied on static solutions to
construct stationary cyclic symmetric classes of solutions, namely the electric
and magnetic stationary families.
Let us consider the general metric

g = gtt dt2 + 2gtφ dtdφ + gφφ dφ2 + grr dr2 ,

and accomplish here a SL(2, R) transformations of the Killingian coordinates t

and φ
t = αt̃ + β φ̃, φ = γ t̃ + δ φ̃, Δ := αδ − βγ = 0. (11.307)

The transformed metric components are given by

gt̃t̃ = α2 gtt + 2αγ gtφ + γ 2 gφφ , gt̃φ̃ = αβ gtt + (αδ + βγ) gtφ + γδ gφφ ,
gφ̃φ̃ = β 2 gtt + 2βδ gtφ + δ 2 gφφ , grr = grr , (11.308)

while under the considered transformations the electromagnetic field ten-

sor (11.13) becomes
⎡ ⎤
0 b̃ − √
c̃grr
−g̃
1 ⎢ ⎢
⎥
⎥
(F ) = √
α̃β̃
⎢ −b̃ 0 ã ⎥,
−g̃ ⎣ ⎦
c̃g
√ rr
−g̃
−ã 0
g̃ = det(gμ̃ν̃ ), (11.309)
216 Einstein–Maxwell Solutions

where the new constant are given in terms of the original ones through
αa+ γ b
ã = , a = δã − γ b̃,
αδ − βγ
βa+δb
b̃ = , b = −βã + αb̃,
αδ − βγ
c
c̃ = . (11.310)
αδ − βγ
Notice that

gt̃t̃ gφ̃φ̃ − gt̃2φ̃ = (gtt gφφ − gtφ

2
)(αδ − βγ)2 = −F Δ2 ,

therefore, in concrete applications it is more useful to use normalized transfor-

mations with Δ = α δ − β γ = 1.
The electromagnetic tensor is form-invariant under the above-mentioned
SL(2, R) transformations if the field constants a, b, and c are identified by
means of Eq. (11.310). This property, on its turn, yields to the form-invariance
of the electromagnetic energy–momentum tensor Tμ ν = 1/(4π)(Fμσ F νσ −
1/4δμν Fτ σ F τ σ ), and consequently to the form-invariance of the Einstein–Maxwell
equations.
Therefore, starting with an electromagnetic solution in which a single elec-
tric (b = 0) or magnetic (a = 0) field is present, by accomplishing the
above-mentioned SL(2, R) transformations, one can generate solutions with both
electric and magnetic fields b̃ = 0, ã = 0 present. Conversely, if one originally has
had a solution endowed with both constant parameters a and b then, via trans-
formations, one could achieve a branch of solutions with one single parameter. At
this level, one may argue that one deals with one specific solution in its different
coordinate representations. But there exists a second point of view in (2 + 1)
gravity: to end with a new solution one has to change the variety, i.e., the topol-
ogy, requiring the ranges of change of the new variable to be, for instance, the
same as the ranges of the original variables. This procedure can be considered
as a generating solution technique and it has been used to construct station-
ary solutions starting from static solutions as we shall show in the forthcoming
sections.
For the metric (11.15), subjected to the above mentioned SL(2, R) transfor-
mations, one gets
F 2
gt̃t̃ = −α2 + H (α W + γ) ,
H
F
gt̃φ̃ = −α β + H (δ + β W ) (γ + α W ),
H
F 2
gφ̃φ̃ = −β 2 + H (β W + δ) ,
H
1
grr = , (11.311)
F
 11.11 Transformed Electrostatic b = 0 Solutions 217

hence, the expressions of the new structural functions are given in the form

F
H̃ = −β 2 + H(δ + β W )2 ,
H
F
W̃ H̃ = −α β + H(δ + β W )(γ + α W ), F̃ = F. (11.312)
H

The transformed electromagnetic field tensor F μ̃ν̃ , as it should be, exhibits its
form-invariant property
⎡ ⎤
0 b̃ − Fc̃
 μ̃ν̃  ⎢ ⎥
F =⎢
⎣ −b̃ 0 ã ⎥,
⎦ (11.313)
c̃
F −ã 0

where as before the new field constant parameters are related to the old ones
according to (11.310).
Although the metrics generated via SL(2, R) transformations should be given
as subsections in this section, because of the important place that various of
these solutions occupy in the field, I have decided to report them in separate
sections together with their properties.

11.11 Transformed Electrostatic b = 0 Solutions

Starting with the general electrostatic Maxwell solution (11.47) with metric

F 2 1
g=− dt + dr2 + Hdφ2 ,
H F
H(r)
F (r) = 4 2 2 K0 + H(r) − b2 l2 ln H(r) ,
C1 l
H(r) = C1 r + C0 , (11.314)

under normalized SL(2.R) transformations

α β γ δ
t = √ t̃ + √ φ̃, φ = √ t̃ + √ φ̃, Δ = αδ − βγ = 0 (11.315)
Δ Δ Δ Δ

(in the general (non-normalized) case the same expressions hold except for the
absence of Δ, set simply Δ = 1), the new metric, the rotated one, acquires the
form
⎡ 2 2
⎤
− αΔ HF
+ γΔ H 0 − αβ F
+ δγ
H
⎢ Δ H Δ
⎥
⎢ ⎥
(gμ̃ν̃ ) = ⎢ 0 1
0 ⎥, (11.316)
⎣ F ⎦
2 2
− αβ F
Δ H + ΔH
δγ
0 − βΔ HF
+ δΔ H
218 Einstein–Maxwell Solutions

the electromagnetic field tensor becomes

⎡ δb
⎤
0 √ 0
Δ
 μ̃ν̃  ⎢⎢ − √δ b γb
⎥
⎥
F =⎢ 0 √ ⎥, (11.317)
⎣ Δ Δ ⎦
0 − √γ Δ
b
0
while the electromagnetic energy–momentum tensor amounts to
⎡ 2 2 ⎤
1 (α δ+β γ)b
− 8π 0 1 γ αb

 ν̃  ⎢ ⎥
HΔ 4π H Δ
⎢ 2 ⎥
Tμ̃ = ⎢ 0 − 8π H
1 b
0 ⎥. (11.318)
⎣ ⎦
2 2
1 (α δ+β γ)b
− 4π H Δ
1 δβb
0 8π HΔ

Explicitly, the new metric is given by the nonzero components

α2 1 γ2 αβ 1 δγ
gt̃t̃ = − + H(r), gt̃φ̃ = − + H(r),
Δ H(r) grr Δ Δ H(r) grr Δ
β2 1 δ2
gφ̃φ̃ = − + H(r),
Δ H(r) grr Δ
1 C12 l2
grr = ,
4 H(r) [K0 + H(r) − b2 l2 ln H(r)]
H(r) = C1 r + C0 . (11.319)
For general SL(2, R) transformations, with non-vanishing entries, the electro-
magnetic field tensor F μν allows for the presence of both electric and magnetic
fields, corresponding to new b and a different from zero.
If one accomplishes the transformation of the dependent variable r to the
radial (polar) coordinate ρ, arc = ρ dφ, one chooses
H(r) = C1 r + C0 = ρ2 , C1 = 2. (11.320)

11.11.1 Stationary Electromagnetic Solution

In particular, for the SL(2, R) transformation
t = t̃ − ω φ̃, φ = φ̃, α = 1, β = −ω, γ = 0, δ = 1, Δ = 1, (11.321)
one obtains a new solution, the rotated one, with metric components
4
gt̃t̃ = − K0 + H(r) − b2 l2 ln H(r) ,
C12 l2
4 1
gt̃φ̃ = ω K0 + H(r) − b2 l2 ln H(r) , gφ̃φ̃ = H(r) − ω 2 ,
C12 l2 H(r) grr
1 C12 l2
grr = ,
4 H(r) [K0 + H(r) − b2 l2 ln H(r)]
H(r) = C1 r + C0 . (11.322)
 11.11 Transformed Electrostatic b = 0 Solutions 219

The electromagnetic field tensor is given by

⎡ ⎤ ⎡ ⎤
1 b2
0 b 0 − 8π 0 0
 μ̃ν̃  ⎢ ⎥  ν̃  ⎢ ⎢ H
⎥
⎥
F =⎢ ⎥
⎣ −b 0 0 ⎦ , Tμ̃ = ⎢ 0 − 8π
1 b2
0 ⎥.
⎣ H ⎦
0 0 0 1 ωb2 1 b2
4π H 0 8π H

Therefore, by means of a SL(2, R) transformation applied to the static electric

cyclic symmetric (2 + 1) Einstein–Maxwell solution one can generate a unique
electromagnetic stationary cyclic symmetric solution in the sense of the structure
of the field tensor F μν , which is equal in all respects to the electromagnetic
solution determined by the metric (11.265). It is worth mentioning that Clément
(1993) reported a solution, Eq. (24), belonging to this class.

11.11.2 Clément Spinning Solution

The so-called Clément’s spinning charged BTZ solution, derived in Clément
(1996), deserves special attention. It arises as a result of a SL(2, R) transforma-
tion of the electrostatic solution given in terms of the radial coordinate ρ → r.
Here the main Clément results are reproduced in a way quite close to the cited
work. Setting C1 = 2, which is equivalent to t → t C1 /2, accomplishing the
coordinate transformation H(r) = C1 r+C0 → r2 , and introducing the definitions
r0 = exp(K0 /(2b2 l2 )), and b2 = 4π G Q2 , the metric (11.314) becomes

dr2
g = −F (r)dt2 + + r2 dφ2 ,
F (r)
K0 r2 r2 r
F (r) = 2 + 2 − b2 ln r2 = 2 − 8πGQ2 ln ,
l l l r0
√ r
A = 2 Q π G ln dt. (11.323)
r0
To establish the range of values of r0 allowing the existence of a black hole, let
us consider F (r) in the form

r2 k r2
F (r) = (1 − ln ), k = 4πGQ2 l2 , (11.324)
l2 r2 r02
2
the factor (1 − rk2 ln rr2 ) vanishes in the set of points rh determined through the
0
LambertW function, LambertW(x) exp(LambertW(x)) = x, namely

rh2 = −k LambertW(−r02 /k), (11.325)

which is positive for r02 = k exp(−1), 0 <  ≤ 1, or explicitly

r02 ≤ 4πGQ2 l2 /e. (11.326)

220 Einstein–Maxwell Solutions

Subjecting the metric (11.323) and the vector potential A to the transformation
at uniform angular velocity
ω ω
t → t − ωφ, φ → φ − 2
t, α = 1, β = −ω, γ = − 2 , δ = 1, (11.327)
l l
one arrives at the metric
ω2 2 2 r2 dr2
g = −(F − 4
r )dt + 2ω(F − 2 )dtdφ + (r2 − ω 2 F )dφ2 + ,
l l F
r2 r2
F = 2 − 4πGQ2 ln 2 ,
l r0
√ r2
A = Q π G ln 2 (dt − ωdφ). (11.328)
r0

One could arrive at this result by using the metric components (11.319) with
transformation coefficients from (11.327) and setting C1 = 2, Δ = 1.
By choosing the axial symmetry as fundamental, the metric (11.328) can be
brought to the form

F(r) 2 dr2 2
g = −r2 (1 − ω 2 /l2 )2 dt + + H(r) (dφ + W(r)dt) ,
H(r) F(r)2
r2 r2 F 2 − r2 /l2 ω r2
F = F = 2 − 4πGQ2 ln 2 , W = ω = −4πGQ2 ln 2 ,
l r0 H H r0
2 2
ω r
H = r2 − ω 2 F = r2 (1 − 2 ) + ω 2 4πGQ2 ln 2 . (11.329)
l r0

The Clément spinning charged BTZ solution is endowed with three parameters
Q, r0 , and ω. It allows for a black hole interpretation.
Alternatively, introducing the scaling transformation r = l/¯l×r̄, the definitions
¯l2 = l2 − ω 2 , |ω| < l, and r̄0 = ¯l/l × r0 , the proper Clément (1996) solution,
dropping the bar from r → ρ, is given in its canonical representation as
1
ds2 = −N (ρ)2 dt2 + dρ2 + K(ρ)2 [dφ + W (ρ)dt]2 ,
L(ρ)2
ρ2
K(ρ)2 = H(ρ) := ρ2 + 4π G ω 2 Q2 ln ( 2 ),
ρ0
ρ 2
4π G (l − ω ) Q2
2 2
ρ2
L(ρ)2 = F (ρ) := 2 − ln ( ),
l l2 ρ2
0
F (ρ)
L(ρ) = F (ρ), K(ρ) = H(ρ), N (ρ) = ρ ,
H(ρ)
4π G Q2 ρ2
W (ρ) = −ω ln ( 2 ),
H(ρ) ρ0
√ ρ
A = 2 Q π G ln (dt − ωdφ); (11.330)
ρ0
 11.11 Transformed Electrostatic b = 0 Solutions 221

see also Garcı́a (2009), Eq. (11.24). The corresponding electromagnetic fields are
given by
4Q √
Fμν = − π G(δ[μ t δν] r − ωδ[μ φ δν] ρ ),
ρ
G Q2 l2 + ω 2 t ν G Q2 ω t ν G Q2 l2 + ω 2 φ ν
Tμ ν = − 2 δμ δt − 2 2 δμ δφ + δμ δφ
2l ρ2 l ρ 2 l2 ρ2
G Q2 ω φ ν G Q2 l2 − ω 2 r ν
+ δμ δt − δμ δr . (11.331)
ρ2 2 l2 ρ2
This solution is a black hole if the condition of the form (11.326),
ρ̄20 ≤ 4πGQ2 ¯l2 /e, (11.332)
is fulfilled. It possesses two horizons, at which F (ρ) vanishes, which are roots of
the relation
ρ
ρ2 − ¯l2 8πGQ2 ln = 0, (11.333)
ρ̄0
which are given by the LambertW function, see (11.325). The largest root deter-
mines the event horizon at ρ = ρ+ = ρh , while the inner one is a Cauchy horizon
at ρ = ρ− , with ρ+ > ρ− > ρ̄0 . Since the metric function H changes sign for a
certain value ρ = ρc < ρ̄0 , similarly as the rotating BTZ solution, thus there are
closed timelike curves in the region inside the radius ρc . It is apparent that the
metric and the electromagnetic field are singular at ρ = 0.
The length rescaling was chosen in such a manner that H → ρ2 at spatial
infinity and g(11.330) → gBTZ . According to Clément, one may formally define
mass and angular momentum parameters M (ρ1 ) and J(ρ1 ) by identifying, at a
given scale ρ = ρ1 , the values of the structural functions with the corresponding
BTZ values. Nevertheless, the mass and angular momentum defined in this way
occur to be ρ1 -dependent and diverge logarithmical as ρ1 → ∞.

Mass, Energy and Momentum

The evaluation of the surface energy and momentum densities yields
1 L(ρ)  2 
(ρ, 0 ) = − ρ + 4π G ω 2 Q2 − 0 ,
πρ K(ρ)2
ρ2 1
j(ρ) = −4 G ω Q2 ρ[1 − ln ( 2 )] , (11.334)
ρ0 K(ρ)
while the integral quantities can be evaluated from the generic expressions
J(ρ) = 2 π K(ρ) j(ρ), E(ρ) = 2 π K(ρ) (ρ), M (ρ) = N (ρ) E(ρ) − W (ρ)J.
In this manner one arrives at
ρ2
J(ρ) = −8π G ω Q2 (1 − ln ( )),
ρ20
L  2 
E(ρ, 0 ) = −2 ρ + 4π G ω 2 Q2 − 2 π K(ρ) 0 ,
ρK
222 Einstein–Maxwell Solutions

ρ2 π Gω 2 Q2 ρ2
M (ρ, 0 ) = −2 − 8 + 8π G Q2
ln ( )
l2 l2 ρ20
−2 π K(ρ) N (ρ) 0 . (11.335)

The evaluation of the corresponding functions for the base energy 0 = 0 yield
at spatial infinity ρ → ∞

G ω Q2 G ωQ2 ρ
j(ρ → ∞) ≈ −4 +8 ln ( ),
ρ ρ ρ0
ρ
J(ρ → ∞) ≈ −8π G ω Q2 (1 − 2 ln ( )),
ρ0
1 G ω 2 Q2 G Q2 ρ
(ρ → ∞, 0 = 0) ≈ − −4 2
+ 4 2 (l2 + ω 2 ) ln ( ),
πl lρ lρ ρ0
2ρ π G ω 2 Q2 π G Q2 l ρ
E(ρ → ∞, 0 = 0) ≈ − − 8 +8 ln ( ),
l lρ ρ ρ0
ρ2 π Gω 2 Q2 ρ
M (ρ → ∞, 0 = 0) ≈ −2 2 − 8 2
+ 16π G Q2 ln ( ). (11.336)
l l ρ0

Using in the expressions(11.334) and (11.335) as the reference energy density

2
the quantity 0 = − πρ1
−M0 + ρl2 , which at the spatial infinity behaves as
0|∞ (M0 ) ≈ − π1l + lM0
2π ρ2 , the series expansions of the corresponding quantities
at ρ → ∞ result in

l M0 G ω 2 Q2
(ρ → ∞, 0|∞ (M0 )) ≈ − 2
−4
2π ρ lρ2
G Q2 ρ
+ 4 2 (l2 + ω 2 ) ln ( ),
lρ ρ0
l M0 π G ω 2 Q2
E(ρ → ∞, 0|∞ (M0 )) ≈ − −8
ρ lρ
π G Q2 2 ρ
+8 (l + ω 2 ) ln ( ),
lρ ρ0
π Gω 2 Q2
M (ρ → ∞, 0|∞ (M0 )) ≈ −M0 − 8
l2
l + ω2
2
ρ
+ 8π G Q2 ln ( ). (11.337)
l2 ρ0

Comparing with the energy characteristics of the BTZ solution, one concludes
that a mass parameter M similar to the BTZ mass is absent; instead, a term
in the mass function due to the product of the rotation ω and the charge Q is
present. Notice that E(ρ) and M (ρ) logarithmically diverge at spatial infinity.
The momentum parameter is due to the product of ω Q, and hence is not a free
parameter.
 11.11 Transformed Electrostatic b = 0 Solutions 223

Cotton Tensor
The Cotton characterization of this solution is given by
⎡ 2 2
⎤
1 ω (F (ρ)l +ρ ) d3 (F (ρ)ω2 +ρ2 )l2 d3
F (ρ) 0 − 18 ρ (l2 −ω2 ) dρ 3 F (ρ)
⎢ 8 ρ (l2 −ω2 ) dρ3 ⎥
 α ⎢ ⎥
Cβ = ⎢ ⎢ 0 0 0 ⎥,
⎥
⎣ 4 2 2
⎦
1 (F (ρ)l +ω ρ ) d3 ω (F (ρ)l2 +ρ2 ) d3
8 l2 ρ (l2 −ω 2 ) dρ3 F (ρ) 0 − 18 ρ (l2 −ω2 ) dρ 3 F (ρ)

(11.338)
where
ρ2    
F (ρ) = + 4 Q2 π G ln ρ20 − 4 Q2 π G ln ρ2 ,
l2
d3 Q2 π G
3
F (ρ) = −16 (11.339)
dρ ρ3
The eigenvalue problem yields

λ1 = 0;
2
V1 = [0, V 2 , 0], Vμ V μ = V 2 , V1 = S1,
−F (ρ)Q2 π G
λ2 = 2 ;
ρ3
V1 ωρ+ −F (ρ)l2
1
V2 = [V , 0, ] = Z,
l2 ρ + −F (ρ)ω
−F (ρ)Q2 π G
λ3 = −2 ;
ρ3
V1 ωρ− −F (ρ)l2
1
V3 = [V , 0, ] = Z̄. (11.340)
l2 ρ − −F (ρ)ω

It is apparent that the Cotton tensor is of Type I: {S, Z, Z̄}.

Kamata–Koikawa Limit
It should be pointed out that Clément (1993) also reported the so-called self-dual
solution, published later in Kamata and Koikawa (1995, 1997). By accomplishing
the limiting transition ω → ±l ⇒ ¯l → 0, of the metric structural func-
tions (11.330), while the other parameters Q and r̄0 remain fixed, one arrives
then at the metric (11.330) with structural functions and vector field

r2 l r r
F = 2
, W = ∓ 8πGQ2 ln , H = r2 + l2 8πGQ2 ln ,
l H r̄0 r̄0
r
A = Q ln (dt ∓ l dφ). (11.341)
r̄0
224 Einstein–Maxwell Solutions

Notice that this solution does not possess an horizon; in the limiting transition
¯l → 0, the horizon does not survive since it disappears below ¯l = (4πGQ2 )1/2 ,
as quoted by Clément.
The proper KK representation of this one-parameter solution is achieved by
accomplishing the radial transformation and scaling of parameters
√
r2 = rKK
2
− r0KK
2
, r0KK = (4πGQ2 l2 )1/2 , r0 = r0KK / e, (11.342)
and the subscripts are self-explanatory. It is worth also noticing that a derivation
and analysis of the KK solution has been accomplished in Cataldo and Salgado
(1996), too.

11.11.3 Martı́nez–Teitelboim–Zanelli Solution

Martı́nez, Teitelboim and Zanelli – see Martı́nez et al. (2000) – reported a gen-
eralization of the BTZ black hole spacetime equipped with an electric charge Q,
the mass M and the angular momentum J. The main features of this charged
black hole, among others, following the quoted paper, are: the total M, J and Q,
which are boundary terms at infinity; the extreme black hole can be thought of
as a particle moving with the speed of light; and the inner horizon of the rotat-
ing uncharged black hole is unstable under the perturbation of a small electric
charge. According to the quoted reference, this electrically charged black hole is
pathological in the sense it exists for arbitrary values of the mass and that there
is no upper bound on the electric charge.
The starting point is the electrostatic metric (11.314) given in terms of the
polar coordinate r,
1
H(r) = C1 r + C0 → r2 , C1 = 2, K0 /l2 → −M̃ , b2 → Q̃2 , (11.343)
4
2
therefore F (r) = rl2 − M̃ − 14 Q̃2 ln r2 . Using the metric components (11.319) with
transformation coefficients from the “rotation boost” transformation
1 1 ω
t→ (t − ωφ) , φ → φ− 2 t , (11.344)
1 − ω /l
2 2 1 − ω /l
2 2 l
one arrives at the metric

r2 1 Q̃2 ω dtdφ Q̃2
g=− 2 − (M̃ + ln r 2
) dt 2
− 2 (M̃ + ln r2 )
l 1 − ω 2 /l2 4 1 − ω 2 /l2 4

ω2 Q̃2 2 dr2
+ r2 + (M̃ + ln r 2
) dφ + .
1 − ω 2 /l2 4 r2 /l2 − M̃ − 14 Q̃2 ln r2
(11.345)
The electromagnetic field tensor is given by
Q̃
Fμν = δ[μ t δν] r + ω l2 δ[μ t δν] φ . (11.346)
r 1 − ω 2 /l2
 11.11 Transformed Electrostatic b = 0 Solutions 225

The angular momentum, charge, and mass can be evaluated via quasi-local def-
initions, see below; the presence of logarithmic terms in the structural metric
functions yields to divergences at infinity of the energy–momentum quantities.
As pointed out by the authors, the divergence in the mass can be handled
by enclosing the system in a large circle of radius r0 in which will be bound
M (r0 ) – the energy within r0 – and the electrostatic energy outside r0 given
by −Q2 ln r0 /2, thus the total mass (independent of r0 and finite) is given by
M̃ = M (r0 ) − Q2 ln r0 /2.
Dropping tilde and replacing r → ρ, this metric can be brought to the canonical
form
1
ds2 = −N (ρ)2 dt2 + dρ2 + K(ρ)2 [dφ + W (ρ)dt]2 ,
L(ρ)2
l2 ω 2 Q2
K(ρ)2 = H(ρ) = ρ2 + 2 (M + ln ρ2 ),
l − ω2 4
ρ2 Q2
L(ρ)2 = F (ρ) = 2 − M − ln ρ2 ,
l 4
L(ρ) L(ρ)
N (ρ) = ρ =ρ ,
H(ρ) K(ρ)
ω l2 1 Q2
W (ρ) = − (M + ln ρ2 ). (11.347)
l2 − ω H(ρ)
2 4

Mass, Energy and Momentum

The evaluation of the surface energy and momentum densities yields
 
1 L l2 ω 2 Q2
(ρ, 0 ) = − 2ρ + 2 − 0 ,
2π K 2 l − ω 2 2ρ
 
ρ l2 ω L 1 2 Q2
j(ρ) = M + Q ln ρ − , (11.348)
π l2 − ω 2 N K 2 2 4
while the integral quantities can be evaluated from the generic expressions for
{J(ρ), E(ρ), M (ρ)}. The evaluation of the corresponding functions with 0 = 0
behave at infinity according to
1 [2 M (l2 + ω 2 ) − ω 2 Q2 ] l Q2 l2 + ω 2
(ρ → ∞, 0 = 0) ≈ − + + ln ρ,
πl 4 π(l2 − ω 2 )ρ2 4 πρ2 l2 − ω 2
ω l2 4 M − Q2 ω Q2 l2
j(ρ → ∞) ≈ + ln ρ,
4πρ l2 − ω 2 2πρ(l2 − ω 2 )
ω l2 4 M − Q2 ω Q2 l2
J(ρ → ∞) ≈ + ln ρ
2 l2 − ω 2 l2 − ω 2
ω Q2 l2
= J(MTZ Eq.82) + 2 ln ρ,
l − ω2
2ρ l 2M l2 − ω 2 Q2 l3 Q2
E(ρ → ∞, 0 = 0) ≈ − + + ln ρ,
l 2ρ l2 − ω 2 2ρ(l2 − ω 2 )
2ρ2 4 M l2 − ω 2 Q2 l2 Q2
M (ρ → ∞, 0 = 0) ≈ − 2 + + ln ρ. (11.349)
l 2(l2 − ω 2 ) l2 − ω 2
226 Einstein–Maxwell Solutions

ρ2
Using in the expressions (11.348) and 0 = − πρ
1
−M0 + l2 , as reference
energy density, which at the spatial infinity behaves as 0|∞ ≈ − π1l + 2π
lM0
ρ2 , the
series expansions of the corresponding quantities at ρ = infinity result in
l l ω2
(ρ → ∞, 0|∞ ) ≈ (M − M 0 ) + (4 M − Q2 )
2π ρ2 4 π(l2 − ω 2 )ρ2
l Q2 l2 + ω 2
+ ln ρ,
4 πρ2 l2 − ω 2
l lω 2
E(ρ → ∞, 0|∞ ) ≈ (M − M0 ) + (4 M − Q2 )
ρ 2ρ (l2 − ω 2 )
l Q2 l2 + ω 2
+ ln ρ,
2 ρ l2 − ω 2
ω2 Q2 l2 + ω 2
M (ρ → ∞, 0|∞ ) ≈ M − M0 + (4 M − Q2
) + ln ρ
2(l2 − ω 2 ) 2 l2 − ω 2
Q2 l2 + ω 2
= −M0 + M(MTZEq.81) (ω → ω/l) + ln ρ.
2 l2 − ω 2
(11.350)

Notice that the charges Q used above differs from Q(MTZ,Eq.83) ,

l
Q(MTZEq.83) = √ Q (11.351)
l2 − ω2
Therefore, comparing with the energy characteristics of the BTZ solution, one
concludes that the parameter M can be considered as the BTZ mass, and the
energy and mass functions logarithmically diverges at spatial infinity.

Field, Energy–Momentum, and Cotton Tensors

The Maxwell field tensor for this MTZ solution is given by
⎡ 1 √ Ql ⎤
0 2 ρ l2 −ω 2 L2 0
⎢ ⎥
⎢ 2 2 ⎥
(F α β ) = ⎢ 12 ρ √QlL 0 − 12 ρω√lL Q
⎥,
⎣ l 2 −ω 2 l 2 −ω 2
⎦
0 1 √ωQ 0
2 lρ l2 −ω 2 L2

ρ2 1  
L2 = 2
− m − Q2 ln ρ2 , (11.352)
l 4
while its eigenvalues and the corresponding eigenvectors amount to

λ1 = 0; V1 = [V 3 ω, 0, V 3 ],
l2 − ω 2 2 3 2
Vμ V μ = ρ V , V1 = S1,
l2
√
1 Q V 1 l2 − ω 2 2 1 ω
λ2 = − ; V2 = [V 1 , L , V 2 ],
2 ρ l l
Vμ V μ = 0, V2 = N2
 11.11 Transformed Electrostatic b = 0 Solutions 227
√
1 Q V1 l2 − ω 2 2 1 ω
λ3 = ; V3 = [V 1 , − L , V 2 ],
2 ρ l l
Vμ V μ = 0, V3 = N3,
Type: {S, N, N }. (11.353)
The energy–momentum tensor
⎡ 2 2 2 ⎤
1 Q (l +ω ) l2 ω Q2
− 32π ρ2 (l2 −ω 2 ) 0 1
16π ρ2 (l2 −ω 2 )
⎢ ⎥
⎢ ⎥
(T α β ) = ⎢ ⎥,
2

⎢ 0 − 32π
1 Q
ρ2 0 ⎥ (11.354)
⎣ ⎦
2 2 2
ω Q2 1 Q (l +ω )
− 16π
1
ρ2 (l2 −ω 2 ) 0 32π ρ2 (l2 −ω 2 )

possesses the following eigenvalues and eigenvectors

1 Q2
λ1 = ; V1 = [V 3 ω, 0, V 3 ],
32π ρ2
2  
V 3 ρ2 l2 − ω 2
Vμ V μ = , V1 = S1,
l2
1 Q2 l2 V 3 2 3
λ2,3 = − 2
; V2 = [ , V , V ],
32π ρ ω
  2 2
l2 l2 − ω 2 V 3 2 V 2
Vμ V = −
μ
L + ,
ω2 L2
V2 = {T2, N2, S2}, V3 = {T3, N3, S3},
Type: {S, 2N }, and so on. (11.355)
For the Cotton tensor
⎡ ⎤
(ρ2 +l2 L2 )ω Q2 (ρ2 +ω2 L2 )l2 Q2
− 18 ρ4 (l2 −ω2 ) 0 1
⎢ 8 ρ4 (l2 −ω 2 ) ⎥
⎢ ⎥
(C β ) = ⎢
α
⎢ 0 0 0 ⎥,
⎥ (11.356)
⎣ ⎦
(ω ρ +l4 L2 )Q2
2 2
(ρ 2 2 2
+l L )ω Q 2
− 18 ρ4 l2 (l2 −ω2 ) 0 1
8 ρ4 (l2 −ω 2 )

the eigenvalues and the corresponding eigenvectors are

2
V2
λ1 = 0; V1 = [0, V 2 , 0], Vμ V μ = V1 = S1,
L2
1 Q2 −L2 V 1 ω ρ + −L2 l2
1
λ2 = ; V2 = [V , 0, ] = Z,
8 ρ3 l2 ρ + −L2 ω

1 Q2 −L2 V 1 ω ρ − −L2 l2
λ3 =− 1
; V3 = [V , 0, ] = Z̄,
8 ρ3 l2 ρ − −L2 ω
Type I: {S, Z, Z̄}. (11.357)
228 Einstein–Maxwell Solutions

Transformation of the Clément Stationary Solution to the MTZ Metric

Since these two metric representations arise from SL(2, R) transformations
applied on the same static charged metric, there ought to be a link between them;
in fact, time and angular coordinates are related by the scaling transformations
1
{t, φ}C = {t, φ}M T Z , (11.358)
1 − ω 2 /l2
for the same radial coordinate r and angular parameter ω, while the constant
parameters are related according to
{16π G Q2 , 4π G Q2 ln r0 2 }C = {Q̃2 , −M̃ }M T Z . (11.359)
In this way, the Clément (1996) solution (11.328) is transformed into the MTZ,
Martı́nez et al. (2000) solution (11.345).

11.12 Transformed Magnetostatic a = 0 Solutions

The stationary rotating generalization of the magnetostatic metric (11.76)
F 2 1
g=− dt + dr2 + Hdφ2 , F (r) = H(r) h(r),
H F
4
H(r) = 2 2 K0 + h(r) + a2 l2 ln h(r) , h(r) := C1 r + C0 ,
C1 l
is determined by means of the SL(2, R) transformations
α β γ δ
t = √ t̃ + √ φ̃, φ = √ t̃ + √ φ̃, Δ = αδ − γβ,
Δ Δ Δ Δ
which lead to the new metric in the form
⎡ 2 2
⎤
− αΔ HF
+ γΔ H 0 − αβ F
Δ H +
γδ
ΔH
⎢ ⎥
⎢ ⎥
(gμ̃ν̃ ) = ⎢ 0 1
0 ⎥,
⎣ F ⎦
2
δ2
− αβ F
Δ H + ΔH
γδ
0 − βΔ F
H + ΔH

which is accompanied with the electromagnetic field tensor

⎡ ⎤
0 a √βΔ 0
⎢ ⎥
⎢ ⎥
(F τ σ ) = ⎢ −a √βΔ 0 a √αΔ ⎥ . (11.360)
⎣ ⎦
0 −a Δ√α
0
The corresponding Maxwell energy–momentum tensor becomes
⎡ (α δ+β γ)a2 ⎤
γ α a2 H
− 8πΔ H
0
 ν̃  ⎢ ⎥
F 4πΔ F
⎢ a2 H ⎥
Tμ̃ = ⎢ 0 0 ⎥. (11.361)
⎣ 8π F ⎦
δ β a2 H (α δ+β γ)a2 H
− 4πΔ F 0 8π Δ F
 11.12 Transformed Magnetostatic a = 0 Solutions 229

Explicitly, the nonzero metric components are

α2 γ2 1 αβ δγ 1
gt̃t̃ = − h(r) + ,g =− h(r) + ,
Δ Δ h(r) grr t̃φ̃ Δ Δ h(r) grr
β2 δ2 1
gφ̃φ̃ = − h(r) + ,
Δ Δ h(r) grr
1 C12 l2
grr = , h(r) = C1 r + C0 . (11.362)
4 h(r) [K0 + h(r) + a2 l2 ln h(r)]

11.12.1 Stationary Magneto-Electric Solution

In particular, for the SL(2, R) transformation
ω ω
t = t̃, φ = − 2
t̃ + φ̃, α = 1, β = 0, γ = − 2 , δ = 1, (11.363)
l l
one obtains a new solution with metric components
ω2 1 ω 1
gt̃t̃ = −h(r) + 4
, gt̃φ̃ = − 2 ,
l h(r) grr l h(r) grr
1 1 C12 l2
gφ̃φ̃ = , grr = ,
h(r) grr 4 h(r) [K0 + h(r) + a2 l2 ln h(r)]
h(r) = C1 r + C0 , (11.364)

which is equal to the constant W = ω stationary magneto-electric solution

(11.255). Therefore, by means of a SL(2, R) transformation applied to the mag-
netostatic cyclic symmetric (2 + 1) Einstein–Maxwell solution, one can generate
a unique electromagnetic stationary cyclic symmetric solution in the sense of the
structure of the field tensor F μν . Clément (1993) reported a field belonging to
this class of solutions; see Eq. (Cl.23).

11.12.2 Dias–Lemos Magnetic BTZ–Solution Counterpart

Dias and Lemos (2002) published a rotating magnetic solution in (2+1) gravity:
the magnetic counterpart of the spinning charged BTZ solution, i.e., a point
source generating a magnetic field. Also, it was established that both the static
and rotating magnetic solutions possess negative mass and that there is an upper
bound for the intensity of the magnetic field source and for the value of the
angular momentum.
A simple representation of this solution can be achieved from our transformed
magnetic metric (11.362) by setting
ω ω
t = t̃, φ = − 2
t̃ + φ̃, α = 1, β = −ω, γ = − 2 , δ = 1, Δ = 1,
l l
C1 = 2, h(r) = C1 r + C0 → r2 ,
230 Einstein–Maxwell Solutions

obtaining
ω2 F dr2
g = −(r2 − 4
F )dt2 + (F − ω 2 r2 )dφ2 − 2ω( 2 − r2 )dtdφ + ,
l l F (r)
K0 r2
F = 2
+ 2 + a2 ln r2 . (11.365)
l l
The proper Dias–Lemos representation uses a more involved definition of the
transformed r-coordinate and parameterizations of the SL(2, R) transforma-
tions, namely

h(r) = C1 r + C0 → (ρ2 + r+ 2
− ml2 )/l2 , C1 = 2/l2 , χ2 = a2 l4 ,
ω
t = 1 + ω 2 t̃ − l ω φ̃ φ = − t̃ + 1 + ω 2 φ̃;
l
dropping tildes, one has
 
ω2  
g = − h − 2 m l2 + χ2 ln |h| dt2
l
ω
−2 ω 2 + 1 m l2 + χ2 ln |h| dt dφ
l  
+[h l2 + (ω 2 + 1) m l2 + χ2 ln |h| ]dφ2
l2 ρ2 dρ2
+ 2 − ml2 )[ρ2 + r 2 + χ2 ln |h|] ,
(ρ2 + r+ +
h = (ρ2 + r+
2
− ml2 )/l2 , (11.366)

with vector potential

1 ω
A= χ ln |h(ρ)|[− dt + 1 + ω 2 dΦ]. (11.367)
2 l
Notice that in the above mentioned representation, the equation
2
r+ + χ2 ln |(ρ2 + r+
2
)/l2 − m| = 0, (11.368)

used in the Eq. (3.2) of Dias and Lemos (2002) was not used here. According to
these authors, this rotating magnetic spacetime is null and timelike geodesically
complete, and as such horizonless.
It it noteworthy to point out that by subjecting the static Hirschman–Welch
metric (11.101) to the SL(2, R) transformation
ω
t→ 1 + ω 2 t − ω lφ, ρ → ρ, φ → − t+ 1 + ω 2 φ,
l
one arrives at the Dias and Lemos (2002) solution too, see also Garcı́a (2009),
Eq. (11.41), given in the canonical representation by metric
1
ds2 = −N (ρ)2 dt2 + dρ2 + K(ρ)2 [dφ + W (ρ)dt]2 ,
L(ρ)2
H(ρ) := (ρ2 + r+
2
− ml2 )/l2 ,
 11.12 Transformed Magnetostatic a = 0 Solutions 231

H(ρ)  2 2 + χ2 ln H(ρ),
L(ρ) = ρ + r+
ρ

K(ρ) = 2 + ω 2 l2 m + (1 + ω 2 )χ2 ln H(ρ),
ρ2 + r+
√
L(ρ) ω 1 + ω 2 [ml2 + χ2 ln H(ρ)]
N (ρ) = ρ , W (ρ) = − . (11.369)
K(ρ) l K(ρ)2
Notice that this metric, in the case of vanishing charge χ = 0, yields to an alter-
native coordinate representation of the rotating BTZ solution, with parameter
ω, namely
 2 
ρ + r+ 2   ρ2 l2
ds2 = − 2
− 1 + ω 2
m l 2
dt2 + 2  2 − m l2
 dρ2
l (ρ + r+ ) ρ2 + r+
2

−2 ω m l 1 + ω 2 dφ dt + (ρ2 + r+ 2 + l2 ω 2 )dφ2 , (11.370)

which differs from the standard BTZ solution representations.
Accomplishing in the above metric the transformations
t → l2 t, ρ → ρ2 /l2 − ml2 ω 2 − r+ 2 + M
and identifying the parameters according to
2 2 2
M 2 (1 + ω )ω l
m= , M = J 2 /4 →
1 + 2 ω2 (2ω 2 + 1)2
√ √
1 M l ± M 2 l2 − J 2 M 2 l2 − J 2
2
ω = √ ,m= ,
2 M 2 l2 − J 2 l3
one arrives at the BTZ solution counterpart representation
 2 −1  2 
ρ J2 ρ
ds2 = −ρ2 dt2 + 2 + M + 2 dρ2 − J dφ dt + 2 + M dφ2 .
l 4ρ l

On the other hand, by replacing ρ → ρ2 − m l2 ω 2 − r+ 2 one arrives at the

middle of the road metric
 2 
ρ  
ds2 = − 2 − 1 + 2 ω 2 m dt2 − 2 ω m l 1 + ω 2 dφ dt + ρ2 dφ2
l
 2 −1
ρ l2 m2 ω 2 (1 + ω 2 )
+ 2 − (1 + 2 ω )m +
2
dρ2 , (11.371)
l ρ2
which, identifying
M
m= , l2 m2 ω 2 (1 + ω 2 ) = J 2 /4 →
1 + 2 ω2
√ √
1 M l ± M 2 l2 − J 2 M 2 l2 − J 2
2
ω = √ ,m= ,2 1 + ω 2 ω ml → J,
2 M 2 l2 − J 2 l
gives rise to the standard description of the stationary BTZ black hole metric.
Therefore, as the vacuum limit of the DL metric (11.369) one may consider the
232 Einstein–Maxwell Solutions

rotating BTZ solution counterpart, and consequently one may think of it as the
reference vacuum solution in the evaluation of the quasi-local energy, momentum
and mass.

Mass, Energy and Momentum

The surface energy and momentum densities are given by
L
(ρ, 0 ) = −ρ (ρ2 + r+
2
− ml2 + (1 + ω 2 )χ2 ) − 0 ,
π l2 K 2 H
ρ L
j(ρ, 0 ) = ω 1 + ω 2 [m l2 − χ2 + χ2 ln H], (11.372)
π l N K2
while the integral quantities amount to
ρ L
J(ρ, 0 ) = 2ω 1 + ω2 [m l2 − χ2 + χ2 ln H]
l NK
2
= ω 1 + ω 2 [m l2 − χ2 + χ2 ln H],
l
ρ L
E(ρ, 0 ) = −2 P (ρ) − 2π K0 ,
l2 K H
ρ NL
M (ρ, 0 ) = −2 2 P (ρ) − W J − 2π N K0 = 2 m
l HK
2 ω2
− 2 (ρ2 + χ2 + r+2
) + 2 2 [ml2 − χ2 + χ2 ln H] − 2π N K0 ,
l l
P (ρ) := ρ2 + r+2
− ml2 + (1 + ω 2 )χ2 . (11.373)

The evaluation of the main parts of above functions, i.e., the corresponding
functions independent of 0 behave at infinity according to
ω ρ
j(ρ → ∞) ≈ 1 + ω 2 [m l2 − χ2 + 2χ2 ln ( )],
lπρ l
ω ρ
J(ρ → ∞) ≈ 2 1 + ω 2 [m l2 − χ2 + 2χ2 ln ( )],
l l
1 ml2 − 2χ2 χ2 ρ
(ρ → ∞, 0 = 0) ≈ − + + ln ( )
πl 2π l ρ2 π l ρ2 l
m l 2 − χ2 χ2 ρ
+ ω2 [ 2
+2 ln ( )],
πl ρ π l ρ2 l
2ρ ml2 − r+
2
− 2χ2
E(ρ → ∞, 0 = 0) ≈ − +
l lρ
ω2 ρ
+[m l2 − 2χ2 + 2χ2 ln ( )],
lρ l
1 2
M (ρ → ∞, 0 = 0) ≈ 2m − 2 2 (ρ + r+ 2
+ χ2 )
l
ω2 ρ
+ 2 2 [m l2 − χ2 + 2χ2 ln ( )]. (11.374)
l l
 11.12 Transformed Magnetostatic a = 0 Solutions 233

Using in  the expressions (11.373) as reference energy density the quantity

2
0 = − πρ ρl2 − M0 , which at the spatial infinity behaves as 0|∞ ≈ − π1l + 2π
1 lM0
ρ2 ,
the series expansions of the corresponding quantities at ρ = infinity result in
l 1 ρ
(ρ → ∞, 0|∞ ) ≈ (m − M0 ) − (χ2 − χ2 ln ( ))
2π ρ2 2π l ρ2 l
2
ω ρ
+ [ml2 − χ2 + 2χ2 ln ( )],
π l ρ2 l
2 2
l χ χ ρ
E(ρ → ∞, 0|∞ ) ≈ (m − M0 ) − 2 + 2 ln ( )
ρ lρ lρ l
2
ω ρ
+ 2 [m l2 − χ2 + 2χ2 ln ( )],
lρ l
2 2
χ χ ρ
M (ρ → ∞, 0|∞ ) ≈ m − M0 − 2 2 + 2 2 ln ( )
l l l
ω2 ρ
+2 2 [m l2 − χ2 X + 2χ2 ln ( )]. (11.375)
l l
Therefore, comparing with the energy characteristics of the BTZ solution, one
concludes that the mass logarithmically diverges at spatial infinity. For vanish-
ing rotation parameter ω one recovers the static solution in the representation
of Hirschman–Welch and certainly the corresponding energy quantities. The
parameter m can be considered as the BTZ mass.

Field, Energy and Cotton Tensors

To determine the algebraic types of the electromagnetic field, energy–momentum,
and Cotton tensors it is more convenient to work with the DL metric in the form
⎡   √   ⎤
−h − ω 2 h − L2 0 l 1 + ω 2 ω h − L2
⎢ ⎥
g=⎢ ⎣ 0 ρ2
h L2 l 2
0 ⎥ , (11.376)
⎦
√    2 
l 1+ω ω h−L
2 2
0 l −ω h + L + L ω
2 2 2 2

where
χ2 ln (h) + ρ2 ρ2 + Mg
L2 = , h = 2
, Mg = r+ − ml2 . (11.377)
l2 l2
In this representation, the electromagnetic field tensor becomes
⎡ ⎤
0 − lρ3 Lχ2ωh 0
⎢ √ ⎥
⎢ 2
χ 1+ω 2 L2 ⎥
(F α β ) = ⎢ − ω χlρL 0 ⎥, (11.378)
⎣ √
ρ
⎦
2
0 − ρ χL2 1+ω
l4 h
0

with the following eigenvalues and their corresponding eigenvectors

234 Einstein–Maxwell Solutions

ω h 2
λ1 = 0; V1 = [V 1 , 0, √ V 1 ], Vμ V μ = − 2
V 1 , V1 = T1,
l 1+ω 2 1 + ω
√
χ ωρ 1 + ω2 ρ
λ2 = −i √ ; V2 = [i 2 √ V , V , i 2 √ V 2 ], V2 = Z,
2 2
hl 2 lL h L l2 h
√
χ ωρ 1 + ω2 ρ
λ3 = i √ ; V3 = [−i 2 √ V , V , −i 2 √ V 2 ], V3 = Z̄,
2 2
hl2 lL h L l2 h
Type : {T, Z, Z̄}. (11.379)

As far as to the electromagnetic energy momentum tensor is concerned, its matrix

is given by
⎡ χ2 (1+2 ω 2 ) 2
√ ⎤
1 ω χ 1+ω 2
− 81π 0
⎢ l2 (ρ2 +Mg)
⎥
2
4 π l(ρ +Mg)
⎢ ⎥
(T β ) = ⎢ ⎥,
2
α 1 χ
⎢ 0 8 π l2 (ρ2 +Mg) 0 ⎥ (11.380)
⎣ √
⎦
2 2
ω χ2 1+ω 2 1 χ (1+2 ω )
− 81π l3 (ρ2 +Mg) 0 8 π l2 (ρ2 +Mg)

with the following eigenvalues and their corresponding eigenvectors

1 χ2 ω
λ1 = − 2 2
; V1 = [V 1 , 0, √ V 1 ],
8 π l (ρ + Mg) l 1 + ω2
h 2
Vμ V μ = − V 1 , V1 = T1,
1 + ω2
1 χ2 ωl
λ2 = 2 2
; V2 = [ √ V 3 , V 2 , V 3 ],
8 π l (ρ + Mg) 1 + ω2
l2 L2 3 2 ρ2 2
V μ Vμ = V + V 2 , V2 = S2,
1 + ω2 h L2 l 2
1 χ2 ωl
λ3 = ; V3 = [ √ Ṽ 3 , Ṽ 2 , Ṽ 3 ],
8 π l2 (ρ2 + Mg) 1 + ω2
l2 L2 ρ2
V μ Vμ = ( Ṽ 3 2
) + (Ṽ 2 )2 , V3 = S3,
1 + ω2 h L2 l 2
Type : {T, 2 S}. (11.381)

This tensor structure corresponds to that of a perfect fluid energy momentum

tensor, but this time for the state equation: energy = pressure. Again, the solu-
tions generated from this metric by using coordinate transformations possesses
this perfect fluid feature because of the invariance of the eigenvalues.
The Cotton tensor for stationary cyclic symmetric gravitational field is
given by
⎡ 2 ω
√
1+ω 2 (h+L2 ) (ω2 h+L2 +L2 ω2 ) ⎤
χ2
− χ2 h2 l5
0 h2 l4
⎢ 2
⎥
⎢ ⎥
(C α β ) = ⎢ 0 0 0 ⎥. (11.382)
⎣ √
⎦
2 (h+ω2 h+L2 ω2 ) χ2 ω 1+ω 2 (h+L2 )
− χ2 h2 l6
0 2 h2 l 5
 11.13 Transformed Cataldo Hybrid Static Solution 235

Searching for its eigenvectors, one arrives at

ρ2 12
λ1 = 0; V1 = [0, V 2 , 0], Vμ V μ = 2 2V , V1 = S,
h L l
i χ2 L
λ2 = ;
2 h3/2 l5
√ √ √
V 3 l −i L h + ω 1 + ω 2 h + ω 1 + ω 2 L2
V2 = [ , 0, V 3 ], V2 = Z,
h + ω 2 h + L2 ω 2
i χ2 L
λ3 = − ;
2 h3/2 l5
√ √ √
V 3 l i L h + ω 1 + ω 2 h + ω 1 + ω 2 L2
V3 = [ , 0, V 3 ], V3 = Z̄,
h + ω 2 h + L2 ω 2
Type I : {T, Z, Z̄}. (11.383)
The eigenvectors V2 and V3 are complex conjugated while the vector V1, asso-
ciated to the zero eigenvalue, is the only physically meaningful spacelike direction
in this case.
The data of the stationary rotationally symmetric solutions generated by
means of SL(2, R) transformations can be represented schematically by the
table:

Table 11.12.1 Stationary electromagnetic solutions

Integration and SL(2, R) transformations

References via Eqns. Λ = ±1/l2

electrostatic seed SL–T § 11.11 −1/l2

Clément (1996) SL–T § 11.11.2 ±1/l2
Martı́nez et al. (2000) SL–T § 11.11.3 −1/l2
magnetostatic seed SL–T § 11.12 −1/l2
Dias and Lemos (2002) SL–T § 11.12.2 −1/l2
electromagnetic § 11.1.2 −1/l2
Kamata and Koikawa (1995) Int §11.6.7 ±1/l2
Ayón–Beato et al. (2005) Int § 11.7 ±1/l2
Garcı́a (2009) and Int § 11.9 ±1/l2
Garcia–Diaz (2013) Int (11.277), (11.9.3) ±1/l2

11.13 Transformed Cataldo Hybrid Static Solution

Applying the SL(2, R) transformation,
α β γ δ
t = √ t̃ + √ φ̃, φ = √ t̃ + √ φ̃, Δ = αδ − γβ,
Δ Δ Δ Δ
to the Cataldo azimuthal electrostatic solution (11.115) one arrives at a new
stationary solution having the static BTZ solution as a limit, namely
236 Einstein–Maxwell Solutions

α2 γ2 αβ
g = −( H(−) − H(+) )dt2 + 2(− H(−)
Δ Δ Δ
 2 −1
γδ β2 δ2 ρ
+ H(+) )dtdφ + (− H(−) + H(+) )dφ + 2 − M2
dr2 ,
Δ Δ Δ l
 2 (1−√α0 )/2  2 (1+√α0 )/2
√ ρ √ ρ
H(+) := ρ1+ α0 −M , H(−) := ρ1− α0 −M .
l2 l2
(11.384)

The electromagnetic tensors are given by

√
M 1 − α0
Fμν = δ[μ t δν] φ ,
2
M 2 (1 − α0 )
Tμ ν = (−δμ t δt ν + δμ ρ δρ ν − δμ φ δφ ν ). (11.385)
32πρ2 ρ2 /l2 − M

In the seed hybrid static metric α0 has been used instead of the original α to
avoid confusion with the parameters appearing in the SL(2, R) transformations.
Therefore the transformed electromagnetic tensors stay unchanged. Hence, by
means of an SL(2, R) transformation applied to the hybrid static cyclic symmet-
ric (2+1) Einstein–Maxwell solution, one generates a family of hybrid, stationary
cyclic symmetric solutions with structurally unique field tensors Fμν and Tμ ν .
In particular, one could choose the rotation boost transformation

1 1 ω
t→ (t − ωφ) , φ → φ− t ,
1− ω 2 /l2 1− ω 2 /l2 l2

where the parameter ω can be related to the angular momentum constant.

The generalization of the Cataldo static solution, via SL(2, R) transforma-
tions, is given by

1
ds2 = −N (ρ)2 dt2 + dρ2 + K(ρ)2 [dφ + W (ρ)dt]2 ,
L(ρ)2
ρ(ρ2 − M l2 )1/2  2 √α0 √α0 2 √
K(ρ)2 = δ l ρ (ρ − M l2 )− α0 /2
lΔ 
√ √ √
−β 2 l− α0 − α0
ρ (ρ2 − M l2 ) α0 /2
,
ρ2 ρ2 − M l2
N (ρ)2 = , Δ := (αδ − βγ) = 0,
l2 K(ρ)2
1 ρ ρ2 − M l2  √ √ √
W (ρ) = − 2
αβ l− α0 ρ− α0 (ρ2 − M l2 ) α0 /2
Δl K(ρ)
√ √ √ 
− γδ l α0
ρ α0 (ρ2 − M l2 )− α0 /2 ,
ρ2
L(ρ)2 = − M. (11.386)
l2
 11.13 Transformed Cataldo Hybrid Static Solution 237

11.13.1 Mass, Energy and Momentum

The corresponding surface densities are

1 M l2 √α0 √ M l2
(ρ, 0 ) = − β 2 (1 − 2 ) 2 (2 − (1 − α0 ) 2
2π l 1 − M l2 /ρ2 ρ ρ

√ M l2 √α0 √ M l2
− δ 2 l2 α0 (1 − 2 )− 2 (2 − (1 + α0 ) 2 )
ρ ρ
 −1
M l 2 √α0 √ M l2 √α0
× β 2 (1 − 2 ) 2 − δ 2 l2 α0 (1 − 2 )− 2 − 0 , (11.387)
ρ ρ
while the local momentum is
√
α0 M
j(ρ) = βδ , (11.388)
πΔ K(ρ)
therefore the product βδ is related with the rotation properties of the considered
solution. The integral quantities amount to
√
α0 M
J(ρ) = 2 π K(ρ) j(ρ) = 2βδ ,
Δ
E(ρ) = 2 π K(ρ) (ρ),
M (ρ) = N (ρ) E(ρ) − W (ρ) J(ρ). (11.389)

The evaluation of energy and mass functions independent of 0 behave at

infinity as
√ √
1 α0 l M δ 2 l2 α0 + β 2
(ρ → ∞, 0 = 0) ≈ − + √ ,
πl 2πρ2 δ 2 l2 α0 − β 2
√
√ δ 2 l2 α0 − β 2
−3/2− α0 /2
E(ρ → ∞, 0 = 0) ≈ −2ρ l √
Δ
√ √ √ √
√ M [( α0 − 1)β 4 + 2β 2 δ 2 l2 α0 − ( α0 + 1)δ 4 l4 α0 ]
−l 1/2− α0 /2
√ √ ,
2ρ Δ (δ 2 l2 α0 − β 2 )3/2

ρ2
M (ρ → ∞, 0 = 0) = M − 2
l2
 √ √ 
√ δ 2 l2 α0 + β 2 √ βδ αβ − γδ l2 α0
+M α0 2 2√α + 2 α0 √ . (11.390)
δ l 0 − β2 Δ δ 2 l2 α0 − β 2
Using in expressions (11.387)–(11.389) as reference energy density at the spatial
infinity the quantity which behaves as 0|∞ ≈ − π1l + 2πl M0
ρ2 , the series expansions
of the corresponding quantities at ρ → infinity result in
√
l √ l α0 M β2
(ρ → ∞, 0|∞ ) ≈ ( α 0 M − M 0 ) − √ ,
2π ρ2 πρ 2 δ 2 l α0 − β 2
2
238 Einstein–Maxwell Solutions
√
√ δ 2 l2 α0 − β 2 √
E(ρ → ∞, 0|∞ ) ≈ l1/2− α0 /2
√ ( α0 M − M0 )
ρ Δ
√
√
1/2− α0 /2 β 2 α0 M
+ 2l √ ,
ρ δ 2 l2 α0 − β 2
√
M (ρ → ∞, 0|∞ ) ≈ α0 M − M0
√ √
α0 M β γδ 2 l2 α0 − 2αβδ + β 2 γ
− √ . (11.391)
Δ δ 2 l2 α0 − β 2
Therefore, comparing with the energy characteristics of the BTZ solution, one
√
concludes that role of the mass parameter is played by the product α0 M . At
spatial infinity the mass function is finite, and the energy density and global
energy approach infinity as fast as 1/ρ2 and 1/ρ correspondingly.

11.14 Summary on Electromagnetic Maxwell Solutions

In the framework of the (2+1)-dimensional Einstein–Maxwell theory with cos-
mological constant, different families of exact solutions for cyclic symmetric sta-
tionary (static) metrics have been derived. For the static classes and also hybrid,
static and stationary families, their uniqueness is proven by the integration pro-
cedure used. The completeness and relationship of all uniform electromagnetic
Fμμ;α = 0, and constant invariant Fμν F μν = 2γ solutions is achieved. The
uniqueness of the stationary families of solutions has been partially established;
various specific branches of solutions in the general case are determined via a
straightforward integration. In this systematic approach all known electromag-
netic stationary cyclic symmetric solutions are properly identified. It seems to
be a rule that electrically charged solutions allow for a black hole interpretation,
while for the magnetic classes such a black hole feature seems to be absent.
Their energy–momentum densities and global energy–momentum–mass quan-
tities have been evaluated using the Brown–York approach. As reference
characteristics those corresponding to the stationary or static BTZ–AdS with
parameter M0 solutions have been used. The electric and magnetic solutions,
and their generalizations through SL(2, R) transformations exhibit at the spatial
infinity ρ → ∞ the following generic behavior

J(ρ → ∞) ≈ αJ J + βJ ln ρ,
l Q2
(ρ → ∞, 0|∞ (M0 )) ≈ (αM M − α M 0
M 0 ) + αQ
2π ρ2 2π ρ2
a A
+ J+ ln ρ,
2π ρ π ρ2
l Q2 b B
E(ρ → ∞, 0|∞ (M0 )) ≈ (βM M − βM0 M0 ) + β2 + J + 2 ln ρ,
ρ ρ ρ ρ
M (ρ → ∞, 0|∞ (M0 )) ≈ γM M − γM0 M0 + γQ Q + c J + C ln ρ,
2
 11.14 Summary on Electromagnetic Maxwell Solutions 239

where αJ , βM ,. . . , γQ are constant numerical factors related to the physical

parameters: J momentum, M mass, . . . , Q electromagnetic charge.
The momentum, energy and mass of the hybrid solutions behaves at spatial
infinity ρ → ∞ as follows
J(ρ → ∞) ≈ αJ J + βJ ,
l Q2 a
(ρ → ∞, 0|∞ (M0 )) ≈ 2
(αM M − αM0 M0 ) + αQ + J,
2π ρ 2π ρ2 2π ρ2
l Q2 b
E(ρ → ∞, 0|∞ (M0 )) ≈ (βM M − βM0 M0 ) + βQ + J,
ρ ρ ρ
M (ρ → ∞, 0|∞ (M0 )) ≈ γM M − γM0 M0 + γQ Q + c J,
2

where the charge Q is related to the electromagnetic parameter α.

Moreover, the eigenvectors for their electromagnetic field, Cotton and energy–
momentum tensors have been explicitly determined; the static and stationary
Peldan electric classes, the Martı́nez–Teitelboim–Zanelli and the Clément solu-
tions exhibit the following algebraic types:
Field : {S, N, N }, Energy : {S, 2T }, {S, 2N }, {S, 2S},
Cotton : {S, Z, Z̄},
while the static and stationary Peldan magnetic families, the Hirschmann–Welch
and the Dias–Lemos solutions exhibit the following algebraic types:
Field : {T, Z, Z̄}, Energy : {T, 2S}; Cotton : {S, Z, Z̄}.
The Garcı́a solution allows for the set of types:
Field : {T, Z, Z̄}, {S, Z, Z̄}, {T, 2N }, {S, 2N }, {3N },

Energy : {T, 2T }, {T, 2S}, {T, 2N } , · · · , {S, 2N }, {3N },

Cotton : {S, Z, Z̄}.

The hybrid Cataldo and Ayón–Cataldo–Garcı́a solutions fall into the types:
Field : {S, N, N }, Energy : {S, 2T }, {S, 2S}, {S, 2N },

Cotton : {S, Z, Z̄}.

The Kamata–Koikawa belongs to types:
Field : {3S}, Energy : {3S}, {3N }, Cotton : {3S}, {3N }.
Finally, the Matyjasek–Zaslavski solution exhibits the types:
Field : {S, N, N }, Energy : {S, 2T }, {S, 2S}, {S, 2N },

Cotton : {0}.
Recall that algebraic structures {T, 2S} are thought of as perfect fluids.
 12
Black Holes Coupled To Nonlinear
Electrodynamics

In (3+1) gravity it is well known that the vacuum plus a cosmological constant Λ,
i.e., the (anti) de Sitter–Kottler (1918) solution, is a regular non-asymptotically
flat solution (the scalar curvature is equal to 4Λ and all the invariants of the
conformal Weyl tensor are zero.) On the other hand, Einstein–Maxwell elec-
trovacuum asymptotically flat metrics endowed with timelike and spacelike
symmetries do not allow for the existence of regular black hole solutions. In
order to be able to derive regular (black hole) solutions one has to enlarge the
class of electrodynamics to nonlinear ones; as an example, the regular solution
by Ayón–Beato and Garcı́a (1998), which is a solution to gravitational–nonlinear
electromagnetic fields; the first examples belonging to this class are the Borde
(1997) model and the Ayón–Garcı́a regular charged static black hole.
In (2+1) gravity, in the vacuum case, all solutions are locally Minkowski (the
Riemann tensor is zero); the extension to the vacuum plus cosmological con-
stant allows for the existence of the static and the rotating anti-de Sitter regular
black holes; see Bañados et al. (1992). The static (2+1)-charged black hole with
cosmological constant is singular (when radial coordinate goes to zero the curva-
ture and the Ricci square invariants blow up). Similarly, as in (3+1) gravity, one
may search for regular solutions in (2+1) gravity incorporating nonlinear elec-
tromagnetic fields to which one imposes the weak energy conditions in order
to have physically plausible matter-field distributions. One may look for re-
gular solutions with nonlinear electromagnetic fields of the Born–Infeld type
(Born and Infeld, 1934); Salazar et al. (1987); Salazar et. al (1984); Gibbons
and Rasheed (1995); Fradkin and Tseytlin (1985); Deser and Gibbons (1998),
and/or electrodynamics of wider spectra (Cataldo and Garcı́a, 1999, 2000;
Cataldo et al., 2000).
The Born–Infeld electrodynamics (Born and Infeld, 1934) is free from certain
singularities appearing in the classical Maxwell electromagnetic field theory. Non-
linear electromagnetic Lagrangians, in particular the Born–Infeld Lagrangian,
arise in open string theory (the low-energy effective action for an electromagnetic
 12.1 Nonlinear Electrodynamics in (2 + 1) Dimensions 241

field is precisely the Born–Infeld action; see Fradkin and Tseytlin, 1985); string
theory has emerged as the most promising candidate for the consistent quanti-
zation of gravity. In particular, the open string theory has Born–Infeld coupled
vector fields, but it is not clear that this remains the case after compactification
to a 3D space with negative cosmological constant Λ. It is worth mentioning
that there is renewed interest in Born–Infeld theory in various contexts (Frolov
et al., 1996 and Salazar et al., 1987).
Using our experience in the determination of exact solutions in the standard
Einstein gravity with nonlinear electromagnetic fields, in this chapter we deal
with (2 + 1) solutions, mostly static, having nonlinear electrodynamics sources
depending on a single electromagnetic invariant. Stationary generalizations of
the derived static solutions can be constructed using SL(2, R) transformations.

12.1 Nonlinear Electrodynamics in (2 + 1) Dimensions

In this section, we are using electromagnetic Lagrangian L(F ) depending upon
a single invariant F = 1/4F ab Fab , which we demand in the weak field limit to
be equal to the Maxwell Lagrangian

L(F ) −→ −F/4π,

the corresponding energy–momentum tensor has to fulfill the weak energy con-
ditions: for any timelike vector ua , ua ua = −1 (we are using signature – + +)
one requires
Tab ua ub ≥ 0, qa q a ≤ 0,

where q a = Tba ub . This invariant F can be expressed in terms of the electric (vec-
tor) and magnetic (scalar) fields: in a Lorentzian frame, for an observer moving
with the 3-velocity v a , the electric and the magnetic fields are correspondingly
defined as
1
Ea = −Fab v b , B = − abc Fbc v a , (12.1)
2
where Latin indices, to keep the notation used in the original publications, run
the values 0, 1, 2 and abc is the totally anti-symmetric Levi–Civita symbol with
012 = 1, usually the v a is oriented along the time coordinate, i.e., v a = δta , with
such a choice
Ea = F0a , B = −F12 . (12.2)

Thus the invariant can be expressed by F ≡ 14 F ab Fab = 12 (B 2 − E 2 ).

In general one can construct a (2+1)-Einstein theory coupled with nonlinear
electrodynamics starting from the action
  
√ 1
S= −g (R − 2Λ) + L(F ) d3 x, (12.3)
16π
242 Black Holes Coupled To Nonlinear Electrodynamics

with the electromagnetic Lagrangian L(F ) unspecified explicitly at this stage.

We are using units in which c = G = 1. The ambiguity in the definition of
the gravitational constant (there is no Newtonian gravitational limit in (2+1)
dimensions) allows us to maintain the factor 1/16π in the action to keep the
parallelism with (3+1) gravity. Varying this action with respect to gravitational
field gives the Einstein equations

Gab + Λgab = 8πTab ,

Tab = gab L(F ) − Fac Fb c L,F , (12.4)

while the variation with respect to the electromagnetic potential Aa entering in

Fab = Ab,a − Aa,b , yields the electromagnetic field equations
 
∇a F ab L,F = 0, (12.5)

where L,F stands for the derivative of L(F ) with respect to F .

12.2 General Nonlinear Electrostatic Solution

In this section a (2 + 1) static family of (black hole) metrics coupled to nonlinear
electric fields is derived; the source to the Einstein equations is a nonlinear
electrodynamics, satisfying the weak energy conditions, which in the weak field
limit becomes the (2+1)-Maxwell field.
Concrete solutions to the dynamical equations above we present for the static
metric
dr2
ds2 = −f (r)dt2 + + r2 dΩ2 , (12.6)
f (r)
where f (r) is an unknown function of the variable r.
Assume that the electromagnetic field Fμν possesses non-vanishing compo-
nents E := Ftr and B := FΩr . It occurs that the scalar magnetic field B,
vanishes as a consequence of the Einstein equations; the Ricci tensor components,
evaluated for the metric (12.6), yield the following relation

A := Rtt + f 2 Rrr = 0, (12.7)

while the evaluation of the same relation using the electromagnetic energy-
momentum gives
f
A = −8πL,F ( B)2 . (12.8)
r
Therefore, the scalar magnetic field B should be equated to zero, B = 0, thus
the only case to be treated is just the one with the electric field E,
 
Fab = E(r) δat δbr − δar δbt . (12.9)
 12.2 General Nonlinear Electrostatic Solution 243

The Maxwell equations reduce to

   
d d f
(rEL,F ) = 0, BL,F =0 . (12.10)
dr dr r
Therefore, Maxwell equations allow for the integral
q
E(r)L,F = − . (12.11)
4πr
where q is an integration constant chosen in that way in order to obtain the
Maxwell limit.
The invariant F , which reduces now to

2F = −E 2 (r), (12.12)

permits to express the electric field E in term of F . Thus, using (12.12), the
derivative, LF can be expressed as a function of r as follows
q
L,r = E,r . (12.13)
4πr
The Einstein’s equations equivalently can be written as

Rab = 8π (Tab − T gab ) + 2Λgab . (12.14)

From (12.4) using (12.9) and (12.12) the trace becomes

T = 3L(F ) + 2E 2 (r)L,F . (12.15)

As was pointed out above, the Lagrangian L(F ) must satisfy: (i) correspondence
to Maxwell theory, i.e. L(F ) −→ −L/4π, and (ii) the weak energy conditions:
Tab ua ub ≥ 0 and qa q a ≤ 0, where q a = Tba ub for any timelike vector ua ; in our
case the first inequality requires

− (L + E 2 L,F ) ≥ 0, (12.16)

which can be stated equivalently as

q
L ≤ EL,E −→ L ≤ E. (12.17)
4πr
The norm of the energy flux qa is always less than or equal to zero; for ua
√
along the time coordinate, ua = δta / −gtt , one has the inequality qa q a =
−(L + L,F E 2 )2 ≤ 0.
As far as the Einstein equations are concerned, the Rtt (= −f 2 Rrr ) and RΩΩ
components yield respectively the equations
f,r  
f,rr + = −4Λ + 16π 2L(F ) + E 2 L,F , (12.18)
r
 
f,r = −2Λr + 16πr L(F ) + E 2 L,F . (12.19)
244 Black Holes Coupled To Nonlinear Electrodynamics

If one replaces f,r from (12.19) and its derivative f,rr into (12.18) one arrives,
taking into account the equation (12.13), at an identity. Therefore one can forget
the equation (12.18) and integrate the relevant Einstein equation (12.19):

f (r) = −M − Λr2 + 16π r L(F (r)) + E 2 L,F dr. (12.20)

Summarizing, we have obtained a wide class of solutions, depending on a

Lagrangian L(E), given by:
the metric
dr2
ds2 = −f (r)dt2 + + r2 dΩ2 , (12.21)
f (r)
the structural function
   
E,r
f (r) = −M − (Λ − 2C)r2 + 4q r dr − E dr, (12.22)
r
which is obtained from (12.20) by using (12.13) and (12.12), where C is a constant
of integration, and the Lagrangian L(E) is constrained to
q
L,r = E,r . (12.23)
4πr
We recall that the Lagrangian and the energy momentum tensor have to fulfill
the conditions (12.17) quoted above.
In the forthcoming sections, various particular classes are presented for
different choices of the Lagrangian functions.

12.2.1 Static Charged Peldan Solution

As a particular example belonging to the linear Maxwell theory, one has the
static charged Peldan (1993) solution characterized by the function
r2
− gtt = f = −M + − 2q 2 ln r, (12.24)
l2
the Lagrangian and the electric field
1 1 2 1 q2 q
L(E) = − F = E = , E(r) = , (12.25)
4π 8π 8π r2 r
where C in (12.22) has been equated to zero and Λ = −1/l2 . It is worth pointing
out that the static charged Peldan black hole is singular at r = 0.

12.3 Cataldo–Garcı́a Nonlinear EBI Charged Black Hole

A nonlinear charged version of the (2 + 1) AdS black hole solution is derived.
The source to the Einstein equations is a Born–Infeld electromagnetic field,
which in the weak field limit becomes the (2+1)-Maxwell field. The obtained
 12.3 Cataldo–Garcı́a Nonlinear EBI Charged Black Hole 245

Einstein–Born–Infeld solution for certain range of the parameters (mass, charge,

cosmological and the Born–Infeld constants) represents the static circularly
symmetric Cataldo and Garcı́a (1999) black hole.
The Born–Infeld nonlinear electrodynamics Lagrangian is given by
 
b2 F
L(F ) = − 1+2 2 −1 , (12.26)
4π b
where the constant b is the Born–Infeld parameter. Notice that this Lagrangian
reduces to the Maxwell one in the limit when b2 −→ ∞, L(F ) = −F/4π. The
field equations of the Einstein–Born–Infeld (EBI) theory, amount to
  
Fac Fb c F
Gab + Λgab = 2 − b gab ( 1 + 2 2 − 1) ,
2
(12.27)
1 + 2F/b2 b
together with the electromagnetic field equations
 
F ab
∇a = 0. (12.28)
1 + 2F/b2
As a concrete solution of the Einstein–Born–Infeld dynamical equations we
present a static self-consistent solution with metric (12.6). The invariant F is
given by 2F = −E 2 (r). The electromagnetic field equations (12.5) yield
 
rE(r)
∂r = 0, (12.29)
1 − E 2 /b2
which integrates as
q
E(r) = , (12.30)
r2 + q 2 /b2
where q is an integration constant having the meaning of the charge, as one could
expect. One also arrives at this result by integrating the master equation (12.23).
In the Maxwell limit, we obtain from the last expression the right E = q/r in
(2+1) dimensions. The Born–Infeld field is characterized by a charge density
distribution ρe , which can be evaluated from the Maxwell equations divE  =

∇E = 2πρe , which in the considered case amount to

 = 1 d
divE (rE(r)) = 2πρe , (12.31)
r dr
substituting here E(r) from (12.30) one obtains
qr02
ρe = , (12.32)
2πr(r2+ r02 )3/2
where r0 = q/b. It is easy to verify that the surface integral of ρe is equal to q,
in fact  ∞  ∞
dr
ρe dA = qr02 2 + r 2 )3/2
= q. (12.33)
0 0 (r 0
246 Black Holes Coupled To Nonlinear Electrodynamics

It is worth pointing out the regular behavior of the vector E of the electric field
and the surface charge distribution ρe ; the same regular behavior one encounters
for the static spherically symmetric electric field in (3+1) Born–Infeld theory.

12.3.1 Static Cyclic Symmetric EBI Solution

As far as the Einstein equation are concerned, the Rtt and RΩΩ components yield
respectively the equations
 
f,r 4q 2 r
f,rr + = −4Λ − − 8b 2
− 1 , (12.34)
r r r2 + q 2 /b2 r2 + q 2 /b2
 
4q 2 r
f,r = −2Λr − − 4b r
2
− 1 . (12.35)
r2 + q 2 /b2 r2 + q 2 /b2
The general integral of the RΩΩ –equation is given by

f (r) = −M − (Λ − 2b2 )r2 − 2b2 r r2 + q 2 /b2 − 2q 2 ln(r + r2 + q 2 /b2 ).

(12.36)

From this last expression one sees that there is a contribution of the Born–Infeld
field to the term with the cosmological constant. The Lagrangian and the electric
field are given by
   
b2 F b2 r
L(F ) = − 1+2 2 −1 =− −1 ,
4π b 4π r2 + q 2 /b2
q
E(r) = . (12.37)
r + q 2 /b2
2

Invariants
Having the explicit metric, one can easily calculate the curvature tensor
components:
2b2 r
R0110 = 2b2 − Λ − , (12.38)
r2 + q 2 /b2
  
q2
R0202 = f (r) Λr − 2b r + 2b r
2 2 2 2
r2 + 2 , (12.39)
b

and
  
−1 q2
R1221 = f (r) Λr − 2b r + 2b r
2 2 2 2
r2 + 2 . (12.40)
b

It is surprising that the covariant metric and curvature components do not

exhibit a singular behavior in the neighborhood of the origin at r = 0. Never-
theless, this solution is singular at r = 0 in the sense that its invariant curvature
 12.3 Cataldo–Garcı́a Nonlinear EBI Charged Black Hole 247

characteristics – Ricci and Kretschmann scalars – such as the Ricci scalar and
the Ricci square blow up at r = 0, (the Riemann square does not need to be
evaluated since in (2+1) dimensions the Riemann tensor is given in terms of
the Ricci tensor, curvature scalar and the metric tensor). Additionally, in (2+1)
dimensions one considers the behavior of the invariant det(Rab )/ det(gab ), thus
one has to evaluate the invariants (Weinberg, 1972)
det(Rab )
R, Rab Rab ,
det(gab )
at critical points. In our case these three invariants are given as
4(2q 4 + 5q 2 b2 r2 + 3b4 r4 )
R = 6Λ − 12b2 + 3/2
, (12.41)
b2 r (r2 + q 2 /b2 )

8b2 (4b2 − 2Λ) 3r2 + 2q 2 /b2
Rab Rab = 3(4b − 2Λ) −
2 2
r r2 + q 2 /b2
 
r2 3(r2 + q 2 /b2 )2
+8b4 2 + 2 + , (12.42)
r + q 2 /b2 r2
and
  2
det Rab 2Λ − 4b2 2b2 (2r2 + q 2 /b2 )
= 4b2
−2Λ + 4b −
2
.
det gab r2 + r3 r2 + q 2 /b2 r r2 + q 2 /b2
(12.43)

Since these scalars go to infinity at r → 0, we conclude that they are singular at

this point.

12.3.2 Cataldo–Garcı́a Black Hole to EBI

This solution is a black hole. To establish this assertion one has to demonstrate
the existence of horizons, which require the vanishing of the gtt component, i.e.,
f (r) = 0. The roots of this equation give the location of the horizons (inner and
outer in our case). Since this equation is a transcendent one, we are not able,
as is usual for charged (2+1) black holes, to express the roots analytically, even
for the charged static Peldan solution; the roots are expressed in terms of the
Lambert W (x) function. To overcome this difficulty we study the extreme case,
in which the derivative of ∂r (f (r)) = 0 gives
2qb
rextr = √ >0 (12.44)
Λ2 − 4b2 Λ
for Λ < 0. Now entering rextr into f (r) = 0 one obtains a relation between
mass, charge, cosmological constant and the Born–Infeld parameter, which can
be solved explicitly for the mass (the extreme one):
248 Black Holes Coupled To Nonlinear Electrodynamics
  
q Λ − 4b 2
Mextr = −2q 2 ln . (12.45)
b Λ

We have an extreme black hole if Λ < 0, Mextr > 0 and q 2 < b2 ; this last con-
straint arises from Λ < 4b2 q 2 /(q 2 − b2 ) when one demands Mextr > 0. Fixing
the values of the Mextr for given values of q, b and Λ, one has a black hole
solution with inner and outer horizons when M > Mextr . For M < Mextr one
has a soliton solution, i.e., there are no horizons at all and we have a naked
singularity. It may occur that for certain values of the parameters there is
only one positive root of the equation f (r) = 0: the horizon rh > 0. In such
a case one has also a black hole solution. A similar analysis can be carried
out for other rextr which arises for Λ > 0. At infinity, for weak electromag-
netic field this BI solution asymptotically behaves as the static charged Peldan
solution.
Moreover, for Λ = −1/l2 this BI solution at infinity behaves as anti-de Sitter
spacetime. At the origin r = 0, this BI solution is singular. If one requires
additionally the vanishing of the cosmological constant, one arrives at a solution
reported in Gott et al. (1986).
As regards the analytical extension of our solution, one may follow step by
step the procedure presented in standard textbooks (for instance, Wald, 1984)
to determine the Kruskal–Szekeres  coordinates. First one has to integrate for
the tortoise r∗ coordinate: r∗ = 1/f (r)dr, which in our case has no expression
in terms of elementary functions; next one defines the null coordinate u and v
by u = t − r∗ , v = t + r∗ ; in these coordinates the studied metric acquires the
form
ds2 = −f (r)dudv + r2 dΩ2 , (12.46)

where r has to be interpreted as function of u and v, r∗ = r∗ (u, v). Further, one

introduces null Kruskal–Szekeres coordinates U = −e−αu and V = eβV where
α and β are to be chosen appropriately, finally one introduces the Kruskal–
Szekeres coordinates T = (U + V )/2, X= (V-U)/2, arriving at the Kruskal
extension.
If one were interested in the thermodynamics of the obtained solution one
would need to evaluate the temperature of the black hole, which is given in
terms of its surface gravity by Visser (1992); Brown et al. (1994):


kB TH = k. (12.47)
2π
For a spherically symmetric (and for circularly symmetric in (2+1) dimensions)
system the surface gravity can be computed via (for our signature)
 
1 ∂r gtt
k = − lim √ , (12.48)
r→r+ 2 −gtt grr
 12.4 Regular Black Hole Solution 249

where r+ is the outermost horizon. For our solution we have from (12.22), (12.47
and (12.48) that
  
 q2
kB T = −2(Λ − 2b )r+ − 4b r+ + 2 .
2 2 2 (12.49)
4π b

Since in our case there is no analytical expression of r+ in terms of elementary

functions, one cannot give a parameter-dependent expression of (12.49). It is easy
to check that when q = 0, T in (12.49) reduces to the BTZ temperature. In the
extremal case (12.44), the temperature vanishes in (12.49). The entropy can be
trivially obtained using the entropy formula S = 4πr+ . Other thermodynamic
quantities such as heat capacity and chemical potential can be computed as
in Brown et al. (1994). We recall that most of these quantities in the literature
are evaluated for metrics given in terms of polynomial functions.
Notice that the 4D Einstein–Born–Infeld counterpart – the Kottler–Born–
Infeld black hole; Salazar et. al (1984); Gibbons and Rasheed (1995) – can be
given by the metric (12.6) with
2
f (r) = 1 − 2M/r − (Λ/3 − 2b2 /3)r2 − b2 r4 + q 2 /b2
3

q2 r dr
−2 , (12.50)
3r r + q 2 /b2
4

where now dΩ2 = dθ2 + sin2 θdφ2 . As in the (2+1) case, there is also a contribu-
tion to the cosmological constant term of the nonlinear field. The corresponding
electric field is given by
q
E(r) = . (12.51)
r + q 2 /b2
4

Notice that the electric field in this case is regular everywhere. This gravitational
field asymptotically behaves as the Kottler charged solution, with the structural
function and electromagnetic field of the form
 
2M Λ 2 q2 1
f (r) = 1 − − r + 2 +O ,
r 3 r r6
 
q 1
E(r) = 2 + O .
r r3
By canceling Λ one obtains an asymptotically flat solution.

12.4 Regular Black Hole Solution

Following Ayón–Beato and Garcı́a (1998), a static class of regular everywhere
solution can be constructed; it is given by the metric (12.6) with structural
function of the form
f (r) = −M − Λr2 − q 2 ln(r2 + a2 ) (12.52)
250 Black Holes Coupled To Nonlinear Electrodynamics

where M , a, q and Λ are free parameters. The corresponding Lagrangian and

the electric field are given by
q 2 (r2 − a2 ) r3
L(r) = , E(r) = q . (12.53)
8π (r2 + a2 )2 (r2 + a2 )2
L(r) and the electric field satisfy the weak energy conditions (12.17). To express
the Lagrangian in terms of F or equivalently E, one has to write r in terms of E
by solving the quartic equation for r(E), this will give rise an explicit r containing
radicals of E, which introduced in L(r), finally will bring L as function of E.
The expression L(E) is not quite illuminating, thus we omit it here.

12.4.1 Regularity
To establish that this solution is regular one has to evaluate the curvature Ricci
and Kretschmann scalars. The non-vanishing curvature components, which are
regular at r = 0, are given by:
q 2 (a2 − r2 )
R0110 = + Λ, (12.54)
(r2 + a2 )2
 2 2 
q r
R0202 = −f (r) + Λr 2
, (12.55)
r2 + a2
 2 2 
q r
R1212 = f (r)−1 + Λr 2
, (12.56)
r2 + a2
where 0,1,2 stand respectively for t, r and Ω.
Evaluating the invariants R, and Rab Rab one has
2q 2 (r2 + 3a2 )
R= + 6Λ (12.57)
(r2 + a2 )2
r4 + 2r2 a2 + 3a4
Rab Rab = 12Λ2 + 4q 4
(r2 + a2 )4
8Λq 2 (3a2 + r2 )
+ . (12.58)
(r2 + a2 )2
Since the metric, the electric field and these invariants behave regularly for all
values of r, we conclude that this solution is curvature regular everywhere. Nev-
ertheless, for solutions without any horizon or black hole solutions with an inner
and outer horizons, at r = 0 a conical singularity may arise. At r = 0 the function
f (r) becomes f (0) = −M − q 2 ln(a2 ). Thus for M positive, M > 0, and a in the
range 0 < a < 1, the value of f (0) will be f (0) = −M + q 2 ln(1/a)2 , which will
be positive, say f (0) := β 2 , if ln(1/a)2 > M/q 2 . In such a case, for 0 < β < 1 the
solutions will show angular deficit since the angular variable Ω, which originally
runs 0 ≤ Ω < 2π will now run 0 ≤ Ω < 2βπ; the parameter a can be expressed in
terms of β, q and M as a2 = exp[−(β 2 +M )/q 2 ]. For β = 1, there will be no angu-
lar deficit, the ratio of the perimeter of a small circle around r = 0 to its radius,
 12.4 Regular Black Hole Solution 251

as this last tends to zero, will be 2π. If one allows M to be negative, M < 0,
and a to take values in the interval 0 < a < 1, then f (0) will be always posi-
tive; in this case one can adopt the following parametrization: −M = β 2 cos2 α,
q 2 ln(1/a)2 = β 2 sin2 α, therefore f (0) = β 2 . One will have angular deficit if
0 < β < 1, and for β = 1 the resulting (2+1) spacetime will be free of singular-
ities. Another possibility with positive f (0) = β 2 arises for M < 0, and a > 1;
f (0) can be parameterized as −M = β 2 cosh2 α, q 2 ln(1/a)2 = β 2 sinh2 α. Again
the values taken by β will govern the existence of angular deficit, for β = 1
the solutions will be regular. If f (0) is negative, f (0) =: −β 2 , the character of
the coordinates t and r changes, the coordinate t becomes spacelike, while r is
now timelike and one could think of the singularities, if any, as causal structure
singularities because they could arise at the “time” r = 0. In what follows we
shall treat the parameter a as a free one, having in mind the above restrictions
to have solutions free of conical singularities.

12.4.2 Horizons
To establish that this solution represents a black hole, one has to demonstrate
the existence of horizons, which require the vanishing of the gtt component, i.e.,
f (r) = 0. The roots of this equation give the location of the horizons (inner and
outer in our case). The roots – at most four – of the equation f (r) = 0 can be
expressed in terms of the Lambert W (r) function
 2 
Λa2 − M Λ Λa − M 1
r1,2,3,4 = ±[exp( − LW [ exp ]) − a2 ] 2 .
q2 q2 q2
There arise various cases which depend upon the values of the parameters: four
real roots (two positive and two negative roots: the negative roots have to be
ignored), two complex and two real roots, two complex and one real positive
root (the extreme case), and four complex roots (no black hole solutions). This
analytical expression for the Lambert function can be used in all calculations,
recall that Lambert function fulfills the following equation

ln(LW (x)) + LW (x) = ln(x).

Analytically one can completely treat the extreme black hole case; for it, the
derivative of f (r) has to be zero, ∂r (f (r)) = 0 , at the rextr , this gives

q2
rextr = −a2 − >0 (12.59)
Λ
for Λ < 0. From this expression one concludes that the following inequality holds:
a2 < −q 2 /Λ. Entering now rextr into f (r) = 0 one obtains a relation between the
parameters involved, which can be solved explicitly for the mass (the extreme
one):
252 Black Holes Coupled To Nonlinear Electrodynamics
  
2 2 −Λ
Mextr = a Λ + q 1 + ln , (12.60)
q2
this Mextr varies its values depending on the values given to the parameters a,
q and Λ. We have an extreme black hole characterized by negative cosmological
constant, Λ < 0, and positive extreme mass, Mextr > 0, if the parameter a is
restricted by the inequality a2 < −(q 2 (1 + ln(−Λ/q 2 )))/Λ.
For other values of the mass M , one distinguishes the following branches:
if M > Mextr one has a black hole solution, and if M < Mextr there are no
horizons.

12.4.3 Thermodynamics
Similarly, as this question was treated in the previous section, here one can also
evaluate the temperature (12.47), of the black hole through its surface gravity
(12.48). For our present solution we have from (12.22), (12.47) and (12.48) that
 
 q 2 r+
kB T = −Λr+ − 2 . (12.61)
2π r+ + a2

The entropy can be trivially obtained using the entropy formula S = 4πr+ .
To achieve the maximal extension of our regular black solutions one has to
proceed step by step through the procedure presented above, determining first
the Kruskal–Szekeres coordinates.

12.5 Coulomb-Like Black Hole Solution

The electromagnetic tensor for a Coulomb potential in (3 + 1) Maxwell electro-
dynamics is trace free, T = Tab g ab = 0, Thus, it is of some interest to establish
which solution arises if one requires the vanishing of the trace T in (2 + 1)
electrodynamics. Demanding the vanishing of

T = Tab g ab = 3L(F ) − 4F L,F , (12.62)

one obtains
L = C |F |3/4 , (12.63)

where C is a constant of integration, and bars denote moduli. Taking into account
(12.12) one gets that
L = C E 3/2 . (12.64)

Entering this L into (12.13) one gets that E = (q 2 /6πC)2 1/r2 . By choosing now
C = |q|/6π we arrive at the electric field
q
E(r) = , (12.65)
r2
 12.5 Coulomb-Like Black Hole Solution 253

which coincides with the standard Coulomb field for a point charge of the
Maxwell theory in the Minkowski space.
The Lagrangian in this case is given by

q2 |q| 3/2
L= = E . (12.66)
6πr3 6π
It is easy to check that this Lagrangian satisfies the weak energy conditions:
qa q a ≤ 0, where q a = Tba ub for any timelike vector ua and on the other hand

q2
− (L + E 2 L,F ) = ≥ 0. (12.67)
12πr3
We rewrite the Einstein’s equations equivalently as

Rab = 8πTab + 2Λgab , (12.68)

where it has been considered that the trace of Tab is equal to zero, T = 0. The
Einstein equations for Rtt (= −f 2 Rrr ) and RΩΩ components yield respectively
the equations

f,r 2q 2
f,rr + = −2Λ + 3 , (12.69)
r 3r
4q 2
f,r = −2Λr − 2 . (12.70)
3r
It is easy to show that equation (12.69), by virtue equation the Maxwell equa-
tions is just an identity. Therefore the only Einstein equation to be integrated
is (12.70), which gives
4q 2
f (r) = C0 − Λr2 + , (12.71)
3r
where C0 is a constant of integration. We will see now that the constant C0
can be expressed in terms of asymptotic values of the mass. For the circularly
symmetric metric (12.6) the quasilocal energy E(r) and the quasilocal mass M (r)
at a radial boundary r can be shown to be respectively

E = 2( f0 (r) − f (r)), (12.72)

M (r) = f (r)E(r), (12.73)

where f0 (r) = g0rr (r) is a background metric function which determines the zero
of the energy. The function f0 (r) can be obtained simply by setting constants of
integration of our solution (12.71) to some special values that then specify the
reference spacetime. We set in (12.71) q = C0 = 0 as the background and as
a consequence it is the vacuum anti-de Sitter spacetime. The same background
function was used in Brown et al. (1994); Chan (1996) for analogous calcula-
tions. Now for Λ = −1/l2 < 0 we have that f0 (r) = r/l (for a de Sitter
254 Black Holes Coupled To Nonlinear Electrodynamics

spacetime Λ > 0) and then the quasi-local energy and the quasi-local mass are
given respectively by

r r2 4 q2
E(r) = − C0 + 2 + , (12.74)
l l 3 r
  
r r2 4 q2 r2 4 q2
M (r) = 2 C0 + 2 + − 2 C0 + 2 + . (12.75)
l l 3 r l 3 r

As r −→ ∞, the analogous ADM mass is defined to be M =: M (∞). In our case

we see from (12.74) and (12.75) that E(∞) vanishes and M (∞) =: M = −C0
respectively. Then we get that the constant C0 has the sense of the asymptotic
observable mass M > 0 and we can write
4q 2
f (r) = −M − Λr2 + . (12.76)
3r
In the following we analyze the obtained solutions. The case Λ > 0 corresponds
to an asymptotically de Sitter spacetime. For Λ < 0, one is dealing with an
asymptotically anti-de Sitter spacetime. For vanishing cosmological constant,
Λ = 0, one has an asymptotically flat solution coupled with a Coulomb-like field.

12.5.1 Horizons for the Coulomb-Like Solution

To establish the existence of horizons, one has to require the vanishing of the gtt
component, i.e., f (r) = 0. The roots of this equation are
h M
r1 = − , (12.77)
3Λ h √  
h M i 3 h M
r2 = − + + + , (12.78)
6Λ 2h 2 3Λ h
√  
h M i 3 h M
r3 = − + − + , (12.79)
6Λ 2h 2 3Λ h
where
    1/3
2 M 3 + 12q 4 Λ
h= 18q + 3 3 Λ2 . (12.80)
Λ

These equations give the location of the horizons (if there are any). Since the
coordinate r ranges over positive values from 0 to infinity, we exclude the roots
which are negative. Because the positive character of M > 0, the complex or
real character of these roots depends crucially on Λ values. For Λ > 0 there is
only one real root. For Λ < 0 there are two possibilities: two complex roots and
one real; or three real roots. To obtain only
√ real roots we must cancel the term
h/3Λ + M/h. Then we have that h = ± −3ΛM . If M /Λ + 12q 4 = 0 then one
3
 12.5 Coulomb-Like Black Hole Solution 255

has an extreme black hole, in such case M = Mextr = −(12q 4 Λ)1/3 and the roots
become
 1/3  1/3
2q 2 2q 2
r1 = 2 , r2 = r3 = rextr = − . (12.81)
3Λ 3Λ

From these expressions we see that for Λ > 0 there is not a positive rextr ,
although the horizon r1 > 0 the behavior of the f function is quite peculiar, for
r > r1 , f (r) < 0, while for r < r1 , f (r) > 0. For Λ < 0, we have a positive
extreme horizon (since r1 < 0 it does no represent an horizon).
For Λ > 0 (and M > 0) we have that α > 0 and then we always have one
real root r1 > 0 and two complex roots r2 and r3 . From (12.71), hence f (r) is a
decreasing function for r > 0.
For Λ < 0 the roots r2 and r3 can be real or complex roots (in this case always
r1 < 0). For M 3 + 12q 4 Λ > 0 or q 2 > −M/3Λ/2 we have that these roots are
complex. For M 3 + 12q 4 Λ < 0 or 0 ≤ q 2 < −M/3Λ/2 we can write the real
roots as
 ⎛ ⎡ ⎤ ⎞
−M 1 2q 2
2π
r± = −2 cos ⎝ arccos ⎣  ⎦± ⎠. (12.82)
3Λ 3 −M
3 3
3Λ

In this case these roots represent the horizons of a black hole; the outer horizon
(the event horizon) is r+ , while the inner horizon is r− . When q = 0 we have
that r− = 0 and r+ = −M/Λ, which is the horizon of the static BTZ metric.
Using the negative cosmological constant as Λ = −1/l2 one gets
⎛  ⎛ ⎡ ⎤ ⎞⎞
 ⎝ 16M 1 2q 2
2π
kB T = − cos ⎝ arccos ⎣  ⎦+ ⎠⎠
4π 3l2 3 M 3 l2 3
3
⎛ ⎡ ⎤ ⎞
q2 1 2q 2
2π
− cos−2 ⎝ arccos ⎣  ⎦+ ⎠. (12.83)
M l2 3 M 3 l2 3
3

It is easy to check that when q = 0, T in (12.83) reduces to the BTZ temperature.

In the extremal case (12.81), the temperature vanishes in (12.83). The maxi-
mal extension of black solutions can be achieved by using the Kruskal–Szekeres
approach.
Another interesting case arises when Λ = 0, yielding a charged asymptotically
flat (2+1)-dimensional black hole with a cosmological horizon at

4q 2 3 M 2
rH = ,T =− < 0. (12.84)
3M 16π q 4
256 Black Holes Coupled To Nonlinear Electrodynamics

12.6 Stationary Nonlinear Electrodynamics Black Holes

The standard way to search for solutions within the class of stationary cyclic
symmetric metrics:
1
ds2 = −N (ρ)2 dt2 + dρ2 + K(ρ)2 [dφ + W (ρ)dt]2 ,
L(ρ)2
for nonlinear electrodynamics determined by a Lagrangian L(F ) depending upon
a single invariant F = 1/4F ab Fab , to which one demands in the weak field limit
to become the Maxwell Lagrangian L(F ) −→ −F/4π, consists in solving the
Einstein equations
Gab + Λgab = 8πTab ,
with
Tab = gab L(F ) − Fac Fb c L,F ,
and the electromagnetic field equations
 
∇a F ab L,F = 0,
where L,F stands for the derivative of L(F ) with respect to F . The field Fab =
Ab,a − Aa,b , where Aa is the electromagnetic potential Aa . Although this task is
solvable for particular cases, in general is time consuming.
Thus, as in the search for Maxwell stationary (2+1) solutions, it is easier to
generate solutions via SL(2, R) transformations applied to solutions given by
(diagonal) static metrics. Consider the static metric
g = gtt dt2 + gφφ dφ2 + gρρ dρ2 ,
and accomplish here a SL(2, R) – transformations of the Killingian coordinates
t and φ
t = αt̃ + β φ̃, φ = γ t̃ + δ φ̃, Δ := αδ − βγ = 0,
then the transformed metric components will be
gt̃t̃ = α2 gtt + γ 2 gφφ , gt̃φ̃ = αβ gtt + γδ gφφ ,
gφ̃φ̃ = β 2 gtt + δ 2 gφφ , gρρ = gρρ .

Requiring the new coordinate φ̃ to be cyclic, 0 ≤ φ̃ ≤ 2π, one is generating

stationary solutions from static ones. In this manner one generates the stationary
nonlinear solution with arbitrary L(F (ρ)), the stationary Cataldo–Garcı́a black
hole to EBI, the stationary generalization of the regular and of the Coulomb-like
solutions. Explicit expressions are easily derivable.
 13
Dilaton Field Minimally Coupled to (2 + 1)
Gravity

Using the Schwarzschild coordinate frame for a static cyclic symmetric met-
ric in (2 + 1) gravity coupled minimally to a dilaton logarithmically depending
on the radial coordinate in the presence of an exponential potential together
with a Maxwell electric field, by solving first-order linear Einstein equations,
the general solution is derived and identified with the Chan–Mann dilaton solu-
tion, and, in the charged case, with the Chan–Mann charged dilaton solution.
In these coordinates, a new stationary dilaton solution is obtained; it does not
allow for a de Sitter–Anti-de Sitter limit at spatial infinity, where its structural
functions increase indefinitely. On the other hand, it is horizonless and allows for
a naked singularity at the origin of coordinates; moreover, one can identify at a
large radial coordinate a (quasi-local) mass parameter and in the whole space a
constant angular momentum.
Via a general SL(2, R) transformation, applied on the static cyclic symmetric
metric, a family of stationary (charged) dilaton solutions has been generated. A
particular SL(2, R) transformation is identified, which gives rise to the rotat-
ing Chan–Mann (charged) dilaton solution. All the exhibited solutions have
been characterized by their quasi-local energy, mass, and momentum through
their series expansions at spatial infinity. The algebraic structure of the Ricci–
energy-momentum, and Cotton tensors is given explicitly. A summary of results
is presented in section 13.9.

13.1 Scalar Field Minimally Coupled to Einstein Gravity

The literature on stationary rotating scalar field solutions is rather scarce among
them Chan (1997), Chan and Mann (1996); most of the known solutions of this
class are static. The static cyclic symmetric metric can always be described by
three structural functions appearing in the metric

ds2 = −A(r)2 dt2 + B(r)2 dr2 + C(r)2 dt2 ,

258 Dilaton Field Minimally Coupled to (2 + 1) Gravity

leaving still a freedom in the choice of the r–coordinate, which can be used to
fix the metric structure, for instance:
the Schwarzschild coordinate frame: ds2 = −A(r)2 dt2 + B(r)2 dr2 + r2 dt2 , or
the gtt = −1/grr coordinate frame: ds2 = −F (r)2 dt2 + dr2 /F (r)2 + H(r)2 dt2 . In
one of the well-known works on dilaton Chan and Mann (1994), this last metric
was used to argue that its use simplifies the calculations; this is partially true
when it is also assumed that H(r) = rN/2 ; nevertheless, the field equations to be
integrated are of the second order. The r-gauge freedom allows one to fix only one
structural function and leave two structural metric functions undetermined in
the class of static circularly symmetric metric in (2+1) gravity; the Einstein field
equations yield further constraints fixing the dependence on r of the remaining
structural functions.
In the Schwarzschild coordinate frame, gθθ = r2 , the equations to be inte-
grated, for a dilaton logarithmically depending on r and associated to an
exponential potential, are first-order equations for each of the (different) struc-
tural functions. Thus, one can claim on its generality and uniqueness under the
restrictions imposed on the dilaton field. It should be pointed out that in the
Chan–Mann papers the solutions were derived under the ansatz gθθ = γ 2 rN . In
this section, the integration of the field equations is done for the static cyclic
symmetric metric in the r-gauge fixed according to gtt = −1/grr = F (r), and
gθθ = H(r); it happens that the resulting two possible solutions can be identified,
via parameter choice, with the static Chan and Mann (1994) solution.
On the other hand, dealing with stationary cyclic symmetric metrics for
logarithmically dependent dilaton field, there exists a single solution for the
Schwarzschild coordinate gθθ = r2 –gauge.
The action to be considered in this work dealing with (2+1)-dimensional
gravity is given by

√
S = d3 x −g R − 4 ∇μ Ψ ∇μ Ψ + 2 ebΨ Λ , (13.1)

where Ψ is the massless minimally coupled scalar field, and R is the scalar
curvature. The variations of this action yield the dynamical equations
EQμν := Rμν = 4 ∇μ Ψ ∇ν Ψ − 2gμν e b Ψ Λ,
b
∇μ ∇μ Ψ + Λ e b Ψ = 0, (13.2)
4
where at this stage Λ and b are arbitrary parameters. Details about the derivation
of the Einstein–dilaton equations and analysis of the known solutions can be
found in Garcia–Diaz and Gutierrez–Cano (2014b).

13.2 Static Black Hole Coupled to a Scalar Ψ(r) = k ln(r)

The static cyclic symmetric metric in the (2 + 1) Schwarzschild coordinate frame
is given by
 13.2 Static Black Hole Coupled to a Scalar Ψ(r) = k ln(r) 259

dr2
g = −N (r)2 dt2 + + r2 dφ2 . (13.3)
L(r)2
The scalar field equation for the dilaton Ψ(r) = k ln(r) becomes
L2 d N L dL 1
F EQ := + + bΛ rbk = 0. (13.4)
r N dr r dr 4k
The simplest Einstein equation is EQ3 3
L2 d N L dL
EQ3 3 = − − + 2 Λ rbk = 0, (13.5)
N r dr r dr
thus, from EQ3 3 + F EQ, one gets the constant relation

rbk Λ (b + 8 k) = 0 → b = −8k. (13.6)

On the other hand, equation EQ1 1 − EQ2 2 − EQ3 3 gives a first-order equation
for L2 , namely
d 2 L2 2
L + 4 k2 − 2 Λr1−8 k = 0 (13.7)
dr r
with integral
Λ
L(r)2 = r−4 k C1 + r2−8 k
2 2

1 − 2 k2
 
Λ
4 k2
r r−8 k .
2
2
= r C1 + (13.8)
1−2 k 2

The remaining Einstein equation arises from −(EQ1 1 − EQ2 2 + EQ3 3 ) r/(2 L2 ),
and yields
1 dN Λ 2 k2
− 2 r(1−8 k ) − 2 = 0. (13.9)
N dr L r
Finally, substituting L(r)2 from (13.8) into the above equation (13.9) for N (r),
and integrating one obtains
 
2 4 k2 Λ 2
2
N (r) = CN r C1 + r , N (r)2 = CN 2 r8 k L(r)2 . (13.10)
2
1−2 k 2

2 2
2 N , −EQ1
d 1 L d
The remaining equation with second derivatives of dr = N r dr N +
2 2

N dr dr + N dr 2 −2 Λ r
L dL dN L d N bk
= 0, is fulfilled for the determined structural functions.
Thus, the general static solution can be given as
2 dr2
g = −CN 2 r8 k L(r)2 dt2 + + r2 dφ2 ,
L(r)2
 
Λ
4 k2
r r−8 k ,
2
2 2
L(r) = r C1 +
1 − 2 k2
Ψ(r) = k ln (r). (13.11)
260 Dilaton Field Minimally Coupled to (2 + 1) Gravity

endowed with three significant parameters: mass −C1 , cosmological constant

Λ → ± l12 , and the dilaton parameter k. The constant CN can be absorbed by
scaling the coordinate t.
The constant Λ can be equated to minus the standard Λs = ± l12 ; indeed, by
setting in (13.11)

1 2
α , r → rα1/(4 k ) , φ → φ α−1/(4 k ) ,
2 2
Λ=± 2
l
C1 → C1 α1+1/(2 k ) , CN → CN α−(1+1/(4 k )) ,
2 2
(13.12)

one arrives at the metric (13.11) with structural functions with Λ = ± l12 .
Notice that the Λ used by in Chan–Mann, when considered as a cosmological
constant, differs from the standard cosmological constant Λs = ± l12 = −Λ, where
+ and − stand correspondingly for de Sitter and anti-de Sitter (AdS).
If Λ (1 − 2 k 2 ) > 0 then one has

a) dS horizonless: Λ < 0 ∧ {k < − √12 , k > √12 },C1 > 0,

b) dS cosmological singularity: Λ < 0 ∧ {k < − √12 , k > √1 },C1
2
< 0,
c) AdS horizonless: Λ > 0 ∧ {− √12 < k < √12 },C1 > 0,
d) AdS black hole: Λ > 0 ∧ {− √12 < k < √12 },C1 < 0,

If Λ (1 − 2 k 2 ) < 0 then one has

e) dS cosmological singularity: Λ < 0 ∧ {− √12 < k < √12 },C1 > 0,

f) AdS event horizon: Λ > 0 ∧ {k < − √12 , k > √12 },C1 > 0.

Notice that for vanishing dilaton parameter k = 0, the derived solution reduces
to the BTZ one for C1 = −M , CN = 1, and Λ = 1/l2 , on the other hand, for
opposite in sign constants one gets the cosmological dS solution.
Accomplishing in the general above solution (13.11) the transformations and
constant parameterizations:

1 2−N
r→r N/2
, φ → β θ, k = ± , CN = 2β/N, 2 β 2 C1 = −M N,
2 N

one gets the Chan–Mann static solution: Chan (1997); Chan and Mann (1996)

dr2
ds2 = −U (r)dt2 + β 2 + β 2 rN dθ2 ,
U (r)
2M 1− N 8Λβ 2
U (r) = β 2 F (r) = − r 2 + rN ,
N (3 N − 2)N
1
Ψ(r) = ± N (2 − N ) ln (r), 0 < N < 2, N = 2/3. (13.13)
4
 13.2 Static Black Hole Coupled to a Scalar Ψ(r) = k ln(r) 261

13.2.1 Quasi Local Momentum, Energy, and Mass

To characterize non-asymptotically flat solutions one uses the Brown–York for-
malism, see Brown and York Jr. (1993); Brown et al. (1994), of quasi–local
momentum, energy, and mass quantities. The case to be treated in detail corre-
sponds to the black hole solution with parameters Λ > 0, C1 = −M, 1 − 2k 2 =
B 2 , 0 < B < 1, CN = 1. Evaluating the quasi–local characteristics as functions
of r, one obtains for the energy and mass
 √
2 B 2 −2 M B2 Λ
(r) = −r 1 − 2 B2 − 0 ,
r Λ πB
 √
M B2 Λ
E(r) = −2 r2 B −1 1 − 2 B 2
2
− 2πr0 ,
r Λ B
2 √ 
r2B Λ M B2
M (r) = 2 M − 2 2 Λ − 2 r 1 − 2B 2 π 0 . (13.14)
B B r Λ
The series expansion of energy and mass functions independent of 0 behave at
infinity, r → ∞, as √
BM Λ 2B 2
(r → ∞, 0 = 0) ≈ √ − 2 r ,
2 Λr π r Bπ
2
√
BM Λ 2B 2
E(r → ∞, 0 = 0) ≈ √ − 2 r ,
r Λ rB
Λ 2
M (r → ∞, 0 = 0) = 2 M − 2 2 r2B . (13.15)
B
The reference energy density to be used in this evaluation is the one correspond-
ing to the anti-de Sitter metric with parameter M0 ,
 √
1 r2 1 l M0 Λ M0
0 (M0 ) = − − M0 , 0|∞ (M0 ) ≈ − + =− + √ .
π r l2 π l 2π r2 π 2 Λr2 π
The series expansions of the corresponding quantities at r → ∞ result in
√ 2 B 2 −2 √
B M − M0 Λr Λ
(r → ∞, 0|∞ (M0 )) ≈ √ − + ,
2π Λr 2 Bπ π
 
r2 B −1
2
√ B M − M0
E(r → ∞, 0|∞ (M0 )) ≈ −2 + 2r Λ+ √ ,
B r Λ
2
Λ r2 B Λ r2 M0
M (r → ∞, 0|∞ (M0 )) ≈ −2 + 2 − + 2M
B2 B B
2
−B M r(2−2 B ) . (13.16)
For B = 1, which corresponds to vanishing dilaton parameter k, one arrives at
the static BTZ metric, and at its quasi local energy and mass; compere with
the energy–mass characteristics of the BTZ solution. Therefore, one concludes
that role of the mass parameter is played by M . At spatial infinity all physical
quantities are infinite; in particular, the field Ψ increases logarithmically.
262 Dilaton Field Minimally Coupled to (2 + 1) Gravity

13.2.2 Classification of the Energy–Momentum and Cotton Tensors

The Ricci tensor, determining in (2+1) dimensions the Riemann curvature ten-
sor, or equivalently, for this class of dilatons, the energy–momentum tensor
matrix (T μ ν ) is described by
⎡ ⎤
−2 Λ r−8 k
2
0 0
⎢ ⎥
⎢ (4 k2 −1) ⎥
⎢ 4 k 2 r−2−4 k C1 − 2r−8 k Λ (2 k2 −1) ⎥,
2 2
0 0
⎣ ⎦
−2 Λ r−8 k
2
0 0
and is algebraically characterized by the following eigenvectors:
2 2
λ1,2 = −2 Λ r−8 k : V1, 2 = [V 1 , 0, V 3 ], Vμ V μ = −V 1 N 2 + V 3 r2 ,
2

δV1N 2  
V3 = , Vμ V μ = (V 1 ) N 2 δ 2 − 1 ;
r
δ 2 > 1, V1 = S1, δ = 1, V1 = N1, δ 2 < 1, V1 = T1
 2 
2 −2−4 k2 −8 k2 4 k − 1
λ3 = 4 k r C1 − 2Λ r = T 2 2 : V3 = [0, V 2 , 0],
(2 k 2 − 1)
2
V2
Vμ V μ = , V3 = S3. (13.17)
L2
Hence, depending on the sign of the norm, one will have spacelike, δ 2 > 1, null,
δ 2 = 1, or timelike, δ 2 < 1, eigenvectors: V1, 2 = {S, N, T}. Therefore, in the
case of a double root one may choose different vector components determining,
for instance, one spacelike vector S and the other timelike T or null N one.
Therefore, one may have the algebraic Ricci types: {S, 2S}, {S, 2N }, {S, 2T } and
{S, (S, T )}, {S, (T, T )}, . . . , {S, (T, N )}.
The matrix of the Cotton tensor is given by
⎡ ⎤
0 0 C 13
⎢ ⎥
(C μ ν ) = ⎢ ⎣ 0 0 0 ⎥
⎦,
C 31 0 0
C1 2  
k 1 − 2 k 2 r−2−8 k ,
2
C 13 =
CN
 
C 31 = −C1 k 2 1 − 2 k 2 r−4 L(r)2 ,
C1 2 4  2
k 1 − 2 k 2 r−6−8 k L(r)2 < 0.
2
C 13C 31 = − (13.18)
CN
The characteristic equation allows for one real and a pure imaginary eigenvalues.
Therefore, the Cotton tensor is characterized algebraically by
V2
λ1 = 0 : V1 = [0, V 2 , 0], Vμ V μ = , V1 = S1,
L2
λ2 = λ̄3 = C 13C 31 :
 13.3 General Static Chan–Mann Solution 263
√
C 31
V2 = Z = [V , 0, V = √ 1 V 1 ], V3 = Z̄.
1 3
(13.19)
C 3
Thus, the algebraic type of the Cotton tensor is Type I : {S, Z, Z̄}.

13.3 General Static Chan–Mann Solution

With the purpose of establishing the uniqueness of the Chan–Mann solution,
its derivation is accomplished in a coordinate frame such that the static cyclic
symmetric line element is given by

dr2
g = −F (r)dt2 + + H(r)dφ2 .
F (r)
The Einstein-field equations (13.2) for this metric are
   2 
F d d 1 d
EQtt = −2 Λ ebΨ F + H F+ F F ,
4 H dr dr 2 dr2
 d 2 d2 d2
Λ ebΨ 1 dr H 1 dr 2H 1 dr 2F
EQrr = 2 + − −
F 4 (H)2 2 H 2 F
d  d  2
1 dr F dr H d
− −4 Ψ ,
4 HF dr
   d 2  2 
1 d d 1 F dr H 1 d
EQφφ = 2 Λ e H −
bΨ
F H+ − H F
2 dr dr 4 H 2 dr2
(13.20)

while the dynamical equation for the scalar Ψ amounts to

   2  d  d
d d d 1 F dr H dr Ψ 1
EQF = F Ψ+ Ψ F+ + bΛ ebΨ . (13.21)
dr dr dr2 2 H 4
The combination EQtt /F (r) + EQrr F (r) yields
 2  2
d2 H dH dΨ
2H − + 16 H 2 = 0, (13.22)
dr2 dr dr
√
which, introducing the new function Z = H, and substituting Ψ = k ln(r),
gives rise to an Euler equation
d2 k 2 Z (r)
2
Z (r) + 4 =0 (13.23)
dr r2
√
with solutions of the form rα , α2 − α + 4 k 2 = 0, α± = 1/2 ± 1/2 1 − 16 k 2 .
Therefore
2
H(r) = (Z+ rα+ + Z− rα− ) , Z± = const. (13.24)
264 Dilaton Field Minimally Coupled to (2 + 1) Gravity

The field
√ equation (13.21), introducing a new function Y (r) through F =
Y (r)/ H and replacing Ψ = kln(r), becomes
d Y (r) √
4k ( ) + bΛ rb k H = 0, (13.25)
dr r
with first integral
 √
4 k Y (r) = −r bΛ rb k Hdr + C0 r, (13.26)

and using here H(r) from (13.24), one gets

√
2
bΛ Z+ rbk+5/2+1/2 1−16 k
Y (r) = C0 r −  √ 
2k 2 bk + 3 + 1 − 16 k 2
√
2
bΛ Z− rbk+5/2−1/2 1−16 k
−  √ . (13.27)
2k 2 bk + 3 − 1 − 16 k 2
Consequently, the structural function F (r) is
Y (r)
F (r) = √ √ . (13.28)
Z+ r1/2+1/2 1−16 k2 + Z− r1/2−1/2 1−16 k2

The remaining equations to be satisfied√ impose some constraints on Z± , and

1+ 1−16 k2
on the values of b, namely b− := − for Z+ = 0, Z− = 0 and b+ :=
√ k
−1+ 1−16 k2
k for Z − = 0, Z+ 
= 0. Correspondingly, one gets the solutions in the
form
dr2
g∓ = −dt2 F (r)∓ + + dφ2 H(r)∓ ,
F (r)∓
√
C0 1 ± 1 √1−16 k2
2
r1∓ 1−16 k Λ
F (r)∓ = r2 2 +8 √ ,
Z∓ 4 ∓ 4 1 − 16 k 2 − 48 k 2
√
1−16 k2
H(r)∓ = Z∓ 2 r1∓ , Ψ(r) = k ln(r). (13.29)

13.3.1 Regular F (r)+ Function for the Metric g+

The function F (r)+ would be regular (with respect to the set of parameters)
for certain values of the dilaton parameter k: if √
k is restricted to the range
0 < k < 1/4, in F+ the denominator D+ := 4 + 4 1 − 16 k 2 − 48 k 2 does not
approach zero in this whole range of k, D+ > 0, therefore F (r)+ is regular for
0 < k < 1/4. At k = 0 the metric is well-behaved and recognizable as the static
BTZ metric with parameters C0 = −M , Λ = 1/l2 , and Z+ = 1. √ On the other
hand, introducing a new constant ν related to the dilaton, ν = 1 − 16 k 2 , 0 <
ν < 1 , one rewrites the structural functions as
 13.3 General Static Chan–Mann Solution 265

C0 (1− ν)/2 r1+ν Λ

F (r)+ = r +8 ,
Z+ (ν + 1)(3ν + 1)
 (3ν+1)/2
8 Λ Z+
rh = − ,
C0 (3ν + 1)(ν + 1)
√
1 − ν2
2 1+ν
H(r)+ = Z+ r , Ψ(r) = ± ln(r), 0 < ν < 1. (13.30)
4
It is apparent that these structural functions for ν = 1 yield the functions of
the static BTZ metric, under minor arrangements. Other interesting solution
contained in this metric representation is the one determined by ν = 0, namely
 1/2
C0 1/2 8 Λ Z+
F (r)+ = r + 8 Λ r, rh = − ,
Z+ C0
1
H(r)+ = Z+ 2 r, Ψ(r) = ± ln(r), (13.31)
4
which is identifiable with the string solution.
On the other hand, if one were adopting the choice
1
ν =1+ 1 − 16 k 2 → k = ν(2 − ν), 1 < ν < 2, (13.32)
4
then the structural functions would be
C0 1−ν/2 8Λ
F (r)+ = r + rν , H(r)+ = Z+
2 ν
r ,
Z+ (3 ν − 2)ν
1
Ψ(r) = ± ν(2 − ν) ln (r), 1 < ν < 2. (13.33)
4

13.3.2 Chan–Mann Solution

The function F (r)− exhibits a √
singularity in its dependence on k: the reality of
the radical in the powers of r, 1 − 16 k 2 , imposes
√ to k to ranges 02 < k < 1/4,
√ in F− the denominator D− := 4 − 4 1 − 16 k − 48 k riches 0 at
moreover 2

k = ± 2/6, and at k = 0, therefore depending on the range of k, F (r)− splits

into two branches:
√ √
a) F (r)− for k within − 2/6 < k < 0, and 0 < k < 2/6, has D− < 0,
consequently, for Λ √> 0 and C0 > 0, (Z∓ = 1), the function F (r)− becomes zero
at r = (− C 0 D− (1∓ 1−16 k 2 )/2
8ΛZ∓ ) .
√ √
b) F (r)− for k within −1/4 < k < − 2/6, and 2/6 < k < 1/4, has D− > 0.
One may introduce a new dilaton constant, say N , through

N =1− 1 − 16 k 2 → k = ± N (2 − N )/4,
266 Dilaton Field Minimally Coupled to (2 + 1) Gravity

consequently, this correspondence establishes the following relationship in the

ranges of values of the dilaton constants
√ √
2 2 2 2
0<k< →0<N < , < k < 1/4 → < N < 1, (13.34)
6 3 6 3
with the structural metric functions and dilaton field in the form
C0 1− N 8Λ
F (r) = r 2 + rN , H(r) = Z−2 N
r ,
Z− (3 N − 2)N
1
Ψ(r) = ± N (2 − N ) ln (r). (13.35)
4
Further, by setting C0 = −2M /(N β), Z− = β, and scaling the coordinate
t → βt, and adding the ranges of values of the dilaton constants N and ν, from
(13.34) and (13.32) respectively, one arrives at the solution
dr2
ds2 = −U (r)dt2 + β 2 + β 2 rN dθ2 ,
U (r)
2M 1− N 8Λβ 2
U (r) = β 2 F (r) = − r 2 + rN ,
N (3 N − 2)N
1
Ψ(r) = ± N (2 − N ) ln (r), 0 < N < 2, N = 2/3,
4
known as the Chan–Mann black hole solution for minimally coupled scalar field;
see Chan (1997); Chan and Mann (1996). In this way, the uniqueness of the solu-
tion for a dilaton of the form Ψ(r) = k ln (r) coupled to a exponential potential
exp(b Ψ) has been established avoiding guesses about a special structure of the
metric. It describes a black hole with horizon at
  2
M (3N − 2) 3N −2
rh = ,
4β 2 Λ
for N = 2/3, 0 < N < 2, M > 0 and Λ > 0.

13.4 Stationary Solution Coupled to Ψ(r) = k ln(r)

The (2 + 1)-dimensional stationary cyclic symmetric metric in the Schwarzschild
coordinate frame {t, r, φ} can be given as
dr2
g = −N (r)2 dt2 + + r2 (dφ + W (r)dt)2 . (13.36)
L(r)2
In this frame the scalar field equation becomes
1 dN 1 d L bΛ rbk+1
+ + = 0. (13.37)
N dr L dr 4 k L2
The simplest Einstein equation is EQ3 1 :
d2 W
3 1 dL 1 dN
dr 2
dW
=− − + , (13.38)
dr
r L dr N dr
 13.4 Stationary Solution Coupled to Ψ(r) = k ln(r) 267

with first integral

dW N
= W0 3 , (13.39)
dr r L
Replacing dL
dr from (13.37) and
dW
dr from (13.39) into the Einstein equations,
from EQ3 3 one gets
bΛ rbk W0 2
− 2 4 + 8 Λ rbk = 0, (13.40)
k r
which is fulfilled if
 
4 Λ 2 k2 − 1
b = − = 0, W0 = 2
2
, (13.41)
k k2
thus, for this class the dilaton field ought always to exist, b = 0 and k = 0, there
is no room for a vacuum plus cosmological
 constant
 field. Moreover,
 onehas to
distinguish two branches: Λ > 0 ∧ 2 k 2 − 1 > 0, and Λ < 0 ∧ 2 k 2 − 1 < 0.
Replacing b into the field equation one gets
1 d 1 d Λ
N+ L − 2 3 2 = 0. (13.42)
N dr L dr k r L
The remaining EQ1 1 − EQ2 2 Einstein equation yields
1 d 1 d k2
− N+ L+4 = 0. (13.43)
N dr L dr r
Adding these two last equations, one arrives at a linear first-order equation for L2
d 2 L2 k 2 Λ
L +4 = 2 3, (13.44)
dr r k r
with integral
Λ 1
L(r)2 = r−4 k C1 +
2
. (13.45)
2 k 2 (2 k 2 − 1) r2
Finally, substituting this result for L(r)2 into the first-order linear equa-
tion (13.43) for N (r) and integrating one obtains
2
N (r)2 = CN 2 r8 k L(r)2 . (13.46)
2
d
The remaining Einstein equation with second derivatives of dr 2 N is fulfilled for

this set of functions. On the other hand, the equation for W (13.39) becomes
d
W = W0 CN r4 k −3
2
(13.47)
dr
with general integral
W0 CN
r4 k −2 + W1
2
W =
2(2 k − 1)
2

Λ C
√ N r4 k −2 + W1 .
2
→ W (r) = (13.48)
2 k2 − 1 2 k
268 Dilaton Field Minimally Coupled to (2 + 1) Gravity

The parameter W1 can be equated to zero by a coordinate translation. Thus,

this stationary solution coupled to a dilaton can be given by

2 dr2 2
g = −CN 2 r8 k L(r)2 dt2 + + r2 (dφ + W (r)d t) ,
L(r)2
Λ 1
L2 = r−4 k C1 +
2
,
2 k 2 (2 k 2 − 1) r2

Λ C
√ N r4 k −2 ,
2
W =
2 k − 1 2k
2

Ψ(r) = k ln(r). (13.49)

Recall
 that in the derivation of this solution √one has to fit the condition
Λ 2 k 2 − 1 > 0, together with k = 0, k = 1/ 2. Moreover, having in mind
2
that gtt = −CN 2 C1 r4 k , the signature {−, +, +} imposes some constraints on
the range of values of the constant:
assuming that Λ is related with the standard cosmological constant Λs then
Λs = ± l12 = −Λ, where + and − stand correspondingly for de Sitter and anti-de
Sitter (AdS) cases.
 
If Λ < 0, then 2 k 2 − 1 < 0 with

● dS horizonless: Λ < 0 ∧ {− √12 < k < √12 } ∧ k = 0,C1 > 0,

 
If Λ > 0, then 2 k 2 − 1 > 0 and arises

● AdS horizonless Λ > 0 ∧ {k < − √12 , k > √1

2
},C1 > 0.

Thus the allowed dilaton solutions are horizonless.

13.4.1 Momentum, Energy, and Mass for a Rotating Dilaton

To characterize non-asymptotically flat solutions one uses the Brown–
York formalism of quasi-local momentum, energy, and mass quantities.
The expressions for the momentum density and the quasi-local momentum
are
2 Λ (2 k 2 − 1) 2 Λ (2 k 2 − 1)
j= , J(r) = = W0 . (13.50)
2π k r k
Evaluating the quasi-local characteristics as functions of r one obtains for the
energy and mass
L(r)
(r) = − − 0 , E(r) = −2 L(r) − 2 π r 0 ,
πr
Λ
r4 k −2 − 2 CN r4 k L(r) π 0 .
2 2
M (r) = −2 CN C1 − 2 CN (13.51)
2 k −1
2
 13.4 Stationary Solution Coupled to Ψ(r) = k ln(r) 269

The series expansions of energy and mass functions independent of 0 behave at

infinity r → ∞ as
√ √ √
2kC1 2 k 2 − 1 2Λ
(r → ∞, 0 = 0) ≈ − √  2 4 − √ ,
2π Λ r k 2 π k r 2 k2 − 1
2
   √
C1 k 2 2 k 2 − 1 2Λ
E(r → ∞, 0 = 0) ≈ − 1 + √ ,
Λ r4 k2 −2 r k 2 k2 − 1
Λ
r4 k −2 .
2
M (r → ∞, 0 = 0) = −2 CN C1 − 2 CN (13.52)
2 −1
k2
The constant value in M (r → ∞) one may identify with a mass, namely M =
−C1 CN , for CN = −1. Notice that by scaling the coordinate t, without any loss of
generality, one can set CN = ±1. Nevertheless, as has been established above, the
range that the constant may assume imposes certain constraints, in particular the
positiveness of C1 , the obtained stationary solution cannot have the appropriate
mass M = −C1 , instead one has to recur to CN = −1 to overcome the difficulty.
At spatial infinity all physical quantities are infinite, except for the constant
quasi global momentum J, in particular, the mass function increases as a power
of r, while the dilaton field Ψ logarithmically increases.

13.4.2 Classification of the Energy–Momentum and Cotton Tensors

The Ricci tensor, determining in (2+1) dimensions the Riemann curvature ten-
sor, or equivalently, for this class of dilatons, the energy–momentum tensor is
described by the matrix
⎡ 2Λ ⎤
− r4 0 0
⎢ ⎥
(T μ ν ) = ⎢
⎣ 0
4 k2 L(r)2
r2 − 2Λ
r4 0 ⎥ ⎦, (13.53)

0 0 − 2rΛ
4

and is algebraically characterized by the following eigenvectors:

Λ  2 2
λ1,2 = −2 4
: V1, 2 = [V 1 , 0, V 3 ], Vμ V μ = r2 V 3 + W V 1 − V 1 N 2 ;
r
Vμ V μ > 1, V1 = S1, Vμ V μ = 0, V1 = N1, Vμ V μ < 1, V1 = T1,
Vμ V μ > 1, V2 = S2, Vμ V μ = 0, V2 = N2, Vμ V μ < 1, V2 = T2,
k 2 L2 Λ (V 2 )2
λ3 = 4 2
− 2 4 = T 2 2 : V3 = [0, V 2 , 0], Vμ V μ = , V3 = S3.
r r L2
(13.54)

In the case of a double root, one may choose different vector components
determining, for instance, one spacelike vector S and the other timelike T
or null N one, and other possible combinations. Therefore, one may have
270 Dilaton Field Minimally Coupled to (2 + 1) Gravity

the algebraic Ricci–energy–momentum types: {S, 2S}, {S, 2N }, {S, 2T } and

{S, (S, T )}, {S, (T, T )}, . . . , {S, (T, N )}.
The matrix of the Cotton tensor is given by
⎡ ⎤
C 11 0 C 13
⎢ ⎥
(C μ ν ) = ⎢
⎣ 0 0 0 ⎥,
⎦
C 31 0 −C 1 1
C1 √
2Λk 2 k 2 − 1 r−4 k −4 ,
2
C 1 1 = −C 3 3 =
2
 
k 2 (2 k 2 − 1) C1 Λ
r−4 k −2 ,
2
C 13 =−  2 4 +
CN rk (2 k − 1) r k
2 2 2

 
C 3 1 = k 2 2 k 2 − 1 CN C1 2 r−4 k −4 .
2
(13.55)

The characteristic equation allows for one real and pure imaginary eigenvalues.
Consequently, the Cotton tensor is algebraically characterized by

(V 2 )2
λ1 = 0 : V1 = [0, V 2 , 0], Vμ V μ = , V1 = S1,
L2
 
= i k 2 C1 2 k 2 − 1 r−4 k −3 L(r) :
2
λ2 = λ̄3 = (C 1 1 )2 + C 1 3 C 3 1
V2 = Z = [V 1 , 0, V 3 ], V3 = Z̄ = [V̄ 1 , 0, V̄ 3 ]. (13.56)

Hence, the algebraic type of the Cotton tensor is Type I: {S, Z, Z̄}.

13.5 Stationary Dilaton Solutions Generated via SL(2, R)

Transformations
Subjecting the metric (13.11) to a general SL(2, R) transformation

t = α T + β Φ,
φ = γ T + δ Φ, Δ := αδ − βγ. (13.57)

one arrives at the stationary metric equipped with rotation

   
g = − α2 N (r)2 − γ 2 r2 d T 2 − 2 β α N (r)2 − δ γ r2 dr dΦ
dr2  
+ + δ 2 r2 − β 2 N (r)2 dΦ2 ,
L(r)2
 
Λ
4 k2
r r−8 k , N (r)2 = CN
2
2 2 2
L(r) = r C1 + L(r)2 ,
1−2 k 2

Ψ(r) = k ln (r). (13.58)

 13.5 Stationary Dilaton Solutions Generated 271

or, in the standard representation used to evaluate energy, mass, and momentum,
one has
dr2
g = −N(r)2 dT 2 + + K(r)2 (dΦ + W(r)dT )2 ,
L(r)2
N 2 r2 2
N(r)2 = (α δ − β γ) ,
− β2N 2
δ 2 r2
K(r)2 = δ 2 r2 − β 2 N 2 ,
Λ
L(r)2 = r−4 k C1 + r−8 k +2 ,
2 2

1 − 2 k2
δ γ r2 − β α N 2
W(r) = , β = 0 = δ,
δ 2 r2 − β 2 N 2
 
2 Λ 2 8 k2
N 2 := CN2
r4 k C1 + r 2
= CN r L(r)2 . (13.59)
1−2 k 2

For the AdS (Λ = 1/l2 )–black hole branch, the

√ constants appearing√in the struc-
√
tural functions
√ will be replaced by C N → 1 − 2 k 2 = B, k = 1 − B 2 / 2,
0 < k < 1/ 2, 1 > B > 0, thus

N2B −4+4 B 2
N (r)2 = NB2 = r2 r−2 B C1 B 2 + Λ , L(r)2 = L2B =
2
r . (13.60)
B2
The quasi-local momentum J(r) is constant in the whole spacetime, namely

J(r) = J(r → ∞) = 2 β δ B 3 C1 = W0 = J0 . (13.61)

The quasi-local energy is given by

√
r2 B −1 Λ + C1 r−2 B 2 B 2
2

δ 2 − β 2 Λ − β 2 C1 B 2 r−2 B
2
E(r) = −2
B δ 2 − β 2 Λ − β 2 C1 B 2 r−2 B 2
+β 2 C1 r−2 B B 4 ,
2
(13.62)

while the quasi-local mass amounts to

2Δ Λ + C1 B 2 r−2 B
2

M (r) = −   β 2 C1 B 2 (B 2 − 1)
B δ 2 − Λβ 2 − β 2 C1 B 2 r−2 B 2
2
 β α Λ − δ γ + β α C1 B 2 r−2 B
2

+r2 B (δ 2 − Λβ 2 ) + J0
δ 2 − β 2 Λ − β 2 C1 B 2 r−2 B 2
−2 0 π Δ r2 Λ + C1 B 2 r−2 B 2 . (13.63)

The reference energy density to be used in this evaluation is that which

corresponds to the anti-de Sitter metric with parameter M0 ,
√ √
−M0 + r2 Λ Λ M0
0 (M0 ) = − , 0|∞ (M0 ) ≈ − + 1/2 √ .
πr π Λπ r2
272 Dilaton Field Minimally Coupled to (2 + 1) Gravity

Then the series expansion of the mass at spatial infinity, r → ∞, amounts to

  2
Δ C1 B δ 2 − β 2 Λ + B 2 β 2 Λ Δ r2 B Λ
M (r → ∞) ≈ −2 − 2
δ2 − β 2 Λ B
(β α Λ − δ γ)
+J0 + M (0|∞ ),
δ2 − β 2 Λ
2
M (0|∞ ) = M (r → ∞, 0|∞ (M0 )) ≈ Δ C1 B 2 r2−2 B
+2 Δ Λ r2 − Δ M0 .

When the dilaton field and the rotation vanish, B = 1, β = 0, the generated
solution reduces to the AdS metric and one identifies C1 = −M , or com-
paring with the quasi-local BTZ mass MBT Z (r → ∞) = M − M0 , the same
conclusion is achieved. Frequently, in the literature one encounters the SL(2, R)
transformation
T Φ ω T Φ
t=  −ω  ,φ=−  + , (13.64)
1− ω2
1− ω2 l2 1 − ω2
1− ω2
l2 l2 l2 l2

correspondingly, the structural functions (13.59) assume the form

2 
l − ω 2 r2 N 2
N(r) = − 2 2 2 , L(r) = L(r), N = N (r),
l (ω N − r2 )
   
ω N 2 l2 − r 2 l2 ω 2 N 2 − r 2
W(r) = − 2 2 2 , K(r) = − . (13.65)
l (ω N − r2 ) l2 − ω 2
In this parametrization, the quasi-local momentum becomes
l 2 B 3 C1
J(r) = −2 ω = J0 ,
l2 − ω 2
and vanishes as soon the rotation ω becomes zero.

13.5.1 Sub-Class of Rotating Dilaton Black Holes

Requiring the fulfillment of the relationship
CN 2 Λ
γδ =βα , (13.66)
(1 − 2 k 2 )
one obtains a sub-family of rotating dilaton solutions
 
8 κ k2 2 β 2 CN 2 Λ2 2 2
g = −α CN r
2 2
L − 2r dt2 − 2β α CN 2 C1 r4 k dr dφ
δ (1 − 2 k )
2 2

dr2 2
+ 2 + δ 2 r2 − β 2 r8 k CN 2 L2 dφ2 ,
L 
Λ
r−8 k , Ψ(r) = k ln (r),
2 2
L(r) = r4 k C1 +
2
r 2
(13.67)
1 − 2 k2
 13.5 Stationary Dilaton Solutions Generated 273

in which the term with power r2 in W or equivalently in gT Φ , in the metric

(13.58), cancels out. On the other hand, in the limit of vanishing dilaton param-
eter k = 0, under the coordinate transformation and new parameterizations

β 2 C1 CN 2 + ρ2 2 M + M2 − J0 2 Λ ρ2 − J0 2
r2 = = ,
δ 2 − β 2 CN 2 Λ 4 δ 2 M 2 − J0 2 Λ
δ J0 M+ M2 − J 0 2 Λ
α= ,β= , C1 = − , (13.68)
CN 2 δ CN C1 2 δ2
this sub-branch becomes the rotating BTZ metric
The transformations above (13.68) may be used in the metric (13.67) to obtain
the studied case in terms of the physical BTZ parameters M and J0 and the
dilaton parameter k. Nevertheless, the expressions are quite involved due to the
r → ρ transformation.

13.5.2 Rotating Chan–Mann Dilaton Black Hole

Moreover, the metric (13.67) gives rise to the Chan–Mann rotating dilaton
solution. Straightforwardly by subjecting the static seed metric (13.11) to the
following SL(2, R) transformation together with a transformation of the radial
coordinate
√ √
3 N − 2 N (2 A T + ω Φ)
t= √ , r = ρN/2 ,
2 B 3 N 2 A2 − 2 N A2 − 2 Λ ω 2
√ √
B A N 3 N − 2 Φ + 4 √N √Λ3ωN −2 T
φ= √ , (13.69)
3 N 2 A2 − 2 N A2 − 2 Λ ω 2
where N is the new dilaton parameter. Do not mix up with the structural
function N (r), together with the introduction of new constants through

1 2−N B2
k=± = 1, CN = 2 ,
2 N N
 2 2 
3N A − 2 N A2 − 2 Λ ω 2 N
C1 = ,
4 (3 N − 2) AB 2
(13.70)

one arrives at the Chan and Mann (1996) metric, Eq. (3.1a), namely
 
ρN Λ B 2
g = − Aρ1− N/2 + 8 dT 2
(3 N − 2) N
  −1
ρN Λ ρ1− N/2 Λ ω2
+ 8 + A − 2 dρ2
(3 N − 2) N B2 A (3 N − 2) N
 
ω 2 ρ1− N/2
−ω ρ1− N/2 dT dΦ + ρN B 2 − dΦ2 (13.71)
4A
274 Dilaton Field Minimally Coupled to (2 + 1) Gravity

where r of Eq. (3.1a) is replaced by ρ, and B stands for β of Eq. (3.1a) of Chan
and Mann (1996), to avoid confusion.
It calls one’s attention to the sophisticated modification of the function gρρ ,
which under the SL(2, R) remains invariant and could be used (the original in
relation to the static solution) for the adequate evaluation of the energy and
mass.

13.6 Dilaton Coupled to Einstein–Maxwell Fields

The previous approach to determine Einstein–dilaton solutions in (2 + 1) grav-
ity is extended to the Einstein–Maxwell–dilaton case; thus, the Schwarzschild
coordinate frame is used to determine static cyclic symmetric metrics to (2 + 1)
Einstein equations coupled to an electric Maxwell field and a dilaton logarith-
mically depending on the radial coordinate in the presence of an exponential
potential. The general solution is derived and identified with the Chan–Mann
charged dilaton solution. Via a general SL(2, R) transformation, applied on the
obtained charged dilaton metric, a family of stationary dilaton solutions has
been generated; these solutions possess five parameters: dilaton and cosmolog-
ical constants, charge, momentum, and mass for some values of them. All the
exhibited solutions have been characterized by their quasi-local energy, mass,
and momentum through their series expansions at spatial infinity. The struc-
tural functions determining these solutions increase with the radial coordinate;
hence, they do not exhibit an dS–AdS behavior at infinity. Moreover, the alge-
braic structure of the Maxwell field, energy–momentum, and Cotton tensors is
given explicitly.

13.6.1 Einstein–Maxwell-Scalar Field Equations

The action to be considered in this work dealing (2+1)-dimensional gravity is
given by
  
√ B
S = d3 x −g R − ∇μ Ψ ∇μ Ψ + 2 ebΨ Λ − e−4 a Ψ F 2 , (13.72)
2
where Λ, b are arbitrary at this stage parameters, Ψ is the massless mini-
mally coupled scalar field, R is the scalar curvature, and F 2 = Fμ ν F μ ν the
electromagnetic invariant. The variations of this action yield the dynamical
equations
B  
Rμν = ∇μ Ψ ∇ν Ψ − 2gμν e b Ψ Λ + 2 e−4 a Ψ Fμ α Fν α − gμ ν F 2 ,
2
B μ
∇ ∇μ Ψ + b e b Ψ Λ + 2 a e−4 a Ψ F 2 = 0,
2  
∇μ e−4 a Ψ Fμ ν = 0. (13.73)
 13.7 Static Charged Solution Coupled to Ψ(r) = k ln(r) 275

13.7 Static Charged Solution Coupled to Ψ(r) = k ln(r)

The static cyclic symmetric metric in the (2 + 1) Schwarzschild coordinate frame
is given by
dr2
g = −N (r)2 dt2 + + r2 dφ2 . (13.74)
L(r)2
t r
The electromagnetic field equations for the tensor field Fμν = 2Ftr δ[μ δν] , and
the dilaton Φ(r) = k ln(r) becomes
d Ftr L r−4 a k+1 N
EQF = → Ftr = Q r4 ak−1 . (13.75)
dr N L
The simplest Einstein equation is R11 + R22 L2 N 2 , which yields
1 d 1 d 1 Bk 2
N− L− = 0, (13.76)
N dr L dr 2 r
thus one gets
2
N (r) = CN L (r) rB k /2
. (13.77)
On the other hand, the equation R33 gives a first-order equation for L2 = Y (r),
namely
d 1 Bk 2 Y (r) r4 ak Q2
Y (r) + +2 − 2 Λ rbk+1 = 0 (13.78)
dr 2 r r
integrating one obtains
r4 ak Q2 Λ r2+bk
+ r−1/2 Bk C1 .
2
L(r)2 = Y (r) = −4 2
+ 4 2
(13.79)
Bk + 8 ak 4 + Bk + 2bk
Substituting this expression of Y (r) into the remaining scalar field equation
d 1 Bk 2 Y (r) r4 ak aQ2 bΛ rbk+1
Y (r) + −8 +2 = 0, (13.80)
dr 2 r Bkr Bk
one arrives at relationships between constants, namely
1
a=− , b = −B k. (13.81)
4B k
Therefore, the general charged dilaton static solution can be given as
2 dr2
g = −CN 2 rB k L(r)2 dt2 + + r2 dφ2 ,
L(r)2
 
r2 Λ Q2 4
r−B k , k 2 = ,
2 2
L(r)2 = rB k /2 C1 + 4 + 4
4 − B k2 Bk 2 B
QCN −B k2 /2
δν] , Ftr = QCN r−1/2 Bk −1 = −At,r , At = 2
2
t r
Fμν = 2Ftr δ[μ r ,
B k2
Ψ(r) = k ln (r), (13.82)
endowed with four relevant parameters: in particular, one may identify the mass
M = −C1 , cosmological constant Λ → ± l12 , dilaton parameter k, and the charge
276 Dilaton Field Minimally Coupled to (2 + 1) Gravity

Q. The constant CN can be absorbed by scaling the coordinate t, thus it can be

equated to unit, CN → 1. Moreover, one has to set the charge Q to zero, Q = 0,
when looking for the limiting solutions for vanishing dilaton k = 0, which are just
the dS and AdS solutions with parameters C1 = ±M respectively, and CN = 1.
There is no static electrically charged limit of this solution for a vanishing dilaton
field.
The constant Λ can be equated to minus the standard cosmological constant
Λs = ± l12 ; indeed, by setting in (13.82)
1 2
α , r → rα2/(B k ) , φ → φ α−2/(B k ) , Q → Q α(1+2/(B k )) ,
2 2 2
Λ=±
l2
C1 → C1 α1+4/(B k ) , CN → CN α−(1+2/(B k )) ,
2 2
(13.83)
one arrives at the metric (13.82) with Λ = ± l12 . Notice that the Λ used in
Chan–Mann works, when considered as a cosmological constant, differs from
the standard cosmological constant Λs = ± l12 = −Λ, where + and − stand
correspondingly for de Sitter and anti-de Sitter (AdS).
If B > 0, Λ (4 − B k 2 ) > 0, then one has
√√
a) dS horizonless: Λ < 0 ∧ {k < − √B2 , k > √B },C1 > 0,
2 √ √
b) dS cosmological singularity: Λ < 0 ∧ {k < − √B2 , k > √B2 },C1 < 0,
√ √
c) AdS horizonless: Λ > 0 ∧ {− √B2 < k < √B2 },C1 > 0,
√ √
d) AdS black hole: Λ > 0 ∧ {− √B2 < k < √B2 },C1 < 0.

If B > 0, Λ (4 − B k 2 ) > 0, then one has

√ √
e) dS cosmological singularity: Λ < 0 ∧ {− √B2 < k < √B },C1
2
> 0,
√ √
f) AdS event horizon: Λ > 0 ∧ {k < − √B2 , k > √B },C1
2
> 0.
The norm of the normal vector nμ = ∇μ r to the surface r = r0 = const. is given
by g rr (r0 ) = L(r0 )2 . It becomes a null vector at the horizon rh of the black hole
solution with B > 0, k 2 < 4/B, C1 < 0, and Λ = l12 which is determined as the
outer root of g rr (r) = L(r)2 = 0, namely
2 r2 Λ Q2
rB k /2
C1 + 4 +4 = 0. (13.84)
4−Bk 2 Bk 2
The electromagnetic invariant of this solution amounts to
Q2 C1 2 −2(1+ Bk2 )
Fμν F μν = −2 r .
CN 2
Accomplishing in the general above solution (13.82) the r-transformation and
constant parameterizations:
√ 
2 2−N
r→r N/2
, k = ±√ ,
B N
 13.7 Static Charged Solution Coupled to Ψ(r) = k ln(r) 277

one gets, modulo minor constants arrangements, the Chan and Mann (1994,
1996) static solution:
N2
ds2 = −C2N U (r)dt2 + dr2 + rN dφ2 ,
4 U (r)
2N Λ N
U (r) = C1 r1−N/2 + rN + 4 Q2 ,
(3 N − 2) 2−N
N
δν] , Ftr = QCN rN/2−2 = −At,r
t r
Fμν = 2Ftr δ[μ
2
N QCN N/2−1
→ At = r ,
2−N

N (2 − N )
Ψ(r) = ± ln (r), 0 < N < 2, N = 2/3. (13.85)
2B

13.7.1 Quasi-Local Mass, Momentum, Energy for Charged Dilaton

To characterize non-asymptotically flat solutions one uses the Brown–York for-
malism of quasi-local momentum, energy, and mass quantities; see 1.3.1. The
evaluation of the mass for the studied charged dilaton solution yields
2 2
M (r, 0 ) := −2CN r1/2 Bk L2 − 2π r1+1/2 Bk L 0 ,
2
r(4− Bk )/2 CN Λ
M (r, 0 = 0) = −2 CN C1 − 8
4 − Bk 2
−1/2 Bk2 2
r CN Q
−8 . (13.86)
Bk 2
Comparing with the energy characteristics of the BTZ solution, one concludes
that role of the mass parameter is played by C1 CN = −M . Recall that CN can
be equated to 1. At spatial infinity all physical quasi-local quantities are infinite,
in particular, the field Ψ logarithmically increases.

13.7.2 Algebraic Classification of the Field, Energy–Momentum,

and Cotton Tensors
The Maxwell field tensor is given by
⎡ ⎤
− CNQL2 r−1−1/2 Bk
2
0 0
⎢ ⎥
⎢ ⎥
(F μ ν ) = ⎢ −r−1+1/2 Bk L2 Q CN
2
0 0 ⎥, (13.87)
⎣ ⎦
0 0 0
and is algebraically characterized by the following eigenvectors:
λ1 = r−1−Bk Q : V1,2 = [V 1 , −r1/2 Bk L2 CN V 1 , 0], Vμ V μ = 0, V1 = N1,
2 2

λ2 = −r−1−Bk Q : V2 = [V 1 , r1/2 Bk L2 CN V 1 , 0], Vμ V μ = 0, , V2 = N2,

2 2

2
λ3 = 0 : V3 = [0, 0, V 3 ], Vμ V μ = V 3 r2 , V3 = S3, (13.88)
278 Dilaton Field Minimally Coupled to (2 + 1) Gravity

therefore, its type is {N, N, S}. For this class of charged dilatons, the energy–
momentum tensor is described by
T μ ν = 2Q2 r−2(1+ Bk ) δ μ φ δ φ ν ,
2
(13.89)
and is algebraically characterized by the following eigenvectors:
2
12 2 B k2 V2
λ1,2 = 0 : V1, 2 = [V , V , 0], N := Vμ V
1 2 μ
= −V CN r 2
L(r) + 2 ,
L
V1, 2 = T1, 2, N1, 2, S1, 2,
2
λ3 = 2 Q2 r−2−2 Bk : V3 = [0, 0, V 3 ], Vμ V μ = V 3 r2 , V3 = S3.
2
(13.90)
Hence depending on the sign of the norm N one will have spacelike, N > 0, null,
N = 0, or timelike, N < 0, eigenvectors: V1, 2 = {S, N, T}. Therefore, in the
case of a double root one may choose different vector components determining,
for instance, one spacelike vector S and the other timelike T or null N one.
Therefore, for the Maxwell energy tensor one may have the algebraic types:
{2S, S}, {2N, S}, {2T, S} and {(S, T ), S},{(T, T ), S},. . .
The Cotton tensor amounts to
C μ ν = C 1 3δμ tδφ ν + C 3 1δμ φδtν ,
1    
B C1 k 2 4 − Bk 2 r−Bk + 16 Q2 2 + Bk 2 r−3/2 Bk
2 2
C 13 = 2
32 CN r
CN  
r−Bk /2 C1 2 B 2 k 4 4 − Bk 2
2
C 31 = −
32 r4 Bk 2
 
+12 r−Bk Q2 C1 Bk 2 4 + Bk 2
2

 
+4 B 2 r2−Bk Λ C1 k 4 + 64 r−3 Bk /2 Q4 2 + Bk 2
2 2

 2

2 2 + Bk
+64 Br2−3 Bk /2 Λ Q 2 k 2 . (13.91)
(4 − Bk 2 )

Depending on the signs of C1 and Λ = ±1/l2 the components of C 1 3 and C 3 1 may

be positive or negative; consequently their product could be positive. Therefore,
the Cotton tensor, for both C 1 3 and C 3 1 , positive or negative, is characterized
algebraically by
2
V2
λ1 = 0 : V1 = [0, V 2 , 0], Vμ V μ = 2 , V1 = S1,
L
 3 1/2
1 3 1 1 C 1
λ2 = C 3 C 1 : V2 = [V , 0, V ],
C 13
2
N = Vμ V μ = −V 1 (N 2 C 1 3 − r2 C 3 1 )/C 1 3 ,
V2 = T2, N2, S2,
 1/2
C 31
λ2 = − C1 3 C3 1 : V3 = [V , 0, −V
1 1
],
C 13
2
Vμ V μ = −V 1 (N 2 C 1 3 − r2 C 3 1 )/C 1 3 ,
V3 = T3, N3, S3. (13.92)
 13.8 Stationary Charged Dilaton Generated via SL(2, R) 279

Depending on the sign of the norm N one has spacelike, N > 0, null, N = 0,
or timelike, N < 0, eigenvectors, denoted by: V1, 2 = {S, N, T}. There-
fore, when the eigenvalues are real, the Cotton tensor allows for types I :
{2S, S},{2N, S},{2T, S} and {(S, T ), S}, {(T, T ), S},. . .
In the spacetime region where C 1 3 and C 3 1 are of opposite signs, one of the
eigenvalues becomes imaginary and the following scheme arises
V2
λ1 = 0 : V1 = [0, V, 0], Vμ V μ = , V1 = S1,
L2 √
C 31
λ2 = λ̄3 = C 13C 31 : V2 = Z = [V , 0, √ 1 V 1 ], V3 = Z̄,
1
(13.93)
C 3
thus, the algebraic type of the Cotton tensor is {S, Z, Z̄}.

13.8 Stationary Charged Dilaton Generated via SL(2, R)

Subjecting the metric (13.82) to a general SL(2, R) transformation

t = α T + β Φ,
φ = γ T + δ Φ, Δ := αδ − βγ,

one arrives at the stationary metric equipped with rotation

2 dr2
g = − α2 CN 2 rBk L2 − γ 2 r2 d T 2 +
L2
2 2
−2 β α CN 2 rBk L2 − δ γ r2 d T d Φ + δ 2 r2 − β 2 CN 2 rBk L2 dΦ2 ,
 
r2 Λ Q2
r−B k .
2 2
L(r)2 = rB k /2 C1 + 4 + 4 (13.94)
4−Bk 2 Bk 2

Using the standard notation, this charged rotating dilaton solution is given as

dr2
g = −N(r)2 d T 2 + + K(r)2 (dΦ + W(r)d T )2 ,
L(r)2
N 2 r2 2
N(r)2 = (α δ − β γ) ,
− β2N 2
δ 2 r2
 
2 r2 Λ Q2 2 B k2
N 2 := CN2
rB k /2 C1 + 4 + 4 = CN r L(r)2 ,
4 − B k2 Bk 2
 
r2 Λ Q2
B k2 /2
r−B k ,
2
L(r) = r
2
C1 + 4 +4
4 − B k2 Bk 2
K(r)2 = δ 2 r2 − β 2 N 2 ,
δ γ r2 − β α N 2
W(r) = , β = 0 = δ,
δ 2 r2 − β 2 N 2
Ψ = k ln(r),
T r Φ r
Fμν = 2FT r δ[μ δν] + 2FΦ r δ[μ δν] ,
280 Dilaton Field Minimally Coupled to (2 + 1) Gravity
 
d QCN −B k2 /2
FT r = −AT,r = α Ftr = −α 2 r ,
dr B k2
Ftr = QCN r−1/2 Bk −1
2
,
 
d QCN −B k2 /2
FΦ r = −AΦ,r = β Ftr = −β 2 r ,
dr B k2
QCN −B k2 /2
At = 2 r . (13.95)
B k2
It is worth noticing that the structure of the electromagnetic energy tensor
amounts to
⎡ 2 ⎤
γ 0 γδ
2 ⎢ ⎥
(Tμν ) = 2 Q2 r−2 Bk ⎢⎣ 0 0 0 ⎦.
⎥ (13.96)
γδ 0 δ2
For the AdS (Λ = 1/l2 )–black hole branch, the
√ constants appearing√in the struc-
√
tural functions
√ will be replaced by C N → 1 − 2 k 2 = B, k = 1 − B 2 / 2,
0 < k < 1/ 2, 1 > B > 0, thus
N2B −4+4 B 2
N (r)2 = NB2 = r2 r−2 B C1 B 2 + Λ , L(r)2 = L2B =
2
r . (13.97)
B2

13.8.1 Quasi-Local Mass and Momentum

The evaluation of the quasi-local momentum J(r) yields
1   δ CN β Q2 − Bk2 /2
J(r) = β δ CN C1 4 − Bk 2 + 8 r , (13.98)
2 Bk 2
hence, for positive B > 0, the contribution of the electromagnetic field Q = 0 to
the momentum, at spatial infinity, disappears and one has
1  
J(r → ∞) = β δ CN C1 4 − Bk 2 =: J0 . (13.99)
2
The evaluation of the quasi local mass yields
1 CN Δ ML 1 β δ C N MJ
M (r) = − ,
2 r Bk (Bk − 4) D
3 2 2 2 B 2 k4 r2 D
  
ML := −4 Bk 2 C1 r3 Bk 2 − 4 P + Λ β 2 CN 2 Bk 2
 2 2
+4 rQ2 β 2 CN 2 C1 Bk 2 − 4 Bk 2 + 16 Bk 2 r5−1/2 Bk Λ P
2  2 2  
+r1+1/2 Bk C1 2 β 2 CN 2 B 2 k 4 Bk 2 − 4 − 16 Q2 r3−1/2 Bk Bk 2 − 4 P,
 
MJ = 16 Bk 2 Q2 r−Bk /2 − B 2 C1 k 4 δ γ Bk 2 − 4
2

 
+4 Λ CN 2 α β Bk 2 − 4 r2
  2 2
−64β α r−1/2 Bk Q4 CN 2 Bk 2 −4 +β αB 2 k 4 C1 2 CN 2 r1/2 Bk Bk 2 −4
2
 13.8 Stationary Charged Dilaton Generated via SL(2, R) 281
  
+4 β α Bk2 C1 CN 2 Q2 Bk 2 − 4 Bk 2 − 8 ,
 
β 2 CN 2 Q2 Bk 2 − 4  
− β 2 CN 2 r( Bk −4)/2 C1 Bk 2 − 4 ,
2
D = P −4
Br2 k 2
(13.100)

where P = δ 2 Bk 2 − 4 δ 2 + 4 β 2 CN 2 Λ. The order zero in the series expansion of

M (r) is given by

Λ β 2 CN 2 Bk 2 + δ 2 Bk 2 − 4 δ 2
M0 = 2 M Δ
δ 2 Bk 2 − 4 δ 2 + 4 β 2 CN 2 Λ
Bk 2 δ γ + 4 β α CN 2 Λ − 4 δ γ
−J0 2 2 ,
δ Bk − 4 δ 2 + 4 β 2 CN 2 Λ
(13.101)

where CN C1 has been replaced by −M . Comparing with the quasi-local BTZ

mass MBT Z = 2 M − 2 r2 /l2 ; when the rotation vanishes, β = 0, Δ = αδ → 1,
implies that M0 → −CN C1 = M . It should be pointed out that, due to the
presence of terms with positive powers of r in the series expansion of M(r), it
increases as r → ∞, like the MBT Z does.
Frequently, in the literature one encounters the SL(2, R) transformation
T Φ ω T Φ
t=  −ω  ,φ=−  + ,
1− ω2
1− ω2 l2 1 − ω2
1− ω2
l2 l2 l2 l2

correspondingly, the structural functions (13.95) assume the form

2 
l − ω 2 r2 N 2
N(r) = 2 2
2
, L(r) = L(r), N = N,
l (r − ω 2 N 2 )
 22   
ω N l − r2 l2 r 2 − ω 2 N 2
W(r) = 2 2 , K(r) =
2
. (13.102)
l (r − ω 2 N 2 ) l2 − ω 2
In this parametrization, the quasi-local momentum becomes
1 ω l2
J0 = − C1 CN (4 − B k 2 ),
2 l2 − ω 2
and vanishes as soon the rotation ω becomes zero.
The generated stationary metric (13.59) gives rise, among others, to a charged
generalization of Chan–Mann rotating dilaton solution; see below, Section 13.8.3.
With this result we are giving an answer to the remark contained in the Con-
clusions of Chan and Mann (1996): “Although the static charged black solutions
of (1.1) exist, Chan and Mann (1994), at present we are unable to generalize our
spinning solution to charged cases. This endeavor is complicated by the fact that
when one adds Maxwell fields to a spinning solution, both electric and magnetic
fields must be present. . . ”
282 Dilaton Field Minimally Coupled to (2 + 1) Gravity

13.8.2 Algebraic Classification of the Field, Energy–Momentum,

and Cotton Tensors
The Maxwell electromagnetic field tensor matrix (F μ ν ) amounts to
⎡ −1−3/2 Bk2
⎤
0 − δ r ΔCN L2 Q 0
⎢ ⎥
⎢ ⎥
⎢ −α L2 Q r−1−1/2 Bk CN
2
0 −β L2 Q r−1−1/2 Bk CN
2
⎥,
⎢ ⎥
⎣ 2
⎦
γ r −1−3/2 Bk Q
0 ΔCN L2 0

and has the following eigenvectors

2
βV3 (V3 ) r2 Δ2
λ1 = 0 : V1 = [− , 0, V 3 ], Vμ V μ = ,
α α2
δV2 γV2
λ2 = Q r−1−Bk : V2 = [−r− Bk /2 , V 2 , r− Bk /2
2 2 2

2
],
Δ L CN Δ L2 CN
Vμ V μ = 0, V2 = N2,
δV2 − Bk2 /2 γ V
2
λ3 = −Q r−1−Bk : V3 = [r− Bk /2
2 2
, V 2
, −r ],
Δ L2 CN Δ L2 CN
μ
Vμ V = 0, V3 = N3. (13.103)

Correspondingly, its type is {S, N, N }.

The electromagnetic energy–momentum tensor is given by
⎡ ⎤
βγ 0 βδ
Q2 r−(2+2 Bk ) ⎢ ⎥
2

(T μ ν ) = −2 ⎢ 0 0 ⎥
Δ ⎣ 0 ⎦, (13.104)
−α γ 0 −α δ
and is algebraically characterized by:

λ1,2 = 0 : V1, 2 = [V 1 , V 2 , −V 1 γ/δ],

2 2
CN 2 L2 rBk Δ2 1 2 V 2
N = Vμ V = −
μ
V + 2 ,
δ2 L
N > 0, V1, 2 = S1, 2, N = 0, V1, 2 = N1, 2, N < 0, V1, 2 = T1,
βV3
λ3 = 2 Q2 r−(2+2 Bk ) : V3 = [−
2
, 0, V3 ],
α
(V3 )2 r2 Δ2
Vμ V μ = , V3 = S3. (13.105)
α2
In the case of a double root, depending on the sign of the norm Vμ V μ ,
one will have spacelike, N > 0, null, N = 0, or timelike, N < 0, eigen-
vectors: V1, 2 = {S1, 2, N1, 2, T1, 2}. Therefore, one may choose different
vector components determining, for instance, one spacelike vector S and the
other timelike T or null N one. Therefore, one may have the algebraic Ricci
types: {2S, S}, {2N, S}, {2T, S} and {(S, T ), S}, {(T, T ), S}, . . . , {(T, N ), S},
 13.8 Stationary Charged Dilaton Generated via SL(2, R) 283

where parenthesis is used to stand out the multiplicity of the root under con-
sideration. The matrix of the Cotton tensor for the rotating charged solution is
given by
⎡ ⎤
−α β C 3 1 +γ δ C 1 3 −β 2 C 3 1 +δ 2 C 1 3
0
⎢ Δ Δ
⎥
⎢ ⎥
(C μ ν ) = ⎢ 0 0 0 ⎥, (13.106)
⎣ ⎦
− −α C 1Δ+γ C 3 0 − −α β C Δ
2 3 2 1 3 1
1 +γ δ C 3

where
1  
B C1 k 2 4 − Bk 2 r−Bk
2
C 13 =
32 CN r2
 
+16 Q2 2 + Bk 2 r−3/2 Bk ,
2


CN  
r−Bk /2 C1 2 B 2 k 4 4 − Bk 2
2
C 31 =− 4 2
32 r Bk
 
+12 r−Bk Q2 C1 Bk 2 4 + Bk 2
2

 
+4 B 2 r2−Bk Λ C1 k 4 + 64 r−3 Bk /2 Q4 2 + Bk 2
2 2

 2

2−3 Bk2 /2 2 2 2 + Bk
+64 Br ΛQ k . (13.107)
(4 − Bk 2 )

Because of the presence of C1 and Λ = ±1/l2 in C 1 3 and C 3 1 , these tensor

components may be positive or negative quantities in some spacetime regions.
Therefore, in general, the Cotton tensor can be characterized algebraically by
three real eigenvalues
2
V2
λ1 = 0 : V1 = [0, V 2 , 0], Vμ V μ = , V1 = S1,
L2
λ2 = C 1 3 C 3 1 : V2 = [V 1 , 0, V 3 ],
 √ 
V 1 −α β C 3 1 + γ δ C 1 3 − C 3 1 C 1 3 Δ
−V =3
,
δ2 C 1 3 − β 2 C 3 1
2   √ 
V 1 Δ2 C 3 1 r2 − C 1 3 N 2 δ 2 C 1 3 + β 2 C 3 1 + 2 δ β C 3 1 C 1 3
Vμ V μ = 2 ,
(δ 2 C 1 3 − β 2 C 3 1 )
V2 = T2, N2, S2,
λ3 = − C 1 3 C 3 1 : V 3 = [V 1 , 0, V 3 ]
 √ 
V 1 −α β C 3 1 + γ δ C 1 3 + C 3 1 C 1 3 Δ
−V =3
,
δ2 C 1 3 − β 2 C 3 1
2    √ 
μ V 1 Δ2 C 3 1 r2 − C 1 3 N 2 δ 2 C 1 3 + β 2 C 3 1 − 2 δ β C 3 1 C 1 3
Vμ V = 2 ,
(δ 2 C 1 3 − β 2 C 3 1 )
V3 = T3, N3, S3. (13.108)
284 Dilaton Field Minimally Coupled to (2 + 1) Gravity

In this case the Cotton tensor is Type I with particular eigenvector subclasses:
{T, T, S}, {T, S, S}, . . . , {S, S, S}.
For one real and two complex conjugate eigenvalues, one has
V2
λ1 = 0 : V1 = [0, V, 0], Vμ V μ = , V1 = S1,
L2
λ2 = λ̄3 = C 13C 31 :
 √ 
V 1 −α β C 3 1 + γ δ C 1 3 − C 3 1 C 1 3 Δ
V 2 = Z = [V , 0, V ], V = −
1 3 3
,
δ2C 1 3 − β 2 C 3 1
V 3 = Z̄. (13.109)
Thus, the algebraic type of the Cotton tensor is Type I: {S, Z, Z̄}.

13.8.3 Particular Stationary Charged Dilaton via SL(2, R)

Transformation
Requiring the fulfillment of the relation
CN 2 Λ
γ δ = −4 αβ (13.110)
(B k 2 − 4)
the term with power r2 in W or equivalently in gT Φ disappears. The generalized
rotating charged Chan–Mann metric is given by
 
2 C N
4 2 2 2
α Λ r dr2
g = − CN 2 α2 L2 rBk − 16 β 2 2 d T 2
+
δ 2 (Bk 2 − 4) L2
CN 2 2
−2α β r B k /2 C1 Bk 2 + 4 Q2 d T d Φ
Bk 2
2
+ δ 2 r2 − β 2 CN 2 rBk L2 dΦ2 ,
 
r2 Λ Q2
B k2 /2
r−B k , Ψ = k ln(r),
2
2
L = r C1 + 4 +4
4−Bk 2 Bk 2

T r Φ r
Fμν = 2FT r δ[μ δν] + 2FΦ r δ[μ δν] , FT r = α Ftr , FΦ r = β Ftr ,
Ftr = QCN r−1−B k
2
/2
. (13.111)
Switching the rotation β = 0, the above-presented metric reduces to the Chan–
Mann charged dilaton metric.

13.9 Summary of Dilaton Minimally Coupled to Gravity

Using the Schwarzschild coordinate frame for a static cyclic symmetric metric
in (2 + 1) gravity coupled minimally to a dilaton logarithmically depend-
ing on the radial coordinate in the presence of a exponential potential and
 13.9 Summary of Dilaton Minimally Coupled to Gravity 285

an electrical Maxwell field the general solutions of the Einstein–dilaton and

Einstein–Maxwell–dilaton equations are derived.
In the Einstein–dilaton case, by solving first-order linear equations, the gen-
eral solution is derived and identified with the Chan–Mann dilaton solution.
Moreover, for completeness, the Chan–Mann solution is also derived in its orig-
inal coordinate frame avoiding any ansatz. A new stationary dilaton solution is
obtained in the Schwarzschild coordinate frame; it does not allows for an AdS
limit at spatial infinity, where its structural functions increase indefinitely; it is
horizonless and singular at the origin of coordinates; moreover, one can identify
at infinity a mass parameter and, in the whole space, a constant (quasi-local)
angular momentum.
In the Einstein–Maxwell–dilaton branch, the static solution is equivalent to
the Chan–Mann charged dilaton static solution. Via a SL(2, R) transformation
of the Killing coordinates, applied on the derived dilaton and charged static
cyclic symmetric metric, families of stationary dilaton solutions has been gen-
erated; they are equipped with five relevant parameters interpretable as dilaton
parameter, charge, momentum, cosmological constant, and mass for some values
of them. A particular SL(2, R) transformation is identified, which gives raise
to the charged generalization of the rotating Chan–Mann dilaton solution. At
spatial infinity all these solutions do not allow for an AdS–dS limit; their struc-
tural functions increase indefinitely as the radial coordinate increases. There
exists a horizon, structurally common to the full class of solutions, determin-
ing their black hole character for a range of physical parameters. This families
of solutions, the static and stationary ones, have been characterized by their
quasi-local energy, mass, and momentum through their series expansions at spa-
tial infinity. The algebraic classifications of the electromagnetic field, Maxwell
energy–momentum, and Cotton tensors are established. The electromagnetic
field tensor belongs to the type {S, N, N }. The Maxwell energy–momentum ten-
sor is of types: {2S, S},{2N, S},{2T,S},{(S, T ), S},. . . The Cotton tensor exhibits
various possibilities. For real roots it falls into type I:{T, T, S}, {T, S, S},· · · For
a complex root the Cotton tensor is of the Type I: {S, Z, Z̄}.
 14
Scalar Field Non-Minimally Coupled
to (2+1) Gravity

A stationary cyclic symmetric black hole solution for a scalar field non-minimally
coupled to (2+1)-dimensional gravity is derived and analyzed; its quasi-local
momentum, energy and mass are evaluated. They are asymptotically similar to
the ones of the rotating anti–de Sitter black hole solution with a contribution
due to the angular momentum parameter. The algebraic structure of the Ricci,
energy-momentum, and Cotton tensors is established.

14.1 Einstein Equations for Non-Minimally Coupled Scalar Field

The action for a dilaton non-minimally coupled to (2+1)-dimensional gravity is
given by
  
1 3 √ 1 2
S= d x −g (R + 2 ) − ∇μ Ψ ∇ Ψ − ζR Ψ ,
μ 2
(14.1)
2 κ l
where Λ = −1/l2 stands for the (negative) cosmological constant, Ψ is the mass-
less non-minimally coupled scalar field, ζ is the coupling constant, and R the
scalar curvature. The variations of this action yield the dynamical equations

1 1
Gμ ν − 2 δμ ν = κ ∇μ Ψ ∇ν Ψ − δμ ν ∇σ Ψ ∇σ Ψ
l 2

 
+ ζ Ψ2 − ∇μ ∇ν Ψ2 + Gμ ν Ψ2 ,

Ψ2 = ζ R Ψ, (14.2)
where  = ∇ ∇μ stands for the Laplace–Beltrami operator.
μ

Within (2+1)-dimensional Einstein theory, the general form of a stationary

cyclic symmetric line element – allowing for a timelike Killing vector ∂t and a
spacelike Killing vector field ∂φ – can be given as
dρ2
g = −N (ρ)2 dt2 + + K(ρ)2 (dφ + W (ρ)dt)2 .
L(ρ)2
 14.2 Stationary Black Hole for a Non-Minimally Coupled Scalar Field 287

14.1.1 Martinez–Zanelli Black Hole Solution with Tμ μ = 0

In the static case, W = 0, and for the class of metrics allowing for the polar
variable ρ, K = ρ, and restricted also to N (ρ) = L(ρ), the Einstein equations
(14.2) and the dynamical equation for the scalar Ψ allow for a unique solution;
see Ayón-Beato et al. (2000),
2
(ρ + m) (ρ − 2 m)
N (ρ)2 = L(ρ)2 = , K(ρ) = ρ, W (ρ) = 0,
l2 ρ
√ 
m
ΨM Z (ρ) = 8 , (14.3)
κ (ρ + m)
known as the Martı́nez and Zanelli (1996) black hole solution for a scalar field
non-minimally coupled to gravity . This solution arises for a coupling constant
ζ = 1/8, which yields the vanishing of the trace Tμ μ = 0 and consequently the
scalar curvature becomes constant R = 6 l−2 . Moreover, it describes a black hole
with horizon at ρ+ = 2 m, and presents a curvature singularity at the origin
ρ = 0.

14.2 Stationary Black Hole for a Non-Minimally Coupled

Scalar Field
Subjecting the above metric to the SL(2, R) transformation
l lω
t= √ T−√ Φ,
l −ω
2 2 l − ω2
2

lω l
φ = −√ T+√ Φ,
l −ω
2 2 l − ω2
2

ρ = ρ, (14.4)
one arrives at the metric
2 
l − ω 2 ρ3 − 3 l2 ρ m2 − 2 l2 m3 2 ω m2 (3 ρ + 2 m)
g=− dt − 2 dt dφ
(l − ω ) ρ l
2 2 2 (l2 − ω 2 ) ρ
 
3 ω 2 ρ m2 + 2 ω 2 m3 + l2 − ω 2 ρ3 2 ρ l2 2
+ dφ + 2 dρ . (14.5)
(l2 − ω 2 ) ρ (ρ − 2 m) (ρ + m)
Choosing the parameters as
lR− lJ M 1/2
ω=− =− , R± = M l ± M 2 l2 + J 2 , m = l, (14.6)
J R+ 31/2
one brings (14.5) to a representation close to the one of the BTZ solution at
spatial infinity ρ → ∞, namely
dr2
g = −N (ρ)2 dt2 + + K(ρ)2 (dφ + W (ρ)dt)2 ,
L(ρ)2
ρ2 M 3/2 l
L(ρ)2 = − M − 2 ,
l2 33/2 ρ
288 Scalar Field Non-Minimally Coupled to (2+1) Gravity

1 M 1/2 l2 R− L(ρ)
K(ρ)2 = ρ2 − lR− − 3/2 , N (ρ) = ρ ,
2 3 ρ K(ρ)
 
J M 1/2 l
W (ρ) = − 1 + 2 ,
2 K(ρ)2 33/2 ρ
√
2 Ml
Ψ =8 √ √ . (14.7)
κ 3ρ + M l

1/2
This solution represents a black hole with horizon at ρ = 2 m = 2 M
31/2
l.

14.2.1 Quasi-Local Momentum, Energy, and Mass

To characterize non asymptotically flat solutions one uses the Brown–York for-
malism of quasi-local momentum, energy, and mass quantities, see Section 1.3
and the summary (1.3.1), which for the stationary cyclic symmetric metric (14.7)
gives the physical quasi-local momenta
√ √
3ρ + 3 Ml J
jφ (ρ) = √  √ √ ,
√
π 2 ρ 18 ρ3 − 9 lR− ρ − 2 3 M l2 R−
√
J Ml J
j(ρ → ∞) ≈ + √ ,
2π ρ 2 3π ρ2
√
Ml J
J(ρ) = J + √ , J(ρ → ∞) ≈ J, (14.8)
3ρ
while for the energy and mass one gets
√ √ √
9 ρ3 − 9 M l2 ρ − 2 3M 3/2 l3 18 ρ3 + 3 M l2 R−
(ρ) = − √ √ ,
3 l πρ3/2 18 ρ3 − 9 lR− ρ − 2 3 M l2 R−
√ √ √ √
2 9 ρ3 − 9 M l2 ρ − 2 3M 3/2 l3 18 ρ3 + 3 M l2 R−
E(ρ) = −  √ √ ,
9 lρ2 18 ρ3 − 9 lR− ρ − 2 3 M l2 R−
 √  √ √
4 9 ρ3 − 9 M l2 ρ − 2 3M 3/2 l3 18 ρ3 + 3 M l2 R−
M (ρ) = − √ √
9 l2 ρ 18 ρ3 − 9 lR− ρ − 2 3 M l2 R−
√ √ √ √
4J 2 9 ρ + 2 3 M l 3 ρ + 3 M l
+ √ √ . (14.9)
3ρ 18 ρ3 − 9 lR− ρ − 2 3 M l2 R−

The series expansion of energy and mass functions independent of 0 behave at

infinity ρ → ∞ as
 14.2 Stationary Black Hole for a Non-Minimally Coupled Scalar Field 289
√
1 M 2 l2 + J 2
(ρ → ∞, 0 = 0) ≈ − + ,
lπ 2π ρ2
√
ρ Ml M 2 l2 + J 2
E(ρ → ∞, 0 = 0) ≈ −2 + + ,
l 2ρ 2ρ
√
ρ2 M 2 l2 + J 2
M (ρ → ∞, 0 = 0) = −2 2 + M + . (14.10)
l l
The reference energy density to be used is
the one corresponding to the AdS
2
metric with parameter M0 , 0 (M0 ) = − πρ 1
l2 − M0 , 0|∞ (M0 ) ≈ − π l + 2π ρ2 ,
ρ 1 l M0

the series expansions of the corresponding quantities at ρ → infinity result in

√
M 2 l2 + J 2 − lM0 M − M0 1 J2
(ρ → ∞, 0|∞ (M0 )) ≈ 2
≈ 2
+ ,
2π ρ 2πρ 4 π ρ2 M l
√
M 2 l2 + J 2 − lM0 M − M0 1 J2
E(ρ → ∞, 0|∞ (M0 )) ≈ ≈ +
ρ ρ 2ρ M l
√
2 2
M l +J 2 J2
M (ρ → ∞, 0|∞ (M0 )) ≈ − M0 ≈ M − M0 + . (14.11)
l 2 M l2
Where the third column in the above approximations hold for M l  J. There-
fore, comparing with the energy characteristics of the BTZ solution – see
Section 4.1.1, (4.10) – one concludes that role of the mass parameter is played
by M , and of the momentum parameter by J. At spatial infinity all physical
quantities are finite; in particular, the field Ψ vanishes.

14.2.2 Algebraic Classification of the Ricci, Energy–Momentum,

and Cotton Tensors
The Ricci tensor, determining in (2+1) dimensions the Riemann curvature
tensor, is described by the matrix (Rμ ν )
⎡ ⎤
− −2 l ρ +lρ3m(l2+2
2 3 2 3
ω 2 ρ3 +2 ω 2 m3 3ω m3
−ω 2 )l2 0 ρ3 (l2 −ω 2 )
⎢ ⎥
⎢ ⎥
⎢ 0 2 ρ3 −m3
0 ⎥,
⎢ l2 ρ3 ⎥
⎣ ⎦
3 2 l ρ +2 l m −2 ω ρ +ω m
2 3 2 3 2 3 2 3
− ρ3 (l32ω−ω
m
2 )l2 0 ρ (l −ω )l
3 2 2 2

(14.12)
and is algebraically characterized by the following eigenvectors:
2  
ρ3 + m3 3 3 μ (V 3 ) ρ2 l2 − ω 2
λ1 = 2 2 3 : V1 = [ω V , 0, V ], Vμ V = , V1 = S1,
l ρ l2
2 ρ3 − m3 V1ω
λ2,3 = 2 3
: V2, 3 = [V 1 , V 2 , 2 ],
l ρ l
2 4 2  2
μ −(V ) (ρ − 2 m) (ρ + m) l − ω 2 + (V 2 ) ρ2 l6
1 2
Vμ V = 2 , (14.13)
l4 ρ (ρ − 2 m) (ρ + m)
290 Scalar Field Non-Minimally Coupled to (2+1) Gravity

hence depending√on the sign of this norm one will have spacelike (V 1 = 0),
2
null (V 2 = ± l2 − ω 2 (ρ + m) (ρ − 2 m) V 1 /(l3 ρ)), or timelike (V 2 = 0)
eigenvectors, and also when
√ 2
1 δ l − ω (ρ + m) (ρ − 2 m) V
1
2 2 V1ω
V2, 3 = [V , , ] = {S, N, T}, (14.14)
l3 ρ l2

where δ may assume the values greater than unit δ > 1, equal to unit δ = 1
and less than unit δ < 1, determining correspondingly spacelike, null, or timelike
eigenvectors. Thus, in the case of a double root one may choose different vec-
tor components determining, for instance, one spacelike vector S and the other
timelike T or null N one. Therefore, one may have the algebraic Ricci types:
{S, 2S}, {S, 2N }, {S, 2T } and {S, (S, T )}, {S, (T, T )}, . . . , {S, (T, N )}.
For the energy–momentum tensor matrix (T μ ν )
⎡ ⎤
(l2 m2 +2 ω2 m2 +ω2 ρ2 −l2 ρ2 )m 3

κ l2 ρ2 (ρ+m)(l2 −ω 2 ) 0 − κ ρ2 (ρ+m)(l
3ω m
2 −ω 2 )
⎢ ⎥
⎢ ⎥
⎢ 0 (m−ρ)m
0 ⎥
⎢ ρ2 κ l2 ⎥
⎣ ⎦
3ω m3 (2 l 2
m +ω m2 −ω 2 ρ2 +l2 ρ2 )m
2 2

κ l2 ρ2 (ρ+m)(l2 −ω 2 ) 0 − κ l2 ρ2 (ρ+m)(l2 −ω 2 )

one has the eigenvectors

   
m 2 m2 + ρ2 ρ2 ω V3 V3 2 ρ2 l2 − ω 2
λ1 = − 2 2 : V1 = [− 2 μ
, 0, ρ V3 ], Vμ V = ,
κ l ρ (ρ + m) l2 l2
V1 = S1,
(m − ρ) m 1 2 V ω
1
λ2 = : V2, 3 = [V , V , ],
κ ρ2 l2 l2
2 4 
μ −(V1 )2 (ρ − 2 m) (ρ + m) l2 − ω 2 + (V2 )2 ρ2 l6
Vμ V = 2 . (14.15)
l4 ρ (ρ − 2 m) (ρ + m)

Therefore, as in the case of the algebraic structure of the Ricci tensor, one may
distinguishes spacelike (V1 = 0), null
2
V2 = ± l2 − ω 2 (ρ + m) (ρ − 2 m) V1 /(l3 ρ),

or timelike (V2 = 0) eigenvectors, and also when

√ 2
δ l2 − ω 2 (ρ + m) (ρ − 2 m) V1 V1 ω
V2, 3 = [V1 , , 2 ] = {S, N, T},
l3 ρ l
(14.16)

for δ > 1, = 1, < 1. The same comment about the double root and its vectors
applies in this case. Consequently, one may have the algebraic energy–momentum
types: {S, 2S}, {S, 2T }, {S, 2N } and {S, (S, T )}, . . . , {S, (T, N )}.
 14.2 Stationary Black Hole for a Non-Minimally Coupled Scalar Field 291

The existence of this relationship between the eigenvectors of the Ricci and
energy–momentum tensors is due to the tensor relation
κ ν (2 ρ − m) ν
Rμ ν + Tμ − δμ = 0.
ρ l2 ρ
For the Cotton tensor matrix (C μ ν )
⎡ ⎤
3ω m3 (−2 ρ3 +3 ρ m2 +2 m3 ) 3m3 (−l2 ρ3 −ω 2 ρ3 +3 ω 2 ρ m2 +2 ω 2 m3 )
− 0
⎢ 2l2 ρ6 (l2 −ω 2 ) 2l2 ρ6 (l2 −ω 2 ) ⎥
⎢ ⎥
⎢ 0 0 0 ⎥,
⎢ ⎥
⎣ ⎦
3m (−l ρ +3 l ρ m +2 l m −ω ρ )
3 2 3 2 2 2 3 2 3
3ω m 3
(−2 ρ3
+3 ρ m2 +2 m3 )
− 2l4 ρ6 (l2 −ω 2 ) 0 2l2 ρ6 (l2 −ω 2 )

the characteristic equation allow for one real and a pure imaginary eigenvalues,
therefore, the Cotton tensor is characterized algebraically by
(V2 )2 ρ l2
λ1 = 0 : V1 = [0, V2 , 0], Vμ V μ = 2 , V1 = S1,
(ρ − 2 m) (ρ + m)
ρ (ρ − 2 m) (ρ + m) m3
3i
λ2 = λ̄3 = :
2ρ5 l3
 
1 m3 ω 6 ρ3 − 9 m2 ρ − 6 m3 − 2 λ l2 ρ6 (l2 − ω 2 ) 1
V2 = Z = [V , 0, V ],
3 [(l2 + ω 2 )ρ3 − ω 2 m2 (3 ρ + 2 m)] m3
V3 = Z̄. (14.17)
Thus, the algebraic type of the Cotton tensor is given by Type I: {S, Z, Z̄}.
 15
Low-Energy (2+1) String Gravity

In this chapter the dynamical equations for an n-dimensional heterotic string

theory of the Horowitz type are shown explicitly in the string frame and in
the Einstein frame too. In particular, the dynamical equations of the three-
dimensional string theory are also given. A detailed derivation of these equations
has been accomplished by Garcia–Diaz and Gutierrez–Cano (2014a); this work is
followed in part. The relation of the Horowitz–Welch and Horne–Horowitz string
black hole solution is exhibited. The Chan–Mann charged dilaton solution is
derived and the subclass of string solutions is explicitly identified. The stationary
generalization, via SL(2, R) transformations, of the static (2+1) Horne–Horowitz
string black hole solution is given.

15.1 n-Dimensional Heterotic String Dynamical Equations

In this section we reproduce the field equations “for a part of the low energy
action” of Horowitz (1992), to an n-dimensional heterotic string theory described
by a metric gμν , a scalar field Φ, a Maxwell field Fμν , and a three-form Hμνλ .
The three-form H is related to the two-form potential B and a gauge field Aμ
through H = dB − a A ∧ dF , where a is a constant to be adjusted at the
end for final results. In this text, n is used instead of D to denote the number
of dimensions. Moreover, Λ is reserved for the standard cosmological constant,
whereas ΛH , and ΛCM = −Λ, denote the Λs used by Horowitz (1992) and Chan
and Mann (1996), respectively.

15.1.1 String Frame


The heterotic string action for dimension n, S = d n xL, is given by
  
√ 1
S = d n x −ge−2Φ R − 2Λ + U (Φ) + 4(∇Φ)2 − F 2 − H 2 . (15.1)
12
Variations with respect to the metric, the scalar Φ field, the totally anti-
symmetric H field, and the electromagnetic F field lead to the dynamical field
equations
 15.1 n-Dimensional Heterotic String Dynamical Equations 293

1
Gμν − gμν (−2Λ + U ) + 2∇μ ∇ν Φ − 2gμν ∇2 Φ + 2gμν (∇ Φ)2
2
1 1 1
−2 (Fμα Fνβ g αβ − gμν F 2 ) − (Hμαβ Hν αβ − gμν H 2 ) = 0, (15.2a)
4 4 6
1 1 dU
4∇2 Φ − 4(∇ Φ)2 + R − 2 Λ − F 2 − H 2 + U (Φ) − = 0, (15.2b)
6 2 dΦ

∇σ (e−2Φ H μνσ ) = 0, (15.2c)

  1
∇λ e−2Φ F λ + a e−2Φ Fνλ H νλ = 0. (15.2d)
12
The last equation differs in sign from Horowitz’s corresponding equation (2.10b).
By contracting (15.2a) one evaluates R,
 
2 n n−4 2 n−6 2
R= nΛ − U − 2(n − 1)∇2 Φ + 2n(∇Φ)2 + F + H ,
n−2 2 2 24
(15.3)
which, replaced again into (15.2a) and (15.2b), allows one to rewrite the set of
dynamical equations as
1
Rμν = −2∇μ ∇ν Φ + 2Fμα Fνβ g αβ + Hμαβ Hν αβ
 4 
2 1 1 1
+ gμν Λ − U − ∇ Φ + 2(∇Φ)2 − F 2 − H 2 ,
2
(15.4a)
n−2 2 2 12

1 n − 2 dU
2∇2 Φ − 4(∇Φ)2 − 2Λ + F 2 + H 2 + U + = 0, (15.4b)
6 4 dΦ

∇σ (e−2Φ H μνσ ) = 0, (15.4c)

  1
∇λ e−2Φ F λ + a e−2Φ Fνλ H νλ = 0. (15.4d)
12

15.1.2 Einstein Frame

On the other hand, to pass to the Einstein frame description of this low energy
string theory, one accomplishes a conformal transformation of the form
g̃μν = e2σ gμν , g̃ μν = e−2σ g μν → g̃ = e2 nσ g, dim = n, (15.5)
which transforms the action S (15.1), considered as the barred one, taking into
account that F̃ 2 = g̃ μα g̃ νβ Fμν Fαβ = e−4σ g μα g νβ Fμν Fαβ = e−4σ F 2 , and H̃ 2 =
e−6σ H 2 , to the form
 
√
S= d n x −ge(n σ−2 σ−2Φ) e2σ R̃ + e2 σ (−2Λ + U ) + 4(∇Φ)2
1 
−e−2σ F 2 − e−4σ H 2 . (15.6)
12
294 Low-Energy (2+1 ) String Gravity

Thus, one may use the conformal transformed curvature scalar

e2 σ R̃ = R − 2(n − 1)g μν ∇ν ∇μ σ − (n − 1)(n − 2)g μν ∇μ σ∇ν σ, (15.7)

and choose
2 n−2
n σ − 2 σ − 2Φ = 0 → σ = Φ, Φ = σ. (15.8)
n−2 2
Substituting these relations in the action above (15.6), one arrives at
 
n √ 4
S= d x −g R − (∇Φ)2 − 2e4 Φ/(n−2) Λ + V (Φ)
n−2

−4 Φ/(n−2) 2 1 −8 Φ/(n−2) 2
−e F − e H , (15.9)
12
where the divergence has been dropped from this action
√ √
−2(n − 1) −gg μν ∇ν ∇μ σ = −2(n − 1)( −gσ ,ν ),ν ,

and denoted
e2σ U (Φ) = V (Φ).

The extremum of S is achieved along the dynamical equations:

 
1 4 1
Gμν = −gμν e 4Φ/(n−2)
Λ + gμν V (Φ) + ∇μ Φ ∇ν Φ − gμν ∇α Φ ∇ Φα
2 n−2 2
1 1 1
+ 2 e−4Φ/(n−2) (Fμσ Fν σ − gμν F 2 )− e−8Φ/(n−2) (Hμαβ Hν αβ − gμν H 2 ),
4 4 6
(15.10a)

2 −8Φ/(n−2) 2 n − 2 d V
8 ∇ν ∇ν Φ − 8 Λe4Φ/(n−2) + 4 e−4Φ/(n−2) F 2 + e H + = 0,
3 2 dΦ
(15.10b)

∇σ e−8Φ/(n−2) H αβσ = 0, (15.10c)

a −8Φ/(n−2)
∇λ e−4Φ/(n−2) F λ + e Fαβ H αβ = 0. (15.10d)
12
Replacing in (15.10a) the scalar curvature R,
4 n − 4 −4Φ/(n−2) 2 1 n − 6 −8Φ/(n−2) 2
R= (∇Φ)2 + e F + e H
n−2 n−2 12 n − 2
n
− (−2Λ e4Φ/(n−2) + V ),
n−2
one rewrites (15.10a) as
 15.2 Dynamical Equations in (2+1) String Gravity 295

2Λ 1 4
Rμν = gμν e4Φ/(n−2) − gμν V (Φ) + ∇μ Φ ∇ν Φ
n−2 n−2 n−2
 
1
+ e−4Φ/(n−2) 2 Fμσ Fν σ − gμν F 2
n−2
 
1 2 1
+ e−8Φ/(n−2) 2 Hμαβ Hν αβ − gμν H 2 . (15.11)
4 3n−2

15.2 Dynamical Equations in (2+1) String Gravity

In the three-dimensional case, the above Einstein action (15.9) reduces to
  
3 √ −4Φ 2 1 −8Φ 2
S = d x −g R − 2e Λ − 4(∇Φ) − e
4Φ 2
F − e H + V (φ) ,
12
(15.12)
and the Einstein frame dynamical equations (15.10) become
 
Rμν = 2Λ gμν e4Φ + 4 ∇μ Φ ∇ν Φ + e−4Φ 2 Fμσ Fν σ − gμν F 2
 
1 1
+ e−8Φ Hμαβ Hν αβ − gμν H 2 − gμν V (Φ). (15.13a)
2 3

2 −8Φ 2 1 d V
8 ∇2 Φ − 8 Λe4Φ + 4 e−4Φ F 2 + e H + = 0, (15.13b)
3 2 dΦ
 
∇σ e−8Φ H αβσ = 0, (15.13c)
 −4Φ λ  a −8Φ
∇λ e F + e Fαβ H αβ = 0. (15.13d)
12
One can recover the string dynamical equations by using the conformal inverse
relations (see Eisenhart, 1966),
˜ 2 , e−2 σ σ̃ ,k = σ ,k ;k − (n − 2)(∇σ)
σ,i;j = σ̃,i;j + 2σ,i σ,j − g̃ij (∇σ) ˜ 2,
;k
W W̃ ,k ˜ 2 ,
Rij = R ij + g̃ij σ̃;k + (n − 2) σ̃,i;j + σ,i σ,j − g̃ij (∇σ)
W W̃
e−2 σ R = R + 2(n − 1)σ̃;k
,k ˜ 2,
− (n − 1)(n − 2)(∇σ) (15.14)

where tilde is used to denote that covariant differentials are constructed with Γ̃s
or contravariant tensor components are built with g̃ μν . For σ = 2Φ and n = 3
one gets
˜ 2,
Φ,μ;ν = Φ̃,μ;ν + 4Φ,μ Φ,ν − 2g̃μν (∇Φ)
Φ;α ;α = e4Φ Φ̃;α
;α ˜ 2 ,
− 2(∇Φ)
˜ 2
;α + 4Φ,μ Φ,ν − 4g̃μν (∇Φ) ,
Rμν = R̃μν + 2Φ̃,μ;ν + 2g̃μν Φ̃;α
˜ 2 ,
;α − 8 (∇Φ)
R = e4Φ R̃ + 8 Φ̃;α (15.15)
296 Low-Energy (2+1 ) String Gravity

and conformally transforming the above dynamical equations (15.13), using rela-
tions (15.15), one gets the barred dynamical equations of the (2+1) string theory
under consideration
  
−2Φ ˜ 1 2
S = d x −g̃e
3
R̃ − 2Λ + 4(∇Φ) − F̃ − H̃ + U (φ) ,
2 2
12
(15.16a)

1
R̃μν + 2∇μ ∇ν Φ − 2Fμα Fνβ g̃ αβ − Hμαβ Hνγλ g̃ γα g̃ λβ
 4 
H̃ 2 ˜ ˜
+ g̃μν −2Λ + U + F̃ +
2
+ 2∇ Φ − 4(∇Φ)
2 2
= 0, (15.16b)
6

2∇ ˜ 2 + F̃ 2 − 2Λ + 1 H̃ 2 + U (Φ) + 1 d U = 0,
˜ 2 Φ − 4(∇Φ) (15.16c)
6 4 dΦ
˜ σ (e−2Φ H̃ μνσ ) = 0,
∇ (15.16d)

˜ λ e−2Φ F̃ λ + a 1 e−2Φ F̃νλ H̃ νλ = 0.

∇ (15.16e)
12

15.3 Horne–Horowitz Black String

Horne and Horowitz (1992) published an exact string black hole solution in three
dimensions for the full string theory of Section 15.1, endowed with mass, axion
charge per unit length, the asymptotic value of the dilaton, and a cosmological
constant. This string solution is given by
S M 2 Q2 M −1 Q2 −1 k dr2
g = −(1 − )dt + (1 − )dx2 + (1 − ) (1 − ) ,
r Mr r Mr 8r2
Q 1 k
Hrtx = 2 , φ = ln r + ln . (15.17)
r 2 2
The identification of the functions appearing in the action (15.1) and equations
(15.2) with the ones of Horne and Horowitz (1992) corresponds to φ = −2Φ, and
k = −2Λ = l2 , k = 2 l , where l has the dimension of length.
8 4 2

Accomplishing in the above mentioned metric a conformal transformation

E S 2 S
g μν = e−2σ g μν = k 2r g μν , where superscripts E and S stand for Einstein and
String respectively, one arrives at the corresponding solution in the Einstein
frame, namely
E k r2 Q2 M −1 Q2 −1 k 2 dr2
g= (1 − )dx2 + (1 − ) (1 − ) ,
2 Mr r Mr 16
k r2 M 2 Q 1 1 k
− (1 − )dt , Hrtx = 2 , Φ = − ln r − ln , (15.18)
2 r r 2 4 2
fulfilling the dynamical equations (15.10) for dimension n = 3.
 15.3 Horne–Horowitz Black String 297

By accomplishing a SL(2, R) transformation of the Killing coordinates

t = α T + β φ, Δ := αδ − βγ.
x = γ T + δ φ,

one arrives at a stationary HH-string solution

 
E k r2 M 2 Q2 2
g =− (1 − )α − (1 − )γ dT 2
2 r Mr
 2

Q M
+ k r2 (1 − )γδ − (1 − )αβ dT dφ
Mr r
2
 2

kr Q M 2 M Q2 −1 k 2 dr2
+ (1 − )δ 2 −(1 − )β dφ2 + (1− )−1 (1− ) ,
2 Mr r r Mr 16
Q 1 1 k 8 4
Hrtx = 2 , Φ = − ln r − ln , = −2Λ = 2 . (15.19)
r 2 4 2 k l
In the literature one frequently encounters the SL(2, R) transformation
T φ ω T φ
t=  −ω  ,x=− 2
 + ,
1− ω2
1− ω2 l 1− ω2
1− ω2
l2 l2 l2 l2

which yields
   
(r − M ) rM − Q2 rk l2 − ω 2
g=− 2 dT 2
2 l (−ω 2 rM + ω 2 M 2 + rM − Q2 )
M −1 Q2 −1 k 2 dr2
+ (1 − ) (1 − )
r Mr 16 
l2 kr −ω 2 rM + ω 2 M 2 + rM − Q2
+
2M (l2 − ω 2 )
  2  2
ω −l M r + l2 M 2 + rM − Q2
× dφ − 2 dT ,
l (−ω 2 rM + ω 2 M 2 + rM − Q2 )
Q 1 1 k
Hrtx = , Φ = − ln r − ln . (15.20)
r2 2 4 2
The evaluation of the quasilocal mass, energy and momentum is done using the
Brown–York approach; this yields
 2 
2 M −Q
2
J(r → ∞) = 2ω l 2 , (15.21a)
(l − ω 2 ) M

1 (M − ω Q) (M + ω Q)
(r → ∞) = −1/2 − 1/4 − 0 , (15.21b)
rπ l2 M l2 π (ω − 1) (ω + 1) r2
√
2 1 − ω2
E(r → ∞) = √ − 2π K(r)0 , (15.21c)
l2 − ω 2
298 Low-Energy (2+1 ) String Gravity

MBY (r → ∞) = 4r + 4 l (−2r2 + M0 l2 ), (15.21d)

1 1 M0 l
0 = − + . (15.21e)
π l 2 π r2

ω is interpreted as a rotating parameter and the mass function increases as the

radial coordinate approach spatial infinity. Moreover, various pathologies take
place at this location.

15.4 Horowitz–Welch Black String

Horowitz and Welch (1993) published an exact string black hole solution in three
dimensions for the low-energy string theory
  
3 √ −2Φ 1 2
S = d x −ge R − 2Λ + 4(∇Φ) − H , Λ = −2/kHW , (15.22)
2
12
of Section 15.1, endowed with mass, angular momentum, axion charge per unit
length, and a negative cosmological constant. This string solution is given by a
modified BTZ black hole to a (2 + 1) string theory with vanishing scalar Φ and
electromagnetic Fαβ fields; in this last case the field equations (15.4) become
1
Rμν − Hμαβ Hν αβ = 0, (15.23a)
4
1
− 2Λ + H 2 = 0, (15.23b)
6
∇σ (H μνσ ) = 0. (15.23c)
The totally anti-symmetric tensor Hμνα has to be proportional to the volume
three form μνα , because of the equation (15.23c), the proportionality factor
ought to be a constant, hence one may choose
1
Hμνσ = α μνσ . (15.24)
l
Taking into account the properties of 
ανσ βνσ = −2δ β α , (15.25)
therefore
α2 β α2
Hανσ H βνσ = −2 δ α , H 2
= −6 . (15.26)
l2 l2
Consequently, from (15.23a) one has
1 2
Rμν − Hμαβ Hν αβ = 0 → Rμν = − 2 gμν , α = 2 (15.27)
4 l
1 α2 2
− 2Λ + H 2 = 0 → Λ = − 2 = − 2 , kHW = l2 . (15.28)
6 2l l
 15.4 Horowitz–Welch Black String 299

On the other hand, H = dB, thus

2
Hαβγ = 3B[αβ,γ] = αβγ . (15.29)
l
As is concluded in Horowitz and Welch (1993): “Thus every solution to three
dimensional general relativity with negative cosmological constant is a solution
to low energy string theory with: Φ = 0, Hαβγ = 2l αβγ , and Λ = − l22 .”
In particular, in Horowitz and Welch (1993) it is established that the BTZ
black hole metric
r2 r2 J2
g = −( 2
− M )dt2 − J dt dφ + r2 dφ2 + ( 2 − M + 2 )−1 dr2 , (15.30)
l l 4l
in the presence of an anti-symmetric B field
r2
Bφ t = , H = dB, (15.31)
l
is a solution of the string theory with a zero scalar field Φ.
By a target space duality transformation, Eq. (13) of Horowitz and Welch (1993),
which referred to Buscher (1993), means that from a given solution (gμν , Bμν , Φ)
independent on one variable, say x, one generates a new solution (g̃μν , B̃μν , Φ̃)
with
1 Bxα
g̃xx = , g̃xα = ,
gxx gxx
1
g̃αβ = gαβ − (gxα gxβ − Bxα Bxβ ),
gxx
gxα gx[α Bβ] x
B̃xα = , B̃αβ = Bαβ − 2 ,
gxx gxx
1
Φ̃ = Φ − ln gxx , (15.32)
2
where α and β run over all directions except x. Applying this transformation to
expressions (15.30) and (15.31), along the coordinate symmetry φ, one gets Eq.
(14) of Horowitz and Welch (1993), namely
S J2 2 1 r2 J2
g = (M −
2
)dt2 + dt dφ + 2 dφ2 + ( 2 − M + 2 )−1 dr2 , (15.33)
4r l r l 4r
which, once diagonalized by means of a SL(2, R) coordinate transformation
l l
t = − t̃ +  x̃,
2 − r2
r+ 2 − r2
r+
− −
2 2
r+ r−
φ=  t̃ −  x̃, (15.34)
2 − r2
r+ 2 − r2
r+
− −

and the r-coordinate transformation

r2 = lr̃ (15.35)
300 Low-Energy (2+1 ) String Gravity

yields the string solution derived in Horne and Horowitz (1992); see also Section
15.3. Dropping primes, it becomes

S M 2 Q2 M −1 Q2 −1 l2 dr2
g = −(1 − )dt + (1 − )dx2 + (1 − ) (1 − ) ,
r Mr r Mr 4r2
Q 1
Bxt = , φ = − ln (r l), (15.36)
r 2

where M = r+ 2
/l and Q = J/2.
The identification of the functions appearing in the action (15.1) and equations
(15.2) with the ones of Horne and Horowitz (1992),
  
√ 1
S= d 3 x −geφ R − 2Λ + (∇Φ)2 − H 2 , Λ = −4/kHH , (15.37)
12

requires that φ = −2ΦHW , and kHH = 2 kHW , kHW = l2 , Λ = −2/l2 , where the
subscripts are in correspondence with the initial of the author’s family name.

15.5 Chan–Mann String Solution

Chan and Mann (1994, see also Chan and Mann, 1996), derived a class of solu-
tions to dilaton minimally coupled to (2 + 1) Einstein–Maxwell gravity. There is
a subclass of solutions allowing an interpretation from the viewpoint of the low
energy (2 + 1) string theory for specific values of the charged dilaton solution.
First, we derive the dilaton solution. Next, assigning specific values to constants
characterizing the charged dilaton, the correspondence with the string theory
developed in the previous section is established.

15.5.1 Einstein–Maxwell-Scalar Field Equations

The Chan and Mann (1994) action for a (2+1)-dimensional gravity is given by
  
√ B −4 a Ψ 2
S= d x −g R − ∇μ Ψ ∇ Ψ + 2 e ΛCM − e
3 μ bΨ
F ,
2

where the parameters ΛCM , b are arbitrary at this stage, Ψ is the massless
minimally coupled scalar field, R is the scalar curvature, and F 2 = Fμ ν F μ ν
the electromagnetic invariant. The variations of this action yield the dynamical
equations

B  
Rμν = ∇μ Ψ ∇ν Ψ − 2gμν e b Ψ ΛCM + 2 e−4 a Ψ Fμ α Fν α − gμ ν F 2 ,
2
B μ
∇ ∇μ Ψ + b e b Ψ ΛCM + 2 a e−4 a Ψ F 2 = 0,
2  
∇μ e−4 a Ψ Fμ ν = 0.
 15.5 Chan–Mann String Solution 301

15.5.2 Static and Stationary Black String Solutions

The static cyclic symmetric metric in the (2 + 1) Schwarzschild coordinate frame
was derived in Section 13.7. It is given explicitly by (13.82), namely
dr2
gE = −r2 L(r)2 dt2 + + r2 dφ2 ,
L(r)2
 
L(r)2 = r C1 − 2 r2 Λs + 2 Q2 r−2 ,
Q Q
t r
Fμν = 2Ftr δ[μ δν] , Ftr = 2 = −At,r → At = ,
r r
Ψ(r) = −1/2 ln (r). (15.38)
This string solution was derived for the first time in Chan and Mann (1994), for
B = 8, k = −1/2, a = 1, b = 4; it fulfills the Einstein string equations (15.13) for
Λ = Λs = ± l12 = −ΛCM , in the case of vanishing H.
Under the conformal transformation
1
g̃μν = e4 Ψ(r) gμν = 2 gμν (15.39)
r
it becomes
  dr2
gS = − r C1 − 2 r2 Λs + 2 Q2 r−2 dt2 +   + dφ2 ,
r C1 − 2 r2 Λs + 2 Q2
Q Q
t r
Fμν = 2Ftr δ[μ δν] , Ftr = = −At,r → At = ,
r2 r
Ψ(r) = −1/2 ln (r); (15.40)
this is a solution of the equations (15.4) of the (2 + 1) string theory.
Moreover, subjecting the metric (15.40) to SL(2, R) transformations of the
Killing coordinates
t = α τ + β θ, φ = γ τ + δ θ
one arrives at a rotating charged string solution, namely
   
gS = − α2 L2 /r2 − γ 2 dτ 2 − 2 αβ L2 /r2 − γ δ dθ dτ
  dr2
+ δ 2 − β 2 L2 /r2 dθ2 + 2 ,
L
Q Q
τ r
Fμν = 2α Ftr δ[μ δν] − 2β Ftr δ[μ
θ r
= −At,r → At = ,
δν] , Ftr :=
2
r r
Ψ(r) = −1/2 ln (r), L2 := r C1 − 2 r2 Λs + 2 Q2 . (15.41)
In particular when ΛCM = 1/l2 , AdS branch, and the usual choice of the SL(2, R)
transformations is adopted
τ θ
t= −ω ,
1− 1 − ω 2 /l2
ω 2 /l2
ω τ θ
φ=− 2 + , (15.42)
l 1 − ω /l
2 2 1 − ω 2 /l2
302 Low-Energy (2+1 ) String Gravity

where ω stands for the rotation parameter, this metric can be written as
L2 1 − ω 2 /l2
gS = − dτ 2
r2 1 − ω 2 + L2 /r2
 2
1 − ω 2 L2 /r2 ω 1 − ω 2 /l2 dr2
+ dθ − dτ + 2, (15.43)
1 − ω /l
2 2 l 1 − ω L /r
2 2 2 2 L
and is endowed with four parameters: the mass M = −C1 , charge Q, rotation
ω, and cosmological constant Λs = ±1/l2 . It is worth pointing out that string
solutions can be found in various dimensions; see, for instance, the works of
Witten (1991); Mandal et al. (1991); Maki and Shiraishi (1993), among others.
 16
Topologically Massive Gravity

In this part of the book we deal with exact solutions to the Einstein topologically
massive gravity equations. However, since the material to be included only rep-
resents twenty per cent of the whole book’s subject matter, we prefer to present
this content in the form of chapters devoted to a concise but, we hope, complete
(in the range of the possibilities) exposition of the exact solutions in topologically
massive gravity (TMG) in three dimensions in the case of vacuum in the presence
of a cosmological constant Λ of both signs. Thus, this chapter has an introduc-
tory character, while the next three chapters deal with very specific families of
the existing Petrov-type Cotton solutions in TMG.
The extension of the 3D Einstein gravity to other field theories to provide
them with certain degrees of freedom (a massive spin 2 graviton) was proposed
more than thirty-five years ago by Deser, Jackiw and Templeton; see Deser et al.
(1982a): “Three-dimensional massive gauge theories,” which is known as the
TMG. It includes a Chern–Simons term constructed from connections (Cot-
ton tensor) with broken parity invariance; see also its extended version, with
a detailed analysis, in Deser et al. (1982b). A cosmological constant was intro-
duced in these three-dimensional theories by Deser (1984). The sign in front of
the curvature scalar has been chosen opposite to the standard one of the 4D
Einstein gravity to yield, in the limit of the linearized theory, to the existence of
a spin 2 graviton with positive energy. A modern treatment of these aspects in
cosmological massive gravity appeared recently in Carlip et al. (2009).
As far as the determination of exact solutions to vacuum equations with a
cosmological constant in TMG is concerned, through this long period, various
classes of solutions have been reported apart from the trivial Minkowski flat
spacetime, the de Sitter (A)dS3 cosmology, and the BTZ black hole – confor-
mally flat (zero Cotton tensor) solutions – for any value of the coupling mass
parameter μ. Among these classes one can mention homogeneous spaces, pp-
waves metrics, cosmological solutions, and Kerr–Schild metrics. Fortunately, a
systematic up-to-date classification of the existing families of vacuum solutions
304 Topologically Massive Gravity

in TMG has been done recently by Chow, Pope, and Sezgin (CPS): see Chow
et al. (2010a), referred to as CPSa, and Chow et al. (2010b), referred to as CPSb,
for Kundt spacetimes. Thus, I shall take advantage of this progress and incor-
porate their results in this book; I shall work out the derivation of the various
classes reported in the literature using a presentation pattern similar to CPS and
exhibit their characterization by means of the Garcı́a et al. (2004) Petrov–Segré
classification of the Cotton tensor and other relevant features.
There are other kinds of generalizations of the TMG that pursue other
purposes. For instance, implications in the AdS3 /CF T2 correspondence. The
construction of a new massive gravity theory, NMG, preserving the parity invari-
ance and possessing a single massive spin 2 field has been achieved by Bergshoeff,
Hohm and Townsend (see Bergshoeff et al., 2009); this theory is constructed
with combinations of quadratic invariants F (R, Rμν Rμν ). The generalization of
this last theory took place with the publication of “minimal massive gravity,”
which has positive energy spin 2 graviton and positive central charges for the
“asymptotic AdS-boundary algebra.” These theories experience nowadays a lot
of activity; many researchers are engaged in their developments, and the list of
references on these topics is quite vast: a Living review concerned these devel-
opments not only in three dimensions but also in higher dimensions, “Massive
gravity,” has been published recently by de Rham (2014).

16.1 Chern–Simons Action and Field Equations of TMG

The Chern–Simons action
   
1 √ 1 2
I= dx3 −g R − 2Λ + λμν Γρ λσ ∂μ Γσ ρν + Γσ μτ Γτ νρ ,
κ μ 3
(16.1)

where κ = 16πG, in the presence of a matter distribution, under variations,

yields the Einstein topological massive gravity (E-TMG) equations
1 1
Eαβ := Rαβ − R gαβ + Λ gαβ + Cαβ = κ Tαβ ,
2 μ
1
Cαβ = α μν ∇μ (Rνβ − R gνβ ), (16.2)
4
where αμν is the Levi–Civita pseudotensor, and Cαβ is the symmetric, Cαβ =
Cβα , traceless, Cαα = 0, Cotton (pseudo-)tensor.
Notice that in the vacuum case the E-TMG equations (16.2) can be
represented equivalently by the system of equations

R = 6 Λ, (16.3a)

1
Sα β + Cα β = 0, (16.3b)
μ
 16.2 Exact Vacuum Solutions of TMG with Λ 305

where Sα β is the traceless Ricci tensor,

R
Sα β := Rα β − gα β , S α α = 0. (16.4)
3
Practically, as was stated previously, in this text we follow, with minor changes,
the CPS reviews, Chow et al. (2010a and 2010b); the cosmological constant
is mostly denoted by its standard notation Λ. Nevertheless, when using the
CPS representations of the metrics reported in their reviews, they are pre-
sented as in the CPS works, where they first kept the same coordinates as in
the originals that appeared in the literature, with possible minimal rearrange-
ments, and later found coordinate transformations that brought the solutions
to a canonical form. In the CPS publications, any cosmological constant only
appears via its parameter m, which has dimensions of mass. Moreover, the leit-
motifs pursued in the original publications are quoted at the beginnings of the
corresponding paragraphs to show the diversity of problems yielding a similar
answer.

16.2 Exact Vacuum Solutions of TMG with Λ

In the CPS reviews on exact solutions to E-TMG equations, namely Chow
et al. (2010a and 2010b), it was established that almost all local metrics reduce
to four (five in our subclassification) particular families of solutions:

1. General triaxially squashed vacuum solutions (Λ = 0)

2. Timelike biaxially squashed AdS3 solutions
3. Spacelike biaxially squashed AdS3 solutions
4. AdS3 non-covariantly and covariantly constant type N wave solutions
5. Kundt dS3 –AdS3 solutions
1. Details in Chapter 17. A systematic search of squashed solutions in TMG
dates back to the works by Nutku and Baekler (1989), and, independently, by
Ortiz (1990), (NBO), dealing with the derivation of homogeneous spacetimes of
various Bianchi types (BT) in vacuum without a cosmological constant – the
so-called triaxially squashed spacetimes, although 3D Bianchi-type VIII spaces
were determined by Vuorio (1985) from a different point of view, and studied in
detail some years before the NBO publications by Percacci, Sodano, and Vuorio;
see Percacci et al. (1987). In the context of Einstein-TMG solutions, one of the
first works, to our knowledge, where the term “squashed space” appeared is the
publication by Nutku and Baekler (1989). Moreover, in this work one finds all
the building material to construct Bianchi spacetimes to vacuum Einstein-TMG
equations; as an introduction to this topic one has to address them; they posed
the problem of finding homogeneous, anisotropic solutions of the Bianchi types
to E-TMG equations.
306 Topologically Massive Gravity

Later Nutku (1993), in the same Bianchi-type VIII framework but now
with a cosmological constant, derived timelike biaxially squashed and spacelike
spacetimes.
All these squashed spacetimes are Petrov type D, and exhibit the constant
scalar invariant (CSI) property, i.e., all scalar polynomial curvature invariants
are constant.
2. Details in Section 17.5. Timelike biaxially squashed AdS3 spacetime can
be regarded as a timelike fibration over H, or as a stationary cyclic symmetric
spacetime decomposed with respect to the timelike Killing vector. This spacetime
is Petrov type D.
3. Details in Section 17.6. Spacelike biaxially squashed AdS3 spacetime can
be regarded as a spacelike fibration over AdS2 , or as a stationary cyclic sym-
metric spacetime decomposed with respect to the spacelike Killing vector. This
spacetime is Petrov type D.
4. Details in Chapter 18. In the spirit of CPS, the AdS3 pp-wave solutions are
generalizations of pp-waves to include a negative cosmological constant; these
solutions are Petrov type N. Nevertheless, we prefer to restrict the name of
pp-waves to those solutions possessing a covariantly constant kμ;ν = 0 null con-
gruence k as they are defined in the standard 4D Einstein gravity. In TMG there
is room for a wider class of solutions with a non-covariantly constant kμ;ν = 0
null congruence k of Cotton type N, to be denoted as TN-wave solutions.
5. Details in Chapter 19. In general, Kundt spacetimes are defined by metrics
admitting a null geodesic vector field that is shear-free, twist-free, and expansion-
free. Therefore, in three dimensions, because the twist and shear trivially vanish,
a “Kundt spacetime is simply one that admits an expansion-free null geodesic
congruence.” These spacetimes allow for Petrov type II, D, III, N, and O solu-
tions. Moreover, all the existing Kundt solutions of TMG are CSI – constant
scalar invariants – spacetimes for which all polynomial scalar (Ricci) curvature
invariants are constant; these late metric structures were studied for the first time
by Coley, Hervik, and Pelavas (see Coley et al., 2006 and Coley et al., 2008) in
three dimensions in a wider context.
Many of these solutions “have been independently rediscovered as solutions of
TMG several times. The literature is rather fragmented, using different coordi-
nate systems that are not obviously related.” Thus, adopting the aim of CPS,
a review establishing the relationships, if any, between families of solutions
becomes pertinent; the metrics are presented in the same coordinates as the orig-
inal literature, although certain rearrangements have been made. Each metric is
then transformed to a canonical form.
 17
Bianchi-Type (BT) Spacetimes in TMG; Petrov
Type D

17.1 Generalities on Bianchi-Type (BT) 3D Spaces

The classification of the Bianchi-type 3D spaces – from I to IX – was accom-
plished more than hundred years ago on the basis of the Lie groups that
characterize them. The classification of homogeneous 3D Bianchi-type spaces
can be found in Stephani et al. (2003), Chapter 13 and 14, Table 13.4.
A quite complete outline of the main concepts needed when studying the sym-
metries of a manifold is detailed in Ortiz (1990), §2: vector field on a manifold,
exponential map, Lie derivative, invariance of the metric tensor, Killing vector
field, Lie group formed by the set of all symmetry generators, commutators of
generators, Lie algebras, structure constants, right invariant vector field ↔ right
invariant 1-form, left invariant vector field ↔ left invariant 1-form, transitive
group, isometry group, homogeneous spacetime, space of left invariant 1-forms,
decomposition of the matrix of structure constants, and classification of 3D Lie
algebras – Bianchi types.
Let the set of r independent Killing vectors {ξA α , A = 1, · · · , r} forms a basis
of a Gr group, then the generators of this symmetry group, defined as
∂
KA = ξA α ,
∂xα
fulfill the commutation rule

[KA , KB ] = −C N A B KN , C N A B = −C N B A ,

and the Jacobi identities,

[KA , [KB , KC ]] + [KB , [KC , KA ]] + [KC , [KA , KB ]] = 0.

Moreover, the quantities C N A B , called constants of structure of the group, fulfill

the Lie identity
C N M [ L C M A B] = 0.
308 Bianchi-Type (BT ) Spacetimes in TMG; Petrov Type D

If by linear transformations T of the r basis Killing operators KA one can convert

one set of structure constants into one another, then the corresponding groups
that they represent are isomorphic. On the contrary, if there is no such transfor-
mation, the groups are non-isomorphic. If a basis of generators is given by the
set {KA }, then a linear combination of them by means of an invertible constant
matrix T A B̃ also determines a basis;
KÃ = T N Ã KN ,
such that the new structure constants obey the following transformation law
C L̃ M̃ Ñ = T A M̃ T B Ñ T L̃ D C D A B ,

where T L̃ D is inverse to T A M̃ .
Returning to the classification of 3D homogeneous spaces, to describe these
spaces one uses the left invariant 1–forms σ α , α = 1, 2, 3 which fulfill the Maurer–
Cartan structure equations,
1 α
dσ α = C βγ σ β ∧ σ γ (17.1)
2
where the structure constants C α βγ = −C α γβ obey the Lie identity
C α β[μ C β νγ] = 0. (17.2)
Each Bianchi type is characterized by a specific set of structure constants,
which in turn determines the corresponding group. The classification of homo-
geneous 3D Bianchi-type spaces can be found also in the textbook by Stephani
(1990).
Following Nutku and Baekler (1989), the triad description of the homogeneous,
anisotropic cosmological models in TMG keep a close parallelism to the Maurer–
Cartan structure description. By means of a linear transformation between the
1-forms σ a and the triad 1-forms ω a
ω a = λa α σ α ,
λa α = λa δ a α no sum in a, λa = constants, (17.3)
one determines the 3D spacetime metric with Lorentzian signature defined
through
ds2 = ηab ω a ω b , ω a = λa σ a , no sum in a,
ηab = diag(−1, 1, 1). (17.4)
From the first Cartan structure equations
dω a + Γa b ∧ ω b = 0, (17.5)
one determines the connection 1-forms, Γa b = Γa bs ω s , which, used in the second
Cartan structure equations
1 a
Θa b := dω a b + ω a s ∧ ω s b = R bcd ω c ∧ ω d , (17.6)
2
 17.2 Nutku–Baekler–Ortiz “Timelike” BT VIII Spacetime 309

gives rise to the Riemann curvature tensor

Rs abc = ∂b Γs ac − ∂c Γs ab + Γv ac Γs bv − Γv ab Γs cv . (17.7)
Because of the constancy of the structure constants C α βμ in the case of 3D
Bianchi-type spacetimes, the structure constants are related according to
Cm ns = λm α λβ n λγ s C α βγ , λm α λα n = δ m n , λm α λβ m = δ β α , (17.8)
thus the connections are
1
Γa bc = (−C a bc + ηbm η an C m cn − ηcm η na C m nb ) (17.9)
2
and consequently in the Riemann tensor components the derivatives of Γs drop
out. Moreover, its contraction yields the Ricci tensor and a subsequent contrac-
tion leads the crucial important formula for the scalar curvature R in terms of
polynomials of the scale factors λa , as explicitly shown by Nutku and Baek-
ler (1989) and Ortiz (1990) (NBO for short), and Nutku (1993). We derive R
in the forthcoming paragraphs for specific metrics. It is clear that the curvature
invariants constructed on constant Riemann tensor are also constant objects, i.e.,
scalar constant invariants. Moreover, the invariants built on covariant derivatives
of the Riemannian curvature are constant scalars too.
Nutku and Baekler (1989) and Ortiz (1990) established that the only Bianchi
types allowed by the E-TMG equations are the Bianchi Type VIII, and III space-
times. These classes are derived here in some detail. In the original papers, differ-
ent Bianchi types were integrated for both Euclidean and Lorentzian signatures.

17.2 Nutku–Baekler–Ortiz “Timelike” BT VIII Spacetime

The first variant of Bianchi type VIII spacetime, the so-called NBO “time-
like” triaxilly squashed BT VIII spacetime, arises for the metric, in coordinates
{ψ, θ, φ}, and orthonormal triad given by
2 2 2
ds2 = −λ20 σ (0) + λ21 σ (1) + λ22 σ (2) = ηab ω (a) ω (b) ,
ω (a) = λa σ (a) , ω (a) = ω (a) α dxα , {λ0 , λ1 , λ2 } = constants,
σ (0) = dψ + sinh θ dφ,
σ (1) = − sin ψdθ + cos ψ cosh θdφ,
σ (2) = cos ψdθ + sin ψ cosh θdφ,
dσ (1) = σ (2) ∧ σ (0) , dσ (2) = σ (0) ∧ σ (1) , dσ (0) = −σ (1) ∧ σ (2) . (17.10)
We denote the triad components enclosed within a parenthesis to avoid confusion
with coordinate components.
The E–TMG equations determine the constraints on the scale factors: among
them, they yield the scalar curvature
(λ1 + λ0 − λ2 ) (λ1 + λ0 + λ2 ) (λ2 − λ1 + λ0 ) (−λ2 − λ1 + λ0 )
R= . (17.11)
2λ0 2 λ2 2 λ1 2
310 Bianchi-Type (BT ) Spacetimes in TMG; Petrov Type D

In vacuum TMG, the vanishing of the curvature scalar R requires the fulfillment
of the relations
λ0 ± λ1 ∓ λ2 = 0 → λ0 + λ1 + λ2 = 0; (17.12)
this last relation can be always adopted because of the triad {ω (a) } is defined up
the signs of λa . The equations E (a) (b) = 0 lead, taking into account the relation
above, to the condition
λ0 2 + λ1 2 + λ2 2 λ1 2 + λ2 λ1 + λ2 2
μ=− =2 . (17.13)
λ0 λ1 λ2 λ1 λ2 (λ1 + λ2 )
These Bianchi type VIII solutions are called triaxially squashed spacetimes of
the TMG theory; it is apparent that they are unique to their class; they were
derived in Nutku and Baekler (1989), Eq. (4.1), and Ortiz (1990) (type(a) with
a = 0). They are characterized by a Petrov type D Cotton tensor: its eigenvectors
are given by
L1 = −L+ λ1 ; VI α = ω (1) α , spacelike,
L2 = −L+ λ2 ; VII α = ω (2) α , spacelike,
L0 = −L+ λ0 ; VIII α = ω (0) α , timelike, (17.14)
where the positive L+ is given by
μ λ0 2 + λ1 2 + λ2 2 λ1 2 + λ2 λ1 + λ2 2
L+ = −2 =2 = 4 . (17.15)
λ0 λ1 λ2 λ0 2 λ2 2 λ1 2 2
(λ1 + λ2 ) λ2 2 λ1 2
In terms of its eigenvectors, the Cotton tensor is expressed as
Cα β = L+ λ0 ω (0) α ω (0) β − L+ λ1 ω (1) α ω (1) β − L+ λ2 ω (2) α ω (2) β , (17.16)
hence, its constant triad components C(a) (b) = ω α (a) Cα β ω β (b) are
C(a) (b) = L+ λ0 δ (0) (a) δ (0) (b) − L+ λ1 δ (1) (a) δ (1) (b) − L+ λ2 δ (2) (a) δ (2) (b) . (17.17)

17.2.1 Nutku Timelike Biaxially Squashed Metric

Nutku extended the results of NBO to the presence of a cosmological constant
Λ. In such case the scalar curvature is equal to 6Λ, R = 6 Λ; in Nutku (1993),
ΛN utku = −Λstandard = −Λ. Now one leaves (17.12), and instead requires λi =
±λj , i = j. In particular, without loss of generality one may set
λ2 = λ1 , (17.18)
which, substituted into E α β , yields
E ψ φ = 2 μ λ1 2 λ0 2 − 2 μ λ1 4 + 3 λ0 3 − 3 λ0 λ1 2 = 0. (17.19)
Solving for λ0 one obtains
2
λ 0 = − μ λ1 2 , (17.20)
3
 17.3 Nutku–Baekler–Ortiz “Spacelike” Squashed BT VIII Spacetime 311

together with the trivial solutions λ0 = ±λ1 . The substitution of λ2 = λ1 into

the scalar curvature R = 6Λ leads to

(λ0 − 2 λ1 ) (λ0 + 2 λ1 ) − 12 λ1 4 Λ = 0, (17.21)

which, substituting λ0 from (17.20), gives rise to the equations

1 μ
μ2 λ1 2 − 9 − 27 Λ λ1 2 = 0 → λ1 = 3 , λ0 = −6 . (17.22)
μ2 − 27 Λ μ2 − 27 Λ
Finally, this Nutku (1993) solution, Eq. (18), can be given by the metric
 
9 4μ2
2
ds = 2 − 2 2 2 2 2
(dψ + sinh θdφ) + dθ + cosh θdφ , (17.23)
μ − 27Λ μ − 27Λ
which is characterized by a Petrov type D Cotton tensor
⎡ ⎤
−1 0 0
⎢ ⎥ 1  2 
(C α β ) = L ⎢ ⎥
⎣ 0 2 3 sinh θ ⎦ , L := 9 μ μ + 9 Λ . (17.24)
0 0 −1

Vuorio Timelike Biaxially Squashed Metric

Moreover, it contains the timelike biaxially squashed solution in the case of
vanishing cosmological constant. To derive this subclass of solutions from the
vanishing condition of the scalar curvature (17.11) one uses here the relation
λ2 = λ1 , thus
R(λ2 = λ1 ) = 0 → λ0 = ∓ 2λ1 . (17.25)

Replacing these λs in equation (17.13) with μ, one arrives at

3 3
μ=± → λ1 = ± , (17.26)
λ1 μ
which determines a metric of the form (17.23)
9
ds2 = −4 (dψ + sinh θdφ)2 + dθ2 + cosh2 θdφ2 , (17.27)
μ2
for Λ = 0 known as the Vuorio (1985) solution.

17.3 Nutku–Baekler–Ortiz “Spacelike” Squashed BT VIII Spacetime

In this paragraph, a second branch of Bianchi type VIII solutions is derived
– the NBO “spacelike” triaxilly squashed BT VIII spacetime: the metric and
orthonormal triad used in the coordinates {ψ, θ, φ} are
2 2 2
ds2 = −λ20 σ (0) + λ21 σ (1) + λ22 σ (2) = ηab ω (a) ω (b) , ω (a) = λa σ (a) ,
σ (0) = cosh ψdθ + sinh ψ cos θdφ,
312 Bianchi-Type (BT ) Spacetimes in TMG; Petrov Type D

σ (1) = sinh ψdθ + cosh ψ cos θdφ,

σ (2) = dψ + sin θ dφ,
{λ0 , λ1 , λ2 } = constants. (17.28)

where the left invariant 1-forms σ (a) fulfill the Maurer–Cartan equations

dσ (1) = σ (2) ∧ σ (0) , dσ (2) = σ (0) ∧ σ (1) , dσ (0) = −σ (1) ∧ σ (2) .

It should be pointed out that this metric structure (17.28) corresponds to

another real cut of the metric (17.10), when considered as a complex one: making
there the complex transformations and identifications

θ → iθ, ψ → iψ, φ → φ, σ (0) → iσ (2) , σ (1) → σ (1) , σ (2) → iσ (0) ,

ω (0) → iω (2) , ω (1) → ω (1) , ω (2) → iω (0) , λ0 → λ2 , λ1 → λ1 , λ2 → λ0 ,

one arrives at the metric structure (17.28).

The E-TMG equations, among others, yield the scalar curvature
(λ0 + λ1 + λ2 ) (λ0 + λ1 − λ2 ) (−λ2 + λ0 − λ1 ) (λ2 + λ0 − λ1 )
R= . (17.29)
2λ2 2 λ0 2 λ1 2
In vacuum TMG, the vanishing of the curvature scalar R requires the fulfillment
of the relations λ0 ± λ1 ∓ λ2 = 0, which one fixes to be

λ0 + λ1 + λ2 = 0, (17.30)

because the triad is defined up the signs of λa . The equation E ψ ψ = 0 leads,

taking into account the relation above, to the coupling constant
λ0 2 + λ 1 2 + λ 2 2 λ1 2 + λ1 λ2 + λ2 2
μ= = −2 . (17.31)
λ0 λ1 λ2 λ1 λ2 (λ1 + λ2 )
These λs (17.30) and μ (17.31) determine a second family of Bianchi type VIII
solutions, called triaxilly squashed spacetimes of the TMG theory; it is clear
that they are unique. They are characterized by a Petrov type D Cotton tensor.
Searching for its eigenvectors one solves the equations

C α β V β A − L VA α = 0, det(C α β − L δ α β ) = 0,

which lead to

L1 = −λ1 L− , VI α = ω (1) α , spacelike,

L2 = −λ2 L− , VII α = ω (2) α , spacelike,
L0 = −λ0 L− , V0 α = ω (0) α , timelike, (17.32)

where the negative constant L− is equal to

μ λ2 2 + λ2 λ1 + λ1 2
L− = −2 = −4 2 2 . (17.33)
λ1 λ2 λ0 λ2 (λ2 + λ1 ) λ1 2
 17.3 Nutku–Baekler–Ortiz “Spacelike” Squashed BT VIII Spacetime 313

They allow to express Cα β as

Cα β = λ0 L− ω (0) α ω (0) β − λ1 L− ω (1) α ω (1) β − λ2 L− ω (2) α ω (2) β ; (17.34)

hence, its constant triad components C(a) (b) = ω α (a) Cα β ω β (b) are

C(a) (b) = λ0 L− δ (0) (a) δ (0) (b) − λ1 L− δ (1) (a) δ (1) (b) − λ2 L− δ (2) (a) δ (2) (b) . (17.35)

17.3.1 Spacelike Biaxially Squashed Metric; Nutku Solution

Counterpart
Nutku (1993) extended the results of NBO to the presence of a cosmological con-
stant Λ and derived the timelike biaxially squashed metric (17.23), nevertheless
no mention is made of this second spacelike possibility, although in the publica-
tion by Nutku and Baekler (1989) the metric structure (17.28) is reported under
the number (4.6). For that reason we propose to call this second branch “Nutku
solution counterpart”.
In the presence of a cosmological constant Λ, the scalar curvature fulfills R =
6 Λ; now one has to abandon (17.30) and require λi = ±λj , i = j. Without loss
of generality, one may demand
λ0 = λ1 (17.36)

which, substituted into E α β , yields the component E ψ φ of the form


3 λ2
E φ : 3 λ 1 λ 2 − 3 λ 2 − 2 μ λ2 λ 1 + 2 μ λ1 = 0 → λ 1 = −
ψ 2 3 2 2 4
. (17.37)
2μ

Its substitution into the remaining TMG equations leads to a single constraint,
namely
μ 1
λ2 μ2 − 27 λ2 Λ + 6 μ = 0 → λ2 = −6 , λ1 = 3 , (17.38)
μ2 − 27 Λ μ2 − 27 Λ
hence, the Nutku solution counterpart can be given by the metric
 
9 4μ2
ds2 = 2 −dθ2 + cos2 θdφ2 + 2 (dψ + sin θdφ)2 . (17.39)
μ − 27Λ μ − 27Λ
Although this solution is not explicitly given in Nutku (1993), it can straightfor-
wardly be derived by complex coordinate transformations from Nutku (1993),
Eq. (18). This solution is characterized by a Petrov type D Cotton tensor.

Spacelike Biaxially Squashed Vuorio Solution Counterpart

Moreover, it contains the biaxially squashed spacelike solution in the case of
vanishing cosmological constant, i.e., the metric (17.39) for Λ = 0. For complete-
ness, this subclass of solutions is derived from the vanishing condition of the
314 Bianchi-Type (BT ) Spacetimes in TMG; Petrov Type D

scalar curvature (17.29); as the first step one uses the equality λ2 = λ1 , next its
substitution in R gives

R(λ2 = λ1 ) = 0 → λ0 = ∓ 2λ1 , (17.40)

finally, replacing these λs in the equation (17.31) for μ one arrives at

3 3
μ=± → λ1 = ± , (17.41)
λ1 μ
which determines a metric of the form (17.39) for Λ = 0,
9
ds2 = −dθ2 + cos2 θdφ2 + 4 (dψ + sin θdφ)2 , (17.42)
μ2
which can be named “Vuorio solution counterpart.”

17.4 Nutku–Baekler–Ortiz Solutions of Bianchi Type III

Another branch of solutions, see Nutku and Baekler (1989), Eq. (4.10), and Ortiz
(1990) (type(a) with a = 0), is derived using the Bianchi type III left 1-forms
from the metric
2 2 2
ds2 = −λ21 σ (1) + λ22 σ (2) + λ23 σ (3) = ηab ω (a) ω (b) , ω (a) = λa σ (a) ,
σ (1) = eα θ (cosh θ dx + sinh θ dy),
σ (2) = eα θ (sinh θ dx + cosh θ dy),
σ (3) = dθ, (17.43)

where σ (a) are Bianchi III left invariant 1-forms that satisfy

dσ (1) = ασ (3) ∧ σ (1) + σ (3) ∧ σ (2) , dσ (2) = ασ (3) ∧ σ (2) + σ (3) ∧ σ (1) ,
dσ (3) = 0, C (1) (3)(1) = α = C (2) (3)(2) , C (1) (3)(2) = 1 = C (2) (3)(1) , (17.44)

which corrects the relations in the text line following Eqs. (A.28) and (A.56)
of Chow et al. (2010a), denoted for short CPSa. The scalar curvature for the
vacuum TMG is given by
2 λ1 2 λ2 2 + 12 λ1 2 α2 λ2 2 − λ1 4 − λ2 4
R=− =0
2 λ1 2 λ2 2 λ3 2
from which one isolates α2
2 2
1 (λ1 − λ2 ) (λ1 + λ2 )
α2 =
12 λ1 2 λ2 2
with two possible branches
√
1 3 (λ1 + λ2 ) (λ1 − λ2 )
α=± .
6 λ2 λ1
 17.4 Nutku–Baekler–Ortiz Solutions of Bianchi Type III 315

Nevertheless, one can always choose the sign of α to be positive, α = |α| ; if α

were negative, α = −|α|, then by changing θ → −θ, y → −y one gets the case
with positive α. From the equation E θ θ one obtains the expression of μ

3 λ1 2 + λ2 2
μ= .
2 λ3 λ1 λ2
Substituting μ into E z y one gets

15 λ1 2 λ2 4 − 15 λ1 4 λ2 2 − λ2 6 + λ1 6 = 0,

with solutions
√
λ1 = 2 + 3 λ2 , (17.45a)

or
√
λ1 = 2 − 3 λ2 . (17.45b)

Evaluating α and μ one gets

6 6
α = ±1 → α = 1, μ = → λ3 = . (17.46)
λ3 μ
Therefore, one distinguishes only two possible branches of Bianchi type III
solutions without a cosmological constant.

17.4.1 Nutku–Baekler–Ortiz BT III Timelike Solution with Λ = 0

Part of the name used above has been borrowed from CPSa to designate in a
definite manner this class of solutions; in a similar way we proceed to denote the
second spacelike branch of solutions. For the choice
√ 6
α = 1, λ1 = 2 + 3 λ, λ2 = λ, λ3 = , (17.47a)
μ

one obtains the metric given in Chow et al. (2010a) as (A.28), namely
 √ 2 2
 36 2 36
ds2 = λ2 −(2 + 3)2 σ (1) + σ (2) + 2 σ (3) = 2 dθ2
μ μ
 √ 
+ λ2 e2θ −(2 + 3)2 (cosh θ dx + sinh θ dy)2 + (sinh θ dx + cosh θ dy)2
√
3 + 2 3 2 √ 2
=− λ 3(dx − dy) + 2e2θ (dx + dy)
6
√
3 + 2 3 2 4θ 36
+ λ e (dx + dy)2 + 2 dθ2 . (17.47b)
6 μ
316 Bianchi-Type (BT ) Spacetimes in TMG; Petrov Type D

Making the coordinate transformations CPSa(A.29),

√ √
3+2 3 μ 3+2 3 μ
√ λ(x − y) → τ, √ λ(x + y) → x , e−2θ → y  , (17.47c)
6 2 3 6
dropping primes, one gets
 
9 dx 2 dx2 + dy 2
ds2tl = 2 −λ (dτ +
2
) + , (17.47d)
μ y y2

which coincides with the solution given by Eq. (A.11) of Chow et al. (2010a)
with Λ = 0, CPSa(A.11).

17.4.2 Nutku–Baekler–Ortiz BT III Spacelike Solution with Λ = 0

On the other hand, for the second available choice
√ 6
α = 1, λ1 = 2 − 3 λ, λ2 = λ, λ3 = , (17.48a)
μ

one obtains the solution given by

 √ 2 2
 36 2 36
ds2 = λ2 −(2 − 3)2 σ (1) + σ (2) + 2 σ (3) = 2 dθ2
μ μ
 √ 2 
+ λ e −(2 − 3) (cosh θ dx + sinh θ dy) + (sinh θ dx + cosh θ dy)2
2 2θ 2

√
2 3 − 3 2 √ 2
= λ 3(dx − dy) − 2e2θ (dx + dy)
6
√
2 3 − 3 2 4θ 36
− λ e (dx + dy)2 + 2 dθ2 , (17.48b)
6 μ

coinciding with CPSa (A.56). Making the coordinate transformations CPSa-

(A.57),
√ √
2 3−3 μ 2 3−3 μ
√ √ λ(x + y) → t, √ λ(x − y) → z, e−2θ → x, (17.48c)
3 3 2 6 2
one arrives at Λ = 0.
 
9 dt 2 dx2 − dt2
ds2sl 2
= 2 λ (dz + ) + , (17.49)
μ x x2

which corresponds to (A.46) with Λ = 0 of Chow et al. (2010a).

All these homogeneous Bianchi type VIII and III solutions – the unique three
dimensional spacetimes of this kind in TMG – are presented in a succinct manner
in the following table:
 17.5 Timelike Biaxially Squashed Metrics 317

Table 17.4.1 Bianchi Type (BT) VIII and BT III TMG solutions

References Bianchi T Eqs.text Eqs. in Ref. Rel. to CPSa

Nutku and Baekler (1989) TriAtl:(17.10) (4.1) (A.27)

BT VIII TriAsl:(17.28) (4.6) (A.54)
Ortiz (1990) TriAtl:(17.10) (a = 0) (A.28)
BT VIII TriAsl:(17.28) ↓ (5.3) (A.56)
Nutku and Baekler (1989) sl:(17.47b) (A.27)
BT III tl:(17.48b) (4.10) (A.27)
Ortiz (1990) sl:(17.47b) (a = 0) (A.28)
BT III tl:(17.48b) ↓ (5.5) (A.28)
Nutku (1993); Λ BiA sl: (17.39) (A.48)
BT VIII; Λ BiA tl: (17.23) (18) (A.12)
CPSa; Chow et al. (2010a) 17.5.4
tl – timelike, sl – spacelike TriA – triaxial BiA – biaxial T – type

17.5 Timelike Biaxially Squashed Metrics

17.5.1 Representation of the Vacuum Biaxially Squashed Solutions
Most of the known exact solutions in TMG belong to the biaxially squashed –
timelike or spacelike – homogeneous anisotropic Bianchi type VIII families of
spacetimes. Fortunately, the identification and sorting of the known solutions
in the timelike and spacelike classes have been accomplished by the familiar
Chow, Pope and Sezgin; see CPSa: Chow et al. (2010a), who made a complete
classification of the existing solutions. Moreover, they classify in that work the
pp-wave solutions too.
In (2 + 1)-dimensional gravity any stationary cyclic symmetric metric can be
given as
g = gtt dt2 + 2gtφ dt dφ + gφφ dφ2 + grr dr2 , (17.50)
where gαβ depend on the coordinate r, gαβ (r).
Without loss of generality one can choose a “spacelike” representation of the
stationary cyclic symmetric (2 + 1) metric developed with respect to the cyclic
symmetry ∂φ ,
 
√ √ gt φ
mα = δ α φ / gφ φ , m = gφ φ dφ + dt (17.51)
gφ φ
obtaining
 2
gt φ gt2φ − gtt gφ φ 2
g = gφ φ dφ + dt − dt + grr dr2 , (17.52)
gφ φ gφ φ
or, in a more familiar way,
F (r) 2 dr2 2
g=− dt + + H(r) [dφ + W (r)dt] . (17.53)
H(r) F (r)
318 Bianchi-Type (BT ) Spacetimes in TMG; Petrov Type D

On the other hand, if one chooses the stationary symmetry ∂t

 
gt φ
k = δ t / |gt t |, k = |gt t | d t +
α α
dφ , (17.54)
gt t
as the fundamental Killing field, the stationary cyclic symmetric (2 + 1) metric
can be written in the “timelike” representation
 2
gt φ gtt gφ φ − gt2φ 2
g = −|gt t | d t + dφ + dφ + grr dr2 , (17.55)
gt t gt t
or in the more standard way,
F (r) 2 dr2
g=− [dt − ω(r) dφ] + h(r) dφ2 + . (17.56)
h(r) F (r)
Vuorio (1985), one of the first researchers looking for exact solutions in TMG,
used the stationary “rotationally” symmetric ansatz for the metric, i.e., the
metric (17.56) up to coordinate redefinition.
Although the general timelike biaxially squashed Nutku solution has been
derived in the previous paragraph 17.2.1, it is enlightening to exhibit its deriva-
tion from the point of view of the above-detailed stationary cyclic symmetric
metric structure. Consider the middle of the road metric element in coordinates
{φ, θ, τ }
λ2 λ1
ds2 = − (dτ + cosh θ dφ)2 + (dθ2 + sinh2 θ dφ2 ), (17.57)
4 4
where the scale factors λ1 and λ2 are constants, and seek for solutions to the
Einstein-TMG equations (16.2). The relevant E-TMG equations amount to:
Eφτ : (3 λ2 3/2 − μ λ1 λ2 ) (λ2 − λ1 ) = 0 → μ λ1 = 3 λ2 , (17.58)

substituting λ1 into the equation Eφφ = Eθθ ,

 
Eφφ : − λ2 + Λ λ1 2 μ λ1 + 4 λ2 (λ2 − λ1 ) = 0, (17.59)
one arrives at
 2  μ2 9
μ − 27 Λ λ2 = 12 μ → λ2 = 144 =4 λ2 ,
(μ2 − 27 Λ)
2 μ2 − 27 Λ
2
4μ 36
λ2 := , λ1 = 2 . (17.60)
μ2 − 27 Λ μ − 27 Λ
To obtain these solutions for λs one could use the equation arising from the van-
ishing of the contraction E α α = 0 of the vacuum T α β = 0 E–TMG equations:
4 λ1 + 3 Λ λ1 2 = λ2 .
The timelike biaxially squashed dS–AdS solution can be given by the metric
9
ds2tl = −λ2 (dτ + cosh θ dφ)2 + dθ2 + sinh2 θdφ2 ,
μ2 − 27Λ
4μ2
λ2 := 2 . (17.61)
μ − 27 Λ
 17.5 Timelike Biaxially Squashed Metrics 319

It is characterized by the constant scalar curvature

R = 6Λ, (17.62)
the Cotton tensor
μ(μ2 + 9Λ) α θ
C αβ = − δ θ δ β + δ α φ δ φ β − 2δ α τ δ τ β − 3 cosh θ δ α τ δ φ β , (17.63)
9
and the traceless Ricci tensor Sμν = Rμν − 13 R gμν , which, as it should be, for
all vacuum E-TMG solutions plus Λ the relation between the Cotton and the
traceless Ricci tensor
1
Sαβ + C αβ = 0
μ
holds. These tensors allow for an alternative representation through the metric
tensor and a unit vector, say K α , of the form
μ2 + 9Λ α
Sβα = (δ β + 3 K α Kβ ) (17.64)
9
where the unit timelike vector K α , is given by
μ2 − 27 Λ α 6μ 6μ
Kα = δ τ , Kα = − 2 cosh θδα φ − 2 δα τ . (17.65)
6μ μ − 27 Λ μ − 27 Λ

17.5.2 Eigenvectors of the Cotton Tensor; Triad Formulation

This paragraph is devoted to the search of the eigenvectors of the Cotton tensor,
and, consequently, of the traceless Ricci tensor because of (17.5.1). Using the
orthonormal triad 1-forms
Θ(a) = Θ(a) α dxα ≡ h(a) α dxα , a = 1, 2, 3, α = θ, φ, τ,
the metric is given as
2 2 2
ds2 = g(a)(b) Θ(a) Θ(b) = Θ(1) + Θ(2) − Θ(3) ,
   
where the constant triad metric g(a)(b) = diag(1, 1 − 1) = g (a)(b) .
For the timelike squashed metric (17.61) under consideration, the basis 1-forms
are
3 3 sinh θ
Θ(1) = dθ, Θ(2) = dφ,
μ − 27 Λ
2 μ2 − 27 Λ
6μ cosh θ 6μ
Θ(3) = 2 dφ + 2 dτ, (17.66)
μ − 27 Λ μ − 27 Λ
therefore the matrix of the tensor h(a) α , Θ(a) α = h(a) α dxα , amounts to
⎡ ⎤
0 3 √ 21 0
μ −27 Λ
⎢ ⎥
⎢ ⎥
(a)
h α =⎢ 3 ⎢ √ sinh θ
0 0 ⎥, (17.67)
μ2 −27 Λ ⎥
⎣ ⎦
6 μμ2cosh
−27 Λ
θ
0 μ
6 μ2 −27 Λ
320 Bianchi-Type (BT ) Spacetimes in TMG; Petrov Type D

with inverse matrix h(a) α , ∂(a) = ∂(a) α ∂ ∂xα = h(a) α ∂ ∂xα , given by
⎡ √ 2 ⎤
μ −27 Λ
0 0
⎢ 3 ⎥
  ⎢ √ 2 √ 2 ⎥
h(a) = ⎢
α μ −27 Λ
⎢ 3 sinh θ 0 −
μ −27 Λ cosh θ ⎥ .
⎥ (17.68)
⎣ 3 sinh θ ⎦
μ −27 Λ
2
0 0 6μ

These tensors are used to construct the triad tensor (constant) components. In
the case of the Cotton tensor given by coordinate components (17.63), using

C (a) (b) = h(a) α C α β h(b) β , (17.69)

one gets its triad components

⎡ ⎤
−1 0 0
μ  ⎢ ⎥
C (a) (b) = 9 Λ + μ2 ⎢
⎣ 0 −1 0 ⎥
⎦, (17.70)
9
0 0 2

from which it becomes

 apparent which are the eigenvalues of the Cotton tensor,
λS = 29 μ μ2 + 9 Λ = −2λD , and its Petrov type D character.
Accordingly, one may write the triad components of the Cotton tensor in a
very simple way
(1) (1) (2) (2) (3) (3)
C(a)(b) = λD δ(a) δ(b) + λD δ(a) δ(b) − λS δ(a) δ(b) , (17.71)

hence its coordinate components Cαβ = C(a)(b) Θ(a) α Θ(b) β become

Cαβ = λD Θ(1) α Θ(1) β + λD Θ(2) α Θ(2) β − λS Θ(3) α Θ(3) β . (17.72)

Proceeding in the standard way in the search of the eigenvectors of the Cotton
tensor certainly one should arrive at the same result.

17.5.3 Complex Extension Toward the Spacelike Squashed Metric

Accomplishing in the metric (17.61) the transformation

cosh θ = x → θ = arccoshx

one gets the timelike metric in the form

 
9 dx2
2
dstl = 2 −λ (dτ + x dφ) + 2
2 2
+ (x − 1)dφ .
2 2
(17.73)
μ − 27Λ x −1
From this metric (17.73), by “complex transformations”

x = i y, τ = i z, , φ = t, (17.74)
 17.5 Timelike Biaxially Squashed Metrics 321

one arrives at the spacelike biaxially squashed metric

 
9 dy 2
2
dssl = 2 2 2
λ (dz + y d t) + 2 − (y + 1)d t .
2 2
(17.75)
μ − 27Λ y +1
Moreover, the transformation
y = sinh ρ
brings the above metric to the spacelike biaxially squashed form
9
ds2sl = λ2 (dz + sinh ρ dt)2 + dρ2 − cosh2 ρ dt2 . (17.76)
μ2 − 27Λ
From the complex point of view, these timelike and spacelike biaxially squashed
metric structures can be considered as two real cuts of a complex metric. There-
fore, from this perspective, any solution of a specific class, should it be timelike
or spacelike, would have a counterpart. This procedure is well known in Maxwell
electrodynamics: from electric solutions one gets their magnetic counterpart via
complex relationships.
At this stage, it is worth noticing the real timelike cut arising from (17.76) by
means of
z → iτ , t → iφ, (17.77)
namely
9
ds2tl = −λ2 (d τ + sinh ρ dφ)2 + dρ2 + cosh2 ρ dφ2 . (17.78)
μ2 − 27Λ
Similarly, from (17.61), via
τ → iz, φ → it, θ → ρ (17.79)
one gets another spacelike squashed metric representation
9
ds2sl = λ2 (dτ + cosh ρ dt)2 + dρ2 − sinh2 ρ dt2 . (17.80)
μ2 − 27Λ

17.5.4 Alternative Metric Representation of dstl 2

To present in “canonical” coordinates the timelike biaxially squashed metrics,
even when this family is reducible to a single solution of TMG, there are given
various forms of the spacelike 2D metric. Following Chow et al. (2010a), these
representations are related to the coordinates used in the 2D “metric ds22 of the
hyperbolic space H2 with squared radius L2 = 9/(μ2 − 27Λ2 ), L > 0, which is
the upper leaf X 0 ≥ L of the hyperboloid
−(X 0 )2 + (X 1 )2 + (X 2 )2 = −L2
in the 3D flat spacetime with metric
ds2 = −(dX 0 )2 + (dX 1 )2 + (dX 2 )2 . CPSa(A.2).
322 Bianchi-Type (BT ) Spacetimes in TMG; Petrov Type D

There are several coordinate systems to be chosen:

For the choice CPSa(A.4):

X 0 = L cosh θ, X 1 = L sinh θ cos φ, X 1 = L sinh θ sin φ, (17.81a)

one gets H2 in polar coordinates, CPSa(A.5):

 
ds2 2 = L2 dθ2 + sinh2 θdφ2 , (17.81b)

which, by means of the transformation:

cosh θ = Ar2 + B, φ → φ/(2AL2 ),

can be brought to H2 in polynomial representation CPSa(A.13),

4L2 A2 r2 (Ar2 + B)2 − 1 2
ds2 2 = dr2 + dφ . (17.81c)
(Ar + B) − 1
2 2 4L2 A2
For the choice CPSa(A.6):
y L2 + x2 x y L2 − x2
X0 = + , X1 = L , X2 = − + , (17.81d)
2 2y y 2 2y
one gets H2 in Poincaré coordinates, CPSa(A.7):
L2  2 
ds2 2 = 2
dx + dy 2 , (17.81e)
y
which, by means of the transformation:
φ 1
x= 2
,y= 2
2AL (Ar + B)
can be brought to the polynomial representation CPSa(A.14)
4L2 A2 r2 (Ar2 + B)2 2
ds2 2 = dr 2
+ dφ . (17.81f)
(Ar2 + B)2 4L2 A2
The choice CPSa(A.8):

X 0 = L cosh θ cosh φ, X 1 = L sinh θ, X 2 = L cosh θ sinh φ, (17.81g)

leads to H2 in polar coordinates CPSa(A.9):

 
ds2 2 = L2 dθ2 + cosh2 θdφ2 , (17.81h)

which, by means of the transformation:

sinh θ = Ar2 + B, φ → φ/(2AL2 ),

can be brought to the polynomial representation CPSa(A.15)

4L2 A2 r2 (Ar2 + B)2 + 1 2
ds2 2 = 2 2
dr2 + dφ . (17.81i)
(Ar + B) + 1 4L2 A2
 17.5 Timelike Biaxially Squashed Metrics 323

In Chow et al. (2010a), to facilitate comparison with the solutions reported in

the literature, the timelike squashed solution is represented in various alternative
metric forms, correspondingly CPSa(A.10), CPSa(A.11), and CPSa(A.12) for

4μ2
λ2 := , (17.82a)
μ2 − 27Λ
explicitly
9
ds2tl = −λ2 (dτ + cosh θ dφ)2 + dθ2 + sinh2 θdφ2 , (17.82b)
μ2 − 27Λ
 
9 dx 2 dx2 + dy 2
ds2tl = 2 −λ2 (dτ + ) + , (17.82c)
μ − 27Λ y y2
9
ds2tl = 2 −λ2 (dτ + sinh θ dφ)2 + dθ2 + cosh2 θdφ2 . (17.82d)
μ − 27Λ
Another possibility is achieved by transformations of the type cosh = Ar2 + B,
which yields CPSa(A.16), namely

μ 2 r2
ds2tl = −(dt − r dφ)2 + dr2 + F (r)dφ2 ,
3 F (r)
1 2
F (r) = (μ − 27Λ)r4 + k1 r2 + k0 . (17.82e)
36
In the forthcoming subsections the identification of the solutions reported by
different authors is carried out following the pattern given in Chow et al. (2010a),
citing their equation for reference.

Vuorio Solution; Λ = 0
Vuorio (1985) reported the first stationary rotationally symmetric solu-
tion (2.21), CPSa(A.25), to the E-TMG equations, namely:
9  2

ds2 = 2 − (dt + 2(1 − cosh σ) dθ) + dσ 2 + sinh2 σ dθ2 . (17.83)
μ
Carrying out the coordinates transformations and constants redefinitions

t → 2τ + 2φ, θ → −φ, σ → θ, Λ → 0,

one brings the above solution to CPSa(A.10) with Λ = 0.

Percacci–Sodano–Vuorio Solution; Λ = 0
Percacci et al. (1987) considered stationary solutions possessing a timelike Killing
vector with constant scalar twist; their solution (3.20), CPSa(A.26), is given by
 2
2 1 1 1
ds2 = −3 dx2 + e(μ x /3) dx0 + (d x1 )2 + e(2 μ x /3) (d x0 )2 . (17.84a)
3 3
324 Bianchi-Type (BT ) Spacetimes in TMG; Petrov Type D

Doing the coordinates transformations

μ μ μ
√ x2 → τ, √ x0 → x, exp − x1 → y, (17.84b)
2 3 3 3 3
one arrives at CPSa(A.11) with Λ = 0.

Nutku–Baekler and Ortiz BT III Timelike Solutions; Λ = 0

Nutku and Baekler (1989), and Ortiz (1990) determined solutions formed from
left-invariant 1-forms of Bianchi type III spaces. A detailed derivation of them
is given here in 17.4, 17.4.1. The metric of these solutions – (4.10) of Nutku and
Baekler (1989) and (type(a) with a = 0) of Ortiz (1990) – is given by (17.47b);
it can be reduced to CPSa(A.11) with Λ = 0.

Clément Solutions; Λ = 0
Clément (1992a), and Clément (1992b) too, considered stationary rotationally
symmetric solutions. His solution (4.4), CPSa(A.30), is given by the metric
 μ 2 μ
ds2 = − dt + 2cdθ − 2c cosh( r)dθ + dr2 + c2 sinh2 ( r) dθ2 . (17.85a)
3 3
Making the coordinates transformations and constants redefinitions
μ μ μc
(t + 2c θ) → τ, r → θ  , − θ → φ, (17.85b)
6 3 3
dropping prime, one arrives at (17.82b), CPSaA(24), with Λ = 0.
The Clément’s solution (4.5), CPSa(A.31),
 μ 2 μ
ds2 = − dt − 2c sinh( r)dθ + dr2 + c2 cosh2 ( r) dθ2 , (17.86a)
3 3
subjected to the coordinates transformations
μ μ μc
t → τ, r → θ  , − θ → φ, (17.86b)
6 3 3
dropping prime, gives rise to (17.82d), CPSa (A.12), for Λ = 0.
On the other hand, the solution (4.7) by Clément (1992a) has various possible
choices of signs:
Two choices give
c 2
ds2 = − dt ∓ 2d e±μ r/3 dθ + dr2 + c d e±2μ r/3 dθ2 , (17.87a)
d
which is CPSa(A.32). Subjecting (17.87a) to the coordinates transformations
CPSa(A.32)

c μ√
t → τ, ∓ c dθ → x, e∓ μ r/3 → y, (17.87b)
d 3
one arrives at (17.82c), CPSa(A.11), with Λ = 0.
 17.5 Timelike Biaxially Squashed Metrics 325

Another two choices of signs lead to

 2
2 ±μ r/3 c ±2μ r/3 2
ds2 = −3 c d dθ ± e dt + dr2 + e dt , (17.88a)
3d 3d
see CPSa(A.33). Making the coordinates transformations

√ μ c
3c d θ → τ, ± t → x, e∓ μ r/3 → y, (17.88b)
3 3d
one arrives at (17.82c), CPSa(A.11), with Λ = 0.

Nutku Solutions
Nutku (1993) published a black hole solution (25), CPSa(A.21), to the E–TMG
equations given by
 2
2J − M 2μ r2 − 3J/μ r2 6
ds2 = − dt − dθ + dr2 + F (r)dθ2 ,
6 2J − M F (r) 2J − M
1 2 1 1 J2
F (r) = (μ − 27Λ)r4 − M r2 + . (17.89a)
36 6 4 μ2
Making the coordinates transformations CPSa(A.22)
  √
2J − M 3 1 J  6
t+ √ θ→t, √ θ → φ,
6 2 2J − M μ 2J − M
M 1 J2
− → k1 , → k0 , (17.89b)
6 4 μ2
dropping prime, one arrives at the solution (17.82e), CPSa(A.16).
Moreover, Nutku generalized the Λ zero Vuorio (1985) solution (see 17.5.4)
to include the cosmological constant; the Nutku (1993) solution (3.11), after
relabelling ψN → τ corresponds to (17.82d), CPSa(A.12). On the other hand,
Nutku succeeded in deriving his solution (3.17), which corresponds to (17.82e),
CPSa(A.16), for k1 = 1 and k0 = 0 and θN → φ.

Gürses “Gödel Type” Solution

Gürses (2008) considered “solutions that are of Gödel type, which are defined by
his equations: (42), (43), (44), (46), and part of the unlabelled equation following
(39),” as discussed, (A.17), in Chow et al. (2010a), namely
√ 2 e2 1
ds2 = − [ a0 dt + u2 (r, θ)dθ + u1 (r, θ)dr] + 0 r2 ψ(r)dθ2 + dr2 ,
a0 ψ(r)
μ2 − 27Λ 2 b1
ψ(r) = r + b0 + 2 . (17.90a)
36 r
CPSa(A.18), where u1 and u2 are constrained by the equation CPSa(A.19)
∂u1 ∂u2 2 μ e0
= + √ r, (17.90b)
∂θ ∂r 3 a0
326 Bianchi-Type (BT ) Spacetimes in TMG; Petrov Type D

and hence of the form CPSa(A.19),

∂ 1 μ e0 ∂
u1 = U (r, θ), u2 = − √ r2 + U (r, θ). (17.90c)
∂r 3 a0 ∂θ
Doing the coordinates transformations and constants redefinitions
√ e0
a0 t + U (r, θ) → t , √ θ → φ,
a0
b0 → k1 , b1 → k0 , (17.90d)

dropping prime, one arrives at the solution (17.82e), CPSa(A.16).

The Gürses solution (17.91a) below is an special case of the above solution
(17.90) for u1 = 0, u2 = − √1a0 (c0 + 13 e0 μ r2 ), c0 = constant.

Gürses Solution
In the work by Gürses (1994) dealing with a perfect fluid (5) of constant pressure
and density as solution of the Einstein plus Λ equations, its interpretation is
established as a vacuum TMG plus Λ solution given by CPSa(A.18):
 2
√ 3c0 + μ e0 r2 e2 1
ds2 = − a0 dt − √ dθ + 0 r2 ψ(r)dθ2 + dr2 ,
3 a0 a0 ψ(r)
μ2 − 27Λ 2 b1
ψ(r) = r + b0 + 2 . (17.91a)
36 r
Accomplishing the coordinates transformations and constants redefinitions
√ c0 e0
a0 t − √ θ → t , √ θ → φ, b0 → k1 , b1 → k0 , (17.91b)
a0 a0
dropping prime, one arrives at the solution (17.82e), CPSa(A.16).

Clément Solution
Clément (1994) considered a “Killing symmetry reduction procedure to
obtain stationary rotationally symmetric solutions”. Clément’s solution (18),
CPSa(A.23), is given as
   2
√ 1 3b 2 μ dρ2 1
ds = −
2
2 adt + √ − ρ dθ + + F (ρ) dθ2 ,
2a μ 3 F (r) 2 a
1 b2
F (ρ) = (μ2 − 27Λ)ρ2 + 4(a − b) ρ + 9 2 . (17.92a)
9 μ
Accomplishing ing the coordinates transformations and constants redefinitions
CPSa(A.24)
√
√ 3b 2 9 b2
2 a t + √ θ → t , ρ → r2 , √ θ → φ, a − b → k1 , → k0 , (17.92b)
μ 2a a 4 μ2
dropping prime, one arrives at (17.82e), CPSa(A.16).
 17.6 Spacelike Biaxially Squashed Metrics 327

Anninos–Li–Padi–Song–Strominger Solution
Anninos et al. (2009) reported “warped” AdS3 black hole solutions of TMG; their
solution (3.4), after accomplishing the redefinitions σ → θ, u → φ, becomes the
solution (17.82d), CPSa(A.12).
The solutions reported in this subsection are gathered in the following table:

Table 17.5.1 Timelike biaxially squashed solutions

References Secs. or (Eqs.) Eqs. in Ref. Rel. to CPSa Λ

Vuorio (1985) (17.83) (2.21) (A.25) −

Percacci et al. (1987) 17.84 (3.20) (A.26) −
Nutku and Baekler (1989) 17.4.1 (4.10) (A.27) −
and Ortiz (1990) (17.47b) (A.28) −
Clément (1992a), (1992b) (17.85) (4.4–7) (A.30–33) −
√
Nutku (1993) (17.89) (25) (A.21)
√
Gürses (2008) (17.90) (42–46) (A.17)
√
Gürses (1994) (17.91) (5) (A.18)
√
Clément (1994) (17.92) (18) (A.23)
√
Anninos et al. (2009) (17.82d) (3.4) (A.12)
√
CPSa: Chow et al. (2010a) 17.5.4 (A.10–15)

17.6 Spacelike Biaxially Squashed Metrics

In this paragraph the integration of the Bianchi spacelike biaxially squashed
solutions is accomplished, starting from the middle-of-the-road metric element
λ1 λ2
ds2 = (dz + sinh ρ dτ )2 + (dρ2 − cosh2 ρ dτ 2 ), (17.93)
4 4
and searching for solutions to the Einstein-TMG equations. The trace of the field
equation for TMG in vacuum reads

4 λ2 + 3 Λ λ2 2 − λ1
Eμμ = = 0 → λ1 = 4 λ2 + 3 Λ λ2 2 , (17.94a)
λ2 2
the equation component

E z τ = λ1 3 λ1 − μ λ2 (−λ2 + λ1 ) = 0 → λ1 = 1/9 μ2 λ2 2 , (17.94b)

hence one arrives from (17.94a) at

36 144 μ2
1/9 μ2 λ2 2 = 4 λ2 + 3 Λ λ2 2 = 0 → λ2 = , λ 1 = 2 . (17.94c)
μ2 − 27 Λ (μ2 − 27 Λ)
The derived solution can be given by the metric in coordinates xα = {z, ρ, τ }
36 μ2 9
ds2 = 2 (dz + sinh ρ dτ )2 + (dρ2 − cosh2 ρ dτ 2 ), (17.95)
(μ2 − 27 Λ) (μ2 − 27 Λ)
328 Bianchi-Type (BT ) Spacetimes in TMG; Petrov Type D

and characterized by the Cotton tensor

1  2 
C αβ = μ μ + 9 Λ [2δ α z δ z β − δ α ρ δ ρ β − δ α τ δ τ β + 3 sinh ρδ α z δ τ β ] , (17.96)
9
and a traceless Ricci tensor κ S α β = Rα β − R α
3 δ β, which can be given also as

μ2 + 9Λ α 1 μ2 − 27 Λ α
Sβα = (δ β − 3 K α Kβ ), K α = ± δ z, (17.97)
9 6 μ
where K α is a spacelike unit vector.

17.6.1 Eigenvectors of the Cotton Tensor; Triad Formulation

As in the previous timelike case, one can represent the studied solution by means
of the triad formalism, in which the metric is given by
2 2 2
ds2 = g(a)(b) Θ(a) Θ(b) = Θ(1) + Θ(2) − Θ(3) (17.98)

where the orthonormal basis triad Θ(a) = h(a) α dxα is defined as

6μ 6 μ sinh (ρ)
Θ(1) α = [ , 0, 2 ], spacelike,
μ2 − 27 Λ μ − 27 Λ
1
Θ(2) α = [0, 3 , 0], spacelike,
μ − 27 Λ
2

cosh (ρ)
Θ(3) α = [0, 0, 3 ], timelike, (17.99)
μ2 − 27 Λ

hence, the triad tensor components h(a) α can be represented by the matrix
⎡ ⎤
μ
6 μ2 −27 Λ 0 6 μμ2sinh(ρ)
−27 Λ
⎢ ⎥
⎢ √ 21 ⎥
(a)
h β =⎢ ⎢ 0 3 0 ⎥. (17.100)
μ −27 Λ ⎥
⎣ ⎦
0 0 3 √cosh(ρ)
2 μ −27 Λ

The directional derivative vectors are:

∂ ∂
∂(a) = ∂(a) α α
= h(a) α α ,
∂x ∂x
μ2 − 27 Λ cosh (ρ)
∂(1) α =[ , 0, 3 ],
6μ μ2 − 27 Λ
1
∂(2) α = [0, μ2 − 27 Λ, 0],
3
sinh (ρ) μ2 − 27 Λ μ2 − 27 Λ
∂(3) α = [− , 0, ], (17.101)
3 cosh (ρ) 3 cosh (ρ)
 17.6 Spacelike Biaxially Squashed Metrics 329

thus, their components can be gathered in the matrix form

⎡ √ 2 ⎤
1 μ −27 Λ
0 0
  1 ⎢ 2 μ ⎥
⎢ ⎥
α
h(a) = μ2 − 27 Λ ⎢ 0 1 0 ⎥. (17.102)
3 ⎣ ⎦
sinh(ρ)
− cosh(ρ) 1
0 cosh(ρ)

From the coordinate components of the Cotton tensor (17.96), via

C (a) (b) = h(a) α C α β h(b) β ,

one determines its triad components

1   
C (a) (b) = μ μ2 + 9 Λ 2δ (a) (1) δ (1) (b) − δ (a) (2) δ (2) (b) − δ (a) (3) δ (3) (b) ,
9
(17.103)
and reciprocally the Cotton tensor coordinate components are

Cαβ = C(a)(b) Θ(a) α Θ(b) β = λs Θ(1) α Θ(1) β + λd Θ(2) α Θ(2) β − λd Θ(3) α Θ(3) β ,
μ  2  μ  2 
λs = 2 μ + 9 Λ , λd = − μ + 9Λ . (17.104)
9 9

17.6.2 Alternative Metric Representation of ds2sl

These representations are related to the coordinates used in the 2D “metric ds22
of AdS2 space, with squared radius

L2 = 9/(μ2 − 27Λ2 ), L > 0.

2D AdS2 with AdS radius L > 0 is the hyperboloid

−(X 0 )2 + (X 1 )2 − (X 2 )2 = −L2

in the flat 3D spacetime with metric

ds2 = −(dX 0 )2 + (dX 1 )2 − (dX 2 )2 . CPSa(A.36).

There are several coordinate systems to be chosen:

For the choice, CPSa(A.37):

X 0 = L cosh ρ cos τ, X 1 = L sinh ρ , X 2 = L cosh ρ sin τ. (17.105a)

one gets ds2 2 in polar coordinates, CPSa(A.38):

 
ds2 2 = L2 dρ2 − cosh2 ρ dτ 2 . (17.105b)

For the choice CPSa(A.39):

t L2 ρ2 − t2 L2 ρ2 − t2
X0 = L , X1 = − , X2 = + . (17.105c)
ρ 2ρ 2ρ 2ρ 2ρ
330 Bianchi-Type (BT ) Spacetimes in TMG; Petrov Type D

one gets ds2 2 in conformally flat coordinates that cover the Poincaré patch
CPSa(A.40):
L2  2 
ds2 2 = 2
dρ − dt2 . (17.105d)
ρ
The choice CPSa(A.41):

X 0 = L cosh ρ , X 1 = L sinh ρ cosh τ, X 2 = L sinh ρ sinh τ. (17.105e)

leads to ds2 2 in coordinates that cover X 0 ≥ L, CPSa(A.42):

 
ds2 2 = L2 dρ2 − sinh2 ρdτ 2 . (17.105f)

The choice CPSa(A.43):

X 0 = L sin τ, X 1 = L cos τ sinh φ, X 2 = L cos τ cosh φ, (17.105g)

gives ds2 2 in coordinates that cover only −L ≤ X 0 ≤ L, CPSa(A.44):

 
ds2 2 = L2 cos2 τ dφ2 − dτ 2 . (17.105h)

To facilitate comparison with the solutions reported in the literature, the space-
like squashed (SLS) solution, following Chow et al. (2010a), is presented in
various alternative forms with minor changes from our side, namely, restoring
the negative cosmological constant Λ and the λ constant, explicitly

4μ2
m2 = −Λ, λ2 = . (17.106)
μ2 − 27Λ
These solutions, corresponding respectively to CPSa(A.45), CPSa(A.47), and
CPSa(A.46), are given by
9
ds2sl = λ2 (dz + sinh ρ dτ )2 + dρ2 − cosh2 ρdτ 2 , (17.107a)
μ2 − 27Λ

9
ds2sl = λ2 (dz + cosh ρ dτ )2 + dρ2 − sinh2 ρdτ 2 , (17.107b)
μ2 − 27Λ
 
9 dt 2 − dt2 + dx2
ds2sl = 2 2
λ (dz + ) + , (17.107c)
μ − 27Λ x x2

and the time-dependent real cut of the complex extension of (17.107a), ρ →

i t, τ → −iφ, which yields
9
ds2sl = λ2 (dz + sin t dφ)2 − dt2 + cos2 tdφ2 , (17.107d)
μ2 − 27Λ
comparable with CPSa(A.48).
 17.6 Spacelike Biaxially Squashed Metrics 331

Hall–Morgan–Perjés SLS Solution; Λ = 0

Hall et al. (1987) searched for solutions of TMG of a particular Petrov–Segré
type; type D to be specific. Their solution (61), CPSa(A.51), is
 2
μ2 2 2 2
ds = 2du dv −
2
v du + dy − μ v du + 2 v f (u) du2 . (17.108a)
9 3
Accomplishing the coordinate transformations
 
u = e−F (u) du, v  = eF (u) v, F (u) := f (u)du, (17.108b)

one sees that f (u) is redundant, and one can equate it to zero, f (u) → 0. Under
such a choice, the second coordinate transformations
μ2  1 μ2 
u → û,  + u → v̂ (17.108c)
18 v 18
bring the metric to the form CPSa(A.52)
 2
36 dû dv̂ 12 dû
ds2 = − + dy + . (17.108d)
μ2 (û − v̂)2 μ û − v̂
Finally, the coordinate changes
1 1 μ 1
(û + v̂) → t, (v̂ − û) → x, y + log| (v̂ − û)| → −z, (17.108e)
2 2 6 2
leads to (17.107c), CPSa(A.46), with Λ = 0.

Nutku–Baekler and Ortiz BT III Spacelike Solutions; Λ = 0

Nutku and Baekler (1989) and independently Ortiz (1990) considered solutions
of Bianchi type III formed from left-invariant 1-forms without a cosmological
constant. The derivation of these solutions is given in 17.4, and 17.4.2, while
the metric is explicitly reported in (17.48b), it can be reduced to (17.107c),
CPSa(A.46), with Λ = 0.

Moussa–Clément–Leygnac SLS Solution; Λ = 0

Moussa et al. (2003) obtained black hole solutions by analytically continuing the
Vuorio (1985) solution, 17.5.4. Their solution (4), CPSa(A.58), is given by
 2 
9  ρ2 − ρ20 2 dρ2 2
ds = 2 −
2
dφ + 2 + 3 dt − (ω + ρ) dφ . (17.109a)
μ 3 ρ − ρ20 3
Making the coordinates transformations
√
ρ ρ 3
√0 φ → τ, → cosh ρ  , (ω φ − t) → z, (17.109b)
3 ρ0 2
dropping prime, one arrives at (17.107b), CPSa(A.47), with Λ = 0; ω = constant.
332 Bianchi-Type (BT ) Spacetimes in TMG; Petrov Type D

Bouchareb–Clément SLS Solution

Bouchareb and Clément (2007) reported their black hole solution (4.1),
CPSa(A.49),
μ2 − 27Λ 2 9 dρ2
ds2 = − (ρ − ρ20 )dφ2 + 2
2
3(μ + 9Λ) μ − 27Λ ρ − ρ20
2
   2
μ2 + 9Λ 1 4μ2
+3 dt − ρ + ω dφ . (17.110a)
4μ2 3 μ2 + 9Λ
Making the coordinates transformations
ρ0 μ2 − 27Λ ρ μ2 + 9Λ μ2 − 27Λ
√ φ → τ, → cosh ρ  , √ (ω φ − t) → z,
3 3 μ2 + 9Λ ρ0 3 4μ2
(17.110b)
dropping prime, one arrives at (17.107b), CPSa(A.47); ω = constant.

Anninos–Li–Padi–Song–Strominger SLS Solution

Anninos et al. (2009) consider “warped” AdS3 black hole solutions of TMG.
After making the coordinate relabellings σ → ρ, u → z, their solution (3.3)
becomes (17.107a), CPSa(A.45).
All the above-reported solutions can be gathered for quick reference in the
form of a table:

Table 17.6.1 Spacelike biaxilly squashed solutions

References Secs. or (Eqs.) Eqs. in Ref. Rel. to CPSa Λ

Hall et al. (1987) (17.108) (61) (A.51) −

Nutku and Baekler (1989) 17.4.2 (A.53) −
and Ortiz (1990) 17.6.2 (A.56)
Moussa et al. (2003) (17.109) (4) (A.58) −
√
Bouchareb and Clément (2007) (17.110) (4.1) (A.49)
√
Anninos et al. (2009) 17.6.2 (3.3) (A.45)
√
CPSa:Chow et al. (2010b) (17.107) (A.34–48)
 18
Petrov Type N Wave Metrics

18.1 Brinkmann-Like 3D Metric

Solutions with a covariantly constant null vector field ka (which is their invariant
characterization), ka k a = 0,

ka;b = 0 → ka;b k b = 0, Rab k a k b = 0, (18.1)

are called plane-fronted gravitational waves with parallel propagated null rays
and denoted as pp-waves. They have been known in 4D Einstein relativity since
long ago; Brinkmann (1923) discovered them in 1923; see Kramer et al. (1980),
§21.5. The condition on the Ricci tensor point on the existence of solutions for
null electromagnetic fields, radiation and vacuum, excluding other field solutions,
among them dS–AdS cosmological constant solutions. This fact establishes a
difference with the 3D wave solutions in TMG.
To get an insight of how the wave solutions with a non-vanishing cosmological
constant arise in the framework of the TMG, let us consider the Brinkmann-like
3D Killing wave metric

ds2 = dρ2 + 2F (ρ, u)dudv + H(ρ, u)du2 . (18.2)

One chooses the null vector field ka in the direction ∂v , thus

k μ = δ μ v , kμ = δ u μ F (ρ, u), k μ kμ = 0. (18.3)

Its covariant differential is

∂F (ρ, u)
kμ;ν = (δ u μ δ ρ ν − δ ρ μ δ u ν ) , (18.4)
∂ρ
consequently, the congruence is geodesic and Killingian one, but it is not covari-
antly constant to shear the name of pp-wave; this takes place only for the
structural function
F (ρ, u) = f (u) → 1, (18.5)
334 Petrov Type N Wave Metrics

but if this were the case, the metric could not allow for a cosmological con-
stant even in TMG; this subclass of solutions with vanishing cosmological
constant can properly be named pp-wave solutions; see Section 18.3 for specific
solutions.
For the class of wave solutions with a Killingian null congruence k, such that
kμ;ν = 0, (18.4), all the scalar invariants representing respectively the twist,
shear, and expansion:
1 μ
ω 2 = k μ;ν k[μ;ν] , σ 2 = k μ;ν kμ;ν − (k μ ;μ )2 , θ = k ;μ , (18.6)
2
vanish; these wave solutions can be called type N Killing waves and denoted as
TN-waves instead of pp-waves. This class of TN-solutions is different from type
N Kundt family of metrics, arising as limit of type III Kundt metric, to be treated
in detail in Chapter 19; it is worth pointing out that the null vector for this Kundt
class is a non-Killingian one; thus this class of wave solutions with a shear-free,
twist-free, expansionless, and non-Killingian null congruence can be called type
N non-Killingian waves and denoted TNnK-wave solutions. Having made these
remarks, we can proceed with the search for the corresponding solutions.

18.2 AdS3 Non-Covariantly Constant TN-Waves

To be consistent with the comments stressed above, we first integrate the cos-
mological case for the metric (18.2), and vector field (18.3). The E ρ ρ component
of the TMG equations is
 2
∂
F (ρ, u) + 4 Λ F 2 = 0 (18.7)
∂ρ
because of the nonlinearity, the equation splits in two branches of AdS solutions,
Λ = −m2 :
F (ρ) = F0 (u) e± 2 m ρ → F (ρ) = e 2 m ρ . (18.8)

The factor F0 (u) can be set equal to the unit, F0 (u) → 1, by a transformation
of the coordinate u, moreover, it is enough to treat the case with m, the branch
of solutions with −m is derivable from the first one simply by replacing there
m → −m. The E v v component yields a dependent equation, which, for the
solutions of (18.8), identically holds.
The only remaining equation is the E v u one, which for the F (ρ) solution (18.8)
becomes
∂3 ∂2 ∂
3
H (ρ, u) + (μ − 3 m) 2 H (ρ, u) − 2 m (μ − m) H (ρ, u) = 0. (18.9)
∂ρ ∂ρ ∂ρ
There arise three different cases depending on the μ(m) relation.
 18.2 AdS3 Non-Covariantly Constant TN-Waves 335

Case μ = ±m:
The general solution of (18.9) for H(ρ, u) is given by

H(ρ, u) = f1 (u) + f2 (u) e2 mρ + f3 (u) e(−μ+m)ρ , (18.10)

which is characterized by the Cotton tensor with components

1  2 
Cαβ = μ μ − m2 e−(μ− m)ρ f3 (u) δ u α δ u β . (18.11)
2
Using the null vector kα , (18.3), which is common for all the cases we are dealing
with: kα = F (ρ) δ u α , is equipped with a non-vanishing covariant derivative

kα;β = 2 me2 mρ δ u [α δ ρ β] , (18.12)

thus one concludes that the shear tensor and the divergence vanish, k(α;β) = 0
and k α ;α = 0, while the rotation tensor (twist) is different from zero k[α;β] =
0, although the shear, rotation, and expansion invariant scalars are all zero.
Consequently, this geeodesic null vector is also a Killing vector k(α;β) = 0. With
the help of this null Killing vector k, one may write the Cotton tensor (18.11)
as
1  
Cαβ = μ μ2 − m2 e−(μ+3 m)ρ f3 (u) kα kβ , k α = δvα , (18.13)
2
which manifestly shows the Petrov type N character of this Cotton tensor pos-
sessing a triple zero eigenvalue, Cαβ k α = 0.
Case μ = m:
For this branch of solution, the E v u equation (18.9) becomes
∂3 ∂2
H (ρ, u) − 2 m H (ρ, u) = 0 (18.14)
∂ρ3 ∂ρ2
hence it general solution can be given as

H (ρ, u) = f1 (u) + f2 (u) ρ + f3 (u) e2 mρ . (18.15)

It is characterized by a Cotton tensor

Cαβ = −f2 (u) m2 δ u α δ u β , (18.16)

or, using the null vector k from (18.3), one can represent it as

Cαβ = −f2 (u) m2 e−4 mρ kα kβ , k α = δvα , (18.17)

from which its Petrov type N property is apparent.

Case μ = −m:
For this branch of solutions, the E v u equation (18.9) amounts to
∂3 ∂2 ∂
H (ρ, u) − 4 m H (ρ, u) + 4 m2 H (ρ, u) = 0, (18.18)
∂ρ3 ∂ρ2 ∂ρ
336 Petrov Type N Wave Metrics

its general solution can be given as

H (ρ, u) = f1 (u) + f2 (u) e2 mρ + f3 (u) e2 mρ ρ (18.19)

This type N-wave gravitational field is characterized by the Petrov type N Cotton
tensor

Cαβ = −f3 (u) m2 e2 mρ δ u α δ u β = −f3 (u) m2 e−2 mρ kα kβ , k α = δvα , (18.20)

where the null vector k from (18.3) has been used.

18.2.1 AdS3 TN-Waves with Λ = 0

Summarizing, the general type N wave solutions with a geodesic Killing
null vector shear-free, twist-free, expansionless, and non-covariantly constant
congruence, denoted as TN-wave, are given by the metrics

μ = ±m : ds2 = dρ2 + 2e2mρ dudv + [e(m−μ)ρ f1 + e2mρ f2 + f3 ]du2 , (18.21a)

μ = m : ds2 = dρ2 + 2e2mρ dudv + [ρ f1 + e2mρ f2 + f3 ]du2 , (18.21b)

μ = −m : ds2 = dρ2 + 2e2mρ dudv + [ρ e2mρ f1 + e2mρ f2 + f3 ]du2 , (18.21c)

√
where m = −Λ, with Λ being a non-vanishing negative cosmological con-
stant; some of the arbitrary functions fi := fi (u) can be removed via coordinate
transformations. Nevertheless, it is convenient to include them when making
comparisons with the solutions existing in the literature; these representations
are equivalent to Eqs. (4.1)–(4.3) of Chow et al. (2010a).
In the literature about exact solutions for vacuum E-TMG equations in the
presence of a cosmological constant one find the same solution discovered by
different authors and given in different coordinate systems. Fortunately, the task
of identifying them has been accomplished in Chow et al. (2010a) (CPSa), and
their results are reproduced in this text.

18.2.2 Nutku TN-Wave Solution

Nutku (1993) considered several solutions with a cosmological constant. The
Nutku’s wave solution (16–17) is given by the metric
1  x 
1+μ/m
ds2 = 2 −2dudv + d x − 2 m ḧ + c [m x + h]
2
du2 , (18.22)
[m x + h]
CPSa(A.59), where c = constant and h(u) is an arbitrary function. Accomplish-
ing the coordinates transformations
1 ḣ
− log(m x + h(u)) → ρ, −v + 2 (m x + h(u)) → v  ,
m m
 18.2 AdS3 Non-Covariantly Constant TN-Waves 337

dropping prime, one arrives at (18.21a), CPSa(4.1), with

1
f1 = −c, f2 = (2hḧ + ḣ2 ), f3 = 0.
m2

18.2.3 Clément TN-Wave Solution

Clément (1994) considered a Killing symmetry reduction procedure to obtain
stationary rotationally symmetric solutions. The Clément solution (21) can be
written as
   2
1 dρ2 dθ2 M dθ
2
ds = + 2m ρ dt − 2 +
2 2
1 + cρ (1−μ/m)/2
dt − ,
4m2 ρ2 m 2 m
(18.23)
CPSa(A.60), where c and M are constant. Making the coordinates transforma-
tions  
1  θ 1 θ
logρ → ρ , t − → u, 2 t + → 2 v,
2m m m m
one arrives at (18.21a), or CPSa(4.1), with
Mc M
f1 = , f2 = 0, f3 = .
2 2

18.2.4 Ayón–Hassaı̈ne TN-Wave Solution

Ayón-Beato and Hassaı̈ne (2005) derived AdS waves by using the general AdS
wave ansatz. Their solutions given by equations (A3), (A5), and (A4) are
reported in CPSa correspondingly as CPSa(A.61)–CPSa(A.63):
1  
ds2 = 2 2 dy 2 − 2dudv − (m y)1+μ/m F1 (u)du2 , μ = ± m, (18.24a)
m y
1
ds2 = dy 2 − 2dudv − y 2 log(−m y)F1 (u)du2 , μ = m, (18.24b)
m2 y 2
1
ds2 = dy 2 − 2dudv − log(−m y)F1 (u)du2 , μ = − m, (18.24c)
m2 y 2
where F1 (u) is an arbitrary function. Doing the coordinates transformations
1
− log(−m y) → ρ, −v → v  ,
m
dropping prime, one arrives at (18.21a), (18.21b), (18.21c), or CPSa(4.1–4.3),
CPSa(A.11), with
f1 = −F1 (u), f2 = 0 = f3 .
Considering a correspondence between Cotton gravity with a conformally cou-
pled scalar field and TMG, see Deser et al. (2005) in this respect; Ayón-Beato
and Hassaı̈ne (2006) had previously found AdS waves.
338 Petrov Type N Wave Metrics

18.2.5 Ölmez–Sarioğlu–Tekin TN-Wave Solution

Ölmez et al. (2005) considered supersymmetric solutions. Their solution (9) can
be given as
 
ds2 = dρ2 + 2e∓ 2mρ du dv + β2 e∓(μ+m)ρ + β1 e∓ 2mρ + β0 dv 2 , (18.25)

CPSa(A.64), where {β0 (v), β1 (v), β2 (v)} are arbitrary functions. Carrying out
the coordinates transformations
u → v  , v → u  , ∓ρ → ρ  ,
dropping primes, one arrives at (18.21a), CPSa(4.1), with
f1 = β2 , f2 = β1 , f3 = β0 .

18.2.6 Dereli-Sarioğlu TN-Wave Solution

Dereli and Sarioğlu (2001) considered supersymmetric solutions. Their solution
(3.27–3.36) is a special case of the above solution of (18.25) with βi = constants
or (18.21a), CPSa(4.1), with constants fi .

18.2.7 Carlip–Deser–Waldron–Wise TN-Wave Solution

 ±m,
Carlip et al. (2008) derived an AdS wave solution (3.14 ), for μ =
CPSa(A.66) given by
1  
ds2 = 2 2 dz 2 + 2 dx+ dx− + 2h(x+ ) (m z)1+μ/m (dx+ )2 , (18.26)
m z
where h(x+ ) is an arbitrary function. Subjecting (18.26) to the coordinates
transformations
1
− log(m z) → ρ, x+ → u, x− → v,
m
for μ = ±m, one arrives at (18.21a), CPSa(4.1), with
2h
f1 = , f2 = 0 = f3 .
m2

18.2.8 Gibbons–Pope–Sezgin TN-Wave Solution

Gibbons et al. (2008) considered supersymmetric solutions of TMG. They found
that all such solutions are AdS waves; their solutions are just the ones reported
above (18.21a), (18.21b), and (18.21c), or CPSa((4.1), (4.2), (4.3)).

18.2.9 Anninos–Li–Padi–Song–Strominger TN-Wave Solution

Anninos et al. (2009) considered “warped” AdS3 black hole solutions of TMG.
Their solution (3.7), for μ = −3m, can be written as CPSa(A.65)
 18.2 AdS3 Non-Covariantly Constant TN-Waves 339
 
21 2 +1 −
− 2
ds = 2 2 du + dx dx + 2 (dx ) . (18.27)
m u u
Making the coordinates transformations
1 x+
− log(m u) → ρ, m2 x− → u  , → v,
m 2m2
dropping prime, one arrives at (18.21a), CPSa(4.1) with
μ = −3m, f1 = 1, f2 = 0 = f3 .

18.2.10 Garbarz–Giribet–Vásquez TN-Wave Solution

Garbarz et al. (2009) obtained solutions for the special values of μ = ±m. Their
solution (1.5), for μ = m, is given by
 −2   
r2 κ2 M κ2 M 1
ds = 2 r −
2 2
dr − m r −
2 2 2
dt − 2 dφ
2 2
m 2m2 2m2 m
  2 2
 2
   2
r κ M κ M dφ
+ k log 2 − 2 2 + dt − , (18.28)
r0 2m r0 2 m
CPSa(A.67), where M, r0 , κ, k are constants. Accomplishing the coordinates
transformations
 
κ2 M 1 1
m2 r 2 − 2
→ e2mρ , t − φ → u, t + φ → −2 v,
2m m m
one arrives at (18.21b), CPSa(4.2), with
κ2 M
f1 = 2km, f2 = 0, f3 = − 2klog(mr0 ).
2
Their second solution (3.18), for μ = −m, is given as

r2 κ2 M
−2
κ2 M
 
ds2 = m2 r2 − 2m2 dr2 − m2 r2 − 2m2 dt2 − m12 dφ2
  2
κ2 M r2 κ2 M 2
+ k r2 − 2m2 log r02
− 2m2 r02
+ κ 2M dt + dφ m , (18.29)

CPSa(A.68), where M, r0 , κ, k are constants. Carrying out the coordinates

transformations
 
κ2 M 1 1
m r −
2 2
→ e2mρ , t − φ → u, t + φ → −2 v,
2m2 m m
one arrives at (18.21c), CPSa(4.3), with
2k 2k κ2 M
f1 = , f2 = − 2 log(mr0 ), f3 = .
m m 2
These type N waves solutions, with a Killing geodesic non-covariantly constant
null congruence, can be shown in the form of a table:
340 Petrov Type N Wave Metrics

Table 18.2.1 Type N solutions with a non-covariantly constant null vector

TN-wave solutions; Authors Equations Ref. Λ

√
Chow et al. (2010a) (18.21) (4.1)–(4.3)
√
Nutku (1993) (18.22) (16–17)
√
Clément (1994) (18.23) (21)
√
Ayón-Beato and Hassaı̈ne (2005) (18.24) (A.3–5)
√
Ölmez et al. (2005) (18.25) (9)
√
Dereli and Sarioğlu (2001) (18.25) βi const.
√
Carlip et al. (2008) (18.26) (3.14)
√
Gibbons et al. (2008) (18.21a)–(18.21c)
√
Anninos et al. (2009) (18.27) (3.7)
√
Garbarz et al. (2009) (18.28), (18.29) (1.5), (3.18)

18.3 pp-Wave Solutions; Λ = 0

For a covariantly constant null congruence k, (18.1), for the wave metric (18.2),
the TMG equations considerably simplify: first, the cosmological constant Λ van-
ishes, Λ=0; second, from the E ρ ρ component equation, (18.7), one establishes
that
d
F (ρ) = 0 → F (ρ) = F0 → 1 (18.30)
dρ
while the E v u component (18.9) yields
∂3 ∂2
3
H (ρ, u) + μ 2 H (ρ, u) = 0, (18.31)
∂ρ ∂ρ
with general solution
H (ρ, u) = f3 (u) + f2 (u) ρ + f1 (u) e−μ ρ . (18.32)
Therefore the metric for pp-wave solutions is given by
 
ds2 = dρ2 + 2 du dv + f1 (u)e−μ ρ + ρ f2 (u) + f3 (u) du2 , (18.33)
coinciding with CPSa(A.70); the functions f2 (u) and f3 (u) can be equated to
zero via transformations, but here are maintained as they stand for comparison
with solutions reported in the literature. This solution is characterized by the
Petrov type N Cotton tensor
1
Cαβ = f1 (u)μ3 e−μ ρ kα kβ , k α = δvα , kα = δαu . (18.34)
2
where k is a null vector.

18.3.1 Martinez–Shepley pp-Wave Solution; Λ = 0

Martinez and Shepley (1986) happen to be the first researches to look for pp-
wave solutions of TMG, as one realizes from an unpublished preprint of them
with zero cosmological constant.
 18.3 pp-Wave Solutions; Λ = 0 341

18.3.2 Aragone pp-Wave Solution; Λ = 0

Aragone (1987) reported his (11) solution by the metric
4N02 2
ds2 = dx2 + ċ(u) du2 − 2du dv − 2N0 du dx, (18.35)
(c(u) − μ v)2 μ
CPSa(A.71). Subjecting (18.35) to the coordinates transformations
 
c(u) c(u)
v− → eμρ/2 , N0 x + N0 v − → u ,
μ μ
 −1
4 c(u)
−u+ 2 v− → v ,
μ μ
dropping primes, one arrives at CPSa(A.70) with
4
f1 = , f2 = 0 = f3 .
μ2

18.3.3 Percacci–Sodano–Vuorio pp-Wave Solution; Λ = 0

Percacci et al. (1987) considered stationary solutions for which the timelike
Killing vector has a constant scalar twist. The solution given by their equation
(3.19), is
 2  1 2
2 1 0 2 μx
ds2 = (dx 1 2
) − dx dx − (dx0 )2 , (18.36)
μx1 2 8
CPSa(A.72). In the original expression of the solution (3.19), there are functions
ωi not specified explicitly, but subjected to a certain equation; in CPSa(A.72)
these functions are chosen as ω1 = 0 and ω2 = 16/μ3 (x2 )2 ; any other choice is
equivalent by redefinition of x0 .
Making the coordinates transformations
x0
→ u, x1 → e−μρ/2 , −x2 → v,
4
one arrives at (18.33), CPSa(A.70), with
μ2
f1 = − , f2 = 0 = f3 .
4

18.3.4 Hall–Morgan–Perjés pp-Wave Solution; Λ = 0

Hall et al. (1987) reported a pp-wave solution (46) of the form
ds2 = du2 + 2 dx dr − 2e−μ u f (x) dx2 , (18.37)
CPSa(A.73), where f (x) is an arbitrary function. Doing the coordinate transfor-
mations
u → ρ, x → u  , r → v,
342 Petrov Type N Wave Metrics

dropping prime, one arrives at (18.33), CPSa(A.70), with

f1 = −2f, f2 = 0 = f3 .

18.3.5 Dereli–Tucker pp-Wave Solution; Λ = 0

Dereli and Tucker (1988) considered solutions with a pp-wave-like ansatz. The
solution in their equations (2.14) and (2.22) can be given by
   
1 1 1
ds2 = dx2 + 2du dv + 2 2 eμ x f1 + f3 − f1 x + f2 − 2 f1 du2 , (18.38)
μ μ μ
CPSa(A.74), where f1 (u), f2 (u), and f3 (u) are arbitrary functions. Carrying out
the coordinates transformations x → −ρ, one arrives at (18.33), CPSa(A.70), but
with the replacements of f1 → 2f1 /μ2 , f2 → 2(f1 /μ − f3 ), f3 → 2(f2 − f1 /μ2 )
there.

18.3.6 Deser–Steif pp-Wave Solution; Λ = 0

Deser and Steif (1992) considered an impulsive pp-wave solution. Their solution
(2.1) of the equation
1
Gμν + Cμν = −κ2 Tμν , Tuu = Eδ(y)δ(u) (18.39)
μ
is given as
   
1
ds2 = dy 2 − du dv + C + B y + 2 Eκ2 y − (1 − e−μ y ) θ(y)δ(u) du2 ,
μ
(18.40)
CPSa(A.76), where C(u), and B(u) are arbitrary functions.
For y > 0 and δ(u) → 1, making the coordinates transformations
y → ρ, v → −v  /2,
dropping prime, one arrives at (18.33), CPSa(A.70), with
f1 = 2κ2 E/μ, f2 (u) = 2κ2 E + B(u), f3 = −2κ2 E/μ + C(u).

18.3.7 Clément pp-Wave Solution; Λ = 0

Clément (1994) considered, in the class of stationary rotationally symmetric
solutions, a pp-wave metric in his equation (4.15), which amounts to
ds2 = dr2 ± 2σ0 (dt − ω0 dθ) dθ − σ0 (a + br + c e∓ μ r ) (dt − ω0 dθ)2 , (18.41)
see CPSa(A.75), where a, b, c, ω0 , σ0 are constants. Accomplishing the coordi-
nates transformations
± r → ρ, t − ω0 θ → u, ± ω0 θ → v,
one arrives at (18.33), CPSa(A.70), with
f1 = −c ω0 , f2 = ∓ b ω0 , f3 = −a ω0 .
 18.3 pp-Wave Solutions; Λ = 0 343

18.3.8 Cavaglia pp-Wave Solution; Λ = 0

Cavaglia (1999) reported various classes of pp-waves. The solutions in equations
(23)–(26) of Cavaglia (1999) are given by the metric
2
ds2 = −(h(u − v)) du dv − H(u + v)(du + dv) dφ, (18.42)
⎧
⎨ 1) μ tanh 4 (u − v)α − β ,
⎪ α 1

h(u − v) = 2) − α
μ tanh 4 (u − v)α − β ,
1
⎪
⎩ 3) α 1 (u − v)α − β −1 ,
μ 4

where α and β are constants, CPSa(A.77). Making the coordinates transforma-

tions

H(u + v)d(u + v) → u  , φ → −2 v 

together with the transformation involving ρ,

   2
1 α
for 1) : cosh (u − v)α − β → eμρ/2 , f1 = −f3 = , f2 = 0,
4 2μ H
   2
1 α
for 2) : cos (u − v)α − β → e μρ/2
, −f1 = f3 = , f2 = 0,
4 2μ H
 2
1 α
for 3) : (u − v)α − β → eμρ/2 , f1 = − , f2 = 0 = f3 ,
4 2μ H
dropping primes, one arrives at (18.33), CPSa(A.70).

18.3.9 Dereli–Sarioğlu pp-Wave Solution; Λ = 0

Dereli and Sarioğlu (2001) considered supersymmetric solutions. The solution,
in their equations (43),(44), and (45), is obtained as a limit of their more general
solution with a non-vanishing cosmological constant reviewed in Ölmez et al.
(2005).

18.3.10 Garcı́a–Hehl–Heinicke–Macı́as pp-Wave Solution; Λ = 0

Garcı́a et al. (2004) constructed a pp-wave solution (2.137)–(2.141)

ds2 = dy 2 + dx2 − dt2 − (C + Ay + Beμ y )(dt − dx)2 , (18.43)

CPSa(A.78). where A, B, C are constants. Carrying out the coordinates trans-

formations
y → −ρ, t − x → u, t + x → −v,

one arrives at (18.33), CPSa(A.70), with

f1 = −B, f2 = A, f3 = −C.
344 Petrov Type N Wave Metrics

18.3.11 Macı́as–Camacho pp-Wave Solution; Λ = 0

Macı́as and Camacho (2005) considered Kerr–Schild solutions of TMG without
a cosmological constant. They gave two solutions explicitly. The solution in their
equation (63) is reported in Chow et al. (2010a) as CPSa(A.79), namely
  2
a μ(ξ+y0 u) y02
ds = dξ − 2du dv + 2 c + b(ξ + y0 u) + e
2 2
dv + y0 dξ + du ,
μ 2
(18.44)
where y0 is a constant. Making the coordinates transformations
ξ + y0 u → −ρ, v + y0 ξ + y02 u/2 → u , u → −v  ,
dropping primes, one arrives at (18.33), CPSa(A.70), with
2a
f1 = , f2 = −2b, f3 = 2c.
μ
The second solution of their equation (72) amounts to
1  2
ds2 = dξ 2 − 2du dv − c + α(ξ + 1/μ) + γeμ ξ dv + y0 dξ + y02 du/2 ,
μ
(18.45)
CPSa(A.80), where c = c(v) and γ = γ(v) are arbitrary functions, and α is a
constant. Accomplishing the coordinates transformations
ξ → −ρ, −v → u  , u → v  ,
dropping primes, one arrives at (18.33), CPSa(A.70), with
γ(u) α C(u) + α/μ
f1 = − , f2 = , f3 = − .
μ μ μ
All the references of these pp-wave solutions can be gathered in a table:

Table 18.3.1 Type N solutions with a covariantly constant null vector; pp-wave
solutions

References Eqs. Eqs. to Ref. Λ

Martinez and Shepley (1986) 18.3.1 −

Aragone (1987) (18.35) (11) −
Percacci et al. (1987) (18.36) (3.19) −
Hall et al. (1987) (18.37) (46) −
Dereli and Tucker (1988) (18.38) (2.14), (2.22) −
Deser and Steif (1992) (18.40) (7) −
Clément (1994) (18.41) (4.15) −
Cavaglia (1999) (18.42) (23)–(26) −
Dereli and Sarioğlu (2001) 18.3.9 (43)–(45) −
Garcı́a et al. (2004) (18.43) (20.137–141) −
Macı́as and Camacho (2005) (18.44), (18.45) (63), (72) −
CPSa: Chow et al. (2010a) (18.33) (A.70) −
 19
Kundt Spacetimes in TMG

In general, Kundt spacetimes are defined as those spacetimes that admit a null
geodesic vector field which is shear-free, twist-free, and expansion-free. In three
dimensions, because of this dimensionality, the twist and shear trivially vanish,
therefore a Kundt spacetime is simply one that admits an expansion-free null
geodesic congruence. These spacetimes allow for Petrov type II, D, III, N, and O
solutions as we shall see in the next paragraphs. Moreover, except for a general
type II branch of metrics, all the existing Kundt solutions are CSI-constant
scalar invariants – spacetimes for which all polynomial scalar (Ricci) curvature
invariants are constant.

19.1 Null Geodesic Vector Field

In general relativity Ehlers (1961) introduced the frame of reference description
of the kinematics of a fluid uα ; the frame is constructed on timelike trajectories
to which the timelike vector field uα is tangent, and a local spacelike hypersurface
orthogonal to those trajectories. In (2 + 1) dimensions the equivalent description
was worked out by Barrow et al. (1986).
Similarly, when dealing with light trajectories, one has at one’s disposal a null
vector field k α – the tangent vector to the light rays – and one can decompose
the spacetime with respect to a frame constructed on the basis of the null vector
k α but, this time, using a second null vector lα such that k α lα = −1, to define
the metric projection tensor

bμν = gμν + kμ lν + kν lμ ; k μ kμ = 0 = lμ lμ , k μ lμ = −1,

bμν k ν = 0 = bμν lν , (19.1)

hence, the sub-manifold with metric bα β is orthogonal to the directions k α and

lα , and consequently spanned by spacelike vectors; for a (3 + 1)-dimensional
spacetime the sub-manifold is a 2D orthogonal hypersurface, while for a
346 Kundt Spacetimes in TMG

(2 + 1) spacetime it is a one-dimensional curve. There is a full parallelism in

the formulation of scalar invariants in (3 + 1) and (2 + 1) spacetime geometries,
except for minor changes in numerical factors due to the difference in dimensions.
Thus let us consider a null vector field k μ , which is required to be geodesic and
affinely parametrised, therefore
kμ;ν k ν = 0, k μ kμ = 0. (19.2)
Instead of considering the decomposition of kμ;ν into its kinematical parts with
respect to the spacetime metric gα β , as one commonly does for the timelike fluid
vector field, defining the shear σμν and rotation ωμν tensors, and the expansion
scalar θ according to
2θ 1
kμ;ν = ωμν + σμν + gμν , θ = k μ ;μ ,
D 2 
1 1 4θ
ωμν := (kμ;ν − kν;μ ), σμν := kμ;ν + kν;μ − gμν , (19.3)
2 2 D
one can do it equivalently using its projection onto the sub-manifold by means
of bαβ
 
1
kα;β := bμ α bν β kμ;ν = kα;β + kα kσ;β lσ + kβ kγ;σ lγ lσ
2
 
1
+kβ kα;σ lσ + kα kγ;σ lγ lσ ,
2
to define the expansion θ, the twist ωμ;ν , and the shear σμ;ν tensors according
with
2θ 1 1
kμ;ν = ωμ;ν + σμ;ν + bμν , θ = g μν kμ;ν = k μ ;μ ,
D−2 2 2
1 1 4θ
ωμν := (kμ;ν − kν;μ ), σμν := (kμ;ν + kν;μ − bμν ),
2 2 D−2
bμ μ = g μ μ − 2 = D − 2. (19.4)
These definitions we shall adopt from now on. They depend on the choice of the
vector field lα . Nevertheless the scalars of expansion θ, rotation ω 2 , and shear
σ 2 depend only on kμ ; these last two invariants are defined as follows
(k μ ;μ )2
ω 2 = ωμν ω μν = k μ;ν k[μ;ν] , σ 2 = σ μν σμν = k μ;ν kμ;ν − . (19.5)
D−2
In the particular case of 3D spacetime geometry, as has been stated previously,
the hypersurface to which k and l are orthogonal is one-dimensional and space-
like. Thus there exits a spacelike vector field mα , mα mα = 1, such that one may
identify the projection tensor with its tensor product, namely bαβ = mα mβ .
Because of the dimension 1 of the direction to which m is tangent, there is
no room for the rotation and shear; as stated in CPS (Chow et al., 2010b)
the shear and the twist (rotation) trivially vanish, consequently, “every null
 19.2 General Kundt Metrics 347

geodesic congruence of a three-dimensional spacetime is shear-free and twist-

free.” Hence, in three dimensions, “a Kundt spacetime is simply one that admits
an expansion-free null geodesic congruence.”
D-dimensional Kundt spacetimes are defined by metrics admitting a null
geodesic vector field that is shear-free σ 2 = 0, and twist-free ω 2 = 0, and
expansion-free, θ = 0.
Similarly, D-dimensional Robinson–Trautman are defined as those spaces
admitting a null geodesic vector field that is shear-free σ 2 = 0, and twist-free
ω 2 = 0, but possessing expansion, θ = 0. Consequently, a three-dimensional
Robinson–Trautman spacetime admits an expanding, k μ ;μ = 0, null geodesic
congruence, kμ;ν k ν = 0 .

19.2 General Kundt Metrics

Despite TMG being a higher-derivative theory, for Kundt metrics, as quoted in
Chow et al. (2010b), the field equations are still linear in curvature, and “so
the calculations and results bear some similarity to those for Kundt metrics of
four-dimensional Einstein gravity.”
In this text, the term metric is used to denote a geometrical structure – line
element – equipped with a metric tensor, i.e., a set of metric structural func-
tions, while the term solution refers to that set of metric structural functions
which fulfill a set of field equations, and consequently there is no freedom in the
structural functions. Nevertheless, referring to a specific solution, it is given by
the metric and a set of particular functions.
The four-dimensional Kundt metric, borrowed from the Kramer–Stephani et
al. books (see Kramer et al., 1980 and Stephani et al., 2003, §27), is given
by:
2  
ds2 = 2 dζ dζ̄ − 2du dv + W dζ + W̄ dζ̄ + Hdu . (19.6)
P
where P (ζ, ζ̄, u), and H(ζ, ζ̄, u) are real functions, and W = W (ζ, ζ̄, u) is a
complex function. This metric is invariant under the following coordinate trans-
formations and the associated transformations of the functions P , W , and H
functions:
ζ  = f (ζ, u) :
P  = P 2 f,ζ f¯,ζ̄ , W  = W/f,ζ , +f¯,u /(P 2 f,ζ f¯,ζ̄ ),
2

1  ¯ 
H = H − f ,u f,u /P 2
+ W f ,u
¯ + W̄ f¯,u f,ζ ,
f, ζ̄ (19.7a)
f,ζ f¯ ,ζ̄

v  = v + g(ζ, ζ̄, u) :
P  = P, W  = W − g,ζ , H  = H − g,u , (19.7b)
348 Kundt Spacetimes in TMG

v
u = h(u), v  = :
h,u
 
W 1 h,u,u
P  = P, W  = , H = H + . (19.7c)
h,u h,u 2 h,u

A special subclass of metrics – Kundt vacuum type III and N metrics (R12 =
∂ζ ∂ζ̄ ln P + Ψ2 + Ψ̄2 = 0) – arises for

∂ζ ∂ζ̄ ln P = 0 → P P,ζ,ζ̄ − P,ζ P,ζ̄ = 0.

In such a case, always, via the transformation (19.7a), P can be brought to

P = 1; the demonstration goes as follows

∂ζ ∂ζ̄ ln P = 0 → ln P = F (ζ, u) + F̄ (ζ̄, u) → P = eF (ζ,u) eF̄ (ζ̄,u)

→ P  = e2F (ζ,u) e2F̄ (ζ̄,u) f,ζ f¯,ζ̄ ,
2

hence, if one requires P  → 1, then

2

∂f (ζ, u)
e2F (ζ,u) f,ζ → 1, = e−2F (ζ,u) ,
∂ζ
 ζ

f (ζ, u) ≡ ζ (ζ, u) = e−2F (ζ,u) dζ,
ζ0

achieving in this way P  = 1.

19.2.1 3D Kundt Metric

A dimensional reduction of the 4D Kundt metric (19.6) is achieved by considering
the sub-manifold with coordinates
√ {v, u, Reζ}, i.e., the slices z = 0 of the complex
coordinate ζ = (r + iz)/ 2
√
ζ =→ r/ 2, u → −u, together with , H(v, u, ζ) → −f (v, u, r)/2,
√
W (v, u, ζ) + W̄ (v, u, ζ̄) → 2W (v, u, r), (19.8)

(instead of the coordinate ρ used in CPSb, Chow et al. (2010b), we prefer to

denote it by r), gives rise to the general 3D Kundt metric in the form
1
ds2 = dr2 + 2 dudv + f (v, u, r) du2 + 2 W (v, u, r) du dr. (19.9)
P 2 (v, u, r)
The transformation (19.7a) becomes

r = r (r, u) :
     2
P  = P r,r , W  = W/r,r
− r ,u /(P r,r ) ,
 
f  = f + (r ,u ) /(P r,r

)2 − W r ,u /(r,r

2
). (19.10)
 19.2 General Kundt Metrics 349
r
In particular, the change r (u, r) = 1/P (u, r) dr, can be used to bring,
dropping primes, the Kundt metric to

ds2 = dr2 + 2du dv + f (v, u, r)du2 + 2W (v, u, r)du dr. (19.11a)

The null geodesic vector k μ is given by

k μ = δ μ v , kμ = δ u μ (19.11b)

with its covariant derivative being

1 ∂ 1 ∂
kμ;ν = f (v, u, r) δ u μ δ u ν + W (v, u, r) (δ u μ δ r ν + δ r μ δ u ν ), (19.11c)
2 ∂v 2 ∂v
hence, k is geodesic and expansion-free as can be established straightforwardly
form the above equation, additionally, it is shear–free and rotationless. Moreover
this congruence is, in general, non-Killingian; for a single pp-wave case it becomes
a Killing null field.
When using numbers to denote coordinates, the correspondence is {v, u, r}
= {1, 2, 3}. Most of the functions are denoted with Latin characters.
For the general Kundt metric (19.11a) the TMG equations read:
1 ∂3
E21 = − W (v, u, r) = 0, (19.11d)
2μ ∂v 3
integrating one gets

W (v, u, r) = v 2 W2 (u, r) + vW1 (u, r) + W0 (u, r), (19.11e)

Replacing this W (v, u, r) into

 
3 2 1 ∂2W ∂3f
E 1 =E 3 = 2μ + =0 (19.11f)
4μ ∂v 2 ∂v 3
one can integrate f (v, u, r) as

f (v, u, r) = F3 (u, r) v 3 + F2 (u, r) v 2 + F1 (u, r) v + F0 (u, r) ,

2μ
F3 (u, r) := − W2 (u, r) . (19.11g)
3
On the other hand the evaluation of scalar curvature R, which for this vacuum
TMG theory is equal to 6Λ, R = 6Λ, gives
 2
∂2 3 ∂ ∂2 ∂2
R= f − W − 2 W W + 2 W = 6Λ. (19.11h)
∂v 2 2 ∂v ∂v 2 ∂v∂r
Substituting into (19.11h) the obtained metric functions one gets
 
45 2 6 ∂ 15 6
− 2 v 2 (F3 ) + v 6 F3 + F3 − F 3 W1 − W 0 F 3
2μ μ ∂r μ μ
∂ 2
+2 F2 + 2 W1 − 3/2 (W1 ) − 6Λ = 0, (19.11i)
∂r
350 Kundt Spacetimes in TMG

therefore, because of the variable independence on v, one has

F3 = 0 → W2 = 0, (19.11j)
consequently, solving for F2 , one obtains
3 2 ∂
F2 (u, r) = 3 Λ + (W1 ) − W1 , (19.11k)
4 ∂r
which substituted in the remaining TMG equation E 2 2 yields
 
∂2 5 ∂
E22 : W 1 + μ − W 1 W1
∂r2 2 ∂r
 
3 2 1
+ 3 Λ + (W1 ) − μW1 W1 − 2 Λ μ = 0 (19.11l)
4 2
which, worked out in the CPSb fashion, can be written in the “subjective” form
 
∂ ∂ 1 3 ∂ 1
W1 − W12 − 2Λ + (μ − W1 )( W1 − W12 − 2Λ) = 0. (19.11m)
∂r ∂r 2 2 ∂r 2
As we shall see in a forthcoming paragraph, this equation plays a fundamental
role in the algebraic type properties of the resulting solutions.
Notice that one can always set W0 (u, r) = 0 by means of a coordinate
transformation of the coordinate v; in fact
v → v + Φ(u, r); (19.11n)
the gur -component becomes
gur → gur = ∂r Φ(u, r) + Φ(u, r) W1 + W0 + v W1 , (19.11o)
thus, requiring
∂r Φ(u, r) + Φ(u, r) W1 + W0 = 0, (19.11p)
which occurs if
 
 
Φ (u, r) = −e W1 (u,r)dr
W0 (u, r) dr + φ (u) e− W1 (u,r)dr , (19.11q)

one achieves the purpose

W0 (u, r) → 0, gur = v W1 (u, r) . (19.11r)

19.3 3D Canonical Kundt Metric

In this section, the main geometrical features of the Kundt metrics derived by
Chow et al. (2010b) in the framework of the vacuum topologically massive gravity
are exhibited and analyzed in detail.
Summarizing the results derived previously, one begins with the general
canonical Kundt metric which can be written as
ds2 = dr2 + 2du dv + (v 2 F2 + vF1 + F0 )du2 + 2 v W1 du dr, (19.12a)
 19.3 3D Canonical Kundt Metric 351

where Fi = Fi (u, r), i = 0, 1, 2, and W1 = W1 (u, r). These structural functions

ought to fulfill the TMG equations: the “key” equation
 
∂ ∂ 1 3 ∂ 1
W1 − W12 − 2Λ + (μ − W1 )( W1 − W12 − 2Λ) = 0; (19.12b)
∂r ∂r 2 2 ∂r 2
next, the algebraic equation to determine F2
3 2 ∂
F2 (u, r) = 3 Λ + (W1 (u, r)) − W1 (u, r) , (19.12c)
4 ∂r
∂
a linear ordinary second-order equation for F1 or first-order equation for ∂r F1
 
∂2 1 ∂
F 1 + μ − W 1 F1
∂r2 2 ∂r
 
5 ∂ ∂2
+ W1 − μ W1 − 3 W1 = 0, (19.12d)
2 ∂u ∂u∂r
and finally, a linear ordinary third-order equation to determine F0
  2  
∂3 3 ∂ 7 ∂ 1 ∂
3
F 0 + μ + W1 2
F0 − 3 Λ − W1 + W1 − W1 μ
2
F0
∂r 2 ∂r 2 ∂r 4 ∂r
     
  3 9 ∂
+ 4 Λ + W1 2 μ − W1 − μ − W1 W1 F 0
2 2 ∂r
∂2 1 ∂ 1 ∂W1 ∂ 2 W1
=− F1 − F 1 F1 + F1 . (19.12e)
∂u∂r 2 ∂r 2 ∂u ∂u2
The covariant derivative of the geodesic expansionless null vector k α = δ α v in
this representation amounts to
   
3 ∂ 1
kα;β = v W1 −
2
W1 + 3 Λ + F1 δ u α δ u β + W1 δ u (α δ r β) . (19.13)
4 ∂r 2
Notice that Kundt metrics in general possess a geodesic non-Killingian null con-
gruence k, only in the pp-wave limit, which takes place for {W1 = 0, F1 = 0, Λ =
0}, the covariant derivative kα;β vanishes.
The evaluation of the traceless Ricci tensor S μ ν , modulo the field equations
(19.12), yields

S μ ν = S 1 1 (δ μ v δ v ν + δ μ u δ u ν − 2δ μ r δ r ν ) + S 1 2 δ μ v δ u ν + S 1 3 δ μ v δ r ν + S 3 2 δ μ r δ u ν
(19.14a)
 
1 1 ∂ 1 2
S 1 1 = S 2 2 = − EVC = − W1 − (W1 ) − 2 Λ ,
2 2 ∂r 2
S 3 3 = EV C = −2S 1 1 ,
 
1 1 ∂ ∂
S 1 3 = − EVC v (3 W1 − 2 μ) + F1 − W1 ,
2 2 ∂r ∂u
 
1 ∂ ∂
S 3 2 = vμ EVC + F1 − W1 ,
2 ∂r ∂u
352 Kundt Spacetimes in TMG
 
1 1 ∂ ∂
S 2 = − EVC μ (3 W1 − 2 μ) v −
1 2
F1 − W1 (W1 − 2 μ) v
4 4 ∂r ∂u
 2 
1 ∂ ∂ ∂
− 2
F0 + W1 F0 + F0 W1 ,
2 ∂r ∂r ∂r
∂ 1 2
EVC = W1 − (W1 ) − 2 Λ. (19.14b)
∂r 2

19.3.1 Petrov Classification of the Cotton and Traceless Ricci

Tensors
It is apparent that S 1 1 plays the role of the eigenvalue of the matrix tensor (S α β );
if S 1 1 = 0 then the tensor S α β is Petrov type II with a possible degeneration to
Petrov type D: the traceless Ricci tensor matrix S, via elementary (similarity)
algebraic transformations – addition of one column to a second one multiplied
by a factor (scalar or function other than zero and addition of a row to another
one multiplied by a factor and multiplication of a column by a factor other than
zero) – can be brought into its equivalent Jordan matrix form, namely
⎡ 1 ⎤ ⎡ 1 ⎤
S 1 S12 S13 S 1 S̃ 12 0
⎢ ⎥ ⎢ ⎥
S := ⎢ ⎣ 0 S11 0 ⎥ ∼ JS = BSA = ⎢ 0
⎦ II ⎣ S11 0 ⎥∼
⎦
0 S32 −2 S 1 1 0 0 −2 S 1 1
⎡ ⎤
S 1 0
⎢ ⎥ S32 S13 + 2 S11 S12
JSII =⎢
⎣ 0 S 0 ⎥⎦ , S̃ 1
2 := , (19.15)
2S 1 1
0 0 −2 S
where the multiplication by the matrix A adds in S the third column multiplied
by S 3 2 /(2S 1 1 ) to the second one to eliminate the term in the intersection of
the second column and the third row, T 2 3 → 0, while the multiplication by the
matrix B adds in the resulting matrix the third row multiplied by S 1 3 /(2 S 1 1 )
to the first one to eliminate T 1 3 → 0, these matrices are given by
⎡ ⎤ ⎡ ⎤
1 0 0 1 0 S 1 3 /(2S 1 1 )
⎢ ⎥ ⎢ ⎥
A=⎢ ⎣ 0 1 0 ⎥ ⎢
⎦, B = ⎣ 0 1 0 ⎥.
⎦ (19.16)
0 S 3 2 /(2S 1 1 ) 1 0 0 1
Above, it has been assumed S 1 1 = 0, thus the equivalent Jordan matrix JS to
S is of Petrov type II for S̃ 12 = 0, and Petrov type D if S̃ 12 vanishes, S̃ 12 = 0,
with Jordan form
⎡ 1 ⎤
S 1 0 0
⎢ ⎥
JSD = BSA = ⎢ ⎣ 0 S11 0 ⎥.
⎦ (19.17)
0 0 −2 S 1 1
 19.3 3D Canonical Kundt Metric 353

Table 19.3.1 Petrov types of general Kundt metrics

S = ∂r W1 − W1 2 /2 − 2Λ
S α β :{S 1 1 = S 2 2 = S, S 3 3 = −2S, S 1 2 , S 1 3 , S 3 2 }

Petrov types Eigenvalues Jordan Matrix Λ

No type I
√
II S = 0 JSII , {S, S 1 2 }
√
D S = 0 JSD , {S, S 1 2 = 0}
√
III S=0 JSIII , {S = 0 = S 1 2 , S 1 3 , S 3 2 }
√
N S=0 JSN , {S = 0 = S 1 2 , S 1 3 , S 3 2 = 0}
√
O S=0 {0}

On the other hand, if the eigenvalue vanishes, S 1 1 = 0, then the tensor S α β is

Petrov type III with possible degeneration to Petrov type N or O (conformally
flat).
⎡ ⎤ ⎡ ⎤
0 S12 S13 0 S12 0
⎢ ⎥ ⎢ ⎥
S := ⎢⎣ 0 0 0 ⎥⎦ ∼ JSIII = T S = ⎣ 0
⎢ 0 S̃ 23 ⎥
⎦
0 S32 0 0 0 0
3 1
S 2S 3
S̃ 23 := − , (19.18)
S12
where the transformation matrix T ,
⎡ ⎤
0 0 S 1 2 /S 3 2
⎢ ⎥
T =⎢⎣ −S 2 /S 2
3 1
0 1 ⎥,
⎦ (19.19)
0 1 0

adds rows multiplied by certain specific factors and permute the resulting second
by the third row. It is assumed that S 1 2 = 0. If S̃ 23 := −S 3 2 S 1 3 /S 1 2 is different
from zero, S̃ 23 = 0 then the metric is of Petrov type III. On the contrary, if
S̃ 23 = 0, then the spacetime is of Petrov type N,
⎡ ⎤ ⎡ ⎤
0 S12 0 0 1 0
⎢ ⎥ ⎢ ⎥
JSN = T S = ⎢ ⎣ 0 0 0 ⎥ ⎢
⎦ ∼ ⎣ 0 0 0 ⎦.
⎥ (19.20)
0 0 0 0 0 0

Finally, if additionally S 1 2 = 0 the metric becomes conformally flat or Petrov

type O.
Summing up, for the 3D Kundt metrics the Petrov types of the traceless Ricci
and Cotton tensors are presented in the table above.
For the Cotton tensor – a symmetric and traceless tensor – a similar algebraic
classification to the one presented above for the traceless Ricci tensor takes place;
354 Kundt Spacetimes in TMG

instead of the components S α β replace them by the Cotton components C α β with

eigenvalue C 1 1 .
As is well known, for vacuum spacetimes the E-TMG equations are equivalent
to the scalar curvature equation and the vanishing relation between the traceless
Ricci and Cotton tensors:
1
R = 6 Λ, Sα β + Cα β = 0.
μ
These tensors shear the same eigenvalues up to the factor −μ and consequently
the same Petrov types. In particular, these relations hold for the canonical metric
(19.12a).
The general Petrov type II Kundt metric with a variable W1 (u, r) fulfilling
EVC = 0 and equation (19.12b) is unknown. The class of Petrov type II with
W1 = 2μ/3 has been completely integrated. For Petrov type III Kundt met-
rics EVC ought to vanish, EVC = 0; all solutions has been derived and fully
characterized.
Invariants. As far as to the Riemann invariants is concerned, one has:
Riemann–Ricci invariants
R = Rα β , (19.21a)

3
Rαβ Rαβ = EVC2 + 12 Λ2 , (19.21b)
2
 
5
Rαβ;γ Rαβ;γ = (3 W1 (u, r) − 2μ) − μ EV C 2
2 2
(19.21c)
8
traceless Ricci tensor S α β invariants
3
Sαβ S αβ = EVC2 , (19.22a)
2
S σ α Sσβ S αβ = 3/4 EV C 3 , (19.22b)

9
S σ α Sσβ S ν α Sν β = EVC4 , (19.22c)
8
3
Sαβ;γ S αβ;γ = W1 (u, r) (3 W1 (u, r) − 4 μ) EVC2 , (19.22d)
4
3
Sαβ;γ S αβ; S γ  = − W1 (u, r) (3 W1 (u, r) − 4 μ) EVC3 (19.22e)
8

Constant Scalar Invariant Spacetimes

Coley and collaborators accomplished a systematic study of “spacetimes for
which all polynomial scalar invariants constructed from the Riemann curva-
ture and its covariant derivatives are constant and are called constant scalar
invariant (CSI) spacetimes,” see Coley et al. (2006). In Coley et al. (2008) it
was established that a 3D CSI spacetime is either locally homogeneous or is
 19.3 3D Canonical Kundt Metric 355

a Kundt spacetime, and all 3D CSI spacetimes were determined. The proof of
this assertion is based on a theorem stating, roughly speaking, “that on an open
neighborhood in which the Segré type does not change and all Ricci invariants
are constant (consequently its eigenvalues are constant too) there exits a frame
such that all the components of the Ricci tensor are constants and of the canon-
ical (Segré) form in any dimension.” An extensive use of the Segré forms and of
triad frames in which the Ricci tensor components are constant, together with
the constancy (vanishing) of the invariants constructed from covariant derivatives
of the Ricci tensor allow to establish this powerful result in three dimensions.
Extensions to higher dimensions have been carried out by Coley et al. (2009a
and 2009b).
The invariants of Sαβ , equivalently, of the Cotton tensor Cαβ , involv-
ing covariant differentiations, contain, as factor, the function W1 (u, r) EV C
to some power. Thus, there are type II metrics, unknown at present, with
EV C = 0 and with a function W1 (u, r) depending on its variables and con-
strained to fulfill the key equation (19.12b). Moreover, type II metrics with
constant EV C fulfilling (19.12b) arise for W 1(u, r) = 2μ/3; the invariants
become constants and the solutions become constant scalar invariant (CSI)
spacetimes.
If the Cotton tensor is of type III, its eigenvalues are zero, EV C = 0, together
with its invariants. Moreover, the Cotton tensor invariants, or the curvature Ricci
invariants, become constants, hence all Kundt type III metrics are constant scalar
invariant spacetimes.

19.3.2 Sub-Branch W1 (r) of the General Kundt Metric in TMG

In Chow et al. (2010b) the W1 (r) case is studied in detail. It is argued that by
means of a transformation r → r + g(u) one can eliminate one of the integration
constants from the two that appear in the solution of the second-order equa-
tions (19.12b); this is true but in no way this means that this solution will be
independent of the coordinate u, considered as a parameter, for a partial differ-
ential equation of the form of the equation (19.12b). As an example, consider
the third-order differential equation (19.11e), explicitly
∂3
W (v, u, r) = 0 → W (v, u, r) = v 2 W2 (u, r) + v W1 (u, r) + W0 (u, r)
∂v 3
which, by means of coordinate transformations and function redefinitions, one
can achieve W0 (u, r) = 0, but still W1 (u, r) and W2 (u, r) remain functions of
their arguments.
An important general integral, within this framework, is given by the
integration of the function F1 (u, r), equation (19.12d),
   
∂2 1 ∂ 5 ∂ ∂2
2
F 1 + μ − W 1 F 1 = μ − W 1 W1 + 3 W1 ,
∂r 2 ∂r 2 ∂u ∂u∂r
356 Kundt Spacetimes in TMG

namely
    
5 ∂W1 ∂ 2 W1 dr
F1 (u, r) = { μ − W1 +3 }EI(r)dr
2 ∂u ∂u∂r EI(r)
  1
+f11 (u) EI(r)dr + f10 (u) , EI(r) := e (−μ+ 2 W1 )dr , (19.23)

where the function W1 (u, r) participates actively in the integrals, and, in general
for W1 (u, r) depending on r and u too, it should contribute substantially with
nonzero terms to the function F1 (u, r).
Therefore, extending these considerations to the case of the key equation
(19.12b) for W1 (u, r), a second-order equation, from the two parameters of inte-
gration depending on u one would be able, by transformations, to exclude one
of them, but not both at the same time, thus the remaining integration func-
tion will depend on the coordinate u. The general solution of the key equation
(19.12b) still remains unknown; it is a challenge to overcome.
The case W1 (r) is a sub-branch of Petrov type II Kundt metrics; even in this
case the general solution is unknown, although the families of solutions of type
II and D have been completely integrated in the case of constant W1 = 2μ/3
which are CSI spacetimes.

19.3.3 Kundt Metric Structure for W1 (r)

The Kundt metric for W1 (r) becomes

ds2 = dr2 + 2du dv + v 2 F2 + vF1 + F0 du2 + 2 v W1 du dr. (19.24a)

where the structural functions W1 (r), and Fi (u, r), i = 0, 1, 2, ought to satisfy
the following equations:
 
d d 1 2 3 d 1
W1 − W1 − 2Λ + (μ − W1 )( W1 − W12 − 2Λ) = 0, (19.24b)
dr dr 2 2 dr 2

3 2 d
F2 (r) = W1 (r) − W1 (r) + 3 Λ, (19.24c)
4 dr
 
∂2 1 ∂
F 1 + μ − W1 F1 = 0, (19.24d)
∂r2 2 ∂r

  2  
∂3 3 ∂ 7 ∂ 1 ∂
3
F 0 + μ + W1 2
F0 − 3 Λ − W1 + W1 − W1 μ
2
F0
∂r 2 ∂r 2 ∂r 4 ∂r
      
3 9 ∂ 3
+ 4 Λ μ − W1 − μ − W1 W1 + W1 2 μ − W1 F 0
2 2 ∂r 2
∂2 1 ∂
=− F1 − F 1 F1 . (19.24e)
∂u∂r 2 ∂r
 19.4 Type II CSI Kundt Metric; W1 = 2μ/3 357

The equation (19.24d) integrates as

 
F1 (r, u) = f11 (u) exp( (W1 (r)/2 − μ)dr)dr + f12 (u), (19.25)

which coincides with the the homogeneous solution of (19.23) for W1 (r). By a
coordinate transformation, one can send f12 (u) → 0, as stated in Chow et al.
(2010b).

19.4 Type II CSI Kundt Metric; W1 = 2μ/3

A particular type II metric, S = −EV C/2 = 0 arises for:
2 2
type II: W1 = μ, S = (μ2 + 9Λ) = 0. (19.26)
3 9
The corresponding type II metric gives rise to CSI solutions. Explicitly one has,

4μ
ds2 = dr2 +2du dv+ v 2 F2 (r) + vF1 (u, r) + F0 (u, r) du2 + v du d r, (19.27a)
3
see Chow et al. (2010b), Eqn. (A.32), or more compactly using for this kind of
citation the notation CPSb(A.32). The structural functions ought to satisfy the
following equations
1
F2 (r) = μ2 + 3 Λ, (19.27b)
9

∂2 2 ∂
F1 + μ F1 = 0, (19.27c)
∂r2 3 ∂r
 
∂3 ∂2 5 ∂ ∂2 1 ∂
3
F 0 + 2μ 2
F 0 + μ − 3Λ F0 = − F1 − F 1 F1 . (19.27d)
∂r ∂r 9 ∂r ∂u∂r 2 ∂r
The solution of (19.27c) is

F1 (u, r) = f11 (u)e−2μ r/3 + f12 (u), (19.28)

which can be set equal to zero, F1 (u, r) = 0, via coordinate transformations:

1
v = ṽ − f11 (u)e−2μ r/3 .
2(μ2 /3 + 3Λ)
Dropping tilde, the Petrov type II SCI Kundt metric amounts to CPSb(A.32)
 2
2 1 2 
ds = d r + μ v du + 2du dv −
2
μ − 27Λ v 2 du2 + F0 (u, r) du2 (19.29)
3 9
subjected to the equation for F0 (u, r)
 
∂3 ∂2 5 ∂
3
F0 + 2μ 2 F0 + μ − 3Λ F0 = 0, (19.30)
∂r ∂r 9 ∂r
358 Kundt Spacetimes in TMG

which possesses the general solution

√ 2 √ 2
F0 = f01 (u) e−μ e 4 μ +27 Λ r/3 +f02 (u) e−μ e− 4 μ +27 Λ r/3 +f03 (u) , (19.31)

allowing for a trigonometric representation in dependence of the relationship

μ(Λ) and the sign of the radical 4 μ2 + 27 Λ; the corresponding sub-classes are
reported in the forthcoming paragraphes.
Evaluating the traceless Ricci tensor one arrives at

S μ ν = S v v (δ μ v δ v ν + δ μ u δ u ν − 2δ μ r δ r ν ) + S v u δ μ v δ u ν + S r u δ μ r δ u ν , (19.32a)
1 2  1 2 
Svv = μ + 9 Λ = 0, S r u = −2 vμ μ + 9Λ ,
9   9
2
1 ∂ 2 ∂
Svu = − F0 + μ F0 . (19.32b)
2 ∂r2 3 ∂r
In general, within this branch of solutions with constant W1 = 2μ/3, the
spacetime is of type II for S v u = 0,
⎡ v ⎤
S v 1 0
⎢ ⎥
(JII ) = ⎢
⎣ 0 Svv 0 ⎥ , F0 (u, r) = 0.
⎦
0 0 −2S v v

For S v u = 0, which occurs for F0 (u, r) = 0, the solution is of the type D.

Returning to type II solutions, their explicit expressions depend on the rela-
tionship between μ(Λ) and also on the sign of Λ. Correspondingly, one may
distinguish the following branches:

19.4.1 Negative Cosmological Constant; Λ = −m2

For this negative cosmological constant, the metric is given by
 2
2 1 
ds = d r + μ v du +2du dv− μ2 + 27 m2 v 2 du2 +F0 (u, r) du2 . (19.33)
2
3 9
with structural function F0 (u, r), in correspondence with CPSb(3.5)–(3.7), of the
form:

27 2 4 μ2
2
μ > m ,γ = − 3 : f0i := f0i (u), i = 1, 2, 3,
4 9 m2
F0 = e−μ r cosh(γ m r)f01 + e−μ r sinh(γ m r)f02 + f03 , (19.34a)

27 2 4 μ2
μ2 < m , γ = 3− :
4 9 m2
F0 = e−μ r cos(γ m r)f01 + e−μ r sin(γ m r)f02 + f03 , (19.34b)
27
μ2 = m2 : F0 = r e−μ r f01 + e−μ r f02 + f03 , (19.34c)
4
 19.5 Type D CSI Kundt Solutions; W1 = 2μ/3, F0 = 0 359

where the constants of integration f0i (u) depending on the variable u are denoted
by f0i := f0i (u), i = 1, 2, 3. Via a coordinate transformation the arbitrary
function f03 (u) can be set zero, f03 (u) = 0.

19.4.2 Positive Cosmological Constant; Λ = m2

The metric in this case amounts to
 2
2 1 
ds = d r + μ v du +2du dv − μ2 − 27m2 v 2 du2 +F0 (u, r) du2 . (19.35)
2
3 9

with a single structural function F0 (u, r):


2 27 2 4 μ2
μ > m ,γ= +3:
4 9 m2
F0 = e−μ r cosh(γ m r)f01 (u) + e−μ r sinh(γ m r)f02 (u) + f03 (u), (19.36)

the function f03 (u) can be equated to zero, f03 (u) → 0, by coordinate
transformations. This solution corresponds to the CPSb(3.12) one.

19.4.3 Zero Cosmological Constant; Λ = 0

The simplest case is determined by metric
 2
2 1
ds = d r + μ v du + 2du dv − μ2 v 2 du2 + F0 (u, r) du2 ,
2
(19.37)
3 9

with the structural function F0 (u, r) of the form:

F0 = e−μ r/3 f01 + e−5 μ r/3 f02 + f03 . (19.38)

where f0i = f0i (u). The function f03 can be equated to zero, f03 → 0, by
coordinate transformations. This solution is just the CPSb(3.11) one.

19.5 Type D CSI Kundt Solutions; W1 = 2μ/3, F0 = 0

The type D CSI Kundt solution is given by the metric
2 1 2 
ds2 = (dr + μ v du)2 + 2du dv − μ − 27 Λ v 2 du2 , (19.39a)
3 9
and it is characterized by the type D traceless Ricci–Cotton tensors, S μ ν +
C μ ν /μ = 0,

S μ ν = S v v (δ μ v δ v ν + δ μ u δ u ν − 2δ μ r δ r ν ) + S r u δ μ r δ u ν , (19.39b)
360 Kundt Spacetimes in TMG

Table 19.5.1 Type II and Type D SCI Kundt solutions

S = ∂r W1 − W12 /2 − 2Λ = 0 and W1 (u, r) = 2μ/3

II W1 (u, r) μ(m) Eqs. Rel. CPS Λ = ±m2

II general W1 (u, r) unknown μ (19.12b) (2.4–15) Λ

II W1 = 2μ/3 μ (19.29) (3.4) Λ
II W1 = 2μ/3 μ2 > 27m2 /4 (19.34a) (3.5) <0
II W1 = 2μ/3 μ2 < 27m2 /4 (19.34b) (3.6) <0
II W1 = 2μ/3 μ2 = 27m2 /4 (19.34c) (3.7) <0
II W1 = 2μ/3 μ2 > 27m2 /4 (19.35) (3.12) >0
II W1 = 2μ/3 μ (19.37) (3.11) 0
D W1 = 2μ/3 μ (19.39a) (3.17) <0

1 2  2  
Svv = μ + 9 Λ = 0, S r u = − vμ μ2 + 9 Λ ,
9 9
⎡ v ⎤
S v 0 0
⎢ ⎥
JD =⎢
⎣ 0 Svv 0 ⎥.
⎦ (19.39c)
0 0 −2S v v
Making the coordinate transformation
18 1
u= ũ, v = ,
(μ2 − 27 Λ) ṽ − ũ
one brings (19.39a) to the form
 2
36 dṽ dũ 12μ dũ
ds2 = − + dr − , (19.39d)
μ2 − 27 Λ (ũ − ṽ)2 μ2 − 27 Λ ũ − ṽ
which, by the subsequent transformations
1 1 μ2 − 27 Λ
t= ũ + ṽ, x = ṽ − ũ, z = r − ln(ṽ − ũ), (19.39e)
2 2 6μ
becomes the: biaxially squashed spacelike limit
  2
2 9 dx2 − dt2 4μ2 dt
ds = 2 + 2 dz + , (19.39f)
μ − 27 Λ x2 μ − 27 Λ x

where the 2D sector AdS2 is written using Poincaré coordinates (t, x).
In a concise way all these solutions can be visualized in the table above.

19.6 Petrov Type III Kundt Metrics

If S 1 1 ≡ S v v = S = −EV C/2 = 0, the traceless Ricci S α β and Cotton C α β
tensors are Petrov type III; the triple eigenvalue is equal to zero, S 3 = 0. Thus
W1 (u, r) ought to fulfill
 19.7 Type III Kundt Solution; Λ = 0, W1 = 0 361

∂ 1 2
W1 (u, r) − (W1 ) − 2 Λ = 0, (19.40)
∂r 2
which guarantees the fulfillment of the key equation (19.24b). The solutions of
this equation (19.40) depend on r − f (u), thus, by a shift of the coordinate
r → r + f (u) one brings the dependence to r. One distinguishes the following
branches of solution depending on the sign of the cosmological constant Λ:
Case Λ = 0. For zero cosmological constant Λ = 0, there are two cases

Λ = 0 : W1 (u, r) = 0, (19.41a)
2
Λ = 0 : W1 (u, r) = − . (19.41b)
r
Case Λ = −m2 . For negative cosmological constant Λ = −m2 , one distinguishes
three different classes

Λ = −m2 : W1 (r) = −2 m, (19.41c)

Λ = −m : W1 (u, r) = −2 m coth (m r) ,
2
(19.41d)
Λ = −m2 : W1 (u, r) = −2 m tanh (mr) . (19.41e)

Case Λ = m2 . For positive cosmological constant Λ = m2 there is a single family

of solutions
Λ = m2 : W1 (r) = −2 m cot (mr) . (19.41f)

These structural functions W1 coincide, modulo the identification of σ with the

cosmological constant Λ, with the functions Wx1 , fulfilling (28) of Coley et al.
(2008) (equivalent to (19.40)), listed in the paragraph 3.2.2. Segré types {3},
{(21)}, and {(1, 11)} of Coley et al. (2008).
Each of these cases will be treated in detail in the forthcoming sections; first
one evaluates the function F2 (u, r), next one proceeds to integrate the function
F1 (u, r) to end with the integration of the function F0 (u, r).

19.7 Type III Kundt Solution; Λ = 0, W1 = 0

We start the derivation of the Petrov type III Kundt solutions handling the
easiest case of Λ = 0 and W1 from (19.41a): evaluating F2 from equation (19.24c),
one gets
W1 (u, r) = 0 → F2 = 0. (19.42)

Replacing these functions into the F1 – equation (19.24d) – one arrives at

∂2 ∂
F1 (u, r) + μ F1 = 0 (19.43)
∂r2 ∂r
with solution

F1 (u, r) = f12 (u) + f11 (u) e−μ r → F1 (u, r) = f11 (u) e−μ r , (19.44)
362 Kundt Spacetimes in TMG

by coordinate transformations is achieved f12 (u) → 0. Next, the equation

(19.24e) for F0 (u, r) becomes
 
∂ ∂2 df11 1
e μ r 2 F0 = μ + μ e−μ r f11 2 . (19.45)
∂r ∂r du 2
Taking into account that
  
(1 + μ r) e−μ r r x
(2 + μ r) e−μ r
e−μ r rdr = − , ze−μ z dzdx =
μ2 μ3
one arrives at
 
1 df11
F0 (u, r) = 2 + f01 (u) e−μ r + f02 (u) r + f03 (u)
μ2 du
1 1 df11
− 2 f11 2 e−2 μ r + e−μ r r , (19.46)
8μ μ du
which, by means of coordinate transformations, can be brought to the form
1 1 df11
F0 (u, r) = f01 (u) e−μ r − f11 2 e−2 μ r + e−μ r r . (19.47)
8μ2 μ du
Finally, the resulting metric can be given as
ds2 = dr2 + 2du dv + v f11 (u) e−μ r + F0 (u, r) (19.47) du2 . (19.48)
This solution is characterized by a Cotton tensor Cαβ and a traceless Ricci tensor
Sαβ fulfilling the relation Sαβ + μ1 Cαβ = 0, expressible as
C α β = C v u δαv δu β + C v r δαv δr β + C r uδαr δu β , (19.49a)
 
μ −μ r 2 −μ r df11 df11
C u=− e
v
f11 e − 2 μ v f11 − 2 μ f01 + 4
2 2
− 2μr ,
4 du du
(19.49b)
2
μ
Cvr = Cru = f11 e−μ r . (19.49c)
2

19.7.1 Type N pp-Wave Limit

For vanishing function f11 one gets a type N metric
ds2 = dr2 + 2du dv + f01 (u) e−μ r du2 (19.50a)
characterized by a Petrov type N Cotton tensor
μ3 −μ r
C α β = C v u k α kβ , C v u = e f01 , k α = δ α v . (19.50b)
2

19.8 Type III Kundt Solution; Λ = 0, W1 = −2/r

Evaluating the function F2 , (19.24c), for W1 given by (19.41b), one gets
 19.8 Type III Kundt Solution; Λ = 0, W1 = −2/r 363

2 1
W1 (u, r) = − → F2 = 2 . (19.51)
r r
Substituting this function into the F1 (u, r)–equation (19.24d) one gets
 
∂ μ r ∂F1
e r = 0, (19.52)
∂r ∂r
therefore, integrating

e−μ r
F1 (u, r) = − dr f11 (u) =: Ei (1, μ r) f11 (u), (19.53)
r
where Ei(x) is the exponential integral,
 r −μ x
e d e−μ r
Ei (1, μ r) = − dx, Ei (1, μ r) = − . (19.54)
x dr r
For the determined structural functions, the F0 (u, r)–equation (19.24e) becomes
  2 
∂3 3 ∂ 2 ∂ 2
F0 + μ − F0 − F0 + 2 F0 = Rhs(u, r),
∂r3 r ∂r2 r ∂r r
1 1 df
f11 2 Ei (1, μ r)e−μ r +
11 −μ r
Rhs(u, r) = e . (19.55)
2r r du
Since the solution of the homogeneous equation is proportional to r, explicitly

F0 H (u, r) = r e−μ r f01 (u) + r2 f02 (u) + rf03 (u) (19.56)

then, searching for the solution to the non-homogeneous equation, it is more

convenient to introduce an auxiliary function Φ(u, r) through

F0 (u, r) = Φ (u, r) r. (19.57)

In terms of the function Φ the equation (19.55) becomes

 
∂ ∂2 eμ r
eμ r 2 Φ = Rhs , (19.58)
∂r ∂r r
consequently
 r s  z 
eμ t
ΦN H (u, r) = Rhs (u, t) dt e−μ z dz ds,
t
ΦH = e−μ r f01 (u) + rf02 (u) + f03 (u) , (19.59)

accomplishing a partial integration one obtains


1 2 1 2
Φ = f11 2 (Ei (1, μ r)) dr − (f11 ) r Ei (1, 2 μ r)
4 2
 r s
μ df11
− f11 2
Ei (1, μ z)e−μ z dz ds + r Ei (1, μ r)
2 du
−2 μ r
1 2 e df11 e−μ r
+ (f11 ) − . (19.60)
4 μ du μ
364 Kundt Spacetimes in TMG

Finally the corresponding F0 (u, r) function can be given as

F0 (u, r) = r Φ(19.60) + re−μ r f01 (u) + r2 f02 (u) + r f03 (u) . (19.61)

Solution via the Variation of Parameters (VP)

Consider an ordinary non-homogeneous third-order equation of the form
d3 X(r) d2 X(r) dX(r)
3
+ p2 (r) + p1 (r) + p0 (r)X(r) = Rhs(r) (19.62)
dr dr2 dr
and assume that a set of three independent solutions {X1 (r), X2 (r), X3 (r)} of the
homogeneous equation has been determined; then the non-homogeneous (NH)
solution is sought as a linear combination of them,

XN H = C1 X1 + C2 X2 + C3 X3

where the coefficients Ci = Ci (r), i = 1, 2, 3 are considered at this stage as

d
functions of the variable r and that their derivatives dr Ci (r) = Ci  fulfill the set
of equations:
Rhs  
C1  X1  + C2  X2  + C3  X3  = Rhs, C3  = X1 X2  − X2 X1  , (19.63a)
W
Rhs  
C1  X1 + C2  X2 + C3  X3 = 0, C1  = X2 X3  − X3 X2  , (19.63b)
W
Rhs  
C1  X1  + C2  X2  + C3  X3  = 0, C2  = X3 X1  − X1 X3  , (19.63c)
W
 
where W = det [Xi ], [Xi  ], [Xi  ] is the Wronskian of the system. Integrating
the derivatives Ci  , one gets the functions Ci (r) and consequently one determines
the non-homogeneous solution, which, added to the homogeneous one, leads to
the general solution.
One may use the variation of parameters procedure to derive the solution of
∂3 ∂2
3
Φ (u, r) + μ 2 Φ (u, r) = rhsΦ ,
∂r ∂r
1 −μ r df11 1
rhsΦ = 2 e + 2 Ei (1, μ r) f11 2 e−μ r . (19.64)
r du 2r
The independent solutions to the homogeneous equation for Φ are given by
* +
Φ1 = 1, Φ2 = r, Φ3 = e−μ r . (19.65)

The solution for Φ to the non-homogeneous equation is sought in the form

ΦN H (u, r) = Φ1 C1 (u, r) + Φ2 C2 (u, r) + Φ3 C3 (u, r). (19.66)

Via the variation of the parameters Ci (u, r), their derivatives are
∂C1 rhsΦ (1 + μ r) ∂C2 rhsΦ ∂C3 rhsΦ eμ r
=− , = , = .
∂r μ2 ∂r μ ∂r μ2
 19.8 Type III Kundt Solution; Λ = 0, W1 = −2/r 365

Integrating these equations, one arrives at


e−μ r df11 f11 2 Ei (1, μ r) e−μ r f11 2 2
C1 (u, r) = 2 − dr + (Ei (1, μ r)) ,
μ r du 2 μ2 r2 4μ
 −μ r  
df11 e 1 2 Ei (1, μ r) e−μ r
C2 (u, r) = − + Ei (1, μ r) + f11 dr,
du μr 2 μ r2
1 df11 1  
C3 (u, r) = − 2 + 2
f11 2 e−μ r − Ei (1, μ r) (μ r + 1) .
μ r du 2μ r
Taking into account that the integral
 −μ r
e Ei (1, μ r) e−μ r Ei (1, μ r) 1 2
dr = − + (Ei (1, μ r)) μ
r2 r 2
e−2 μ r
−2 Ei (1, 2 μ r) μ +
r
one obtains the final expression of Φ(u, r), namely
1  −μ r μr 2

Φ= −e Ei (1, μ r) + (Ei (1, μ r)) − Ei (1, 2 μ r) (μ r − 1) f11 2
μ 4
df11 1 −2 μ r 2 1 df11 −μ r
+ Ei (1, μ r) r + e f11 − e
du 2μ μ du
+f01 (u)e−μ r + f02 (u) r + f03 (u),
F0 (u, r) = r Φ(u, r), (19.67)
the arbitrary integration functions f02 (u) and f03 (u) of the variable u can be
canceled by coordinate transformations. The metric is given by
4v v2
ds2 = dr2 + 2du dv − du dr + 2 du2 + [vEi(1, μ r)f11 + r Φ(u, r)(19.67)] du2 .
r r
(19.68)
This solution is characterized by the Cotton and traceless Ricci tensors, fulfilling
Sαβ + μ1 Cαβ = 0, of the form
C αβ = C v u δαv δuβ + C r u δαr δu β + C v r δαv δr β ,
μ
C v u = − (1 + μ r) f11 2 Ei (1, μ r) e−μ r
4
v μ (μ r + 1) μ μ df11 −μ r μ3
+ 2
f11 e−μ r + f11 2 e−2μ r − e + rf01 e−μ r ,
2 r 4 2 du 2
μ
Cvr = Cru = f11 e−μ r . (19.69)
2r
Generically, this metric is Petrov type III (f11 = 0).

19.8.1 Type N Limit

The type N branch arises for f11 = 0: its metric is given by
4v v2
ds2 = dr2 + 2du dv − du dr + 2 du2 + r f01 (u) e−μ r du2 , (19.70a)
r r
366 Kundt Spacetimes in TMG

with Cotton tensor components

μ3
C αβ = rf01 e−μ r k α kβ , k α = δ α v , k α kα = 0. (19.70b)
2
Flat spacetime occurs for f01 = 0(= f11 ):
4v v2 r2
ds2 = dr2 + 2du dv − du dr + 2 du2 → ds2 = dr2 + 4 du dṽ,
r r (u − ṽ)2
where it has been used the transformation v = 2r2 /(u − ṽ).

19.9 Type III Kundt Solution; Λ = −m2 , W1 = −2m

In the presence of Λ and constant structural function W1 , the only possible case
is the one with negative Λ, Λ = −m2 ; the equation (19.40) is solved by (19.41c),
on the other hand, the equation for F2 , (19.24c), gives F2 = 0, thus

W1 = −2m → F2 = 0, (19.71a)

while the equation (19.24d) for F1 amounts to

∂ 2 F1
+ (μ + m)F1 = 0. (19.71b)
∂r2
The solutions of this equation depend on the relationship between μ and m, and
determine the features of the whole corresponding metrics; in the forthcoming
paragraphs each possible branch is determined on its own account. In the deriva-
tion of the various branches of solutions one may adopt the following pattern:
first, integrate the general case μ = ± m, next particularize the derived results
and key equations to the sub-cases μ = ± m and proceed to integrate them.

19.9.1 Type III Kundt Solution; Λ = −m2 , W1 = −2m, μ = ±m

For this class of solutions, the function F1 fulfilling (19.71b) is

F1 (u, r) = e−(μ+m)r f11 (u) + f12 (u), f12 (u) → 0, (19.72a)

which, substituted in the equation (19.24e) for F0 , yields

  
r(−μ+m) ∂ (m+μ)r ∂ −2 m r ∂F0
e e e
∂r ∂r ∂r
 
1 df11 −(μ+m)r 2 −2 (μ+m)r
= (μ + m) 2 e + f11 e , (19.72b)
2 du
with a general solution
e−(m+μ)r df11
F0 (u, r) = e(m−μ)r f01 (u) + e2mr f02 (u) + f03 (u) −
2m(μ + 3m) du
 19.9 Type III Kundt Solution; Λ = −m2 , W1 = −2m 367

e−2 (m+μ)r
− f11 2 . (19.72c)
8(μ + 2m)(μ + 3m)
Finally one may write the corresponding Kundt solution as
 
ds2 = dr2 + 2dudv − 4m vdu dr + v e−(μ+m)r f11 + F0 (u, r) du2 ,
e−(m+μ)r df11 e−2 (m+μ)r f11 2
F0 = e(m−μ)r f01 − − , (19.72d)
2m(μ + 3m) du 8(μ + 2m)(μ + 3m)
where f02 and f03 have been canceled by means of coordinate transformations.
This solution is characterized by a Petrov type III Cotton tensor
C αβ = C v u δαv δu β + C r u δαr δu β + C v r δαv δr β ,
1 μ (μ + m) −2 (μ+m)r v
f11 2 + μ (μ + m) f11 e−(μ+m)r
2
Cvu = − e
4 3m + μ 2
1 μ (μ + m) −(μ+m)r df11 1  
− e + μ μ2 − m2 f01 e(−μ+m)r
4 m du 2
v r 1 −(μ+m)r
C r = C u = μ (μ + m) f11 e . (19.72e)
2

Type N Limit
For f11 = 0 this solution becomes a Petrov type N spacetime with metric
ds2 = dr2 + 2dudv − 4m vdu dr + e(m−μ)r f01 (u) du2 , (19.73a)
and Cotton tensor
1  
C α β = μ μ2 − m2 f01 e(−μ+m)r k α kβ , k α = δ α v . (19.73b)
2

19.9.2 Type III Kundt Solution; Λ = −m2 , W1 = −2m, μ = m

For this class of solutions, the function F1 fulfilling (19.71b) is
F1 (u, r) = e−2 m r f11 (u), f12 (u) → 0, (19.74a)
which, substituted in the equation (19.24e) for F0 , yields
  
∂ ∂ ∂F0 df11 −4 m r
e2 m r e−2 m r = 2m e + mf11 2 e−6 m r (19.74b)
∂r ∂r ∂r du
with general solution
e−2 m r df11 e−4 m r
F0 (u, r) = r f01 (u) + e2mr f02 (u) + f03 (u) − − f11 2 . (19.74c)
8 m2 du 96 m2
Finally one may write this branch of Kundt solutions as
ds2 = dr2 + 2dudv − 4m vdu dr + v e−2 m r f11 (u) + F0 (u, r) du2 ,
e−2 m r df11 e−4 m r
F0 = r f01 (u) − − f11 2 , (19.74d)
8 m2 du 96 m2
368 Kundt Spacetimes in TMG

where f02 and f03 have been canceled by coordinate transformations. This solu-
tion is characterized by the type III Cotton and traceless Ricci tensor of the
form

C αβ = C v uδαv δu β + C r u δαr δu β + C v r δαv δr β ,

m m −2 m r df11
C v u = − e−4 m r f11 2 + 2 v m3 f11 e−2 m r − e
8 2 du
−m2 (f01 + f11 ) , C v r = C r u = m2 f11 e−2 m r . (19.74e)

Type N Limit
Nevertheless, for f11 = 0 this spacetime becomes a type N solution given by

ds2 = dr2 + 2dudv − 4m vdu dr + r f01 (u)du2 , (19.75a)

with a type N Cotton tensor

C α β = −m2 f01 k α kβ , k α = δ α v , k α kα = 0. (19.75b)

Moreover, for f01 = 0 the metric describes an AdS spacetime – a type O metric
field – given by

ds2 = dr2 + 2dudv − 4m vdu dr = dr2 + 2v du d(ln v − 2mr) = dr2 + 2v du dV,

which can be reached via the coordinate transformation V = ln v − 2mr, v =

eV +2mr .

19.9.3 Type III Kundt Solution; Λ = −m2 , W1 = −2m, μ = −m

For this class of metrics with μ = −m, the function F1 fulfilling (19.71b),
∂2
∂r 2 F1 (u, r) = 0, is given by

F1 (u, r) = r f11 (u). (19.76a)

The evaluation of F2 , (19.24c), yields F2 = 0, therefore, using these results in

F0 , (19.24e), leads to
∂ 3 F0 ∂F0 ∂ 2 F0 1 2 df11
3
+ 4 m2 − 4 m 2 = − (f11 ) r − , (19.76b)
∂r ∂r ∂r 2 du
with general solution
r df11 r(2 + mr)
F0 = (r f01 (u) + f02 (u)) e2 m r + f03 (u) − 2
− f11 2 . (19.76c)
4m du 16 m3
Finally this Petrov type III Kundt solution can be written as

ds2 = dr2 + 2dudv − 4m vdu dr + [v r f11 (u) + F0 (u, r)] du2 ,

r df11 r(2 + mr)
F0 = re2mr f01 (u) − 2
− f11 2 , (19.76d)
4m du 16 m3
 19.10 Type III Kundt Metric; Λ = −m2 , W1 = −2 m coth(m r) 369

where the functions f02 (u) and f03 (u) have been eliminated via coordinate
transformations.
This solution is characterized by the Cotton and traceless Ricci tensor of the
form

C αβ = C v uδαv δu β + C r u δαr δu β + C v r δαv δr β ,

1 df11 1 1 + 2mr 2
C v u = −m2 e2 m r f01 − − (f11 ) ,
4 du 16 m
1
C v r = C r u = mf11 . (19.76e)
2

Type N Limit
Again, as in the previous branch of solutions, one finds the type N subcase arising
for f11 = 0, namely

ds2 = dr2 + 2dudv − 4m vdu dr + re2mr f01 (u)du2 , (19.77a)

with a type N Cotton tensor

C α β = −m2 e2 m r f01 k α kβ , k α = δ α v . (19.77b)

Moreover, for f01 = 0 the metric describes an AdS Type O spacetime.

19.10 Type III Kundt Metric; Λ = −m2 , W1 = −2 m coth(m r)

The equation for vanishing eigenvalue S, (19.24b), which guarantees the Petrov
type III of the Cotton tensor, for negative cosmological constant Λ = −m2 ,
allows for a solution of the form (19.41d), and evaluating the equation for F2
(19.24c) one gets
m2
W1 (u, r) = −2 m coth (m r) → F2 = . (19.78)
sinh2 (m r)
Replacing these functions into the F1 equation (19.24d) one arrives at
∂2 ∂
2
F1 (u, r) + [m coth (m r) + μ] F1 (u, r) = 0 (19.79)
∂r ∂r
with solution

m e−μ r
F1 (u, r) = f11 (u) dr = f11 ξ(r), (19.80)
sinh (m r)
where it has been introduced the auxiliary function ξ(r),

dξ me−μ r m e−μ r
= → ξ (r) = dr. (19.81)
dr sinh (m r) sinh (m r)
370 Kundt Spacetimes in TMG

The key equation (19.24e) for F0 (u, r) becomes

∂3 ∂2
F 0 + [μ − 3m coth (m r)] F0
∂r3 ∂r2
∂
−2 m μ coth (m r) + 2 m − 3 m coth2 (m r) F0
∂r
+2 m2 μ coth2 (m r) + 3 m coth (m r) − 3 m coth3 (m r) − μ F0 = RHS,
 
1 df11 d
RHS := − 2 + f11 2 ξ (r) ξ (r) . (19.82)
2 du dr
Following the standard procedure of solving non-homogeneous linear ordinary
differential equations by means of the variation of parameters of the determined
general homogeneous solution, which in the present case amounts to
 
F0 = sinh (m r) C1 (u) e−μ r + C2 (u) e−m r + C3 (u) em r , (19.83)
one should arrive at an algebraic system of equations for the first-order deriva-
tives with respect to the variable r of Ci (u, r) considered at this stage of function
of r too, beside of u. The integration of these derivatives, which depend on the
RHS function, allows one to construct the non-homogeneous solution which
added up to the homogeneous one, gives the sought general solution. Neverthe-
less, the structure of the homogeneous solution above points out to a better
function choice, namely of an auxiliary function , Φ (u, r), in terms of which
F0 (u, r) = sinh (m r) Φ (u, r) . (19.84)
The key equation (19.82) in terms of Φ becomes
∂3 ∂2 ∂ RHS
3
Φ + μ 2 Φ − m2 Φ − m2 μ Φ = =: Rhs. (19.85)
∂r ∂r ∂r sinh (m r)
In what follows we proceed to determine the three different branches of solutions
in the specific cases of μ = ± m, μ = − m, and μ = m separately.

19.10.1 Type III Kundt Solution; Λ = −m2 ,

W1 = −2m coth(m r), μ = ±m
Solution Through the Variation of Parameters
The homogeneous solution to the homogeneous equation (19.85) (Rhs → 0), one
searches in the form of ΦH ek r arriving at the characteristic equation
k 3 + μ k 2 − m2 k − m2 μ = 0 = (k + μ)(k − m)(k + m)
with solutions k = −μ, −m, m, hence ΦH is given by
Φ (r) = C1 e−μ r + C2 em r + C3 e−m r . (19.86a)
Thus the set of independent solutions amounts to
* +
Φl = e−μ r , Φ2 = em r , Φ3 = e−m r (19.86b)
 19.10 Type III Kundt Metric; Λ = −m2 , W1 = −2 m coth(m r) 371

Looking for the solution to the non-homogeneous equation (Rhs = 0) – the

non-homogeneous solution – one assumes it to be of the form

Φ (r) = C1 (u, r) Φl + C2 (u, r) Φ2 + C3 (u, r) Φ3 , (19.86c)

where u enters here as a parameter.

The variation of the parameters Ci (u, r), within the VP procedure, yields

∂C1 Rhs eμ r m f11 2 ξ (r) + 2 du
d
(f11 )
= 2 , C1 = dr,
∂r μ − m2 2 (m2 − μ2 ) sinh2 (mr)

∂C2 1 Rhs e−m r 1 f11 2 ξ (r) + 2 dud
(f11 )
= , C2 = − e−(μ+m) r 2 dr,
∂r 2 m (m + μ) 4 (m + μ) sinh (m r)
 (m−μ) r
∂C3 1 Rhs em r 1 e f11 2 ξ (r) + 2 du
d
(f11 )
= , C3 = − dr,
∂r 2 m (m − μ) 4 (m − μ) sinh (m r)
2

where it has been used dξ/dr from (19.81).

Notice the presence of μ ± m in the denominators of the above expressions,
which clearly point on the degeneracy of these quantities when dealing with
the sub-branches μ = ±m; this is the reason how the quoted cases has to be
integrated separately for the Φ-equation (19.85). Continuing with the general
case μ = ±m, one may factorize the terms in the form
 −μ r 
df11 e dr
ΦN H =
du m2 − μ2 sinh2 (mr)
 −(m+μ)r  
1 emr e 1 e−mr e(m−μ)r
− dr − dr
2 (m + μ) sinh2 (mr) 2 (m − μ) sinh2 (mr)
   (m−μ)r
1 e−μr ξ (r) 1 e−m r e ξ (r)
+f11 2 dr − dr
2 m2 − μ2 sinh2 (mr) 4 (m − μ) sinh2 (mr)
 −(m+μ)r 
1 emr e ξ (r)
− dr , (19.86d)
4m+μ sinh2 (m r)
whereas the homogeneous solution is given by

ΦH (u, r) = f01 (u) e−rμ + f02 (u) em r + f03 (u) e−m r , (19.86e)

which together determine the general metric

ds2 = dr2 + 2du dv + (v 2 F2 + vF1 + F0 )du2 − 4 m v coth(m r) du dr,

m2
F2 (u, r) = ,
sinh2 (m r)

m e−μ r
F1 (u, r) = f11 (u)ξ(r), ξ(r) = dr,
sinh(m r)
F0 (u, r) = sinh(m r) [ΦH (u, r) + ΦN H (u, r)(19.86d)] . (19.86f)
372 Kundt Spacetimes in TMG

This solution is characterized by the Cotton:

C αβ = C v uδαv δu β + C v r δαv δr β + C r uδαr δu β ,
μ  2 
Cvu = μ − m2 f03 (u) e−μ r sinh (m r)
2
 r  y  
μ m2 −μ r 1 e−μ z
− 2
f11 sinh (m r) e dz dy
4 sinh2 (m y) sinh (m z)
μmv m cosh(m r) + μ sinh(m r) μ df11 −μ r
− f11 e−μ r + e cosh (m r) ,
2 sinh2 (m r) 2 du
mμ e−μ r
Cru = Cvr = − f11 . (19.86g)
2 sinh (m r)
It should be pointed out that via the variations of parameters, the general
solution is determined by a linear combinations of linear (single) integrals.

Type N Limit
The solution presented above (19.86f) allows for a type N Kundt subcase with
metric
ds2 = dr2 + 2du dv + (v 2 F2 + F0 )du2 − 4 m v coth(m r) du dr,
m2
F2 (u, r) = , F0 (u, r) = sinh(m r) f03 (u) e−rm , (19.87a)
sinh2 (m r)
with a type N Cotton tensor, JCN ,
μ  2 
C α β = C v u k α kβ , , C v u = μ − m2 f03 (u) e−μ r sinh (m r) , k α = δ α v .
2
(19.87b)

Multi Exponent–Integral Representation of Φ; General Procedure

For a constant coefficients’ equation involving up to n, i = 1, · · · , n, order
derivatives
n
di
ai i φ(r) = f (r), i = 0, 1, · · · , n (19.88)
i
dr

where without loss of generality one may assume the highest coefficient an = 1,
beside the VP procedure based on the determination of the roots of the char-
acteristic equation, one can search for an expression of the equation of the
form
  
d d d d α1 r
eαn r eαn−1 r · · · (e φ(r)) = f (r)e(α1 +α2 +···+αn )r , (19.89)
dr dr dr dr
d
with n in number derivative operators dr . Comparing the coefficients ai of (19.88)
di
with those of (19.89) multiplying derivatives dr i φ of the same order in both

equations, one obtains a system of equations for αi in terms of the known ai , of

the form
 19.10 Type III Kundt Metric; Λ = −m2 , W1 = −2 m coth(m r) 373

1 = 1,
nα1 + (n − 1)α2 + · · · + 1 αn = an−1 ,
........................................................
α1 (α1 + α2 ) · · · (α1 + α2 + · · · + αn−1 + αn ) = a0 , (19.90)
whose solutions determine the exponents. Finally, integrating (19.89), one arrives
at a compact expression for the main part of the non-homogeneous solution
 r   y  
−α1 r −α2 t
φN H (r) = e e ··· f (z)e (α1 +α2 +···+αn ) z
dz dy · · · dt.
(19.91)
The general solution for the equation (19.88) with constant coefficients is just the
sum of φN H (r) and the n-independent solutions to the homogeneous equation,
which are of the exponential form eαi r multiplied by powers of r in the case of
degeneracy.
This procedure will be used extensively in the derivation of the solutions to
the non-homogeneous equations, thus we shall refer to this subsection as 19.10.1.

Multi Exponent–Integral Representation of Φ(u, r)

Returning to the search of the non-homogeneous solution to the equation (19.85)
written in multi-exponent form
  
∂ ∂ ∂ αr
eγr eβr (e Φ(u, r)) = Rhs(u, r) e(α+β+γ)r (19.92a)
∂r ∂r ∂r
or accomplishing the differentiations
∂3 ∂2   ∂
3
Φ + (γ + 3 α + 2 β) 2 Φ + 2 γ α + β 2 + 3 α2 + 4 β α + γ β Φ
∂r  ∂r  ∂r
+ γ β α + γ α2 + 2 β α2 + β 2 α + α3 Φ = Rhs (u, r) . (19.92b)
Comparing the coefficients multiplying derivatives of the same order in the
expanded equation (19.92b) with the corresponding of (19.85), one gets the
system
2 β + 3 α + γ = μ,
γ β + 4 β α + 3 α2 + 2 γ α + β 2 = −m2 ,
γ β α + γ α2 + β 2 α + α3 + 2 β α2 = −m2 μ, (19.92c)
with the following set of solutions
{α = ∓ m, β = ± 2 m, γ = ∓ m + μ} , sum := α + β + γ = μ,
{α = μ, β = −μ ∓ m, γ = ± 2 m} , sum = ± m,
{α = ∓ m, β = μ ± m, γ = ± m − μ} , sum = ± m, (19.92d)
where the upper (lower) signs match with the upper (lower). The substitution
of the found exponents in the expression
374 Kundt Spacetimes in TMG
 r  t  y
Φ(u, r) = e−αr (e−βt (e−γy (Rsh(u, z) ez(γ+β+α) )dz)dy)dt, (19.92e)

determines an specific representation of the sought non-homogeneous solution.

For instance, for the first set of exponents, one has
        
df11 m r r −2m t t −(μ−m) y y eμ z dξ
Φ=− e e e dz dy dt
du sinh (m z) dz
        
f11 2 m r r −2m t t −(μ−m) y y eμ z ξ (z) dξ
− e e e dz dy dt,
2 sinh (m z) dz
(19.92f)

which determines in turn the non-homogeneous F0 (u, r) function through

F0N H (u, r) = sinh(mr) Φ(u, r). (19.92g)

Consequently, the general solution

 
F0 (u, r) = sinh(m r) e−μ r f01 + em r f02 + e−m r f03 + Φ(u, r) . (19.92h)

Of course, this expression can be simplified in various manners; for instance one
dξ
can partially integrate the first term in Φ(u, r) using the derivative dr from
(19.81).
Finally, the metric becomes

ds2 = dr2 + 2du dv + m2 v 2 cosech2 (mr)du2 − 4 m v coth(m r) du dv

+[vξ(r)f11 (u) + F0 (u, r)(19.92h)]du2 . (19.92i)

VP of the CPS Representation of F0 (u, r) through φ

The Chow et al. (2010b) representation of the solution for F0 (u, r) is achieved
via an alternative definition of the function F0 (u, r) through a new auxiliary
function φ(u, r) according to
F0 (u, r) F0 (u, r)
φ(u, r) = −2mr
, Φ(u, r) = = 2 e−mr φ(u, r), (19.93a)
1−e sinh(mr)
which brings the equation (19.85), (19.24e), to the form
∂3 ∂2 ∂ RHS
φ + (μ − 3 m) 2 φ + 2 m (m − μ) φ = = Rhs(u, r)
∂r 3 ∂r ∂r (1 − e−2 mr )
 
me−(μ+m)r f11 2 ξ (r) + 2 du
d
f11
=− 2 . (19.93b)
(1 − e−2 mr )
The solution to the homogeneous equation is given by

φH (u, r) = C1 (u) + C2 (u) e2 mr + C3 (u) e(−μ+m)r

≡ C1 (u)φ1 + C2 (u) φ2 + C3 (u) φ3 . (19.93c)
 19.10 Type III Kundt Metric; Λ = −m2 , W1 = −2 m coth(m r) 375

The non-homogeneous solution is searched in the form of

φN H (r) = C1 (u,r)φ1 + C2 (u, r) φ2 + C3 (u, r) φ3 (19.93d)
via the variation of the parameters Ci (u, r), where u plays the role of a parameter.
Thus their first derivatives are
∂C1 Rhs ∂C2 e−2mr ∂C3 e−(m−μ)r
= , = Rhs, =− 2 Rhs.
∂r 2m(m − μ) ∂r 2m(m + μ) ∂r (m − μ2 )
(19.93e)
Therefore the general φ(u, r) solution to the key equation (19.93b) via the
variation of parameters method is

1 r(m−μ) mer(m−μ) 2 du d
f11 + f11 2 ξ (r) e−r(m−μ)
φNH = e dr
4 sinh2 (m r) (m2 − μ2 )
 r(m−μ) d
1 e 2 du f11 + f11 2 ξ (r) e−2 m r
− e2 m r dr
8 sinh2 (m r) (μ + m)
 r(m−μ) d
1 e 2 du f11 + f11 2 ξ (r)
− dr, (19.93f)
8 sinh2 (m r) (m − μ)
which is a linear (single) integral representation of the solution under consider-
ation. Of course, this representation is far from the triple integral CPS Chow et
al. (2010b) representation (19.94j) of the next paragraph 19.10.2.

19.10.2 Multi Exponent–Integral Representation of φ

To obtain a CPS Chow et al. (2010b) representation of the solution to the
key equation (19.93b) in terms of the function φ one has to follow the multi-
exponent–integral procedure previously detailed in 19.10.1. Introducing an
∂
auxiliary function H = ∂r φ we rewrite the equation (19.93b) as
 
∂ βr ∂
e (e H(u, r)) = Rhs(u, r) e(α+β) r ,
αr
∂r ∂r
 r  t
H(u, r) = e−α r e−β t Rhs(u, z) e(α+β) z dzdt, (19.94a)

accomplishing the derivation one gets

∂2 ∂
H + (β + 2 α) H + α (β + α) H = Rhs. (19.94b)
∂r2 ∂r
Comparing the coefficients of (19.94b) with those of (19.93b) one gets the system
{α (β + α) = −2 m (μ − m) , β + 2 α = μ − 3 m} (19.94c)
with solutions
[[α = −2 m, β = m + μ], [α = μ − m, β = −m − μ]]. (19.94d)
376 Kundt Spacetimes in TMG

For the first set of solutions one obtains

 r  t
2m r −(μ+m) t
H(u, r) = e e Rhs(u, z)e(μ−m) z dzdt. (19.94e)

The integral for φ is

 
 r  t  y m 2 dfdu11 + f11 2 ξ (z)
1
φ(u, r) = − e2 m t e−(m+μ)y dzdydt, (19.94f)
4 sinh2 (mz)
which leads to the CPS F0 (u, r) from (19.93a),

F0 (u, r) = 2 e− m r sinh (mr) φ(u, r). (19.94g)

Incidentally an integration of 1/sinh2 (m r) in the term with factor du d

f11 can be
carried out, hence one may write
 r  t −(μ+m) z
1 df11 2mt e cosh (m z)
φ= e dzdt
2 du sinh (m z)
 r  t  y
m ξ (z)
− f11 2 e2 m t e−(μ+m)y dzdydt; (19.94h)
4 sinh2 (m z)
moreover, one has the relation
e−r(μ+m) cosh (m r) e−μ r
= − e−(μ+m) r
sinh (m r) sinh (rm)
d
therefore, the integration of this term in the factor of du f11 yields
 −r(μ+m) 
e cosh (m r) e−μ r e−μ r−m r
dr = dr +
sinh (m r) sinh (m r) μ+m
which leads to the sought solution φ in the form

φ = φH + φN H = f03 (u) + f02 (u)e2mr + f01 (u)e(m−μ)r

 r
1 df11 
+ dt e2 m t ξ(t )
2 m du
 r  t  t
m  ξ(t )
− f11 2 e2mt dt dt e−(m+μ)t dt , (19.94i)
4 sinh2 (mt )
which coincides, modulo the identification ξ → F , with the triple integral CPS
solution representation given by the expression

F0 (m−μ)r 2mr 1 df11 r 2mt
= e f01 + e f02 + f 03 + e F (t)dt
1 − e−2mr 2m du ∞
  t  t 2m(t−t ) (m−μ)t
m 2 r   e e
− f11 dt dt dt F (t ), (19.94j)
4 ∞ ∞ ∞ sinh (mt )
2
 19.10 Type III Kundt Metric; Λ = −m2 , W1 = −2 m coth(m r) 377

which determines the metric CPSb(A.32),

ds2 = dr2 + 2du dv + (v 2 F2 + vF1 + F0 )du2 − 4 m v coth(m r) du dr,


m2 m e−μ r
F2 (u, r) = 2 , F1 (u, r) = f11 (u)ξ(r), ξ(r) = dr,
sinh (m r) sinh(m r)
F0 (u, r) = (1 − e−2mr ) φ(19.94i). (19.94k)

19.10.3 Type III Kundt Solution; Λ = −m2 ,

W1 = −2m coth(m r), μ = −m
Solution F0 through Φ via Variation of Parameters
The equation (19.85) for the function Φ(u, r) in the case μ = −m becomes

∂3 ∂2 ∂ RHS
3
Φ − m 2 Φ − m 2 Φ + m3 Φ = = Rhs
∂r ∂r ∂r sinh (m r)
2 d
1 mem r f11 ξ (r) + 2 du f11
=− 2 . (19.95a)
2 sinh (m r)

Since the characteristic equation (k + m)(k − m)2 = 0 allows for a double root,
then the homogeneous solution is

Φ( H) (u r) = C1 e−m r + C2 em r + C3 r em r . (19.95b)

Using the VP procedure for the non-homogeneous proposal

ΦN H (r) = C1 (u,r) e−m r + C2 (u,r) em r + C3 (u,r) r em r (19.95c)

one arrives at first-order equations for Ci (u,r)

∂ Rhs em r ∂ Rhs (2 m r + 1) e−m r ∂ e−m r Rhs

C1 = 2
, C2 = − 2
, C3 = .
∂r 4m ∂r 4m ∂r 2m
d 
Hence, factorizing terms with du f11 and f11 2 one may write the non-
homogeneous solution as
  
1 df11 −m r e2m r 2mr + 1
ΦN H (u, r) = − e dr − e mr
dr
4 du m sinh2 (m r) m sinh2 (m r)
 
mr 1
+2 r e dr
sinh2 (m r)
  
1 e2 m r ξ (r) (2 mr + 1) ξ (r)
− f11 2 e−m r 2 dr − e mr
dr
8 m sinh (mr) m sinh2 (mr)
 
ξ (r)
+2 r em r dr ,
sinh2 (mr)
 
ξ (r) = ln 1 − e−2 m r = −m r + ln (sinh (m r)) . (19.95d)
378 Kundt Spacetimes in TMG

Multi Exponent–Integral Representation of Φ

Let us search for a multi-exponent–integral representation, see 19.10.1, of the
solution to the equation (19.95a). Comparing the coefficients of the derivatives
of the same order one arrives at the system

γ + 3 α + 2 β = −m,
2γ α + β 2 + 3 α2 + 4 β α + γ β = −m2 ,
γ β α + γ α 2 + 2 β α 2 + β 2 α + α 3 = m3 (19.96a)

with solutions

{α = m, β = −2 m, γ = 0} , {α = −m, β = 2 m, γ = −2 m, } ,
{α = −m, β = 0, γ = 2 m} . (19.96b)

The function Φ in this approach is expressible as

 r  t  y
Φ = e−α r e−β t e−γ y e(γ+β+α)z Rhs (u, z) dzdydt, (19.96c)

thus, for each set of exponential constants one gets its corresponding exponent–
integral representation.

VP of the CPS Representation of φ

The CPS representation of the solution for F0 (u, r) is achieved via the definition
of the function F0 (u, r) in terms of φ(u, r) as
F0 (u, r) 1
φ(u, r) = −2mr
→ φ(u, r) = emr Φ(u, r). (19.97a)
1−e 2
The equation (19.95a), (19.24e), for φ becomes

∂3 ∂2 2 ∂ f11 2 ξ (r) + 2 du
d
f11
φ − 4m φ + 4 m φ = −m 2 = Rhs. (19.97b)
∂r 3 ∂r 2 ∂r (1 − e−2 mr )
The solution to the homogeneous equation is given by

φH (u, r) = C1 (u) + C2 (u) e2 mr + C3 (u) e2 mr r

≡ C1 (u)φ1 + C2 (u) φ2 + C3 (u) φ3 . (19.97c)

The non-homogeneous solution is sought in the form of

φN H (u, r) = C1 (u,r)φ1 + C2 (u, r) φ2 + C3 (u, r) φ3 (19.97d)

via the variation of parameters Ci (u, r), where u plays the role of a parameter.
The first derivatives of Ci (u, r) are
∂C1 1 Rhs ∂C2 1 (2 m r + 1) e−2 m r ∂C3 1 e−2 m r
= , = − Rhs, = Rhs.
∂r 4 m2 ∂r 4 m2 ∂r 2 m
 19.10 Type III Kundt Metric; Λ = −m2 , W1 = −2 m coth(m r) 379

Therefore the general non-homogeneous φN H (u, r) solution to the key equation

(19.97b) via the variation of parameters is
1 ,    2 m r  2 
φN H = f11
2
e2mr
− 1 ln e − 1 + 1 − 1
16 m2     +
−2 dilog e2 m r e2 m r − 4 ln e2 m r − 1 m re2 m r − 4 m re2 m r − 2
1 df11  2 m r   
+ 2 e − 1 ln e2 m r − 1 − m r e2 m r (19.97e)
4m du
where it has been used in the relation

     
2 m ln e2 mr − 1 dr = dilog e2 mr + 2 m r ln e2 mr − 1

for the dilog function, which is presented in details in the next equations (19.99).
The above-integrated solution (19.97e) can be given in the CPS form,
F0 (u, r) (f11 )2  2 rm    2
= e −1 1 + ln e2 rm − 1 −1
1 − e2 rm 16m2
 
  1 df11  2 rm   
+2e2m r L2 e2m r + 2
e − 1 ln e2 rm − 1 .
4m du
(19.97f)
Finally, this family μ = −m of Petrov type III CSI spacetimes is given by the
metric
ds2 = dr2 + 2du dv + (v 2 F2 + vF1 + F0 )du2 − 4 v m coth(m r) du dr,
m2
F2 (u, r) = ,
sinh2 (m r)
 
F1 (u, r) = f11 (u)ξ(r), ξ (r) = ln e2 m r − 1 = m r + ln (sinh (m r)) ,
F0 (u, r) = (1 − e−2 rm ) [φH (u, r) + φN H (u, r)(19.97e)] . (19.97g)
This solution is characterized by the Cotton:
C αβ = C v u δαv δuβ + C v r δαv δr β + C r u δαr δuβ ,
  m3 e2 mr 1 2 mr df11
C v u = −m2 e2 mr − 1 f03 + 2 v f11 2 − 2 me
(e 2 mr − 1) du
1  
− m f11 2 ln e2 mr − 1 + e2 mr ,
4
e2 mr
C r u = C v r = m2 2 mr f11 . (19.97h)
e −1

Type N Limit
It allows for a type N Kundt solution with metric
ds2 = dr2 + 2du dv + (v 2 F2 + F0 )du2 − 4 v m coth(m, r) du dr,
m2
F2 = , F0 = 2f03 r sinh(m r) e−rm (19.98a)
sinh2 (m r)
380 Kundt Spacetimes in TMG

with a type N Cotton tensor, JCN ,

 
C α β = C v u δ α v δ u β , C v u = −m2 e2 mr − 1 f03 (u) . (19.98b)

The Dilogarithm dilog(x) and the Logarithmic Integral Li(x) Functions

By definition, the dilogarithm function dilog(x) is
 x
ln (t) d ln (x)
dilog (x) = dt, dilog (x) = , (19.99a)
1 1 − t dx 1−x

hence

  r  
dilog 1 + e± 2 m r = ∓ 2m ln 1 + e± 2 mt dt,
d    
dilog e±2 m r + 1 = ∓2 m ln e±2 m r + 1 , (19.99b)
dr

   
dilog e2 m r + dilog e−2 m r + 2 m2 r2 = 0,
   
dilog 1 + e−2 m r + dilog 1 + e2 m r + 2 m2 r2 = 0, (19.99c)


     
2 m ln e2 mr − 1 dr = dilog e2 mr + 2 m r ln e2 mr − 1 . (19.99d)

The logarithmic integral Li(x), denoted in CPS by Li2 (x), is defined as


ln (|1 − x|)
Li(x) = − dx (19.99e)
x
hence depending on the range of value assumed by the variable of integration, one
would have different functional relations with the dilog(x) and ln(x) functions:

● for x ≤ 1
 
Li(x) = dilog(1 − x), x ≤ 1; Li(e−2 m r ) = dilog 1 − e−2 mr ,
 
Li(−e2 m r ) = dilog 1 + e2 mr , (19.99f)

● while for x > 1

Li(x) = − dilog(x) − ln(x − 1) ln(x) + π 2 /6, x > 1;

   
Li(1 + e−2 mr ) = −dilog 1 + e−2 mr + 2 mr ln 1 + e−2 mr + π 2 /6
 
= −dilog 1 + e−2 mr − 4 m2 r2 + 2 mr ln(e2 mr + 1) + π 2 /6,
(19.99g)

see Abramowitz and Stegun (1965).

 19.10 Type III Kundt Metric; Λ = −m2 , W1 = −2 m coth(m r) 381

Multi Exponent–Integral Representation of φ

To obtain a CPS Chow et al. (2010b) representation of the solution φ to (19.97b)
one has to follow the exponent–multiple integral procedure previously detailed
in 19.10.1 or simple rewrite the equation (19.97b) as
 
∂2 −2m r ∂
e φ(u, r) = e−2m r Rhs (19.100a)
∂r2 ∂r
The integral for φ is
 r  t y
φ (u, r) = e2 m t e−2 m z Rhs (u, z) dz dy dt

which arises from (19.94f) for μ = −m, namely

    d 
1 r 2 m t t y m 2 du f11 + f11 2 ξ (z)
φ(u, r) = − e dzdydt (19.100b)
4 sinh2 (mz)
which yields to an expression comparable with CPS φ(u, r) giving rise to

F0 2mr 2mr 1 df11 r 2mt
= f01 + e f02 + re f03 + e F (t)dt
1 − e−2mr 2m du ∞
 r  t  t
m e2m t
− f11 2 dt dt dt 2  )
F (t ). (19.100c)
4 ∞ ∞ ∞ sinh (mt

19.10.4 Type III Kundt Solution;

Λ = −m2 , W1 = −2m coth(mr), μ = m
Solution Φ(u, r) through the Variation of Parameters
The equation (19.85) for the function Φ(u, r) in the case μ = m becomes
∂3 ∂2 ∂ RHS
Φ + m Φ − m2 Φ − m3 Φ = = Rhs
∂r3 ∂r2 ∂r sinh (m r)
 
m e−m r df11
=− f11 2 ξ (r) + 2 , (19.101a)
2 sinh2 (m r) du
where
d 2m
ξ(r) = 2 m r , ξ(r) = ln(1 − e−2 m r ). (19.101b)
dr e −1
Since the homogeneous solution is given as a linear combination of the set of
independent functions {em r , e−m r , r em r }, then the non-homogeneous solution
is of the form
ΦN H = C1 (u, r)em r + C2 (u, r)e−m r + C3 (u, r)em r r, (19.101c)
Using the VP procedure one arrives at
∂ e−m r RHS ∂ em r (2 m r − 1) RHS
C1 = , C2 = ,
∂r 4m2 sinh (m r) ∂r 4 m2 sinh (m r)
∂ em r RHS
C3 = − .
∂r 2 m sinh (m r)
382 Kundt Spacetimes in TMG

∂
Consequently, integrating ∂r Ci , the non-homogeneous Φ solution amounts to
1    
ΦN H = − 2
f11 2 −2 dilog e−2 rm + 4 rm ln 1 − e−2 rm
8m
    2 
− e2 rm − 1 1 + ln 1 − e−2 rm − 1 e−rm
 
e−rm df11  2 rm   
+ 2
e − 1 ln e2 rm − 1 − 2 m r e2 rm . (19.101d)
2m du

Multi Exponent–Integral Representation of Φ(u, r)

To determine the structural function Φ subjected to the equation
∂3 ∂2 ∂
Φ + m Φ − m2 Φ − m3 Φ = Rhs, (19.102a)
∂r3 ∂r2 ∂r
one follows the procedure given in 19.10.1.
Comparing the coefficients of the derivatives of the corresponding differential
equations one arrives at an algebraic system for the exponents

γ + 2 β + 3 α = m,
γ β + 2 γ α + β 2 + 4 β α + 3 α2 = −m2 ,
γ β α + γ α2 + β 2 α + 2 β α2 + α3 = −m3 (19.102b)

with the sets of solutions

{α = −m, β = 2 m, γ = 0} ,
{α = m, β = 0, γ = −2 m} ,
{α = m, β = −2 m, γ = 2 m} . (19.102c)

Therefore, one has various solution representations at one’s disposal.

VP of the CPS Representation of F0 (u, r) through φ

If one were using the variations of parameters method to determine the solu-
tion to the non-homogeneous equation one generalize the parameters Ci of the
homogeneous solution

φH (r) = C1 (u) + C2 (u)r + C3 (u) e2 m r ≡ C1 (u)φ1 + C2 (u) φ2 + C3 (u) φ3

(19.103a)
to be dependent on the variable r in such a way that the non-homogeneous
solution is searched in the form of

φN H (r) = C1 (u,r)φ1 + C2 (u, r) φ2 + C3 (u, r) φ3 (19.103b)

via the variation of parameters Ci (u, r), where u plays the role of a parameter.
The derivatives of the Ci are
∂C1 1 em r Rhs (2 m r − 1) ∂C2 1 em r Rhs ∂C3 1 Rhs
= 2
, = − , = ,
∂r 8 m ∂r 4 m ∂r 8 e m r m2
 19.10 Type III Kundt Metric; Λ = −m2 , W1 = −2 m coth(m r) 383

hence the non-homogeneous φN H solution can be given as

 
1 r
φN H = e Rhs (u, r) (2 m r − 1) dr −
mr
em r Rhs (u, r) dr
8m2 4m

1
+ 2 e2 m r Rhs (u, r) e−m r dr, (19.103c)
8m
where the function Rhs(u, r) entering in φ, using ξ from (19.101b), is given by
 d  
2 du f11 + f11 2 ln 1 − e−2 m r me−3 m r
Rhs = −2 2 , (19.103d)
(1 − e−2 m r )

therefore, factorizing dfdu11 and f11 2 in the expression of φ, accomplishing the

integration and using a relation between the dilog functions (19.99) in φN H , one
gets
f11 2 *    
φN H = 2
−2 dilog e2 m r − 4 m r ln e2 mr − 1 + 4 m2 r2
16m
    2 -
+ e2 m r − 1 1 + ln 1 − e−2 m r −1
 
1 df11  2m r   
+ 2 e − 1 ln 1 − e−2 m r . (19.103e)
4m du
In this expression, terms proportional to the independent homogeneous solu-
tions {1, r, e2m r } have been eliminated. On the other hand, using L2 (e2 mr ) from
(19.99), one rewrites (19.103e) as

f11 2 , 2 m r    
−2 m r 2

φN H = e − 1 1 + ln 1 − e − 1
16m2 +
+2L2 (e2 m r ) + 4 m2 r2
 
1 df11  2m r   
+ 2 e − 1 ln 1 − e−2 m r , (19.103f)
4m du
which coincides with the CPS Chow et al. (2010b) form of the solution:

ds2 = dr2 + 2du dv + (v 2 F2 + vF1 + F0 )du2 − 4 v m coth(m r) du dr,

m2
F2 (u, r) = , F1 (u, r) = f11 (u)ξ(r),
sinh2 (m r)
ξ(r) = ln(1 − e−2 m r ) = −2 m r + ln(e2 m r − 1),
F0 (u, r) = (1 − e−2 m r ) [φH (u, r) + φN H (u, r)(19.103e))] ,
φH = e2 mr f03 (u) + rf02 (u) + f01 (u) . (19.103g)

This solution is characterized by the Cotton:

C α β = C v u δαv δu β + C v r δαv δr β + C r uδαr δu β ,

  m3 e2 mr
C v u = −m2 f02 (u) e−2 mr e2 mr − 1 − 2 vf11 2
(e2 mr − 1)
384 Kundt Spacetimes in TMG

1 df11 1  
+ me−2 mr + m f11 2 ln 1 − e−2 mr + e−2 mr ,
2 du 4
m2
Cru = C v r = − 2 mr f11 . (19.103h)
e −1

Type N Limit
It allows for a type N Kundt solution with metric

ds2 = dr2 + 2du dv + (v 2 F2 + F0 )du2 − 4 v m coth(m, r) du dr,

m2
F2 (u, r) = , F0 (u, r) = 2 f02 r sinh(m r) e−rm , (19.104a)
sinh2 (m r)
with a type N Cotton tensor, JCN ,
 
C α β = C v u k α kβ , C v u = −m2 f02 (u) 1 − e−2 mr , k α = δ α v . (19.104b)

Multi Exponent–Integral Representation of φ

The Chow–Pope–Sezgin (Chow et al., 2010b) representation of the solution for
F0 (u, r) is achieved via an alternative definition of the function F0 (u, r) through
φ(r), namely
1 mr F0 (u, r)
φ(u, r) = e Φ(u, r) = , (19.105a)
2 1 − e−2mr
thus, replacing Φ (r) = 2 e−rm φ (r) in (19.102a), one gets
 2

∂ −2m r ∂ 1 RHS
e φ = e−m r Rhs, Rhs = (19.105b)
∂r ∂r2 2 2 sinh(m r)
Incidentally, it arises from the equation (19.93b) for μ = m. Integrating the
equation (19.105b) one obtains the general solution for φ (r) as
  
1 r t 2 m y y −m z
φ= e e Rhs (z) dzdydt + C3 e2 m r + C2 r + C1 .
2
Since the function ξ(r) can explicitly be integrated in terms of ln, (19.101b), one
can rewrite φ (u, r) as
 r t  y 
df11   dz dy dt
φ = −m e2 m y 2 + f11 2 ln 1 − e−2 m z 2
du (e2 m z − 1)
+C3 (u) e2 m r + C2 (u) r + C1 (u). (19.105c)

19.11 Type III Kundt Metric; Λ = −m2 ; W1 = −2 m tanh(m r)

Deriving the various possible branches of solutions, one has the case (19.41e),

W1 (r) = −2m tanh(mr).

 19.11 Type III Kundt Metric; Λ = −m2 ; W1 = −2 m tanh(m r) 385

We follow here the same pattern used in the previous sections: first the general
case of μ = ± m is treated; next, the derived results and key equations are
specialized to handle the sub-cases μ = ± m.
Evaluating the equation (19.24c) for F2 one gets

m2
F2 = − . (19.106)
cosh2 (m r)
Replacing these functions into the equation (19.24d) for F1 , one arrives at

∂2 ∂
F1 (u, r) + [μ + m tanh (m r)] F1 (u, r) = 0, (19.107)
∂r2 ∂r
with solution

m e−μ r
F1 (u, r) = f11 (u) dr = f11 ξ(r), (19.108)
cosh (m r)
where it has been introduced the auxiliary function ξ(r),

dξ me−μ r e−μ r
= → ξ (r) = m dr. (19.109)
dr cosh (m r) cosh (m r)
The equation (19.24e) for F0 (u, r) becomes

∂3 ∂2
F 0 + [μ − 3 m tanh (m r)] F0
∂r3 ∂r2
∂
+2 m −2 m + 3 m tanh2 (m r) − μ tanh (m r) F0
  ∂r
+2 m2 tanh2 (m r) − 1 (μ − 3 m tanh (m r)) F0
 
1 d d
=− 2
2 f11 + f11 ξ (r) ξ (r) =: RHS . (19.110)
2 du dr
Following the standard procedure for solving non-homogeneous linear ordinary
differential equations by means of the variation of parameters of the homogeneous
solution
 
F0H = cosh( m r) C1 (u) e−μ r + C2 (u) em r + C3 (u) e−m r , (19.111)

one arrives at an algebraic system of equations for the first-order derivatives

of Ci (u, r) with respect to r. Nevertheless, the structure of the homogeneous
solution above points out to a better function choice, namely
F0 (u, r)
F0 (u, r) = Φ (u, r) cosh (m r) , Φ (u, r) = , (19.112)
cosh (m r)
where u enters as a parameter. The equation (19.110) in terms of Φ becomes

∂3 ∂2 ∂ RHS
3
Φ + μ 2 Φ − m2 Φ − m2 μ Φ = = Rhs, (19.113)
∂r ∂r ∂r cosh (m r)
386 Kundt Spacetimes in TMG
 
2 df11
1 f11 ξ + 2 du dξd me−μ r
Rhs = − ,
ξ (r) = . (19.114)
2 cosh (m r) dr dr cosh (m r)
In what follows we proceed to determine the three different branches of solutions
in the specific cases of μ = ± m, μ = − m, and μ = m separately.

19.11.1 Type III Kundt Solution; Λ = −m2 ,

W1 = −2m tanh(mr), μ = ±m
First, the derivation of the solution via the variation of parameters is given. Next,
the determination of the solution in a multi-exponent–integral representation is
done. Finally, the CPS representation of the solution through a key function
introduced in Chow et al. (2010b) is given.

Solution Φ Through the Variation of Parameters

The homogeneous solution to the homogeneous equation (19.113) (Rhs → 0),
one searches in the form of ΦH ek r arriving at the characteristic equation

k 3 + μ k 2 − m2 k − m2 μ = 0 = (k + μ)(k − m)(k + m)

with solutions k = {−μ, −m, m}, hence ΦH is given by Φ (r) = C1 e−μ r +

C2 em r + C3 e−m r . Searching for the solution to the non-homogeneous equation
(Rhs = 0) one assumes it to be of the form

Φ (r) = C1 (u, r) Φl + C2 (u, r) Φ2 + C3 (u, r) Φ3 , (19.115a)

where u enters here as a parameter. The variation of the constants Ci (u, r),
within the variation of parameters procedure, yields
∂ eμ r RHS ∂ 1 1 RHS
C1 = 2 , C2 = e−m r ,
∂r (μ − m2 ) cosh (rm) ∂r 2 m(μ + m) cosh (m r)
∂ 1 1 RHS
C3 = − em r .
∂r 2 m(μ − m) cosh (m r)
Notice the presence of μ ± m in the denominators of the above expressions,
which clearly point to the degeneracy of these quantities when dealing with
the sub-branches μ = ±m; this is the reason why the quoted cases have to be
integrated separately for the equation (19.113).
Thus, the non-homogeneous solution amounts to
 −μ r 
df11 e m
ΦN H (u, r) = dr
du m − μ
2 2
cosh2 (mr)
 −(m+μ)r  
1 emr e 1 e−mr e(m−μ)r
− dr − dr
2 (m + μ) cosh2 (mr) 2 (m − μ) cosh2 (mr)
 19.11 Type III Kundt Metric; Λ = −m2 ; W1 = −2 m tanh(m r) 387
 
1 m e−μr
2 ξ (r)
+ f11 dr
2m −μ
2 2
cosh2 (mr)
 −(m+μ)r  (m−μ)r 
1 emr e ξ (r) 1 e−m r e ξ (r)
− dr − dr , (19.115b)
4 (m + μ) cosh2 (mr) 4 (m − μ) cosh2 (mr)

whereas the homogeneous solution is given by

ΦH (u, r) = f01 (u) e−rμ + f02 (u) em r + f03 (u) e−m r , (19.115c)

which together determine the general metric of this spacetime:

ds2 = dr2 + 2du dv + (v 2 F2 + vF1 + F0 )du2 − 4 v m tanh(m r) du dr,

m2
F2 (u, r) = − ,
cosh2 (m r)

m e−μ r
F1 (u, r) = f11 (u) ξ(r), ξ(r) = dr,
cosh(m r)
F0 (u, r) = cosh(m r) (ΦH (u, r) + ΦN H (u, r)(19.115b)). (19.115d)

This spacetime is characterized by the Cotton tensor

C αβ = C v u δαv δuβ + C v r δαv δr β + C r u δαr δuβ ,

μ  2  μ df11 −μ r
Cvu = μ − m2 e−μ r f01 (u) cosh (mr) − e sinh (mr)
2 2 du
mμv [μ cosh (mr) + m sinh (mr)]
− f11 e−μ r
2 cosh2 (mr)
  r
2
μ m −μ r 1 e−μ z
− e f11 2 cosh (mr) dzdr,
4 cosh2 (mr) cosh (mz)
mμ e−μ r
Cru = Cvr = − f11 . (19.115e)
2 cosh (mr)

Type N Limit
The type N limit Kundt solution is given by the metric

ds2 = dr2 + 2du dv + (v 2 F2 + F0 )du2 − 4 v m tanh(m r)du dr,

m2
F2 = − , F0 = cosh(m r) f01 (u) e−μ r , (19.116a)
cosh2 (m r)

with a type N Cotton tensor, JCN ,

μ  2 
C α β = C v u k α kβ , C v u = μ − m2 e−μ r f01 (u) cosh (mr) , k α = δ α v .
2
(19.116b)
388 Kundt Spacetimes in TMG

Multi Exponent–Integral Representation of Φ

In this approach, see details in 19.10.1, the non-homogeneous solution to the
equation (19.113), which is assumed to be expressible as
  
∂ ∂ ∂ αr RHS
eγr eβr (e Φ(u, r)) = (19.117a)
∂r ∂r ∂r cosh (m r)
is searched in the multi-exponent–integral form.
Comparing the coefficients multiplying derivatives of the same order in
equations (19.113) and (19.117a) one gets the algebraic system
2 β + 3 α + γ = μ,
γ β + 4 β α + 3 α2 + 2 γ α + β 2 = −m2 ,
γ β α + γ α2 + β 2 α + α3 + 2 β α2 = −m2 μ, (19.117b)
with the following set of solutions
{α = ∓m, β = ±2 m, γ = ∓m + μ} , sum = μ,
{α = μ, β = −μ ∓ m, γ = ±2 m} , sum = ±m,
{α = ∓m, β = μ ± m, γ = ±m − μ} , sum = ±m. (19.117c)
The substitution of the found exponents in the multi-exponent–integral expres-
sion for ΦN H (u, r)
 
 r  t  y f11 2 ξ (z) + 2 df11
m du
Φ = − e−α r e−β t e−γ y e(α+β+γ−μ) z dzdydt,
2 cosh2 (m z)
(19.117d)
determines a specific representation of the sought inhomogeneous solution. For
instance, for the first set of exponents, one has
        
df11 m r r −2m t t −(μ−m) y y 1
ΦN H = − e e e dz dy dt
du cosh2 (m z)
        
f11 2 m r r −2m t t −(μ−m) y y ξ (z)
− e e e dz dy dt,
2 cosh2 (m z)
(19.117e)
which gives, in turn, the general F0 (u, r) function through
F0 (u, r) = cosh(mr) e−μ r f01 (u) + em r f02 (u) + e−m r f03 (u) + Φ(u, r) .
(19.117f)
Consequently one gets the general type III Kundt metric
ds2 = dr2 + 2du dv + m2 v 2 cosech2 (mr)du2 − 4mv cosech(mr)du dv
+[vξ(r)F11 (u) + F0 (u, r)]du2 . (19.117g)
Of course this expression can be simplified in various manners; for instance, one
can partially integrate the first integral in Φ(u, r).
 19.11 Type III Kundt Metric; Λ = −m2 ; W1 = −2 m tanh(m r) 389

VP of the CPS Representation of F0 through φ

The Chow–Pope–Sezgin representation of the solution, in terms of new auxiliary
function φ(u, r), is achieved via an alternative definition of the function F0 (u, r)
through a
F0 (u, r) F0 (u, r)
Φ= , Φ = 2 φ e−m r , φ (u, r) = (19.118a)
cosh (m r) 1 + e−2 m r
which transforms the equation (19.113), (19.24e), to
∂3 ∂2 ∂ 1 emr RHS
3
φ + (μ − 3 m) 2 φ + 2 m (m − μ) φ = = Rhs. (19.118b)
∂r ∂r ∂r 2 cosh (mr)
The homogeneous solution, with characteristic polynomial k(k−2m)(k+μ−m) =
0, reads
φH (r) = C1 (u) + C2 (u) e2 mr + C3 (u) e(−μ+m)r
≡ C1 (u)φ1 + C2 (u) φ2 + C3 (u) φ3 (19.118c)
The non-homogeneous solution is searched in the form of
φN H (r) = C1 (u,r)φ1 + C2 (u, r) φ2 + C3 (u, r) φ3 (19.118d)
via the variation of parameters Ci (u, r), where u plays the role of a parameter.
The derivatives of Ci are given by
∂ 1 emr RHS ∂ e−2 m r emr RHS
C1 = , C2 = ,
∂r 4 (m − μ) m cosh (mr) ∂r 4 (m + μ) m cosh (mr)
∂ e−r(m−μ) emr RHS
C3 = − .
∂r 2(m2 − μ2 ) cosh (mr)
Therefore the general φN H (u, r) solution to the key equation (19.118b)
via VP is
F0 (u, r)
φ(u, r) = = C1 (u) + C2 (u) e2 mr + C3 (u) e(m−μ)r
(1 + e−2 mr )
  
1 ξ e−(m+μ)r e2mr ξ e−(μ+3m)r
+f11 −2
dr − 2 dr
(1 + e−2m r ) m − μ
2
2 2(m + μ) (1 + e−2mr )

(m−μ)r ξ me−2 mr
+e dr
(1 + e−2 mr ) m2 − μ2
2
  
df11 1 e−(m+μ) r e2mr e−(μ+3 m)r
+ − dr − 2 dr
du (1 + e−2m r )
2 m−μ m+μ (1 + e−2 mr )

(m−μ)r 1 m e−2 mr
+2e dr , (19.118e)
(1 + e−2 mr ) m2 − μ2
2

which is a single integral representation of the solution while the CPSb(4.54)

representation is multiple integral.
390 Kundt Spacetimes in TMG

Multi Exponent–Integral Representation of φ

To obtain the representation CPSb(4.54) of the solution to the key equation
(19.118b) one has to follow the multi-exponent–integral procedure previously
∂
detailed in 19.10.1. Introducing an auxiliary function H = ∂r φ we rewrite the
equation (19.118b) as
 
∂ βr ∂
EQH := e (e H(r)) = Rhs e(α+β) r ,
αr
∂r ∂r
 
H(r) = e−α r e−β r Rhs(u, r)e(α+β) r drdr. (19.119a)

Accomplishing the derivation one gets

∂2 ∂
H (r) + (β + 2 α) H (r) + α (β + α) H (r) = Rhs(u, r). (19.119b)
∂r2 ∂r
Comparing the coefficients of (19.119b) with those of (19.118b) one obtains the
system

{α (β + α) = −2 m (μ − m) , β + 2 α = μ − 3 m} , (19.119c)

with solutions

[[α = −2 m, β = m + μ], [α = μ − m, β = −m − μ]]. (19.119d)

For the first set of solutions one obtains

 r  y
H(r) = e2m r e−(μ+m) y Rhs(u, z)e(μ−m) z dzdy. (19.119e)

Consequently, φ integrates as
 r  t  y
φ= e−α t e−β y Rhs (z) e(α+β)z dzdydt,

and, for the case under consideration, φ is

 r  t  y
2mt −(m+μ)y
φ= e e Rhs(z)e(μ−m) z dzdydt
 
   df11 2
m r 2 m t t −(m+μ)y y 2 du + f11 ξ (z)
=− e e dzdydt.
4 cosh2 (mz)
Integrating once, one gets
 r  t −(m+μ)z
1 df11 e sinh (m z)
φN H = − e 2mt
dzdt
2 du cosh (m z)
 r  t  y
m ξ (z)
− f11 2 e2 m t e−(m+μ)y dzdydt. (19.119f)
4 cosh2 (m z)
To obtain the CPS Chow et al. (2010b) representation, one accomplishes the
substitution
 19.11 Type III Kundt Metric; Λ = −m2 ; W1 = −2 m tanh(m r) 391

sinh (m r) [2 − (1 + e(−2m)r )] e−μ r

e−(μ+m)r = e−μ r = − e−(μ+m)r ,
cosh (m r) 2 cosh (m r) cosh (m r)
thus
  
e−r(m+μ) sinh (m r) 1 e(m−μ) r
e 2mr
drdr = e2 m r ξ(r) dr − ,
cosh (m r) m μ2 − m2
the last term is proportional to one of the homogeneous solutions, hence, it can
be dropped out. Consequently, one arrives at the solution coincident with the
CPSb (4.54), see Chow et al. (2010b), by doing the identification ξ → F, r → ρ,

F0 (u, ρ) 1 df11 ρ 2mt
φCP S = =e (m−μ)ρ
f01 (u) − e F (t)dt
1 + e−2mρ 2m du ∞
 ρ  t  t  
1 e2m(t−t ) e(m−μ)t
− f11 2 dt dt dt F (t ). (19.119g)
4m ∞ ∞ ∞ cosh2 (mt )

19.11.2 Type III Kundt Solution;

Λ = −m2 , W1 = −2m tanh(mr), μ = −m
Solution Φ Through the Variation of Parameters
The key equation (19.113) for the function Φ(u, r) in the case μ = −m becomes
∂3 ∂2 ∂ RHS
Φ − m Φ − m 2 Φ + m3 Φ = = Rhs
∂r3 ∂r2 ∂r cosh (m r)
 
2 df11 mr
m f11 ξ (r) + 2 du e
=− , ξ(r) = ln(e2 m r + 1). (19.120a)
2 cosh2 (m r)
Since the characteristic equation (k + m)(k − m)2 = 0, allows for a double root,
then the homogeneous solution is ΦH (r) = C1 e−m r + C2 em r + C3 r em r .
Using the VP procedure for the non–homogeneous function

ΦN H (r) = C1 (u,r) e−m r + C2 (u,r) em r + C3 (u,r) r em r (19.120b)

one arrives at first-order equations for Ci (u,r)

∂C1 1 Rhs em r ∂C2 1 Rhs (2 m r + 1) e−m r ∂C3 1 e−m r Rhs
= , = − , = .
∂r 4 m2 ∂r 4 m2 ∂r 2 m
d 
Substituting the expression of Rhs, factorizing the terms with du f11 and f11 2
and integrating one arrives at the non-homogeneous solution ΦN H
e−m r df11  2m r   
ΦN H = − 2
e + 1 ln e2 m r + 1
2 m du
e−m r ,    2mr 2 
− f11
2
e 2mr
+ 1 1 + ln e + 1 − 1
8 m2  +
+2e2 m r dilog e2 m r + 1 . (19.120c)
392 Kundt Spacetimes in TMG

Finally, the metric of this solution can be written as

ds2 = dr2 + 2du dv + (v 2 F2 + vF1 + F0 )du2 − 4 v m tanh(m r) du dr,
m2
F2 (u, r) = − ,
cosh2 (m r)
 
F1 (u, r) = f11 (u) ξ(r), ξ(r) = ln e2 mr + 1 ,
F0 (u, r) = cosh(m r) (ΦH + ΦN H (19.120c)) ,
ΦH = f01 (u) e−mr + f02 (u) emr + f03 (u) emr r. (19.120d)
This spacetime is characterized by the Cotton tensor
C αβ = C v u δαv δu β + C v r δαv δr β + C r u δαr δu β ,
1   m3 e2 mr
C v u = m2 e2 mr + 1 f03 + 2 v f11 2
2 (e2 mr + 1)
m   m df11
+ f11 2 ln e2 mr + 1 + 1 + ,
4 2 du
m2 e2 mr
C r u = C v r = 2 mr f11 . (19.120e)
e +1

Type N Limit
For vanishing function f11 one gets a type N metric
ds2 = dr2 + 2du dv + (v 2 F2 + F0 )du2 − 4m v tanh(m r) du dr,
 
F0 (u, r) = cosh(m r) f01 (u) e−m r + f02 (u) em r + f03 (u) r em r ,
m2
F2 (r) = − , (19.121a)
cosh2 (m r)
characterized by a Petrov type N Cotton tensor
1  
C α β = C v u k α kβ , C v u = m2 e2 mr + 1 f03 , k α = δ α v . (19.121b)
2

Multi Exponent–Integral Representation of Φ

Let us search for a multi-exponent–integral representation, see 19.10.1, of the
solution to the equation (19.120a). One gets the algebraic system
γ + 3 α + 2 β = −m,
2 γ α + β 2 + 3 α2 + 4 β α + γ β = −m2 ,
γ β α + γ α 2 + 2 β α 2 + β 2 α + α 3 = m3 (19.122a)
with solutions
{α = m, β = −2 m, γ = 0} , {α = −m, β = 2 m, γ = −2 m, } ,
{α = −m, β = 0, γ = 2 m} . (19.122b)
 19.11 Type III Kundt Metric; Λ = −m2 ; W1 = −2 m tanh(m r) 393

Since in this approach Φ is expressible as

 r  t  y
ΦN H = e−α r e−β t e−γ y e(γ+β+α)z Rhs (u, z) dzdydt, (19.122c)

for each set of exponents one gets its corresponding multi-exponent–integral

solution representation.

VP of the CPS Representation of φ

The Chow–Pope–Setzgin representation of the solution for F0 (u, r) is achieved
via an alternative definition of the function F0 (u, r); instead of

F0 (u, r) F0 (u, r) −mr

Φ(u, r) = =2 e
cosh(mr) (1 + e−2mr )
one uses φ(u, r) defined as
F0 (u, r) 1
φ(u, r) = → φ(u, r) = emr Φ(u, r). (19.123a)
1 + e−2mr 2
In terms of φ the equation (19.120a), (19.24e), becomes
∂3 ∂2 2 ∂ 1 emr RHS
φ − 4m φ + 4 m φ = = Rhs. (19.123b)
∂r3 ∂r2 ∂r 2 cosh (mr)
The solution to the homogeneous equation is given by

φh (r) = C1 (u) + C2 (u) e2 mr + C3 (u) r e2 mr

≡ C1 (u)φ1 + C2 (u) φ2 + C3 (u) φ3 . (19.123c)

The non-homogeneous solution is sought in the form of

φN H (r) = C1 (u,r)φ1 + C2 (u, r) φ2 + C3 (u, r) φ3 (19.123d)

via the variation of parameters Ci (u, r), where u plays the role of a parameter.
The first derivatives are
∂ 1 Rhs ∂C2 1 (2 m r + 1) e−2 m r Rhs
C1 = , = − ,
∂r 4 m2 ∂r 4 m2
∂ 1 e−2 m r Rhs
C3 = .
∂r 2 m
Therefore the general φ(u, r) solution to the equation (19.123b) via the variation
of parameters method is
F0 (u, r)
φ(u, r) = = f01 (u) + f02 (u) e2 mr + f03 (u) r e2 m r
(1 + e−2 mr )
+C1 (u, r) + C2 (u, r) e2 mr + C3 (u, r) r e2 m r , (19.123e)

explicitly the non-homogeneous solution is given by

394 Kundt Spacetimes in TMG

f11 2 ,  2 m r    2mr 2 

φN H = − e + 1 1 + ln e + 1 − 1
16 m2  +
+2 e2 m r dilog e2 m r + 1
1 df11  2 m r   
− 2
e + 1 ln e2 m r + 1 , (19.123f)
4 m du
which coincides with the CPS solution Chow et al. (2010b)(4.58) if one prop-
2mr
erly identifies
 the
 Li2 function with the dilog appearing here, Li2 (−e ) =
2 mr
dilog 1 + e , (19.99f).

Multi Exponent–Integral Representation of φ

To obtain a multi-exponent–integral representation of equation (19.123b) for
φ(u, r) one simply rewrites it as
 
2
  mr m f 2
ξ (r) + 2 df11
∂ ∂ 1 e RHS 11 du
e2 mr 2 e−2 mr φ = = Rhs = − 2 .
∂r ∂r 2 cosh (mr) (1 + e −2 m r )
(19.124a)
Therefore, the integral for φ(u, r) is
 
 r  t y f11 2 ξ (z) + 2 df11
1 du
φ (u, r) = − m e2 m t dz dy dt. (19.124b)
4 cosh2 (m z)
It can be given as
 r y  t
m ξ (z)
φ (u, r) = − f11 2 e 2mt
dz dydt
4 cosh2 (m z)
 2mr  2mr 
1 df11 ln e +1 e +1
− 2
. (19.124c)
4 du m

19.11.3 Type III Kundt Solution;

Λ = −m2 , W1 = −2m tanh(mr), μ = m
Solution Φ via Variation of Parameters
The equation (19.113) for the function Φ(u, r) in the case μ = m becomes
∂3 ∂2 ∂ RHS
Φ +m Φ − m2 Φ − m3 Φ = = Rhs
∂r3 ∂r2 ∂r cosh (m r)
 
1 d dξ
=− 2 f11 + f11 2 ξ ,
2 cosh (m r) du dr
dξ e−m r  
=m , ξ(r) = − ln 1 + e−2 m r . (19.125a)
dr cosh (m r)
Its homogeneous solution of the form ek r fulfills (k + m)2 (k − m) = 0, because
of the repeated roots, it is
ΦH = C1 em r + C2 e−m r + C3 r e−m r , (19.125b)
 19.11 Type III Kundt Metric; Λ = −m2 ; W1 = −2 m tanh(m r) 395

thus the non-homogeneous solution is sought in the form

ΦN H = C1 (u, r) em r + C2 (u, r) e−m r + C3 (u, r) r e−m r . (19.125c)
Using the VP procedure, one arrives at
∂ e−m r RHS ∂ em r (2mr − 1)) RHS
C1 = 2
, C2 = ,
∂r 4m cosh (m r) ∂r 4m2 cosh (m r)
∂ em r RHS
C3 = − .
∂r 2 m cosh (m r)
The use of
d me−m r me−2 m r  
ξ (r) = =2 , ξ(r) = − ln 1 + e−2 m r ,
dr cosh (m r) 1 + e−2 m r
in the RHS, its substitution into the Ci s and their integrations lead the non-
homogeneous ΦN H in the form
1 ,   2  
ΦN H = − 2 f11 2 1 + ln 1 + e−2 m r − 1 1 + e−2 m r em r
8m   +
+2 e−m r dilog 1 + e−2 m r + 4 m r e−m r + 2 em r
1 df11  m r   
+ 2 e + e−m r ln e2 m r + 1 − 2 mem r r . (19.125d)
2m du
Using the relation (19.99c) between the dilog functions and Li2 (19.99f),
excluding the terms proportional to the homogeneous functions, one finally
arrives at
e−m r ,     
2 −2 m r 2
ΦN H = − 2
(f 11 ) 1 + ln 1 + e − 1 1 + e2 m r
8m +
−2 Li2 (−e2m r ) − 4 m2 r2
e−m r df11  2m r   
+ 2
e + 1 ln e2 m r + 1 . (19.125e)
2 m du
which coincides, up to the factor 2e−m r , with CPSb(4.56).

Multi-Exponent–Integral Representation of Φ
In this approach the search of the solution to the key equation (19.125a),
following 19.10.1, gives rise to the algebraic system
γ + 2 β + 3 α = m,
γ β + 2 γ α + β 2 + 4 β α + 3 α2 = −m2 ,
γ β α + γ α2 + β 2 α + 2 β α2 + α3 = −m3 (19.126a)
with the following set of solutions
{α = −m, β = 2 m, γ = 0} ,
{α = m, β = 0, γ = −2 m} ,
{α = m, β = −2 m, γ = 2 m} , (19.126b)
396 Kundt Spacetimes in TMG

with the corresponding solution representations. For instance, for the first set of
exponents one has
 
mr −2 m r
Φ1 = e e em r Rhs (r) dr drdr.

VP of the CPS Representation of φ(r)

The CPSb representation of the solution for F0 (u, r) is achieved via an alternative
definition of the function F0 (u, r) through φ(r), namely
F0 (u, r) F0 (u, r)
φ(u, r) = ,Φ= = 2 e−mr φ (19.127a)
1 + e−2mr cosh(m r)
thus replacing Φ = 2 e−m r φ in (19.125a) one gets
∂3 ∂2 1 RHS
3
φ − 2 m 2 φ = em r . (19.127b)
∂r ∂r 2 cosh(mr)
If one were using the variations of parameters method to determine the solu-
tion to the non-homogeneous equation one generalizes the parameters Ci of the
homogeneous solution
φH (r) = C1 (u) + C2 (u)r + C3 (u) e2 m r , (19.127c)
to be dependent on the variable r in such a way that the non-homogeneous
solution is search in the form of
φN H (r) = C1 (u,r)φ1 + C2 (u, r) φ2 + C3 (u, r) φ3 (19.127d)
via the variation of parameters Ci (u, r), where u plays the role of a parameter.
The derivatives of the Ci are
∂C1 e r m (2 m r − 1) RHS ∂C2 e r m RHS
= , = − ,
∂r 8 m2 cosh(m r) ∂r 4 m cosh(m r)
∂C3 e− r m RHS
= ,
∂r 8 m2 cosh(m r)
hence, the non-homogeneous φN H solution can be given as
 mr
1 e RHS (2 m r − 1)
φN H = dr
8 m2 cosh (m r)
 mr 
r e RHS 1 2mr RHS
− dr + e e−m r dr. (19.127e)
4m cosh (m r) 8 m2 cosh (m r)
Replacing the structural functions
   
ξ (r) = − ln 1 + e−2 m r , cosh (m r) = 1/2 em r 1 + e−2 m r
in the above equation and integrating one obtains
 
1 df11  2 mr   
φN H = 2
e + 1 ln 1 + e−2 m r + 2 mr + e2 m r
4m du
 19.11 Type III Kundt Metric; Λ = −m2 ; W1 = −2 m tanh(m r) 397

1 ,     
−2 m r 2
− f11
2
e 2mr
+ 1 1 + ln 1 + e − 1 + 4 m r + 2 e2 m r
16 m2  +
+2 dilog 1 + e−2 m r . (19.127f)

One can always eliminate the terms proportional to the independent homoge-
neous solutions {1, r, e2mr }, and use the relation (19.99c) between the dilog and
Li2 (19.99f) and write, without loss of generality, the function φN H (u, r)
1 ,    2 
φN H = − 2
f11 2 e2 m r + 1 1 + ln 1 + e−2 m r −1
16 m   +
−2 Li2 −e2m r − 4 m2 r2
 
1 df11  2 mr   
+ e + 1 ln 1 + e−2 m r , (19.127g)
4 m2 du
i.e., the formula CPSb(4.56). Notice that the relation Φ = 2 e− m r φ holds, as
indeed it should.
The metric of this solution can be written as

ds2 = dr2 + 2du dv + (v 2 F2 + vF1 + F0 )du2 − 4 m v tanh(m r) du dr,

m2
F2 (u, r) = − ,
cosh2 (m r)
 
F1 (u, r) = f11 (u) ξ(r), ξ(r) = − ln 1 + e−2 m r ,
 
F0 (u, r) = 1 + e2 mr [φH (u, r; Ci (u) → f0i ) + φN H (u, r)] . (19.127h)

This spacetime is characterized by the Cotton tensor

C αβ = C v u δαv δu β + C v r δαv δr β + C r uδαr δu β ,

  m3 e2 mr m −2 mr df11
C v u = −m2 1 + e−2 mr f02 − 2 v f11 2 + 2 e
(e2 mr + 1) du
m  
ln 1 + e−2 mr − e−2 mr (f11 ) ,
2
+
4
m2
C r u = C v r = − 2 mr f11 . (19.127i)
e +1

Type N Limit
Its Petrov type N limit is given by the metric

ds2 = dr2 + 2du dv + (v 2 F2 + vF1 + F0 )du2 − 4 m v tanh(m r) du dr,

m2
F2 (u, r) = − ,
cosh2 (m r)
 
F0 (u, r) = r 1 + e2 mr f02 , (19.128a)

and is characterized by the Cotton tensor

 
C α β = C v u k α kβ , C v u = −m2 1 + e−2 mr f02 , k α = δ α v .
398 Kundt Spacetimes in TMG

Multi Exponent–Integral Representation of φ

The CPS representation of the solution for F0 (u, r) is achieved via an alternative
definition of the function F0 (u, r) through φ(r), namely
F0 (u, r) F0 (u, r)
φ(u, r) = −2mr
,Φ= = 2 e−mr φ, (19.129a)
1+e cosh(m r)
thus, replacing Φ = 2 e−m r φ in (19.125a), or F0 (φ) in (19.24e), one gets
 2

mr ∂ −2mr ∂ φ 1 RHS
e e = . (19.129b)
∂r ∂r2 2 cosh(mr)
Incidentally, it arises from the key CPS equation (19.118b) for μ = m.
The first integral of the equation (19.129b) yields

∂2 1 2mr RHS
φ = e e−m r dr + A0 e2 m r ,
∂r2 2 cosh(mr)
therefore, integrating twice, one obtains the general solution for φ (r) in the form
 
 r t  y f11 2 ξ (z) + 2 df11 e−2 m z
1 du
φ=− m e2 m y dzdy dt
4 cosh2 (m z)
+f01 (u)e2m r + f02 (u) r + f03 (u). (19.129c)
The function ξ(r) can be integrated in terms of logarithm, namely
d me−2 m r  
ξ (r) = 2 −2 m r
, ξ(r) = − ln 1 + e−2 m r , (19.129d)
dr 1+e
which substituted in φ yields
 r t  y  
e−4 m z dz df11  −2 m z

φ = −m e2 m y 2 2 − f11
2
ln 1 + e dy dt
(1 + e−2 m z ) du
+C3 (u) e2 m r + C2 (u) r + C1 (u) (19.129e)
this is an integral representation of the solution, although one can integrate it
forward.

19.12 Type III Kundt Metric; Λ = m2 , W1 = −2m cot(m r), μ

The equation (19.40) for vanishing eigenvalue S, which guarantees the Petrov
type III of the Cotton tensor, for positive cosmological constant Λ = m2 , leads
to
W1 (r) = −2 m cot (m r) , from (19.41f). (19.130)
Evaluating the equation (19.24c) for F2 one gets
m2
F2 = 2 . (19.131)
sin (m r)
 19.12 Type III Kundt Metric; Λ = m2 , W1 = −2m cot(m r), μ 399

Replacing these functions into the equation (19.24d) for F1 one arrives at
∂2 ∂
2
F1 (u, r) + (μ + m cot (m r)) F1 = 0 (19.132)
∂r ∂r
with solution

e−μ r
F1 (u, r) = f11 (u) m dr = f11 ξ(r); (19.133)
sin (m r)
where it has been introduced the auxiliary function ξ(r),

dξ me−μ r e−μ r
= → ξ (r) = m dr. (19.134)
dr sin (m r) sin (m r)
The key equation (19.24e) for F0 (u, r) becomes
∂3 ∂2
F 0 (u, r) + (μ − 3m cot (m r)) F0
∂r3 ∂r2
  ∂
+2 m 2 m − μ cot (m r) + 3 m cot2 (rm) F0
  ∂r
+2 m2 F0 1 + cot2 (m r) (μ − 3m cot (m r))
 
1 d 2 d
=− 2 f11 + (f11 ) ξ (r) ξ (r) = RHS. (19.135)
2 du dr
Following the standard VP procedure for solving non-homogeneous linear
ordinary differential equations the solution F0 N H is searched as
F0 = C1 (u, r) e−μ r + C2 (u, r) sin (m r) + C3 (u, r) cos (m r) sin (m r) .
(19.136)
Nevertheless, the structure of the solution above points to a better function
choice, namely
Φ (u, r) = F0 (u, r)/sin (m r), F0 (u, r) = Φ (u, r) sin (m r) . (19.137)
The key equation (19.135) in terms of Φ becomes
∂3 ∂2 ∂ RHS
3
Φ + μ 2
Φ + m2 Φ + Φm2 μ = . (19.138)
∂r ∂r ∂r sin (m r)
In what follows we proceed to determine the single solution of this class.
First the derivation of the solution via the variation of parameters procedure is
done. Next, the solution in a multi-exponent–integral representation is curried
out.

19.12.1 Solution Φ Through VP

The homogeneous solution to the homogeneous equation (19.138) (Rhs → 0),
one searches in the form of Φh ek r arriving at the characteristic equation
k 3 + μ k 2 + m2 k + m2 μ = 0 = (k 2 + m2 )(k + μ)
400 Kundt Spacetimes in TMG

with solutions k = {−μ, im, −i m}, hence ΦH can be given as

ΦH (r) = C1 e−μ r + C2 sin (m r) + C3 cos (m r) ≡ C1 Φl + C2 Φ2 + C3 Φ3 .

(19.139a)
Searching for the solution to the non-homogeneous equation (Rhs = 0) one
assumes the solution to be of the form

ΦN H (u, r) = C1 (u, r) Φl + C2 (u, r) Φ2 + C3 (u, r) Φ3 . (19.139b)

The variation of the parameters Ci (u, r), within the VP procedure, yields
∂ eμ r RHS
C1 = 2 ,
∂r μ + m2 sin (m r)
∂ (m sin (m r) − μ cos (m r)) RHS
C2 = − ,
∂r m (μ2 + m2 ) sin (m r)
∂ (μ sin (m r) + m cos (m r)) RHS
C3 = − . (19.139c)
∂r m (μ2 + m2 ) sin (m r)
Some simplifications arise by replacing ξ(r), or more precisely, its derivative from
(19.134), to get the non-homogeneous solution as

 e−μ r (m sin (m r) − μ cos (m r)) f11 2 ξ + 2 df11

1 du
ΦN H = sin (m r) dr
2 (μ2 + m2 ) sin2 (m r)
 e−μ r (μ sin (m r) + m cos (m r)) f11 2 ξ + 2 df11
1 du
+ cos (m r) dr
2 sin2 (m r) (μ2 + m2 )
 m f11 2 ξ + 2 df11
1 du
− e−μ r dr, (19.139d)
2 sin2 (m r) (μ2 + m2 )
d
or, taking into account that the evaluation of the integrals multiplying du f11
gives
d f11 sin (m r)
ξ (r) ,
du m2
factorizing, one obtains a linear single integral solution for ΦN H
 
d f11 sin (m r) f11 2 −μ r ξ (r) m
ΦN H = ξ (r) + −e dr
du m2 2 ( μ2 + m2 ) sin2 (m r)

(m sin (m r) − μ cos (m r)) ξ (r)
+ sin (m r) e−μ r dr
sin2 (m r)
 −μ r 
e ξ (r) (μ sin (m r) + m cos (m r))
+ cos (m r) dr . (19.139e)
sin2 (m r)
The homogeneous solution can be written as

ΦH = f01 (u) e−rμ + f02 (u) sin(m r) + f03 (u) cos(m r), (19.139f)
 19.12 Type III Kundt Metric; Λ = m2 , W1 = −2m cot(m r), μ 401

which together determine the general solution. The metric of this solution can
be written as

ds2 = dr2 + 2du dv + (v 2 F2 + vF1 + F0 )du2 − 4 m v cot(m r) du dr,


m2 e−μ r
F2 (u, r) = , F1 (u, r) = f11 (u) ξ(r), ξ (r) = m dr,
sin2 (m r) sin (mr)
F0 (u, r) = sin(m r) [φH (u, r) + φN H (u, r)] . (19.139g)

This spacetime is characterized by the Cotton tensor

C αβ = C v u δαv δuβ + C v r δαv δr β + C r u δαr δuβ ,

1   1 df11 −μ r
C v u = μ μ2 + m2 f01 e−μ r sin (mr) + μ e cos (mr)
2  2 du
1 ξ (r)
− μ m (f11 ) e−μ r sin (mr)
2
dr
4 sin2 (mr)
1 [μ sin (mr) + m cos (mr)]
− m μ v f11 e−μ r ,
2 sin2 (mr)
1 e−μ r
Cru = C v r = − mμ f11 . (19.139h)
2 sin (mr)

Type N Limit
Its Petrov type N limit is given by the metric

ds2 = dr2 + 2du dv + (v 2 F2 + vF1 + F0 )du2 − 4 m v cot(m r) du dr,

m2
F2 (u, r) = 2 , F0 (u, r) = f01 (u) e−μ r sin(m r). (19.140a)
sin (m r)
and is characterized by the type N Cotton tensor
1  2 
C α β = C v u k α kβ , C v u = μ μ + m2 f01 e−μ r sin (mr) , k α = δ α v .
2
(19.140b)

19.12.2 Multi Exponent–Integral Representation of Φ

Let us now search the non-homogeneous solution to the third-order equation
(19.138) by guessing that this equation can be written in the form
  
∂ ∂ ∂ αr RHS
eγr eβr (e Φ(u, r)) = e(α+β+γ)r , (19.141a)
∂r ∂r ∂r sin (m r)
hence the solution will be
 r  t  y
RHS(u, z) (α+β+γ) z
Φ(u, r) = e−αr (e−βt (e−γy e dz)dy)dt. (19.141b)
sin (m z)
402 Kundt Spacetimes in TMG

Comparing the coefficients multiplying derivatives of the same order in equation

(19.138) and with the expanded (19.141a) one gets the system
2 β + 3 α + γ = μ,
γ β + 4 β α + 3 α 2 + 2 γ α + β 2 = m2 ,
γ β α + γ α2 + β 2 α + α3 + 2 β α2 = m2 μ,
with the following set of complex solutions
[α = μ, β = −μ + im, γ = −2 im], [α = μ, β = −μ − im, γ = 2 im],
[α = im, β = μ − im, γ = −μ − im],
[α = −im, β = μ + im, γ = −μ + im],
[α = −i m, β = 2 im, γ = μ − im], [α = im, β = −2 im, γ = μ + im].
The substitution of the found exponents in the expression (19.141b) determines
a specific representation of the solution we are looking for. In particular, for the
first set of constant exponents one gets a complex solution representation ΦC of
the CPS Chow et al. (2010b) solution:
 r  t  y
RHS(z) −im z
ΦC = e−μ r eμ t e−im t e2 imy e dzdydt, (19.141c)
sin(m z)
or in term of the trigonometric functions
 r  t
ΦC = e−μ r eμ t (cos (m t) − i sin (m t)) (cos (2 m y) + i sin (2 m y))
 y
RHS (z) (cos (m z) − i sin (m z))
× dzdydt, (19.141d)
sin (m z)
with real part
 r  t  y
RHS (z) cos (m z)
ΦR = e−μ r eμ t sin (m t) sin (2 m y) dzdydt
sin (m z)
 r  t  y
−μ r μt
+e e cos (m t) sin (2 m y) RHS (z) dzdydt
 r  t  y
−e−μ r eμ t sin (m t) cos (2 m y) RHS (z) dzdydt
 r  t  y
RHS (z) cos (m z)
+e−μ r eμ t cos (m t) cos (2 m y) dzdydt,
sin (m z)
(19.141e)
and an imaginary part, which does not contribute to the solution since it fits
the homogeneous equation. Therefore, we work out the real part, arriving at the
compact “cosine” CPS Chow et al. (2010b) representation
  t  y  
m r cos[m (t − 2 y + z)] df11
ΦR = − dt dy eμ(y−z−r) f11
2
ξ + 2 dz;
2 sin2 (m z) du
(19.141f)
 19.12 Type III Kundt Metric; Λ = m2 , W1 = −2m cot(m r), μ 403

df11
interestingly, the factor of du can be simplified up to
sin (m r) ξ (r) df11
,
m2 du
as was done in the previous VP paragraph, coinciding with the expression
of CPS Chow et al. (2010b), which determines in turn the non-homogeneous
F0 (u, r) function. Finally, the metric becomes
ds2 = dr2 + 2du dv + m2 v 2 cosec2 (m r)du2 − 4mv cot(m r)du dr
+[vξ(r)f11 (u) + F0 (u, r)]du2 ,
 r  t  y
F0 (u, r) m cos[m (t − 2 y + z)]
= − f11 2 dt dy eμ (t−z−r) ξdz
sin(m r) 2 sin2 (m z)
sin (mr) ξ (r) df11
+ + e−μr f01 (u) + sin(m r)f02 (u) + cos(m r)f03 (u),
m2 du

e−μ r
ξ (r) = m dr. (19.141g)
sin (m r)
For other choice of exponents, one arrives at other representations of the sought
inhomogeneous solution.
All the references of the solutions of the type III and N Kundt families can be
represented schematically by the table:

Table 12.1.1 Type III and type N Kundt solutions

S = ∂r W1 − W1 2 /2 − 2Λ = 0

W1 (u, r) μ(m) T III Rel. to TN Λ = ±m2 , 0

Secs. CPSb Eqns.

0 μ 19.7 (4.15) – 0
−2/r μ 19.8 (4.39) (19.70) 0
μ = ±m 19.9.1 (4.3) (19.73) <0
-2m μ=m 19.9.2 (4.9) (19.75) <0
μ = −m 19.9.3 (4.13) (19.77) <0
μ = ±m 19.10.1 (4.26) (19.87) <0
−2m coth(2mr) μ = −m 19.10.3 (4.34) (19.98) <0
μ=m 19.10.4 (4.30) (19.104) <0
μ = ±m 19.11.1 (4.53) (19.116) <0
−2m tanh(2mr) μ = −m 19.11.2 (4.57) (19.121) <0
μ=m 19.11.3 (4.55) (19.128) <0
−2m cot(2mr) μ 19.12 (4.44) (19.140) >0
 20
Cotton Tensor in Riemannian Spacetimes

In this chapter a systematic derivation of the Cotton tensor is presented follow-

ing Garcı́a et al. (2004). In 3D spaces the Cotton tensor is prominent as the
substitute for the Weyl tensor. It is conformally invariant and its vanishing is
equivalent to conformal flatness. However, the Cotton tensor arises in the con-
text of the Bianchi identities and is present in any dimension n. In this text, its
irreducible decomposition is performed and the number of independent compo-
nents as n(n2 −4)/3 is determined. Subsequently, its characteristic properties are
exhibited. An algebraic classification of the Cotton tensor in three dimensions is
accomplished. This classification is used throughout this text; for each class of
the derived here solutions the Cotton tensor is evaluated and classified.
The nonlinear coupling of gravity to matter in general relativity presents diffi-
cult technical problems in attempts to understand the gravitational interaction of
elementary particles and strings or to investigate details of the gravitational col-
lapse; see Witten (1991). Progress in the former area has come mainly from treat-
ing quantum fields as propagating on fixed background geometries: Polchinski
(1998), whereas much of the progress in the latter has come from detailed numer-
ical works: Choptuik (1993); Abrahams and Evans (1993); Gundlach (1999).
Exact solutions of the relevant matter–gravity equations can play an impor-
tant role by shedding light on questions of interest in both general relativity and
string theory. One is often interested in certain classes of solutions with spec-
ified asymptotic properties; the most common of them are the asymptotically
flat spacetimes. Recent work in string theory has, via the AdS/CFT conjecture,
highlighted the importance of the asymptotically AdS spacetimes, see Malda-
cena (1998). The AdS/CFT correspondence relates a quantum field theory in d
dimensions to a theory in d + 1 dimensions that includes gravity Gubser et al.
(1998), and Witten (1998). This is the motivation for looking at the conformally
flat spaces and at the spaces of constant curvature. For this reason we decided
to review the subject and to collect some old and new results that are nowadays
important in the context of anti-de Sitter spacetimes and to present them in
 20.1 Bianchi Identities and the Irreducible Decomposition 405

modern language. These results seem presently not to be too well known in the
community.
In the theory of conformal spaces the main geometrical objects to be ana-
lyzed are the Weyl (1918) and the Cotton (1899) tensors. It is well known that
for conformally flat spaces the Weyl tensor has to vanish. However, the Cotton
tensor is only conformally invariant in three dimensions. Any 3D space is con-
formally flat if the Cotton tensor vanishes. If matter is present, the Ricci tensor
is related to the energy–momentum tensor of matter by means of the Einstein
equations. Then the vanishing of the Cotton tensor imposes severe restrictions on
the energy–momentum tensor. The Cotton tensor also plays a role in the context
of the Hamiltonian formulation of general relativity; see Arnowitt et al. (1962).
First the Cotton 2-form in the context of the Bianchi identities is derived.
Subsequently its characteristic properties are described and an irreducible
decomposition with respect to the (pseudo-)orthogonal group is performed. This
allows ones to determine the number of irreducible components in any dimen-
sion. Moreover, in four dimensions, the relation of the Cotton to the Bach tensor
is exhibited. After that it is shown how to derive the Cotton 2-form in three
dimensions by means of a variational procedure. The classification of the Cotton
2-form in three dimensions is accomplished by means of its eigenvalues.
Our notation and conventions for this chapter are taken from Schouten
(1954) and for exterior calculus we refer to Hehl et al. (1995). For quick and
easy reference, we display our conventions for index positions and signs of the
Christoffel symbol, the Riemann tensor, and the Ricci tensor (holonomic indices
i, j, · · · = 0, 1, 2, 3). The sign of the Ricci tensor is the same as those of the Lij
tensor and the Cotton tensor. In particular, the Ricij sign introduces a relative
sign between the Lij tensor and the Weyl tensor in the decomposition of the
curvature:

∇i Tj k = ∂i Tj k − Γij  T k + Γi k Tj  , (20.1)

+Rijk = ∂i Γjk − ∂j Γik + Γim Γjk
    m
− Γjm Γik ,
 m
(20.2)
i
+Ricjk = Rijk , (20.3)
ji
+R = Rij , (20.4)
4
Weylijk = Rijk + g[i|[k L]|j] . (20.5)
n−2
An extensive comparison between the various conventions can be found in Misner
et al. (1973).

20.1 Bianchi Identities and the Irreducible Decomposition

of the Curvature
Let Vn be a Riemannian space of n dimensions. We have the coframe ϑα and its
dual frame eα according to
406 Cotton Tensor in Riemannian Spacetimes

ϑα = ei α dxi , eα = ei α ∂i , eα ϑβ = ei α ei β = δαβ , (20.6)

where xi are local coordinates and the ei α are the n-bein coefficients. Greek
letters α, β, · · · = 1, . . . , n denote anholonomic, Latin letters i, j, · · · = 1, . . . , n
holonomic indices. The symbol  stands for the interior product.
We introduce a metric g according to
g = gαβ ϑα ⊗ ϑβ = gij dxi ⊗ dxj . (20.7)
With the metric at our disposal we also have the Hodge dual operator. Then we
define the η-basis according to
η :=  1 , ηα :=  ϑα , ηαβ :=  (ϑα ∧ ϑβ ) , ηαβγ :=  (ϑα ∧ ϑβ ∧ ϑγ ). (20.8)
Furthermore, we equip our manifold with a symmetric connection (u is an
arbitrary vector),
∇u eα = Γα β (u) eβ , Γα β (∂i ) = Γiα β , Γα β = Γiα β dxi , (20.9)
with dϑβ = −Γα β ∧ ϑα . In a Riemannian space the connection 1-form can be
expressed in terms of the metric and the object of anholonomity Ωα := dϑα and
Ωα := gαβ Ωβ ,
1   1
Γαβ = Γiαβ dxi = dgαβ + e[α dgβ]γ ϑγ +e[α Ωβ] − (eα eβ Ωγ ) ϑγ . (20.10)
2 2
The curvature 2-form reads
1
Rα β := dΓα β − Γα γ ∧ Γγ β = Rμνα β ϑμ ∧ ϑν . (20.11)
2
Since a metric is given, we can lower the second index. Then the curvature 2-form
is antisymmetric Rαβ = −Rαβ . This can be seen if we choose an orthonormal
frame
gαβ = diag(−1, −1, . . . , 1, 1, . . .) , (20.12)
. /0 1 . /0 1
ind n-ind

where ind , the number of negative eigenvalues, denotes the index of the metric.
Then dg = 0 and, according to (20.10), the connection is antisymmetric. In turn,
from (20.11) we can infer the antisymmetry of Rαβ .
The exterior covariant derivative of a p-form Xα β is given by
DXα β = dXα β − Γα γ ∧ Xγ β + Γγ β ∧ Xα γ . (20.13)
The Bianchi identities can be formulated concisely with the help of this definition.
In a Riemannian space the torsion T α = Dϑα vanishes. Thus, the first Bianchi
identity reads
0 = DT α = DDϑα = Rβ α ∧ ϑβ , (20.14)
or, in components,
R[αβγ] δ = 0 . (20.15)
 20.1 Bianchi Identities and the Irreducible Decomposition 407

The first Bianchi identity is a (co-)vector valued 3-form with

 
n n2 (n − 1)(n − 2)
n = (20.16)
3 3!
independent components that imposes the same number of constraint equations
on the components of the curvature. Accordingly, in n-dimensions, the curvature
2-form has
    
n n n n2 (n − 1)(n + 1)
−n = (20.17)
2 2 3 12
independent components. For n = 3, we have six independent components and
for n = 4 (the case of GR) 20 independent components.
The second Bianchi identity is
DRα β = 0 , ∇[λ Rμν]α β = 0 . (20.18)
We now perform the irreducible decomposition of the curvature with respect
to the pseudo-orthogonal group (Hehl et al., 1995):
n=1 Rαβ = 0
n=2 Rαβ = Scalarαβ
(20.19)
n=3 Rαβ =  αβ
Scalarαβ + Ricci
n≥4 Rαβ =  αβ + Weylαβ
Scalarαβ + Ricci
● The Scalarαβ -piece is given by
1
Scalarαβ := − R ϑ α ∧ ϑβ , R := eα Ricα , Ricα := eβ Rα β ,
n(n − 1)
(20.20)
where R is the curvature scalar and Ricα the Ricci 1-form. This piece has 1
independent component and is present in any dimension n > 1. In components
we have
Ricα = Ricμα ϑμ , Ricμα = Rλμα λ , R = Rλμ μλ , (20.21)
and
2
Scalarμναβ = − R gμ[α gβ]ν . (20.22)
n(n − 1)
The Scalar piece enjoys the obvious symmetry
Scalarαβ ∧ ϑβ = 0 , Scalar[μνα]β = 0 . (20.23)
● From dimension 3 onwards the tracefree Ricci piece comes into play,
2 1
 αβ := −
Ricci ϑ[α ∧ Ric
β] , β := Ricβ −
Ric R ϑβ . (20.24)
n−2 n
It has 12 (n + 2)(n − 1) independent components. In index notation this
corresponds to
4 1
 μναβ = −
Ricci  β| ν] ,
g[μ| [α Ric  αβ = Ricαβ −
Ric R gαβ . (20.25)
n−2 n
408 Cotton Tensor in Riemannian Spacetimes

If we contract the first Bianchi identity (20.14), we find

0 = eβ (Rα β ∧ ϑα ) = Ricα ∧ ϑα , (20.26)
since Rα α = 0 in a Riemannian space. Thus, Ricμα ϑμ ∧ ϑα = 0 or
Ricαβ = Ricβα , (20.27)
that is, the Ricci tensor is symmetric. This also implies
 αβ ∧ ϑα = 0 .
Ricci (20.28)
● Finally, in dimension greater than three, the Weyl 2-form emerges according to
Weylαβ := Rαβ − Scalarαβ − Ricci
 αβ . (20.29)
From the construction it is clear that the Weyl 2-form is totally traceless, i. e.,
eα Weylαβ = −eβ Weylαβ = 0 , eα eβ Weylαβ = 0 . (20.30)
This property also explains the vanishing of the Weyl 2-form in three dimen-
sions. An arbitrary antisymmetric tensor-valued 2-form Aαβ = −Aβα =
Aμναβ ϑμ ∧ ϑν /2 in three dimensions has nine independent components. The
condition eα Aαβ = 0 results in three 1-forms, i.e., nine constraint equations
that eventually yield the vanishing of all components.
 αβ ,
According to Heinicke (2001), we can combine Scalarαβ and Ricci
2
 αβ = −
Scalarαβ + Ricci ϑ[α ∧ Lβ] , (20.31)
n−2
with
1
Lα := eβ Rα β − R ϑα , (20.32)
2(n − 1)
i.e., this sum can be expressed in a coherent way in terms of the 1-form Lα . From
Scalarαβ and Ricci αβ it inherits the property
Lα ∧ ϑα = 0 . (20.33)
We may expand Lα in components as
1
Lαβ = Lβα = Ricαβ − R gαβ . (20.34)
2(n − 1)
This tensor is sometimes called Schouten tensor. Also the names rho tensor or
Pαβ can be found in the literature. Then the curvature 2-form can be expressed
as
2
Rαβ = Weylαβ − ϑ[α ∧ Lβ] (20.35)
n−2
or, in components,
4
Weylαβγδ = Rαβγδ + g[α|[γ Lδ]|β] . (20.36)
n−2
 20.2 Cotton 2-Form 409

20.2 Cotton 2-Form

By applying the exterior covariant derivative to (20.35), we obtain the following
representation of the second Bianchi identity,
2
0 = DRαβ = DWeylαβ + ϑ[α ∧ Cβ] , (20.37)
n−2
where we encounter the Cotton 2-form
1
Cα := DLα = Cμνα ϑμ ∧ ϑν (20.38)
2
or, in components,
 
1
Cαβγ = 2 ∇[α Ricβ]γ − ∇[α Rgβ]γ . (20.39)
2(n − 1)
We perform an irreducible decomposition of the Cotton 2-form with respect to
the Lorentz group. We can use the decomposition for the torsion, since this is
also a vector-valued 2-form. Then we have
Cα = (1)
Cα + (2)
Cα + (3)
Cα
(20.40)
= TENCOT + TRACOT + AXICOT ,
with the following counting of their independent components
1 2 1 1
n (n − 1) = n(n2 − 4) + n + n(n − 1)(n − 2) , (20.41)
2 3 6
where
1
(2)
C α := ϑα ∧ (eβ C β ) , (20.42)
n−1
1
C := eα (Cβ ∧ ϑβ ) ,
(3) α
(20.43)
3
C := C α − (2) C α − (3) C α ,
(1) α
(20.44)
or, in components,
2
(2)
Cμν α = − δ α Cν]β β , (20.45)
n − 1 [μ
(3) 1
Cμν α = C[μνβ] g αβ , (20.46)
3!
(1)
Cμν α = Cμν α − (2) Cμν α − (3) Cμν α . (20.47)
TENCOT, TRACOT, and AXICOT are the computer algebra names of the
pieces of the Cotton 2-form, denoting the tensor, the trace, and the axial pieces,
respectively. The number of independent components of these pieces is given in
(20.41). They arise as follows: TRACOT corresponds to a scalar-valued 1-form
C := eα C α with n independent components. In general, a (co-)vector-valued
2-form has
 
n n2 (n − 1)
n = (20.48)
2 2
410 Cotton Tensor in Riemannian Spacetimes

independent components. AXICOT corresponds to a scalar valued 3-form

(C α ∧ ϑα ) and thus has
 
n n(n − 1)(n − 2)
= (20.49)
3 6
independent components. Thus, TENCOT is left with
n2 (n − 1) n(n − 1)(n − 2) n
− − n = (n − 2)(n + 2) (20.50)
2 6 3
independent components.
We now show that in a Riemannian space the trace piece (TRACOT) and the
axial piece (AXICOT) vanish. Hence, only the tensor piece (TENCOT) with its
n(n2 − 4)/3 independent components survives. This insight seems to be new. For
n = 3, we have 5 and for n = 4 (the case of GR) 16 independent components.
In order to see the vanishing of AXICOT, we contract the Cotton 2-form with
the coframe and use (20.33):
ϑα ∧ Cα = ϑα ∧ DLα = −D(ϑα ∧ Lα ) = 0 , (20.51)
or
2
∇[μ Lνα] = 0 .
C[μνα] = (20.52)
3!
The second Bianchi identity leads to a vanishing trace of the Cotton 2-form
(TRACOT), C = eα C α = 0. In order to see this, we contract (20.37) twice:
n−3 α 1
0 = eβ DRαβ = eβ DWeylαβ − C − ϑα ∧ C , (20.53)
n−2 n−2

0 = eα eβ DRαβ = eα eβ DWeylαβ − 2C = −2C , (20.54)

or  
1
Cαβ = ∇α
α
Ricβ α
− R δβα = 0. (20.55)
2
As we see, the second Bianchi identity relates the derivative of the Weyl 2-form
to the Cotton 2-form,
n−3
eβ DWeylα β = Cα . (20.56)
n−2
This formula allows us to rewrite the Einstein equation as a Maxwell-like equa-
tion for the Weyl tensor, see Bini et al. (2001), e. g., the Ricci identity intertwines
the derivative of the Cotton 2-form with the Weyl 2-form,
2
DCα = DDLα = −Rα β ∧ Lβ = −Weylα β ∧ Lβ + ϑ[α ∧ Lβ] ∧ Lβ
n−2
= −Weylα β ∧ Lβ . (20.57)
Consequently, in three dimensions, Cα is a covariantly conserved 2-form, with
DCα = 0. Thus it is a candidate for a conserved current that can be derived
 20.2 Cotton 2-Form 411

by means of a variational procedure. The properties of the Cotton tensor are

summarized in Table 20.2.1.
Something similar emerges in four dimensions. By means of the Ricci identity
and the decomposition of the curvature we have

DD Cα = −Rα β ∧  Cβ
= −Weylα β ∧  Cβ − Ricci
 α β ∧  Cβ − Scalarα β ∧  Cβ .
(20.58)

For p-forms φ, ψ of the same degree, there holds  φ ∧ ψ =  ψ ∧ φ. By means of

ϑα ∧  φ = (−1)p−1  (eα φ) we can prove that Scalarαβ ∧  C α = 0. Performing
a “partial integration” we arrive at
   
DD Cα = −D  Weylα β ∧ Lβ + D Weylα β ∧ Lβ −  Ricci  α β ∧ Cβ . (20.59)

Next, we use the “double duality relations” for the irreducible pieces of the
curvature,
 1 μν
Weylαβ = Weylμν η αβ , (20.60)
2
1

 αβ
Ricci = −Ricci
 μν η μν αβ , (20.61)
2
 1 μν
Scalarαβ = Scalarμν η αβ . (20.62)
2
Together with eqs.(20.37) and (20.31), we obtain
  1 1
D Weylα β ∧ Lβ = η μν α β DWeylμν ∧ Lβ = − η μν α β ϑ[μ ∧ Cν] ∧ Lβ
2 2
1 μβν
=− η α ϑμ ∧ Lβ ∧ Cν
2
= − Ricci

 ν α ∧ Cν +  Scalarν α ∧ Cν
 α ν ∧ Cν .
=  Ricci (20.63)

Substituting this into (20.59) one gets

 
DD Cα = −D Weylα β ∧ Lβ . (20.64)

Thus,
Bα := D Cα +  Weylα β ∧ Lβ =: Bα β ηβ (20.65)

or, in components,
Bαβ = ∇μ Cαμβ + Lμν Weylαμβν , (20.66)

is a covariantly conserved 3-form:

DBα = 0 (∇β Bα β = 0) . (20.67)

412 Cotton Tensor in Riemannian Spacetimes

One recognizes the Bach tensor Bαβ , Bach (1921); Schouten (1954); Penrose and
Rindler (1986); Tsantilis et al. (1996) . From the symmetry properties of Cα , Lα ,
and Weylαβ it follows that
Bα ∧ ϑα = 0 (Bα α = 0) , eα B α = 0 (B[αβ] = 0) . (20.68)
Moreover, it transforms as a conformal density and can be derived from a vari-
ational principle. Since in a conformally flat space the Weyl and the Cotton
tensors vanish, the vanishing of the Bach tensor is also a necessary (but not
sufficient) condition for a 4D space to be conformally flat.

Cα as a Variational Derivative
It is well known Deser et al. (1984); Baekler et al. (1992) that Cα can be obtained
by means of varying the 3D Chern–Simons action
 
1 2
CRR = − Γα β ∧ dΓβ α − Γα β ∧ Γβ γ ∧ Γγ α (20.69)
2 3
with respect to the metric keeping the connection fixed. In order to enforce
vanishing torsion
T α := Dϑα = dϑα − Γβ α ∧ ϑβ (20.70)
and vanishing non-metricity
Qαβ := −Dgαβ = −dgαβ + Γα γ gγβ + Γβ γ gαγ , (20.71)
we have to apply Lagrange multipliers. Then the total Lagrangian reads
L = CRR + λα ∧ T α + λαβ ∧ Qαβ , (20.72)
where λα is a 1-form and λαβ = λβα a symmetric 2-form.
The variation of this Chern–Simons Lagrangian (20.69), which only depends
on the connection, turns out to be
1  
δCRR = −δΓα β ∧ Rβ α + d Γα β ∧ δΓβ α . (20.73)
2
In the next step, we enforce vanishing torsion and non-metricity by means of
respective Lagrange multiplier terms:
L = CRR + λα ∧ T α + λαβ ∧ Qαβ . (20.74)
The variation then yields
δL = δCRR + δλα ∧ T α + λα ∧ δT α + δλαβ ∧ Qαβ + λαβ ∧ δQαβ
 
= −δΓα β ∧ Rβ α + δλα ∧ T α + λα ∧ dδϑα + δΓβ α ∧ ϑβ + Γβ α ∧ δϑβ
1  
+ d Γα β ∧ δΓβ α + δλαβ ∧ Qαβ
2
+λαβ ∧ (−dδgαβ + δΓα γ gγβ + Γα γ δgγβ + δΓβ γ gαγ + Γβ γ δgαγ )
 20.2 Cotton 2-Form 413

1  β 
= −δΓα β ∧ Rβ α + δλα ∧ T α + δλαβ ∧ Qαβ + d Γα ∧ δΓβ α
2
+λα ∧ Dδϑα − δΓβ α ∧ λα ∧ ϑβ − λαβ ∧ Dδgαβ + δΓα β ∧ (λα β + λβ α )
= δλα ∧ T α + δλαβ ∧ Qαβ + δϑα ∧ Dλα + δgαβ Dλαβ
−δΓα β ∧ (Rβ α + λβ ∧ ϑα − λα β − λβ α )
 
1 β
−d −λα ∧ δϑ + Γα ∧ δΓβ − λ δgαβ .
α α αβ
(20.75)
2
The corresponding field equations read
δL
= Tα = 0, (20.76)
δλα
δL
= Qαβ = 0 , (20.77)
δλαβ
δL
= −Rβ α − λβ ∧ ϑα + 2λβ α = 0 , (20.78)
δΓα β
δL
= Dλα = 0 , (20.79)
δϑα
δL
= Dλαβ = 0 . (20.80)
δgαβ
We can solve (20.78) for its symmetric and its antisymmetric parts,

R[αβ] + ϑ[α ∧ λβ] = 0 , (20.81)

−R(αβ) + ϑ(α ∧ λβ) + 2λαβ = 0 . (20.82)

Because of (20.77), the symmetric part of the curvature vanishes,

0 = DQαβ = −DDgαβ = Rα γ gγβ + Rβ γ gαγ = 2R(αβ) . (20.83)

Thus, by means of (20.82)

1
λαβ = − ϑ(α ∧ λβ) . (20.84)
2
According to (20.35), in three dimensions, Rαβ = −2 ϑ[α ∧ Lβ] . We substitute
this into (20.81) and find

λβ = 2 Lβ , λαβ = −ϑ(α ∧ Lβ) . (20.85)

Eventually,
1 δL δL 2 δL
= Cα , = −ϑ(α ∧ C β) , − eβ  = Cα . (20.86)
2 δϑα δgαβ n−1 δgαβ
In the presence of matter, the gravitational field equation is given by δ(L +
Lmat )/δϑα = 0. Hence, the Cotton 2-form can be coupled to the energy–
momentum 2-form of matter
δLmat
Σα := . (20.87)
δϑα
414 Cotton Tensor in Riemannian Spacetimes

This is carried out in the topologically massive gravity model of Deser, Jackiw
and Templeton (DJT); see Deser et al. (1982a, 1982b), where the Lagrangian
(20.72) is enriched by a Hilbert–Einstein term and a cosmological term ( is the
gravitational constant and θ a dimensionless coupling constant):

LDJT = θ CRR + VHE + VΛ + λα ∧ T α + λαβ Qαβ + Lmat

1 αβ Λ
= θ CRR − R ∧ ηαβ − η + λα ∧ T α + λαβ Qαβ + Lmat . (20.88)
2 
Then the field equation (20.79) reads
1
Gα + Ληα + Cα = Σα , (20.89)
μ
where Gα = 12 ηαβγ ∧ Rβγ is the Einstein 2-form and the DJT coupling constant
1/μ = −2θ. The model of topologically massive gravity in Riemann–Cartan
space by Baekler et al. (1992) considers additionally to the Chern–Simons term
for the curvature also a corresponding term for the torsion. This model includes
Einstein, Einstein–Cartan, and the DJT field equations as limiting cases.
In three dimensions the Cotton tensor arises from the variation of the topo-
logical Chern–Simons term. Recently, a similar procedure was proposed by
Jackiw et al. Jackiw and Pi (2003); Guralnik et al. (2003); Jackiw (2006) for
the case n = 4 by starting from the corresponding four-dimensional topological
Lagrangian (1/2) θ Rα β ∧ Rβ α = θ dCRR , where θ is an external, prescribed field.
The variation with respect to the metric yields a four-dimensional Cotton type
tensor that differs from the one in our definition (20.38) with (20.32).
The Einstein (n − 1)-form Gα is equivalent to the 1-form Lα according to

Gα = Lβ ∧ ηβα ; (20.90)

see Heinicke (2001). Hence, we may rewrite the DJT-field equation as a

differential equation for Lα ,

DLα + μ Lβ ∧ ηβα = μ Σα − μΛηα . (20.91)

The Bianchi identities imply full integrability of this system.

In the case of Einstein gravity, the equivalence of Gα and Lα implies a rela-
tion between the Cotton 2-form and the energy–momentum (n − 1)-form in any
dimension. We can solve
Gα + Λ ηα = Σα (20.92)

for Lα and obtain by covariant exterior differentiation

1
Cα = (−1)n−1+ind  D Σα − eα (ϑβ ∧ Σβ ) . (20.93)
n−1
Note that the cosmological constant Λ, which induces a constant curvature term,
drops out.
 20.3 Conformal Correspondence 415

Table 20.2.1 Properties of the Cotton 2-form Cα in arbitrary

dimensions
1
Cα := DLα , Lα := eβ Rα β − 2(n−1) R ϑα Cotton 2-form
ϑα ∧ C α = 0 (axialfree) 1st Bianchi identity
eα C α = 0 (tracefree) contracted 2nd Bianchi identity
2
DWeylαβ = − n−2 ϑ[α ∧ Cβ] 2nd Bianchi identity
DCα = −Weylα β ∧ Lβ Ricci identity
α = Cα + (n − 2) σ,β Weyl β
C conformal transformation
α
1
Cα = D [Σα − n−1 eα (Σβ ∧ ϑβ )] Einstein equation differentiated

We recognize that all vacuum solutions of Einstein’s theory have a vanishing

Cotton 2-form. Therefore, via the Bianchi identity, the Weyl 2-form is diver-
genceless, see (20.56). This property considerably simplifies the classification of
the Weyl tensor. Petrov type D spacetimes with vanishing Cotton 2-form have
been classified in Ferrando and Sáez (2002).

20.3 Conformal Correspondence

The conformal correspondence between two n-dimensional manifolds Vn and V2n
is achieved by means of a conformal transformation of the form Eisenhart (1966);
Schouten (1954)
ĝαβ = exp(2σ)gαβ , ĝ αβ = exp(−2σ)g αβ , (20.94)
where σ is an arbitrary function. In general, a conformal transformation (20.94) is
not associated with a transformation of coordinates, i.e., with a diffeomorphism
of Vn ; both metrics in (20.94) are given in the same coordinate system and frame.
Since these transformations preserve angles between corresponding directions,
the causal structure of the manifold is preserved. As a rule, indices of quantities
with hat , Â, are raised and lowered by means of ĝ αβ or ĝαβ , respectively, those
of untransformed quantities by g αβ or gαβ . The transformed connection reads
 
2 α β = Γα β + δαβ dσ − ϑα σ ,β + σ,α ϑβ =: Γα β + Sα β ,
Γ (20.95)
a comma denotes partial and a semicolon covariant differentiation. If D̂ =
2 α β , the
d + Γα β + Sα β is the exterior covariant derivative with respect to Γ
transformed curvature is
2 α β = dΓ
R 2α β − Γ
2α γ ∧ Γ
2 γ β = Rα β + 2 ϑ[α ∧ Sγ] g γβ , (20.96)
with
1 ,α
Sγ := Dσ,γ − σ,γ dσ + σ σ,α ϑγ . (20.97)
2
By contracting (20.96) with the frame eβ , we infer
2 α = Lα − (n − 2) Sα ,
L (20.98)
416 Cotton Tensor in Riemannian Spacetimes

 α β = Weylα β ,
Weyl (20.99)
2
R = exp(−2σ) [R − 2(n − 1) σ ;α − (n − 1)(n − 2)σ,α σ ] . (20.100)
,α ,α

The Weyl 2-form is conformally invariant since a conformal transformation does

2 onto (20.98)
not act on the trace-free part of the curvature. Application of D
yields the transformation behavior of the Cotton 2-form,
2α = Cα + (n − 2) σ,β Weylα β .
C (20.101)

Thus, in n = 3, where the Weyl 2-form vanishes, the Cotton 2-form becomes
conformally invariant.

20.4 Criteria for Conformal Flatness

In the following paragraphs we investigate the criteria for conformal flatness,
i. e., the possibilities to transform the curvature to zero by means of a conformal
transformation. We basically follow Schouten (1954). Since we have seen that
the curvature 2-form in two, three, and more than three dimensions is built up
rather differently, we have to investigate these cases separately.
n=2
In n = 2 the only non-vanishing curvature piece is the curvature scalar R. Its
behavior under conformal transformation is given by
 
R2 = exp(−2σ) R − 2 σ ,i ;i = 0 . (20.102)

Thus,
2=0 R
R ⇐⇒ σ ,i ;i = . (20.103)
2
This is a scalar wave equation for the conformal factor σ with R as source. Since
the wave equation always has a solution, we conclude that all 2D spaces are
conformally flat.
n≥3
For more than two dimensions we start from (20.35), namely
2
Rαβ = Weylαβ − ϑ[α ∧ Lβ] . (20.104)
n−2
Since the Weyl 2-form is conformally invariant it cannot be transformed to zero
by means of a conformal transformation. Consequently, the vanishing of the Weyl
2-form is a necessary condition for conformal flatness.
The Lα 1-form transforms according to
2 α = Lα − (n − 2) Sα .
L (20.105)

We can transform Lα to zero if there is a function σ such that

Lα = (n − 2) Sα . (20.106)
 20.5 Classification of the Cotton 2-Form in 3D 417

This will impose a differential restriction on Lij . By means of (20.97), we rewrite

the latter equation as a differential equation for σ,i ,
1 ,j 1
Dσ,i = σ,i σ,j ϑj − σ σ,j ϑi + Li . (20.107)
2 n−2
If we apply the covariant derivative to both sides of (20.107), we obtain a
necessary condition for the integrability,
1
− Ri j σ,j = DDσ,i = σ,j Dσ,i ∧ ϑj − σ ,j Dσ,j ∧ ϑi + Ci . (20.108)
n−2
This becomes a necessary and sufficient condition of integrability if the depen-
dence on σ,i can be eliminated; see Schouten (1954, 1958). Thus we substitute
Dσ,i from (20.107) into (20.108):
2 1
− Ri j σ,j = − L[i ∧ ϑj] σ ,j + Ci . (20.109)
n−2 n−2
Using the decomposition (20.35) of the curvature, we finally arrive at

− (n − 2) Weyli j σ,j = Ci . (20.110)

For n = 3, the Weyl 2-form is zero and Cα = 0 is the integrability condition

for the conformal factor. Thus, if the Cotton 2-form is zero, the space is con-
formally flat. Conversely, if the space is conformally flat, there is a conformal
transformation such that R 2α β = 0 ⇔ L 2α = 0 ⇒ C 2α = 0. Since the Cotton
2-form is conformally invariant in three dimensions, we find Cα = 0. Hence, the
vanishing of the Cotton 2-form is the necessary and sufficient condition for a V3
to be conformally flat.
In more than three dimensions the vanishing of the Weyl 2-form is a necessary
condition for conformal flatness. Thus, also in dimensions greater than three,
Cα = 0 is the integrability condition for the conformal factor. However, for
n > 3, the contracted second Bianchi identity (20.53) implies the vanishing of
the Cotton 2-form when the Weyl 2-form is zero. Hence, the vanishing of the
Weyl 2-form is also the sufficient condition for conformal flatness.

20.5 Classification of the Cotton 2-Form in 3D

A vector-valued 2-form in three dimensions has nine independent components,
the same as the number of components of a 3 × 3 matrix. A mapping between
these two can be achieved by means of the Hodge dual. The Hodge dual of a
vector-valued 2-form in three dimensions is a vector-valued 1-form with the same
number of independent components. Its components form a second rank tensor
(“matrix”),
Cαβ := eα  Cβ =  (Cβ ∧ ϑα ) (20.111)
418 Cotton Tensor in Riemannian Spacetimes

or, in components,
 
1
Cα = ∇μ
β
Ricνα − Rgνα η μνβ . (20.112)
4
This alternative representation of the Cotton 2-form, often called Cotton–York
tensor – see York Jr. (1971) (even though it was already discussed explicitly by
ADM Arnowitt et al., 1962) – can only be defined in three dimensions. Sometimes
it appears under the name Bach tensor in the literature; see Christodoulou and
Klainerman (1993), for example This seems to be a misnomer.
The Cotton tensor is tracefree
Cα α = eα  C α =  (C α ∧ ϑα ) = 0 . (20.113)
In the three dimensions, the second Bianchi identity (20.37) amounts to ϑ[α ∧
Cβ] = 0. In view of the definition (20.111), we infer that the Cotton tensor is
symmetric Cαβ = Cβα . Introducing this symmetry explicitly into (20.112), we
obtain the alternative representation
C αβ = C βα = η μν(α ∇μ Ricν β) . (20.114)
We now perform a classification of the Cotton tensor with respect to its
eigenvalues. The corresponding generalized eigenvalue problem reads:
 αβ 
C − λ g αβ Vβ = 0 , C [αβ] = 0 , C αβ gαβ = 0 . (20.115)
By lowering one index, we can reformulate this as ordinary eigenvalue problem
for the matrix Cα β . However, in that case, the symmetry C αβ = C βα is no longer
manifest:
 β 
Cα − λ δαβ Vβ = 0 , Cα α = 0 . (20.116)

20.5.1 Euclidean Signature

The case of Euclidean signature is simple: the generalized eigenvalue problem
reduces to an ordinary one. As a real symmetric matrix C αβ possesses three real
eigenvalues and the eigenvectors form a basis. With respect to this basis, C αβ
takes a diagonal form. Since C αβ is tracefree, the sum of the eigenvalues is zero.
Consequently, we can distinguish three classes:

● Class A
Three distinct eigenvalues: λ1 = λ2 and λ3 = −(λ1 + λ2 ).
● Class B
Two distinct eigenvalues: λ1 = λ2 = 0, λ3 = −2λ1 .
● Class C
One distinct eigenvalue: λ1 = λ2 = λ3 = 0.
In the present context of Euclidean signature, this implies Cαβ = 0.
 20.5 Classification of the Cotton 2-Form in 3D 419

20.5.2 Lorentzian Signature

In the case of an indefinite metric, the roots of the characteristic polynomial
 
det C αβ − λ g αβ = 0 (20.117)

may be complex. Accordingly, the matrix Cα β is no longer symmetric and in the

equivalent ordinary eigenvalue problem
 
det Cα β − λ δαβ = 0 (20.118)

complex eigenvalues occur, too. This point seems to have been overlooked by the
authors of Barrow et al. (1986). Consequently, the classification will not be as
simple as was the case for the Euclidean metric.
In the following, we will present a classification of Cα β . The tracefree condition
(20.116)2 , in orthonormal coordinates, reads explicitly

C1 1 + C2 2 + C3 3 = 0 . (20.119)

Accordingly, we can eliminate C3 3 , e.g., from (20.116)1 . Then the secular

determinant reads
 
 C1 1 − λ C1 2 C1 3 
 

det  −C1 2
C2 − λ
2
C2 3  = 0, (20.120)

 −C 3 C2 3
−C1 − C2 − λ 
2 2
1

with the five matrix elements C1 2 , C1 2 , C1 3 , C2 2 , C2 3 . We compute the determi-

nant and order according to powers of λ,

λ3 + b λ + c = 0 . (20.121)

A cubic polynomial with real coefficients has at least one real root and the
complex roots have to be complex conjugates. Since one eigenvalue is real, types
D and II with only one independent eigenvalue λ1 = λ2 = −2λ3 are always real.
For class I, besides the real eigenvalue, two complex eigenvalues may occur. In
that case, they are complex conjugated. Therefore, class I can be subdivided
into class I with three real eigenvalues, [111], and class I with one real and two
complex conjugated eigenvalues, [1z z̄].
We can now specify simple criteria for deciding to which of these classes the
Cotton tensor Cα β belongs. First determine the eigenvalues.

1. Three different eigenvalues (2 independent)

1. all real ⇒ Class I
2. one real, two complex ⇒ Class I
2. Two different eigenvalues (1 independent λ1 = λ2 = −2λ3 )
1. (Cα β − λ1 δαβ )(Cβ γ + 12 λ1 δβγ ) = 0 ⇒ Class D
2. else ⇒ Class II
420 Cotton Tensor in Riemannian Spacetimes

3. All eigenvalues zero

1. Cα β = 0 ⇒ 0
2. Cα β Cβ γ = 0 ⇒ Class N
3. else ⇒ Class III
The Jordan normal forms of the Cotton tensor read:

“Petrov” Jordan form Segré eigenvalues

type notation λ1 + λ2 + λ3 = 0
⎛ ⎞
λ1 0 0
I ⎝ 0 λ2 0 ⎠ [111] Real; λ1 = λ2
⎛ 0 0 −λ1 − λ2 ⎞
Re z Im z 0
I ⎝ −Im z Re z 0 ⎠ [1z z̄] λ2 = z, λ3 = z̄
⎛ 0 0 −2 Re⎞
z
λ1 0 0
D ⎝ 0 λ1 0 ⎠ [(11)1] λ1 = λ2 = 0, λ3 = −2λ1
0 0 −2λ1
⎛ ⎞
λ1 1 0
II ⎝ 0 λ1 0 ⎠ [21] λ1 = λ2 = 0, λ3 = −2λ1
⎛ 0 0 −2λ⎞1
0 1 0
N ⎝ 0 0 0 ⎠ [(21)] λ1 = λ2 = λ3 = 0
⎛ 0 0 0 ⎞
0 1 0
III ⎝ 0 0 1 ⎠ [3] λ1 = λ2 = λ3 = 0
⎛ 0 0 0 ⎞
0 0 0
O ⎝ 0 0 0 ⎠
0 0 0

This parallels exactly the Petrov classification of the Weyl tensor in four dimen-
sions; see Stephani et al. (2003). This comes about since the Weyl tensor in 4D is
equivalent to a (complex) 3 × 3 tracefree matrix, as Cα β in 3D; for a similar clas-
sification of Cαβ , see Hall and Capocci (1999). A curvature classification in terms
of spinors has been done by Torres del Castillo and Gómez-Ceballos (2003).
 References

Anninos, D., Li, W., Padi, M., Song, W., and Strominger, A. (2009). “Warped AdS3
black holes,” JHEP 0903, 130, arXiv:0807.3040.
Aragone, C. (1987). “Topologically massive gravity in the dreibein light-front gauge,”
Class. Quant. Grav. 4, L1.
Ayón-Beato, E., Cataldo, M., and Garcia, A. A. (2005). “Electromagnetic fields in
stationary cyclic symmetric 2+1 gravity”, In Proceedings of the 10th. PASCOS04
and Pran Nath Fest, Eds. G. Alverson, E. Barberi, P. Nath, M.T. Vaughn (World
Scientific, 2005), p.359.
Ayón-Beato, E., Garcia, A. A., Macias, A., and Perez-Sanchez, J. M. (2000). “Note
on scalar fields nonminimally coupled to (2+1) gravity,” Phys. Lett. B495, 164
[gr-qc/0101079].
Ayón-Beato, E. and Hassaı̈ne, M. (2005). “pp–waves of conformal gravity with self-
interacting source,” Annals Phys. 317, 175. hep-th/0409150.
Ayón-Beato, E. and Hassaı̈ne, M. (2006). “Exploring AdS waves via nonminimal
coupling,” Phys. Rev. D 73, 104001. hep-th/0512074.
Ayón-Beato, E., Martinez, C., and Zanelli, J. (2004). “Birkhoff’s theorem for three-
dimensional AdS gravity,” Phys. Rev. D70, 044027. [hep-th/0403227].
Bañados, M., Henneaux, M. Teitelboim, C., and Zanelli, J. (1993). “Geometry of the
(2+1) black hole,” Phys. Rev. D48, 1506. [gr-qc/9302012].
Bañados, M., Teitelboim, C., and Zanelli, J. (1992). “The black hole in three-
dimensional spacetime,” Phys. Rev. Lett. 69, 1849. [hep-th/9204099].
Barrow, J. D., Burd, A. B., and Lancaster, D. (1986). “Three-dimensional classical
spacetimes,” Class. Quant. Grav. 3, 551–567.
Barrow, J. D., Shaw, D. J., and Tsagas, C. G. (2006). “Cosmology in three dimensions:
Steps towards the general solution,” Class. Quant. Grav. 23, 5291. [gr-qc/0606025].
Bouchareb A. and Clément, G. (2007). “Black hole mass and angular momentum in
topologically massive gravity,” Class. Quant. Grav. 24, 5581. arXiv:0706.0263.
Brown, J. D., Creighton, J., and Mann, R. B. (1994). “Temperature, energy and heat
capacity of asymptotically anti-de Sitter black holes,” Phys. Rev. D50, 6394. [gr-
qc/9405007].
Brown, J. D. and York Jr., J. W. (1993). “Quasilocal energy and conserved charges
derived from the gravitational action”, Phys. Rev. D47, 1407.
Carlip, S. (1998). Quantum Gravity in 2+1 Dimensions, Cambridge, UK: Cambridge
University Press.
Carlip, S., Deser, S., Waldron, A., and Wise, D. K. (2008). “Topologically massive AdS
gravity,” Phys. Lett. B666, 272. arXiv:0807.0486.
Carlip, S., Deser, S., Waldron, A., and Wise, D. K. (2009). “Cosmological topologically
massive gravitons and photons,” Class. Quant. Grav. 26, 075008.
422 References

Cataldo, M. (2002). “Azimuthal electric field in a static rotationally symmetric (2+1)-

dimensional space-time,” Phys. Lett. B529, 143. [gr-qc/0201047].
Cataldo, M. (2004). “Rotating perfect fluids in (2+1)–dimensional Einstein gravity,”
Phys. Rev. D69, 064015.
Cataldo, M., Crisostomo, J., del Campo, S., and Salgado, P. (2004). “On magnetic
solution to (2+1) Einstein–Maxwell gravity,” hep-th/0401189.
Cataldo, M., Cruz, N., del Campo, S., and Garcia, A. (2000). “(2+1)–dimensional black
hole with Coulomb-like field,” Phys. Lett. B484, 154. [hep-th/0008138].
Cataldo, M., del Campo, S., and Garcia, A. (2001). “BTZ black hole from (3+1)
gravity,” Gen. Rel. Grav. 33, 1245. [gr-qc/0004023].
Cataldo, M., and Garcı́a, A. (1999). “Three dimensional black hole coupled to the
Born–Infeld electrodynamics,” Phys. Lett. B456, 28.
Cataldo, M. and Garcia, A. (2000). “Regular (2+1)-dimensional black holes within
nonlinear electrodynamics,” Phys. Rev. D61, 084003. [hep-th/0004177].
Cataldo, M. and Salgado, P. (1996). “Static Einstein–Maxwell solutions in (2+1)-
dimensions,” Phys. Rev. D54, 2971.
Cavaglia, M. (1999). “The Birkhoff theorem for topologically massive gravity,” Grav.
Cosmol. 5, 101. gr-qc/9904047.
Chan, K. C. K. (1996). “Comment on the calculation of the angular momentum for
the (anti)selfdual charged spinning BTZ black hole,” Phys. Lett. B373, 296. [gr-
qc/9509032].
Chan, K. C. K. (1997). “Modifications of the BTZ black hole by a dilaton and scalar,”
Phys. Rev. D55, 3564. [gr-qc/9603038].
Chan, K. C. K. and Mann, R. B. (1994). “Static charged black holes in (2+1)-
dimensional dilaton gravity,” Phys. Rev. D50, 6385, [Erratum: Phys. Rev. D52,
2600. (1995)]. [gr-qc/9404040].
Chan, K. C. K. and Mann, R. B. (1996). “Spinning black holes in (2+1)-dimensional
string and dilaton gravity,” Phys. Lett. B371, 199. [gr-qc/9510069].
Chow, D. D. K., Pope, C. N., and Sezgin, E. (2010a). “Classification of solutions in
topologically massive gravity,” Class. Quant. Grav. 27, 105001.
Chow, D. D. K., Pope, C. N., and Sezgin, E. (2010b). “Kundt spacetimes as solutions
of topologically massive gravity,” Class. Quant. Grav. 27, 105002.
Clément, G. (1976). “Field-theoretic extended particles in two space dimensions,” Nucl.
Phys. B114, 437.
Clément, G. (1985a). “Stationary solutions in three-dimensional general relativity,” Int.
J. Theor. Phys. 24, 267.
Clément, G. (1992a). “Stationary rotationally symmetric solutions in topologically
massive gravity, Class. Quant. Grav. 9, 2615.
Clément, G. (1992b). “Localised solutions in topologically massive gravity,” Class.
Quant. Grav. 9, S35.
Clément, G. (1993). “Classical solutions in three-dimensional Einstein–Maxwell cosmo-
logical gravity,” Class. Quant. Grav. 10, L49.
Clément, G. (1994). “Particle-like solutions to topologically massive gravity,” Class.
Quant. Grav. 11, L115, gr-qc/9404004.
Clément, G. (1996). “Spinning charged BTZ black holes and selfdual particle-like
solutions,” Phys. Lett. B367, 70. [gr-qc/9510025].
Coley, A., Hervik, S., and Pelavas, N. (2006). “On spacetimes with constant scalar
invariants,” Class. Quant. Grav. 23, 3053. gr-qc/0509113.
 References 423

Coley, A., Hervik, S., and Pelavas, N. (2008). “Lorentzian spacetimes with con-
stant curvature invariants in three dimensions,” Class. Quant. Grav. 25, 025008.
arXiv:0710.3903.
Coley, A., Hervik, S., and Pelavas, N. (2009). “Lorentzian spacetimes with con-
stant curvature invariants in four dimensions,” Class. Quant. Grav. 26, 125011.
arXiv:0904.4877.
Coley, A., Hervik, S., Papadopoulos, G. O., and Pelavas, N. (2009). “Kundt space-
times,” Class. Quant. Grav. 26, 105016. arXiv:0901.0394.
Collas, P. (1977). “General relativity in two-and three-dimensional space-times,” Am.
J. Phys. 45, 833.
Cornish, N. J. and Frankel, N. E. (1991). “Gravitation in (2+1)–dimensions,” Phys.
Rev. D43, 2555.
Cornish, N. J. and Frankel, N. E. (1994). “Gravitation versus rotation in (2+1)–
dimensions,” Class. Quant. Grav. 11, 723. [gr-qc/9306012].
Coussaert, O. and Henneaux, M. (1994b). “Self-dual solutions of the 2 + 1 Einstein
gravity with a negative cosmological constant,” [hep-th9407181].
Cruz, N. and Martı́nez, C. (2000). “Cosmological scaling solutions of minimally coupled
scalar fields in three dimensions” Class. Quant. Grav. 17, 2867.
Cruz, N. and Zanelli, J. (1995). “Stellar equilibrium in (2+1)-dimensions,” Class.
Quant. Grav. 12, 975 [gr-qc/9411032].
Dereli, T. and Sarioğlu, O. (2001). “Supersymmetric solutions to topologically mas-
sive gravity and black holes in three dimensions,” Phys. Rev. D 64, 027501.
gr-qc/0009082.
Dereli, T. and Tucker, R. W. (1988). “Gravitational interactions in 2 + 1 dimensions,”
Class. Quant. Grav. 5, 951.
Deser, S. (1984).“Cosmological topological supergravity,” in Quantum Theory of
Gravity, ed. S.M. Christensen, Adam Hilger, London.
Deser, S. and Jackiw, R. (1989). “String sources in (2+1)-dimensional gravity,” Ann.
of Phys. 192, 352.
Deser, S., Jackiw, R., and Pi, S. Y. (2005). “Cotton blend gravity pp waves,” Acta
Phys. Polon. B36, 27. [gr-qc/0409011].
Deser, S., Jackiw, R. and Templeton, S. (1982a). “Three-Dimensional Massive Gauge
Theories,” Phys. Rev. Lett. 48, 975.
Deser, S., Jackiw, R., and Templeton, S. (1982b). “Topological massive gauge theories,”
Ann. of Phys. 140, 372.
Deser, S., Jackiw, R. and ’t Hooft, G. (1984). “Three-dimensional Einstein gravity:
dynamics of flat space,” Ann. of Phys. 152, 220–235.
Deser, S. and Mazur, P. O. (1985). “Static solutions in D = 3 Einstein–Maxwell theory,”
Class. Quant. Grav. 2 L51–L56.
Deser, S. and Steif, A.R. (1992). “Gravity theories with lightlike sources in D = 3,”
Class. Quant. Grav. 9, L153. hep-th/9208018.
Dias, O. J. C. and Lemos, J. P. S. (2002b). “Rotating magnetic solution in three-
dimensional Einstein gravity,” JHEP 01(2002), 006. [hep-th/0201058].
Garbarz, A., Giribet, G., and Vásquez, Y. (2009). “Asymptotically AdS 3 solutions to
topologically massive gravity at special values of the coupling constants,” Phys.
Rev. D 79, 044036. arXiv:0811.4464.
Garcı́a, A. (1999). “On the rotating charged BTZ metric,” hep-th/9909111.
Garcı́a, A. A. (2004). “Stationary circularly symmetric 2+1 rigidly rotating perfect
fluid,” Phys. Rev. D69, 124024.
424 References

Garcia, A. A. (2009). “Three-dimensional stationary cyclic symmetric Einstein–

Maxwell solutions; black holes,” Ann. of Phys. 324, 2004–2050.
Garcia-Diaz, A. (2014). “Hydrodynamic equilibrium of a static star in the presence of
a cosmological constant in 2 + 1 dimensions,” arXiv:1412.5620[gr-qc].
Garcı́a, A. A. and Campuzano, C. (2003). “All static circularly symmetric perfect fluid
solutions of 2+1 gravity,” Phys. Rev. D67, 064014.
Garcı́a, A., Cataldo, M., and del Campo, S. (2003). “Relationship between (2+1) and
(3+1) Friedmann–Robertson–Walker cosmologies,” Phys. Rev. D68, 124022. [hep-
th/0309098].
Garcı́a, A., Hehl, F. W., Heinicke, C., and Macias, A. (2004). “The Cotton tensor in
Riemannian space-times,” Class. Quant. Grav. 21, 1099 [gr-qc/0309008].
Garcia-Diaz, A. A. (2013). “Three-dimensional stationary cyclic symmetric Einstein–
Maxwell solutions; energy,” [gr-qc/0309008].
Garcia-Diaz, A. and Gutierrez-Cano, G. (2014a). “Low energy 2+1 string gravity; black
hole solutions,” arXiv:1412.5618[gr-qc].
Garcia-Diaz, A. and Gutierrez-Cano, G. (2014b). “Dilaton minimally coupled to 2+1
Einstein–Maxwell fields: stationary cyclic symmetric black holes,” arXiv:1412.5621
[gr-qc].
Gibbons, G. W., Pope, C. N., and Sezgin, E. (2008). “The general supersymmetric
solution of topologically massive supergravity,” Class. Quant. Grav. 25, 205005,
arXiv:0807.2613.
Giddings, S., Abbott, J., and Kuchar, K. (1984). “Einstein’s theory in a three-
dimensional space-time,” Gen. Rel. Grav. 16, 751.
Gott, J. R. and Alpert, M. (1984). “General relativity in a (2+1)-dimensional space-
time,” Gen. Rel. Grav. 16, 243.
Gott, J. R., Simon, J. Z., and Alpert, M. (1986). “General relativity in a (2+1)-
dimensional space-time: An electrically charged solution,” Gen. Rel. Grav.
18, 1019.
Gürses, M. (1994). “Perfect fluid sources in 2+1 dimensions,” Class. Quant. Grav. 11,
2585–2587.
Gürses, M. (2008). “Gödel type metrics in three dimensions,” Gen. Rel. Grav.,42, 1413.
arXiv:0812.2576.
Hall, G. S. and Capocci, M. S. (1999). “Classification and conformal symmetry in
three-dimensional space-time,” J. Math. Phys. 40, 1466–1478.
Hall, G. S., Morgan, T., and Perjés, Z. (1987). “Three-dimensional space-times,” Gen.
Rel. Grav. 19, 1137.
Hirschmann, E. W. and Welch, D. L. (1996). “Magnetic solutions to (2+1) gravity,”
Phys. Rev. D53, 5579 [hep-th/9510181].
Horne, J. H. and Horowitz, G. T. (1992). “Exact string solution in three dimensions,”
Nucl. Phys. B368, 444.
Horowitz, G. T. (1992). “The dark side of string theory; Black holes and black strings,”
[arXiv:hep-th/9210119].
Horowitz, G. T. (2005). “Creating naked singularities and negative energy,” Phys.
Scripta T117, 86. [hep-th/0312123].
Horowitz, G. T. and Welch, D. L. (1993). “Exact three-dimensional black holes in string
theory,” Phys. Rev. Lett. 71, 328. [hep-th/9302126].
Jackiw, R. (2006). “4-dimensional Einstein gravity extended by a 3-dimen-
sional gravitational Chern–Simons term,” Int. J. Theor. Phys. 45, 1431.
[gr-qc/0310115].
Kamata, M. and Koikawa, T. (1995). “The electrically charged BTZ black hole with
self (antiself) dual Maxwell field,” Phys. Lett. B353, 196. [hep-th/9505037].
 References 425

Kamata, M. and Koikawa, T. (1997). “2+1-dimensional charged black hole with (anti-)
self dual Maxwell field,” Phys. Lett. B391, 196.
Kogan, I. I. (1992). “About some exact solutions for (2+1) gravity coupled to gauge
fields,” Mod. Phys. Lett. A7, 2341. [hep-th/9205095].
Lubo, M., Rooman, M. and Spindel, P. (1999). “(2+1)-dimensional stars,” Phys. Rev.
D59, 044012 [gr-qc/9806104].
Macı́as, A. and Camacho, A. (2005). “Kerr–Schild metric in topological massive (2+1)
gravity,” Gen. Rel. Grav. 37, 759.
Mann, R. B. and Ross, S. F. (1993). “Gravitationally collapsing dust in 2 + 1
dimensions”, Phys. Rev. D47, 3319.
Martinez, E. A. and Shepley, L. (1986). preprint, University of Texas.
Martı́nez, C., Teitelboim, C. and Zanelli, J. (2000). “Charged rotating black hole in
three space-time dimensions,” Phys. Rev. D61, 104013 [hep-th/9912259].
Martı́nez, C. and Zanelli, J. (1996). “Conformally dressed black hole in (2+1)-
dimensions,” Phys. Rev. D 54, 3830. [gr-qc/9604021].
Matyjasek, J. and Zaslavskii, O. B. (2004). “Extremal limit for charged and rotating
(2+1)–dimensional black holes and Bertotti-Robinson geometry,” Class. Quant.
Grav. 21, 4283 [gr-qc/0404090].
Melvin, M. A. (1986). “Exterior solutions for electric and magnetic stars in 2+1
dimensions,” Class. Quant. Grav. 3, 117-131.
Moussa, K. A., Clément, G., and Leygnac, C. (2003). “The black holes of topologically
massive gravity,” Class. Quant. Grav. 20, L277, gr-qc/0303042.
Nutku, Y. (1993). “Exact solutions of topologically massive gravity with a cosmological
constant,” Class. Quant. Grav. 10, 2657.
Nutku, Y. and Baekler, P. (1989). “Homogeneous, anisotropic three-manifolds of
topologically massive gravity,” Annals Phys. 195, 16.
Ortiz, M. E. (1990). “Homogeneous solutions to topologically massive gravity,” Annals
Phys. 200, 345.
Obukhov, Y. N. (2003). “New solutions in 3-D gravity,” Phys. Rev. D68, 124015.
[gr-qc/0310069].
Ölmez, S., Sarioğlu, O., and Tekin, B. (2005). “Mass and angular momentum of asymp-
totically AdS or flat solutions in the topologically massive gravity,” Class. Quant.
Grav. 22, 4355, gr-qc/0507003.
Peldan, P. (1993). “Unification of gravity and Yang-Mills theory in (2+1)-dimensions,”
Nucl. Phys. B395, 239 [gr-qc/9211014].
Percacci, R., Sodano, P., and Vuorio, I. (1987). “Topologically massive planar universes
with constant twist,” Ann. of Phys. 176, 344.
Rooman, M. and Spindel, P. (1998). “Gödel metric as a squashed anti-de Sitter
geometry,” Class. Quant. Grav. 15, 3241 [gr-qc/9804027].
Saslaw, W. C. (1977). “A relation between the homogeneity of the Universe and the
dimensional of space” Mon. Not. R. Astr. Soc. 179, 659.
Staruszkiewicz, A. (1963). “Gravitation theory in three-dimensional space,” Acta Phys.
Polon. 24, 735.
Torres del Castillo, G. F. and Gómez-Ceballos, L. F. (2003). “Algebraic classification
of the curvature of three-dimensional manifolds with indefinite metric,” J. Math.
Phys. 44, 4374 .
Vuorio, I. (1985). “Topologically massive planar universe,” Phys. Lett. B163, 91.
Gen. Rel. Grav. 23, 181–187.
Zaslavskii, O. B. (1994). “Thermodynamics of 2+1 black holes,” Class. Quant. Grav.
11, L33–L38.
426 References

Additional Literature
Abrahams, A. M. and Evans, C. R. (1993). “Critical behavior and scaling in vacuum
axisymmetric gravitational collapse,” Phys. Rev. Lett. 70, 2980.
Abramowitz, M. and Stegun, I. (1965). Handbook of Mathematical Functions. New
York: Dover Publications Inc.
Adler, R., Bazin, M., and Schiffer, M. (1965). Introduction to General Relativity,
McGraw–Hill.
Arnowitt, R., Deser, S., and Misner, C. W. (1962). “The dynamics of General Rela-
tivity.” In: Gravitation: An Introduction to Current Research, L. Witten (Ed.).
Wiley, New York.
Ashtekar, A., and Krishnan, B. (2004). “Isolated and dynamical horizons and their
applications,” Living Rev. Rel. 7, 10.
Ayón-Beato, E. and Garcı́a, A. (1998). “Regular black hole in general relativity coupled
to a nonlinear electrodynamics,” Phys. Rev. Lett. 80, 5056.
Bach, R. (1921). “Zur Weylschen Relativitätstheorie und der Weylschen Erweiterung
des Krümmungsbegriffs,” Math. Zeitschr. 9, 110–135.
Baekler, P., Mielke, E. W., and Hehl, F. W. (1992). “Dynamical symmetries in
topological 3D gravity with torsion,” Nuovo Cimento. B107, 91–110.
Barrow, J. D., and Saich, P. (1993). “Scalar–field cosmologies,” Class. Quant. Grav.
10, 279.
Bergshoeff, E. A., Hohm, O., and Townsend, P. K. (2009) “Massive gravity in three
dimensions,” Phys. Rev. Lett.102, 201301.
Bertotti, I. (1959). “Uniform electromagnetic field in the theory of general relativity,”
Phys. Rev. 116, 1331–1333.
Bini, D., Jantzen, R. T., and Minutti, G. (2001). “The Cotton, Simon–Mars and
Cotton–York tensors in stationary spacetimes,” Class. Quant. Grav. 18, 4969.
Borde, A. (1997). “Regular black holes an topological change,” Phys. Rev. D50, 7615.
Born, M., and Infeld, L. (1934). “On the quantum theory of the electromagnmetic
field,” Proc. Roy. Soc. (London) A144, 425.
Brinkmann, M. W. (1923). “On Riemann spaces conformal to Euclidean spaces,” Proc.
Natl. Acad. Sci. U.S. 9, 1.
Buchdahl, H. A. (1959). “General relativistic fluid spheres,” Phys. Rev. 116, 1027.
Buscher, T. (1993). “Path integral derivation of quantum duality in nonlinear sigma
models,” Phy. Lett.B201, 466.
Carter, B. (1968). “Hamilton–Jacobi and Schrodinger separable solutions of Ein-
stein’s equations,”Commun. Math. Phys. 10, 280 (1968), (1968). “A new family of
Einstein spaces,” Phys. Lett. A26, 399.
Cotton, E. (1899). “Sur les variétés a trois dimensions,” Ann. Fac. d. Sc. Toulouse (II)
1, 385.
Choptuik, M. W. (1993). “Universality and scaling in gravitational collapse of a
massless scalar field,” Phys. Rev. Lett. 70, 9.
Christodoulou, D. and Klainerman, S. (1993). The Global Nonlinear Stability of the
Minkowski Space. Princeton University Press, Princeton, NJ.
de Rham, C. (2014). “Massive Gravity,” Living Rev.Rel. 17, 7. arXiv:1401.4173
[hep-th].
Deser, S. and Gibbons, G. W. (1998). “Born–Infeld–Einstein actions,” Class. Quant.
Grav. 15, L35.
Ehlers, J. (1961). “Beiträge zur realativistischen Mechanik kontinuierlich Medien,” Abh.
Mainzer Akad. Wiss., Math.–natuwiss,” No, 11, English Translation: (1993) “Con-
tributions to relativistic mechanics of continuous media,” Gen. Rel. Grav. 25,
 References 427

1225–1266. See also, (2002). “Relativistic hydrodynamics,” Gen. Rel. Grav. 34,
2171.
Eisenhart, L. P. (1966). Riemannian Geometry, Princeton University Press. Princeton,
NJ.
Ellis, G. F. R. and Elst, H. (1999). Theoretical and Observational Cosmology, Ed. ML
Lachieze-Rey (Dordrecht: Kluwer) p.1.
Ferrando, J. J. and Sáez, A. J. (2002). “On the classification of Type D spacetimes,”
[gr-qc/0212086].
Fradkin, E. and Tseytlin, A. (1985). “Nonlinear electrodynamics from quantized
strings,” Phys. Lett. B163, 123.
Frolov, V., Hendy, S., and Larsen, A. L. (1996). “Stationary strings and principal Killing
triads in 2+1 gravity,” Nucl. Phys. B 468, 336.
Garcı́a, A. A. (1988). “Comments on a paper by Collinson,” Gen. Rel. Grav. 20, 589.
Gibbons, G. W. and Rasheed, D. A. (1995). “Electric–magnetic duality rotations in
non-linear electrodynamics,” Nucl. Phys. B454, 185.
Gubser, S. S., Klebanov, I. R., and Polyakov, A. M. (1998). “Gauge Theory Correlators
from Non-Critical String Theory,” Phys. Lett. B428, 105 (arXiv: hep-th/9802109).
Gundlach, C. (1999). “Critical Phenomena in Gravitational Collapse,” Living Rev. Rel.
2, 4 (arXiv: gr-qc/0001046).
Guralnik, G., Iorio, A., Jackiw, R., and Pi, S.-Y. (2003). “Dimensionally reduced gravi-
tational Chern–Simons term and its kink,” Ann. Phys. (NY) 308, 222–236. (arXiv:
hep-th/0305117).
Gürses, M. and Gürsey, Y. (1975). “Conformal uniqueness and various forms of the
Schwarzschild interior metric,” Nuovo Cimento 25 B, 786.
Hayward, S. A. (2008). “Dynamics of black holes,” arXiv:0810.0923 [gr-qc].
Hehl, F. W., McCrea, J. D., Mielke, E. W., and Ne’eman, Y. (1995). “Metric–affine
gauge theory of gravity: Field equations, Noether identities, world spinors, and
breaking of dilation invariance,” Phys. Repts. 258, 1.
Heinicke, C. (2001). “The Einstein 3-form Gα and its equivalent 1–form Lα in Riemann–
Cartan spacetime,” Gen. Relat. Grav. 33, 1115.
Israel, R. B. (1966). “Singular hypersurfaces and thin shells in General Relativity,”
Nuovo Cimento. XLIV B, 1.
Infeld, L. and Plebański, J. F. (1960). Motion and Relativity, Pergamon Press, New
York.
Jackiw, R. S. and Pi, Y. (2003). “Chern–Simons Modification of General Relativity,”
arXiv: gr-qc/0308071.
Korn, G. A. and Korn, T. M. (1961). Mathematics Handbook for Scientists and
Engineers: Definitions, Theorems, and Formulas for References and Review,
McGraw–Hill, New York.
Kottler, F. (1918). “Über die physikalischen Grudlagen der Einsteischen Gravitations-
theorie,” Annalen Physik 56, 410.
Kramer, D., Stephani, H., MacCallum, M., and Hertl, E. (1980). Exact Solutions of
the Einstein’s Field Equations, Deutsch. Ver. der Wiss., Berlin.
Landau, L. and Lifshitz, E. (1970). Théorie des Champs, Mir, Moscow,
Liddle, A. R. and Lyth, D. H. (2000). Cosmological Inflation and Large-Scale Structure,
Cambridge University Press.
Lucchin, F. and Matarrese, S. (1985). “Power–law inflation,” Phys. Rev. D32, 1316.
Madsen, M. S. (1986). “An unusual cosmological solutions for Λφ4 theory with broken
symmetry,” Gen. Rel. and Grav. 18, 879.
428 References

Maldacena, J. (1998). “The large N limit of superconformal field theories and

supergravity,” Adv. Theor. Math. Phys. 2, 231 (arXiv: hep-th/9711200).
Mandal, G., Sengupta, A. M. and Wadia, S. R. (1991). “Classical solutions of
2–dimensional string theory,” Mod. Phys. Lett. A 6 1685.
Mak, M. K. and Harko, T. (2000). “Maximum mass–radius ratio for compact general
relativistc objects in Swarzschild–de Sitter geometry,” Mod. Phys. Lett. A15, 2105.
Maki, T. and Shiraishi, K. (1993). “Multi-black hole solutions in cosmological Einstein–
Maxwell dilaton theory,” Class. Quant. Grav. 10, 2171.
Misner, C. W., Thorne, K. S., and Wheeler, J. A. (1973). Gravitation, San Francisco,
Freeman.
Ortı́n, T. (2004). Gravity and Strings, Cambridge University Press.
Oppenheimer, J. R. and Snyder, H. (1939). “On continued gravitational contraction,”
Phys. Rev. 56, 455.
Oppenheimer, J. R. and Volkoff, G. (1939). “On massive neutron cores,” Phys. Rev.
55, 374.
Penrose, R. and Rindler, W. (1986). Spinors and Space-Time. 2 Vols. Cambridge
University Press, Cambridge, UK.
Plebański, J. F. (1964). “The algebraic structure of the tensor of matter”, Acta Phys.
Pol. 26, 963.
Plebański, J. F. (1967). On Conformally Equivalent Riemannian Spaces, Monograph
of CINVESTAV-IPN, México.
Plebański, J. F. (1975). “A class of solutions of Einstein–Maxwell equations,” Ann. of
Phys. (USA) 90,196.
Polchinski, J. (1998). String Theory, 2 vols. Cambridge University Press, Cambridge,
UK.
Ponce de Leon, J. and Cruz, N. (2000). “Hydrostatic equilibrium of a perfect fluid
sphere with exterior higher-dimensional Schwarzschild spacetime,” Gen. Rel. Grav.
32, 1207.
Robinson, I. (1959). “A solution of the Einstein–Maxwell equations,” Bull. Acad. Polon.
Sci., Ser. Math. AStr. Phys. 7, 351.
Salazar, H., Garcı́a, A., and Plebański, J. F. (1984). “Type D solutions of the Einstein
and Born–Infeld nolinear electrodynamics equations,” Nuovo Cimento B84, 65.
Salazar, H., Garcı́a, A., and Plebański, J. F. (1987). “Duality rotations and type D
solutions to Einstein equations with nonlinear electrodynamics sources,” J. Math.
Phys. 28, 2171.
Schouten, J. A. (1954). Ricci Calculus, Springer–Verlag, Berlin.
Schouten, J. A. (1958). “Über die konforme Abbildung n-dimensionaler Mannig-
faltigkeiten mit quadratischer Maßbestimmung auf eine Mannigfaltigkeit euklidis-
cher Maßbestimmung,” Math. Zeit. 11, 58.
Stephani, H. (1990). General Relativity–An Introduction to the Theory of Gravitational
Field, Cambridge University Press, Sec. Ed.
Stephani, H., Kramer, D., MacCallum, M., Honselaers, C. and Herlt, E. (2003). Exact
Solutions to Einstein’s Field Equations, Second Edn. Cambridge University Press.
Straumann, N. (1984). General Relativity and Relativistic Astrophysics, Springer–
Verlag, Berlin.
Tsantilis, E., Puntigam, R. A., and Hehl, F. W. (1996). “A quadratic curvature
Lagrangian of Pawlowski and Ra̧czka: A finger exercise with MathTensor.” In: Rel-
ativity and Scientific Computing, F. W. Hehl, R. A. Puntigam, H. Ruder (Eds.).
Springer, Berlin (arXiv: gr-qc/9601002).
 References 429

Tangherlini, F. R. (1963). “Schwarzschil field in n dimensions and the dimensionality

of the space problem,” Il Nuovo Cimento. 27, 636.
Tonnelat, M. A. (1959). Les principles de la théorie èlectromagnètique et de la relativltè,
Masson et C ie , Editeurs, 120 Boulevard Saint Germain, Paris.
Visser, M. (1992). “Dirty black holes: Thermodynamics and horizon structure,” Phys.
Rev. D46, 2445.
Wald, R. M. (1984). General Relativity, Chicago, University of Chicago Press.
Weinberg, S. (1972). Gravitation and Cosmology: Principles and applications of the
General Theory of Relativity, Wiley, New York.
Witten, E. (1991). “String theory and black holes,” Phys. Rev. D44, 314.
Witten, E. (1998). “Anti de Sitter space and holography,” Adv. Theor. Math. Phys. 2,
253. (arXiv: hep-th/9802150).
Weyl, H. (1918). “Reine Infinitesimalgeometrie,” Math. Zeitschr. 2, 384.
York Jr., J. W. (1971). “Gravitational degrees of freedom and the initial-value
problem,” Phys. Rev. Lett. 26, 1656–1658.
Zarro, C. A. D. (2009). “Buchdahl limit for d–dimensional spherical solutions with a
cosmological constant,” Gen. Rel. Grav. 41, 453.
 Index

algebraic classification, 7 dilaton minimally coupled to Einstein

of the Cotton–York tensor, 7 gravity, 257
Jordan normal forms, 9 action dilaton–gravity, 257
“Petrov” types, 9 dilaton Ψ(r) = k ln(r), 258
Segré notations, 9 field and Cotton tensors, 262
of the energy–momentum tensor, 9 mass, and energy, 261
algebraic Plebański types, 10 general static metric, 263
of the traceless Ricci tensor, 11 Chan–Mann solution, 265
regular F (r)+ function for the metric
g+ , 264
Brown–York energy, mass, and momentum, stationary dilaton Ψ(r) = k ln(r), 266
11 fields and Cotton tensors, 269
cyclic symmetric stationary metric, 11 mass, energy, and momentum, 268
energy density , 13 stationary dilaton via SL(2, R)-trans., 270
hypersurface extrinsic curvature, 12 class of rotating dilaton black holes, 272
momentum tensor, 13 rotating Chan–Mann dilaton black hole,
projection tensor, 12 273
surface momentum density vector, 13 dilaton minimally coupled to
total quasi-local energy, 13 Einstein–Maxwell gravity, 274
total quasi-local mass, 13 action and EM-scalar field equations, 274
total quasi-local momentum, 14 static charged dilaton, 275
fields and Cotton tensors, 277
mass, and energy, 277
cyclic symmetric stationary solutions, 44
stationary charged dilaton generated via
Bañados–Teitelboim–Zanelli, 45
SL(2, R), 279
Brown–York energy, mass, momentum,
fields, and Cotton tensors, 282
45
mass, energy, and momentum, 280
BTZ solution counterpart, 46
particular dilaton, 284
BY characteristics of BTZ counterpart, dilaton non-minimally coupled
47 action for a dilaton, 286
real cut, 47 dilaton field to (2+1) gravity, 286
canonical metric and Einstein equations, Laplace–Beltrami operator, 286
44 Martinez–Zanelli black hole, 287
spacelike Killing vector, 44 algebraic Ricci types, 290
timelike Killing vector, 44 mass, energy, and momentum, 288
Coussaert–Henneaux metric, 47 massless dilaton, 286
symmetries of BTZ solution, 52 dust solutions, 27
Killing equations, 52 Barrow–Shaw–Tsagas anisotropic dust, 30
Killing vectors of BTZ, 54 area expansion, 31
Killingian commutators, 55 dust energy density and shear, 31
time-dependent SO(2) × SO(2) cyclic geometric–mean scale factor, 31
metric, 48 BST diagonal anisotropic dust with Λ, 33
 Index 431

Gödel 3D universes, 34 mass, energy, 170

BST-(t, x, y) cosmology, 34 hybrid stationary Ayon–Cataldo–Garcia
area expansion and shear, 36 (ACG), 191
BST class 1 of solutions, 38 ACG with BTZ limit, 192
BST class 2 of solutions, 37 field and Cotton tensors, 196
BST class 3 of solutions, 39 mass, energy, and momentum, 194
Szekeres solution, 38 magnetostatic; Λ = 0, 159
Cornish–Frankel dust through Heaviside Barrow–Burd–Lancaster, 161
function, 27 Melvin, 161
Giddings–Abott–Kuchař dust, 28 Peldan electrostatic with Λ, 156
Gaussian coordinates, 28 field and Cotton tensors, 158
irrotational dust, 28 mass, energy, 157
isotropic coordinates, 28 Peldan magnetostatic with Λ, 162
static class, 30 field and Cotton tensors, 164
time-dependent class, 29 mass, energy, 163
Rooman–Spindel dust Gödel model, 40 SL(2, R) generated stationary, 215
constancy of the pressure, 40 Clement spinning, 219
incompressible dust Gödel models, 41 Cotton tensor, 223
RS dust, 41 Dias–Lemos magneti BTZ-counterpart,
229
Einstein–Maxwell (EM) fields field and Cotton tensors, 226, 233
static EM equations, 152 Martı́nez–Teitelboim–Zanelli, 224
stationary cyclic symmetric, 143 mass, energy, and momentum, 221, 225,
circularity conditions, 143 232, 237
closed integral curves, 143 stationary hybrid Cataldo, 235
cyclic symmetry, 143 Table 11.11:... electromagnetic solutions,
Hodge star operation, 144 235
Ricci circularity, 143 transformed electrostatic:, 217
stationary symmetry, 143 transformed magnetostatic solutions:,
stationary EM equations, 145 228
stationary metric and Maxwell fields, 145 Table 11.3.1: Electro–Magneto static
electromagnetic tensor, 145 solutions, 168
energy–momentum tensor, 145 uniform field, 173
quadratic invariant, 146, 147 Matyjasek–Zaslavskii electrostatic, 176
Einstein–Maxwell solutions, 142 elementary similarity transformations, 152
constant invariants’, 180
F F = 0, 184 frame of reference
F F = ∓2/l2 , 182 acceleration, 16
field and Cotton tensors, 188 area expansion, 16
Kamata–Koikawa, 185 covariant derivative operator, 15
Kamata–Koikawa, proper, 189 kinematics of the frame, 15
mass, energy, and momentum, 187 perfect fluid, 15
electrostatic; Λ = 0, 153 Barrow–Shaw–Tsagas decomposition,
Deser–Mazur, 155 15, 16
Gott–Simon–Alpern, 155 Bianchi identities, 17
Melvin, 155 fluid conservation laws, 17
Garcia stationary class, 204 projected Riemann tensor, 17
field and Cotton tensors, 210 relative position tensor, 17
Garcia solution with BTZ–limit, 206 rotation scalar, 18
mass, energy, and momentum, 208 shear scalar, 18
Hirschmann–Welch magneto static with Λ, projection tensor, 15
165 shear, 16
mass, energy, 167 3-velocity field, 15
hybrid static Cataldo solution, 168 vorticity, 16
field and Cotton tensors, 171 vorticity scalar, 16
432 Index

Friedmann–Robertson–Walker cosmology; hydrodynamic equilibrium, 80

solutions Buchdahl theorem, 80
barotropic (2 + 1) PF, 111 microscopically stable, 80
barotropic (3 + 1) PF, 110 Buchdahl theorem in d dimensions, 85
comparison between (3+1) and (2+1) Buchdahl’s inequalities, 86
barotropic, 112 Buchdahl’s procedure, 87
FRW dust Einstein Eqs. for a PF, 85
Collas, 111 Lagrange mean value theorem, 86
Cornish–Frankel, 111 main inequality, 89
Gidding–Abbott–Kuchař (GAK), 119 regular center, 86
matching to conical flat space, 119 Schwarzschild, Kasner and Tangherlini
Oppenheimer–Snyder 4D dust, 119 (SKT)-solution, 88
Saslaw, 111 static spherical metric, 85
Mann–Ross dust with Λ, 114 static solution with constant density, 89
asymptotically AdS–FRW dust, 115 matching to SKT, 90
extrinsic curvature for AdS–FRW dust, stellar equilibrium in (2 + 1)D with Λ, 81
117 hydrostatic equilibrium, 83
monotonically decreasing pressure, 83
extrinsic curvature for static BTZ, 117
extrinsic curvature of Kij , 117
matching to static BTZ, 116 integral function used
polytropic (2 + 1) PF, 114 dilogarithm dilog(x), 380
exponential Ei(x), 363
polytropic (3 + 1) PF, 113
logarithmic Li(x), 380
radiation dominated universes:
Cornish–Frankel, 112
Saslaw, 112 main features of 3D gravity, 1
FRW dilaton–matter cosmology; equations black holes in 2 + 1 gravity, 7
asymptotically AdS–BTZ, 7
Einstein Eqs. for (3 + 1)D, 122
field equations of 3D gravity, 2
PF formulation, 123
Einstein equations, 2
standard formulation, 123
postulates, 2
Einstein Eqs. for (2 + 1)D, 123
Riemann spacetime geometry, 2
PF formulation, 124
Riemann tensor through the
standard formulation, 124
energy–momentum tensor, 2
FRW inflaton–dilaton–matter cosmology;
Riemann tensor via Ricci, 2
solutions
fields and matter gravity, 7
barotropic single dilaton, 127 matter locally curves the spacetime, 2
sfFRW barotropic dilaton–matter fields, Newtonian limits, 3
131 discrepancies, 5
(2 + 1) solutions; γ3 = 2Γ3 , 133 Newtonian gravity limit holds, 5
(3 + 1) solutions; γ4 = 2Γ4 , 131 Newtonian theory of gravity, 3
Barrow–Saich solution; γ = 2 Γ, 134 no acceleration in geodesic slow motion,
sfFRW dilaton field alone, 135 5
(2 + 1)D potential V (φf (β) ), 136 the outcome, 5
(3 + 1)D potential V (φf (β) ), 135 Tonnelat’s linearized theory in 4D, 5
Barrow–Burd–Lancaster (BBL) weak 3D gravity, 5
β = 1/2, 137 weak nD gravitational theory, 3
Madsen β = 1/2, 137 no dynamical degrees of freedom, 6
sfFRW scale factor given, 139 no geodesic deviation for dust, 5
(3 + 1) generalization of sBBBL, 141 point particles global effects, 3
second (2 + 1) BBL–sBBL, 139
slow roll spatially flat FRW (sfFRW), 129 nonlinear electrodynamics in (2 + 1)D, 241
(2 + 1) power law, 130 action, 241
(3 + 1) power law, 128 electric (vector) field, 241
Cruz–Martı́nez, 131 electromagnetic Lagrangian, 241
Lucchin–Matarrese, 129 magnetic (scalar) field, 241
 Index 433

Maxwell Lagrangian, 241 radiation p = μ/2, 105

weak energy conditions, 241 RS rotating dust gtt = −1 = −grr , 40
Born–Infeld Lagrangian, 245
Cataldo–Garcı́a solution to EBI, 244 static PF solutions; Λ, 58
analytical extension, 248 conformally flat 3D, 74
black hole, 247 conformally flat 4D, 76
extreme black hole, 248 correspondence of 3D and 4D, 74
horizons, 247 Einstein equations, 67
Lambert W (x) function, 247 general solution, 68
Ricci and Kretschmann invariants, 247 incompressible PF, 66, 68
surface gravity, 248 Oppenheimer–Volkov equation, 67
temperature, 248 state equation
Coulomb-like black hole solution, 252 barotropic law p = γ ρ, 70
nonlinear–electrostatic solution, 242 polytropic law p = Cργ , 71
regular black hole, 249 Table 6.6: Incompressible..., 78
conical singularities, 251 static PF solutions; cosmological, 58
deficit angle, 250 constant density μ0 (incompressible), 59
extreme black hole, 251 Collas, 60
Lambert function, 251 Cornish–Frankel, 61
Ricci and Kretschmann scalars, 250 Giddings–Abott–Kuchař, 60
null geodesic vector field in 3D, 345 Cornish–Frankel polytropic, 62
covariant derivative of, 346 polytropic index, 62
expansion scalar, 346 Oppenheimer–Volkoff pressure equation,
shear tensor of, 346 62
shear scalar, 346 pure radiation p = μ(r)/2, 65
twist tensor of, 346 stiff matter, 64
rotation scalar, 346 string
n-dimensional heterotic string theory, 292
perfect fluid (PF); differentially rotating, 94 conformal transformation, 293
differentially rotating with Λ, 93 dynamical equations, 292, 294
Lubo–Rooman–Spindel Eqs., 102 Einstein frame, 293
plane–fronted null congruence, 333 heterotic string action, 292
Brinkmann pp–wave, 333 string frame, 292
Point particles’ solutions three-form H , 292
Deser–Jakiw–’T Hooft static N , 23 variations, 292
energy and Euler invariant, 23 string in (2 + 1)D, 295
energy–momentum tensor, 24 conformal inverse relations, 295
Staruszkiewicz double, 22 dynamical equations, 300
Staruszkiewicz single, 19 Einstein frame equations, 295
Dirac-delta point source, 21 Horne–Horowitz black string, 296
isotropic coordinates, 22 Horowitz–Welch black string, 296
relationship deficit angle vs mass, 20 static, 301
static cyclic metric, 19 stationary, 297
stationary charged, 301
rigidly rotating PF, 94 string frame equations, 293
barotropic without Λ, 102 string–dilaton Chan–Mann, 300
Garcia rigidly rotating, 96 target space duality transformation, 299
Garcia interior to BTZ, 97
rigidly rotating with J/(2r 2 ), 96 topologically massive gravity (TMG), 303
LRS rigidly rotating, 104 Bianchi type VIII spacetime, 309
barotropic class, 105 NBO “spacelike” triaxilly squashed, 311
constant density, 105 Nutku timelike biaxially squashed, 310
Garcia representation, 105 Nutku–Baekler–Ortiz (NBO) “timelike”
incompressible μ0 , 106 triaxilly squashed, 309
PF star grr = 1, 106 Petrov type D Cotton tensor of, 310
434 Index

spacelike biaxially squashed, 313 H2 in Poincaré coordinates, 322

Type D Cotton tensor of, 312 H2 in polar coordinates, 322
Vuorio limit of, 311 H2 in polynomial representation, 322
Vuorio spacelike limit of, 313 alternative representation of, 321
Bianchi types (BT) 3D spaces, 307 complex extension of, 320
BT III spacetime, 314 hyperbolic space H2 , 321
NBO spacelike with Λ = 0, 316 Table 17.5: Timelike biaxially squashed
NBO timelike with Λ = 0, 315 metrics, 327
Chern–Simons action for, 303, 304 Kundt solutions of
Einstein–TMG equations, 304 multi-exponent–integral, 372
vacuum plus Λ, 304
Table 19.4: Type II and D..., 360
exact solutions based on
Table 19.6: Type III, N, 403
Chow–Pope–Sezgin (CPS), 305
type D, 359
general families of metrics in, 305
type II, 357
covariantly constant type N pp-wave,
305 type III, 360
Kundt dS3 –AdS3 , 305 via variation of parameters (VP), 364
non-covariantly constant type N wave, Petrov type N wave metric of, 333
305 AdS3 non-covariantly constant vector,
spacelike biaxially squashed, 305 334, 336
timelike biaxially squashed, 305 Brinkmann-like Killing, 333
triaxially squashed, 305 covariantly constant null vector
identifying spacelike biaxially squashed pp-wave; Λ = 0, 340
metrics, 327 Table 18.3: pp-wave solution, 344
alternative representation of, 329 Table 18.2: Type N sol. with
Table 17.6: Spacelike biaxially squashed non-covariantly constant vector,
metrics, 332 Λ = 0, 340
identifying timelike biaxially squashed Table 17.4: BT VIII and BT III solutions,
metrics, 317 317