11585_9789811211485_TP.

indd 1 6/3/20 3:00 PM

 Other Related Titles from World Scientific

Advanced Interferometric Gravitational-Wave Detectors

(In 2 Volumes)
Volume I: Essentials of Gravitational-Wave Detectors
Volume II: Advanced LIGO, Advanced Virgo and Beyond
edited by David Reitze, Peter Saulson and Hartmut Grote
ISBN: 978-981-3146-07-5 (Set)
ISBN: 978-981-3146-10-5 (Vol. 1)
ISBN: 978-981-3146-11-2 (Vol. 2)

Topics on Strong Gravity: A Modern View on Theories and Experiments

edited by César Augusto Zen Vasconcellos
ISBN: 978-981-3277-33-5

Modified Gravity: Progresses and Outlook of Theories, Numerical

Techniques and Observational Tests
edited by Baojiu Li and Kazuya Koyama
ISBN: 978-981-3273-99-3

Loop Quantum Gravity for Everyone

by Rodolfo Gambini and Jorge Pullin
ISBN: 978-981-121-195-9

KahFee - 11585 - Interacting [Link] 1 05-03-20 [Link] PM

11585_9789811211485_TP.indd 2 6/3/20 3:00 PM
 Published by
World Scientific Publishing Co. Pte. Ltd.
5 Toh Tuck Link, Singapore 596224
USA office: 27 Warren Street, Suite 401-402, Hackensack, NJ 07601
UK office: 57 Shelton Street, Covent Garden, London WC2H 9HE

British Library Cataloguing-in-Publication Data

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

INTERACTING GRAVITATIONAL, ELECTROMAGNETIC, NEUTRINO AND

OTHER WAVES
In the Context of Einstein’s General Theory of Relativity
Copyright © 2020 by World Scientific Publishing Co. Pte. Ltd.
All rights reserved. This book, or parts thereof, may not be reproduced in any form or by any means,
electronic or mechanical, including photocopying, recording or any information storage and retrieval
system now known or to be invented, without written permission from the publisher.

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

ISBN 978-981-121-148-5 (hardcover)

ISBN 978-981-121-149-2 (ebook for institutions)
ISBN 978-981-121-150-8 (ebook for individuals)

For any available supplementary material, please visit

[Link]

Desk Editor: Ng Kah Fee

Typeset by Stallion Press

Email: enquiries@[Link]

Printed in Singapore

KahFee - 11585 - Interacting [Link] 2 05-03-20 [Link] PM

March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-fm page v

To My Wife and Son

v
 b2530   International Strategic Relations and China’s National Security: World at the Crossroads

This page intentionally left blank

b2530_FM.indd 6 01-Sep-16 [Link] AM

March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-fm page vii

Preface

Einstein’s theory of general relativity is a classical theory of space, time

and gravitation. In competition with other gravitational theories, it is the
one best supported by experiments and observations. This is particularly
true, after the detection of the first gravitational wave by the Laser Inter-
ferometer Gravitational-Wave Observatory in September 14, 2015 (Abbott
et al., 2016), a century-old prediction of Albert Einstein (1916) by using
his by-then new theory of relativity (Einstein, 1915a, 1915b). Soon after,
ten more gravitational waves were detected from binary mergers of not only
two stellar-mass black holes (Abbott et al., 2019) but also two neutron stars
(Abbott et al., 2017), jointly observed by the Virgo Collaboration after
August 2017. The studies of Einstein’s theory in the strong field regime
have been further promoted by the direct detection of the supermassive
black hole, M87, in April 2017 by the Event Horizon Telescope, but with a
final and processed image released only on April 10, 2019 (Akiyama et al.,
2019).
However, over the last several decades, the complexity and the nonlinear
nature of the field equations of the theory have constantly hindered us
from deeply understanding its content and richness. As a result, most of
such studies have been numerical (Baumgarte and Shapiro, 2010), once the
strong field regime is involved.
Nevertheless, the implication of Einstein’s theory of general relativ-
ity (GR) cannot be completely exploited without analytical studies of the
Einstein field equations. Such analytical investigations usually follow two
different paths. One is to find approximate solutions, generally based on
some linear approximations. Such linearized field equations can be well
justified and represent good approximations to the real world, such as the
cosmological linear perturbations (Lyth and Liddle, 2009), and the problem

vii
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-fm page viii

viii Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

of a binary compact system in its inspiral phase (Maggiore, 2008). But, it

is also true that such obtained solutions commonly lose the nonlinear fea-
tures of the field equations, and many interesting properties, such as the
occurrence of spacetime singularities, are due to the nonlinearity of the
field equations. The second path is to find exact solutions, which can rem-
edy the shortage of the former, but again, because of the complexity of
the field equations, exact solutions often have high degree of symmetry,
which would not be present in the real world. Therefore, one must always
take cautions when studying exact solutions and distinguish properties due
to the nonlinearity of the field equations from those due to the assumed
symmetries.
With the above in mind, the search for exact solutions has already
achieved great success in the stationary axisymmetric spacetimes (Stephani
et al., 2009), as a result of the revolutionary new techniques based on the
recent mathematical developments in the theory of nonlinear differential
equations (Gardner et al., 1967). Meanwhile, there have been also successful
attempts to employ these same techniques in the studies of spacetimes with
two commuting space-like Killing vectors, including the ones with plane or
cylindrical symmetry (Griffiths and Podolský, 2009; Bronnikov, Santos and
Wang, 2019).
In this book, we consider only the spacetimes for nonlinearly interact-
ing gravitational plane waves, a subject that has been extensively studied
since the pioneering works of Szekeres (1970, 1972) and Khan and Pen-
rose (1971), and well documented by Griffiths (1991, and republished in
2016). Among the differences between Griffiths’ book and our current one,
two aspects are worth mentioning here. First, in this book, applying the
distribution theory to the spacetimes of colliding plane waves, a technique
first introduced to general relativity by Taub (1980), we develop a complete
description of the Einstein field equations in terms of the Newman–Penrose
formalism (Newman and Penrose, 1962, 1963). Second, in such spacetimes
the frame, which is normally adopted to define the polarization of the inter-
acting gravitational plane waves, is usually rotating along each of the two
colliding wave paths. As a result, defining polarization with respect to this
kind of bases losses its meaning when we compare the polarization of a grav-
itational wave at different locations and times. To overcome this problem,
we first introduce a parallelly-transported frame along the gravitational
wave path, and then project its polarization to this parallelly transported
frame (Wang, 1991b), so that such defined polarization angle has a unique
physical interpretation, and is independent of the choice of coordinates,
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-fm page ix

Preface ix

that is, if such a defined angle changes in one set of coordinates, it will
change in any of others, or vice versa. Once this is made clear, we turn
to the gravitational analog of the electromagnetic Faraday rotation (1846a,
1846b) for colliding gravitational plane waves. Using the Bianchi identities,
we show explicitly that the changes of the polarization of a plane gravita-
tional wave is exactly due to the nonlinear interaction with the oppositely
moving gravitational and/or matter waves (Wang, 1991b). Without such
nonlinear interactions, the polarization of each of the gravitational plane
waves remains constant. This has been further shown by studying exactly
solutions for the spacetimes of colliding plane waves.
Among such solutions are those that describe both colliding purely grav-
itational plane waves and colliding gravitational plane waves coupled with
matter fields. The collision and interaction of such waves are interesting
in several aspects. One of them is that the nonlinearity of general relativ-
ity shows up explicitly in the failure of the principle of superposition. As
a result, colliding plane gravitational waves generically develop spacetime
singularities in the future of the collision. Killing–Cauchy horizons can be
also developed in the interacting region of the two colliding plane waves in
some particular cases (Chandrasekhar and Xanthopoulos, 1986a), but they
are not stable (Yutsever, 1987; Griffiths, 2005), and with small but generic
perturbations they ought to be turned into spacetime singularities (Clarke
and Hayward, 1989).
One may argue that such formations of spacetime singularities are due
to the high symmetry of the spacetime. In particular, with plane symmetry
the incoming waves always have infinitely large amount of energy. There-
fore, in more realistic cases the singularities might be replaced by high
curved regions. However, recently numerical studies of two colliding plane-
fronted massless particle waves in asymptotically flat spacetimes showed
that spacetime singularities are still formed, even the total energy of the
incoming waves is finite, but now they are hidden inside horizons, that is,
now black holes are formed (Pretorius and East, 2018).
The structure of this book is as follows. The introduction for the fun-
damental concepts and material are arranged so that they are contained
in two chapters, Chapters 1 and 2. Specifically, Chapter 1 includes the
introduction of some basic physical quantities, the Newman–Penrose for-
malism, the physical interpretation of the optical scalars, and the Einstein
field equations in terms of distribution-valued tensors. On the other hand,
in Chapter 2, the definition of a single plane gravitational wave is given,
and its polarization angle and amplitude are given explicitly in terms of the
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-fm page x

x Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

Weyl scalars. The spacetime singularities are also studied, and found that
spacetimes of gravitational plane waves are generally always singular when
written in the Baldwin, Jeffery and Rosen coordinates (Baldwin and Jef-
fery, 1926; Rosen, 1937), and are free of singularities only in some particular
cases (Wang et al., 2018). The relevance of such singularities to memory
effects (see, for example, Favata, 2010; Bieri, Garfinkle and Yunes, 2017)
and soft theorems of gravitons (Hawking, Perry and Strominger, 2016, 2017;
Strominger, 2017) is also mentioned.
In Chapter 3, after the Weyl and Ricci scalars, as well as the Bianchi
identities, are written in terms of distributions for the spacetimes of collid-
ing gravitational plane waves, the polarization of a gravitational plane wave
in the interaction region is defined with respect to a parallelly-transported
frame along its wave path, whereby a gravitational analog of the electro-
magnetic Faraday rotation is investigated for various types of collisions. The
nature of the singularities formed in the interaction region of the two collid-
ing plane gravitational wave due to their mutual focus, and some methods
for generating exact solutions are also discussed.
Chapters 4 is devoted to the studies of the collision and subsequent inter-
action of two pure gravitational plane waves, while Chapter 5 is devoted
to the studies of a gravitational plane wave colliding with a matter wave.
The latter can be an impulsive and/or a shock dust shell, an electromag-
netic wave, or a neutrino wave. In these chapters, we present three classes
of exact solutions of the Einstein field equations, which include most of
the known diagonal and non-diagonal solutions found so far for the colli-
sion of gravitational plane waves without or with the presence of matter
fields. The main properties of these solutions and the effects of polarization
of the colliding gravitational plane waves on the formation and nature of
singularities are investigated in detail.
In Chapter 6, we study isometries between the internal spacetimes of
black holes and the interaction regions of two colliding plane waves, and
show that the interiors of all the known black holes have a one-to-one corre-
spondence to the interaction regions of colliding plane waves. These include
the Schwarzschild, Reissner–Nordström, Kerr–Neman–NUT solutions, and
the ones with a cosmological constant. Finally, in Chapter 7, we present
our concluding remarks, and point several directions one can pursue in the
future, including some observations.
This book is based in a large part on the author’s Ph.D. dissertation
(Wang, 1991f), which was written about three decades ago. But, the mate-
rials and topics are still relevant to current studies of gravitational waves,
especially the phenomena due to the nonlinear effects of the Einstein field
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-fm page xi

Preface xi

equations, such as the formation and nature of spacetime singularities

due to the mutual focus of the two interacting gravitational waves, and
the gravitational analog of the electromagnetic Faraday rotation. Another
important topic is the memory effect (Zeldovitch and Polnarev, 1974),
which is not considered in this book, but can be easily extended to such
studies by following the introduction of Section 2.5 of Chapter 2. Memory
effects have been extensively studied recently (Favata, 2010; Bieri, Garfinkle
and Yunes, 2017), as the current generation of gravitational wave detectors
might be able to observe such nonlinear phenomena (Lasky et al., 2016).
Finally, taking this opportunity, I would like to express my gratitude to
my teachers, colleagues and friends, who are so important to me, not only
because of the publication of this book, but also because of the way that
they change and sharp my life. Certainly, first I would like to express my
gratitude to Prof. Dimitri G. Tsoubelis, my Ph.D. supervisor, for leading me
to the field that this small book covers. It is difficult to imagine that without
his constant interest and cooperation the present book would have been
completed. I am greatly indebted to Profs. Basills C. Xanthopoulos and
Charalampos Kolassis for giving me the benefit of consultation and helpful
discussions. My deep thanks go to Prof. Pericles Tsekeris and his family
members for constant support and encouragements. My thanks also go to
Profs. S. Evangelou, P. Kanti, K. Tamvakis, V. Tsikoudi and I. Vergados
for their encouragements and help. The discussions with Dr. Athanasios
Economou on soliton techniques were invaluable, and I appreciated so much
his help and friendship. I also thank Mrs. Kaity Stoghianidis and Dr. John
Rizos for their useful discussions and help. I owe my thanks to Dr. Anna
van Gogh for both her help and friendship.
I have also benefited greatly from numerous discussions with my collab-
orators, Profs. E. Abdalla, J.S. Alcaniz, K. Bamba, M.L. Bedran, G.A.
Benesh, K.A. Bronnikov, Ronggen Cai, Roberto Chan, Gerald Cleaver,
M.F. da Silva, J.C.N. de Araujo, A. De Felice, H.P. de Oliveira, Chikun
Ding, X.-J. Fang, J. Gariel, O. Goldoni, J. Greenwald, A.E. Gumrukcuoglu,
L. Herrera, E.W. Hirschmann, P.R. Holvorcem, M. Ishak, Mubashar Jamil,
J.L. Jing, Klaus Kirsten, D. Langlois, Patricio S. Letelier, J. Lenells, Zhong-
Heng Li, Kai Lin, Hai-Shan Liu, Tan Liu, Guo-Liang Lu, JianXin Lu, Roy
Maartens, Shinji Mukohyama, K. Noui, R. Opher, Supriya Pan, A. Papa-
zoglou, P. Rocha, M. Sami, N.O. Santos, H.J. Schmidt, Qin Sheng, Fu-Wen
Shu, Parampreet Singh, J. Soda, A.F.F. Teixeira, M. Tian, David Wand,
Bin Wang, Zhao-Jun Wang, H. Wei, Zhong-Chao Wu, Jie Yang, W.-Q.
Yang, Shao-Jun Zhang, X. Zhang, Wen Zhao, and Tao Zhu, and my former
and current postdocs and Ph.D. students, Drs. Madhurima Bhattacharjee,
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-fm page xii

xii Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

Jakub Bilski, Carlos Frederico Brandt, Jaime F. da Rocha, Michael Devin,

Antonino Flachi, Jaijai Geng, Yungui Gong, Fei-Hung He, Yongqing Huang,
Bao-Fei Li, Alexandre Yasuda Miguelote, Gil de Oliveira Neto, Jacob Oost,
Hamed Pejhan, Paulo R. Pereira, Filipe de Moraes Paiva, V.H. Satheeshku-
mar, Mohd Shahalam, Manabendra Sharma, Andreas Tziolas, José A.C.
Nogales Vera, Mew-Bing Wan, Xinwen Wang, Qiang Wu, and Messrs Jared
R. Fier, Wen-Cong Gan, Baowen Li, Chao Zhang and Xiang Zhao. I am
also very grateful and thankful to my classmates and lifetime friends, Profs.
Zhaojun Wang and Yannan Yang, for their constant encouragement and
support.
In the preparation of this book, I have been greatly assisted by a num-
ber of colleagues. In particular, I am extremely grateful to Mr. Kah Fee
Ng, Senior Editor from World Scientific Publishing Company, for the most
helpful comments and assistance.
I cannot express adequately my indebtedness to my wife, Yumei, for her
invaluable assistance with the manuscript of this book, as well as for her
time, patience, encouragement, inspiration and unconditional support.
Finally, to those people whose contribution I can no longer quite recall
I offer both my thanks and apology.
Anzhong Wang
China Spring, TX
September, 2019
March 7, 2020 11:29 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-fm page xiii

Contents

Preface vii

List of Figures xvii

1. Fundamentals of Einstein’s Theory of General Relativity 1

1.1 Spacetime Manifolds . . . . . . . . . . . . . . . . . . . . . 2
1.2 Covariant Differentiations, the Riemann Tensor
and Einstein’s Field Equations . . . . . . . . . . . . . . . 3
1.3 Curves, Parallel Transportation and Geodesics . . . . . . 6
1.4 Geodesic Deviations . . . . . . . . . . . . . . . . . . . . . 6
1.5 Decompositions of the Riemann Tensor . . . . . . . . . . 10
1.6 Newman–Penrose Formalism . . . . . . . . . . . . . . . . 12
1.7 Optical Scalars . . . . . . . . . . . . . . . . . . . . . . . . 21
1.8 Matter Fields . . . . . . . . . . . . . . . . . . . . . . . . . 25
1.9 Spacetimes with Distribution-Valued Tensors . . . . . . . 28

2. Plane Gravitational Waves 39

2.1 Null Fields . . . . . . . . . . . . . . . . . . . . . . . . . . 39
2.2 Plane Gravitational Waves . . . . . . . . . . . . . . . . . 41
2.3 Aichelburg–Sexl Plane-Fronted Gravitational Wave . . . 42
2.4 Polarization of Gravitational Plane Waves . . . . . . . . 44
2.5 Singularities of Gravitational Plane Wave Spacetimes and
Memory Effects . . . . . . . . . . . . . . . . . . . . . . . 48

3. Colliding Plane Gravitational Waves 53

3.1 Spacetimes for Colliding Plane Gravitational Waves . . . 53

xiii
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-fm page xiv

xiv Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

3.2 The Basic Differential Equations for Colliding Plane

Gravitational Waves . . . . . . . . . . . . . . . . . . . . . 56
3.3 Different Coordinate Systems . . . . . . . . . . . . . . . . 67
3.3.1 The (t, z)-coordinate system . . . . . . . . . . . . 67
3.3.2 The (η, μ)-coordinate system . . . . . . . . . . . . 68
3.3.3 The (φ, θ)-coordinate system . . . . . . . . . . . . 70
3.4 Gravitational Analog of the Faraday Rotation . . . . . . 70
3.4.1 Vacuum spacetimes . . . . . . . . . . . . . . . . . 76
3.4.2 Spacetimes filled with null dust . . . . . . . . . . 77
3.4.3 Spacetimes filled with a massless scalar field . . . 78
3.4.4 Spacetimes filled with an electromagnetic
field . . . . . . . . . . . . . . . . . . . . . . . . . . 79
3.4.5 Spacetimes filled with a neutrino field . . . . . . . 80
3.5 Singularities in Spacetimes of Colliding Plane
Gravitational Waves . . . . . . . . . . . . . . . . . . . . . 81
3.6 Methods for Generating New Solutions . . . . . . . . . . 83
3.6.1 The Chandrasekhar–Ferrari method . . . . . . . . 85
3.6.2 Inverse scattering method of Belinsky
and Zakharov . . . . . . . . . . . . . . . . . . . . 85

4. Collision of Pure Gravitational Plane Waves 91

4.1 Collisions of Collinearly Polarized Gravitational
Plane Waves . . . . . . . . . . . . . . . . . . . . . . . . . 91
4.2 Collisions of Collinearly and Non-collinearly Polarized
Gravitational Plane Waves . . . . . . . . . . . . . . . . . 102
4.2.1 One-soliton solutions . . . . . . . . . . . . . . . . 103
4.2.2 The nature of singularities formed
after collision . . . . . . . . . . . . . . . . . . . . 106
4.2.3 Specific solutions . . . . . . . . . . . . . . . . . . 108
4.2.4 Polarizations of colliding plane
gravitational waves . . . . . . . . . . . . . . . . . 114
4.3 Collisions of Two Non-collinearly Polarized Gravitational
Plane Waves . . . . . . . . . . . . . . . . . . . . . . . . . 116
4.3.1 Two-soliton solutions . . . . . . . . . . . . . . . . 116
4.3.2 Formation and nature of spacetime
singularities . . . . . . . . . . . . . . . . . . . . . 119
4.3.3 Particular solutions . . . . . . . . . . . . . . . . . 122
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-fm page xv

Contents xv

5. Collisions of Gravitational Waves with Matter Fields 129

5.1 Collision of Impulsive Shells of Null Dust
and Gravitational Plane Waves . . . . . . . . . . . . . . . 130
5.1.1 Colliding impulsive shells with collinearly
polarized gravitational waves . . . . . . . . . . . . 130
5.1.2 Collisions of an impulsive null dust shell and a
non-collinearly polarized gravitational wave . . . 138
5.1.3 Polarizations of colliding plane gravitational
waves when coupled with an impulsive null
dust shell . . . . . . . . . . . . . . . . . . . . . . . 142
5.2 Collisions of Massless Scalar Waves . . . . . . . . . . . . 145
5.3 Collision of Neutrino and Electromagnetic Waves . . . . 154
5.4 Collisions of Two Electromagnetic Plane Waves . . . . . 160
5.5 Other Colliding Plane Waves . . . . . . . . . . . . . . . . 161

6. Isometries to Interiors of Black Holes 163

6.1 The Schwarzschild Black Hole . . . . . . . . . . . . . . . 163
6.2 The Kerr–NUT Black Hole . . . . . . . . . . . . . . . . . 166
6.3 Other Black Holes . . . . . . . . . . . . . . . . . . . . . . 169

7. Concluding Remarks 171

Bibliography 175

Index 197
 b2530   International Strategic Relations and China’s National Security: World at the Crossroads

This page intentionally left blank

b2530_FM.indd 6 01-Sep-16 [Link] AM

March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-fm page xvii

List of Figures

1.1 The hypersurface S is space-like, spanned by the three space-like

vectors, λµ(a) (a = 1, 2, 3). . . . . . . . . . . . . . . . . . . . . . 9
1.2 A null congruence meets So in the circle S. The image of the
circle S on SP is an ellipse. . . . . . . . . . . . . . . . . . . . . 23

2.1 The spacetime for a plane gravitational wave. For a sandwich

wave, the matter can be arranged so that the regions u < 0,
and u > u0 are flat, and the hypersurface u = 0 is the leading
wavefront of the wave. . . . . . . . . . . . . . . . . . . . . . . . 46

3.1 The projection of a spacetime for colliding plane waves onto the
(u, v)-plane. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54

4.1 The soliton structure for the solutions given by Eq. (4.25). . . . 96

5.1 The projection of the colliding null dust spacetime onto the
(u, v)-plane. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150

xvii
 b2530   International Strategic Relations and China’s National Security: World at the Crossroads

This page intentionally left blank

b2530_FM.indd 6 01-Sep-16 [Link] AM

March 7, 2020 9:53 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch01 page 1

Chapter 1

Fundamentals of Einstein’s Theory

of General Relativity

In this chapter, we shall provide some fundamentals that are to be used in

this book. Among these are the definitions of some basic physical quantities
(for example, the Riemann tensor, parallel transport, geodesic deviation,
and the optical scalars), the Newman–Penrose (NP) formalism, and the
Einstein field equations with distribution-valued tensors. For more details
on Einstein’s theory of general relativity, we refer the readers to the text-
books of Wald (1984), D’Inverno (2003), and Carroll (2004). For the NP
formalism, we refer the readers to Newman and Penrose (1962, 1963), Frolov
(1979), Kramer et al. (1980), Griffiths (1991), and Stephani et al. (2009),
while for generalized functions, to Gelfand and Shilov (1964), Challifour
(1972), Vladimirov (1979), Zemanian (1987), and Ding and Ding (2005).
In this book, we shall adopt the following conventions: the metric signa-
ture is (+, −, . . . , −), and the Christoffel symbols, the Riemann, Ricci and
Einstein tensors, and the Ricci scalar are defined, respectively, by
1 ad
Γabc = g (gdc,b + gbd,c − gbc,d ),
2
Ra bcd = Γabd,c − Γabc,d + Γaec Γebd − Γaed Γebc ,
1
Rab = Rc acb , R = g ab Rab , Gab = Rab − Rgab ,
2
where

N −1
∂gab
gab,c ≡ , Rc acb ≡ Rc acb ,
∂xc c=0

1
March 7, 2020 9:53 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch01 page 2

2 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

etc., with N denoting the dimension of the spacetime.1 Then, the Einstein
field equations take the form,
1
Rab − Rgab − Λgab = κTab ,
2
where Λ is the cosmological constant, Tab the energy–momentum tensor,
and κ ≡ 8πG/c4 , with G denoting the Newtonian constant. In addition,
we also define
1 1
X(ab) ≡ (Xab + Xba ), X[ab] ≡ (Xab − Xba ).
2 2

1.1. Spacetime Manifolds

From the mathematical point of view, the fundamental object of Einstein’s
theory of general relativity (GR) is the four-dimensional (4D) spacetime
manifold (Ω, gμν ), where Ω is a connected 4D Hausdoff C ∞ manifold,2 and
gμν is a symmetric Lorentz metric tensor, or simply a metric on Ω. For the
study of differential geometry, we refer the readers to Farnsworth, Fink,
Porter and Thompson (1972) and Westenholz (1981). The points in Ω are
labeled by a general coordinate system (x0 , x1 , x2 , x3 ), often written as xμ
(μ = 0, 1, 2, 3). We use the convention that Greek indices take the values
0, 1, 2, 3, and repeated Greek indices are to be summed over these values
unless specified otherwise.
According to the principle of covariance, all coordinate systems are
equivalent for the description of physical phenomena. Thus, the choice of
coordinate systems is arbitrary. If we go from one coordinate system, say,
xμ , to another, say, xμ , a contravariant vector y μ and a covariant vector
yμ transform, respectively, as
∂xμ ν ∂xν
y μ = y , yμ = yν , (1.1)
∂xν ∂xμ
and a mixed tensor such as y μ νλ as

∂xμ ∂xρ ∂xδ σ

y μνλ = y , (1.2)
∂xσ ∂xν ∂xλ ρδ
etc.

1 Inthis book, we mainly focus on (3 + 1)-dimensional spacetimes, although many for-

mulas developed here can be easily extended to spacetimes with any (higher) dimension.
2 A quantity F is said to be C n if and only if all derivatives of F up to the nth order

exist and are continuous (Hawking and Ellis, 1973).

March 7, 2020 9:53 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch01 page 3

Fundamentals of Einstein’s Theory of General Relativity 3

The contravariant tensor g μν in terms of gμν is defined by

g μν gμλ = δλν , (1.3)
where δνμ is the Kronecker delta, which is unity for μ = ν (no summation
is taken) and zero otherwise. By using g μν and gμν , we can raise and lower
the indices as
y μ ≡ g μν yν , yμ ≡ gμν y ν . (1.4)
We regard tensors derived by such raising and lowering of indices as rep-
resenting the same quantity, since by raising an index and subsequently
lowering it we recover the original tensor.
All the local information about the spacetime is contained in the metric
gμν , which determines the square of the spacetime interval ds between two
infinitesimally separated events or points xμ and xμ + dxμ as
ds2 = gμν dxμ dxν . (1.5)
The contravariant vector dxμ is said to be time-like, space-like or null
according to whether ds2 is positive, negative or zero, respectively. The
spacetime manifold Ω has three space-like and one time-like dimensions.
Since the Einstein field equations contain the second derivatives of the
metric, while the Bianchi identities contain its third derivatives, it is neces-
sary to require gμν to be at least C 3 and xμ = xμ (xμ ) to be at least C 4 , so
that the Einstein field equations are defined everywhere, while the Bianchi
identities are defined at every point of the spacetime manifold. However, as
we shall show in Section 1.9, these conditions can be relaxed in the sense
of distributions (Challifour, 1972; Vladimirov, 1979; Zemanian, 1987; Ding
and Ding, 2005).

1.2. Covariant Differentiations, the Riemann Tensor

and Einstein’s Field Equations
To generalize the ordinary (partial) differentiation to the Riemannian geom-
etry, one introduces an additional structure into the manifold, an affine
connection, ∇, which assigns to each vector field X (≡ X μ ∂μ ) on Ω a dif-
ferential operator, ∇X , which maps an arbitrary vector field Y into a vector
field ∇x Y .
Associated with each metric, we can endow the manifold with a unique
torsion-free connection by requiring
∇g = 0. (1.6)
March 7, 2020 9:53 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch01 page 4

4 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

In a local coordinate basis {∂λ }, Eq. (1.6) can be written in the following
form:

∇∂λ gμν = gμν,λ − gμδ Γδνλ − gδν Γδμλ = 0, (1.7)

where ∂λ ≡ ∂/∂xλ , a comma “,” denotes the partial differentiation with

respect to the indicated argument, and Γλμν (= Γλνμ ) are called the Christof-
fel connection coefficients (symbols), and the connection itself called the
Christoffel connection. From Eq. (1.7) and by using the symmetry of Γλμν ,
we find
1 λδ
Γλμν = g (gμδ,ν + gνδ,μ − gμν,δ ). (1.8)
2
The covariant differentiation for a contravariant or covariant vector is
defined as

Aμ ;ν = Aμ ,ν + Γμνλ Aλ , Aμ;ν = Aμ,ν − Γλμν Aλ , (1.9)

and for a mixed tensor such as Aμ νλ as

Aμ νλ;σ = Aμ νλ,σ + Γμδσ Aδ νλ − Γδνσ Aμ δλ − Γδλσ Aμ νδ , (1.10)

and so on, where a semicolon “;” denotes the covariant differentiation.

Under a coordinate transformation, say, from xμ to xμ , the connection
coefficients, Γλμν , transform as

∂xμ ∂xσ ∂xδ ρ ∂ 2 xσ ∂xμ

Γμ
νλ = ν λ
Γσδ + ν λ . (1.11)
∂x ∂x ∂x
ρ ∂x ∂x ∂xσ
Therefore, Γμαβ is not a tensor. For a covariant vector Aμ , it can be shown
that

Aμ;ν;λ − Aμ;λ;ν = Aδ Rδ μνλ , (1.12)

where Rδ μνλ is the Riemann tensor defined by

Rσμνλ ≡ Γσμλ,ν − Γσμν,λ + Γδμλ Γσδν − Γδμν Γσδλ , (1.13)

and Eq. (1.12) is called the Ricci identity.

The Riemann tensor has the symmetry,

Rσμνλ = −Rμσνλ = −Rσμλν , (1.14a)

Rσμνλ = Rνλσμ , (1.14b)
Rσμνλ + Rσλμν + Rσνλμ = 0, (1.14c)
March 7, 2020 9:53 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch01 page 5

Fundamentals of Einstein’s Theory of General Relativity 5

and satisfies the Bianchi identities

Rσμνλ;ρ + Rσμρν;λ + Rσμλρ;ν = 0. (1.15)

The Ricci tensor Rμλ is defined by

Rμλ ≡ g σν Rσμνλ = Rδ μδλ . (1.16)

From Eqs. (1.14a)–(1.14c) and (1.16), it is easy to show that

Rμλ = Rλμ . (1.17)

The Ricci scalar is defined by

R ≡ g σλ Rσλ = Rλλ . (1.18)

By contracting the Bianchi identities on the pairs of indices μν and σρ, we

find that
 
1
Rμν − gμν R g λν = 0. (1.19)
2 ;λ

The tensor Gμν ≡ Rμν − (1/2)gμν R is called the Einstein tensor.

We are now in a position to write down the Einstein field equations,
which are the fundamental differential equations of GR,
1
Rμν − gμν R − Λgμν = κTμν , (1.20)
2
where κ (≡ 8πG/c4 ) is the Einstein coupling constant, and Λ the cosmo-
logical constant. The tensor Tμν denotes the energy–stress tensor of the
source producing the gravitational field. Without loss of generality, one can
always choose units such that κ = 1.
It must be noted, however, that the form of the Einstein field equations
used by Chandrasekhar (1983) and Pirani (1964) is not consistent with
the requirement that the energy density of matter fields must be positive
(see, for example, Weyl, 1922; Lichnerowicz, 1955; Throne, 1967; Landau
and Lifshitz, 1972; Misner, Thorne and Wheeler, 1973; Wald, 1984; Islam,
1985; D’Inverno, 2003; Carroll, 2004).
The combination of Eqs. (1.19) and (1.20) gives

T μν ;ν = 0, (1.21)

which are the equations for the conservation of energy and momentum
(stress) of the source.
March 7, 2020 9:53 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch01 page 6

6 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

1.3. Curves, Parallel Transportation and Geodesics

A curve in a Riemannian space is defined by points xμ (λ) where xμ are
suitably differentiable functions of the real parameter λ, varying over some
interval of the real line. The curve is time-like, space-like, or null according
to whether its tangent vector (dxμ /dλ) is time-like, space-like or null.
In Euclidean geometry, for an arbitrary vector field X we will say that
X is parallelly transported along the curve if X μ,ν (dxν /dλ) = 0. In a gen-
eral differentiable manifold with a connection, we define analogously that a
vector X is parallelly transported along the curve if its covariant derivative
X μ;ν (dxν /dλ) along this curve is zero, that is, if

dxν   dxν dX μ dxν

X μ;ν = X μ,ν + Γμνδ X δ = + Γμνδ X δ = 0. (1.22)
dλ dλ dλ dλ
A similar definition holds for tensors. Given any curve xμ (λ) with the end
points λ = λ1 and λ = λ2 , the theory of solutions of ordinary differential
equations shows that if Γλμν are suitably differentiable functions of xμ , we
obtain a unique tensor at λ = λ2 by parallelly transporting it from the
point λ = λ1 , along the curve, to the point λ = λ2 (Hawking and Ellis,
1973).
A particular case is the covariant derivative of the tangent vector itself
along the curve xμ (λ). The curve is said to be a geodesic, if the tangent
vector is parallelly transported along this curve, i.e., if

d2 xμ dxν dxδ
2
+ Γμνδ = 0. (1.23)
dλ dλ dλ
When the equation for a geodesic is reduced to the form of Eq. (1.23), we
say that it is affinely parameterized. It should be noted that such defined
affine parameter λ is not unique, and still subjected to the rescaling and
shift of origin: λ̃ = αλ + λ0 , where α and λ0 are constant. Equation (1.23)
also describes the motion of a free particle.

1.4. Geodesic Deviations

A major problem that has to be solved in the study of gravitational radi-
ation is how to identify a gravitational radiation field. The problem arises
because of the principle of the equivalence, which says that the motion
of a test particle in a gravitational field is independent of its mass and
composition. This implies that mechanical phenomena are the same in an
March 7, 2020 9:53 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch01 page 7

Fundamentals of Einstein’s Theory of General Relativity 7

accelerated laboratory as in the earth’s gravitational field, if observations

are confined to a region over which the variation in the earth’s gravita-
tional field is observationally small. Thus, in a local experiment we cannot
distinguish an inertial field from a genuine gravitational one. However, if we
are allowed to carry out non-local experiments, we can distinguish one from
another by observing the variation of the field rather than the field itself.
In GR, this variation is described by the Riemann tensor which specifies
the relative acceleration of neighboring free particles.
Let us consider a one-parameter family of geodesics Γ(w) specified by
the equations

xμ = xμ (λ, w), (1.24)

where we assume xμ to be at least twice continuously differentiable func-

tions of both λ and w. The parameter w varies from one geodesic to another
while λ varies along each of geodesics. For fixed w, we have the geodesic
equations [see Eq. (1.23)],

∂ 2 xμ μ ∂x ∂x
ν δ
= −Γ , xμ = xμ (λ, w). (1.25)
∂λ2 νδ
∂λ ∂λ
We might, in general, identify λ, with the arc length on each of the geodesics.
We prefer, however, to leave λ to be defined just by Eq. (1.25) so that our
following discussions remain also valid for null geodesics.
The family of geodesics gives rise to the vector fields

∂xμ (λ, w)
tμ (λ, w) = , (1.26a)
∂λ
∂xμ (λ, w)
η μ (λ, w) = , (1.26b)
∂w

where tμ (λ, w) is the tangent vector along each geodesic, and η μ (λ, w) is
the vector that describes the deviation of two points on two infinitesimally
near geodesics, which have the same parameter value λ. The vector η μ is
usually called the geodesic deviation vector.
From Eq. (1.26b) we find that the covariant differentiation of η μ along
each geodesic is given by

Dη μ ∂xν ∂η μ ∂xν ∂tμ

≡ η μ;ν = + Γμνδ η δ = + Γμνδ η δ tν . (1.27)
Dλ ∂λ ∂λ ∂λ ∂w
March 7, 2020 9:53 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch01 page 8

8 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

The remarkable fact is that the second differentiation of η μ will bring us

directly to the Riemann tensor. Actually, we have
 
D2 ημ ∂ Dη μ μ ν Dη
δ
= + Γ t
Dλ2 ∂λ Dλ νδ
Dλ
 2 μ 2 ν
∂ ∂ x μ μ δ∂ x
= + Γ t t η
ρ ν δ
+ Γ η
∂w ∂λ2 νδ,ρ νδ
∂λ2
∂ 2 xδ ∂ 2 xδ
+ Γμνδ tν + Γμνδ tν + Γμνδ Γδρσ tν tρ η σ . (1.28)
∂λ∂w ∂λ∂w
By Substituting Eq. (1.25) into Eq. (1.28), we find the well-known geodesic
deviation equations
D2 ημ
= −Rμ νδσ η δ tν tσ = Rμ νδσ tν tδ η σ , (1.29)
Dλ2
where Rμ νλσ is the Riemann tensor given by Eq. (1.13).
To illustrate the physical meaning of the geodesic deviation equations,
let us consider a time-like geodesic, say, C. We introduce an orthogonal
triad of space-like vectors, λμ(a) (a = 1, 2, 3). Throughout this book, we use
the convention that the indices inside parentheses denote tetrad indices,
Roman indices take the values 1, 2, 3, and repeated Roman indices are to
be summed over these values, unless some specific statement to the contrary
is made. These space-like vectors are assumed orthogonal to each other and
to the tangent vector λμ(0) = tμ ,
λμ(α) λν(β) gμν = λμ(α) λμ(β) = ηαβ , (1.30)
where ηαβ denotes the Minkowski metric, given by
⎡ ⎤
1 0 0 0
⎢0 −1 0 0⎥
[ηαβ ] = ⎢
⎣0 0 −1 0 ⎦.
⎥ (1.31)
0 0 0 −1
The tangent vector λμ(0) can be interpreted physically as the four-velocity of
an observer whose worldline is C, and the space-like vectors λμ (a) as rect-
angular coordinate axes used by this observer. For the sake of convenience,
we assume that the orientations of the axes are fixed so that they are not
rotating as determined by local dynamical experiments (see, for example,
Pirani, 1964). This means that the vectors λμ(a) are parallelly transported
along C,
λμ(a);ν tν = 0. (1.32)
March 7, 2020 9:53 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch01 page 9

Fundamentals of Einstein’s Theory of General Relativity 9

Without loss of generality, we also assume that η μ is orthogonal to λμ(0) .

Thus, the tetrad components of the deviation vector η μ are

η (a) = η (a)(σ) λμ(σ) η ν gμν = λ(a)

ν η ,
ν
η (0) = 0, (1.33)

(a)
where λν ≡ η (a)(σ) λμ(σ) gμν , η (μ)(ν) ≡ η μν . The components of η a represent
the spatial coordinates of a particle that moves nearby the observer, who
moves along the geodesic C [sec Fig. 1.1].
(a)
Contracting Eq. (1.29) with λμ and using Eq. (1.32), we find that the
acceleration of the particle relative to the observer is given by

d2 η (a)
= −K (a)(b) η(b) , (1.34)
dτ 2
where

K (a)(b) ≡ Rμνρσ tν tσ λμ(a) λρ(b) , (1.35)

are some of the tetrad components of the Riemann tensor. In writing

Eq. (1.34), we replaced the parameter λ by the proper time τ measured
by the observer using his own clock.
On the other hand, let us consider the same question in the framework of
Newtonian gravitational theory. To be distinguishable, we use t as the time
used by the observer and ζ μ (t) as the coordinate position of the particle
relative to the observer. The gravitational field is described by the Newto-
nian potential φ. If ζ μ (t) is infinitesimal, then the equation of motion for

Fig. 1.1. The hypersurface S is space-like, spanned by the three space-like

vectors, λμ
(a)
(a = 1, 2, 3).
March 7, 2020 9:53 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch01 page 10

10 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

the observer and the particle are given, respectively, by

d2 xa
= −∂ a φ,
dt2
(1.36)
d2 xa d2 ζ a
+ = − {∂ a φ}|x+ζ = −∂ a φ − ζ b ∂ a ∂ b φ,
dt2 dt2
where the derivatives of φ are evaluated at the point xμ . It then follows
that
d2 ζ a
= −K ab ζ b , K ab ≡ ∂ a ∂ b φ. (1.37)
dt2
The condition for the Laplacian potential ∇2 φ = 0 leads to

K aa = 0. (1.38)

The similarity between Eqs. (1.37) and (1.34) is evident. Moreover, we

even have K (a)(a) = 0, whenever the Einstein vacuum field equations are
satisfied.
The above consideration provides additional support for the choice of
the field equation

Rμν = 0, (1.39)

as a description of a free gravitational field.

1.5. Decompositions of the Riemann Tensor

The Riemann tensor Rμνλρ defined by Eq. (1.13) has 20 independent com-
ponents, whereas the Ricci tensor Rμν defined by Eq. (1.16) has only 10.
Physically, it is convenient to decompose the Riemann tensor into three
parts, which are all irreducible representations of the full Lorentz group
(Kramer et al., 1980; Stephani et al., 2009),

Rμνλρ = Cμνλρ + Eμνλρ + Gμνλρ , (1.40)

where
1
Eμνλρ ≡ (gμλ Sνρ + gνρ Sμλ − gνλ Sμρ − gμρ Sνλ ),
2
1
Gμνλρ ≡ (gνρ gμλ − gνλ gμρ ) R, (1.41)
12
1
Sμν ≡ Rμν − gμν R.
4
March 7, 2020 9:53 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch01 page 11

Fundamentals of Einstein’s Theory of General Relativity 11

In Eq. (1.41), the tensor Sμν denotes the traceless part of the Ricci tensor.
The Weyl tensor Cμνλρ is thought of as representing the free gravitational
field (Szekeres, 1965), and has all the symmetries of the Riemann tensor
[see Eqs. (1.14a)–(1.14c)]. In addition, it is also traceless

Cλ μλν = 0. (1.42)

Combining the fact that the Weyl tensor has all the symmetries of the Rie-
mann tensor and Eq. (1.42), we can see that the Weyl tensor has only 10
independent components. At any given point of the spacetime, these com-
ponents are completely independent of the Ricci tensor. Globally, however,
the Weyl and Ricci tensors are not independent, as they are connected by
the Bianchi identities [see Eq. (1.15)]. These identities can be now written
in the form (Kundt and Trumper, 1962; Szekeres, 1966)
1
Cμνσρ; ρ = Rσ[μ;ν] − gσ[μ R;ν] , (1.43)
6
where the square bracket denotes the antisymmetrization,
1
A[μν] ≡ (Aμν − Aνμ ). (1.44)
2
The remarkable analogy between the Bianchi identities of Eq. (1.43) and
the Maxwell equations,

F μν;ν = j μ , (1.45)

suggests that the Bianchi identities represent the interaction between the
free gravitational field and matter. If we define the tensor Jμνσ as
1
Jμνσ ≡ Rσ[μ;ν] − gσ[μ R,ν] , (1.46)
6
we have

Jμνλ; λ = 0, (1.47)

which strongly resembles the equation for the conservation of charges in

electrodynamics

J λ ;λ = 0. (1.48)

Hence, Jμνλ defined by Eq. (1.46) can be considered as representing a mat-

ter current, which consists of the parts of the source that interact with the
free gravitational field. These parts are called gravitationally active, while
March 7, 2020 9:53 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch01 page 12

12 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

the parts of the source that do not contribute to Jμνλ are called gravita-
tionally inert. The propagation of the free gravitational field is independent
of the inert parts of the source.
An equivalent form for the decompositions of Eqs. (1.40) and (1.41) is
given by

1
Rμνλρ = Cμνλρ + [gμλ Rνρ + gνρ Rμλ − gνλ Rμρ − gμρ Rνλ ]
2
1
+ [gμρ gνλ − gμλ gνρ ] R. (1.49)
6
When the Weyl tensor Cμνλδ vanishes, the spacetime is said to be confor-
mally flat.

1.6. Newman–Penrose Formalism

The NP formalism (Newman and Penrose, 1962, 1963) has been used
successfully for both studying the asymptotic properties of a gravita-
tional field (Penrose, 1960, 1963, 1964; Hawking, 1975) and obtaining solu-
tions of the Einstein field equations (see, for example, Kinnersley, 1969;
Talbot, 1969; Lind, 1974; Jagia and Griffiths, 1980; Kramer et al., 1980;
Singh and Grifffiths, 1990; Griffiths and Podolský, 2009; Stephani et al.,
2009). The language of the NP formalism has been generally accepted and
widely used in the literature (see, for example, Pirani, 1964; Davies, 1976a,
1976b; Campbell and Wainwright, 1977; Frolov, 1979; Kramer et al., 1980;
Chandrasekhar, 1983; Stephani et al., 2009). In these books, however, either
the treatment is not consistent, for example, in Chandrasekhar’s book,
Eq. (287) is not consistent with Eqs. (288) and (289), or the definitions of
the Riemann and Ricci tensors (Newman and Penrose, 1962; Frolov, 1979),
or the signature of the metric (Kramer et al., 1980; Stephani et al., 2009) do
not coincide with the ones that we use in this book. Thus, in this section
we provide a complete and consistent form for the NP formalism. For a
more detailed discussion, we refer the readers to the original Newman and
Penrose papers (Newman and Penrose, 1962, 1963), Frolov (1979), Kramer
et al. (1980), Wang (1991f) and Stephani et al. (2009).
The NP formalism is a tetrad formalism but with a special choice of the
tetrad. NP chose the tetrad so that it consists of four null vectors, l, n, m, m,
denoted by

eμ(α) ≡ {lμ , nμ , mμ , mμ }, (1.50)

March 7, 2020 9:53 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch01 page 13

Fundamentals of Einstein’s Theory of General Relativity 13

where lμ and nμ are real, and mμ and mμ are complex conjugates. The
orthogonality properties of these vectors are

lμ lμ = nμ nμ = mμ mμ = mμ mμ = 0,
lμ mμ = lμ mμ = nμ mμ = nμ mμ = 0, (1.51)
l nμ = −m mμ = 1.
μ μ

The tetrad indices can be raised and lowered by η (α)(β) and η(α)(β) , respec-
tively, which are given by
⎡ ⎤
0 +1 0 0
   ⎢+1 0 0 0⎥
η(α)(β) = η (α)(β) = ⎢⎣0
⎥. (1.52)
0 0 −1⎦
0 0 −1 0

It is easy to show that

gμν = η (α)(β) eμ(α) eν(β)

= l μ n ν + l ν n μ − mμ mν − mν mμ , (1.53)
η(α)(β) = g μν
eμ(α) eν(β) = eμ(α) eμ(β) .

The complex spin coefficients γ(α)(β)(γ) are defined by

γ(α)(β)(γ) ≡ e(α)μ;ν eμ(β) eν(γ) , (1.54)

or equivalently

e(α)μ;ν = γ(α)(β)(γ) e(β) (γ)

μ eν . (1.55)

Since [eμ(β) eμ(γ) ] = 0, it can be shown that the spin coefficients satisfy the
;ν
relations

γ(α)(β)(γ) = −γ(β)(α)(γ). (1.56)

Note that the calculations of the spin coefficients does not require the
calculations of the covariant derivatives. In fact, introducing the λ-symbols
via the relations

λ(α)(β)(γ) ≡ γ(α)(β)(γ) − γ(α)(γ)(β)

= [e(α)μ;ν − e(α)ν;μ ]eμ(β) eν(γ)
= [e(α)μ,ν − e(α)ν,μ ]eμ(β) eν(γ) , (1.57)
March 7, 2020 9:53 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch01 page 14

14 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

we find that the spin coefficients are given by

1
γ(α)(β)(γ) = [λ(α)(β)(γ) + λ(β)(γ)(α) − λ(γ)(α)(β) ]. (1.58)
2
As is evident from Eq. (1.57), the calculations of λ(α)(β)(γ) (consequently,
the calculations of γ(α)(β)(γ) ) require only the calculations of partial deriva-
tives. On the other hand, from Eq. (1.57) we find that the λ-symbols have
the properties,

λ(α)(β)(γ) = −λ(α)(γ)(β) . (1.59)

The purpose of the NP formalism is to express the Einstein field equa-

tions in terms of the spin coefficients. To do so, we return to Eq. (1.12). By
projecting the Ricci identity onto the tetrad frame, we have

R(α)(β)(γ)(δ) = Rσμνλ eσ(α) eμ(β) eν(γ) eλ(δ)

= [e(α)μ;ν;λ − e(α)μ;λ;ν ]eμ(β) eν(γ) eλ(δ)

= γ(α)(β)(γ),(δ) − γ(α)(β)(δ),(γ)
(ε) (ε)
+γ(α)(β)(ε) [γ(δ) (γ) − γ(γ) (δ) ]
(ε) (ε)
+γ(α)(ε)(δ) γ(β) (γ) − γ(α)(ε)(γ) γ(β) (δ) , (1.60)

where
(β)
γ(α) (γ) ≡ η (β)(δ) γ(α)(δ)(γ) , γ(α)(β)(γ),(δ) ≡ γ(α)(β)(γ);μ eμ(δ) . (1.61)

Since for any vector we have

A(α),(β) = A(α);ν eν(β) = eν(β) (Aμ eμ(α) );ν

(ε)
= eμ(α) Aμ;ν eν(β) + A(ε) γ(α) (β) , (1.62)

we can define the intrinsic derivative of A(α) in the e(β) -direction as

(ε)
A(α)|(β) ≡ eμ (α)Aμ;ν eν(β) = A(α),(β) − A(ε) γ(α) (β) . (1.63)

(α)
Similarly, we define A |(β) as

(α) (α) (α)

A |(β) ≡ e(α)
μ A ;ν e(β) = A
μ ν
,(β) + A(ε) γ(ε) (β) , (1.64)
March 7, 2020 9:53 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch01 page 15

Fundamentals of Einstein’s Theory of General Relativity 15

and for the more general case we have

(α)
R (β)(γ)(δ)|(ε) ≡ Rσ μνλ;ρ e(α) μ
σ e(β) e(γ) e
ν λ
(δ) e ρ
(ε)

(α) (ζ) (α)

=R (β)(γ)(δ),(ε) +R (β)(γ)(δ) γ(ζ) (ε)
(α) (ζ) (α) (ζ)
−R (ζ)(γ)(δ) γ(β) (ε) −R (β)(ζ)(δ) γ(γ) (ε)
(α) (ζ)
−R (β)(γ)(ζ) γ(δ) (ε) . (1.65)

Therefore, in terms of the intrinsic derivatives, the Bianchi identities (1.15)

take the form
(α) (α) (α)
R (β)(γ)(δ)|(ε) +R (β)(ε)(γ)|(δ) +R (β)(δ)(ε)|(γ) = 0. (1.66)

On the other hand, by projecting Eq. (1.49) onto the tetrad frame, we
find that the relationship among the Riemann, Weyl and Ricci tensors goes
over in the tetrad form without change, and thus is given by

1
R(α)(β)(γ)(δ) = C(α)(β)(γ)(δ) + η(α)(γ) R(β)(δ)
2

+ η(β)(δ) R(α)(γ) − η(β)(γ) R(α)(δ) − η(α)(δ) R(β)(γ)
1 
+ η(α)(δ) η(β)(γ) − η(α)(γ) η(β)(δ) R, (1.67)
6
where

R ≡ η (α)(β) R(α)(β) = 2 R(0)(1) − R(2)(3) . (1.68)

Having written all of the formulas that we need in terms of the tetrad
components, we are now ready to write down the NP equations. However,
before we do so, following NP (Newman and Penrose, 1962), we introduce
the following special notations that considerably simplify the expressions
of the NP equations.
First of all, the spin coefficients are designated by

κ ≡ γ(0)(2)(0) = lμ;ν mμ lν , ν ≡ −γ(1)(3)(1) = −nμ;ν mμ nν ,

ρ ≡ γ(0)(2)(3) = lμ;ν mμ mν , μ ≡ −γ(1)(3)(2) = −nμ;ν mμ mν ,
σ ≡ γ(0)(2)(2) = lμ;ν mμ mν , λ ≡ −γ(1)(3)(3) = −nμ;ν mμ mν ,
τ ≡ γ(0)(2)(1) = lμ;ν mμ nν , π ≡ −γ(1)(3)(0) = −nμ;ν mμ lν ,
March 7, 2020 9:53 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch01 page 16

16 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

1 1
ε≡ [γ(0)(1)(0) − γ(2)(3)(0) ] = [lμ;ν nμ lν − mμ;ν mμ lν ],
2 2
1 1
α≡ [γ(0)(1)(3) − γ(2)(3)(3) ] = [lμ;ν nμ mν − mμ;ν mμ mν ],
2 2
1 1
β≡ [γ(0)(1)(2) − γ(2)(3)(2) ] = [lμ;ν nμ mν − mμ;ν mμ mν ],
2 2
1 1
γ≡ [γ(0)(1)(1) − γ(2)(3)(1) ] = [lμ;ν nμ nν − mμ;ν mμ nν ], (1.69)
2 2
and all other spin coefficients can be obtained from them by using the
symmetry given by Eq. (1.56) and the fact that the complex conjugate of
any quantity can be obtained by replacing the index 2, wherever it appears,
by the index 3, and vice versa.
As mentioned previously, the Weyl tensor has 10 independent compo-
nents at each point of spacetime. In the NP formalism, these components
are specified by five complex “scalars” as follows:
 
Cμνλδ ≡ −4 Ψ2 + Ψ2 [l[μ nν] l[λ nδ] + m[μ mν] m[λ mδ] ]
 
+ 4 Ψ2 − Ψ2 [1[μ nν] m[λ mδ] + m[μ mν] l[λ nδ] ]
− 4{Ψ0n[μ mν] n[λ mδ] + Ψ1 [l[μ nν] n[λ mδ] + n[μ mν] l[λ nδ]
+ n[μ mν] m[λ mδ] + m[μ mν] n[λ mδ] ] − Ψ2 [l[μ mν] n[λ mδ]
+ n[μ mν] l[λ mδ] ] − Ψ3 [l[μ nν] l[λ mδ] + l[μ mν] l[λ nδ]
− l[μ mν] m[λ mδ] − m[μ mν] l[λ mδ] ] + Ψ4 l[μ mν] l[λ mδ]
+ Complex conjugates}, (1.70)

where Ψ0 , . . . , Ψ4 are called the Weyl scalars, and a bar over a letter denotes
the complex conjugate, as mentioned previously.
The various terms in Eq. (1.70) have the following physical interpreta-
tions (Szekeres, 1965). The Ψ0 and Ψ1 terms represent, respectively, the
transverse and longitudinal wave components in the nμ direction, the Ψ2
term a “Coulomb” component, and the Ψ3 and Ψ4 terms the longitudi-
nal and transverse wave components in the lμ direction. By contracting
Eq. (1.70) with appropriate combinations of the null vectors, {l, n, m, m},
we find

Ψ0 ≡ −C(0)(2)(0)(2) = −Cμνλδ lμ mν lλ mδ ,
Ψ1 ≡ −C(0)(1)(0)(2) = −Cμνλδ 1μ nν lλ mδ ,
March 7, 2020 9:53 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch01 page 17

Fundamentals of Einstein’s Theory of General Relativity 17

1
Ψ2 ≡ − [C(0)(1)(0)(1) − C(0)(1)(2)(3) ]
2
1
= − Cμνλδ [lμ nν lλ nδ − lμ nν mλ mδ ],
2
Ψ3 ≡ C(0)(1)(1)(3) = Cμνλδ lμ nν nλ mδ
= −Cμνλδ nμ lν nλ mδ ,
Ψ4 ≡ −C(1)(3)(1)(3) = −Cμνλδ nμ mν nλ mδ . (1.71)

Similarly, the tetrad components of the traceless Ricci tensor, S(α)(β) , can
be written by using the following notation:

1 1 1
Φ00 ≡ S(0)(0) = Sμν lμ lν = Φ00 = R(0)(0) ,
2 2 2
1 1 1
Φ01 ≡ S(0)(2) = Sμν lμ mν = Φ10 = R(0)(2) ,
2 2 2
1 1 1
Φ02 ≡ S(2)(2) = Sμν mμ mν = Φ20 = R(2)(2) ,
2 2 2
1 1
Φ11 ≡ [S + S(2)(3) ] = Sμν (lμ nν + mμ mν ) (1.72)
4 (0)(1) 4
1
= Φ11 = [R(0)(1) + R(2)(3) ],
4
1 1 1
Φ12 ≡ S(1)(2) = Sμν nμ mν = Φ21 = R(1)(2) ,
2 2 2
1 1 1
Φ22 ≡ S(1)(1) = Sμν nμ nν = Φ22 = R(1)(1) ,
2 2 2
and the trace of the Ricci tensor R(α)(β) is defined by

1 1
Λ≡− R = − [R(0)(1) − R(2)(3) ]. (1.73)
24 12
We hope that there will be no confusion between the cosmological constant
Λ used in Eq. (1.20) and the one used in Eq. (1.73), as in this book we shall
mainly consider the cases in which the cosmological constant vanishes.3

3 An interesting mechanism was recently proposed to produce a cosmological constant

by the collision of two plane gravitational and/or electromagnetic waves (Barrabés and
Hogan, 2014, 2015; Halisoy, Mazharimousavi and Gurtug, 2014).
March 7, 2020 9:53 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch01 page 18

18 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

In terms of Φij and Λ, the energy–momentum tensor takes the form,

Tμν = 2{Φ22 lμ lν + Φ00 nμ nν + (Φ11 + 3Λ) (lμ nν + lν nμ )

+ (Φ11 − 3Λ) (mμ mν + mν mμ )
− Φ01 (nμ mν + nν mμ ) − Φ01 (nμ mν + nν mμ )
− Φ12 (lμ mν + lν mμ ) − Φ12 (lμ mν + lν mμ )
+ Φ02 mμ mν + Φ02 mμ mν }. (1.74)

On the other hand, the combination of Eq. (1.67) and Eqs. (1.70)–(1.73)
gives
1
R(0)(1)(0)(1) = C(0)(1)(0)(1) − R(0)(1) + R
6
 
= − Ψ2 + Ψ2 − 2Φ11 + 2Λ,
1
R(0)(1)(0)(2) = C(0)(1)(0)(2) − R(0)(2) = −Ψ1 − Φ01 ,
2
1
R(0)(1)(1)(2) = C(0)(1)(1)(2) + R(1)(2) = Ψ3 + Φ12 ,
2
R(0)(1)(2)(3) = C(0)(1)(2)(3) = Ψ2 − Ψ2 ,
R(0)(2)(0)(2) = C(0)(2)(0)(2) = −Ψ0 ,
1
R(0)(2)(0)(3) = − R(0)(0) = −Φ00 ,
2
1
R(0)(2)(1)(2) = R(2)(2) = Φ02 ,
2 (1.75)
1
R(0)(2)(1)(3) = C(0)(2)(1)(3) − R = Ψ2 + 2Λ,
12
1
R(0)(2)(2)(3) = C(0)(2)(2)(3) − R(0)(2) = Ψ1 − Φ01 ,
2
R(1)(2)(1)(2) = C(1)(2)(1)(2) = −Ψ4 ,
1
R(1)(2)(1)(3) = − R(1)(1) = −Φ22 ,
2
1
R(1)(2)(2)(3) = C(1)(2)(2)(3) − R(1)(2) = Ψ3 − Φ12 ,
2
1
R(2)(3)(2)(3) = C(2)(3)(2)(3) + R(2)(3) + R
6
 
= − Ψ2 + Ψ2 + 2Φ11 + 2Λ.
March 7, 2020 9:53 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch01 page 19

Fundamentals of Einstein’s Theory of General Relativity 19

We also define the operators,

D ≡ lμ ∂μ , Δ ≡ nμ ∂μ , δ ≡ mμ ∂μ , δ ≡ mμ ∂μ . (1.76)
Substituting Eqs. (1.69), (1.75) and (1.76) into Eq. (1.65), we find
 
Dρ − δκ = ρ2 + σσ + ρ(ε + ε) − κ(3α + β − π) − κτ + Φ00 , (1.77a)
Dσ − δκ = σ(ρ + ρ + 3ε − ε) − κ(τ − π + α + 3β) + Ψ0 , (1.77b)
Dτ − Δκ = ρ(τ + π) + σ(τ + π) + τ (ε − ε)
− κ(3γ + γ) + Ψ1 + Φ01 , (1.77c)
Dα − δε = α(ρ + ε − 2ε) + π(ε + ρ) + βσ − βε
− κλ − κγ + Φ10 , (1.77d)
Dβ − δε = σ(α + π) + β(ρ − ε) − κ(μ + γ) − ε(α − π) + Ψ1 , (1.77e)
Dγ − Δε = α(τ + π) + β(τ + π) − γ(ε + ε) − ε(γ + γ)
+ τ π − νκ + Ψ2 + Φ11 − Λ, (1.77f)
Dλ − δπ = π(α + π − β) − λ(3ε − ε) + ρλ + σμ − νκ + Φ20 , (1.77g)
Dμ − δπ = π(π − α + β) − μ(ε + ε) + ρμ
+ σλ − νκ + Ψ2 + 2Λ, (1.77h)
Dν − Δπ = μ(π + τ ) + λ(π + τ ) + π(γ − γ)
− ν(3ε + ε) + Ψ3 + Φ21 , (1.77i)
Δλ − δν = −λ(μ + μ + 3γ − γ) + ν(3α + β + π − τ ) − Ψ4 , (1.77j)
δρ − δσ = ρ(α + β) − σ(3α − β) + τ (ρ − ρ)
+ κ(μ − μ) − Ψ1 + Φ01 , (1.77k)
δα − δβ = γ(ρ − ρ) + ε(μ − μ) + μρ − λσ + αα + ββ
− 2αβ − Ψ2 + Φ11 + Λ, (1.77l)
δλ − δμ = ν(ρ − ρ) + π(μ − μ) + μ(α + β)
+ λ(α − 3β) − Ψ3 + Φ21 , (1.77m)
δν − Δμ = ν(τ − 3β − α) + μ(γ + γ) − νπ + μ2 + λλ + Φ22 , (1.77n)
δγ − Δβ = γ(τ − α − β) − β(γ − γ − μ)
+ μτ − σν − εν + αλ + Φ12 , (1.77o)
March 7, 2020 9:53 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch01 page 20

20 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

δτ − Δσ = τ (τ + β − α) − σ(3γ − γ) + μσ + λρ − κν + Φ02 , (1.77p)

Δρ − δτ = τ (β − α − τ ) + ρ(γ + γ) − μρ
− λσ + νκ − Ψ2 − 2Λ, (1.77q)
Δα − δγ = ν(ρ + ε) − λ(τ + β) + α(γ − μ) + γ(β − τ ) − Ψ3 . (1.77r)

Combining Eqs. (1.65)–(1.69) and Eqs. (1.71)–(1.76), on the other hand,

we find that the Bianchi identities read

δΨ0 − DΨ1 + DΦ01 − δΦ00

= (4α − π)Ψ0 − 2(2ρ + ε)Ψ1 + 3κΨ2 + (π − 2α − 2β)Φ00
+ 2(ε + ρ)Φ01 + 2σΦ10 − 2κΦ11 − κΦ02 , (1.78a)
ΔΨ0 − δΨ1 + DΦ02 − δΦ01
= (4γ − μ)Ψ0 − 2(2τ + β)Ψ1 + 3σΨ2 + (2ε − 2ε + ρ)Φ02
+ 2(π − β)Φ01 + 2σΦ11 − 2κΦ12 − λΦ00 , (1.78b)
 
3 δΨ1 − DΨ2 + 2 (DΦ11 − δΦ10 ) + δΦ01 − ΔΦ00
= 3λΨ0 − 9ρΨ2 + 6(α − π)Ψ1 + 6κΨ3 + (μ − 2μ − 2γ − 2γ)Φ00
+2(α + π + τ )Φ01 + 2(τ − 2α + π)Φ10 + 2(2ρ − ρ)Φ11
+ 2σΦ20 − 2κΦ12 − 2κΦ21 − σΦ02 , (1.78c)
3 (ΔΨ1 − δΨ2 ) + 2 (DΦ12 − δΦ11 ) + δΦ02 − ΔΦ01
= 3νΨ0 − 9τ Ψ2 + 6(γ − μ)Ψ1 + 6σΨ3 + (2α + 2π + τ − 2β)Φ02
+ 2(ρ − ρ − 2ε)Φ12 + 2(μ − μ − γ)Φ01 + 2(τ + 2π)Φ11
+ 2σΦ21 − 2λΦ10 − 2κΦ22 − νΦ00 , (1.78d)
   
3 δΨ2 − DΨ3 + 2 δΦ11 − ΔΦ10 + DΦ21 − δΦ20
= 3κΨ4 − 9πΨ2 + 6(ε − ρ)Ψ3 + 6λΨ1 + (2β + 2τ + π − 2α)Φ20
+ 2(μ − μ − 2γ)Φ10 + 2(ρ − ρ − ε)Φ21 + 2(π + 2τ )Φ11
+ 2λΦ01 − 2σΦ12 − 2νΦ00 − κΦ22 , (1.78e)
 
3 (ΔΨ2 − δΨ3 ) + 2 δΦ12 − ΔΦ11 + DΦ22 − δΦ21
= 3σΨ4 − 9μΨ2 + 6(β − τ )Ψ3 + 6νΨ1 + (ρ − 2ε − 2ε − 2ρ)Φ22
March 7, 2020 9:53 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch01 page 21

Fundamentals of Einstein’s Theory of General Relativity 21

+ 2(π + τ − 2β)Φ12 + 2(β + τ + π)Φ21 + 2(2μ − μ)Φ11

+ 2λΦ02 − 2νΦ01 − 2νΦ10 − λΦ20 , (1.78f)
δΨ3 − DΨ4 + δΦ21 − ΔΦ20
= (4ε − ρ)Ψ4 − 2(2π + α)Ψ3 + 3λΨ2 + (2γ − 2γ + μ)Φ20
+ 2(τ − α)Φ21 + 2λΦ11 − 2νΦ10 − σΦ22 , (1.78g)

ΔΨ3 − δΨ4 − ΔΦ21 + δΦ22

= (4β − τ )Ψ4 − 2(2μ + γ)Ψ3 + 3νΨ2 + (τ − 2α − 2β)Φ22
+ 2(γ + μ)Φ21 + 2λΦ12 − 2νΦ11 − νΦ20 , (1.78h)
DΦ11 − δΦ10 − δΦ01 + ΔΦ00 + 3DΛ
= (2γ + 2γ − μ − μ)Φ00 + (π − 2α − 2τ )Φ01 + (π − 2α − 2τ )Φ10
+ 2(ρ + ρ)Φ11 + σΦ02 + σΦ20 − κΦ12 − κΦ21 , (1.78i)
DΦ12 − δΦ11 − δΦ02 + ΔΦ01 + 3δΛ
= (2β − 2α + π − τ )Φ02 + (2ρ + ρ − 2ε)Φ12 + (2γ − 2μ − μ)Φ01
+ 2(π − τ )Φ11 + νΦ00 + σΦ21 − λΦ10 − κΦ22 , (1.78j)
DΦ22 − δΦ21 − δΦ12 + ΔΦ11 + 3ΔΛ
= (ρ + ρ − 2ε − 2ε)Φ22 + (2π − τ + 2β)Φ12 + (2β − τ + 2π)Φ21
− 2(μ + μ)Φ11 + νΦ01 + νΦ10 − λΦ02 − λΦ20 . (1.78k)

In terms of Φij , Λ and Ψi , the Kretschmann scalar is given by

I ≡ Rαβγδ Rαβγδ
2    
= 8{3(Ψ22 + Ψ2 ) − 4 Ψ1 Ψ3 + Ψ1 Ψ3 + Ψ0 Ψ4 + Ψ0 Ψ4
 
− 4 Φ01 Φ12 + Φ01 Φ12 + 4Φ11 2 + 2Φ02 Φ02
+ 2Φ00 Φ22 + 12Λ2 }. (1.79)

1.7. Optical Scalars

The geometric meaning of the spin coefficients introduced in Section 1.6 is
manifested from the study of the null congruences formed by lμ and nμ ,
March 7, 2020 9:53 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch01 page 22

22 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

respectively. Combining Eqs. (1.55) and (1.69), we find that

lμ;ν = γ(0)(β)(γ) e(β) (γ)
μ eν = (γ + γ)lμ lν + (ε + ε)lμ nν − (α + β)lμ mν

− (β + α)lμ mν − τ mμ lν − κmμ nν + σmμ mν + ρmμ mν

− τ mμ lν − κmμ nν + σmμ mν + ρmμ mν , (1.80)
and
(β) (γ)
nμ;ν = γ(1)(β)(γ) e μ eν = −(γ + γ)nμ lν − (ε + ε)nμ nν + (α + β)nμ mv
+ (β + α)nμ mν + νmμ lν + πmμ nν − λmμ mν − μmμ mν + ν mμ lν
+ π mμ n ν − λ mμ mν − μ mμ mν . (1.81)
Hence, we have
lμ;ν lν = (ε + ε)lμ − κmμ − κmμ ,
(1.82)
nμ;ν nν = −(γ + γ)nμ + νmμ + ν mμ .
Thus, if κ = 0, then lμ defines a null geodesic congruence. If, in addition,
Re(ε) = 0, then lμ is the tangent vector corresponding to an affine param-
eterization. The same holds for the null vector nμ if κ and ε are replaced
by −ν and −γ, respectively. Because of the symmetry between lμ and nμ ,
it is sufficient to consider only the null congruence formed by lμ . And for
the sake of convenience, we assume that this null congruence is an affinely
parameterized geodesic congruence, i.e.
κ = 0 = Re(ε). (1.83)
Let SO and SP be infinitesimal two-dimensional surfaces spanned by m
and m and orthogonal to the null geodesic C at neighboring points O and
P of C, and let S be an infinitesimal circle with center O, lying in SO (see
Fig. 1.2(a)). Suppose that the null geodesic congruence meets SO in the
circle S, then let us observe the image of this null congruence on SP .
We first consider the deviation vector between the null geodesic C and
any other, say, C  , which meets the circle S at O (Fig. 1.2(a)). Without
loss of generality, we assume that η μ lies on SO . Then, from Eqs. (1.27) and
(1.80) we find that the relative velocity of the two null geodesics C and C 
is given by
Dη μ
= η μ;ν lν = lμ;ν η ν
Dλ
= (mν η ν ) [−(α + β)lμ + ρmμ + σmμ ]
+ (mν η ν ) [−(α + β)lμ + ρmμ + σmμ ]. (1.84)
March 7, 2020 9:53 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch01 page 23

Fundamentals of Einstein’s Theory of General Relativity 23

Fig. 1.2. A null congruence meets So in the circle S. The image of the circle S
on SP is an ellipse.

By using m and m, we construct two space-like unit vectors via the relations
mμ + mμ mμ − mμ
E(2)μ = √ , E(3)μ = √ . (1.85)
2 i 2
Then, it is easy to show that

E(2)μ = − sin ϕT μ + cos ϕη μ ,

(1.86)
E(3)μ = cos ϕT μ + sin ϕη μ ,

where ϕ is the angle formed by η μ and E(2)μ [Fig. 1.2(b)], and T μ is the
unit vector tangent to the circle S at O . Thus we have
E(2)μ + iE(3)μ eiϕ
mμ = √ = √ (η μ + iT μ ),
2 2
(1.87)
E(2)μ − iE(3)μ e−iϕ
mμ = √ = √ (η μ − iT μ ).
2 2
March 7, 2020 9:53 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch01 page 24

24 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

Substituting Eq. (1.87) into Eq. (1.84) we find

Dη μ
+ 2 Re[(α + β)mν η ν ]lμ
Dλ
1
= − |η ν ην | {(ρ + ρ)η μ + i(ρ − ρ)T μ
2
+ 2 |σ| [cos 2 (ϕ − ϕ0 ) η μ − sin 2 (ϕ − ϕ0 ) T μ ]}, (1.88)

where

σ ≡ |σ|ei2ϕ0 . (1.89)

Equation (1.88) shows that if ρ and σ vanish then the change of η μ along
C is proportional to lμ . That is, the image of this null congruence on SP is
also a circle, which results from the parallel transport of the circle S from
O to P along C. However, in general, ρ and σ do not vanish. Consequently,
the first term on the right-hand side of Eq. (1.88) describes an expansion
of the circle S along C. The rate of expansion is
ρ+ρ
θ≡− . (1.90)
2
The second term describes a rotation of the circle S with a rate given by
i(ρ − ρ)
ω≡ . (1.91)
2
And the last term on the right-hand side of Eq. (1.88) depends explicitly
on the angle ϕ. It is easy to see that because of this dependence the circle
S goes over into an ellipse, and the minor axis of which forms an angle ϕ0
with respect to E(2)μ [Fig. 1.2(b)].
For the null geodesic congruence formed by nμ , the coefficients −ν, −γ,
−μ and −λ correspond to κ, ε, ρ and σ, respectively. The quantities θ, ω
and σ, as defined above, were first introduced by Sachs (1961, 1962, 1964),
and are called the optical scalars.
Finally, we note that the change of the shape of a null congruence is
due not only to the spacetime curvature, but also to the inertial field. The
latter follows from the fact that even in a flat spacetime the optical scalars
may not vanish because of the choice of the coordinate system. In order to
consider the effects only due to the spacetime curvature, we must consider
the variation of the optical scalars along geodesics (see Section 1.4). That is,
we must consider the acceleration between geodesics, instead of considering
their relative velocity.
March 7, 2020 9:53 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch01 page 25

Fundamentals of Einstein’s Theory of General Relativity 25

1.8. Matter Fields

In this book, in addition to the considerations of exact solutions of the
Einstein vacuum equations, we shall also consider solutions of the Einstein
field equations for the following physically relevant energy–stress tensors.
(α) A massless scalar field: The energy–stress tensor for a massless scalar
field φ takes the form
1
Tμν = φ;μ φ;ν − gμν φ;λ φ;λ , (1.92)
2
where φ satisfies the massless Klein–Gordon equation
φ;μ;ν g μν = 0. (1.93)
(β) A pure radiation field: The energy–stress tensor in this case is given by
Tμν = εκμ κν , κν κν = 0, (1.94)
where ε is non-negative.
Note that the energy–stress tensor for several matter fields has the same
form as Eq. (1.94), for example, a null electromagnetic field, a massless
scalar field, or a neutrino field (Kramer et al., 1980; Stephani et al., 2009).
For the latter cases, however, the corresponding matter field equations must
also be satisfied.
(γ) An electromagnetic field: For an electromagnetic field Fμν , the energy–
stress tensor takes the form
1
Tμν = Fμλ F λ ν − gμν Fρλ F λρ , (1.95)
4
where the antisymmetric tensor Fμν satisfies the Maxwell equations
F[μν;λ] = 0, Fμν;λ g νλ = 0. (1.96)
Introducing the following notations (Newman and Penrose, 1962),
Φ0 ≡ F(0)(2) = Fμν lμ mν , Φ2 ≡ −F(1)(3) = −Fμν nμ mν ,
(1.97a)
1 1
Φ1 ≡ [F(0)(1) − F(2)(3) ] = (Fμν lμ nν − Fμν mμ mν ),
2 2
 
Fμν = 2 −Φ0 n[μ mν] − Φ0 n[μ mν] + Φ2 l[μ mν] + Φ2 l[μ mν]
− 4 Re(Φ1 )l[μ nν] + 4i Im(Φ1 )m[μ mν] , (1.97b)
we find that the Ricci scalars are given by
Φmn = Φm Φn , Λ = 0, (m, n = 0, 1, 2) (1.98)
March 7, 2020 9:53 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch01 page 26

26 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

and that the Maxwell equations read

DΦ1 − δΦ0 = (π − 2α)Φ0 + 2ρΦ1 − κΦ2 , (1.99a)

DΦ2 − δΦ1 = (ρ − 2ε)Φ2 + 2πΦ1 − λΦ0 , (1.99b)
δΦ1 − ΔΦ0 = (μ − 2γ)Φ0 + 2τ Φ1 − σΦ2 , (1.99c)
δΦ2 − ΔΦ1 = (τ − 2β)Φ2 + 2μΦ1 − νΦ0 . (1.99d)

(δ) A massless neutrino field: In general, the case for a massless neutrino
field is much more complicated than the previous ones. This is mainly due
to the fact that a neutrino field is described by a two component spinor φA ,
which satisfies the neutrino Weyl equations
μ
σA Ḃ
φA ;μ = 0, (1.100)
μ
where σA Ḃ
are the complex Pauli spin matrices (Griffiths, 1980, 1991;
Tsoubelis and Wang, 1991), and the spin indices A, B take the values 1, 2.
Then, the energy–stress tensor for a massless neutrino field takes the form

Tμν = i[σμAḂ (φA φḂ ;ν − φḂ φA ;ν ) + σνAḂ (φA φḂ ;μ − φḂ φA ;μ )]. (1.101)

In a spinor basis (OA , LA ), the neutrino spinor φA can be written as

φA = ΦOA + ΨLA , (1.102)

where OA and LA are normalized by the conditions

OA LA = −LA OA = 1. (1.103)

In terms of Φ and Ψ and the spin coefficients, Eq. (1.98) takes the form
(Griffiths, 1976a, 1991; Tsoubelis and Wang, 1991)

DΦ + δΨ = (ρ − ε)Φ + (α − π)Ψ, (1.104a)

δΦ + ΔΨ = (τ − β)Φ + (γ − μ)Ψ. (1.104b)

The Ricci scalars are now given by

Φ00 = i[ΨDΨ − ΨDΨ + κΦΨ − κΨΦ + (ε − ε)ΨΨ],

i
Φ01 = [ΨδΨ − ΨδΨ − ΨDΦ + ΦDΨ − (ρ + ε + ε)ΨΦ + (β − α − π)ΨΨ
2
− κΦΦ + σΦΨ],
March 7, 2020 9:53 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch01 page 27

Fundamentals of Einstein’s Theory of General Relativity 27

Φ02 = −i[ΨδΦ − ΦδΨ + (α + β)ΨΦ + σΦΦ + λΨΨ],

i
Φ11 = [ΦDΦ − ΦDΦ + ΨΔΨ − ΨΔΨ + (ε − ε)ΦΦ + (τ + π)ΨΦ
2
− (τ + π)ΨΦ + (γ − γ)ΨΨ],
i
Φ12 = [ΦδΦ − ΦδΦ − ΨΔΦ + ΦΔΨ + (a − β − τ )ΦΦ − (μ + γ + γ)ΨΦ
2
− νΨΨ + λΦΨ],
Φ22 = i[ΦΔΦ̄ − ΦΔΦ + (γ − γ)ΦΦ + νΦΨ − νΨΦ],
Λ = 0. (1.105)

Equations (1.102) and (1.103) are the basic equations for a neutrino field.

(ε) An isentropic perfect fluid: The energy–stress tensor for a perfect fluid
takes the form

Tμν = (μ + p)uμ uν − pgμν , uμ uν g μν = 1, (1.106)

where uμ is the four-velocity of the fluid, p the pressure, and μ the energy
density (where μ must not be confused with the one used for spin coeffi-
cients). Substituting Eq. (1.106) into Eq. (1.21), we obtain

μ;ν uν + (μ + p)uν;ν = 0,
(1.107)
(μ + p)uμ ;ν uν + (uμ uν − g μν ) p;ν = 0,

which are the conditions imposed on a perfect fluid. In order to completely

describe a perfect fluid, however, Eq. (1.107) has to be supplemented by an
equation of state (Taub, 1956, 1975, 1983). More frequently, the relation
p = p(μ) is prescribed. We call a perfect fluid isentropic if the pressure p is
a function of the energy density μ only. The simplest cases of the isentropic
fluids are those with a “gamma equation of state”

p = (γ − 1)μ, (1.108)

where γ is a constant, and must not be confused with the one used for spin
coefficients.
In general, matter fields need to satisfy some energy conditions. The
energy conditions for a neutrino field are discussed by Griffiths (1980).
These include the weak, dominant and strong energy conditions (Hawking
and Eillis, 1973).
March 7, 2020 9:53 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch01 page 28

28 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

(a) The weak energy condition: This condition says that the energy den-
sity measured by any observer must be non-negative. Mathematically, it is
equivalent to saying that for any time-like vector uμ we must have

T μν uμ uν ≥ 0. (1.109)

Equation (1.109) is also true even for any null vector kμ .

(b) The dominant energy condition: The dominant energy condition is

stronger than the weak energy condition. Besides the requirement of
Eq. (1.109), it also requires that for any observer the local energy flow
vector q μ (≡ T μν uμ ) be non-space-like, i.e.

q μ qμ ≥ 0. (1.110)

(c) The strong energy condition: This is basically stated that the expan-
sion of a time-like geodesic congruence with zero vorticity will monotoni-
cally decrease along a geodesic. Mathematically, this is equivalent to require
Rμν uμ uν ≥ 0. Then, by the Einstein field equations, this implies that
 
1 Λ
Tμν u u ≥
μ ν
T− uν uν , (1.111)
2 κ
where uμ denotes the tangential vector of the given time-like geodesic, and
T ≡ T ν ν . It is normally said that the energy-momentum tensor Tμν satisfies
the strong energy condition, if it obeys Eq. (1.111) for Λ = 0 (Hawking and
Eillis, 1973).

1.9. Spacetimes with Distribution-Valued Tensors

The study of singular surfaces in GR started with the pioneering work of
Lanczos (1922, 1924). Since then, this problem has been widely investi-
gated. In particular, in 1966 Israel first gave a complete analysis of the
case in which there was a discontinuity of the first or second derivatives of
the metric across a non-null (time-like or space-like) hypersurface (Israel,
1966, 1967). It was shown that the singular part of the energy–stress tensor
is related to the jump of the second fundamental form across the singu-
lar hypersurface. If the jump is zero, we obtain the junction conditions
first proposed by Darmois (1927). Later on, O’Brien and Synge (1952)
and Lichnerowicz (1955; see also Papapetrou and Hamoufi, 1968) consid-
ered the same problem and proposed other junction conditions. However,
Bonnor and Vichers (1981) showed that Darmois’ junction conditions are
March 7, 2020 9:53 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch01 page 29

Fundamentals of Einstein’s Theory of General Relativity 29

equivalent to Lichnerowicz’s, and implied by, but not equivalent to, O’Brien
and Synge’s.
The non-null surface case has attracted a lot of interest with various
motivations since 1980s. In particular, the phase transitions that might
have occurred in the early epoch of the universe resulted in the formation
of topological defects, domain walls, cosmic strings, monopoles and tex-
tures, according to various field theories, including a wide variety of grand
unified theories (GUT’s) (Zel’dovish, Kobzarev and Okun, 1976; Kibble,
1976; Vilenkin, 1981). The time-varying gravitational fields associated with
the formation of cosmic strings will create particles that have contributions
to the average energy density of the Universe. These contributions could
be especially important in the early Universe (Zel’dovish, 1980; Vilenkin,
1985; Xanthopoulos, 1986a, 1986b, 1987; Economou and Tsoubelis, 1988a,
1988b; Tsoubelis, 1989a; Letelier and Wang, 1995; Wang and Santos, 1996;
Wang and Nogales, 1997; Nogales and Wang, 1998; Bronnikov, Santos and
Wang, 2019). In particular, they were once believed to provide a mecha-
nism to produce the cosmic microwave background (CMB) and large-scale
structure observed in our Universe (Vilenkin and Shellard, 2000). However,
later observations of CMB ruled out cosmic strings formed in the context
of symmetry breaking in GUT’s as the sources of the cosmological pertur-
bations (Bennett et al., 1996; Spergel et al., 2007), and led to an upper
bound (Ade et al., 2016),
Gμ
 10−7 , (1.112)
c2
where μ denotes the string’s tension.
Nevertheless, the subject has attracted recently lots of attention again
in the framework of string/M-Theory, the so-called cosmic superstrings
(Dvali and Vilenkin, 2004; Copeland, Myers and Polchinski, 2004), which
were formed before inflation took place and stretched to macroscopic length
scales in the inflationary phase. During the subsequent epochs, a compli-
cated network of various string elements forms (Chernoff and Tye, 2018; and
references therein). The main phenomenological consequence of a string
network is the emission of GWs (Ringeval and Suyama, 2017; and refer-
ences therein), generating bursts at cusps, kinks and junctions, as well as a
stochastic gravitational wave background. Low tension strings are natural
in string/M-theory, and can easily satisfy the observational bounds given
above (Chernoff, Flanagan and Wardell, 2018).
On the other hand, domain walls played an important role in the the-
ory of inflation (Linde, 1984; Ipser and Sikivie, 1984; Brandenberger, 1985;
March 7, 2020 9:53 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch01 page 30

30 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

Laguna-Castillo and Matzner, 1986; Berezin, Kuzmin and Tkachev, 1987;

Hill, Schramm and Fry, 1989; Wang, 1991c, 1991d, 1991e, 1992a, 1992b,
1992c, 1992d, 1992e, 1992f, 1993, 1994; Schmidt and Wang, 1993; Letelier
and Wang, 1993a, 1993b, 1993c; Khorrami and Mansouri, 1994; Letelier
and Wang, 1995; Wang and Letelier, 1995a, 1995b; Paiva and Wang, 1995).
Later, to solve the long-standing hierarchy problem, brane world scenarios
were proposed in the late 1990s (Arkani-Hamed, Dimopoulos and Dvali,
1998, 1999; Randall and Sundrum, 1999a, 1999b), and have been inten-
sively studied since then. As a matter of fact, the field has been so exten-
sively investigated that it is very difficult to provide a list of references, so
here we simply refer readers to the review articles of the field (Maartens,
2004; Wands, 2006; Maartens and Koyama, 2010). However, most of these
works have been phenomenological in nature, although there have also been
studies in which the models are built in the framework of string/M-theory
(Horava and Witten, 1995, 1996; Lukas et al., 1999; Goldberger and Wise,
1999; Kachru, Schulz and Trivedi, 2003; Braun and Ovrut, 2006; Gray,
Lukas and Ovrut, 2007; Devin et al., 2009; Wu, Gong and Wang, 2009;
Wang, 2010; Wang and Santos, 2010). Another important application of
these string/M-theory inspired models is to provide a mechanism to produce
the late cosmic acceleration of the Universe without provoking the presence
of dark energy (Roy, 2003; Townsend and Wohlfarth, 2003; Wohlfarth, 2003;
Gong, Wang and Wu, 2008; Wang and Santos, 2008; Wu et al., 2008; Wu,
Gong and Wang, 2009; Devin et al., 2009).
Gravitational collapse and formation of spacetime singularities is
another area in which time-like thin shells have been heavily used to model
collapsing matter sources (Barrabés, Israel and Letelier, 1991; Aposto-
latos and Thorne, 1992; Barrabés et al., 1992; Echeverria 1993; Letelier
and Wang, 1994; Holvorcem, Letelier and Wang, 1995; Wang and Lete-
lier, 1995c; Wang and de Oliveira, 1997; Pereira and Wang, 2000a, 2000b;
Wu et al., 2003; Herrera and Santos, 2005; Nakao et al., 2007; Rocha et al.,
2008a, 2008b; Tziolas and Wang, 2008; Tziolas, Wang and Wu, 2009; Nakao
et al., 2009; Chan et al., 2011; Sharma et al., 2011). As a matter of fact,
this used to be one of the main motivations to study singular surfaces in
the early time of GR.
In the case where the surface of discontinuity is null, the concept of the
second fundamental form becomes invalid, and the methods used for the
non-null surface case cannot be used. Dautcourt (1964) and Penrose (1972)
first studied the null case, and the latter used spinor techniques. The null
March 7, 2020 9:53 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch01 page 31

Fundamentals of Einstein’s Theory of General Relativity 31

case is interesting, because it relates to gravitational shock and impulsive

waves, and to null dust shells consisting of massless particles (Wang and
de Oliveira, 1997; Pereira and Wang, 2002; Barrabés and Hogan, 2003).
Specially, it relates to the collision and interaction of plane gravitational
waves and matter (Griffiths, 1991), which is one of the main subjects of
this book.
In order to give a unified description for the above two cases, Taub
(1980) introduced the formalism of distribution theory (see also Panta-
leo, 1979; Greenwald et al., 2013). Later on, Clarke and Dray (1987) and
Barrabés (1989) generalized the notions of the first and second fundamental
forms to arbitrary surfaces.
In review of the wide range of applications of singular surfaces mentioned
above, in this section we shall provide a general and systematical descrip-
tion, which is applicable for both non-null and null surfaces. Although such
a presentation is restricted only to GR, its generalizations to other theories
of gravity are straightforward.
In the rest of the section, we are mainly concerned with spacetimes
whose curvature tensors contain Dirac delta functions with supports on
submanifolds or even at isolated events (for more general cases, see, for
example, Greenwald et al., 2013). Since the Riemann curvature tensor
includes the second derivatives of the metric and is quadratic in the first
derivatives, if we require the metric to be continuous and the first and
second derivatives of the metric to have finite jumps across the singu-
lar surface, then the Riemann curvature tensor will contain Dirac delta
functions.
To begin with, we assume that we are given a spacetime manifold
(Ω, gμν ) and a surface Σ, described by a function ϕ (xμ ) as

Σ = {xμ : ϕ(xμ ) = 0}. (1.113)

The surface Σ divides Ω into two open regions, Ω+ = {xμ : ϕ (xμ ) > 0} and
Ω− = {xμ : ϕ (xμ ) < 0}. We assume that: (i) the restrictions gμν
±
≡ gμν |Ω±
3 0
are at least C and (ii) gμν is C across Σ.
Note that for the sake of convenience in the following we consider only
the cases in which the spacetime has only one singular surface, but the
extension to the cases with multiple singular surfaces is straightforward.
The normal vector to Σ is defined as
∂ϕ
kμ ≡ = ϕ;μ . (1.114)
∂xμ
March 7, 2020 9:53 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch01 page 32

32 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

We define a step function θ(ϕ) on Ω by4

⎧
⎪
⎪ 1 ϕ > 0,
⎪
⎪
⎨1
θ(ϕ) = ϕ = 0, (1.115)
⎪
⎪ 2
⎪
⎪
⎩
0 ϕ < 0.

Then, we have

∂θ(ϕ)
= δ(ϕ)kμ , (1.116)
∂xμ
where δ(ϕ) denotes the Dirac delta function with support on Σ. Thus, for
a test function f of compact support, we have
  
√
δ(ϕ), f  ≡ −g δ(ϕ)f dV = f dS = − f dS, (1.117)
Ω ∂Ω− ∂Ω+

where dS is the invariant volume element induced on the hypersurface Σ.

The nth derivative δ (n) (ϕ) (n ≥ 1) of δ(ϕ) is defined in a standard way,
 
(n) (n−1) ∂f
δ (ϕ), f  = − δ (ϕ), , (1.118)
∂ϕ

and the following relations are valid (Gelfand and Shilov, 1964):

∂H(ϕ) ∂ϕ ∂δ (n) (ϕ) ∂ϕ (n+1)

= δ(ϕ), = δ (ϕ), (n = 0, 1, 2, . . . ),
∂xλ ∂xλ ∂xλ ∂xλ
ϕδ (n) (ϕ) = −nδ (n−1) (ϕ), ϕn δ (n) (ϕ) = (−1)n n!δ(ϕ), (n = 1, 2, . . . ).
(1.119)

In addition, if F is a function defined in a neighborhood of Σ, we define

the distribution F δ (n) (ϕ) by letting it act on a test function f (Greenwald
et al., 2013) as follows:

Fδ (n) (ϕ), f  = δ (n) (ϕ), f F . (1.120)

4 Note that instead of defining a step function as Eq. (1.115), we can use the Heaviside

step function, H(ϕ), which is unity for the non-negative arguments and otherwise zero.
But all the following results are valid for both of them, if we just simply replace one
by another. In fact, the specific value of θ(ϕ) at ϕ = 0 is irrelevant in the sense of
distributions, and does not affect the following results at all.
March 7, 2020 9:53 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch01 page 33

Fundamentals of Einstein’s Theory of General Relativity 33

Then, we can see that the product F δ(ϕ) is well defined whenever F is
C 0 , and it depends only on the restriction F |Σ of F to Σ. More generally,
the product F δ (n) (ϕ) is well defined, provided that F is C n . Clearly, such
defined F δ (n) (ϕ) depends only on the values of F and its partial derivatives
of the mth order evaluated on Σ, where m ≤ n.
By using the continuity of gμν , we can write gμν as
+ −
gμν = θ(ϕ)gμν + (1 − θ(ϕ))gμν . (1.121)

Then, for a test function f we find that the distribution derivative of gμν is
  + −
 
∂λ gμν , f  = ∂λ θ(ϕ)gμν + (1 − θ(ϕ))gμν ,f
+ −
= {θ(ϕ)gμν,λ + (1 − θ(ϕ))gμν,λ }, f 
+ −
+ {gμν ∂λ θ(ϕ) + gμν ∂λ (1 − θ(ϕ))}, f . (1.122)

Since
 + −
 
gμν ∂λ θ(ϕ) + gμν ∂λ (1 − θ(ϕ)) , f
 +
  −

= ∇λ θ(ϕ), gμν f + ∇λ (1 − θ(ϕ)), gμν f
 
  +    − 
=− ∇λ gμν f dV − ∇λ gμν f dV
Ω+ Ω−
 
 +  −
=− gμν f dSλ − (gμν f )dSλ
∂Ω+ ∂Ω−

 + −

=− gμν − gμν f dSλ = 0, (1.123)
∂Ω+

we have
+
∂λ gμν , f  = {θ(ϕ)gμν,λ + (1 − θ(ϕ))gμν,λ }, f . (1.124)

Equation (1.123) can be simply written as

+ −
gμν,λ = θ(ϕ)gμν,λ + (1 − θ(ϕ))gμν,λ . (1.125)

On the other hand, from the definition of the connection coefficients Γμνλ
we find that

Γμνλ = θ(ϕ)Γ+μ −μ
νλ + (1 − θ(ϕ))Γνλ , (1.126)

where the superscripts “±” always refer to the quantities calculated in Ω± .

If T is a vector field in Ω, and in addition if T and its derivatives have finite
March 7, 2020 9:53 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch01 page 34

34 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

discontinuities across Σ, we define distributions as follows (Taub, 1980):

(T μ ) ≡ θ(ϕ)T +μ + (1 − θ(ϕ))T −μ ,
D
(1.127a)
 μ D
T ;ν ≡ θ(ϕ)T +μ ;ν + (1 − θ(ϕ))T −μ ;ν . (1.127b)

For distribution-valued vector and tensor fields, we define the covariant

differentiation as

[(T μ )D ];ν ≡ [(T μ )D ],ν + (T λ )D Γμνλ , (1.128)

where Γμνλ is given by Eq. (1.126).

If we define the symbol [F ]− to be the discontinuity in the function F
at Σ, i.e.

[F ]− = lim F − lim F, y ∈ Σ, (1.129)

x→y + x→y −

where y ± indicate the limits to be taken in Ω± , respectively, we find

 D −
[(T μ )D ];ν = T μ ;ν + [T μ ] kν δ(ϕ) − [T λ ]− [Γμνλ ]− θ(ϕ)(1 − θ(ϕ)).
(1.130)

It must be noted that, since for any given test function f , we have

θ(ϕ)(1 − θ(ϕ)), f  = [θ(ϕ)(1 − θ(ϕ))f ]dV
Ω

= (1 − θ(ϕ))f dV = 0, (1.131)
Ω+

so we can set

θ(ϕ)(1 − θ(ϕ)) = 0, (1.132)

in the sense of distributions. However, in what follows we prefer to keep

such terms (Taub, 1980).
On the other hand, in the neighborhood of Σ, the metric tensor g ± μν
can be written as

g ± μν = g ± μν (ϕ (xμ ), xμ ). (1.133)

Thus, we have

±  ± 
∂gμν 1 ∂ 2 gμν   
±
gμν = 0
gμν +  ϕ+  ϕ2 + O ϕ3 . (1.134)
∂ϕ ϕ=0 2 ∂ϕ2 
ϕ=0
March 7, 2020 9:53 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch01 page 35

Fundamentals of Einstein’s Theory of General Relativity 35

Then, it is easy to show that

−
[gμν,λ ] = kλ γμν ,
−
(1.135)
[gμν,λδ ] = kλ,δ γμν + kλ γμν,δ + kδ γμν,λ + kλ kδ γ̂μν ,

− −
where γμν ≡ [gμν,ϕ ] and γ̂μν ≡ [gμν,ϕϕ ] . Hence, we have

− 1
[Γμνλ ] = [kν γ μ λ + kλ γ μ ν − k μ γνλ ]. (1.136)
2
From the definition of the Riemann tensor given by Eq. (1.13), on the
other hand, we find

Rσ μνλ = θ(ϕ)R+σ μνλ + (1 − θ(ϕ))R−σ μνλ + δ(ϕ)H σ μνλ

+ θ(ϕ)(1 − θ(ϕ))I σ μνλ , (1.137)

where
− −
H σ μνλ ≡ kν Γσμλ − kλ Γσμν ,
− − (1.138)
− −
I σ μνλ ≡ [Γσλδ ] Γδμν − [Γσνδ ] Γδμλ ,

and R±σ μνλ are the Riemann tensor calculated, respectively, in Ω± . Then,
the Ricci tensor is given by

Rμν = θ(ϕ)R+ μν + (1 − θ(ϕ))R− μν + δ(ϕ)Hμν + θ(ϕ)(1 − θ(ϕ))Iμν ,

(1.139)

where

R± μν ≡ R±σ μσν ,
1 
Hμν ≡ H σ μσν = kμ kλ γ λ ν + kν kλ γ λ μ − k λ kλ γμν − kμ kν γ λ λ ,
2
1
Iμν ≡ I σ μσν = kμ kν γ λδ γλδ + 2kλ kδ γ λ μ γ δ ν − (kμ γ λ ν + kν γ λ μ )kλ γ δ δ
4
  
− 2γμδ γ δ ν − γ δ δ γμν kλ k λ .
(1.140)
The Ricci scalar is given by

R ≡ Rν ν = θ(ϕ)R+ + (1 − θ(ϕ))R− + δ(ϕ)H + θ(ϕ)(1 − θ(ϕ))I, (1.141)

March 7, 2020 9:53 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch01 page 36

36 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

where
R± ≡ R±ν ν ,
H ≡ H ν ν = kμ kν γ μν − k ν kν γ μ μ ,
1  1 μ δ 
I ≡ Iν ν = γμδ γ δ ν − γ δ δ γμν k μ k ν + γμ γ δ − γ δμ γμδ kν k ν .
2 4
(1.142)
Now, we are ready to write down the generalized Einstein field equations,
which take the form,
 
1 1
Rμν − gμν R = GD μν + δ(φ) Hμν − gμν H
2 2
 
1
+ θ(ϕ)(1 − θ(ϕ)) Iμν − gμν I
2
= Tμν
D
+ δ(ϕ)τμν + θ(ϕ)(1 − θ(ϕ))Jμν , (1.143)
where
GD μν ≡ θ(ϕ)G+ μν + (1 − θ(ϕ))G− μν ,
(1.144)
T D μν ≡ θ(ϕ)T + μν + (1 − θ(ϕ))T − μν ,
and τμν and Jμν denote the energy–stress tensors with supports only on
the surface Σ. Equation (1.143) can be written as
G± μν = T ± μν , (1.145a)
1
Hμν − gμν H = τμν , (1.145b)
2
1
Iμν − gμν I = Jμν . (1.145c)
2
From Eq. (1.145b), we find
1    
τμν = [γ δ δ k λ kλ gμν − kμ kν + kμ γ λ ν + kν γ λ μ kλ
2
− k λ kλ γμν − gμν kδ kλ γ δλ ], (1.146)
which is applicable for any kind of thin shells. Contracting it with k ν we
obtain
τμν k ν = 0. (1.147)
In the case of a null shell, Eq. (1.146) reduces to
1
τμν = [(kμ γ λ ν + kν γ λ μ )kλ − γ δ δ kμ kν − gμν kδ kλ γ δλ ]. (1.148)
2
March 7, 2020 9:53 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch01 page 37

Fundamentals of Einstein’s Theory of General Relativity 37

And it is easy to show that in this case τμν is traceless, τν ν = 0.

On the other hand, from Eq. (1.128) it can be shown that the Bianchi
identities take the form
Rσ μνλ;ρ + Rσ μρν;λ + Rσ μλρ;ν = θ(ϕ)(1 − θ(ϕ))Aσ μνλρ , (1.149)
where Aσ μνλρ is a tensor defined on Σ. Thus, the right-hand side of
Eq. (1.149) vanishes everywhere except on Σ, and
Tσ μνλρ , θ(ϕ)(1 − θ(ϕ))Aσ μνλρ  = 0, (1.150)
1
for any C tensor Tσ μνλρ
of compact support in Ω.
Combining Eqs. (1.137) and (1.149) we find
 
1
Rμν − g μν R = θ(ϕ)(1 − θ(ϕ))dμ , (1.151)
2 ;ν

where dμ is a vector defined on Σ. Thus, Eq. (1.149) implies that

   
1
Tμ , Rμν − g μν R = 0, (1.152)
2 ;ν

for any C 1 vector Tμ of compact support. As a consequence of Eqs. (1.143),

(1.147) and (1.151), we obtain
 
1 μνR
R − g
μν
= θ(ϕ)T +μν ;ν + (1 − θ(ϕ))T −μν ;ν
2 ;ν

+ δ(ϕ)(τ μν ;ν + [T μν ]− kν ) + θ(ϕ)(1 − θ(ϕ))J μν ;ν

= θ(ϕ)(1 − θ(ϕ))dμ , (1.153)
or equivalently
T ±μν ;ν = 0, (1.154a)
μν −
τ μν
;ν = −[T ] kν , (1.154b)
J μν ;ν = dμ , (1.154c)
which are the generalized equations for the conservation of energy and stress
of the sources.
 b2530   International Strategic Relations and China’s National Security: World at the Crossroads

This page intentionally left blank

b2530_FM.indd 6 01-Sep-16 [Link] AM

March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch02 page 39

Chapter 2

Plane Gravitational Waves

In this chapter, we investigate the plane gravitational wave, which is dis-

tinguished by both its intuitive physical significance and mathematical
simplicity. We shall show how this class of null fields fits into the more
general frame of pure radiation fields. In particular, in Section 2.1, we
give a general description of null fields, and then in Section 2.2 we restrict
ourselves to plane gravitational waves. In Section 2.3, as an example, we
consider the Aichelburg–Sexl solution (Aichelburg and Sexl, 1971), which
represents the gravitational wave produced by a massless particle. The
solution was obtained by first making a Lorentz boost to the spherically
symmetric Schwarzschild vacuum solution and then taking the massless
limit. In Section 2.4, we study the polarization of a plane gravitational
wave and express the polarization angle explicitly in terms of the Weyl
scalars. Finally, in Section 2.5, we study the singularities of the gravita-
tional plane wave spacetimes, and show explicitly that all such spacetimes
are physically singular at the focusing surface, except for only two particu-
lar cases, in which the distortions of a freely falling observer remains finite
when approaching this focusing point (Wang et al., 2018). Its relevance to
the gravitational memory effects (Favata, 2010; Bieri, Garfinkle and Yunes,
2017) and “soft-graviton” theorems (Hawking, Perry and Strominger, 2016,
2017; Strominger, 2017) is discussed.

2.1. Null Fields

A plane gravitational wave is a null field. To study null fields, we first review
the general properties of a Petrov-type N field (Petrov, 1955), since the null
fields belong to the latter.

39
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch02 page 40

40 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

Theorem 2.1 (Ehlers and Kundt, 1962). In a Petrov-type N

spacetime, there exists an orthogonal tetrad λμ(α) so that the Weyl tensor
takes the form
(1) (2)
Cμνλσ = m(1) (2)
μν mλσ − mμν mλσ , (2.1)
where
(1) (2)
mμν ≡ 2k[μ λ(2)ν] , mμν ≡ 2k[μ λ(3)ν] ,
(2.2)
kμ ≡ λ(0)ν + λ(1)ν , kμ k μ = kμ λμ(2) = kμ λμ(3) = 0.

A null field is a Petrov-type N vacuum field. Thus, according to

Theorem 2.1, we have
(1) (2)
Rμνλσ = Cμνλσ = m(1) (2)
μν mλσ − mμν mλσ , (2.3)
for a null field.
Further properties of a null field can be summarized in the following
theorems.
Theorem 2.2 (Ehlers and Kundt, 1962). The null vector kμ , uniquely
determined (up to a sign) by Eq. (2.3), forms a congruence of shear-free
geodesics,
σ = 0, (2.4)
and satisfies the relations
μ
Rνλσ kμ = 0. (2.5)

A null field is physically interpreted as a pure radiation field in analogy

to the electromagnetic pure radiation fields. This interpretation has been
further justified by the study of asymptotic behaviors of the fields at large
distance from sources (Sachs, 1960–1962).
To see the physical meaning of such a null field, let us consider the
geodesic deviations. Suppose that an observer is moving in such a null field
by following a time-like geodesic with his four-velocity uμ . Without loss of
generality, we assume that the observer moves perpendicular to the two-
dimensional surface spanned by λμ(2) and λμ(3) , i.e. λμ(2) uμ = λμ(3) uμ = 0.
Substituting Eq. (2.3) into Eq. (1.29) and replacing tμ by uμ , we obtain
D2 ημ
= −(kλ uλ )2 [λμ(2) λν(2) − λμ(3) λν(3) ]ην , (2.6)
Dτ 2
where τ is the proper time used by the observer. Recalling the discussions
presented in Sections 1.4 and 1.7, we can see that a circle of relative accel-
erations (with respect to the observer) goes over into an ellipse, with the
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch02 page 41

Plane Gravitational Waves 41

minor axis along λμ(2) . Taking into account the direction of relative accel-
erations, we find that λμ(a) are characterized as eigendirections (Ehlers and
Kundt, 1962). We call λμ(2) the direction of polarization with respect to uμ
(Ehlers and Kundt, 1962). The spatial projection of k μ is orthogonal to
the plane of relative accelerations, and (uμ kμ )2 is the magnitude of rela-
tive accelerations for neighboring test particles. That is, k μ determined by
Eq. (2.3) describes the direction of the propagation of the null field, and
(uμ kμ )2 represents the strength of the null field (measured by the observer
uμ ). Thus, considering Eq. (1.88) we have the following theorem.

Theorem 2.3 (Ehlers and Kundt, 1962). Null fields are character-
ized as purely transverse vacuum fields. There exists a null vector k μ such
that relative accelerations and relative rotations of inertial directions are
orthogonal to the spatial projection of k μ .

2.2. Plane Gravitational Waves

A null field with a non-expanding and non-rotating null geodesic congruence
(ray) is called a plane-fronted gravitational wave, or in short, a pp-wave.
The term “plane-fronted” is characterized by the vanishing of expansion
and rotation of the null geodesics.
According to the definition of a pp-wave and Theorem 2.2, we find that
the expansion, rotation and shear of a pp-wave are zero

θpp = ωpp = σpp = 0. (2.7)

Theorem 2.4 (Ehlers and Kundt, 1962). In a pp-wave spacetime, there

exists a covariant constant vector k̂ μ , which is collinear with the ray vector
k μ defined by, Eq. (2.3),

k μ = k 1/2 k̂ μ . (2.8)

The above theorem implies that in a pp-wave spacetime the null

geodesics formed by k̂ μ (or k μ ) are parallel to each other. Introducing a
function û such that

k̂μ = û,μ , (2.9)

then the metric for a pp-wave can be written in the following form (Ehlers
and Kundt, 1962)

ds2 = 2dûdZ + 2 Re(f )(dû)2 − (dX)2 − (dY )2 , (2.10)

March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch02 page 42

42 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

where {xμ } = {û, Z, X, Y } and f = f (X, Y, û) is a complex function ana-

lytic with respect to X and Y and satisfies the following conditions:
f,ζζ = keiθ , ζ ≡ X + iY, (2.11)
where f,ζ ≡ ∂f /∂ζ, and so on. The functions k and θ are real. Following the
discussions carried out in Section 2.1, we can show that for the metric (2.10)
the function k is the amplitude of the pp-wave, and θ the polarization angle,
relative to the dX-axis. When θ is constant, we say the pp-wave is linearly
polarized.
A pp-wave is called a plane gravitational wave, when k and θ are the
functions of û only. Thus, a plane gravitational wave is a pp-wave with
constant amplitude and polarization in every wavefront. From Eq. (2.11)
we find that for a plane gravitational wave the function f satisfies
f,ζζζ = 0. (2.12)
Then, it can be shown that the function f now takes the form,
Re(f ) = (X 2 − Y 2 )H+ (û) + 2XY H× (û), (2.13)
where the functions H+ (û) and H× (û) determine the type of a plane grav-
itational wave, for example, when H× (û) = 0, the plane gravitational
wave is linearly (or collinearly) polarized. In addition, if H+ (û) = H(û),
where H(û) is the Heaviside step function, the wave is a linearly polar-
ized shock wave, and if H+ (û) = δ(û), it is an impulsive wave (Penrose,
1968). The linearly polarized plane gravitational waves were first studied
by Brinkman (1923) and their physical interpretations were presented by
Robinson (Ehlers and Kundt, 1962). Later on, Rosen (1937) studied the
same problem but by using a different form of the metric. The general
case was studied by Bondi (1957), Bondi, Pirani and Robinson (1959), and
Jordan, Ehlers and Kundt (1960). The global properties for a plane grav-
itational wave spacetime was not studied until Penrose (1965) who first
discussed the focusing effects. Later, Belinsky (1980), Carr and Verdaguer
(1984), Ibañez and Verdaguer (1983, 1986), and Verdaguer (1987) stud-
ied the plane gravitational waves by using soliton technique (Belinsky and
Verdaguer, 2001), developed by Belinsky and Zakharov (1978, 1979). See
also Bondi and Pirani (1989).

2.3. Aichelburg–Sexl Plane-Fronted Gravitational Wave

It is well known that the spherically symmetric Schwarzschild vacuum solu-
tion represents a gravitational field produced by a point-like particle with
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch02 page 43

Plane Gravitational Waves 43

mass m. In the isotropic coordinates, it is given by (D’Inverno, 2003),

(1 − A)2 2
ds2 = dt − (1 + A)4 (dx2 + dy 2 + dz 2 ), (2.14)
(1 + A)2

with A ≡ m/(2r), where r ≡ x2 + y 2 + z 2 . Starting with this form of
metric, Aichelburg and Sexl (1971) were able to obtain the gravitational
field produced by a massless particle, after first making a Lorentz boost
and then taking the massless limit. In doing so, they were able to show
that such an obtained solution takes exactly the same form as that given
by Eq. (2.10) for a plane-fronted gravitational wave.
To show the above claim, let us first consider the Lorentz boost along
the x-direction,

t̄ = γ(t + vx), x̄ = γ(x + vt), ȳ = y, z̄ = z, (2.15)

where γ[≡ 1/(1−v 2 )1/2 ] is the Lorentz factor. Then, the above metric takes
the form,

ds2 = (1 + A)2 (dt̄ 2 − dx̄2 − dȳ 2 − dz̄ 2 )

  2 
2 4 1−A
− γ (1 + A) − (dt̄ − vdx̄)2 , (2.16)
1+A

where
m p(1 − v 2 )
A≡ = , (2.17)
2r 2[(x̄ − v t̄ ) + (1 − v 2 )(ȳ 2 + z̄ 2 )]1/2
2

with m ≡ p(1 − v 2 )1/2 . The above metric is not well defined along the light
cone when we take the limit v → 1. To overcome this problem, Aichelburg
and Sexl introduced the new coordinates t and x via the relations,

x − vt = x̄ − v t̄, y  = ȳ, z  = z̄,

 (2.18)
x + vt = x̄ + v t̄ − 4p ln[ (x̄ − t̄)2 + (1 − v 2 ) − (x̄ − t̄)],

which lead the metric (2.16) to take the form,


2 2 2 2 1
ds2 = dt − dx − dy  − dz  − 4p 
(x − vt ) + ρ2 (1 − v 2 )
  2


1
− (dt − dx )2 , (2.19)
(x − vt )2 + (1 − v 2 )
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch02 page 44

44 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

2 2
where ρ2 ≡ y  + z  . Using the limit,
 
1 1
lim  −
v→1 (x − vt )2 + ρ2 (1 − v 2 ) (x − vt )2 + (1 − v 2 )
= −2ln(ρ)δ(x − t ), (2.20)
we find that
2 2 2 2
lim ds2 = dt − dx − dy  − dz 
v→1
2 2
+ 4pδ(t − x ) ln(y  + z  )(dt − dx )2 , (2.21)
which takes precisely the form of Eq. (2.10) by setting (û, Z, X, Y ) =
(t − x , (t + x )/2, y  , z  ).

2.4. Polarization of Gravitational Plane Waves

By means of a coordinate transformation (Hoenselaers and Ernst, 1990), it
can be shown that the metric given by Eqs. (2.10) and (2.13) is brought to
the form
ds2 = 2e−M dudv − e−U {eV cosh W (dx2 )2 − 2 sinh wdx2 dx3
+ e−V cosh W (dx3 )2 }, (2.22)
where M , U , V and W are functions of either the null coordinate u or v.
When W = 0, Eq. (2.22) reduces to the form for the linearly polarized
plane gravitational wave, studied by Rosen (1937).
The spacetime for a plane gravitational wave is described by the met-
ric (2.22) in terms of the coordinates (u, v, x2 , x3 ). We first consider the
cases in which M , U , V and W are functions of u only. We choose the null
tetrad as (Szekeres, 1972; Griffiths, 1991; Wang, 1991f)
lμ = B{0, 1, 0, 0},
nμ = A{1, 0, 0, 0}, (2.23)
mμ = {0, 0, ζ 2 , ζ 3 },
mμ = {0, 0, ζ 2 , ζ 3 },
where
 
e(U−V )/2 W W
ζ2 ≡ √ cosh + i sinh ,
2 2 2
  (2.24)
3 e(U+V )/2 W W
ζ ≡ √ sinh + i cosh ,
2 2 2
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch02 page 45

Plane Gravitational Waves 45

and

M = ln(AB). (2.25)

From Eq. (1.69), it can be shown that the non-vanishing spin coefficients
are given by
1 1
μ = − AU,u , γ= A[(ln B),u − (1/2)iV,u sinh W ],
2 2
(2.26)
1
λ = A[V,u cosh W + iW,u ],
2
where U,u ≡ ∂U/∂u, and so on. Then, from Eqs. (1.77a)–(1.77r) we find
that the non-vanishing Weyl and Ricci scalars are given, respectively, by

Ψ4 = −Δλ − λ(μ + μ + 3γ − γ), (2.27a)

Φ22 = −Δμ − μ(γ + γ) − λλ. (2.27b)

In the vacuum case, we have the Einstein field equation

Φ22 = 0, (2.28)

where Φ22 is given explicitly in terms of the metric coefficients by Eq. (3.8d)
in Chapter 3.
On the other hand, from Eqs. (1.82) and (2.26) we see that in the
present case the null vector nμ defines a null geodesic congruence, and if B
is chosen to be constant, then the null geodesics are affinely parameterized.
Thus, when M , U , V and W are functions of the null coordinate u only,
the Petrov type N plane gravitational wave represented by Ψ4 propagates
along the null geodesics (see Fig. 2.1).
The Weyl tensor given by Eq. (1.70) now takes the following form:

Cμνλσ = −4{Ψ4l[μ mν] l[λ mσ] + Ψ4 l[μ mν] l[λ mσ] }. (2.29)
(1) (2)
If we define mμν and mμν as (Wang, 1991f)

m(1)
μν = 2l[μ E(2)ν] , m(2)
μν = 2l[μ E(3)ν] , (2.30)

where E(2)μ and E(3)μ are given by Eq. (1.85), we find that Eq. (2.29) can
be written in the following form:
1 (1) (2) (2)
Cμνλσ = − {[m(1)
μν mλσ − mμν mλσ ](Ψ4 + Ψ4 )
2
(2) (1)
+ i[m(1) (2)
μν mλσ + mμν mλσ ](Ψ4 − Ψ4 )}. (2.31)
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch02 page 46

46 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

Fig. 2.1. The spacetime for a plane gravitational wave. For a sandwich wave,
the matter can be arranged so that the regions u < 0, and u > u0 are flat, and
the hypersurface u = 0 is the leading wavefront of the wave.

Making a coordinate transformation in the (E(2)μ , E(3)μ )-plane

 
E(2)μ = cos ϕ4 E(2)μ + sin ϕ4 E(3)μ ,
 
(2.32)
E(3)μ = − sin ϕ4 E(2)μ + cos ϕ4 E(3)μ ,

we find that
(1) (1) (2)
mμν = cos ϕ4 mμν + sin ϕ4 mμν ,
(2) (1) (2)
(2.33)
mμν = − sin ϕ4 mμν + cos ϕ4 mμν ,
and
e+μνλσ = cos 2ϕ4 e+μνλσ + sin 2ϕ4 e×μνλσ ,
(2.34)
e×μνλσ = − sin 2ϕ4 e+μνλσ + cos 2ϕ4 e×μνλσ ,
where
(1) (1) (2) (2)
e+μνλσ ≡ mμν mλσ − mμν mλσ ,
(1) (2) (2) (1)
(2.35)
e×μνλσ ≡ mμν mλσ + mμν mλσ .
 
Thus, in terms of E(2)μ and E(3)μ , Eq. (2.31) reads

Cμνλσ = −{[cos 2ϕ4 Re(Ψ4 ) + sin 2ϕ4 lm(Ψ4 )]e+μνλσ

+ [sin 2ϕ4 Re(Ψ4 ) − cos 2ϕ4 lm(Ψ4 )]exμνλσ }. (2.36)
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch02 page 47

Plane Gravitational Waves 47

If we choose the angle ϕ4 such that

sin 2ϕ4 Re(Ψ4 ) − cos 2ϕ4 Im(Ψ4 ) = 0, (2.37)
or equivalently
Im(Ψ4 )
tan 2ϕ4 = , (2.38)
Re(Ψ4 )
we obtain
Cμνλσ = −(Ψ4 Ψ4 )1/2 e+μνλσ . (2.39)

Defining the null vector ˆ

lμ as
ˆlμ = (Ψ4 Ψ4 )1/4 lμ , (2.40)
Eq. (2.39) becomes
Cμνλσ = −ê+μνλσ . (2.41)
Hence, following Eq. (2.6) we find that for a time-like geodesic congruence
the geodesic deviation is given by
D2 ημ μ μ
= (l̂λ uλ )2 [E(2) ν
E(2) − E(3) ν
E(3) ]ην
Dτ 2
μ μ
= (Ψ4 Ψ4 )1/2 (lλ uλ )2 [E(2) ν
E(2) − E(3) ν
E(3) ]ην . (2.42)
Equation (2.42) shows that the relative accelerations of geodesics are pro-
portional to (Ψ4 Ψ4 )1/2 , which does not relate to any observer. Thus,
(Ψ4 Ψ4 )1/2 represents the absolute amplitude of the relative accelerations
of neighboring test particles. The angle ϕ4 is the polarization angle of the
plane gravitational wave with respect to the basis, (E(2)μ , E(3)μ ).
For the spacetimes described by the metric (2.22), it is easy to show
that
μ μ
E(2);ν lν = 0 = E(3);ν lν . (2.43)
That is, the basis (E(2)μ , E(3)μ ) is parallelly transported along the null
geodesics defined by nμ . Since Ψ4 is a function of u only, we have
 
1 lm Ψ4
ϕ4,v = tan−1 = 0. (2.44)
2 Re Ψ4 ,v
It follows that the polarization angle ϕ4 is constant along the path that the
plane gravitational wave follows, which is consistent with the definition of
a plane gravitational wave given in Section 2.2.
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch02 page 48

48 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

In a similar fashion, it can be shown that the non-vanishing Weyl and

Ricci scalars are Ψ0 and Φ00 , when M , U , V and W are functions of the null
coordinate v only, and that the geodesic deviation for a time-like geodesic
congruence is given by

D2 ημ μ μ
= (Ψ0 Ψ0 )1/2 (nλ uλ )2 [E(2) ν
E(2) − E(3) ν
E(3) ]ην , (2.45)
Dτ 2
but now with the angle ϕ0 being defined as
Im(Ψ0 )
tan 2ϕ0 = − . (2.46)
Re(Ψ0 )
Similar to Eq. (2.43), we have
μ μ
E(2);ν nν = 0 = E(3);ν nν . (2.47)

Thus, in the latter case the metric (2.22) describes a plane gravitational
wave represented by Ψ0 with the polarization angle ϕ0 , which moves along
the null geodesic congruence defined by lμ . The angle ϕ0 is constant along
the path, along which the Ψ0 -wave propagates.
When the angle ϕ0 (ϕ4 ) is constant everywhere, we say the correspond-
ing plane gravitational wave is constantly polarized.

2.5. Singularities of Gravitational Plane Wave Spacetimes

and Memory Effects
The memory effects of GWs have been attracted lots of attention (see, for
example, Favata, 2010; Bieri, Garfinkle and Yunes, 2017; Zhang, Duval,
Gibbons and Horvathy, 2017; and references therein), especially after the
recent observations of several GWs emitted from remote binary systems
of either black holes or neutron stars (Abbott et al., 2019). Such effects
might be possibly detected by LISA (Favata, 2010) or even by current
generation of detectors, such as the advanced LIGO and Virgo (Lasky
et al., 2016). Recently, such investigations gained new momenta due to the
close relations between asymptotically symmetric theorems of soft gravi-
tons and GW memory effects (Hawking, Perry and Strominger, 2016, 2017;
Strominger, 2017).
The characteristic feature of these effects is the permanent displacement
of a test particle after a burst of a GW passes (Braginsky and Grishchuk,
1985; Christodoulou, 1991; Blanchet and Damour, 1992; Thorne, 1992;
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch02 page 49

Plane Gravitational Waves 49

Harte, 2013). In addition, the passage of the GW affects not only the
position of the test particle, but also its velocity. In fact, the change of
the velocity of the particle is also permanent (Souriau, 1973; Braginsky
and Thorne, 1987; Bondi, 1957; Bondi and Pirani, 1989; Grishchuk and
Polnarev, 1989; Zhang, Duval, Gibbons and Horvathy, 2018).
When far from the sources, the emitted GWs can be well approximated
by plane GWs. The spacetimes for plane GWs can be cast in various forms,
depending on the choice of the coordinates and gauge-fixing, as showed in
Sections 2.2 and 2.4. The form of Eq. (2.22) was originally due to Baldwin,
Jeffery and Rosen (BJR) (Baldwin and Jeffery, 1926; Rosen, 1937). Despite
its several attractive features, the system of the BJR coordinates is often
singular within a finite width of a wave, and when studying the asymp-
totic behavior of the spacetime, extensions beyond this singular surface are
needed.
In this section, we point out that there exist actually two kinds of sin-
gularities in plane gravitational wave spacetimes, one represents coordinate
singularities, which can be removed by proper coordinate transformations,
and the other represents really spacetime singularities, and physical quanti-
ties, such as distortions of test particles, become infinitely large when such
singularities are approaching (Wang et al., 2018). Therefore, in the latter
these singularities already represent the boundaries of the spacetimes and
extensions beyond them are not only impossible but also not needed. Since
gravitational memory effects and soft graviton theorems are closely related
to the asymptotical behaviors of plane GW spacetimes, in the latter the
spacetimes cannot be used to study such properties.
In GR, there are powerful Hawking–Penrose theorems (Hawking and
Ellis, 1973) from which one can see that spacetimes with “physically rea-
sonable” conditions are singular. However, the theorems did not tell the
nature of the singularities, and Ellis and Schmidt (1977) classified them
into two different kinds: spacetime curvature singularities and coordinate
singularities. The former is real and cannot be removed by any coordinate
transformations of the kind,

xμ → x = ζ μ (xν ),
μ
μ, ν = 0, 1, 2, 3, (2.48)

while the latter is coordinate dependent, and can be removed by proper

coordinate transformations. One typical example is the coordinate singu-
larity of the Schwarzschild solution in the spherical coordinates at the
Schwarzschild radius r = 2M G/c2 .
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch02 page 50

50 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

Spacetime curvature singularities are further divided into two sub-

classes: scalar curvature singularities and non-scalar curvature singulari-
ties. If any of the 14 independent scalars (Campbell and Wainwright, 1977),
constructed from the four-dimensional Riemann tensor Rμνλ σ
and its deriva-
tives, is singular, then the spacetime is said singular, and the corresponding
singularity is a scalar one. If none of these scalars is singular, spacetimes
can be still singular. In particular, tidal forces and/or distortions (which are
the double integrals of the tidal forces), experienced by an observer, may
become infinitely large (Ori, 2000; Nolan, 2000; Hirschmann, Wang and Wu,
2004; Sharma, Tziolas, Wang and Wu, 2011). This kind of singularities is
usually referred to as non-scalar curvature singularities.
In the spacetimes of plane GWs, all the 14 independent scalars vanish
identically (Stephani et al., 2009), so in such spacetimes the singularities
can be either non-scalar (but real spacetime) singularities or coordinate sin-
gularities. By studying tidal forces and distortions of freely falling observers,
Wang et al. (2018) recently showed that the spacetimes are not singular only
in some particular cases. To show this, for the sake of simplicity, we can
consider only the diagonal case, in which W = 0 in the metric Eq. (2.22).
Then, the only non-vanishing component is Ruu , given by
1 2 2
Ruu = U  − (U  + V  ). (2.49)
2
Thus, in the vacuum spacetimes, we have Ruu = 0, which yields

χ + ω 2 χ = 0, (2.50)

where
1 
χ ≡ e−U/2 , ω≡ V . (2.51)
2
Then, from Eq. (2.50) we can see that, for any given initial value, χ0 > 0,
there always exists a moment, say, u = us at which χ vanishes,

χ(us ) = 0 or U (us ) = +∞, (2.52)

that is, a singularity of the metric (2.22) appears at u = us , which is surely

not a scalar singularity, as mentioned above, all the 14 independent scalars
made of the Riemann tensor now vanish identically. Does this mean that
the singularity must be a coordinate one? The answer is not always affir-
mative. This is because spacetimes can still have non-scalar singularities, as
mentioned above. The non-scalar spacetime singularities can be indicated
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch02 page 51

Plane Gravitational Waves 51

by, for example, the divergence of distortions of a freely falling observer,

which are the twice integrations of the tidal force with respect to the proper
time of the observer (Ori, 2000; Nolan, 2000; Hirschmann, Wang and Wu,
2004; Sharma, Tziolas, Wang and Wu, 2011).
To show this, one can first write χ(u) near the focusing point u = us as

χ(u) ≡ e−U(u)/2 = (u − us )α χ̂(u), (2.53)

where α > 0, and χ̂(us ) = 0. The function χ̂(u) in general takes the form,
∞
χ̂(u) = χn (u − us )n , (2.54)
n=0

with χ0 = 0. Then, one can consider the tetrad, eμ(a) (a = 0, 1, 2, 3), defined
by
1 μ 1 μ
eμ(0) = γ0 δuμ + δ , eμ(1) = γ0 δuμ − δ ,
2γ0 v 2γ0 v (2.55)
U −V U +V
eμ(2) = e 2 δyμ , eμ(3) = e 2 δzμ ,

which satisfies the relations,

eμ(α) eν(β) gμν = ηαβ , eμ(α);ν eν(0) = 0, (2.56)

that is, they are unit orthogonal vectors and parallelly transported along the
time-like geodesics, defined by eμ(0) ≡ dxμ /dλ, where λ denotes the proper
time of the time-like geodesics, so that they form a freely falling frame (Ori,
2000; Nolan, 2000; Hirschmann, Wang and Wu, 2004; Sharma, Tziolas,
Wang and Wu, 2011). Projecting the Riemann tensor onto this frame, one
obtains some non-zero components of the Riemann tensor R(a)(b)(c)(d) . If
one integrates them twice along the time-like geodesics, which gives the
distortions, one finds that such integral always diverges as u → us , except
for the cases (Wang et al., 2018),
1
(i) α = or (ii) α = 1. (2.57)
2
Therefore, all the plane GW spacetimes are singular physically at the focused
point u = us , exceptions are only the ones with α = 1/2 or 1. As a result, all
the plane GW spacetimes cannot be used to study memory effects and soft
graviton theorems, except these two particular cases, as only these space-
times can be possibly extended to null infinity, whereby memory effects and
soft graviton theorems can be studied.
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch02 page 52

52 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

The above results, although very simple, may have profound implica-
tions on the studies of gravitational memory effects (Favata, 2010; Bieri,
Garfinkle and Yunes, 2017) and “soft-graviton” theorems (Hawking, Perry
and Strominger, 2016, 2017; Strominger, 2017), as both of them are con-
cerned with the asymptotic behavior of the spacetimes at infinities. But
when the spacetimes become singular at the finite focusing point, such
infinities do not exist. Therefore, only the non-singular spacetimes are
relevant to them.
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch03 page 53

Chapter 3

Colliding Plane Gravitational Waves

In this chapter, we provide a general description for the spacetimes of col-

liding plane gravitational waves in vacuum or coupled to matter fields.
Specifically, in Section 3.1 the spacetimes for two colliding plane gravita-
tional waves are discussed, and the connection coefficients are explicitly
given in terms of the metric coefficients M, U, V and W , and their deriva-
tives. Then, in Section 3.2 the spin coefficients, Weyl and Ricci scalars,
and Bianchi identities are given in terms of distributions, by assuming that
across the wavefronts of the two incoming gravitational waves, the functions
M, U, V and W are only continuous, that is, C 0 . The different coordinate
systems are discussed in Section 3.3, and following it, in Section 3.4, the
polarization of an interacting gravitational wave is defined with respect to
a parallelly transported frame along the wave path, and then the gravi-
tational analogy of the well-known Faraday rotation in electrodynamics is
studied in detail. In Section 3.5, the nature of spacetime singularities formed
due to the mutual focus of the two colliding plane gravitational waves is
investigated, and finally in Section 3.6 several methods for generating new
solutions of the Einstein field equations are reviewed.

3.1. Spacetimes for Colliding Plane Gravitational Waves

In Sections 2.2 and 2.3, we discussed the spacetimes for plane gravitational
waves. It was shown that when the functions M, U, V and W depend only
on the null coordinate u, the metric (2.22) describes a plane gravitational
wave moving along u = constant hypersurfaces. While when they depend
on the null coordinate v, the metric describes a plane gravitational wave
moving along v = constant hypersurfaces. By properly choosing the two

53
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch03 page 54

54 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

null coordinates, the collision of two such plane gravitational waves in the
Minkowski background can be always considered as head-on collision. Then,
the spacetimes can be arranged as follows.
For u < 0 and v > 0 (see Fig. 3.1), we assume the metric to be that
of a plane wave described by the metric (2.22) with M, U, V and W being
functions of v only. Without loss of generality, we can always choose the
coordinate v so that the hypersurface v = 0 represents the wavefront. For
a sandwich wave, we may further assume that the spacetime is flat (cor-
respondingly, M, U, V and W are constants) in the region v > v0 . In the
region v < 0 and u > 0, on the other hand, the metric is again assumed to
be that of a plane gravitational wave, but with M, U, V and W being func-
tions of u only, and the u = 0 hypersurface is the wavefront. For a sandwich
wave, it is further assumed that the region u > u0 is flat. Assuming that
the collision occurs in a flat background, then in the region u, v < 0 the
spacetime is Minkowski, since in this region the two coming waves have not
arrived, yet. In the region u > 0 and v > 0 where the two plane waves
interact, the metric coefficients M, U, V and W become functions of both
u and v. Hence, a characteristic initial value problem has been set up with
the data posed on a pair of null hypersurfaces u = 0 and v = 0 intersecting
in a space-like two-dimensional surface u = 0 = v. In principle, these initial
conditions uniquely determine the geometry in the wave interacting region
u, v > 0 (Szekeres, 1972; Xanthopoulos, 1986a; Hauser and Ernst, 1989a,
1989b, 1990). For the sake of convenience, the above various regions are
numbered as follows (see Fig. 3.1):

Fig. 3.1. The projection of a spacetime for colliding plane waves onto the (u, v)-
plane.
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch03 page 55

Colliding Plane Gravitational Waves 55

Region I (u, v < 0): This is the region in which the two colliding plane
waves have not arrived, yet, so the spacetime in this region is flat, and the
functions M, U, V and W are constants.
Region II (u < 0, v > 0): This is the region in which a plane grav-
itational wave propagates toward the right-hand side with the v = 0
hypersurface as its wavefront. The functions M, U, V and W depend on v
only.
Region III (u > 0, v < 0): In this region, an opposite moving plane grav-
itational wave is incident, and the u = 0 hypersurface is its wavefront. In
this region M, U, V and W are functions of u only.
Region IV (u, v > 0): In this region two incoming waves interact, and
M, U, V and W are functions of both u and v.

In addition to these four regions, there are also two null hypersurfaces,
Σu and Σv , defined as Σu ≡ {xμ : u = 0} and Σv ≡ {xμ : v = 0},
respectively. At the space-like two-dimensional surface u = 0 = v, the
two plane waves collide. Across the hypersurfaces u = 0 and v = 0, the
Riemann curvature tensor, in general, suffers a shock and or an impulsive
type of discontinuity. For the collision of two sandwich waves, Regions II
and III reduce, respectively, to u < 0 and 0 < v < v0 , and 0 < u < u0 and
v < 0.
Corresponding to the metric (2.22), the non-vanishing connection coef-
ficients Γμαβ defined by Eq. (1.8) are given by

1 M V −U
Γuuu = −M,u , Γu22 = e (e cosh W ),v ,
2
1 1 M −V −U
Γu23 = − eM (e−U sinh W ),v , Γu33 = e (e cosh W ),v ,
2 2
1 M V −U
Γvvv = −M,v , Γv22 = e (e cosh W ),u ,
2
1 1 M V −U
Γv23 = − eM (e−U sinh W ),u , Γv33 = e (e cosh W ),u ,
2 2
1 1
Γ22u = (cosh2 W V,u − U,u ), Γ22v = (cosh2 W V,v − U,v ),
2 2
1 1
Γ33u = − (cosh2 W V,u + U,u ), Γ33v = − (cosh2 W V,v + U,v ),
2 2
1
Γ23u = − e−V (sinh W cosh W V,u + W,u ),
2
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch03 page 56

56 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

1
Γ23v = − e−V (sinh W cosh W V,v + W,v ),
2
1
Γ32u = eV (sinh W cosh W V,u − W,u ),
2
1
Γ32v = eV [sinh W cosh W V,v − W,v ). (3.1)
2
When W = 0, the two incoming gravitational waves have fixed polarization
directions, and are said to be collinearly (linearly) polarized. Otherwise,
they are said to be non-collinearly (nonlinearly) polarized.

3.2. The Basic Differential Equations for Colliding Plane

Gravitational Waves
Assuming that the metric coefficients appearing in Eq. (2.22) are functions
of u and v, we find that, corresponding to the choice of the null tetrad (2.23),
the spin coefficients defined by Eq. (1.69) are given by

1 1 1
ρ= BU,v , μ = − AU,u , ε = − B[2(ln A),v + iV,v sinh W ],
2 2 4
1 1
γ = A[2(ln B),u − iV,u sinh W ], σ = − B(V,v cosh W − iW,v ), (3.2)
4 2
1
λ = A(V,u cosh W + iW,u ),
2
and

κ = ν = τ = π = α = β = 0. (3.3)

The vanishing of the spin coefficients κ and ν implies [see Eq. (1.82)] that
the null vectors nμ and lμ defined by Eq. (2.23) still define null geodesic
congruences even in the interacting region (Region IV), as in the single
plane gravitational wave case. Moreover, if the function A is chosen to be
constant, then the null geodesics defined by lμ are affinely parameterized,
while when the function B is chosen to be constant, the null geodesics
defined by nμ are affinely parameterized.
The plane gravitational waves represented by Ψ0 and Ψ4 propagate
along the null geodesic congruences defined, respectively, by nμ and lμ .
Then, Eqs. (3.2), (1.90) and (1.91) show that after the collision the two null
geodesic congruences are no longer shear-free, although their rotations are
still zero. Therefore, after the collision, the two plane gravitational waves
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch03 page 57

Colliding Plane Gravitational Waves 57

will in general pick up shear. Substituting Eq. (3.3) into Eqs. (1.77a)–
(1.77r), we find

Dρ = (ρ2 + σσ) + ρ(ε + ε) + Φ00 , (3.4a)

Dσ = σ(2ρ + 3ε − ε) + Ψ0 , (3.4b)
Dγ − Δε = −γ(ε + ε) − ε(γ + γ) + Ψ2 + Φ11 − Λ, (3.4c)
Dλ = ρλ + σμ − λ(3ε − ε) + Φ20 , (3.4d)
Dμ = ρμ + σλ − μ(ε + ε) + Ψ2 + 2Λ, (3.4e)
Δλ = −λ(2μ + 3γ − γ) − Ψ4 , (3.4f)
0 = μρ − λσ − Ψ2 + Φ11 + Λ, (3.4g)
−Δμ = μ2 + λλ + μ(γ + γ) + Φ22 , (3.4h)
−Δσ = μσ + λρ − σ(3γ − γ) + Φ02 , (3.4i)
Δρ = −(ρμ + σλ) + ρ(γ + γ) − Ψ2 − 2Λ, (3.4j)

while from Eqs. (1.71)–(1.72) we find that

Ψ1 = Ψ3 = Φ01 = Φ10 = Φ12 = Φ21 = 0. (3.5)

From Eqs. (3.4a)–(3.4j), we obtain

Ψ0 = Dσ − σ(2ρ + 3ε − ε), (3.6a)

1
Ψ2 = [Dγ − Δ(ε + ρ) + (ρ + ε)(γ + γ) + γ(ε + ε) − 2λσ], (3.6b)
3
Ψ4 = −Δλ − λ(2μ + 3γ − γ), (3.6c)
Φ00 = Dρ − (ρ2 + σσ) − ρ(ε + ε), (3.6d)
1
Φ11 = [Dγ − Δε + γ(ε + ε) + ε(γ + γ) + λσ − ρμ], (3.6e)
2
Φ02 = −Δσ + σ(3γ − γ) − μσ − λρ, (3.6f)
2
Φ22 = −Δμ − μ(γ + γ) − μ − λλ, (3.6g)
1
Λ= [Δ(ε − 2ρ) − Dγ − γ(ε + ε)
6
+(2ρ − ε)(γ + γ) − λσ − 3μρ]. (3.6h)
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch03 page 58

58 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

Substituting Eq. (3.2) into Eqs. (3.6a)–(3.6h), we find

1
Ψ0 = − B 2 {V,vv cosh W + (M,v − U,v )V,v cosh W + 2 sinh W V,v W,v
2
− i[W,vv + (M,v − U,v )W,v − sinh W cosh W V,v 2 ]}, (3.7a)

1
Ψ2 = AB M,uv − U,uv + W,u W,v + cosh2 W V,u V,v
6

3
+ i cosh W (V,u W,v − V,v W,u ) , (3.7b)
2
1
Ψ4 = − A2 {V,uu cosh W + (M,u − U,u )V,u cosh W + 2 sinh W V,u W,u
2
+ i[W,uu + (M,u − U,u )W,u − sinh W cosh W V,u2 ]}, (3.7c)

and

1 2
Φ00 = B {2U,vv − U,v 2 + 2U,v M,v − W,v 2 − cosh2 W V,v 2 }, (3.8a)
4
1
Φ11 = AB{2M,uv + U,u U,v − W,u W,v − cosh2 W V,u V,v }, (3.8b)
8
1
Φ02 = AB{2 cosh W V,uv − cosh W (U,u V,v + V,u U,v )
4
+ 2 sinh W (V,u W,v + V,v W,u ) − i[2W,uv − (U,u W,v + W,u U,v )
− 2 sinh W cosh W V,u V,v ]}, (3.8c)
1 2
Φ22 = A {2U,uu − U,u 2 + 2U,u M,u − W,u 2 − cosh2 W V,u 2 }, (3.8d)
4
1
Λ = − AB{2M,uv + 4U,uv − 3U,u U,v − W,u W,v
24
− cosh2 W V,u V,v }. (3.8e)

On the other hand, from Eqs. (1.78a)–(1.78k), (3.3) and (3.5) we find
that the Bianchi identities now read

ΔΨ0 + DΦ02 = (4γ − μ)Ψ0 + 3σΨ2 + (2ε − 2ε + ρ)Φ02

+ 2σΦ11 − λΦ00 , (3.9a)
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch03 page 59

Colliding Plane Gravitational Waves 59

3DΨ2 − 2DΦ11 + ΔΦ00 = −3λΨ0 + 9ρΨ2 + (μ + 2γ + 2γ)Φ00

− 2ρΦ11 − 2σΦ20 + σΦ02 , (3.9b)
3ΔΨ2 − 2ΔΦ11 + DΦ22 = 3σΨ4 − 9μΨ2 − (ρ + 2ε + 2ε)Φ22
+ 2μΦ11 + 2λΦ02 − λΦ20 , (3.9c)
DΨ4 + ΔΦ20 = (ρ − 4ε)Ψ4 − 3λΨ2 − (μ + 2γ − 2γ)Φ20
− 2λΦ11 + σΦ22 , (3.9d)
and
DΦ11 + ΔΦ00 + 3DΛ = 2(γ + γ − μ)Φ00
+ 4ρΦ11 + σΦ02 + σΦ20 , (3.10a)
ΔΦ11 + DΦ22 + 3ΔΛ = 2(ρ − ε − ε)Φ22
− 4μΦ11 − λΦ20 − λΦ02 . (3.10b)
Equations (3.7a)–(3.10b) are the basic differential equations for colliding
plane gravitational waves. To solve them for a given source usually follows
two different paths. One is to pose the initial data on the two half hypersur-
faces u = 0, v ≥ 0 and u ≥ 0, v = 0, and then solve the corresponding ini-
tial value problem (Szekeres, 1972; Yurtsever, 1988a, 1988b, 1989; Griffiths,
1991). However, this method is not practical in obtaining exact solutions,
especially in the non-collinear case (W = 0). Another method which was
pioneered by Khan and Penrose (1971) is essentially to work backward in
time. That is, we first find a solution that is valid in the interaction region
(Region IV), and then extend it back to the pre-collision regions (Regions
I–III) by means of the following substitutions:
u → uH(u), v → vH(v), (3.11)
in the metric coefficients, where H(x) denotes the Heaviside function,

1, x > 0,
H(x) = (3.12)
0, x < 0.

However, such an extension can guarantee the metric coefficients only be C 0

across the null hypersurfaces u = 0 and v = 0. To be physically acceptable,
we must impose some conditions on these coefficients.
We assume that all the metric coefficients are at least C 3 in the inter-
action region (Region IV), and C 0 across the hypersurfaces u = 0 or v = 0.
Then, as shown in Section 1.9, the Einstein field equations (1.77a)–(1.77r)
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch03 page 60

60 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

as well as the Bianchi identities (1.78a)–(1.78k) hold in the sense of distri-

butions. For any given function f , which is at least C 3 in Region IV and
C 0 across the hypersurfaces u = 0 or v = 0, it can be written in the form
(Wang, 1991f, 1992a),1

f (uH(u), vH(v)) = f (u, v)H(u)H(v) + f (0, v)[1 − H(u)]H(v)

+f (u, 0)[(1 − H(v)]H(u), (3.13)

where

f (0, v) = lim f (u, v), f (u, 0) = lim f (u, v). (3.14)

u→0+ v→0+

Thus, for any given test function F (u, v), we have

∂u f (uH(u), vH(v)), F (u, v)

= ∂u {f (u, v)H(u)H(v) + f (0, v)[1 − H(u)]H(v)
+ f (u, 0)[1 − H(v)]H(u)}, F (u, v)
= f,u (u, v)H(u)H(v) + f,u (u, 0)[1 − H(v)]H(u), F (u, v)

+ F (u, v)f (u, v)H(v)[∂u H(u)]dV
Ω

+ F (u, v)f (0, v)H(v)[∂u [1 − H(u)]]dV
Ω

+ F (u, v)f (u, 0)(1 − H(v)[∂u H(u)]dV
Ω

= f,u (u, v)H(u)H(v) + f,u (u, 0)[1 − H(v)]H(u), F (u, v)

 
− ∂u [F (u, v)f (u, v)H(v)]dV − ∂u [F (u, v)f (0, v)H(v)]dV
Ω+u Ω−u

− ∂u [F (u, v)f (u, 0)[1 − H(v)]]dV
Ω+u

= f,u (u, v)H(u)H(v) + f,u (u, 0)[1 − H(v)]H(u), F (u, v)

1A similar expression was also applied to domain walls (Wang, 1991c, 1992e, 1993;
Schmidt and Wang, 1993; Letelier and Wang, 1994) and brane worlds in string/M-theory
(Gong, Wang and Wu, 2008; Tziolas and Wang, 2008; Wang and Santos, 2008; Wu et al.,
2008; Wu, Gong and Wang, 2009; Tziolas, Wang and Wu, 20009; Devin et al., 2009; Wang
and Santos, 2010; Wang, 2010; Sharma et al., 2011), which turns out to be very useful
in the studies of singular surfaces.
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch03 page 61

Colliding Plane Gravitational Waves 61


− F (0, v)H(v)[f + (0, v) − f − (0, v)]dSu
∂Ω+u

− F (0, v)f (0, 0)[1 − H(v)]dSu , (3.15)
∂Ω+u

where F (u, v) is at least a C 1 -function with compact support on Ω. Since

f is C 0 across the hypersurface u = 0, we have f + (0, v) = f − (0, v), and
Eq. (3.15) becomes

∂u f (uH(u), vH(v)), F (u, v)

= f,u (u, v)H(u)H(v) + f,u (u, 0)[1 − H(v)]H(u), F (u, v)

− F (0, v)f (0, 0)[1 − H(v)]dSu . (3.16)
∂Ω+u

If we further assume that

f (0, 0) = 0, (3.17)

then Eq. (3.16) can be written as

∂u f (uH(u), vH(v)), F (u, v)

= f,u (u, v)H(u)H(v) + f,u (u, 0)[1 − H(v)]H(u), F (u, v), (3.18)

or equivalently

f,u (uH(u), vH(v)) = f,u (u, v)H(u)H(v) + f,u (u, 0)[1 − H(v)]H(u)
= f,u (u, vH(v))H(u). (3.19)

In a similar fashion, it can be shown that

f,v (uH(u), vH(v)) = f,v (uH(u), v)H(v),

f,vu (uH(u), vH(v)) = f,uv (u, v)H(v)H(u),
f,uu (uH(u), vH(v)) = f,uu (u, v)H(v)H(u) + f,uu (u, 0)[1 − H(v)]H(u)
+ δ(u)f,u (0, v)H(v) + δ(u)f,u (0, 0)[1 − H(v)]
= f,uu (u, vH(v))H(u) + δ(u)f,u (0, vH(v)),
f,vv (uH(u), vH(v)) = f,vv (u, v)H(v)H(u) + f,vv (0, v)[1 − H(u)]H(v)
+ δ(v)f,v (u, 0)H(u) + δ(v)f,v (0, 0)[(1 − H(u)]
= f,vv (uH(u), v)H(v) + δ(v)f,v (uH(u), 0), (3.20)
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch03 page 62

62 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

where

f,u (0, vH(v)) = lim+ f,u (u, vH(v)),

u→0
(3.21)
f,v (uH(u), 0) = lim+ f,v (uH(u), v).
v→0

Therefore, if we assume that

M (0, 0) = U (0, 0) = V (0, 0) = W (0, 0) = 0, (3.22)

then the non-vanishing spin coefficients given by Eq. (3.2) can be written
as
1
ρ= B(uH(u), v)U,v (uH(u), v)H(v),
2
1
μ = − A(u, vH(v))U,u (u, vH(v))H(u),
2

1
ε = − B(uH(u), v) [ln A(uH(u), v)],v
2

i
+ V,v (uH(u), v) sinh W (uH(u), v) H(v),
2

1
γ = A(u, vH(v)) [ln B(u, vH(v))],u
2 (3.23)

i
− V,u (u, vH(v)) sinh W (u, vH(v)) H(u),
2
1
σ = − B(uH(u), v){V,v (uH(u), v) cosh W (uH(u), v)
2
− iW,v (uH(u), v)}H(v),

1
λ= A(u, vH(v)){V,u (u, vH(v)) cosh W (u, vH(v))
2
+ iW,u (u, vH(v))}H(u).

Similarly, from Eqs. (3.7a)–(3.8e) we find

Ψ0 (uH(u), vH(v)) = ΨIV Im

0 (uH(u), v)H(v) + Ψ0 (uH(u))δ(v),

Ψ2 (uH(u), vH(v)) = ΨIV

2 (u, v)H(u)H(v), (3.24)
Ψ4 (uH(u), vH(v)) = ΨIV
4 (u, vH(v))H(u) + ΨIm
4 (vH(v))δ(u),
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch03 page 63

Colliding Plane Gravitational Waves 63

and

Φ00 (uH(u), vH(v)) = ΦIV Im

00 (uH(u), v)H(v) + Φ00 (uH(u))δ(v),

Φ11 (uH(u), vH(v)) = ΦIV

11 (u, v)H(u)H(v),

Φ02 (uH(u), vH(v)) = ΦIV

02 (u, v)H(u)H(v), (3.25)
Φ22 (uH(u), vH(v)) = ΦIV
22 (u, vH(v))H(u) + ΦIm
22 (vH(v))δ(u),

Λ(uH(u), vH(v)) = ΛIV (u, v)H(u)H(v),

where ΨIV IV
i (u, v) and Φij (u, v) are the corresponding Weyl and Ricci scalars
in Region IV, obtained from Eqs. (3.7a)–(3.8e), and ΨIm Im
0 . . . Φ22 are the
impulsive part of the Riemann tensor with support on the hypersurfaces
u = 0 and v = 0, respectively, and defined by

1 2
ΨIm
0 (uH(u)) ≡ − B (uH(u), 0)[V,v (uH(u), 0) cosh W (uH(u), 0)
2
− iW,v (uH(u), 0)],
(3.26)
1
ΨIm
4 (vH(v)) ≡ − A2 (0, vH(v))[V,u (0, vH(v)) cosh W (0, vH(v))
2
+ iW,u (0, vH(v))],

and

1 2
ΦIm
00 (uH(u)) ≡ B (uH(u), 0)U,v (uH(u), 0),
2
(3.27)
1 2
ΦIm
22 (vH(v)) ≡ A (0, vH(v))U,u (0, vH(v)).
2

The terms ΦIm Im

00 and Φ22 are usually interpreted as representing impulsive
shells of null dust with support, respectively, on the hypersurfaces v = 0 and
u = 0 (Dray and ’t Hooft, 1986; Taub, 1988a; Tsoubelis, 1989b; Tsoubelis
and Wang, 1990, 1991, 1992).
From the assumption that the metric coefficients are at least C 3 in
the interaction region, we can see that ΨIV IV 1
i (u, v) and Φij (u, v) are C in
Im Im Im Im 2
that region, while Ψ0 , Φ00 , and Ψ4 , Φ22 are C , respectively, on the
hypersurfaces v = 0, and u = 0. Then, from Eqs. (3.24) and (3.25) we can
see that Ψ2 , Φ11 , Φ02 and Λ, in general, have step function discontinuities
across either the hypersurface u = 0 or v = 0.
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch03 page 64

64 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

On the other hand, it can be shown that for any given C 1 -test function
F (u, v) we have

[1 − H(u)]H(u), F (u, v) = F (u, v)[1 − H(u)]H(u)dV
Ω

= F (u, v)[1 − H(u)]dV = 0,
Ω+u

H 2 (u), F (u.v) = F (u, v)H(u)H(u)dV
Ω
 
= F (u, v)H(u)dV = F (u, v)dV,
Ω+u Ω+u

2
[1 − H(u)] , F (u, v) = F (u, v)[1 − H(u)][1 − H(u)]dV
Ω

= F (u, v)[1 − H(u)]dV
Ω−u

= F (u, v)dV, (3.28)
Ω−u

or equivalently

[1 − H(u)]H(u) = 0, H 2 (u) = H(u), [1 − H(u)]2 = [1 − H(u)], (3.29)

in the sense of distributions. From Eqs. (3.19)–(3.20) and (3.29), we find

ΨIV IV IV
4 (u > 0, vH(v)) = Ψ4 (u > 0, v > 0)H(v) + Ψ4 (u > 0, 0)[1 − H(v)],

ΨIm Im Im
4 (vH(v)) = Ψ4 (v > 0)H(v) + Ψ4 (0)[1 − H(v)], (3.30)

etc., where

ΨIV IV
4 (u > 0, 0) = lim+ Ψ4 (u > 0, v),
v→0
(3.31)
ΨIm
4 (0) = lim+ ΨIm
4 (v).
v→0

Since ΨIV Im 1 2
4 (u > 0, v > 0) and Ψ4 (v > 0) are at least C and C , respec-
tively, we can see that Ψ4 (u > 0, vH(v)) and Ψ4 (vH(v)) are C 0 across
IV Im

the hypersurface v = 0.
Similarly, it can be shown that ΦIV Im
22 (u > 0, vH(v)) and Φ22 (vH(v))
are C 0 , too, across the hypersurface v = 0, while ΨIV0 (uH(u), v > 0),
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch03 page 65

Colliding Plane Gravitational Waves 65

ΨIm IV Im 0
0 (uH(u)), Φ00 (uH(u), v > 0) and Φ00 (uH(u)) are C across the hyper-
surface u = 0. Thus, we have

D[Ψ4 (uH(f u), vH(v))], F (u, v)

≡ lμ ∂μ [Ψ4 (uH(u), vH(v))], F (u, v)
= B∂v [Ψ4 (uH(u), vH(v))], F (u, v)
= B∂v {ΨIV IV
4 (u, v)H(u)H(v) + Ψ4 (u, 0)H(u)[1 − H(v)]

+ΨIm Im
4 (v)H(v)δ(u) + Ψ4 (0)[1 − H(v)]δ(u)}, F (u, v)

= H(u)H(v)B∂v [ΨIV Im
4 (u, v)] + H(v)δ(u)B∂v [Ψ4 (v)], F (u, v)

D[Ψ2 (uH(u), vH(v))], F (u, v)

= D[ΨIV
2 (u, v)H(u)H(v)], F (u, v)

= H(u)H(v)D[ΨIV
2 (u, v)]

+ ΨIV
2 (u, v)H(u)DH(v), F (u, v), (3.32)

or equivalently

D[Ψ4 (uH(u), vH(v))] = H(u)H(v)D[ΨIV Im

4 (u, v)] + H(v)δ(u)D[Ψ4 (v)],

D[Ψ2 (uH(u), vH(v))] = H(u)H(v)D[ΨIV IV

2 (u, v)] + BH(u)δ(v)Ψ2 (u, 0).
(3.33)

Substituting Eq. (3.23) into Eqs. (3.9a)–(3.10b), and taking Eq. (3.33)
into account, we finally obtain

ΔΨIV IV IV IV IV
0 + DΦ02 = (4γ − μ)Ψ0 + 3σΨ2 + (2ε − 2ε + ρ)Φ02

+ 2σΦIV IV
11 − λΦ00 ,

3DΨIV IV IV IV IV IV
2 − 2DΦ11 + ΔΦ00 = −3λΨ0 + 9ρΨ2 + (μ + 2γ + 2γ)Φ00

− 2ρΦIV IV IV
11 − 2σΦ20 + σΦ02 ,
(3.34)
3ΔΨIV IV IV IV IV IV
2 − 2ΔΦ11 + DΦ22 = 3σΨ4 − 9μΨ2 − (ρ + 2ε + 2ε)Φ22

+ 2μΦIV IV IV
11 + 2λΦ02 − λΦ20 ,

DΨIV IV IV IV IV
4 + ΔΦ20 = (ρ − 4ε)Ψ4 − 3λΨ2 − (μ + 2γ − 2γ)Φ20 ,

− 2λΦIV IV
11 + σΦ22 , (u, v > 0),
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch03 page 66

66 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

and

DΦIV IV
11 + ΔΦ00 + 3DΛ
IV
= 2(γ + γ − μ)ΦIV IV
00 + 4ρΦ11

+σΦIV IV
02 + σΦ20 ,
(3.35)
ΔΦIV IV
11 + DΦ22 + 3ΔΛ
IV
= 2(ρ − ε − ε)ΦIV IV
22 − 4μΦ11

−λΦIV IV
20 − λΦ02 , (u, v > 0),

in the interacting region (Region IV),

ΔΨIm IV Im Im
0 + BΦ02 = (4γ − μ)Ψ0 − λΦ00 ,

ΔΦIm IV IV Im Im
00 + 3BΨ2 − 2BΦ11 = (μ + 2γ + 2γ)Φ00 − 3λΨ0 , (3.36)
ΔΦIm IV
00 + BΦ11 + 3BΛ
IV
= 2(γ + γ − μ)ΦIm
00 , (u > 0, v = 0),

along the null surface v = 0 but with u > 0, and

DΨIm IV Im Im
4 + AΦ20 = −(4ε − ρ)Ψ4 + σΦ22 ,

DΦIm IV IV Im Im
22 + 3AΨ2 − 2AΦ11 = −(ρ + 2ε + 2ε)Φ22 + 3σΨ4 , (3.37)
DΦIm IV
22 + AΦ11 + 3AΛ
IV
= −2(ε + ε − ρ)ΦIm
22 , (u = 0, v > 0),

along the null surface u = 0 with v > 0. Inside Regions I–III, the Bianchi
identities are satisfied identically.
Equation (3.34) shows that the shock parts of the two plane gravita-
tional waves represented, respectively, by ΨIV IV
0 and Ψ4 , interact with each
other through the Coulomb field Ψ2 . The components ΦIV
IV IV
02 and Φ11 of the
matter field interact with both Ψ0 and Ψ4 . That is, Φ02 and ΦIV
IV IV IV
11 are
gravitationally active to both Ψ0 and Ψ4 (Szekeres, 1965). Similarly, ΦIV
IV IV
00
is gravitationally active only to ΨIV IV IV
0 , and Φ22 only to Ψ4 . The component
ΛIV is gravitationally inert to both ΨIV 0 and Ψ4 .
IV

On the other hand, from Eqs. (3.36) and (3.37) we can see that the
component ΦIV Im Im Im
02 is gravitationally active to both Ψ0 and Ψ4 , Φ00 only to
Im Im Im
Ψ0 , and Φ22 only to Ψ4 , while all the other components of the matter
field are gravitationally inert to both ΨIm 0 and Ψ4 .
Im

From Eqs. (3.36) and (3.37) we can also see that the interaction between
purely gravitational impulsive waves is carried out through the Coulomb
field ΨIV2 (Khan and Penrose, 1971; Nutku and Halil, 1977; Wang, 1991f,
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch03 page 67

Colliding Plane Gravitational Waves 67

1992a). Thus, it is necessary for the spacetime to be curved in the interac-

tion region (Region IV) for the collision of two purely gravitational impul-
sive waves with support only on the hypersurfaces u = 0 and v = 0,
although in the pre-collision regions (Regions I–III), the spacetime is flat.

3.3. Different Coordinate Systems

In the studies of spacetimes for colliding plane gravitational waves, several
different coordinate systems are usually used. A successful choice of coordi-
nate systems is of great importance both for solving the NP equations and
for studying the properties of the solutions.
One of the most suitable coordinate system for colliding plane waves
is the (u, v, x2 , x3 ) coordinates, since in this system the two plane gravita-
tional waves move along the u = constant and v = constant hypersurfaces,
respectively, and the spacetime is automatically divided into four regions
[see Fig. 3.1]. However, experience tells us that this coordinate system is
not always convenient, especially when we try to find exact solutions of the
Einstein field equations. In this section, we shall introduce several coordi-
nate systems (Ferrari, 1988; Griffiths, 1991; Wang, 1991f), each of which
meets our special need. Since the metric coefficients depend only on the null
coordinates u and v, the choice of the coordinates x2 and x3 is the same
in all these regions. Thus, in the following we consider only the different
choices of the coordinates x0 and x1 .

3.3.1. The (t, z)-coordinate system

The (t, z)-coordinates are commonly used to study the singularities formed
after the collision (Yurtsever, 1987, 1988a, 1989; Feinstein and Ibañez, 1989;
Griffiths, 1991), and to obtain exact solutions (Ferrari and Ibañez, 1987a,
1987b, 1989a, 1989b; Ferrari, Ibañez and Bruni, 1987a, 1987b; Wang, 1991a;
Tsoubelis and Wang, 1992) by means of the soliton technique (or inverse
scattering method) of Belinsky and Zakharov (Belinsky and Zakharov,
1978, 1979; Belinsky and Verdaguer, 2001).
The coordinates t and z are given by
t ≡ 1 − u2n − v 2m , z ≡ u2n − v 2m , (3.38)
or inversely
 
2n 1−t+z 2m 1−t−z
u= , v= , (3.39)
2 2
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch03 page 68

68 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

where n and m are two constants. In addition, in this coordinate system a

special gauge is usually chosen

e−U = t. (3.40)

In terms of t and z, the metric (2.22) takes the form

ds2 = f (t, z)(dt2 − dz 2 ) − t{eV cosh W (dx2 )2 − 2 sinh W dx2 dx3

+e−V cosh W (dx3 )2 }, (3.41)

where now V, W are functions of t and z, and the function f (t, z) is

defined by

e−M
f≡ . (3.42)
8nmu2n−1 v 2m−1

3.3.2. The (η, μ)-coordinate system

The (η, μ)-coordinate system was first used by Chandrasekhar and Ferrari
(1984) for the spacetimes for colliding plane gravitational waves and later
used by several others, see, for example, Tsoubelis and Wang (1989, 1992),
Griffiths (1991), Wang (1991a), and references therein. In this coordinate
system, the Einstein field equations for the functions V and W take the
form

(1 − |E|2 ){[(1 − η 2 )E,η ],η − [(1 − μ2 )E,μ ],μ }

= −2Ē[(1 − η 2 )(E,η )2 − (1 − μ2 )(E, μ)2 ], (3.43)

which was originally found by Ernst (1968a, 1968b), where the Ernst poten-
tial E is defined via the relations

Z −1 eV sinh W eV
Z ≡ χ + iq2 , E≡ , χ≡ , q2 ≡ , (3.44)
Z +1 cosh W cosh W

and η, μ are given by

η = un X + v m Y, μ = un X − v m Y,
(3.45)
X = (1 − v 2m )1/2 , Y = (1 − u2n )1/2 .
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch03 page 69

Colliding Plane Gravitational Waves 69

Taking the first and second derivatives of η and μ with respect to u and v,
we obtain the following useful relations:
nun−1 mv m−1
η,u = (XY − un v m ), η,v = (XY − un v m ),
Y X
n(n − 1)un−2 n2 v m u2n−2
η,uu = (XY − un v m ) − ,
Y Y3
m(m − 1)v m−2 m2 un v 2m−2
η,vv = (XY − un v m ) − ,
X X3
nmun−1 v m·1
η,uv = − η, (3.46)
XY
nun−1 mv m−1
μ,u = (XY + un v m ), μ,v = − (XY + un v m ),
Y X
n(n − 1)un−2 n2 v m u2n−2
μ,uu = (XY + un v m ) + ,
Y Y3
m(m − 1)v m−2 m2 un v 2m−2
μ,vv =− (XY + un v m ) − ,
X X3
nmu v
n−1 m−1
μ,uv = μ.
XY
In addition, we also have

1 − η 2 = (XY − un v m )2 , 1 − μ2 = (XY + un v m )2 . (3.47)

Combining Eqs. (3.46) and (3.47), we find

dη 2 dμ2 (η,u du + η,v dv)2 (μ,u du + μ,v dv)2
− = −
1 − η2 1 − μ2 1 − η2 1 − μ2
4nmu v
n−1 m−1
= dudv. (3.48)
XY
On the other hand, from Eqs. (3.38) and (3.45) we obtain

t = (1 − η 2 )1/2 (1 − μ2 )1/2 , z = ημ. (3.49)

Thus, we have
tη tμ
t,η = − , t,μ = − , z.η = μ, z,μ = η, (3.50)
1 − η2 1 − μ2
and
 dη 2 dμ2
dt2 − dz 2 = η 2 − μ2 2
− . (3.51)
1−η 1 − μ2
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch03 page 70

70 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

Substituting Eqs. (3.44), (3.49) and (3.51) into Eq. (3.41), we obtain
dη 2 dμ2
ds2 = (η 2 − μ2 )f (η, μ) 2
−
1−η 1 − μ2

− [(1 − η 2 )(1 − μ2 )]1/2 [χ(dx2 )2 + χ−1 (dx3 − q2 dx2 )2 ]. (3.52)

3.3.3. The (φ, θ)-coordinate system

To calculate the Weyl and Ricci scalars, it was found convenient to work in
the φ and θ coordinates (Chandrasekhar and Ferrari, 1984; Chandrasekhar
and Xanthopoulos, 1985a, 1986a), which are defined by

η = cos φ, μ = cos θ. (3.53)

In this coordinate system, the metric (3.52) takes the form

ds2 = (cos2 φ − cos2 θ)f (φ, θ)(dφ2 − dθ2 )

− sin φ sin θ{χ(dx2 )2 − χ−1 (dx3 − q2 dx2 )2 }. (3.54)

It can be shown that the following relations hold:

φ+θ θ−φ
un = cos , v m = sin ,
2 2 (3.55)
t = sin φ sin θ, z = cos φ cos θ.

3.4. Gravitational Analog of the Faraday Rotation

In 1846, Faraday (1846a, 1846b) discovered that a magnetic field could
affect the propagation of light. Suppose that a uniform magnetic field is
imposed in a medium, and let a linearly polarized light beam propagate
along the line of the force, then it was found that this linearly polarized
light beam emerges with its plane of polarization rotated through an angle
proportional to the magnetic field and the thickness of the medium. In
electrodynamics, this phenomenon is known as the Faraday rotation (or
Faraday effect). Because of the similarity between the electromagnetic and
gravitational fields, we expect that a similar phenomenon also occurs in
a gravitational field. Stark, Connors and Piran (Stark and Connors, 1977;
Connors and Stark, 1977; Connors, Piran and Stark, 1980) first investi-
gated the problem by observing the polarization vector of an X-ray emitted
from an accretion disk surrounding a Kerr black hole, and found that the
polarization vector rotates in the gravitational field of a black hole. Later,
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch03 page 71

Colliding Plane Gravitational Waves 71

Ishihara, Takahashi and Tomimatsu (1988) studied the gravitational lens-

ing effect caused by a Kerr black hole via the propagation of a polarization
vector along a light ray. In particular, they found a rotation of the plane
of polarization due to the presence of the black hole’s spin. The rotation
angle is proportional to the mass and the line-of-sight component of the
angle momentum of the black hole.
Later, Piran, Safier and Stark (Piran and Safier, 1985; Piran, Safier,
and Stark, 1985; Piran, Safier, and Katz, 1986) studied the cylindrically
symmetric spacetimes (Bronnikov, Santos and Wang, 2019). Starting from
a cylindrical line element, they first defined the “+” and “×” modes of
polarizations of a cylindrical gravitational wave, and then by using both
analytic approximations and numerical calculations they showed that a
conversion between the “+” and “×” modes occurs. If an outgoing (or
ingoing) cylindrical wave is linearly polarized, its polarization vector rotates
as it propagates. This is another gravitational analog of the electromagnetic
Faraday rotation, but with cylindrical symmetry. It should be noted that
the decomposition of the “+” and “×” modes considered by these authors
is coordinate-dependent.
In this section, by using the definition given in Section 2.3, we shall
provide a coordinate-independent analysis for the changes of polarizations,
but now with plane symmetry (Wang, 1991b, 1991f).
As mentioned in Section 3.2, each of the null vectors lμ and nμ defines
a null geodesic congruence. However, from Eq. (3.1) it can be seen that the
μ μ
space-like vectors E(2) and E(3) are parallelly transported along the null
geodesics defined by nμ or the ones by lμ only in Regions II and III, but
not in the interaction region (Region IV). As a matter of fact, now we have
μ 1 μ
E(2);ν lν = sinh W V,ν lν E(3) ,
2 (3.56)
μ 1 μ
E(3);ν lν = − sinh W V,ν lν E(2) ,
2
and
μ 1 μ
E(2);ν nν = sinh W V,ν nν E(3) ,
2 (3.57)
μ 1 μ
E(3);ν nν = − sinh W V,ν nν E(2) .
2
Thus, the angles φ0 and φ4 defined, respectively, by Eqs. (2.46) and (2.38)
have physical meaning only locally in the interaction region (Region IV). In
order to compare the polarization of a plane gravitational wave at different
times and places, we have to find a parallelly transported basis along the
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch03 page 72

72 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

wave path and then define the polarization angle relative to this parallelly
transported basis. In this way, we can see that the changes of polariza-
tion of the wave has an absolutely physical meaning and independent of
the choice of the coordinates. For example, if the changes (relative to the
parallelly transported basis) are zero, it means that the polarization vector
μ
E(2) defined in Section 2.3 is parallel to the parallelly transported basis.
Clearly, such defined parallelism is independent of the coordinates.
To find a parallelly transported basis along the null geodesics defined
μ μ
by lμ , we make a rotation in the (E(2) , E(3) )-plane as we did in Eq. (2.32)
μ μ(0)
but with the angle φ4 now denoted by φ4 , and E(2) and E(3) by λμ(2) and
λμ(3) , respectively. Then, we find
1 (0)
λμ(2);ν lν = (sinh W V,ν − 2ϕ4,ν )lν λμ(3) ,
2
(3.58)
1 (0)
λμ(3);ν lν = − (sinh W V,ν − 2ϕ4,ν )lν λμ(2) .
2
(0)
Therefore, if the angle φ4 is chosen so that
(0)
sinh W V,v − 2ϕ4,v = 0, (3.59)
the space-like orthogonal vectors and λμ(2) λμ(3)
are parallelly transported
along the null geodesics (or the Ψ4 -wave path) defined by lμ , and the
difference
(0)
θ4 ≡ ϕ4 − ϕ4 , (3.60)
defines the angle between the polarization direction of the Ψ4 -wave and the
λμ(2) direction.
μ μ
Similarly, if the basis (E(2) , E(2) ) is rotated so that the rotated angle
(0)
ϕ0 satisfies
(0)
sinh W V,u − 2ϕ0,u = 0, (3.61)
the vectors λμ(2) and λμ(3) are parallelly transported along the Ψ0 -wave path,
and the angle
(0)
θ0 ≡ ϕ0 − ϕ0 , (3.62)
determines the polarization direction of the Ψ4 -wave relative to the
(λμ(2) , λμ(3) )-basis. From Eqs. (3.59) and (3.61), we can see that the angle
(0) (0)
φ4 (φ0 ) is constant in Region III (Region II). Without loss of generality,
(0) (0)
we can choose φ4 (φ0 ) to be zero in Region III (Region II), such that the
μ μ μ μ
(λ(2) , λ(3) )-basis coincides with the (E(2) , E(3) )-basis in these regions.
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch03 page 73

Colliding Plane Gravitational Waves 73

On the other hand, Eq. (3.24) shows that the Ψ0 - and Ψ4 -waves, in gen-
eral, consist of two parts. We call the δ-function part the impulsive part, and
the H-function part the shock part, although the latter still includes three
different cases: smooth wave, shock wave and the wave with unbounded
wavefront (Tsoubelis and Wang, 1989). The treatment for these three cases,
however, is the same for the present problem, so we shall not distinguish
them in the following discussions. We shall use θ0sh and θ0Im to denote the
polarization angles for the shock and impulsive parts of Ψ0 , and θ4sh and θ4Im
for the shock and impulsive parts of Ψ4 , respectively. Thus, in Region II we
have θ0sh = θ0sh (v) and θ0Im = constant. That is, in Region II the shock part
in general is non-collinearly polarized, while the impulsive part is always
collinearly polarized. Along the wave path, the two angles θ0sh and θ0Im do
not change in this region
sh Im
θ0,u = 0 = θ0,u . (3.63)
Of cause, in Regions I and III the angles θ0sh and θ0Im vanish, since Ψ0
vanishes there.
Similarly, for the Ψ4 -wave, we have
sh
θ4,u = 0 = θ4Im,u , (3.64)
in Region III, and θ4sh and θ4Im vanish in Regions I and II, since Ψ4 vanishes
in these two regions.
In the interaction region (Region IV), the situation is different. From
the Bianchi identities (3.34)–(3.37), we find that
1
AΨIV
0,u = {A[4(ln B),u + U,u − 2i sinh W V,u ]Ψ0
IV
2
−3B(cosh W V,v − iW,v )ΨIV IV
2,v − 2BΦ02,v

+B(U,v − 2i sinh W V,v )ΦIV

02

−2B(cosh W V,v − iW,v )ΦIV

11

−A(cosh W V,u − iW,u )ΦIV

00 },
1
BΨIV
4,v = {B[4(ln A),v + U,v + 2i sinh W V,v ]ΨIV
4
2
−3A(cosh W V,u + iW,u )ΨIV IV
2 − 2AΦ20,u

+A(U,u + 2i sinh W V,u )ΦIV

20

−2A(cosh W V,u + iW,u )ΦIV

11

−B(cosh W V,v + iW,v )ΦIV

22 }, (u, v > 0), (3.65)
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch03 page 74

74 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

and
1
AΨIm
0,u = {A[4(ln B),u + U,u − 2i sinh W V,u ]ΨIm
0
2
−2BΦIV Im
02 − A(cosh W V,u − iW,u )Φ00 }, (u > 0, v = 0),
(3.66)
1
BΨIm
4,v = {B[4(ln A),v + U,v + 2i sinh W V,v ]ΨIm
4
2
−2AΦIV Im
02 − B(cosh W V,v − iW,v )Φ22 }, (u = 0, v > 0).

It is now convenient to introduce the following “scale-invariant” quan-

tities via the relations (Szekeres, 1972; and Griffiths, 1976b),
(0) (0) (0)
Ψ0 = B 2 Ψ0 , Ψ2 = AB Ψ2 , Ψ4 = A2 Ψ4 ,
(0) (0) (0)
Φ00 = B 2 Φ00 , Φ11 = ABΦ11 , Φ02 = ABΦ02 , (3.67)
(0)
Φ22 = A2 Φ22 , Λ = ABΛ(0) .

Since from now on only the “scale invariant” terms are used, we shall drop
all of the super-indices “0” from these quantities without causing any confu-
sion. Then, substituting Eq. (3.67) into Eqs. (3.65) and (3.66), respectively,
we find
1
ΨIV
0,u = {[U,u − 2i sinh W V,u ]ΨIV
0
2
−3(cosh W V,v − iW,v )ΨIV IV
2 − 2Φ02,v

+(U,v − 2M,v − 2i sinh W V,v )ΦIV

02

−2(cosh W V,v − iW,v )ΦIV

11

−(cosh W V,u − iW,u )ΦIV

00 },
(3.68)
1
ΨIV
4,v = {[U,v + 2i sinh W V,v ]ΨIV
4
2
−3(cosh W V,u + iW,u )ΨIV IV
2 − 2Φ20,u

+(U,u − 2M,u + 2isinh W V,u )ΦIV

20

−2(cosh W V,u + iW,u )ΦIV

11

−(cosh W V,v + iW,v )ΦIV

22 }, (u, v > 0),
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch03 page 75

Colliding Plane Gravitational Waves 75

and
1
ΨIm
0,u = {[U,u − 2i sinh W V,u ]ΨIm
0
2
−2ΦIV Im
02 − (cosh W V,u − iW,u )Φ00 }, (u > 0, v = 0),
(3.69)
1
ΨIm
4,v = {[U,v + 2i sinh W V,v ]ΨIm
4
2
−2ΦIV Im
02 − (cosh W V,v − iW,v )Φ22 }, (u > 0, v = 0).
Hence, we find that
sh 1
θ0,u =− {3[cosh W V,v Im(ΨIV IV IV IV
0 Ψ2 ) + W,v Re(Ψ0 Ψ2 )]
4(ΨIV IV
0 Ψ0 )

+ 2 Im(ΨIV IV IV IV
0 Φ20,v ) + (2M,v − U,v )Im(Ψ0 Φ20 )

+ 2ΦIV IV IV
11 [cosh W V,v Im(Ψ0 ) + W,v Re(Ψ0 )]

+ ΦIV IV IV
00 [cosh W V,u Im(Ψ0 ) + W,u Re(Ψ0 )]

− 2 sinh W V,v Re(ΨIV IV

0 Φ20 )}, (3.70)
sh 1
θ4,v = {3[cosh W V,u Im(ΨIV IV IV IV
4 Ψ2 ) − W,u Re(Ψ4 Ψ2 )]
4(ΨIV IV
4 Ψ4 )

+ 2 Im(ΨIV IV IV IV
4 Φ02,u ) + (2M,u − U,u )Im(Ψ4 Φ02 )

+ 2ΦIV IV IV
11 [cosh W V,v Im(Ψ4 ) − W,v Re(Ψ4 )]

+ ΦIV IV IV
22 [cosh W V,v Im(Ψ0 ) − W,v Re(Ψ4 )]

+ 2 sinh W V,u Re(ΨIV IV

4 Φ02 )}, (u, v > 0),
and
1
Im
θ0,u =− {[cosh W V,u Im(ΨIm Im Im
0 ) + W,u Re(Ψ0 )]Φ00
4(ΨIm Im
0 Ψ0 )

+ 2 Im(ΨIm IV
0 Φ20 )}, (u > 0, v = 0),
(3.71)
1
Im
θ4,v = {[cosh W V,v Im(ΨIm Im Im
4 ) − W,v Re(Ψ4 )]Φ22
4(ΨIm Im
4 Ψ4 )

+ 2 Im(ΨIm IV
4 Φ02 )}, (u > 0, v = 0).
Equations (3.70) and (3.71) show that, due to the interaction between
the two incoming plane gravitational waves and the interaction with the
matter fields, the polarization directions of the Ψ0 - and Ψ4 -waves got
changed relative to the parallelly transported basis along their wave paths.
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch03 page 76

76 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

In other words, the polarization direction of a plane gravitational wave is

no longer parallelly transported along the wave path because of the above
two kinds of interaction (Wang, 1991b). The change of polarizations of
a plane gravitational wave caused by the interaction with another oppo-
sitely moving plane gravitational wave is the exact analog of the well-known
electromagnetic Faraday rotation, but having the oppositely moving plane
gravitational wave as the medium and the magnetic field. We call the effect
caused by the nonlinear interaction between the plane gravitational waves
and matter fields the deflection effect, and the effect caused by the interac-
tion between the two plane gravitational waves the gravitational Faraday
rotation (or gravitational Faraday effect).
When W = 0, all the Weyl and Ricci scalars are real [see Eqs. (3.7a)–
(3.8e), (3.26) and (3.27)]. Consequently, from Eqs. (3.70) and (3.71) we find
sh sh Im sh
θ0,u = θ4,v = θ0,u = θ4,v = 0, (W = 0). (3.72)

That is, in the collinear case the polarizations of colliding plane gravita-
tional waves do not change.
To further illustrate the properties of the polarization of colliding plane
gravitational waves, in the rest of this section we restrict ourselves to several
specific cases (with W = 0), which are interesting from the point of view
of physics. Because of the symmetry shared by the two plane gravitational
waves, it is sufficient to consider only one of them, say, the Ψ0 -wave. In
addition, since we are now working in Region IV, we do not make any
more specific statements about it in the following, and understand all the
following results valid only in this region (plus its two boundaries u = 0, v >
0 and u > 0, v = 0).

3.4.1. Vacuum spacetimes

When the spacetime is vacuum, the corresponding collision is a purely grav-
itational one, and the Ricci scalars are zero,

Φij = 0, Λ = 0. (3.73)

Then, Eq. (3.70) reduces to

sh 3
θ0,u =− {cosh W V,v Im(ΨIV IV
0 Ψ2 )
4(ΨIV IV
0 Ψ0 )

+ W,v Re(ΨIV IV
0 Ψ2 )}. (3.74)
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch03 page 77

Colliding Plane Gravitational Waves 77

sh
Thus, if Ψ2 = 0, then we have θ0,u = 0. That is, if Ψ2 = 0, then the
polarization of a plane gravitational wave does not change. However, it was
shown that the “Coulomb” part Ψ2 appears necessarily in the interaction
region (Region IV), due to the nonlinearity of the Einstein field equations
(Szekeres, 1972). Thus, the change of polarizations of two colliding purely
gravitational waves is purely due the nonlinear interaction between the two
plane gravitational waves.
On the other hand, from Eq. (3.68) we find

Im
θ0,u |v=0 = 0. (3.75)

That is, the impulsive plane gravitational wave does not change its polar-
ization after collision, when no matter is present in the spacetime. From
Eq. (3.36) we can see that in the present case the ΨIm
0 -wave component
does not interact with other components.

3.4.2. Spacetimes filled with null dust

When a spacetime is filled with a null dust, the energy–stress tensor can
be written in the form (Chandrasekhar and Xanthopoulos, 1986b; and
Tsoubelis and Wang, 1990, 1991)

Tμv = ε1 lμ lν + ε2 nμ nν , (3.76)

which is the superposition of two pure radiation fields given by Eq. (1.94),
where ε1 and ε2 are non-negative. Equation (3.76) represents a pair of
oppositely moving null dust clouds with the energy density ε1 and ε2 ,
respectively, each of which is separately conserved (Taub, 1988a; Tsoubelis
and Wang, 1991). The corresponding non-vanishing Ricci scalars are
given by
ε2 ε1
Φ00 = , Φ22 = . (3.77)
2B 2 2A2

On the other hand, Eq. (3.25) shows that, like Ψ0 and Ψ4 , the compo-
nents Φ00 and Φ22 , in general, consist of two parts: the H-function part and
the δ-function part. The latter is supported on the hypersurfaces u = 0 or
v = 0, and usually interpreted as impulsive shells of null dust. When atten-
tion is restricted to the inside of the interaction region, only the H-function
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch03 page 78

78 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

part remains, and from Eq. (3.70) we find

sh 1
θ0,u =− {3[cosh W V,v Im(ΨIV IV
0 Ψ2 )
4(ΨIV IV
0 Ψ0 )

+ W,v Re(ΨIV IV IV IV
0 Ψ2 )] + Φ00 [cosh W V,u Im(Ψ0 )

+ W,u Re(ΨIV
0 )]}. (3.78)
sh
Unlike in the vacuum case, now θ0,u can be different from zero even when
IV
Ψ2 = 0, because of the presence of the last term in the right-hand side
of Eq. (3.78), which represents the interaction between ΨIV IV
0 and Φ00 [see
Eq. (3.34)].
It was shown that, when null dust is present the collision of two plane
gravitational waves does not require the Coulomb field Ψ2 to appear neces-
sarily in the interaction region (Tsoubelis and Wang, 1990). Thus, a plane
gravitational shock wave can change its polarization due to the deflection
effect. On the other hand, Eq. (3.71) now becomes

Im 1
θ0,u =− {[cosh W V,u Im(ΨIm
0 )
4(ΨIm Im
0 Ψ0 )
+ W,u Re(ΨIm Im
0 )]Φ00 }. (3.79)

Obviously, when ΦIm00 is different from zero, the polarization of the impulsive
part of Ψ0 changes after collision because of the interaction between ΨIm 0
and ΦIm
00 [see Eq. (3.36)].

3.4.3. Spacetimes filled with a massless scalar field

For a massless scalar field, the energy–stress tensor is given by Eq. (1.92)
with the potential φ satisfying Eq. (1.93). Since in the present case all of the
metric coefficients are functions of u and v only, without loss of generality,
we assume that φ = φ(u, v). Then, Eq. (1.93) becomes (Tsoubelis and
Wang, 1991),

2φ,uv − U,u φ,v − U,v φ,u = 0. (3.80)

The non-vanishing Ricci scalars are given by

1 2 1
Φ00 = φ,v , Φ22 = φ2,u ,
2 2 (3.81)
1 1
Φ11 = φ,u φ,v , Λ = − φ,u φ,v .
4 12
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch03 page 79

Colliding Plane Gravitational Waves 79

Substituting Eq. (3.81) into Eq. (3.70), we find

sh 1
θ0,u =− {6[cosh W V,v Im(ΨIV IV
0 Ψ2 )
8(ΨIV IV
0 Ψ0 )

+ W,v Re(ΨIV IV
0 Ψ2 )]

+ φ,v [cosh W (φ,u V,v + φ,v V,u )Im(ΨIV

0 )

+ (φ,u W,v + φ,u W,v ) Re(ΨIV

0 )]}, (3.82)

which shows that a plane gravitational shock wave changes its polarization
due to both the nonlinear interaction between the two plane gravitational
waves and the interaction with the massless scalar field φ.
When the spacetime is filled only with a massless scalar field, we have

ΦIm Im
00 = 0 = Φ22 , (3.83)

or equivalently

U,u |u=0 = 0 = U,v |v=0 . (3.84)

Combining Eqs. (3.84) and (3.71) we find

Im
θ0,u = 0, (3.85)

which means that an impulsive plane gravitational wave does not change
its polarization when it passes through a massless scalar field, since in this
case there is no interaction between the gravitational impulsive wave and
the massless scalar field [see Eqs. (3.36) and (3.37)].
Note that Eq. (3.84) is also the condition under which the hypersurfaces
u = 0 and v = 0 are free of matter (Tsoubelis and Wang, 1989, 1990).

3.4.4. Spacetimes filled with an electromagnetic field

When an electromagnetic field is null, its energy–stress tensor takes the form
of a pure radiation field, which has been already discussed in Subsection
3.4.2. Thus, in this subsection we consider only the non-null case (Tsoubelis
and Wang, 1991). Since the component φ1 defined by Eq. (1.100) is zero
in Regions II and III, from Eqs. (1.99a) and (1.99d) it can be shown that
it must be also zero in Region IV. If we introduce the “scale invariant”
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch03 page 80

80 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

quantities
(0) (0)
Φ0 = BΦ0 , Φ2 = AΦ2 , (3.86)
and drop the superscript “0”, then the non-vanishing Ricci scalars are
ΦIV
00 = Φ0 Φ0 , ΦIV IV
02 = Φ0 Φ2 = Φ20 , ΦIV
22 = Φ2 Φ2 , (3.87)
and the Maxwell field equations (1.99a)–(1.99d) reduce to
2Φ0,u = (U,u − i sinh W V,u )Φ0 − (cosh W V,v − iW,v )Φ2 ,
(3.88)
2Φ2,u = (U,v + i sinh W V,v )Φ2 − (cosh W V,u + iW,u )Φ0 .
Note that in the present case all of the Ricci scalars have only the shock
part, otherwise, the Maxwell potentials Φi (or equivalently, the electromag-
netic field tensor Fμν ) will contain the square roots of δ-function, which is
not acceptable physically. Then, from Eqs. (3.70), (3.87) and (3.88) we find
sh 1
θ0,u =− {3[cosh W V,v Im(ΨIV IV
0 Ψ2 )
IV IV
4(Ψ0 Ψ0 )
+ W,v Re(ΨIV IV IV
0 Ψ2 )] + 2 Im(Ψ0 Φ0,v Φ2 )
− sinh W V,v Re(ΨIV IV
0 Φ0 Φ2 ) + 2M,v Im(Ψ0 Φ0 Φ2 )}. (3.89)
Thus, similar to the last two cases, a plane gravitational shock wave can
change its polarization when it interacts with an electromagnetic field.
On the other hand, for the impulsive part of Ψ0 , Eq. (3.71) becomes

Im Im(ΨIm
0 Φ0 Φ2 )
θ0,u =− . (3.90)
2(ΨIm Im
0 Ψ0 )

It follows that a gravitational impulsive wave can also change its polariza-
tion due to the interaction between the impulsive wave and the electromag-
netic field component ΦIV 02 .

3.4.5. Spacetimes filled with a neutrino field

When a neutrino field is present, in general it requires the spin coefficients
to be different from zero (Griffiths, 1976a, 1976b). Consequently, the metric
for a neutrino field generally is not reducible to the simple form of Eq. (2.22)
(Szekeres, 1972; Griffiths, 1976b; Tsoubelis and Wang, 1991). However,
when one of the spinor components Ψ and Φ are zero [see Eq. (1.102)],
the metric does take the form of Eq. (2.22) (Griffiths, 1976b; Tsoubelis and
Wang, 1991). Then, Eq. (1.105) shows that the only non-vanishing Ricci
scalar is either Φ00 (when Φ = 0) or Φ22 (when Ψ = 0). But this corresponds
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch03 page 81

Colliding Plane Gravitational Waves 81

to the case for a pure radiation field, which we have already discussed in
Subsection 3.4.2.

3.5. Singularities in Spacetimes of Colliding Plane

Gravitational Waves
The study of colliding plane gravitational waves was initially motivated by
Penrose’s conjecture that the focusing effect of single plane waves should
cause the colliding waves to interact strongly and eventually produce space-
time singularities (Penrose, 1965). Later on this conjecture was verified
by several exact solutions (Szekeres, 1970, 1972; Khan and Penrose 1971;
Nutku and Halil, 1977) and various theorems (Tippler, 1980; Centrella and
Matzner, 1982; Matzner and Tippler, 1984). Thus, until 1986 it was believed
that the collision of plane gravitational waves inevitably produced singu-
larities in the future of the interaction. However, it were Chandrasekhar
and Xanthopoulos who first pointed out that such singularities were not
imperative to occur, and, instead, that Killing–Cauchy (or simply Cauchy)
horizons may take place (Chandrasekhar and Xanthopoulos, 1986a). In par-
allel to the Chandrasekhar–Xanthopoulos work, Ferrari and Ibañez (1987a,
1988) found that this was also the case when the two plane gravitational
waves are collinearly polarized. Following Chandrasekhar and Xanthopou-
los, and Ferrari and Ibañez, the studies of singularities in the colliding
plane gravitational wave spacetimes have attracted a great deal of atten-
tion. Yurtsever (1988a) first found that there exists an abundance of exact
colliding collinearly polarized plane gravitational wave solutions, which cre-
ate Cauchy horizons instead of spacetime curvature singularities. It was also
shown that those Cauchy horizons are not stable in the full nonlinear the-
ory against small but “generic” perturbations of the initial data,2 and that
“generic” initial data always produce space-like curvature singularities with-
out Cauchy horizons. Tsoubelis and Wang (1989) and Feinstein and Ibañez
(1989) obtained the same results by considering various solutions. Later,
Yurtsever (1989) generalized his previous studies to cover the non-collinear
case, and found that similar results also hold. Specifically, it was shown that
at a fixed value of the space-like coordinate z [see Eq. (3.41)] the metric for

2 A rigorous proof of the instability for the case in which the initial waves have con-

stant aligned polarizations had to wait until 2005, when Griffiths first showed that such
horizons are indeed unstable with respect to bounded perturbations of the initial waves
(Griffiths, 2005). However, such a proof is still absent for the non-collinear case.
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch03 page 82

82 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

vacuum solutions takes the form of a “generalized inhomogeneous” Kasner

vacuum solution,

ds2 ≈ ε0 (z)dτ 2 − ε1 (z)τ 2p1 (z) dz 2

− ε2 (z)τ 2p2 (z) (dX(z)
2 2
) − ε3 (z)τ 2p3 (z) (dX(z)
3 2
) , (3.91)

as t → 0 in both of the collinear and non-collinear cases, where the expo-

nents Pa (z) satisfy the Kasner relations

p1 (z) + p2 (z) + p3 (z) = p1 (z)2 + p2 (z)2 + p3 (z)2 = 1, (3.92)

where τ is monotonically related to the time-like coordinate t, εμ (z) are reg-

2 3
ular functions at the fixed point z, and X(z) and X(z) are in general asymp-
totically z-dependent linear combinations of the coordinates x2 and x3 . The
exponent p1 (z) is always referred to the z-axis, while p2 (z) and p3 (z) are
2 3
referred to the asymptotic Kasner axes X(z) and X(z) , respectively. More-
over, it was further shown that the asymptotic Kasner nature of the metric
implies the corresponding asymptotic Kasner behavior of the spacetime cur-
vature, which in turn implies that the solutions that are free of curvature
singularities on the “focusing hypersurface” are the ones that their asymp-
totic Kasner form is that of a degenerate Kasner solution. It can be shown
that the latter case always corresponds to

p1 (z) = 0. (3.93)

On the other hand, Tsoubelis and Wang studied the effect of polariza-
tions of colliding plane gravitational waves on the formation and nature
of singularities (Tsoubelis and Wang, 1992). Specifically, it was found that
some astigmatic singularities are turned into anastigmatic singularities, or
vice versa, due to the interaction between different polarization modes of
colliding plane gravitational waves. Moreover, it was also found that all
of the solutions that are free of singularities in the collinear case are so
in the non-collinear case, but inversely not. That is, due to the interac-
tion between different polarization modes some singularities are turned into
Cauchy horizons.
In addition, Chandrasekhar and Xanthopoulos (1987a) studied the
effects of sources on the Cauchy horizons, and found that the coupling
of gravitational waves with an electromagnetic field does not affect in any
essential way the development of the Cauchy horizons, if the polarizations
of the colliding gravitational waves are not parallel. However, if the polar-
izations are parallel, the space-like singularity that occurs in the vacuum is
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch03 page 83

Colliding Plane Gravitational Waves 83

transformed into a horizon followed by a three-dimensional time-like singu-

larity by the merest presence of the electromagnetic field. In addition, the
couplings of the gravitational waves with a “stiff” fluid and null dust affect
the development of Cauchy horizons and singularities in radically different
ways: the “stiff” fluid affects the development decisively in all cases but
qualitatively in the same way, while null dust prevents the development of
Cauchy horizons and allows only the development of space-like singulari-
ties. Tsoubelis and Wang (1991), on the other hand, showed that when the
null dust is confined on null hypersurfaces, the conclusion for the null dust
is quite different from the one given above, and found that the collision
of plane gravitational waves does not necessarily develop a space-like cur-
vature singularity in the interaction region, when coupled with null dust
shells. Except the ones developed after the collision, singularities also form
in the pre-collision regions. The latter are usually attributed to the “back-
reaction”, and have not been well understood so far. Later, Konkowski and
Helliwell (1989, 1991) studied three typical cases and found that in the col-
liding impulsive wave spacetimes and colliding sandwich wave spacetimes
the singularities in the pre-collision regions are quasi-irregular, and that in
the colliding thick gravitational wave spacetimes they are non-scalar cur-
vature singularities. But, they argued that all these singularities are not
stable, and that the presence of matter or matter fields will convert these
singularities into scalar curvature ones.

3.6. Methods for Generating New Solutions

The Einstein field equations with a tow-dimensional group G2 of isometries
acting orthogonally transitively on non-null orbits are nonlinear partial
differential equations in two independent variables (Kramer et al., 1980;
Stephani et al., 2009). When the orbits are time-like, the corresponding
Einstein field equations are elliptical and the spacetimes are stationary
axisymmetric, whereas when the orbits are space-like, the corresponding
Einstein field equations are hyperbolic and the spacetimes are either cylin-
drically or plane symmetric. The plane symmetric case includes cosmolog-
ical models as well as colliding plane gravitational waves.
Since the pioneering work of Geroch (1971, 1972), a fair amount
of labor has been devoted to the above cases, and several methods
have been developed and applied to the construction and analysis of
properties of solutions for these corresponding partial differential equa-
tions (MacCallum, 1984; Stephani et al., 2009). Among these methods
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch03 page 84

84 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

are the Harrison Bäcklund transformations (Harrison, 1978, 1980), the

Hoenselaers–Kinnersley–Xanthopoulos transformations (Hoenselaers Kin-
nersley and Xanthopoulos, 1979a, 1979b; Xanthopoulos, 1979, 1981),
the Neugebauer–Kramer involution (Neugebauer, 1979, 1980a, 1980b;
Kramer and Neugebauer, 1980, 1984), the Hauser–Ernst formulation of the
Riemann–Hilbert problem (Hauser and Ernst, 1979a, 1979b, 1980a, 1980b,
1981; Palenta and Meinel, 2017), the inverse scattering method (or soliton
technique) of Belinsky and Zakharov (Belinsky and Zakharov, 1978, 1979;
Belinsky and Verdaguer, 2001), the Chandrasekhar–Ferrari method (Chan-
drasekhar and Ferrari, 1984), and more recently the initial value problem of
Alekseev and Griffiths for colliding gravitational and electromagnetic waves
(Alekseev and Griffiths, 2001, 2004). See also Griffiths and Santano-Roco
(2002) for the initial value problem of colliding collinearly polarized plane
gravitational waves.
The usefulness of the above-mentioned methods is mostly restricted to
the vacuum case or to the case in which the Ricci tensor vanishes on the
subspace spanned by the Killing vectors (Belinsky, 1979, 1980; Kitching-
ham, 1986).
Progress in the direction of relaxing the above-mentioned restrictions
started with the work of Belinsky and Ruffini (1980). They first extended
the inverse scattering method (ISM) of Belinsky and Zakharov (BZ) to the
five-dimensional vacuum stationary axisymmetric case and showed that,
through the Jordan–Kaluza–Klein dimensional reduction procedure, their
results are relevant to the construction of exact solutions corresponding
to a scalar–vector–tensor theory. Later, following Belinsky and Ruffini’s
approach, Ibañez and Verdaguer (1986) constructed some cosmological
solutions representing solitonic perturbations of the Friedmann–Lemaı́tre–
Robertson–Walker (FLRW) four-dimensional background with an effective
ultra-relativistic equation of state for the matter content. Diaz, Gleiser
and Pullin (1987, 1988) further extended ISM of BZ to a class of perfect
fluid solutions. As an example, they obtained finite perturbations of the
solitonic type for the FLRW flat spacetime with a perfect fluid satisfying
the “gamma” law of Eq. (1.108) for the equation of state. Yet, motivated
by the five-dimensional representation of the Brans–Dicke–Jordan theory,
Bruckman (1986, 1987) extended ISM (BZ) to D-dimensional stationary
axisymmetric spacetimes.
In this section, we concern ourselves only with ISM (BZ) and
Chandrasekhar–Ferrari method in the vacuum case for both their utility
and simplicity. For the extended ISM (BZ), we refer the readers to the
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch03 page 85

Colliding Plane Gravitational Waves 85

above-mentioned papers, the review presented by Verdaguer (1984) and

the monograph of Belinsky and Verdaguer (2001). For other methods, the
original works cited above will be the excellent references to look for.

3.6.1. The Chandrasekhar–Ferrari method

The Chandrasekhar–Ferrari method (Chandrasekhar and Ferrari, 1984)
is based on the reciprocal relations that exist between stationary
axisymmetric spacetimes and plane symmetric spacetimes. In particular,
Chandrasekhar and Ferrari showed that in vacuum the metrics for plane
symmetric spacetimes satisfy the same Ernst equation (3.43) as do the met-
rics for stationary axisymmetric spacetimes. Therefore, a correspondence
exists between solutions of these two classes of spacetimes in the sense that
they can be obtained from the same Ernst potential. By using this corre-
spondence Chandrasekhar and Ferrari showed that the Schwarzschild and
Kerr solutions correspond, respectively, to the Khan–Penrose (1971) and
Nutku–Halil (1977) solutions. Following the same line, Chandrasekhar
and Xanthopoulos have obtained many new solutions for either collid-
ing purely gravitational plane waves (Chandrasekhar and Xanthopoulos,
1986a), or for colliding plane gravitational waves coupled with electromag-
netic field (Chandrasekhar and Xanthopoulos, 1985a, 1987b), or for collid-
ing plane gravitational waves coupled with null dust (Chandrasekhar and
Xanthopoulos, 1986b), or for colliding plane gravitational wave coupled
with a “stiff” fluid (Chandrasekhar and Xanthopoulos, 1985b, 1985c).

3.6.2. Inverse scattering method of Belinsky and Zakharov

When ISM (BZ) is used to obtain new (exact) solutions of the Einstein
vacuum equations with plane symmetry, the metric is written in the form
of Eq. (3.41). The corresponding vacuum equations are given by (Carr and
Verdaguer, 1984; Verdaguer, 1984),

A,t − B,z = 0, (3.94)

and
1 1
(ln f ),t = − + Tr(A2 + B 2 ),
t 4t
(3.95)
1
(ln f ),z = Tr(AB),
2t
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch03 page 86

86 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

where the matrices A, B, and g are defined, respectively, by

eV cosh W − sinh W
g=t , (3.96)
− sinh W e−V cosh W
A ≡ tg,t g −1 , B = tg,z g −1 , (3.97)
where Tr(Y ) denotes the trace of the indicated matrix Y , and g −1 is the
inverse of g.
It can be shown that the integrability conditions for Eq. (3.95) are
automatically satisfied if the matrix g is subject to Eq. (3.94). Thus, once
a solution of g is found, the function f can be obtained by integrating out
Eq. (3.95) directly.
The main idea of ISM (BZ) is to look for a linear eigenvalue problem that
has the nonlinear equations (3.94) as the integrability conditions. Solving
the linear problem will produce, by an appropriate procedure, solutions to
the original nonlinear equations.
The linear problem associating to Eq. (3.94) is
2λt tA + λB
∂t − ∂λ ψ = − 2 ψ,
λ2 − t2 λ − t2
(3.98)
2λ2 tB + λA
∂z − 2 ∂λ ψ = − 2 ψ,
λ − t2 λ − t2
where λ is a complex “spectral” parameter, and ψ = ψ(λ, t, z) is a 2 × 2
matrix which satisfies the “initial condition”
ψ(λ = 0, t, z) = g(t, z). (3.99)
Equations (3.97) and (3.98) consist of a genuine linear differential equation
system for the unknown “wave-function” ψ with Eq. (3.99) as the initial
condition.
Suppose that ψ (0) is a particular solution of Eqs. (3.97)–(3.99) for a
given g (0) (the seed solution). Then, as Belinsky and Zakharov showed, the
ansatz
ψ = Xψ (0) , (3.100)
gives a new solution of ψ, subject to the assumption that the “scattering”
matrix X(λ, t, z) has only single poles. By using the initial condition (3.99),
we in turn obtain a new solution of g.
The key point in using ISM (BZ) is to integrate the differential equation
system (3.98) to get an explicit solution ψ (0) for the given seed g (0) . Then,
the new solutions will be completely determined by the number and the
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch03 page 87

Colliding Plane Gravitational Waves 87

nature (real or complex) of the poles. Each pole is characterized by the

so-called pole trajectory given by

μk = wk − z ± (wk − z)2 − t2 , (k = 1, 2, 3, . . . , N ), (3.101)

where N is the number of poles, wk are arbitrary (real or complex) param-

eters, and μk ’s are the solutions of the equations
2tμk 2μ2k
μk,t = , μk,z = . (3.102)
t − μ2k
2 t2 − μ2k
Then, the new solutions are given by
⎡ ⎤
(0) 2 
N 
N
f = cph f (0) t−N /2 |μk |N +1 ⎣ (μk − μl )−2 ⎦ det(Γkl ),
k=1 l,k=1;k>l

N 
  
  
N
 μk  (0) −1 (j) (k)
gab =   gab − Γjk Na Nb , (a, b = 2, 3), (3.103)
t
k=1 j,k=1

where
3
 (j) (0) (k)
na gab nb
Γjk ≡ ,
μj μk − t2
a,b=2

3
 (k) (0)
n g
Na(j) ≡ b ab
(without summation for the index k), (3.104)
μk
b=2
3
 (k) (k)
n(j)
a ≡ mb Mab (without summation for the index k),
b=2

M (k) ≡ [ψ (0) (μk , t, z)]−1 ,

(0) (k)
where Cph is an arbitrary real constant, and mb ’s are arbitrary real or
(k)
complex parameters. If we start with real poles, then the parameters mb
also have to be real. If we start with a complex pole μk , its complex conju-
gate is also a pole. Thus, complex poles always go in pairs, and in this case
(k)
we can set μk+N/2 = μk . The complex parameters mb will then satisfy
the relations

ma (k+N/2) = m̄(k)
a . (3.105)

It must be noted that the term in the square bracket in the first equation
of Eq. (3.103) should be replaced by unit when N = 1.
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch03 page 88

88 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

From the above general description of ISM (BZ), we can see that a com-
plete solution is obtained only after the system (3.98) is explicitly solved
for a given seed g (0) . However, as noticed by various authors, this in gen-
eral is not trivial. The solutions for Ψ(0) were found for several simple
diagonal seeds (Jantzen, 1980) and for a few non-diagonal seeds (Belinsky
and Francaviglia, 1982; Letelier, 1986, 1989; Kitchingham, 1986; Cespedes
and Verdaguer, 1987). Therefore, some further simplifications are usually
adopted. A common assumption is that the matrices Ψ(0) and g (0) are diag-
onal. With the latter assumption, Economou and Tsoubelis (1989; see also
Letelier, 1986) found that the corresponding expressions of Eqs. (3.103)
and (3.104) can be written in a very simple form in terms of the determi-
nants of four N × N matrices. The Economou–Tsoubelis results come from
the following observations. Integrating the trace of Eq. (3.98) along a pole
trajectory μk , we have

det(ψ (0) ) = 2wk μk (without summation for the index k), (3.106)

which allows us to write the diagonal matrix ψ (0) in the form

ψ (0) (λ, t, z) = diag{ψk , 2wk μk ψk −1 }

(without summation for the index k), (3.107)

where ψk satisfies the equations,

t (0)
(ln ψk ),ζ = (ln g22 ),ζ , (3.108)
t − μk
t (0)
(ln ψk ),ξ = (ln g22 ),ξ , (3.109)
t + μk

with

ζ ≡ t + z, ξ = t − z. (3.110)

Introducing the following quantities:

(0)
μk Qk g22
σk ≡ , sk ≡ σk (without summation for the index k),
t ψk2
sj sk+1
Δjk ≡ (without summation for the indexes j and k),
σj σk − 1
(3.111)
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch03 page 89

Colliding Plane Gravitational Waves 89

where Qk are arbitrary constants, Economou and Tsoubelis found that the
new solutions can be written in the form
N 
 (0)
g22 = |σk | L(−1) L−1
(0) g22 ,
k=1
 

N
g23 = t |σk | LL−1 (0),
k=1


N
(0)
g33 = |σk |−1 |L(+1) L−1 (0)g22 , (3.112)
k=1
⎡ ⎤
(0) 2 
N 
N
f = Cph f (0) L(0) t−N /2
|σk |N ⎣ (σk − σ1 )−2 ⎦
k=1 l,k=1; k>l


N
× |sk |−1 ,
k=1

where
(σ σ )d s s +1
L(d) ≡ det Δ(d) , Δ(d)jk ≡ j σkj σk −1j k
, (d = 0, ±1),
  (3.113)
s
L ≡ det Δ(0) − det Δ(0)jk + σj σj k .

Later, Tsoubelis and Wang found (Wang, 1991a; Tsoubelis and Wang,
1992) that the integration of Eq. (3.108) will become simpler, if we work
in the coordinates η and μ introduced in Section 3.3. Actually, introducing
the quantity Σ(0) via the relation
ψ (0) (λ, t, z) = (2wλ)1/2 diag{Σ(0) , Σ(0)−1 }, (3.114)
it was found that the system (3.98) (not restricted only along the pole
trajectories) reduces to

[(1 − η 2 )∂η + λ∂μ − 2λη∂λ ] ln Σ(0) = 1 − η 2 V (0) ,η ,
(3.115)
[(1 − μ2 )∂μ + λ∂η − 2λμ∂λ ] ln Σ(0) = (1 − μ2 )V (0) ,μ ,
and the “initial condition” of Eq. (3.115) becomes
(0)
Σ(0) (λ = 0, t, z) = eV . (3.116)
The integrability conditions for Eq. (3.115) is the Einstein vacuum field
equation for the seed function V (0) ,
[(1 − η 2 )V (0) ,η ],η − [(1 − μ2 )V (0) ,μ ], μ = 0, (W (0) = 0). (3.117)
 b2530   International Strategic Relations and China’s National Security: World at the Crossroads

This page intentionally left blank

b2530_FM.indd 6 01-Sep-16 [Link] AM

March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch04 page 91

Chapter 4

Collision of Pure Gravitational Plane Waves

In this chapter, we shall present three classes of exact solutions of the

Einstein vacuum field equations, which represent collisions of two purely
gravitational plane waves. Specifically, in Section 4.1 we present a three-
parameter class of diagonal solutions, which represents a variety of models
of collision of two collinearly polarized gravitational plane waves. The main
properties of these solutions are investigated. In Section 4.2, one of the
above two colliding constantly polarized gravitational plane waves is gen-
eralized to a non-collinearly polarized one by adding one soliton into the
above diagonal solutions. The resulting four-parameter class of one-soliton
solutions in general represents the collision of a collinearly polarized and
a non-collinearly polarized gravitational plane waves. The effects of the
polarization of colliding plane gravitational waves on the formation and
nature of singularities in the interacting region are studied. Finally, in Sec-
tion 4.3, a five-parameter class of two-soliton solutions is presented, which
include almost all the known (both diagonal and non-diagonal) solutions of
colliding gravitational plane waves. The reciprocal relations between these
two-soliton solutions and the ones for stationary axisymmetric case are
also investigated. The effects of polarizations of these colliding gravita-
tional waves on the formation and nature of spacetime singularities due
to the mutual focusing of the two colliding purely gravitational waves are
discussed.

4.1. Collisions of Collinearly Polarized Gravitational

Plane Waves
The Einstein vacuum equations for colliding plane gravitational waves
can be obtained from Eqs. (3.8a)–(3.8e) by simply setting all the Ricci

91
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch04 page 92

92 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

scalars zero
Φij = 0, Λ = 0. (4.1)
The conditions for which the two null hypersurfaces u = 0 and v = 0 are
free of matter are
U,u (u = 0, v) = 0 = U,v (u, v = 0). (4.2)
For the special gauge of Eq. (3.40), these conditions together with Eq. (3.38)
imply
1
n, m > . (4.3)
2
In terms of η and μ, the Einstein vacuum equations (Chandrasekhar
and Ferrari, 1984) can be written in the form,
   
Re(Z){[ 1 − η 2 Z,η ],η − [ 1 − μ2 Z,μ ],μ }
2   2
= (1 − η 2 ) (Z,η ) − 1 − μ2 (Z,μ ) , (4.4)
and
μ η 2  
2
Γ,η + 2
Γ,μ = − Z,η Z ,μ + Z ,η Z,μ , (4.5a)
1−μ 1−η (Z + Z) 2

3 1
2ηΓη + 2μΓ,μ = 2
+
1−η 1 − μ2
4   2   2
− 2
[ 1 − η 2 |Z,η | + 1 − μ2 |Z,μ | ],
(Z + Z)
(4.5b)
where
 1/2
1 − η2
f≡ eΓ . (4.6)
η 2 − μ2
Equation (4.4) is equivalent to the Ernst equation (3.43) with the Ernst
potential E defined by Eq. (3.44). In terms of the Ernst potential E,
Eqs. (4.5a)–(4.5b) can be written as
μ η 2  
2
Γ,η + 2
Γ,μ = − 2 E,η E,μ +E,η E,μ , (4.7a)
1−μ 1−η 2
(1 − |E| )
3 1
2ηΓ,η + 2μΓ,μ = 2
+
1−η 1 − μ2
4   2   2
− 2 [ 1 − η 2 |E,η | + 1 − μ2 |E,μ | ].
(1 − |E|2 )
(4.7b)
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch04 page 93

Collision of Pure Gravitational Plane Waves 93

It can be shown that the above equations are equivalent to the ones
given by Eqs. (3.94)–(3.97).
On the other hand, Chandrasekhar and Ferrari (1984) showed that, as
we did in the stationary axisymmetric case, if we introduce a potential Q
for the function q2 in the manner,
1 − μ2 1 − η2
Q,η = q2,μ , Q,μ = q2,η , (4.8)
χ2 χ2
then the complex function Z + defined by

Z + ≡ P + iQ (4.9)

satisfies the same equation (4.4), where

 1/2  1/2 −1
P ≡ 1 − η2 1 − μ2 χ . (4.10)

Thus, in contrast to the stationary axisymmetric case, we can obtain the

Ernst equation either directly from the metric coefficients χ and q2 or from
the potential Q. When the approaching waves have aligned constant polar-
izations, we can globally set

q2 = 0 (or W = 0). (4.11)

Then, the Einstein vacuum equations of Eqs. (4.4), (4.5a) and (4.5a)
reduce to

χ{[(1 − η 2 )χ,η ],η − [(1 − μ2 )χ,μ ],μ }

   
= 1 − η 2 (χ,η )2 − 1 − μ2 (χ,μ )2 , (4.12)

and
μ η χ,η χ,μ
2
Γ,η + 2
Γ,μ = − , (4.13a)
1−μ 1−η χ2
   
3 1 1 − η 2 χ2,η + 1 − μ2 χ2,μ
2ηΓ,η + 2μΓ,μ = + − . (4.13b)
1 − η2 1 − μ2 χ2
In terms of V [see Eq. (3.44)], Eq. (4.12) takes the form
     
1 − η 2 V,η ,η − 1 − μ2 V,μ ,μ = 0. (4.14)

An obvious solution of Eq. (4.14) is given by (Stoyanov, 1979)

  
V1 = a ln 1 − η 2 1 − μ2 , (4.15)

where a is an arbitrary constant.

March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch04 page 94

94 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

Separation of variables, on the other hand, leads to another class of solu-

tions (Chandrasekhar and Xanthopoulos, 1986a; Griffiths, 1987; Tsoubelis
and Wang, 1989),
∞

V2 = [An Pn (η)Pn (μ) + Bn Pn (η)Qn (μ) + Cn Qn (η)Pn (μ)
n=0

+ Dn Qn (η)Qn (μ)], (4.16)

where An , Bn , Cn and Dn are arbitrary constants, and Pn (η) and Qn (η)

are the Legendre functions of the first and second kind, respectively.
Note that Feinstein and Ibañez (1989) gave another expression for
the general solutions of V . However, a complete solution of the Einstein
field equations is obtained only after the integration of Eqs. (4.13a) and
(4.13b) for the function Γ is carried out explicitly. But, for the general
solutions,

V = V1 + V2 , (4.17)

such an integration has not been completely solved yet, and only for some
particular choice of V , the corresponding solutions of Γ have been found
(Griffiths, 1987; Tsoubelis and Wang, 1989; 1992; Li, 1989; Wang, 1991a).
In the following, let us consider the solutions (Tsoubelis and Wang,
1989),
  
V = a ln 1 − η 2 1 − μ2 − 2δ1 Q0 (η)P0 (μ) − 2δ2 P0 (η)Q0 (μ)
   1−η 1−μ
= a ln 1 − η 2 1 − μ2 + δ1 ln + δ2 ln , (4.18)
1+η 1+μ
where δ1 and δ2 are two constants related to the parameters n and m
introduced in Section 3.3.
In order to complete the integration of Eqs. (4.13a) and (4.13b), we
introduce the quantity Σ by
 
η 2 − μ2
Γ = ln + Σ. (4.19)
(1 − η 2 )3/4 (1 − μ2 )1/4

Substituting the above expression into Eqs. (4.13a) and (4.13b), we obtain
μ η
Σ,η + Σ,μ = −V,η V,μ , (4.20a)
1 − μ2 1 − η2
   
2ηΣ,η + 2μΣ,μ = − 1 − η 2 (V,η )2 − 1 − μ2 (V,μ )2 . (4.20b)
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch04 page 95

Collision of Pure Gravitational Plane Waves 95

It can be shown that for the function V given by Eq. (4.18), the system
of Eqs. (4.20a) and (4.20b) has the following solution (Tsoubelis and Wang,
1989):
2 2
Σ = −δ+ ln(η + μ) − δ− ln(η − μ) + (a + δ1 )2 ln(1 − η)
+ (a − δ1 )2 ln(1 + η) + (a + δ2 )2 ln(1 − μ)
+ (a − δ2 )2 ln(1 + μ) + ln C0 , (4.21)
where C0 is an arbitrary constant, and δ± are defined by
δ± ≡ δ1 ± δ2 . (4.22)
To summarize the above results, we obtain the following solutions:
C0
f = 2 2 (1 − η)
b1
(1 − μ)b2 (1 + η)c1 (1 + η)c2 ,
(η + μ) (η − μ)δ−
δ+

  1/2
e−U = 1 − η 2 1 − μ2 ,
(4.23)
  a 1−η
δ1
1−μ
δ2
e = 1 − η 2 1 − μ2
V
,
1+η 1+μ
W = 0,
where
1 1
bA = (a + δA )2 − , cA = (a − δA )2 − , (A = 1, 2). (4.24)
4 4
The solutions of Eq. (4.23) include most of the known diagonal (W = 0)
solutions. For example, when a = 0, we obtain Szekeres’ family of the
colliding plane gravitational wave solutions (Szekeres, 1972), which includes
the Szekeres solution (Szekeres, 1970), in which we have n = m = 2, and the
Khan–Penrose solution (Khan and Penrose, 1971), in which n = m = 1,
where n and m are defined below in Eq. (4.28). For any given a, when
δ1 = 1 and δ2 = 0, we obtain the Ferrari–Ibañez solutions (Ferrari and
Ibañez, 1987a).
In terms of t and z, the function V given by Eq. (4.23) can be written as
1+z+ (−1 − z)2 − t2
V = 2a ln t − δ+ ln
t

1−z+ (1 − z)2 − t2
− δ− ln . (4.25)
t
The last two terms in the right-hand side of Eq. (4.25) correspond to the
soliton structure, while the first term can be thought of as producing a
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch04 page 96

96 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

Fig. 4.1. The soliton structure for the solutions given by Eq. (4.25).

homogeneous expansion and is mainly “responsible” for the development

of the spacetime curvature singularities as t → 0+ (Feinstein and Ibañez,
1989). When δ− = 1, there is a single pole located at the point z = 1
[see Fig. 4.1]. When δ− = 2, there is a double pole located at that point,
and when δ− takes continuous values, there are “generalized” poles located
at that point (Verdaguer, 1984; Feinstein and Charach, 1986; Oliver and
Verdaguer, 1989; Carot and Verdaguer, 1989). Similarly, the δ+ term in
Eq. (4.25) corresponds to the poles located at the point z = −1. The
contribution of the solitonic terms to the divergence of the curvature as
t → 0+ is similar to that of the homogenous term.
The expression of Eq. (4.25) makes the solutions valid in the region
outside of the two light-cones

(1 − z)2 − t2 = 0, (−1 − z)2 − t2 = 0, (4.26)

and Eq. (3.49) further restricts the region of validity of the solutions to that
between the two light-cones [Region IV in Fig. 4.1].
Thus, the Khan–Penrose substitutions of Eq. (3.11) actually extend
these solutions defined in the triangle ABC beyond the lines AC and BC
to Regions I–III, as shown in Fig. 4.1.
On the other hand, combining Eqs. (3.42) and (4.23) we find

e−M = 8nmu2n−1 v 2m−1 f

2 2
un(2−δ+ )−1 v m(2−δ− )−1
= 2 2
X δ+ Y δ−
× (1 − η)b1 (1 − μ)b2 (1 + η)c1 (1 + μ)c2 . (4.27)
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch04 page 97

Collision of Pure Gravitational Plane Waves 97

In writing Eq. (4.27) we have set 8nm2−(δ+ +δ− ) C0 = 1 by using the arbi-
trariness of C0 . Equation (4.27) shows that the application of the Khan–
Penrose substitutions Eq. (3.11) will result in singularity at u = 0 or v = 0,
unless the parameters n and m are chosen so that
1 1
n= 2 , m= 2 . (4.28)
2 − δ+ 2 − δ−
Equations (4.3) and (4.28) are the restrictions on the choice of the soliton
parameters δ1 and δ2 . The solutions which do not satisfy these conditions
are not acceptable for the collision of purely plane gravitational waves.
However, further considerations from the point of view of physics (Tsoubelis
and Wang, 1989) show that the conditions of Eq. (4.3) are too weak and
that for 1/2 < m, n < 1 the extension obtained above cannot guarantee the
integrability of the Weyl scalars corresponding to the waves under collision
(Szekeres, 1972), as it will become clear in the following. Therefore, we
impose the conditions

n, m ≥ 1, (4.29)

so as to make the Weyl scalars integrable (Szekeres, 1972; Tsoubelis and

Wang, 1989). Then, the extended solutions in the pre-collision regions are
given as follows.
In Region I, where u, v < 0, we have [see Eq. (3.22)],

M = U = V = 0, (4.30)

and the corresponding metric takes the form

 2  2
ds2 = 2dudv − dx2 − dx3 , (4.31)

which is flat.
In Region II, where u < 0, and 0 < v < 1, we have
b1 +c2 +(1−n)/2n−1/2
e−M = (1 + v m )b2 +c1 +(1−n)/2n−1/2 (1 − v m ) ,
e−U = 1 − v 2m ,
2a+δ−
(4.32)
eV = (1 + v m )2a−δ− (1 − v m ) ,
W = 0.

Finally, in Region III, where 0 < u < 1, and v < 0, we have

c1 +c2 +(1−m)/2m−1/2 b1 +b2 +(1−m)/2m−1/2
e−M = (1 + un ) (1 − un ) ,
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch04 page 98

98 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

e−U = 1 − u2n ,
2a−δ+ 2a+δ+
eV = (1 + un ) (1 − un ) , (4.33)
W = 0.
Substituting Eq. (4.32) into Eqs. (3.7a)–(3.7c), we find that the only
non-vanishing Weyl scalar is Ψ0 , given by
 
ΨII
0 (v) = −e
2U
{2am2 1 − 4a2 v 4m−2 − 12a2 m2 δ− v 3m−2
− 6am(2m − 1)v 2m−2 − m(m − 1)δ− v m−2 }. (4.34)
Similarly, ψ4III is the only non-vanishing Weyl scalar in Region III, and is
given by
ΨIII
4 (u) = −e
2U
{2an2 (1 − 4a2 )u4n−2 − 12a2 n2 δ+ u3n−2
− 6an(2n − 1)u2n−2 − n(n − 1)δ+ un−2 }. (4.35)
In Region IV, all three of the Weyl scalars Ψ0 , Ψ2 and Ψ4 do not vanish,
but the corresponding expressions are very complicated, so we shall not
present them explicitly here. Anyhow, as far as the behavior of the Weyl
scalars across the hypersurfaces u = 0 and v = 0 is concerned, we only need
to know their limits as u → 0+ and v → 0+ , which are given as follows.
When u → 0+ , we find
ΨIV II
0 (0, v) → Ψ0 (v),
nmδ+ n−1 m−1
ΨIV
2 (0, v) → u v (2av m + δ− ),
X3
(4.36)
n n−2
ΨIV
4 (0, v) → u [(n − 1)δ+ X + 6a(2n − 1)un
X2
+ 3(2n − 1)δ− v m un ],
and when v → 0+ , we have
m m−2
ΨIV
0 (u, 0) → v [(m − 1)δ− Y + 6a(2m − 1)v m
Y2
+ 3(2m − 1)δ+ v m un ],
(4.37)
nmδ− n−1 m−1
ΨIV
2 (u, 0) → u v (2aun + δ+ ),
Y3
ΨIV II
4 (u, 0) → Ψ4 (u).

Combining the above results with Eqs. (3.24) and (3.26), it can be seen
that the Weyl scalars behave as follows when we cross the null hypersurface
separating region A from region B (A → B):
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch04 page 99

Collision of Pure Gravitational Plane Waves 99

(i) IV → II:
Ψ0 is continuous,
Ψ2 = ΨIV
2 H(u), (4.38)
nδ+ n−1
Ψ4 = ΨIV
4 H(u) + u δ(u),
X
(ii) IV → III:
mδ− m−1
Ψ0 = ΨIV
0 H(v) + v δ(v),
Y
Ψ2 = ΨIV
2 H(v), (4.39)
ψ4 is continuous,
(iii) II → I:
Ψ0 = ΨII
0 H(v) + mδ− v
m−1
δ(v),
(4.40)
Ψ2 , Ψ4 are continuous,
(iv) III → I:
Ψ2 , Ψ4 are continuous,
(4.41)
Ψ4 = ΨIII
4 H(u) + nδ+ u
n−1
δ(u).
Equations (4.34) and (4.40) show clearly that the parameter m determines
the type of the wave incident in Region II. Particularly, by observing the
behavior of Ψ0 across the hypersurface v = 0 in the direction II → I, we
can distinguish the following cases.
(a) m = 1. In this case, we have
Ψ0 = 6aH(v) + δ− δ(v). (4.42)
Therefore, when a = 0, the gravitational wave incident in Region II has the
form of an impulsive + shock wave. When a = 0, only the impulsive part
remains.
(b) 1 < m < 2. Then, we have
Ψ0 = m(m − 1)δ− v m−2 H(v), (4.43)
which means that the incoming gravitational wave has an unbounded wave-
front of the form v γ , with γ ∈ (0, 1).
(c) m = 2. Then, we have
Ψ0 = 2δ− H(v), (4.44)
which corresponds to a shock wave.
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch04 page 100

100 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

(d) m > 2. Then, we find that Ψ0 is continuous across this hypersurface.

Thus, the wavefront in this case is smooth.
Replacing v by u and m by n (and, therefore, δ− by δ+ ) in Eqs. (4.42)–
(4.44), we obtain the type of the wave incident in Region III. It is then
obvious that, by properly choosing the values of the parameters a, m, and
n, we can have a variety of situations, for example, an impulsive wave
collides with a shock wave, or an impulsive + shock wave, or a wave with
smooth wavefront, and so on.
Having finished the studies of the behavior of the Weyl scalars across the
hypersurfaces u = 0 and v = 0, we now turn to study the behavior of these
scalars near the hypersurface t = 0 (or equivalently, η = 1, or u2n + v 2m =
1), since this will reveal the singular properties of the spacetimes as the
result of the collision of two pure plane gravitational waves. To this purpose,
let us first return to the expression (3.52), from which we find that the
determinant of the metric is given by
 
det [gμν (η, μ)] = − 1 − η 2 e2Γ
(1 − η)2b1 (1 − μ)2b2 (1 + η)2c1 (1 + μ)2c2
= −C02 2 2 . (4.45)
(η + μ)2δ+ −2 (η − μ)2δ− −2
Thus, as t → 0+ or η → 1− , the metric becomes singular, unless
b1 = 0, (4.46)
which is equivalent to
2 1
(a + δ1 ) = , (4.47)
4
as one can see from Eq. (4.24).
Detailed calculations show that Eq. (4.47) is also the sufficient condition
for which the Weyl scalars remain bounded as t → 0+ . Specifically, when
Eq. (4.47) holds, the non-vanishing Weyl scalars have the following limits:
ΨIV
2 → mn 2
−(b2 +c1 +c2 +2) −2(nc2 +1) −2(mb2 +1)
u v

1 2
× − (a − δ2 ) − 4au2n (δ2 + δ1 v 2m ) ,
4

3mu1−2nv 2m−1 IV
ΨIV
0 → − Ψ2 , (4.48)
2n(a + δ1 )
3nu2n−1v 1−2m IV
ΨIV
4 → − Ψ2 ,
2m(a + δ1 )
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch04 page 101

Collision of Pure Gravitational Plane Waves 101

as t → 0+ . Therefore, when Eq. (4.47) holds, no curvature singularities

develop on the “focusing hypersurface” t = 0, and instead, a Cauchy hori-
zon is developed. Thus, in this case the metric is extendible across the
hypersurface t = 0. However, as pointed out by Yurtsever (1988a, 1989),
the geometry of the spacetime beyond this hypersurface is not uniquely
determined by the initial data posed on the two intersecting characteristic
hypersurfaces u = 0 and v = 0. The spacetime can be smoothly extended
across the Cauchy horizon in infinitely different ways.
In addition, Griffiths (2005) rigorously proved that such Cauchy hori-
zons are not stable, and will be turned into spacetime singularities by gen-
eral bounded perturbations of the initial waves.
From Eq. (4.48) it can be shown that
 2
9 ΨIV
2 → ΨIV IV
0 Ψ4 , (4.49)

as t → 0+ . Then, following the theorem given by Chandrasekhar and Xan-

thopoulos (1986a) that, when the relation
 2
9 ΨIV
2 = ΨIV IV
0 Ψ4

holds, the corresponding solutions are Petrov type D, we find from

Eq. (4.49) that the above solutions become Petrov type D as t → 0+ .
It is worth noting in this regard that the extendible colliding plane gravita-
tional wave models obtained by Chandrasekhar and Xanthopoulos (1986a),
and Ferrari and Ibañez (1988) are also Petrov type D.
On the other hand, if we follow Yurtsever’s approach (1988a, 1989), we
find that the solutions given by Eqs. (4.23) and (4.24) have the limit,
(0) (0) (0)  2 (0)  2
ds2 ≈ ε0 dτ 2 − ε1 τ 2p1 dz 2 − ε2 τ 2p2 dx2 − ε3 τ 2p3 dx3 , (4.50)

as t → 0+ , where
2
+3/4
τ = tα , α = a + δ1 , (4.51)

and
α2 − 14 α − 12 α + 12
p1 ≡ , p2 ≡ − , p3 ≡ , (4.52)
α2 + 34 α2 + 34 α2 + 34
(0)
and εμ are functions of z only. It is easy to show that the pk defined
by Eq. (4.52) satisfy the Kasner relations Eq. (3.92), but now the pk
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch04 page 102

102 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

are independent of the coordinate z. Note that the condition (3.93) is

equivalent to
1
α=± , (4.53)
2
which are exactly the conditions given by Eq. (4.47).
Equations (4.50)–(4.53) show clearly that the homogeneous expansion
parameter a, together with the solitonic parameter δ1 , completely deter-
mines the singularity behavior of the solutions on the focusing hypersurface
t = 0. Specifically, when α < −1/2, p1 and p2 are greater than zero, while p3
is less than zero. Thus, the corresponding singularities are astigmatic (Yurt-
sever, 1988a, 1989). That is, the physical three-values get squashed in the z
and x2 directions and stretched in the x3 direction. When α → ∞, p1 → 1,
p2 → 0+ , and p3 → 0− . Thus, in this case the solutions approach the degen-
erate Kasner solution. When α = −1/2, as shown above, the corresponding
solutions are free of curvature singularities on the focusing hypersurface
t = 0. When −1/2 < a ≤ 0, then we have p1 < 0. Consequently, the corre-
sponding singularities on t = 0 become anastigmatic, which means that the
physical three-values now become stretched in the z direction and squashed
in the x2 and x3 directions. Because of the symmetric dependence of the
exponents pk on a, it is easy to see that the same results can be obtained for
α ≥ 0, if x2 and x3 are exchanged (and, therefore, p2 and p3 are exchanged).
Finally, we note that other solutions, which represent the collisions
of purely collinearly polarized gravitational plane waves, were also found
by several authors (Ferrari and Ibañez, 1987b; Griffiths, 1987; Yurtsever,
1988b; Li, 1989; Tomita, 1998). However, from the point of view of physics,
these solutions share similar physical properties of the ones just considered
above.

4.2. Collisions of Collinearly and Non-collinearly Polarized

Gravitational Plane Waves
Solutions which represent the collision of non-collinearly polarized gravita-
tional plane waves were first found by Nutku and Halil (1977) in generalizing
the Khan–Penrose solution (Khan and Penrose, 1971) to the non-collinear
case. The Nutku–Halil solution represents the collision of two constantly but
not collinearly polarized impulsive gravitational plane waves. Later on, sev-
eral other solutions were found (Chandrasekhar and Xanthopoulos, 1986a;
Ferrari Ibañez and Bruni, 1987a; Ernst, Garcia and Hauser, 1987a; Halilsoy,
1988a,1988b; Tsoubelis and Wang, 1992). Except for the Tsoubelis–Wang
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch04 page 103

Collision of Pure Gravitational Plane Waves 103

solutions, which represent the collision of a collinearly with a non-collinearly

polarized gravitational plane wave, all other solutions mentioned above rep-
resent the collisions of two non-collinearly polarized gravitational plane
waves.
In this section, we shall present a four-parameter class of solutions
obtained by introducing one soliton into the diagonal solutions given by
Eqs. (2.31) and (2.32), which in general represents the collision of two plane
gravitational waves, one of which is collinearly polarized, while the other is
non-collinearly polarized (Tsoubelis and Wang, 1992).

4.2.1. One-soliton solutions

As mentioned in Section 3.6, the crucial point in using the inverse scattering
method of Belinsky and Zakharov is to integrate the system (3.115) for a
given (diagonal) seed solution. But, this in general is not an easy task.
Fortunately, for the seed solutions given by Eqs. (4.23) and (4.24), after a
simple but tedious integration, the following solution can be obtained:
(2λw)a+δ1
Σ(0) (λ, η, μ) = . (4.54)
[(1 + η)(1 + μ) + λ]δ+ [(1 + η)(1 − μ) − λ]δ−
Then, the remaining task is no more than some algebraic calculations. By
choosing the pole trajectory as
μ1 ≡ 1 − z − (1 − z)2 − t2 = (1 − η)(1 + μ), (4.55)
which is equivalent to introducing one soliton (or pole) at the point z = 1,
we find that the one-soliton solutions are given by
  1/2 A (0)
χ = 1 − η 2 1 − μ2 χ ,
B
(η − μ)2δ− +1 (0) 2
q2 = −2q [χ ] , (4.56)
B
   
(1) (1 − η)b1 (1 − μ)b2 (1 + η)c1 (1 + μ)c2
f = Cph 2 2
A,
(η + μ)δ+ (η − μ)(δ− +1)
where
A ≡ (1 − η)2(a+δ1 ) (1 + μ)2(a−δ2 ) + q 2 (η − μ)4δ
− (1 + η)2(a−δ1 ) (1 − μ)2(a+δ2 ) ,
(4.57)
B ≡ (1 − η)2(α+δ1 ) (1 + μ)2(a−δ2 ) (1 + η)(1 − μ)
+ q 2 (η − μ)4δ − (1 + η)2(a−δ1 ) (1 − μ)2(a+δ2 ) (1 − η)(1 + μ),
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch04 page 104

104 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

with

b A ≡ (a + δA ) (a + δA − 1) , c A ≡ (a − δA ) (a − δA − 1), (4.58)
 
where q ≡ 4δ1 −a Q1 is another arbitrary constant.
Note that when η → μ the functions A and B defined by Eq. (4.57)
become unbounded unless δ− ≥ 0. Therefore, for the extension of the above
solutions beyond the hypersurfaces η = ±μ (equivalent to the u = 0 and
v = 0 hypersurfaces), we restrict the solutions of Eqs. (4.56)–(4.58) only to
the cases in which δ− ≥ 0. For δ− < 0, we write these solutions in the form

A (0)
χ = [(1 − η 2 )(1 − η 2 )]1/2 χ ,
B
(η − μ)−2δ− +1 (0) 2
q2 = −2q [χ ] , (4.59)
B
   
(1) (1 − η)b1 (1 − μ)b2 (1 + η)c1 (1 + μ)c2
f = Cph 2 A ,
(η + μ)δ+ (η − μ)(δ− −1)2

where

A ≡ (η − μ)−4δ− A, B  ≡ (η − μ)−4δ− B, (4.60)

and A and B are defined by Eq. (4.57).

When we introduce one pole at the point z = −1 [see Fig. 4.1], we
find that the corresponding solutions are given by Eqs. (4.56)–(4.58) but
replacing μ by −μ and δ2 by −δ2 . Without loss of generality, in the following
we shall consider only the solutions in which δ− ≥ 0.
The considerations given in Section 4.1 show that the proper choice of
the parameter n and m for the solutions given by Eqs. (4.56)–(4.58) is
2 −1
n = (2 − δ+ ) , m = [2 − (δ− + 1)2 ]−1 . (4.61)

The restrictions of Eq. (4.29) now imply that

2
√
1 ≤ δ+ < 2, 0 ≤ δ− < 2 − 1. (4.62)

Inserting the Khan–Penrose substitutions Eq. (3.11) into the above solu-
tions and considering Eq. (3.42), we find that the extended solutions in the
pre-collision regions are given as follows.
In Region I, the metric takes the form of Eq. (4.31), which means that
the spacetime is flat.
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch04 page 105

Collision of Pure Gravitational Plane Waves 105

In Region II, the solutions are given by

2a−δ− +1 A
χ = (1 − v m )2a+δ− +1 (1 + v m ) ,
B
q
q2 = −4 v (2δ− +1)m (1 − v m )2(2a+δ− ) (1 + v m )2(2a−δ− ) ,
B
2a(a+δ− −1)+δ− (δ− −2)/2
(4.63)
e−M = A (1 − v m )
2a(a−δ− −1)+δ− (δ− +2)/2
× (1 + v m ) ,
where
2
A = (1 − v m )2(2a+δ− ) + q  v 4δ− m (1 + v m )2(2a−δ− ) ,
B = (1 − v m )2(2a+δ− ) (1 + v m )2 + q 2 v 4δ− m (1 + v m )2(2a−δ− ) (1 − v m )2 ,
q  ≡ 4δ− q. (4.64)
2
−δ+ −(1−δ− )2 (1)
Note that when writing Eq. (4.64) we had set 8mn2 C ph = 1.
In Region III, we have
χ = (1 − un )2a+δ+ (1 + un )2a−δ+ ,
q2 = 0, (4.65)
2 2
e−M = (1 − un ) 2a(a+δ+ )+(δ+ −1)/2
(1 + un ) 2a(a−δ+ )+(δ+ −1)/2
.
Equations (4.63) and (4.65) show that the extended solutions repre-
sent the collisions of two plane gravitational waves. The one incident in
Region II is in general non-collinearly polarized, while the one in Region III
is collinearly polarized. It is interesting to note that the solutions in Region
III are exactly the ones given by Eq. (4.33) in the collinear case. Thus, the
types that the gravitational plane wave incident in Region III may have are
the same ones as discussed in Section 4.1. The types that the wave incident
in Region II may have are given as follows.
(i) m = 1. Then, we have
6a
ΨII→I
0 = 2 [(1 − 6q 2 + q 4 ) − i4q(1 − q 2 )]H(v)
(1 + q 2 )
1
+ [(1 − q 2 ) − i2q]δ(v). (4.66)
1 + q2
Thus, as it is in the collinear case, the incoming gravitational plane wave in
Region II is the type of impulsive + shock. When a = 0, only the impulsive
part remains.
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch04 page 106

106 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

(ii) 1 < m < 2. In this case, we have

ΨII→I
0 → m(m − 1)(1 + δ− )v m−2
− i2m(1 + 2δ− )(2mδ− + m − 1)v m(2δ− +1)−2 , (4.67)
which means that the incoming wave has unbounded wavefront when across
the hypersurface v = 0.
(iii) m = 2. Then, we have
ΨII→I
0 = 2 (1 + δ− ) H(v), (4.68)
which is a shock wave only.
(iv) m > 2. Then, we find that Ψ0 is continuous across the hypersurface
v = 0. So, the incoming gravitational wave has smooth wavefront in this
case.
Combining the above results with the ones obtained in the collinear
case, we find that the values of the parameters n and m uniquely determine
the type of the incoming gravitational plane waves, no matter whether the
waves are constantly (collinearly) polarized or variably (non-collinearly)
polarized. The freedom for the choice of the values of the parameters n and
m leads the above solutions to represent a variety of models, for example, an
impulsive plane gravitational wave collides with a shock wave, or a shock +
impulsive wave, or a wave with a smooth wavefront, etc.

4.2.2. The nature of singularities formed after collision

To study the nature of the singularities of the solutions presented in the
last subsection on the hypersurface t = 0, we begin with their diagonal
limit. Setting q = 0 in Eqs. (4.56)–(4.58), we find that the corresponding
solutions are given by
 δ1  δ
 2
 
2 a 1−η 1−μ 2
χ = 1−η 1−μ ,
1+η 1+μ
q2 = 0, (4.69)
b b c c
(1 − η) (1 − μ) (1 + η) (1 + η)
1 2 1 2
e−M = 2 2 ,
X δ + + Y δ −
where
 2 1  2 1
bA = (a + δA ) − , cA = (a − δA ) − ,
4 4
(4.70)
1 1
δ1 = δ1 + , δ2 = δ2 − , δ  ± = δ1 ± δ2 .
2 2
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch04 page 107

Collision of Pure Gravitational Plane Waves 107

But, after replacing δ1 and δ2 by δ1 and δ2 , respectively, these are the
seed solutions given by Eqs. (4.23) and (4.24). Thus, the asymptotic behav-
ior of the solutions given by Eq. (4.69) is described by Eqs. (4.50)–(4.52) but
with α being replaced by α ≡ a + δ1 = α + 1/2. Specifically, corresponding
to Eq. (4.53), we now have
α = 0, −1. (4.71)
In the non-collinear case, the solutions given by Eqs. (4.56)–(4.58) have
the following limit as t → 0+ .
(a) α < −1/2. In this case, it can be shown that the above solutions have the
same limit as the corresponding diagonal solutions as t → 0+ . Specifically,
the solutions with α = −1 are free of spacetime curvature singularities on
the hypersurface t = 0 in both of the collinear and non-collinear cases.
(b) − 1/2 ≤ α < 0. In this case, the solutions have the limit
ds2 = ε20 dτ 2 − ε21 τ 2p1 dz 2 − ε22 τ 2p2 (dX(z)
2 2
) − ε23 τ 2p3 (dX(z)
3 2
) , (4.72)
as t → 0+ , where τ and pk are given by Eqs. (4.51) and (4.52) with α
being replaced by α = α + 1/2. Here ε2i ’s are functions of z only, and
2 3
X(z) and X(z) are z-dependent linear combinations of x2 and x3 at each
fixed point of z. Equation (4.72) shows that, relative to the corresponding
diagonal solutions, the Kasner exponents pk are not changed. Thus, in the
both collinear and non-collinear cases, the nature of singularities on the
hypersurface t = 0 is the same, and is anastigmatic. However, the Kasner
axes
 2 along
 which the exponents p2 and p3 are defined are rotated from
x , x3 to (X(z)
2 3
, X(z) ) at each fixed point z.
(c) 0 ≤ α ≤ 1/2. Then, the metric takes the form
ds2 = ε30 dτ 2 − ε31 τ 2p1 dz 2 − ε32 τ 2p2 (dX(z)
2 2
) − ε3 3τ 2p3 (dX(z)
3 2
) , (4.73)
as t → 0+ , but now τ and pk are defined by
2
−α+1 α(α − 1)
τ ≡ tα , p1 ≡ ,
α2 − α + 1
(4.74)
α 1−α
p2 ≡ 2
, p3 ≡ 2 .
α −α+1 α −α+1
The comparison of the above limit with the one in the collinear case shows
that the nature of the singularities on the hypersurface t = 0 is changed
in the present case. In the non-collinear case it is anastigmatic, while in
the collinear case it is astigmatic. Moreover, the Kasner axes in the present
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch04 page 108

108 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

case are also rotated. When α = 0, Eq. (4.74) gives p1 = 0. That is, the
solutions with α = 0 are free of curvature singularities on the hypersurface
t = 0.
(d) α > 1/2. Then, we find
 2  2
ds2 = ε40 dτ 2 − ε41 τ 2p1 dz 2 − ε42 τ 2p2 dx2 − ε43 τ 2p3 dx3 , (4.75)
where τ and pk are given by Eqs. (4.74) and (4.75). Equation (4.75) shows
that in the present case the Kasner axes remain the same as these in the
collinear case, but the Kasner exponents are changed. When 1/2 < α < 1,
we have p1 < 0. Thus, the nature of the singularities is anastigmatic (note
that in the collinear case, it is astigmatic). When α > 1, we have p1 > 0,
which means that the nature of the singularities is astigmatic, as it is in
the collinear case. When α = 1, we have p1 = 0. Thus, the solutions with
α = 1 are free of spacetime curvature singularities on the hypersurface
t = 0, too. The above analysis shows that the non-diagonal solutions given
by Eqs. (4.56)–(4.58) are free of spacetime curvature singularities on the
hypersurface t = 0, if any one of the following conditions holds:
(i) a + δ1 = −1, (ii) a + δ1 = 0, or (iii) a + δ1 = 1. (4.76)
Comparing Eq. (4.71) with Eq. (4.76), we find that the solutions with α = 0
or −1 are free of curvature singularities in both of the collinear and non-
collinear cases, whereas the solutions with α = 1 are free of spacetime
curvature singularities only in the non-collinear case. The above difference
should be obviously attributed to the presence of the ×-polarization mode
of the plane gravitational wave moving toward the right-hand side. In other
words, the interaction between different polarization modes can change the
nature of the singularities on the hypersurface t = 0 and turn some space-
like singularities into Cauchy horizons.

4.2.3. Specific solutions

To illustrate further the features of the solutions presented in the last sub-
section, we consider some special cases corresponding to different values of
the free parameters n, m and a.
Case A: δ1 = δ2 = 1/2: In this case, from Eq. (4.61) we find
n = m = 1. (4.77)
The previous analysis shows that in this case the solutions in general rep-
resent the collision of plane gravitational shock + impulsive waves. Setting
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch04 page 109

Collision of Pure Gravitational Plane Waves 109

δ1 = δ2 = 1/2 in Eqs. (4.56)–(4.58) and considering Eq. (3.42), we find that

the solutions are now given by
A
χ = [(1 − η 2 )(1 − μ2 )]a ,
B
[(1 − η 2 )(1 − μ2 )]2a
q2 = −2q (η − μ), (4.78)
B
2
−M [(1 − η 2 )(1 − μ2 )]a −1/4 A
e = ,
XY (1 + η)2a (1 + μ)2a
where the functions A and B now are defined by
   
A ≡ 1 − η 2 (1 − η)2a (1 + μ)2a + q 2 1 − μ2 (1 + η)2a (1 − μ)2a ,
(4.79)
B ≡ (1 + η)2 (1 − η)2a (1 + μ)2a + q 2 (1 + μ)2 (1 + η)2a (1 − μ)2a .

It can be shown that this subclass of solutions belongs to the Ernst–Garcia–

Hauser solutions (Ernst, Garcia and Hauser, 1987a, 1987b, 1988). To study
these solutions as a whole is very complicated. In the following, we consider
only some representative cases.
Case A.1: a = 0. In this case, we find that the non-vanishing Weyl scalars
are given by
1
Ψ0 (u, v) = ΨIV
0 (uH(u), v)H(v) + [(1 − q 2 ) − i2q]δ(v),
(1 + q 2 )Y
1
Ψ4 (u, v) = ΨIV
4 (u, vH(v))H(u) + [(1 + q 4 )(1 − v 2 H(v))
(1 + q 2 ) DX
(4.80)
+ 2q 2 (1 + 3v 2 H(v)) + 14q(1 − q 2 )v 2 H(v)]δ(u),

Ψ2 (u, v) = ΨIV
2 (u, v)H(u)H(v),

where

ΨII IV
0 (v) = Ψ0 (0, v) = 0, ΨIII IV
4 (u) = Ψ4 (u, 0) = 0,
(4.81)
D(v) ≡ [(1 + q 2 )2 (1 − v 2 H(v))2 + 16q 2 v 2 H(v)]1/2 .

Equations (4.80) and (4.81) show that this model represents the collision
of two impulsive plane gravitational waves. Thus, the corresponding solu-
tion is either the Khan–Penrose solution (Khan and Penrose, 1971) or the
Nutku–Halil solution (Nutku and Halil, 1977). A detained analysis of the
polarizations of these two plane gravitational impulsive waves shows that
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch04 page 110

110 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

it is the Nutku–Halil solution, since the polarization angles for these two
incoming impulsive waves are different, although they are constants.
This solution was studied in detail by Chandrasekhar and Ferrari but
with a different form (Chandrasekhar and Ferrari, 1984). It was found that
the solution is singular on the hypersurface t = 0.
Case A.2: a = 1/2. In this case, the non-vanishing Weyl scalars are
given by
1
Ψ0 (u, v) = ΨIV
0 (uH(u), v)H(v) + [(1 − q 2 ) − i2q]δ(v),
Y (1 + q 2 )
1
Ψ4 (u, v) = ΨIV
4 (u, vH(v))H(u) + {[(1 − vH(v))4
CDX
+ 2q 2 [1 + 2v 2 H(v) − 3v 4 H(v)] + q 4 (1 + vH(v))4 ] (4.82)
+ i4qv 2 H(v)[(1 − v)2 − q 2 (1 + v)2 ]}δ(u),
Ψ2 (u, v) = ΨIV
2 (u, v)H(u)H(v),

where
ΨII IV
0 (v) = Ψ0 (0, v)
 
3 1 − v2
=− 1/2
{[16q 4 − (1 − q 2 )4 + (9 − 14q 2 )v]
(1 + q 2 ) C 2 F
− i2q(1 + q 2 )[(1 + q 2 )(5 + v 2 )v − 2(1 − q 2 )(1 + 2v 2 )]
− [(1 + q 2 )(1 + 6v 2 + v 4 ) − 4(1 − q 2 )(1 + v 2 )v]},
(4.83)
3(1 + u)
ΨIII
4 (u) = ΨN
4 (u0) = ,
(1 − u2 )2
C ≡ (1 − vH(v))2 + q 2 (1 + vH(v))2 ,
D ≡ [C 2 + 16q 2 v 2 H(v)]1/2 ,
F ≡ [(1 + q 2 )(1 + 62 + v 4 ) − 4(1 − q 2 )(1 + v 2 )v]3/2 ,
and
(1 + η)2 (1 + μ)2
ΨIV
2 (u, v) = {(1 − η)3 − q 2 (1 − μ)[2 − (1 − η)(η + μ)
XY A2
+ 2η 2 μ − q 2 (1 − 3μ2 + 2μ3 )]
iq(1 + η)(1 + μ)
+
B[(1 − η)3 (1 + μ) + q 2 (1 − μ)3 (1 + η)]
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch04 page 111

Collision of Pure Gravitational Plane Waves 111

× [(1 − η)6 (1 + η)(1 + μ)(2 + η − 3μ)

+ q 4 (1 − η)(1 − μ)3 (6 − 5η + 2η 2 − 7η 3 − μ − 16ημ + 29η 2 μ
+ 2μ2 − 15ημ2 + 6η 2 μ2 − 5η 3 μ2 + 5η 3 − η 2 μ3 )
− q 6 (1 + η)(1 + μ)(1 − μ)6 (2 − 3η + μ)]}. (4.84)
Equations (4.82) and (4.84) show that in this case the solution represents
the collision of two impulsive + shock waves, one of which is constantly
polarized and the other is variably polarized.
From Eqs. (4.71) and (4.76), we find that this solution is free of space-
time singularities on the hypersurface t = 0 only when q = 0, and when
q = 0 it is always singular. This is shown clearly by Eq. (4.84), from which
we can see that when q = 0 the Weyl scalar ΨIV2 remains finite as η → 1, but
IV
when q = 0, Ψ2 becomes unbounded. Actually, setting q = 0, Eq. (4.84)
becomes
1
ΨIV
2 (q = 0) = . (4.85)
XY (1 − η)
√
Case B: δ1 = δ2 = 6/4. In this case, from Eq. (4.61) we find
n = 2, m = 1. (4.86)
Then, the corresponding solutions are given by
 k  k
1 1−η 1−μ A
χ = [(1 − η 2 )(1 − μ2 )]a+ 2 ,
1+η 1+μ B
2q
q2 = − (η − μ)[(1 − η 2 )(1 − μ2 )]2a [(1 − η)(1 − μ)]2k , (4.87)
B
[(1 − η 2 )(1 − μ2 )]ρ A
e−M = ,
X 3/2 Y [(1 + η)(1 + μ)]4ak
but now the functions A and B are defined as
A ≡ (1 − η 2 )2k [(1 − η)(1 + μ)]2a
+ q 2 (1 − μ2 )2k [(1 + η)(1 − μ)]2a ,
(4.88)
B ≡ (1 − η 2 )2k [(1 − η)(1 + μ)]2a (1 + η)(1 − μ)
+ q 2 (1 − μ2 )2k [(1 + η)(1 − μ)]2a (1 − η)(1 + μ),
where
√
6
k= , ρ ≡ (a + k)(a + k − 1). (4.89)
4
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch04 page 112

112 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

This class of solutions in general represents the collision of a variably polar-

ized shock + impulsive gravitational wave with a constantly polarized shock
wave.
Case B.1: a = 0. In this case, we have
1   
Ψ0 (u, v) = ΨIV
0 (uH(u), v)H(v) + 2
1 − q 2 − i2q δ(v),
Y (1 + q )
Ψ4 (u, v) = ΨIV
4 (u, vH(v))H(u), (4.90)
Ψ2 (u, v) = ΨIV
2 (u, v)H(u)H(v),

and
ΨII IV
0 (v) = Ψ0 (0, v) = 0, ΨIII IV
4 (u) = Ψ4 (u, 0) = 4kY
−4
. (4.91)
Equations (4.90) and (4.91) show that in this case the solution represents
the collision of a constantly polarized shock wave and an impulsive wave.
Comparing it with Case A.1, we find that one of the two incoming impul-
sive gravitational waves in the Nutku–Halil solution now is replaced by a
constantly polarized shock wave.
To study the singularity behavior of the solution on the hypersurface
t = 0, we find that the Weyl scalar ΨIV 2 is given by
4ku
ΨIV
2 (u, v) = {B[(1 − η 2 )4k [2 − (1 − k)η 2 − (1 + k)μ2 ]
XY t2 A2 B
− 6kq 2 (η 2 − μ2 )t4k
− q 4 (1 − μ2 )4k [2 − (1 − k)μ2 − (1 + k)η 2 ]]
− iqt2k (1 − η 2 )4k [k −1 (η 2 − μ2 )(1 + η)(1 − μ)
+ (2 + 3η + 6η 2 + 3η 3 ) − μ(3 + 7η + η 2 − η 3 )
− μ2 (1 + 2η + 5η 2 ) + 2μ3 (1 + η)]}. (4.92)
Obviously, as t → 0+ , the Weyl scalar ΨIV2 becomes unbounded. That is,
similar to the Nutku–Halil solution (Chandrasekhar and Ferrari, 1984), a
spacetime singularity is finally developed on the hypersurface t = 0 due to
the mutual focus of the two colliding gravitational waves.
Case B.2: a + δ1 = 1. In this case, the non-vanishing Weyl scalars are
given by
1
Ψ0 (u, v) = ΨIV
0 (uH(u), v)H(v) + [(1 − q 2 ) − i2q]δ(v),
Y (1 + q 2 )
Ψ4 (u, v) = ΨIV
4 (u, vH(v))H(u), (4.93)
Ψ2 (u, v) = ΨIV
2 (u, v)H(u)H(v),
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch04 page 113

Collision of Pure Gravitational Plane Waves 113

and
1
ΨII IV
0 (v) = Ψ0 (0, v) = {[O(v) − q 2 P (v) − q 4 P (−v)
2 (1 − v 2 ) A2 D
+ q 6 O(−v)] − i2q(1 − v 2 )1+2k [Q(v) − 2q 2 Z(v) − q 4 Q(−v)]},
ΨIII IV
4 (u) = Ψ4 (u, 0) = 4Y
−4
[k + 9au3 + 24a2 ku5 + 2a(4a2 − 1)u6 ],
(4.94)

where

O(x) ≡ (1 + x)12k (1 − x)12 3(4 + 7x + 2x2 )

− 2k[6 + 24x + 19x2 − 16kx(1 + 2x − kx)]},

P (x) ≡ (1 + x)8k+3 (1 − x)4k+7 {3(20 + 121x + 22x2 + 7x3 + 6x4 )
− 2k[(30 + 360x + 113x2 + 24x3 + 57x4 )
− 16kx(11 + 10x + x2 + 6x3 ) + 16k 2 x2 (5 + 3x2 )]}, (4.95)
Q(x) ≡ (1 + x)8k (1 − x)8 {3(8 + 23x + 4x2 − 9x3 )
− 4k[(6 + 36x + 19x2 − 12x3 ) − 4kx(5 + 8x − x2 ) + 16k 2 x2 ]},
Z(x) ≡ x(1 − x2 )4(k+1) [3(41 + 9x2 ) − 16k(15 + 3x2 ) + 16k 2 (7 + x2 )],
  4(1+k) 1/2
D(x) ≡ A2 + 16q 2 x2 1 − x2 .

Thus, unlike the last case, the Nutku–Halil impulsive wave incident in
Region II now is replaced by a variably polarized shock + impulsive wave.
The solution in the present case is free of spacetime singularity on the
hypersurface t = 0 only when q = 0, and otherwise it is always singular.
This can be seen from the Weyl scalar ΨIV2 , which is given by

4u
ΨIV
2 (u, v) = {B[J(η, μ) − 6q 2 K(η, μ) − q 4 J(μ, η)]
XY BA2 (1 + η)(1 + μ)
− i2q(1 + η)2k+1 (1 + μ)2k+1 [L(η, μ) + 2q 2 N (η, μ)
+ q 4 L(μ, η)]}, (4.96)

where

J(x, y) ≡ (1 + y)4 (1 − x)3 (1 + x)8k [(1 + x)(x + y)

+ k(1 − x)(2 + 3x − y) − 2k 2 (1 − x)(x − y)],
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch04 page 114

114 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

K(x, y) ≡ (x − y)(1 − x2 )(1 − y 2 )(1 + x)4k+1 (1 + y)4k+1

× [(x + y) − 2k(1 − k)(1 − x)(x − y)],
L(x, y) ≡ (1 + y)3 (1 − x)3 (1 + x)8k {(1 + 2x − 2y − xy)(x + y)
+ k(1 − x)(1 − y)[(2 + 5x − 3y) − 4k(x + y)]}, (4.97)
N (x, y) ≡ (1 − x2 )(1 − y 2 )(1 + x)4k (1 + y)4k {(x + y)[(1 + xy)2
− 2(x2 + y 2 )] + k(1 − x)(x − y)[(1 + x)(1 + y)(2 − x − y)
− 3(x − y)2 + 4k(x − y)2 ]}.
When q = 0, Eq. (4.96) reduces to
4u(1 − μ)
ΨIV
2 (u, v) = [(1 + η)(η + μ) + k(1 − η)(2 + 3η − μ)
XY t2
− 2k 2 (1 − η)(η − μ)], (4.98)
which becomes unbounded as t → 0+ .

4.2.4. Polarizations of colliding plane gravitational waves

To study the polarizations of colliding plane gravitational waves, in this
subsection we consider only two special cases of the solutions presented in
the last subsection.
Case 1: n = m = 1, a = 0. This is the case discussed in Subsection
4.2.3. The non-vanishing Weyl scalars are given by Eq. (4.80). For the
sake of simplicity, we consider only the impulsive part. Thus, according to
Eqs. (2.38) and (2.46) we have
1 2q
ϕIm
0 = tan−1 ,
2 1 − q2
    (4.99)
1 4q 1 − q 2 v 2
ϕIm
4 = tan−1 H(v).
2 (1 + q 4 ) (1 − v 2 ) + 2q 2 (1 + 3v 2 )
On the other hand, from Eq. (4.78) we find that on the hypersurface
v = 0 we have q2 = 0 (or equivalently, W = 0). Consequently, Eq. (3.61)
yields,
Im(0)
ϕ0 = 0. (4.100)
Thus, we have
Im(0) 1 2q
θ0Im = ϕIm
0 − ϕ0 = tan−1 , (4.101)
2 1 − q2
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch04 page 115

Collision of Pure Gravitational Plane Waves 115

which means that the polarization of the impulsive plane gravitational wave
with support on the hypersurface v = 0 does not change even after the
collision.
Substituting Eqs. (4.43) and (4.78) into Eq. (3.59), and then integrating
it, for v > 0 we find,
    
2
Im(0) 1 −1 1 + q v − 1 − q2
ϕ4 = tan
2 2q
    
2 2
1 + q v + 1 − q
− tan−1 + C0 . (4.102)
2q
Im(0)
From the condition that when v → 0, ϕ4 = 0, we have
 
4q 1 − q 2
tan C0 = 2. (4.103)
8 − (3 − q 2 )
Thus, Eqs. (4.102) and (4.103) yield,
   
Im(0) 1 −1 4q 1 − q 2 v 2
ϕ4 = tan H(v). (4.104)
2 (1 + q 4 ) (1 − v 2 ) + 2q 2 (1 + 3v 2 )

Equations (3.60), (4.99) and (4.104), on the other hand, give

θ4Im = 0. (4.105)
This means that the impulsive wave with support on the hypersurface u = 0
has zero polarization angle relative to the parallelly transported basis even
after the collision. Thus, the solution given by Eqs. (4.78) and (4.79) with
a = 0 represents the collision of two impulsive plane gravitational waves
with different polarization angles. This must be the Nutku–Halil solution.
Actually, if we make a rotation in the (x2 , x3 )-plane with the angle given by
1  Im  1 2q
ϕ≡ θ0 − θ4Im = tan−1 , (4.106)
2 4 1 − q2
we find that the corresponding solution takes the exact form of the Nutku–
Halil solution used by Chandrasekhar and Ferrari (1984).
Case 2: n = m = 1, a = 1/2. This case was also considered in the
last subsection. The non-vanishing Weyl scalars are given by Eq. (4.82).
Comparing Eq. (4.80) with Eq. (4.82) we find that the impulsive part of
Ψ0 in the above two cases are the same. From Eq. (4.78) we find that the
impulsive part of Ψ0 has the same polarization angle θ4Im as the one given
in the last case.
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch04 page 116

116 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

Integrating Eq. (3.59) along the hypersurface u = 0, on the other hand,

we find
   
Im(0) 1 −1 4qv 2 (1 − v)2 − q 2 (1 + v)2
ϕ4 = tan H(v).
2 (1 − v)4 + 2q 2 (1 + 2v 2 − 3v 4 ) + q 4 (1 + v)4
(4.107)
Then, Eqs. (2.38), (4.59), (4.82) and (4.107) give
θ4Im = 0. (4.108)
Thus, similar to the last case, the polarization angle for the impulsive part
of Ψ4 is zero. That is, the impulsive part of the Ψ4 wave does not change
its polarization direction when it collides and interacts with the oppositely
moving plane gravitational wave. The above results are consistent with our
general conclusions obtained in Section 3.4, from there we can see that the
reason that the impulsive part of colliding plane gravitational wave does
not change its polarization direction is that in the present case this part
does not interact with any of the others [sec Eq. (3.69)].

4.3. Collisions of Two Non-collinearly Polarized

Gravitational Plane Waves
In the last section, by adding one soliton into the diagonal solutions dis-
cussed in Section 4.1, a four-parameter class of solutions was obtained,
which in general represents the collision of two plane gravitational waves,
one of which is constantly polarized, while the other is variably polarized.
A natural generalization of the above solutions is to add one more soliton
into the seed solutions.

4.3.1. Two-soliton solutions

By choosing the trajectories of the two poles as
μ1 = 1 − z − [(1 − z)2 − t2 ]1/2 = (1 − η)(1 + μ),
(4.109)
μ2 = −1 − z − [(−1 − z)2 − t2 ]1/2 = −(1 − η)(1 − μ),
we find that the two-soliton solutions with the seed given by Eqs. (4.23)
and (4.24) are given by (Wang, 1991a)
A (0) C
χ= χ , q2 = (χ(0) )2 ,
B B
(4.110)
(1 − η)ρ1 (1 − μ)ρ2 (1 + η)r1 (1 + μ)r2
f = C0 2 A,
(η + μ)(δ+ +1)2 (η − μ)(δ− +1)
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch04 page 117

Collision of Pure Gravitational Plane Waves 117

where C0 is a constant,
δ1 δ2
1−η 1−μ
χ(0) = [(1 − η 2 )(1 − μ2 )]a , (4.111)
1+η 1+μ
and
A ≡ (1 − η 2 )(1 − μ)2(a+δ2 ) (1 + μ)2(a−δ2 ) [(1 − η)2(a+δ1 )
+ Q1 Q2 (η − μ)2δ − (η + μ)2δ + (1 + η)2(a−δ1 ) ]2
+ (1 − μ2 )(1 − η)2(a+δ1 ) (1 + η)2(a−δ1 ) [Q1 (η + μ)2δ + (1 + μ)2(a−δ2 )
− Q2 (η − μ)2δ − (1 − μ)2(a+δ2 ) ]2 ,
B ≡ (1 − μ)2(a+δ2 ) (1 + μ)2(a−δ2 ) [(1 + η)(1 − η)2(α+δ1 )
− Q1 Q2 (1 − η)(η − μ)2δ − (η + μ)2δ+ (1 + η)2(a−δ1 ) ]2
(4.112)
+ (1 − η)2(a+δ1 ) (1 + η)2(a−δ1 ) [Q1 (1 − μ)(η + μ)2δ + (1 + μ)2(a−δ2 )
+ Q2 (1 + μ)(η − μ)2δ− (1 − μ)2(a+δ2 ) ]2 ,
C ≡ −2{Q1 (η + μ)2δ+ +1 [(1 + μ)2(a−δ2 ) (1 − η)2(a+δ1 )
+ Q22 (η − μ)4δ− (1 − μ)2(a+δ2 ) (1 + η)2(a−δ1 ) ]
+ Q2 (η − μ)2δ− +1 [(1 − μ)2(a+δ2 ) (1 − η)2(a+δ1 )
+ Q21 (η + μ)4δ+ (1 + μ)2(α−δ2 ) (1 + η)2(a−δ1 ) ]},
where Q1 and Q2 are arbitrary constants, and
1 1
ρA ≡ (a + δA )(a + δA − 2) − , rA ≡ (a − δA )(a − δA − 2) − ,
4 4
(A = 1, 2). (4.113)
Note that when η → ±μ the functions A, B and C defined by Eq. (4.112)
become unbounded unless δ± ≥ 0. Therefore, for the extension of these solu-
tions beyond the hypersurfaces η = ±μ, we restrict the solutions given in
the form of Eqs. (4.110)–(4.111) only to the cases where δ± ≥ 0. Otherwise,
we write these solutions in the following different forms.
(i) δ+ > 0, δ− < 0: In this case, the solutions given by Eqs. (4.110)–(4.111)
are written in the form
A(1) (0) C(1) (0) 2
χ= χ , q2 = (χ ) ,
B(1) B(1)
(4.114)
(1 − η)ρ1 (1 − μ)ρ2 (1 + η)T1 (1 + μ)12
f = C0 A(1) ,
(η + μ)(δ,+1)2 (η − μ)(δ−1)2
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch04 page 118

118 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

where
A(1) ≡ (η − μ)−4δ− A, B(1) ≡ (η − μ)−4δ− B,
C(1) ≡ (η − μ)−4δ− C,
(4.115)
and the functions A, B and C are given by Eq. (4.112).
(ii) δ+ < 0, δ− > 0: Then, in this case we write these solutions in the form,
A(2) (0) C(2) (0) 2
χ= χ , q2 = (χ ) ,
B(2) B(2)
(4.116)
(1 − η)ρ1 (1 − μ)ρ2 (1 + η)r1 (1 + μ)r2
f = C0 A(2) ,
(η + μ)(δ+ −1)2 (η − μ)(δ− +1)2
where the functions A(2) , B(2) and C(2) are related to the functions A, B
and C given by Eq. (4.112) via the relations,
A(2) = (η + μ)−4δ+ A, B(2) = (η + μ)−4δ+ B, C(2) = (η + μ)−4δ+ C.
(4.117)
(iii) δ± < 0: In this case, we have
A(3) (0) C(3) (0) 2
χ= χ , q2 = (χ ) ,
B(3) B(3)
(4.118)
(1 − η)ρ1 (1 − μ)ρ2 (1 + η)r1 (1 + μ)r2
f = C0 A(3) ,
(η + μ)(δ+ −1)2 (η − μ)(δ− −1)2
where
A(3) ≡ (η + μ)−4δ+ (η − μ)−4δ− A, B(3) ≡ (η + μ)−4δ+ (η − μ)−4δ− B,
(4.119)
C(3) ≡ (η + μ)−4δ+ (η − μ)−4δ− C.
Equations (4.110)–(4.119) show that in all the above cases, the function
f takes the form
f˜(η, μ)
f= , (4.120)
(η + μ)α+ (η − μ)α−
2 2
for example, in Eq. (4.110) we have α+ = (δ+ + 1) and α− = (δ− + 1) ,
etc.
The analysis carried out in the last two sections shows that for the
function f having the form of Eq. (4.120) the parameters n and m must be
chosen so that
n = (2 − α+ )−1 , m = (2 − α− )−1 . (4.121)
Without loss of generality, in the following we consider the above solutions
only for the cases in which δ± ≥ 0. The above solutions in the other cases
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch04 page 119

Collision of Pure Gravitational Plane Waves 119

will have similar physical interpretations and properties. For more details,
we refer the readers to Wang (1991a).
When δ± ≥ 0, Eq. (4.121) becomes
n = [2 − (δ+ + 1)2 ]−1 , m = [2 − (δ− + 1)2 ]−1 . (4.122)
As usual, we take the solutions given by Eqs. (4.110)–(4.111) as valid
only in Region IV, then extend them back to the pre-collision regions by
means of the Khan–Penrose substitutions (3.11). Such extended solutions
represent the collisions of purely gravitational waves only in the cases where
n, m ≥ 1, which in the present case implies that
√
0 ≤ δ± < 2 − 1. (4.123)
From the discussion given in the last two sections, we can see that the
extended two-soliton solutions represent a variety of models of the collision
of two pure gravitational plane waves. For example, when n = 1 and m = 2
the solutions represent the collision of a variably polarized shock and a
variably polarized shock + impulsive wave, and so on.
Comparing with the seed solutions, we can see that in the two-soliton
case both of the two constantly (collinearly) polarized gravitational plane
waves of the seed solutions are generalized to variably (non-collinearly)
polarized ones, whereas in the one-soliton case only one of them is gen-
eralized to a variably polarized wave, while the other is still constantly
polarized.

4.3.2. Formation and nature of spacetime singularities

To study the behavior of the spacetime singularities of the above solutions
near the focusing hypersurface t = 0, we start with their collinear limit,
as we did in the one-soliton case. Setting Q1 = Q2 = 0, we find that the
solutions given by Eqs. (4.110)–(4.111) reduce to
δ1 +1
1−η   a 1−η 1−μ
δ2
χ= χ(0) = 1 − η2 1 − μ2 ,
1+η 1+η 1+μ
q2 = 0, (4.124)
b1 b2 ci c2
(1 − η) (1 − μ) (1 + η) (1 + μ)
f = C0 2 2 ,
(η + μ)(δ+ +1) (η − μ)(δ− +1)
where
2 1 2 1
b1 = (a + δ1 + 1) − , b2 = (a + δ2 ) − ,
4 4
(4.125)
2 1 2 1
c1 = (a − δ1 − 1) − , c2 = (a − δ2 ) − .
4 4
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch04 page 120

120 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

But, these are the solutions given by Eqs. (4.23) and (4.24), if we replace
δ1 by δ1 + 1. Thus, replacing α by α = a + δ1 + 1 in Eqs. (4.50)–(4.52), we
obtain the asymptotic behavior of the solutions given by Eqs. (4.124) and
(4.125). Specifically, the conditions of Eq. (4.53) now become

2 1
(a + δ1 + 1) = , (4.126)
4
which is equivalent to
1 3
α=− , or α=− . (4.127)
2 2
On the other hand, in the non-collinear case we have Q1 Q2 = 0, which
allows us to distinguish the following cases.

(α) α < −1. Then, the solutions given by Eqs. (4.110)–(4.111) have the
same limit as their diagonal solutions given by (4.124) as t → 0+ . The
solutions with α = −3/2 are especially free of spacetime singularities on
the hypersurface t = 0 in both collinear and non-collinear cases. When
α < −3/2, we have p1 > 0. That is, the corresponding singularities are
astigmatic. When −3/2 < α < −1, we have p1 < 0, and the corresponding
singularities are anastigmatic.

(β) −1 ≤ α ≤ −1/2. In this case, the solutions have the limit

(2) (2) (2)
ds2 ε0 dτ 2 − ε1 τ 2p1 dz 2 − ε2 τ 2p2 (dX(z)
2 2
)
(2)
− ε3 τ 2p3 (dX(z)
3 2
) , (4.128)

as t → 0+ , where τ and pi are the same as they are defined in the corre-
sponding diagonal case. Equation (4.128) shows that the Kasner exponents
pi and the time τ are not changed relative to the corresponding diagonal
case, but the Kasner axes along which the p2 and p3 are defined are rotated
from (x2 , x3 ) to (X(z)
2 3
, X(z) ). Since in the present case p1 < 0, the nature
of the singularities on the hypersurface t = 0 is anastigmatic, as it is in the
collinear case. When α = −1/2, we have p1 = 0. Hence, the solutions with
α = −1/2 are free of spacetime curvature singularities on t = 0.

(γ) −1/2 < α < 1/2. In this case, the solutions have the limit
(3) (3)
ds2 = ε0 dτ 2 − ε1 τ 2p1 dz 2 − ε(3) 2τ 2p2 (dX(z)
2 2
)
(3)
− ε3 τ 2p3 (dX(z)
3 2
) , (4.129)
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch04 page 121

Collision of Pure Gravitational Plane Waves 121

as t → 0+ , with τ and pi given exactly by Eqs. (4.50) and (4.51). Recall that
in the collinear case these solutions have the limit of Eqs. (4.49) and (4.51).
But, now the same limit is obtained with the replacement of α by α =
α + δ1 + 1. Thus, we find that in the present case the interaction between
different polarization modes of the two colliding plane gravitational waves
changes both the Kasner exponents and axes. The nature of singularities
in the present case is anastigmatic, whereas it is astigmatic in the collinear
case.

(δ) 1/2 ≤ α ≤ 1. Then, the solutions have the limit

(4) (4)  (4) 
ds2 ε0 dτ 2 − ε1 τ 2p 1dz 2 − ε2 τ 2p2 (dX(z)
2 2
)
(4) 
− ε3 τ 2p3 (dX(z)
3 2
) , (4.130)

as t → 0+ , where
2
τ  ≡ t(α−1) +3/4
, (4.131)

and
(α − 1)2 − 14 α − 12 3
−α
p1 ≡ , p2 ≡ , p3 ≡ 2
. (4.132)
(α − 1)2 + 34 (α − 1)2 + 34 (α − 1)2 + 34

It is easy to show that the exponents pi defined by Eq. (4.132) also satisfy
the Kasner relations (3.92).
As in the last case, the Kasner exponents as well as the Kasner axes are
all changed. In contrast to the collinear case, the nature of singularities on
t = 0 is anastigmatic. When α = 1/2, we have p1 = 0. Thus, the solutions
with α = 1/2 are free of spacetime singularities on the hypersurface t = 0
in the non-collinear case.

() α > 1. Then, the metric takes the form

 
(4) (4) (4)
ds2 ε0 dτ 2 − ε1 τ12p dz 2 − ε2 τ 2p2 (dx2 )2
(4) 
− ε3 τ 2p3 (dx3 )2 , (4.133)

as t → 0+ , where τ  and pi are defined by Eqs. (4.131) and (4.132), respec-

tively. Comparing this case with the last one, we find that in the present case
only the Kasner exponents are changed, and that the Kasner axes remain
unchanged relative to those in the collinear case. When α = 3/2, we have
p1 = 0. That is, the solutions with α = 3/2 are also free of spacetime
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch04 page 122

122 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

curvature singularities on the hypersurface t = 0. The nature of the singu-

larities is anastigmatic for 1 < α < 3/2 and astigmatic for α > 3/2, whereas
in the collinear case it is astigmatic for all the case with α > 1.
To summarize the above results, we conclude that, if any of the following
conditions holds:

3 1 1 3
(i) α = − , (ii) α = − , (iii) α = , or (iv) α = , (4.134)
2 2 2 2

then the corresponding solutions are free of spacetime curvature singular-

ities on the hypersurface t = 0. Therefore, these solutions are extendible
across this surface. On the other hand, comparing Eqs. (4.134) and (4.127),
we find that the solutions with α = −1/2 or −3/2 are free of spacetime sin-
gularities on t = 0 in both of the collinear (Q1 = Q2 = 0) and non-collinear
(Q1 Q2 = 0) cases. But, the solutions with α = 1/2 or 3/2 are free of space-
time singularities only in the non-collinear case. For the same reasons as
those given in the one-soliton case, this is attributed to the interaction
between different polarization modes.
Note that if we set the parameter Q1 equal to zero, we shall rediscover
the one-soliton solutions given by Eqs. (4.56)–(4.58) but with δ1,2 replaced
by δ1,2 + 1/2. Taking this fact into account, we find that the solutions with
α = −3/2 are free of spacetime curvature singularities on the hypersurface
t = 0 only for the case where Q1 Q2 = 0, otherwise they will be all singular
[see Eqs. (4.53), (4.76) and (4.134)].

4.3.3. Particular solutions

Let us now turn to some representative cases of the solutions given by
Eqs. (4.110)–(4.111).

Case A: δ1 = δ2 = 0. In this case, we find that the solutions given by

Eqs. (4.110)–(4.111) reduce to

  a A
χ= 1 − η 2 1 − μ2 ,
B
  2a C
q2 = 1 − η 2 1 − μ2 , (4.135)
B

[(1 − η 2 )(1 − μ2 )]a(a−2)−1/4

f = C0 A,
η 2 − μ2
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch04 page 123

Collision of Pure Gravitational Plane Waves 123

where the functions A, B and C are now defined by

A ≡ (1 − η 2 )(1 − μ2 )2a [(1 − η)2a + Q1 Q2 (1 + η)2a ]2

+ (1 − μ2 )(1 − η 2 )2a [Q1 (1 + μ)2a − Q2 (1 − μ)2a ]2 ,
B ≡ (1 − μ2 )2a [(1 + η)(1 − η)2a − Q1 Q2 (1 − η)(1 + η)2a ]2
(4.136)
+ (1 − η 2 )2a [Q1 (1 − μ)(1 + μ)2a + Q2 (1 + μ)(1 − μ)2a ]2 ,
C ≡ −2{Q1(η + μ)[(1 + μ)2a (1 − η)2a + Q22 (1 − μ)2a (1 + η)2a ]
+ Q2 (η − μ)[(1 − μ)2a (1 − η)2a + Q21 (1 + μ)2a (1 + η)2a ]}.

But, these are the solutions found by Ernst et al. (1987a, 1987b, 1988) by
using a different method. When δ1 = δ2 = 0, Eq. (4.122) yields n = m = 1.
Hence, from the previous analysis we can see that this subclass of solutions
in general represents the collision of variably polarized gravitational shock +
impulsive plane waves.

Case A.1: a = 0. If we further set a = 0, we can write the corresponding

solutions in a simple form, by introducing the quantities p, q and l via the
relations,
p−1 l−q l+q
Q1 Q2 = , Q1 = , Q2 = , (4.137)
p+1 p+1 p+1
from which we obtain

p2 + q 2 = 1 + l 2 . (4.138)

Substituting Eq. (4.137) into Eqs. (4.135) and (4.136), and considering the
fact that the parameter a vanishes in this case, we find that the correspond-
ing solution takes the form,
   
p2 1 − η 2 + q 2 1 − μ2
χ= ,
(1 − pη)2 + (l − qμ)2
2(qμ − lpη)
q2 = , (4.139)
(1 − pη)2 + (l − qμ)2
   
 p2 1 − η 2 − q 2 1 − μ2
f = C0 1/4
,
[(1 − η 2 ) (1 − μ2 )] (η 2 − μ2 )
 
where C0 = 4C0 (1 − p)−2 is another arbitrary constant. The solution
given by Eq. (4.139) is the Nutku–Halil solution with the NUT parame-
ter l (Ernst, Garcia and Hauser, 1987b, 1988). Actually, it can be shown
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch04 page 124

124 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

that the corresponding Ernst potential E introduced in Eq. (3.44) is now

given by

1
E= [(pη + lqμ) + i(qμ − lpη)]. (4.140)
1 + l2

Thus, Eq. (4.140) reduces to

E = pη + iqμ, (4.141)

when l = 0, which is the Ernst potential for the Nutku–Halil solution

(Chandrasekhar and Ferrari, 1984).
It is interesting to note that the Kerr stationary axisymmetric solution
(1963) also follows the same Ernst potential E given by Eq. (4.141) (Ernst,
1968a).

Case A.2: a = 1/2. Then, we find that the corresponding solution can be
written in the form

(1 − pη)2 + (l − qμ)2
χ= t,
p2
(1 − η 2 ) + q 2 (1 − μ2 )
   
p(q − lμ) 1 − η 2 − q(p − η) 1 − μ2
q2 = −2 ,
p2 (1 − η 2 ) + q 2 (1 − μ2 )
(4.142)
(1 − pη)2 + (l − qμ)2
f = C 0 ,
(η 2 − μ2 )

where C0 , p, q and l are now defined by (Economou and Tsoubelis, 1989),

Q2 q+l 1 1+p 1−p

= , = , Q2 = ,
Q1 q−l Q1 q−l q−l
(4.143)
4C0
C 0 = .
Q12 (q − l)2

It is easy to prove that p, q and l defined by Eq. (4.143) also satisfy the rela-
tion (4.138). As first noticed by Ernst, Garcia and Hauser (1987b, 1988), the
above solution is the Chandrasekhar–Xanthopoulos solution with the NUT
parameter l. In fact, setting l = 0, we shall rediscover the Chandrasekhar–
Xanthopoulos solution (Chandrasekhar and Xanthopoulos, 1986a). In this
case, the corresponding stationary axisymmetric solution is the Kerr–NUT
solution (Kramer et al., 1980).
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch04 page 125

Collision of Pure Gravitational Plane Waves 125

When q = 0, defining p and q  as

p = −p−1 , q  = −lp−1 , (q = 0), (4.144)

we find that Eq. (4.142) reads

1 + 2p η + η 2
χ= t, q2 = 2q  μ,
1 − η2
(4.145)
 2
  1 + 2p η + η ,
f =C (q = 0),
0
η 2 − μ2

where

 0 = 4C0 ,
C p2 + q 2 = 1, (4.146)
Q21 q 2

which is the Ferrari–Ibañez solution (Ferrari and Ibañez, 1988). Thus, the
Ferrari–Ibañez solution corresponds to the Taub–NUT solution (Hawking
and Ellis, 1973).

Case B: Q1 = 0 (or Q2 = 0). Setting Q1 = 0 in Eqs. (4.110)–(4.111),

we find that the reduced solutions are the one-soliton solutions given by
Eqs. (4.56)–(4.58) with the soliton parameters δ1,2 replaced by δ1,2 + 1/2.
Similarly, if we set Q2 = 0, instead of Q1 = 0, in Eqs. (4.110)–(4.111)
we find that the resulting solutions are also the one-soliton solutions, but
now with the soliton parameters δ1,2 replaced by δ1,2 + 1/2 and μ by −μ.

Case C: Q1 → ∞ (or Q2 → ∞). If we set

C0
C0 = , (4.147)
Q21

and then take the limit Q1 → ∞ (but keep C 0 finite), we find that the
solutions given by Eqs. (4.110)–(4.111) become

A (0)
χ= χ ,
B
(η − μ)2δ− +1  (0) 2
q2 = −2Q2 χ , (4.148)
B
 
(1 − η)ρ1 (1 − μ)ρ2 (1 + η)r1 (1 + μ)r2
f = C0 2 2 A,
(η + μ)(δ+ −1) (η − μ)(δ− +1)
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch04 page 126

126 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

but now the functions A and B are defined by

 
A ≡ 1 − μ2 (1 − η)2(a+δ1 ) (1 + μ)2(a−δ2 )
 
+ Q22 1 − η 2 (η − μ)4δ− (1 + η)2(a−δ1 ) (1 − μ)2(a+δ2 ) ,
(4.149)
B ≡ (1 − μ)2 (1 − η)2(a+δ1 ) (1 + μ)2(a−δ2 )

+ Q22 (1 − η)2 (η − μ)4δ− (1 + η)2(a−δ1 ) (1 − μ)2(a+δ2 ) ,


and rA are given by

 2 1
rA = (a − δA ) − , (A = 1, 2). (4.150)
4
Equations (4.120), (4.121) and (4.148) show that the appropriate choice
of the parameters n and m now are

n = [2 − (δ+ − 1)2 ]−1 , m = [2 − (δ− + 1)2 ]−1 . (4.151)

Following the discussion given in Section 4.2, it can be shown that this
subclass of solutions represents the same type of collision as the one-soliton
solutions given by Eqs. (4.56)–(4.58) (Tsoubelis and Wang, 1992).
In a similar fashion, we find that in the limit Q2 → ∞, the two-soliton
solutions become
A (0)
χ= χ ,
B
(η + μ)2δ+ +1  (0) 2
q2 = −2Q1 χ , (4.152)
B
 
(1 − η)ρ1 (1 − μ)ρ2 (1 + η)r1 (1 + μ)r2
f = C0 2 2 A,
(η + μ)(δ+ +1) (η − μ)(δ− −1)
where A, B and C are defined by
 
A ≡ 1 − μ2 (1 − η)2(a+δ1 ) (1 − μ)2(a+δ2 )
 
+ Q21 1 − η 2 (η + μ)4δ+ (1 + η)2(a−δ1 ) (1 + μ)2(a−δ2 ) ,
(4.153)
B ≡ (1 − μ)2 (1 − η)2(a+δ1 ) (1 − μ)2(a+δ2 )
+ Q21 (1 − η)2 (η + μ)4δ + (1 + η)2(a−δ1 ) (1 + μ)2(a−δ2 ) ,

and C 0 , r 1 and ρ2 are given by

1 1
C 0 ≡ C0 Q21 , r1 = (a − δ1 )2 − , ρ2 = (a + δ1 )2 − . (4.154)
4 4
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch04 page 127

Collision of Pure Gravitational Plane Waves 127

In this case, the appropriate choice of the parameters n and m is

n = [2 − (δ+ + 1)2 ]−1 , m = [2 − (δ− − 1)2 ]−1 . (4.155)
Case D: a = 0. When a = 0, the corresponding solutions can be con-
sidered as the non-collinear generalization of the Szekeres family of collid-
ing collinearly polarized gravitational plane waves, since if we further set
Q1 = 0 = Q2 we shall rediscover the Szekeres solutions.
It must be noted that Halil “found” a class of non-diagonal solutions
(1979) using the harmonic maps of Riemannian manifolds. As the author
himself declared, it “is” a non-collinear generalization of the Szekeres family
of solutions. However, Griffiths (1987) argued that the Halil solutions do
not satisfy the Einstein vacuum equations.
In addition, Hassan, Feinstein and Manko (1990) developed an algo-
rithm for constructing exact solutions of colliding plane gravitational waves.
As an example, they found an Ernst potential E using the diagonal solu-
tions given by Eqs. (4.23) and (4.24) as seed, but unfortunately they have
not given the explicit solution for the function f . This is because in applying
their algorithm one has to integrate the system (4.7a)–(4.7b) directly after
the Ernst potential is given. However, for such a complicated Ernst poten-
tial this is not an easy task. It is expected that the specific Ernst potential
given by Hassan et al. may correspond to the two-soliton solutions with
Q1 = Q2 .
Finally, we would like to mention that colliding axisymmetric pp waves
were studied by Ivanov (1998), and Gürses, Kahya and Karasu (2002) gen-
eralized the collision of gravitational plane waves to high even-dimensional
spacetimes, with the combinations of collinear and noncollinear polarized
gravitational waves, and found that the spacetime singularity structure
depends on the parameters of the solution.
 b2530   International Strategic Relations and China’s National Security: World at the Crossroads

This page intentionally left blank

b2530_FM.indd 6 01-Sep-16 [Link] AM

March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch05 page 129

Chapter 5

Collisions of Gravitational Waves with Matter Fields

In this chapter, we consider the collision of plane gravitational waves with

matter. Specifically, in Section 5.1 two families of plane symmetric solutions
of the Einstein field equations are presented. These solutions represent the
collision of infinitely thin (impulsive) shells of null dust with the same
kind of shells or with constantly (collinearly) or variably (non-collinearly)
polarized gravitational plane waves. The general properties of these solu-
tions are discussed. In Section 5.2, the collision of massless scalar waves
is studied. A method for generating this kind of solutions from a known
vacuum solution is presented. Some relevant topics are discussed. Follow-
ing it, in Section 5.3 we consider solutions that represent the collision of
oppositely moving “null dust” clouds. With specific examples, the inter-
esting questions about the uniqueness of the outcome of the above kind
of collisions and the gravitational “phase transitions” induced from such
collisions are investigated. By solving the corresponding Einstein–Maxwell–
Weyl field equations, the ambiguity is resolved and the gravitational “phase
transitions” from massless particles into massive particles is not allowed in
general relativity.
In Section 5.4, we consider the collision of two electromagnetic plane
waves. In this section, we mainly focus on the well-known Bell–Szekeres
solution (1974), as it already contains the main features of this type of
collisions, although it is quite simple and represents the collision of two
pure collinearly polarized electromagnetic shock plane waves. In particular,
the focusing surface is a Cauchy horizon, instead of a spacetime curvature
singularity, and its extension beyond this surface is not unique (Clarke
and Hayaward, 1989), but it is expected that such a horizon is not stable,
and in more realistic case, it should be replaced by a spacetime singularity
(Chandrasekhar and Xanthopoulos, 1988; Konkowski and Helliwell, 1991).

129
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch05 page 130

130 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

Finally, in Section 5.5 the collisions of other matter waves in Einstein’s

theory as well as in other theories of gravity, including string/M -theory, are
briefly discussed. A very brief comment on the collisions of gravitational
waves in curved backgrounds, such as FRW and AdS, is also provided.

5.1. Collision of Impulsive Shells of Null Dust and

Gravitational Plane Waves
The infinitely thin (or impulsive) shells have attracted a lot of attention
since Israel’s work (Israel, 1966, 1967) was published (Boulware, 1973; Bron-
nikov, 1980; Siegel, 1981; Dray and ’t Hooft, 1985a, 1985b; Tsoubelis, 1989b;
Barrabés and Hogan, 2003; Bronnikov, Santos and Wang, 2019; and refer-
ences therein). One of the main reasons is that impulsive shells can be
considered as good mathematical models for describing dust clouds anal-
ogous to the models used for impulsive gravitational waves. However, the
interaction of two such shells had not been studied until 1986 when Dray
and ’t Hooft first discussed the gravitational effects of colliding planar shells
of matter. It was shown that, quite similar to the Khan–Penrose solution
of two colliding impulsive gravitational waves (Khan and Penrose, 1971),
such a collision develops a curvature singularity, and that a Coulomb-like
gravitational field appears after the collision. The above results together
with the ones obtained from the study of the effect of colliding null dust
clouds on the formation of singularities (Chandrasekhar and Xanthopou-
los, 1987a) led us once again to believe that the collision of impulsive shells
always produces curvature singularities.
However, in 1991 Tsoubelis and Wang found a two-parameter class of
solutions of the Einstein field equations (Tsoubelis and Wang, 1991) and
found that, like the Chandrasekhar–Xanthopoulos vacuum solution (Chan-
drasekhar and Xanthopoulos, 1986a), the collision of such shells does not
inevitably develop curvature singularities, and that, in contrast to the vac-
uum case (Szekeres, 1972), the Coulomb-like gravitational field does not
necessarily appear in the interaction region.

5.1.1. Colliding impulsive shells with collinearly polarized

gravitational waves
In this subsection, let us consider the two-parameter class of solutions found
by Tsoubelis and Wang (1991) in detail. As mentioned above, this class of
solutions represents the collision of null dust shells and collinearly polarized
gravitational plane waves.
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch05 page 131

Collisions of Gravitational Waves with Matter Fields 131

The analysis given in Section 3.2 [especially, see Eqs. (3.25) and (3.27)]
shows that, for any vacuum solution given in the interaction region
(Region IV), the extended solution obtained by the Khan–Penrose sub-
stitutions (3.11) will satisfy the Einstein vacuum equations inside the
pre-collision regions (Regions I–III). But across the hypersurfaces u = 0
and v = 0 the Ricci tensor will suffer the following discontinuities:

2nu2n−1 2mv 2m−1

Rμν = 2m
δ(u)l̂μ ˆlν + δ(v)ñμ n̂ν , (5.1)
1 − v H(v) 1 − u2n H(u)

where

l̂μ ≡ Alμ = δμu , n̂μ ≡ Bnμ = δμv . (5.2)

Using the Einstein field equations given by Eq. (1.20), we immediately

obtain

2nu2n−1 ˆlν + 2mv

2m−1
Tμν = δ(u)l̂ μ δ(v)n̂μ n̂ν . (5.3)
1 − v 2m H(v) 1 − u2n H(u)

It is clear that when n = 1/2 the first term on the right-hand side of
Eq. (5.3) represents an impulsive shell of null dust with support on the
hypersurface u = 0, while when m = 1/2 the second term represents a
similar shell but with support on the hypersurface v = 0. Therefore, when
n = 1/2 and m = 1/2 the solutions given by Eqs. (4.23) and (4.24) will
represent the collision of two impulsive shells of null dust, each of which
may be accompanied by a constantly polarized gravitational plane wave.
When one of the two parameters n and m is equal to 1/2 and the other
one still satisfies the conditions (4.29), the above solutions will represent
the collision of an impulsive shell of null dust with a gravitational plane
wave.
Without loss of generality, let us consider the solutions given by
Eqs. (4.23) and (4.24) for the case where
   
1 1 1
(i) (n, m) = , , or (ii) (n, m) = ,≥ 1 , (5.4)
2 2 2

which implies that in all the cases to be considered there always exists an
impulsive shell of null dust that propagates along the hypersurface u = 0
toward the left in Fig. 3.1. From Eq. (4.28) we can see that n = 1/2 implies
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch05 page 132

132 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

δ+ = 0, or
 
1
δ1 = −δ2 , n = . (5.5)
2

Then, the combination of Eqs. (3.24), (4.23), (4.24) and (5.4) leads to the
following expressions for the non-vanishing Weyl scalars in Region IV,
 3m
2 m−2 b1 (av + δ1 Y 3 )
ΨIV
0 (u, v) = 8m v
t 2

aδ1 [(2a + δ1 )v m + (a + 2δ1 )Y ]
− ,
(v m + Y )2
2b1 mv 2m−1 2aδ1 mv m−1
ΨIV
2 (u, v) = + , (5.6)
t2 Y (v m + Y )2
2b1 (δ1 v 3m + aY 3 )
ΨIV
4 (u, v) =
Y 3 t2
2aδ1 [(2a + δ1 )Y + (a + 2δ1 )v m ]
− .
Y 3 (v m + Y )2

From Eq. (5.6) we can see that as t → 0+ (while u, v = 0), a space-

time curvature singularity develops in all of the models, except for the
case in which Eq. (4.46) holds. In the latter, one of the Killing vectors ∂x
and ∂y becomes null as t → 0+ , as can be seen from Eqs. (4.23), (4.24)
and (5.4). Thus, when Eq. (4.46) holds, the space-like singularity that oth-
erwise is formed on the hypersurface t = 0 gives its place to a Cauchy hori-
zon, beyond which the metric can be extended, although such extensions
are not unique. In addition, following Yurtsever’s arguments (1987) and
Griffiths’ proof for pure gravitational waves, these horizons are expected not
stable, too.
On the other hand, from Eqs. (3.24) and (5.6), it can be shown that in
Region II the only non-vanishing Weyl scalar is Ψ0 , while in Region III the
only non-vanishing one is Ψ4 , which are given, respectively, by
2m
ΨII
0 (v) = {am(4a2 − 1)v 4m−2 + 12δ1 a2 mv 3m−2
(1 − v 2m )2
+ 3a(2m − 1)v 2m−2 + δ1 (m − 1)v 2m−2 }, (5.7)
a(4a2 − 1)
ΨIII
4 (u) = . (5.8)
2(1 − u)2
In Region I, all the Weyl scalars vanish.
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch05 page 133

Collisions of Gravitational Waves with Matter Fields 133

Combining Eqs. (3.24), (3.26) with Eqs. (5.6)–(5.8), we find that the
non-vanishing Weyl scalars have the following behavior when across the
hypersurface u = 0 or v = 0.

I → III:
Ψ0 , Ψ2 are continuous,
(5.9)
ΨI–III = ΨIII (u = 0)H(u) + aδ(u).
4 4

II → IV:

Ψ0 is continuous,
ΨII–IV
2 = ΨIV
2 (u = 0, v)H(u), (5.10)
(a + kv ) m
ΨII–IV
4 = ΨIV
4 (u = 0, v)H(u) + δ(u).
X2
I → II:
Ψ2 , Ψ4 are continuous,
(5.11)
ΨI–II = ΨII (v = 0)H(v) + 2m(kv m−1 + av 2m−1 )δ(v).
0 0

III → IV:

Ψ4 is continuous,
2m(kv m−1 Y + av 2m−1 )
Ψ0III–IV = ΨIV
0 (u, v = 0)H(v) + δ(v), (5.12)
Y2
Ψ2III–IV = ΨIV
2 (u, v = 0)H(v),

where

k ≡ δ1 = −δ2 . (5.13)

To study the above solutions further, let us consider some typical cases.
Case A: m = 1/2. In this case, from Eqs. (4.28) and (5.4) we find

k = 0, (5.14)

for which we find m = 1/2, as it can be seen form Eq. (4.28). Then, setting
n = 1/2 and m = 1/2 in Eq. (5.3), we find that the non-vanishing compo-
nents of Tμν are given by
δ(u) δ(v)
Tuu = , Tvv = , (5.15)
1 − vH(v) 1 − uH(u)
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch05 page 134

134 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

which correspond to two impulsive shells of null dust with support,

respectively, on u = 0 and v = 0. On the other hand, from Eqs. (5.6)–(5.14)
we find,

a(4a2 − 1)H(v) aδ(v)

Ψ0 = + , (5.16)
2(1 − uH(u) − v) 1 − uH(u)
a(4a2 − 1)H(u) aδ(u)
Ψ4 = + , (5.17)
2[1 − u − vH(v)] 1 − vH(v)
(4a2 − 1)H(u)H(v)
Ψ2 = , (5.18)
4[1 − u − v]2

which show that the corresponding solutions represent the symmetric col-
lision of a pair of impulsive shells of null dust, each of which is in general
accompanied by a constantly polarized impulsive + shock gravitational
plane wave. As a result of the collision, a Coulomb-like gravitational field
Ψ2 appears in the interaction region (Region IV) and a spacetime singular-
ity is finally developed on the hypersurface 1 = u + v.
When a = 0, Ψ0 and Ψ4 vanish everywhere, and the corresponding solu-
tion represents the collision of two pure impulsive shells of null dust. This
solution was first found by Dray and ’t Hooft (1986) and shown by Tsoubelis
(1989b) that it belongs to the Szekeres family of solutions (Szekeres, 1972).
When a = ±1/2, Eqs. (5.16) and (5.17) show that the two incoming
gravitational waves are impulsive ones. Thus, in this case the solutions
represent the collision of two impulsive shells of null dust, each of which is
accompanied by an impulsive gravitational plane waves. One of the remark-
able features in this case is that the interaction region (Region IV) is flat
and no Coulomb-like gravitational field appears in this region. Comparing
it, for example, with the Khan–Penrose solution (Khan and Penrose, 1971),
we are led to the conclusion that, when accompanied by null dust shells,
the collision of two impulsive gravitational waves does not necessarily give
rise to a Coulomb-like gravitational field in the interaction region. Since in
the present case Region IV is flat, no spacetime singularity develops on the
hypersurface 1 = u + v.
Before turning to the next case, let us note that the solutions presented
in this case were first obtained by Stoyanov (1979), but with an incor-
rect interpretation that the above solutions represented the collision of two
purely gravitational waves. The fact that Stoyanov’s interpretation cannot
be supported was first noticed by Nutku (1981) and is made clear from the
analysis presented above.
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch05 page 135

Collisions of Gravitational Waves with Matter Fields 135

Case B: m = 1. When m = 1, we find k = ±1/2 and

δ(u)
Tuu = , Tvv = 0, (5.19)
1 − v 2 H(v)

and
1
ΨI4–III = a(4a2 − 1)H(u) + aδ(u), (5.20)
2
ΨI0–II = 6aH(v) + 2kδ(v), (5.21)
a[3v 2 + 12akv + (4a2 − 1)] a + kv
Ψ4II–IV = H(u) + δ(u), (5.22)
2(1 − v 2 )2 1 − v2
6a 2k
Ψ0III–IV = H(v) + √ δ(v). (5.23)
1−u 1−u

Equation (5.19) shows that in this class of solutions, the hypersurface v = 0

is always free of matter, while Eqs. (5.20) and (5.21), on the other hand,
show that the solutions with a = 0, ±1/2 represent the collision of an
impulsive shell of null dust with an impulsive + shock gravitational wave.
The former is accompanied by an impulsive + shock gravitational wave. In
the present case, all the solutions are singular on the hypersurface 1 − u −
v 2 = 0, except for the ones in which a = 0, ±1 [see Eq. (4.46)].

Case B.1: a = 0. In this case, Eqs. (5.20) and (5.21) reduce to ΨI4–III = 0
and ΨI0–II = δ(v), respectively, where  = ±1. In fact, in this case the only
non-vanishing Weyl scalars are

 vH(v)
Ψ0 (u, v) =  δ(v), Ψ4 (u, v) = δ(u). (5.24)
1 − uH(u) 2(1 − v 2 )

Therefore, this model represents the collision of an impulsive gravitational

plane wave incident from the left-hand side along the null hypersurface
v = 0 with an impulsive shell of null dust incident from the right-hand side
along the null hypersurface u = 0. It is very interesting to note that an
impulsive gravitation wave represented by Ψ4 along the null hypersurface
u = 0 is produced right after the collision. The spacetime is flat inside
all the four regions (Regions I–IV) and the only effect of the collision is
the mutual focusing of the wave pulse and the null dust shell, and isolated
spacetime singularities are developed, respectively at the points (u, v) =
(0, 1) and (u, v) = (1, 0). The corresponding metric was first obtained by
Babala (1987) and rediscovered by Tsoubelis (1989b).
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch05 page 136

136 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

Case B.2: a = ±1/2 ≡  /2. In this case, Eqs. (5.20) and (5.21) reduce to
1 
ΨI–III
4 =  δ(u), (5.25)
2
ΨI–II
0 = 3 H(v) + δ(v), (5.26)
where  and  can be chosen independently, always we have || = | | = 1.
Then, we can see that the present model represents the collision of a shock
wave with an impulsive shell of null dust, each of which is accompanied by
an impulsive gravitational wave. In this case, the formation of a spacetime
singularity along the hypersurface u = 1 − v 2 is inevitable.
Case B.3: a = −2k. Since in this case we have
3
ΨI–III
4 = − εH(u) − εδ(u), (5.27)
2
ΨI–II
0 = −6εH(v) + εδ(v), (5.28)
we find that the corresponding solutions represent the collision of two impul-
sive + shock gravitational waves, one of which is accompanied by an impul-
sive shell of null dust. The collision is such that the spacetime is free of
singularity on the hypersurface 1 − u − v 2 = 0. In fact, the Coulomb-like
gravitational field that develops in Region IV after the collision remains
finite as one approaches the hypersurface t = 1 − u − v 2 = 0, and is given by
1
ΨIV
2 (u, v) = − √ √ . (5.29)
[ 1 − u + v]2 1 − u
Thus, we have
1
ΨIV 2
2 (u = 1 − v , v) = − . (5.30)
4v 3
√
Case C: m = 2. In this case, we have k =  3/8 and that
1
ΨI4–III = a(4a2 − 1)H(u) − aδ(u), (5.31)
2
ΨI0–II = 4kH(v). (5.32)
Thus, in the present case the solutions represent the collision of a shock
gravitational wave incident from the left with an impulsive shell of null
dust incident from the right. The latter is accompanied by a gravitational
shock + impulsive wave, provided a = 0, ±1/2. The behavior of this subclass
of models after collision is similar to the ones obtained in Cases A and B
above. Specifically, the development of a spacetime singularity along the
1 − u − v 4 = 0 is inevitable, except for the cases where a = −k ± 1/2.
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch05 page 137

Collisions of Gravitational Waves with Matter Fields 137

When a = 0, one of the two incoming impulsive shells of null dust

appearing in the Dray–’t Hooft model is replaced by a shock gravitational
plane wave. The outcome of the collision is similar to that in the Dray–’t
Hooft model. Specifically, a Coulomb-like gravitational field appears in the
interaction region, and a spacetime singularity develops on the hypersurface
t = 1 − u − v 4 = 0.
When a = ±1/2, the null dust shell is accompanied by an impulsive
gravitational wave. Thus, the corresponding solutions represent the collision
of a shock gravitational wave and an impulsive shell of null dust which is
accompanied by an impulsive gravitational wave. In this case, a spacetime
singularity also always develops on the hypersurface t = 1 − u − v 4 = 0.

Case D: m = 4. The shock wave incident from the left in Case C now
is replaced by a gravitational plane wave with a smooth wavefront. This
follows from the fact that in this case we have

ΨI0–III = Ψ0III–IV = 0. (5.33)

On the other hand, we have

1
ΨI4–II = a(4a2 − 1)H(u) − aδ(u), (5.34)
2

as in the previous cases, and

a(4a2 − 1) + 12a2 kv 4 + 12ak 2 v 8 + k(4k 2 − 1)v 12

Ψ4II–IV = H(u)
2(1 − v 8 )2
a + kv 4
+ δ(u). (5.35)
1 − v8

Comparing Eqs. (5.33) and (5.34) with Eqs. (5.32) and (5.31), we find that
the current models represent the same kind of collisions as the m = 2
models, except that the shock gravitational wave incident in Region II is
now replaced by a gravitational wave with a smooth wavefront. Thus, in
the following we do not go to detail, except for mentioning that in the
present case the second gravitational wave production is illustrated most
clearly, although it arises essentially in all the spacetime models constructed
above. Consider in this direction, the a = 0 case. For this particular model
Ψ4I–III = 0. Therefore, the null dust shell incident from the right is not
accompanied by any gravitational radiation. According to Eq. (5.35), on
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch05 page 138

138 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

the other hand, we have

3kv 12 kv 4
Ψ4II–IV = H(u) + δ(u), (5.36)
8(1 − v 8 )2 1 − v8
which shows that, upon interacting with the gravitational wave pulse
incident from the left, the shell of null dust stimulates the emission of
gravitational radiation in its own direction, besides getting focused as it is
propagating.

5.1.2. Collisions of an impulsive null dust shell and a

non-collinearly polarized gravitational wave
The analysis carried out in the last section shows that the one-soliton solu-
tions given by Eqs. (4.56)–(4.58) will in general represent the collision of
an impulsive shell of null dust with a variably polarized gravitational plane
wave, if the soliton parameters δ1 and δ2 are chosen so that
δ+ ≡ δ1 + δ2 = 0, (5.37)
which leads to
1
, n= (5.38)
2
as follows from Eq. (4.61). Note that the solutions given by Eqs. (4.56)–
(4.58) are for the case δ− ≥ 0. Thus, in this case we always have m ≥ 1
[see Eq. (4.61)]. That is, the hypersurface v = 0 in this case is always free
of matter.
From Eqs. (4.63) and (4.64), on the other hand, we can see that the
solutions in Region II do not depend on δ+ . Consequently, the condition
given by Eq. (5.37) does not have any restriction on the solutions given by
Eqs. (4.63) and (4.64). Therefore, the gravitational plane wave incident in
Region II is the one described below Eq. (4.65).
The plane gravitational wave incident in Region III can be obtained
from Eqs. (5.8) and (5.9) [see Eqs. (4.33) and (4.34)] and is given by
1
Ψ4 (u, v < 0) = a[(4a2 − 1)Y −4 H(u) + 2δ(u)]. (5.39)
2
It is clear from Eqs. (5.3) and (5.39) that, when a = 0, ±1/2, the impulsive
shell of null dust with support on the hypersurface u = 0 is accompanied
generally by a constantly polarized gravitational impulsive + shock plane
wave. When a = 1/2, the shock part of the latter disappears and only the
impulsive gravitational plane wave remains.
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch05 page 139

Collisions of Gravitational Waves with Matter Fields 139

As an example, let us consider the solutions of Eqs. (4.56)–(4.58) with

the parameters δ1 and δ2 set equal to zero, for which Eq. (4.61) yields
1
, m = 1.
n= (5.40)
2
The previous analysis shows that this class of solutions in general repre-
sents the collision of a variably polarized shock + impulsive gravitational
plane wave with an impulsive shell of null dust. The latter may be accom-
panied by a constantly polarized shock + impulsive gravitational wave. The
corresponding solutions are given by
A 2a+1 t4a
χ= t , q2 = −4qvY H(v),
B B (5.41)
A
e−M = t2a(a−1) , e−U = t = 1 − uH(u) − v 2 H(v),
Y
where
A = [Y − vH(v)]4a + q 2 [Y + vH(v)]4a ,
B = [Y − vH(v)]4a [Y + vH(v)]2 (5.42)
+ q 2 [Y + vH(v)]4a [Y − vH(v)]2 ,
Y = [1 − uH(u)]1/2 .
From the expressions of Eqs. (5.41) and (5.42) it can be seen that to study
this class of solutions as a whole is very complicated. In the following, we
consider only several representative cases, which are sufficient for illustrat-
ing the main properties of this class of solutions.
Case 1: a = 0. In this case, we find
ΨIV
i = 0 (i = 0, 2, 4), (5.43)
which means that the spacetime is flat in the interaction region (Region
IV). Hence, from Eqs. (3.24) and (3.26) we find that
1
Ψ0 (u, v) = {(1 − q 2 ) − i2q}δ(v),
(1 + q 2 )Y
vH(v)
Ψ4 (u, v) = {(1 − q 2 )(1 − v 2 ) + i2q(1 + v 2 )}δ(u), (5.44)
2(1 − v 2 )D
Ψ2 (u, v) = 0,
where
D ≡ [(1 + q 2 )2 (1 − v 2 H(v))2 + 16q 2 v 2 H(v)]1/2 . (5.45)
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch05 page 140

140 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

Thus, in this case the solution represents the collision of an impulsive gravi-
tational plane wave and an impulsive shell of null dust. Since the spacetime
is flat in Region IV, the hypersurface t = 0 is free of spacetime curvature sin-
gularity, except at the two focusing points (u, v) = (0, 1) and (u, v) = (1, 0).
The study of polarization of the two impulsive gravitational waves given
by Eq. (5.44) in the next subsection will reveal that this solution is actually
the Babala solution referred as Case B.1 in Section 5.1.
Case 2: a = 1/2. In this case, if we define the functions F (x, y) and G(x, y)
as

F (x, y) ≡ (x − y)5 − 6q 2 x(x2 − y 2 )2 + q 4 (x + y)5 + i2q(x2 − y 2 )

× [(1 + q 2 )(5x2 + y 2 )y − 2(1 − q 2 )(x2 + 2y 2 )x],
(5.46)
G(x, y) ≡ (x − y)4 (x2 + xy + y 2 ) − q 4 (x + y)4 (x2 − xy + y 2 )
− 6q 2 xy(x2 − y 2 )2 − i2q(1 + q 2 )(x2 − y 2 )3 ,

we find that the non-vanishing Weyl scalars in Region IV are given by

3(1 + q 2 )Y
ΨIV
0 (u, v) = F (Y, v),
tA2 D
3(1 + q 2 )v
ΨIV
4 (u, v) = − F̄ (v, Y ), (5.47)
4tA2 DY 2
1
ΨIV
2 (u, v) = G(Y, v),
2t2 A2 Y
but now the function D is defined by

D ≡ [A2 + 16q 2 v 2 Y 2 ]1/2 . (5.48)

Then, from Eqs. (3.24) and (3.26) we find

1
Ψ0 (u, v) = ΨIV
0 (uH(u), v)H(v) + {(1 − q 2 ) − i2q}δ(v),
(1 + q 2 )Y
1 + q2
Ψ4 (u, v) = ΨIV
0 (u, v)H(u)H(v) + {(1 − vH(v))3
2AD
(5.49)
+ q 2 (1 + vH(v))3 + i2qv(1 − v 2 )H(v)}δ(u),
Ψ2 (u, v) = ΨIV
2 (u, v)H(u)H(v).

Taking Eq. (5.38) into account, we find that this solution represents the col-
lision of a variably polarized gravitational shock + impulsive wave with an
impulsive shell of null dust, which is accompanied by a constantly polarized
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch05 page 141

Collisions of Gravitational Waves with Matter Fields 141

gravitational impulsive wave. In this model, Region III is flat, while Region
II is curved due to the presence of the variably polarized gravitational wave.
As t → 0+ , all the non-vanishing Weyl scalars become unbounded. Thus,
the spacetime now is singular on the hypersurface t = 0.
Case 3: a = 1. Then, we find that the non-vanishing Weyl scalars are given
by
1
Ψ0 (u, v) = ΨIV
0 (uH(u), v)H(v) + {(1 − q 2 ) − i2q}δ(v),
(1 + q 2 )Y
1
Ψ4 (u, v) = ΨIV
4 (u, vH(v))H(u) + {(2 + vH(v))(1 − vH(v))8
2tAD
+ q 2 (1 + vH(v))3 [q 2 (2 − vH(v))(1 + vH(v))5
+ 4(1 + 3v 2 H(v))(1 − vH(v))3 ] (5.50)
2 4 2
+ i2qv(1 − v )H(v)[(1 − v) (1 + 4v + v )
+ q 4 (1 + v)4 (1 − 4v + v 2 )]}δ(u),
Ψ2 (u, v) = ΨIV
2 (u, v)H(u)H(v),

where
6 3
ΨIV
0 (u, v) = F (Y, v), ΨIV
4 (u, v) = F (v, Y ),
A2 D 2Y 2 A2 D (5.51)
6
ΨIV
2 (u, v) = G(Y, v),
A2 Y
and the functions D, F and G now are defined by
D(u, v) ≡ [A2 + 16q 2 v 2 t2 Y 2 ]1/2 ,
F (x, y) ≡ (x − y)10 − q 2 (x2 − y 2 )(x − y)4 (5x4 + 30x3 y + 8x2 y 2
+ 2xy 3 + 3y 4 ) − q 4 (x2 − y 2 )(x + y)4 (5x4 − 30x3 y
+ 8x2 y 2 − 2xy 3 + 3y 4 ) + q 6 (x + y)10
+ i4qx[(x − y)7 (y 2 − 2xy − x2 ) + 2q 2 y(x − y)3 (5x2 + y 2 )
(5.52)
− q 4 (x + y)7 (y 2 + 2xy − x2 )],
G(x, y) ≡ (x − y)6 − 12q 2 xy(x2 − y 2 )2 − q 4 (x + y)6
− i2q[(x − y)4 (y 2 + 4xy + x2 )
+ q 2 (x + y)4 (y 2 − 4xy + x2 )].
From Eqs. (5.3), (4.39) and (5.50), we can see that the solution in the
present case represents the collision of a variably polarized gravitational
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch05 page 142

142 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

impulsive + shock wave with the same type wave but constantly polarized.
The latter has an impulsive shell of null dust as its leading front.
As t → 0+ , Eqs. (5.51) and (5.52) show that all the non-vanishing
Weyl scalars keep finite provided q = 0. That is, the spacetime is free of
spacetime curvature singularity on the hypersurface t = 0 when q = 0.
However, when q = 0, it is easy to see that the non-vanishing Weyl scalars
become unbounded as t → 0+ . Thus, in the latter case the spacetime is
singular on t = 0. Consequently, a close relation is established between the
collision involving impulsive shells of null dust, on the one hand, and the
collision involving only gravitational waves, on the other. For the latter
type of collision, the effect of polarization of colliding plane gravitational
waves on the formation of singularities was already studied in Sections 4.2
and 4.3.
Before proceeding to the next section, we note that solutions that
represent the collision of an impulsive shell of null dust and a variably polar-
ized gravitational plane wave were first studied by Feinstein and Senovilla
(1989). Starting with the assumption that the non-diagonal term q2 (in
their notation it is denoted by ω) depends only on a null coordinate v in
the interaction region, they found a unique solution of the Einstein vacuum
equations for a given arbitrary q2 (v). After extending the solutions back
to the pre-collision regions, they found an impulsive shell of null dust to
appear on the hypersurface u = 0. It was the latter that enabled them to
interpret their solutions as representing the collision of a plane gravitational
wave with an impulsive shell of null dust.

5.1.3. Polarizations of colliding plane gravitational waves

when coupled with an impulsive null dust shell
As we showed in Section 3.4, due to the interaction between an impul-
sive gravitational wave and the same kind of shells, the polarization of the
impulsive gravitational wave gets rotated, while the polarization of a shock
gravitational plane wave can be changed by both the nonlinear interaction
with the other plane gravitational wave and the interaction with matter
fields.
In this section, we consider the first two solutions discussed in the last
subsection to illustrate further the polarizations of gravitational impulsive
waves as well as the shock ones. The main reasons that in this subsection we
do not include the study of the third solution discussed in the last subsection
are two-fold. First, the expressions of the non-vanishing Weyl scalars for this
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch05 page 143

Collisions of Gravitational Waves with Matter Fields 143

solution are very complicated. It is difficult to see the physics. Second, from
this solution we shall obtain the similar results about the polarizations of
colliding plane gravitational waves as we shall get from the first two.
With the above in mind, let us first consider Case 1 solution. The non-
vanishing Weyl scalars in this case are given by Eq. (5.44). Since on the
hypersurface v = 0, we have W = 0 [see Eq. (5.41)]. Hence, integrating
Eq. (3.61) we find that

ϕ00 = 0. (5.53)

Combining Eqs. (2.46), (5.44) and (5.53), we find that

 
1 2q
θ0Im = tan−1 . (5.54)
2 1 − q2
Im
Obviously, θ0,u = 0. That is, the polarization angle of the impulsive gravi-
tational wave represented by Ψ0 does not change, since in the present case
ΨIm
0 does not interact with any of the other components of the gravitational
field [see Eq. (3.69)].
On the other hand, integrating Eq. (3.59), we obtain the following solu-
tion:
 
(0) 1 −1 4q(1 − q 2 )v 2
ϕ4 = tan H(v). (5.55)
2 (1 + q 2 )2 − (1 − 6q 2 + q 4 )v 2
From Eqs. (2.38), (5.44) and (5.55), we find
 
Im 1 −1 2q
θ4 = tan H(v). (5.56)
2 1 − q2
It follows that after it is created from the collision, the impulsive gravi-
tational wave, represented by ΨIm 0 , does not change its polarization, too.
On the other hand, from Eqs. (5.54) and (5.56) we can see that the impul-
sive plane gravitational wave, Ψ4 , has the same polarization angle as the
Ψ0 -wave does.
The above observations show that if we make a rotation in the (x2 , x3 )-
plane, the metric should be brought into a diagonal form. In fact, we find
that the metric takes the form
2(1 + q 2 )
ds2 = dudv − [Y − vH(v)]2 (dx2 )2 − [Y + vH(v)]2 (dx3 )2 , (5.57)
Y
after making a coordinate rotation given by Eq. (2.32) with the angle given
by Eq. (5.54). Equation (5.57) is the solution first found by Babala (1987).
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch05 page 144

144 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

Actually, if we replace u by u defined as


 1 − 2(1 − u)1/2 , u ≥ 0,
u = (5.58)
u − 1, u ≤ 0,

Eq. (5.57) will take the exact form used by Babala (1987).
Now let us turn to the solution of Case 2 given in Subsection 5.1.2. We
first note that in this case the Weyl scalars Ψ0 and Ψ4 consist of two parts:
the shock part and the impulsive part. In the following, we consider them,
separately.
Let us first consider the impulsive part. Following the discussion given
in the last case, we find that
 
1 2q
θ0Im = tan−1
,
2 1 − q2
  (5.59)
1 2qv
θ4Im = tan−1 H(v).
2 (1 − v) + q 2 (1 + v)

Thus, the polarization angle for the impulsive part of Ψ0 remains constant
even after the collision. The reason is that, similar to the last case, ΨIm
0
does not interact with any of the other components of both gravitational
and matter fields. However, for the Ψ4 wave the situation is different. The
interaction between the impulsive part of Ψ4 and the impulsive shell of null
dust ΦIm Im
22 is such as to make the polarization angle θ4 change along the
v-axis according to Eq. (5.59).
In a similar fashion, we find that for the shock part of Ψ0 and Ψ4 the
polarization angles θ0sh and θ4sh are given, respectively, by
 
1 I(v, Y )
θ0sh = − tan−1
H(v),
2 J(v, Y )
  (5.60)
1 I(Y, v)
θ4sh = tan−1 H(u)H(v),
2 J(Y, v)

where the functions I(x, y) and J(x, y) are defined by

I(x, y) ≡ 2qy{(x2 − y 2 )(x − y)(x + 5y) + q 2 [(4 + 6q 2 + 4q 4 + q 6 )x4

+ 4(2 − 2q 4 − q 6 )x3 y + 2(4 + 14q 2 + 4q 4 − 3q 6 )x2 y 2
+ 4(6 − 6q 4 + q 6 )xy 3 − (60 − 126q 2 + 60q 4 − 5q 6 )y 4 ]},
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch05 page 145

Collisions of Gravitational Waves with Matter Fields 145

J(x, y) ≡ (x2 − y 2 )2 (x − y) + q 2 {(5 + 10q 2 + 10q 4 + 5q 6 + q 8 )x5

− (3 + 2q 2 − 2q 4 − 3q 6 − q 8 )x4 y + 2(3 + 14q 2 + 14q 4
+ 3q 6 − q 8 )x3 y 2 − 2(13 + 14q 2 − 14q 4 − 13q 6 + q 8 )x2 y 3 (5.61)
− (27 − 42q 2 − 42q 4 + 27q 6 − q 8 )xy 4 + (45 − 210q 2
+ 210q 4 − 45q 6 + q 8 )y 5 }.
Equation (5.60) shows that the plane gravitational shock waves of Ψ0 and
Ψ4 change their polarization directions along each of their own paths. These
changes are due to the presence of the Coulomb-like field Ψ2 in Eq. (3.70),
which is due to the nonlinear interaction between Ψ0 and Ψ4 .

5.2. Collisions of Massless Scalar Waves

The gravitational interaction of colliding null dust clouds has been inves-
tigated by several authors (Chandrasekhar and Xanthopoulos, 1985b,
1985c, 1986b; Taub, 1988a, 1988b, 1990; Ferrari and Ibañez, 1989a, 1989b;
Feinstein, MacCallum and Senovilla 1989; Tsoubelis and Wang, 1991).
Many remarkable features of the interaction have been found. One of these
arose initially from two solutions of the Einstein field equations constructed
by Chandrasekhar and Xanthopoulos (1985b, 1986b). According to the
above authors, one of the two solutions represents the collision of two null
dust clouds (Chandrasekhar and Xanthopoulos, 1985b), each of which is
accompanied by an impulsive gravitational plane wave. The product of the
collision is a perfect fluid with the equation of state, p (pressure) = ε (energy
density). The other solution represents the collision of the same null dust
clouds but with the product being a mixture of null dust moving in opposite
directions (Chandrasekhar and Xanthopoulos, 1986b). The presence of two
different fluids in the interaction region gave rise to the following issues.
One concerns the uniqueness of the collision, and the other concerns the
possibility of a gravitationally induced transformation (or phase transition)
of massless particles into massive ones.
Taub (1988a, 1988b) showed that the different outcome of the above
collision is due to the different assumptions on the Ricci tensor, which
specify the nature of the interaction. In other words, in order to restore
the uniqueness, one has to impose some conditions on the Ricci tensor.
Feinstein, MacCallum and Senovilla (1989), on the other hand, argued
that the ambiguities arise from an incomplete physical treatment of the
problem, in the sense that only the gravitational field is considered in the
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch05 page 146

146 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

Chandrasekhar–Xanthopoulos approach. These ambiguities remain even

after Taub’s points are taken into account. They disappear only when the
underlying physical equations of the matter fields are taken into considera-
tions. They also argued that the transformation of colliding null dust into
other forms of matter is unrealizable in classical general relativity.
Tsoubelis and Wang (1991), however, showed that as far as the unique-
ness is concerned, the approach given by Taub is equivalent to the one
given by Feinstein et al. Actually, if one follows the instructions given by
Feinstein et al. that first specify the matter field and then solve both the
Einstein field equations and the matter field equations, one will find that
this is no more than to provide the conditions imposed by Taub.
The difference between the above two kinds of approaches essentially
reflects the well-known “problem” of the right-hand side of the Einstein
field equations (Hawking and Ellis, 1973) or in Synge’s words, the prob-
lem between the realist and the agonist (Synge, 1965). The realist wants
to connect the energy–stress tensor appearing in the right-hand side of the
Einstein field equations (1.20) with the known matter fields. So, the start-
ing point is first to specify the matter field, and then to solve the coupling
Einstein and matter field equations. The agonist, on the other hand, wants
to wrestle with the difficult mathematical problems arising out of the field
equations. Then, the starting point is to construct the metric (i.e. to solve
the left-hand side of the Einstein field equations) disregarding the physical
constitution of matter. After the metric is given, the energy–stress ten-
sor is read off from the Einstein field equations. However, such obtained
energy-stress tensor does not always physically acceptable, as they com-
monly violate energy conditions (Hawking and Ellis, 1973). To be physi-
cally acceptable, the agonist has to impose some conditions on the metric
or equivalently on the Ricci tensor.
In this section, by considering a class of solutions which represents the
collision of massless scalar waves, we shall show that, besides the above
mentioned problems, the fluid picture has also difficulties in interpreting
some results of the above kind of collisions. But, these difficulties can be
overcome if we consider the collision as the one of two massless scalar waves.
Before proceeding, we first present a theorem, which is very useful in
constructing the above kind of collisions from the known solutions of the
Einstein vacuum equations.
First we notice that the Einstein field equations when coupled with a
massless scalar field are given by Eqs. (3.8a)–(3.8e) with the non-vanishing
Ricci scalars given by Eqs. (3.67) and (3.69). Following Wainwright, Ince
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch05 page 147

Collisions of Gravitational Waves with Matter Fields 147

and Marshman (1979), Tabensky and Taub (1973) and Taub (1988b), we
find that solutions of a massless scalar field can be obtained from vacuum
ones by simply setting
(M, U, V, W ) = (Mv + Ω, Uv , Vv , Wv ), (5.62)
where Mv , Uv , Vv and Wv are solutions of the Einstein vacuum equations,
and the function Ω satisfies
φ,u
Ω,u = , (5.63a)
U,u
φ,v
Ω,v = , (5.63b)
U,v
Ω,uv = φ,u φ,v , (5.63c)
where the scalar field φ satisfies the massless scalar field equation (3.80). If
we consider, on the other hand, the metric
(0) (0) (0)
ds2 = 2e−M dudv − e−U [eV (dx2 )2 + e−V (dx3 )2 ], (5.64)
we find that the corresponding Einstein vacuum equations are given by [see
Eqs. (3.8a)–(3.8e)],
2 (0) (0)
2U,vv − U,v + 2U,v M,v = V,v2 , (5.65a)
(0) (0)
2U,uu − U,u2 + 2U,u M,u = V,u2 , (5.65b)
(0)
2M,uv = U,u U,v , (5.65c)
and
U,uv − U,u U,v = 0, (5.66)
(0)
2V,uv − U,u V,v(0) − U,v V,u(0) = 0. (5.67)
Introducing the function Ω(0) by
M (0) = N (0) + Ω(0) , (5.68)
where
3
N (0) ≡
U − ln |2U,u U,v |, (5.69)
2
we find that Eqs. (5.65a)–(5.65c) can be written as
(0) (0)
(0) V,u (0) V,v
Ω,u = , Ω,v = ,
2U,u 2U,v (5.70)
(0) 1 (0) (0)
Ω,uv = V,u V,v .
2
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch05 page 148

148 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

Comparison of Eqs. (3.80) and (5.63a)–(5.63c) with Eqs. (5.67) and (5.70)
leads to the following theorem.
Theorem (Tsoubelis and Wang, 1991). Let (M (0) , U, V (0) ) be a solution
of the Einstein vacuum equations corresponding to the metric (5.64), Then
(M, U, V, W ) = (Mv + λ2 (M (0) − N (0) ) + C0 , U, Vv , Wv ), (5.71)
(0)
with N given by Eq. (5.69) is a solution
√ of the Einstein equations coupled
(0)
with a massless scalar field φ = λV / 2, where λ and C0 are constant.
Note that when Vv = Wv = 0, the Einstein vacuum equations give the
following solution for the function Mv :
Mv = N (0) + C1 , (5.72)
where C1 is another arbitrary constant.
As an application of the above theorem, we use the vacuum solutions
given by Eqs. (4.23) and (4.24) as the solutions of (M (0) , U, V (0) ), and for
the sake of convenience we set the functions Vv and Wv to be zero
Vv = Wv = 0. (5.73)
Then, the function Mv has the solution given by√Eq. (5.72). By choosing
the parameter λ introduced in Eq. (5.71) to be 2, we finally obtain the
following solutions:
 
1−η
φ = V (0) ≡ a ln[(1 − η 2 )(1 − μ2 )] + δ1 ln
1+η
 
1−μ
+ δ2 ln , (5.74)
1+μ
and
e−U = 1 − u2n − v 2n ,
2 2
e−M = (1 − η)b1 (1 − μ)b2 (1 + η)c1 (1 + μ)c2 X −2δ+ Y −2δ− , (5.75)
V = W = 0,
but now bA and cA are defined by
1 1
bA = 2(a + δA )2 − , cA = 2(a − δA )2 − , (A = 1, 2), (5.76)
4 4
and the parameters n and m are related to δ± via the relations,
1 1
n= 2 ), m= 2 ). (5.77)
2(1 − δ+ 2(1 − δ−
As usual, taking the above solutions as valid only in the interaction region
(Region IV), we can extend them to the pre-collision regions by means
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch05 page 149

Collisions of Gravitational Waves with Matter Fields 149

of the Khan–Penrose substitutions (3.11). Specifically, the pair (u, v) is

replaced by the one (uH(u), vH(v)). However, the results obtained in the
last chapter and the ones in Section 5.1 show that the extended solutions
are physically acceptable only in the case where
1 1
n= , or ≥ 1, and m = , or ≥ 1. (5.78)
2 2
It can be shown that such extended function φ still satisfies the massless
scalar field equation (3.80).
As shown in Section 5.1, the above extension leads an impulsive shell of
null dust to appear on the hypersurface u = 0(v = 0) if n = 1/2 (m = 1/2).
We note that the above three-parameter class of solutions is a member
of the Ferrari–Ibañez four-parameter class of solutions (Ferrari and Ibañez,
1989a), and was described as “stiff matter” solutions. However, the follow-
ing considerations show that such a term does not justice to the physically
rich class of models given by Eqs. (5.74) and (5.75).
To begin with, we first show that a massless scalar field is not always
energetically equivalent to a perfect fluid with the equation of state p = ε.
Writing the energy–stress tensor given by Eq. (1.92) in the form
Tμν = φ,μ φ,ν − εgμν , (5.79)
where
1
ε≡ φ;μ φ;μ , (5.80)
2
we find that the following possibilities in general arise (Tabensky and Taub,
1973).
Case (i): ε > 0. In this case, the vector φ,μ is time-like. If we introduce
the unit vector Tμ by
φ,μ
Tμ ≡ , (5.81)
(2ε)1/2
we find that the metric and the energy–stress tensor can be written in the
form
gμν = Tμ Tν − Zμ Zν − Xμ Xν − Yμ Yν , (5.82)
Tμν = ε(Tμ Tν + Zμ Zν + Xμ Xν + Yμ Yν ), (5.83)
where Xμ , Yμ and Zμ are three orthogonal unit space-like vectors, which,
together with Tμ , consist of an orthogonal tetrad. Then, Eq. (5.83) implies
that the scalar field φ is energetically equivalent to a perfect fluid with
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch05 page 150

150 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

four-velocity proportional to φ,μ and the equation of state, p = ε, which is

often referred to as the “still” fluid.

Case (ii): ε < 0. In this case, φ,μ defines a space-like unit vector, say Zμ ,
via the relation
φ,μ
Zμ = . (5.84)
(−2ε)1/2

With similar considerations, we find that the metric in the present case is
given by Eq. (5.82), while the energy–stress tensor is given by

Tμν = |ε|(Tμ Tν + Zμ Zν − Xμ Xν − Yμ Yν ). (5.85)

Therefore, in this case the scalar field φ is energetically equivalent to an

anisotropic fluid with vanishing heat flow.

Case (iii): ε = 0: In this case, φ,μ is a null vector and Tμν takes the form
that we have been calling a null dust cloud propagating in the direction
defined by φ,μ .
The significance of the above observations will be made clear as we are
going to reveal the fact that in the solutions under considerations all the
above possibilities arise in the interaction region (Region IV) [see Fig. 5.1].

Fig. 5.1. The projection of the colliding null dust spacetime onto the (u, v)-plane.
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch05 page 151

Collisions of Gravitational Waves with Matter Fields 151

In particular, in this region we have

ε = eM Ruv
4nmeM tan ζ tan ξ
= [a sin 2ζ + δ+ cos ζ cos ξ + δ− sin ζ sin ξ]
t2 (sin ζ)1/n (sin ξ)1/m
×[a sin 2ξ + δ− cos ζ cos ξ + δ+ sin ζ sin ξ], (5.86)

where

sin ζ ≡ un , sin ξ ≡ v m . (5.87)

Equation (5.86) shows that to study the above solutions as a whole is

very complicated, so in the following we consider only some representative
cases.

Case A: δA = 0 (n = m = 1/2). In this case, the scalar field φ is given by

φ = a ln[(1 − η 2 )(1 − μ2 )] = 2a ln[1 − uH(u) − vH(v)], (5.88)

while the combination of Eqs. (1.20), (1.92) and (5.88) yields,

4a2 1
Ruu = H(u) + δ(u), (5.89a)
[1 − u − vH(v)]2 1 − vH(v)

4a2 1
Rvv = 2
H(v) + δ(v). (5.89b)
[1 − uH(u) − v] 1 − uH(u)

After the collision, the component Ruv appears, which makes the scalar
field energetically equivalent to a perfect fluid with the energy density and
pressure given by

4a2 4a2
p=ε= 2 = . (5.90)
(1 − u − v)(8a +3)/2 t(8a2 +3)/2

The last equation shows explicitly that this class of solutions develops
a spacetime curvature singularity along the hypersurface t = 0. In fact,
the singularity happens in all the colliding wave models considered in this
section.
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch05 page 152

152 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

Case B: a = δ1 = 0. In this case, we have

 
1−μ
φ = δ ln , δ ≡ δ2 , (5.91)
1+μ
and
Ruu (v < 0) = 4n2 δ 2 [1 − u2n ]−2 u2(n−1) H(u), (5.92a)

Rvv (u < 0) = 4m2 δ 2 [1 − v 2m ]−2 v 2(m−1) H(v), (5.92b)

2
4n2 δ 2 (sin ζ sin ξ)2δ −1 cos2 (ζ + ξ)
Ruv (u, v) = − H(u)H(v), (5.92c)
t2 cos ζ cos ξ
where
1
n=m= ≥ 1. (5.93)
2(1 − δ 2 )
From Eqs. (5.92a)–(5.92c) we find that in this case the solutions represent
the collision of two null dust clouds only. Unlike the last case, the collision
and interaction of the two null dust clouds are such that in the interaction
region (Region IV), the function ε defined by Eq. (5.80) is no longer pos-
itive as follows from Eq. (5.92c). Therefore, in the present case the scalar
field φ in Region IV is energetically equivalent to an anisotropic fluid with
vanishing heat-flow.
Case C: δ12 ≡ δ 2 = 1/2, δ2 = 0. In this case, we have n = m = 1. Restricting
ourselves again to the physically relevant quantities, we find that
 
1−η
φ = 2a ln[1 − u2 H(u) − v 2 H(v)] + δ ln , (5.94)
1+η
and
 2
δ + 2av
Rμν (u < 0) = 4 H(v)n̂μ n̂ν , (5.95a)
1 − v2
 2
δ + 2au
Rμν (v < 0) = 4 H(u)l̂μ ˆlν , (5.95b)
1 − u2
4H(u)H(v)
Rμν (u, v) = [a sin 2ζ + δ cos(ζ − ξ)]
t2 cos ζ cos ξ
[a sin 2ξ + δ cos(ζ − ξ)]. (5.95c)
As shown by Eqs. (5.95a) and (5.95b), the present class of solutions
also represents the collision of two pulses of scalar waves which are free
of impulsive components. On the other hand, Eq. (5.95c) shows that the
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch05 page 153

Collisions of Gravitational Waves with Matter Fields 153

superposition of the incoming waves after collision is such that one cannot
regard the interaction region as consisting of a fluid of a certain type. In
particular, assuming that δa < 0 with |δ/2a| < 1, from Eqs. (5.95a)–(5.95c)
we find that Ruv = 0 along the following two curves in Region IV. The first
is defined by the relation
δ
sin 2ζ = − cos(ζ − ξ), (5.96)
a
and has its origin at the point (u0 , 0) with u0 = |δ/2a|. Similarly, the second
curve is defined by
δ
sin 2ξ = − cos(ζ − ξ), (5.97)
a
and starts at (0, v0 ) where v0 = u0 . Across these curves the sign of
Ruv changes. Therefore, Region IV is split into subregions occupied by
a fluid whose type alternates among the cases (i) and (ii) considered after
Eq. (5.80), as illustrated in Fig. 5.1.
Case D: a = 0, δA = δ = 0. In this case, we find that
(1 − η)(1 − μ)
φ = δ ln , (5.98)
(1 + η)(1 + μ)
and
16n2 δ 2 2 2(n−1)
Ruu = Y u H(u), (5.99a)
t2
4δ 2 2n 1
Rvv = 2 2
u H(u)H(v) + 2n
δ(v), (5.99b)
t Y 1 − u H(u)
8nδ 2 2n−1
Ruv = u H(u)H(v), (5.99c)
t2
which show that in the present case the solutions represent the collision of
a null dust cloud with an impulsive shell of null dust. After the collision,
a current moving along the v = constant lines toward the right-hand side
is developed, which together with the Ruv component given by Eq. (5.99c)
again makes the scalar field in the interaction region energetically equivalent
to a perfect fluid with the equation of state, p = ε.
However, further considerations (Tsoubelis and Wang, 1991) showed
that the origin of the right-hand moving current is difficult to under-
stand if one sticks to the fluid picture, since before the collision there is
no incoming current moving toward the right-hand side as follows from
Eq. (5.99c). Instead, if we consider the question in terms of the scalar field,
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch05 page 154

154 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

from the Klein–Gordon equation (3.80) we can see that this is because of
the backscattering. Actually, in terms of the latter, the present model rep-
resents the collision of a scalar plane wave incident from the right-hand
side, which collides at (u, v) = (0, 0) with an impulsive wave (shell of null
dust). The focusing that results from this encounter is equivalent to the
scalar wave pulse’s entering a spacetime region where the metric depends
on both u and v. But, then the Klein–Gordon equation (3.80) implies that
right moving waves are generated, i.e. backscattering occurs. This explains
the appearance of the right moving current in Region IV in terms of the
null current picture.

5.3. Collision of Neutrino and Electromagnetic Waves

To further discuss the issues of the uniqueness of the initial value problem of
colliding null dust clouds and the gravitational “phase transitions” induced
from the collision of matter fields, in this section we consider the spacetime
described by the metric

ds2 = 2e−M dudv − e−U [(dx2 )2 + (dx3 )2 ], (5.100)

subject to the assumptions

U,u U,v = 0, (5.101)

R22 = R33 = 0, (5.102)
Ruv = Rvu = 0, (5.103)

in order to make the analysis as simple as possible, and meantime the

relevant physics still remains. Then, the combination of Eqs. (3.8a)–
(3.8e), (5.100), (5.103) and the Einstein field equations (1.20) yields

(e−U ),uv = −e−U (U,uv − U,u U,v ) = 0, (5.104)

which, together with Eq. (5.101), gives the following solution:

e−U = α(u) + β(v), (5.105)

where α(u) and β(v) are arbitrary functions of their indicated arguments.
On the other hand, from Eqs. (3.8a)–(3.8e), (4.20a) and (4.20b) we find
that corresponding to the metric (5.100) the condition (5.103) now reads

2M,uv + U,uv = 0. (5.106)

March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch05 page 155

Collisions of Gravitational Waves with Matter Fields 155

For the function U given by Eq. (5.105), we find that Eq. (5.103) has the
general solution
1
M = − U − A(u) − B(v), (5.107)
2
where A, B are other two arbitrary functions of their indicated arguments.
Corresponding to the metric (5.100) with the function M and U given
by Eqs. (5.107) and (5.105), it can be shown that the Ricci tensor Rμν is
given by
α A − α ˆ β  B  − β 
Rμv = l̂μ lν + n̂μ n̂ν , (5.108)
α+β α+β
where the prime denotes the ordinary differentiation with respect to the
indicated argument. Thus, the corresponding energy–stress tensor can be
written in the form
Tμν = Tμν
L
+ Tμν
R
, (5.109)
where
α A − α
Tμν
L
= εL l̂μ ˆlv ≡ l̂μ l̂v , (5.110a)
α+β
β  B  − β 
Tμv
R
= εR n̂μ n̂v ≡ n̂μ n̂ν . (5.110b)
α+β
Equations (5.109) and (5.110a)–(5.110b), together with Eq. (1.21), yield
Tμν
L;ν
= 0 = Tμν
R;ν
. (5.111)
It follows that, assuming both εL and εR be positive, the source given by
Eqs. (5.109) and (5.110a)–(5.110b) consists of a pair of oppositely moving
null dust clouds, one of which has the conserved energy density εL , while
the other has the conserved energy density εR .
In order to consider the solutions representing the collision of two null
dust clouds, let us assume the functions A and B introduced in Eq. (5.107)
such that
A(0) = B(0) = 0. (5.112)
We also assume that α and β, which specify the function U via Eq. (5.105),
are chosen such that
1
α(0) = β(0) = . (5.113)
2
Then, we can see that the conditions (3.22) are now satisfied, and the Khan–
Penrose substitutions (3.11) can be used. Replacing the pair (u, v) by the
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch05 page 156

156 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

one (uH(u), vH(v)) in Eqs. (5.105) and (5.107), we find that the extended
solutions are given by

eA(u)H(u)+B(v)H(v)
e−M = ,
[1 − f (u)H(u) − g(v)H(v)]1/2 (5.114)
e−U = 1 − f (u)H(u) − g(v)H(v),

where

f (u) = α(0) − α(u), g(v) = β(0) − β(v). (5.115)

Consequently, the corresponding energy–stress tensor is now given by

Tμν = eU [(f  − f  A )H(u)l̂μ l̂v + (g  − g  B  )H(v)n̂μ n̂ν

+ f  δ(u)l̂μ l̂v + g  δ(v)n̂μ n̂ν ]. (5.116)

The only non-vanishing Weyl scalar, on the other hand, is given by

1 1
Ψ2 = (M,uv − U,uv ) = − e2U f  g  H(u)H(v), (5.117)
2 4
which means that all of the models under consideration are conformally
flat, except for the ones in which f  g  = 0.
Now let us turn to prove that the two non-impulsive terms appearing in
Eq. (5.116) for Tμν can be interpreted as representing an electromagnetic
plane wave and a neutrino plane wave, respectively. To this purpose, let us
consider the antisymmetric tensor Fμν , which is defined by

Fμν = 2Φ0 n̂[μ mν] + 2Φ0 n̂[μ mν] , (5.118)

where Φ0 = Φ0 (u, v), and mμ is given by

1
mμ = √ e−U/2 (δμ2 + iδμ3 ). (5.119)
2

Provided that it satisfies the Maxwell equation (1.96), Fμν represents a null
electromagnetic field with the energy–stress tensor Tμν given by

1
Tμν
e
= Fμλ F λ ν − gμν Fρλ F λρ = 2Φ0 Φ0 n̂μ n̂ν . (5.120)
4
Comparing Eq. (5.116) with Eq. (5.120), we conclude that the second term
in the right-hand side of Eq. (5.116) can be attributed to an electromagnetic
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch05 page 157

Collisions of Gravitational Waves with Matter Fields 157

wave, if the function Φ0 obtained from

1 U 
Φ0 Φ0 = e (g − g  B  )H(v), (5.121)
2
satisfies the Maxwell equation (1.96) through Eq. (5.118). In fact, choosing
Φ0 such that
 1/2
1 U   
Φ0 = e (g − g B ) H(v), (5.122)
2
we find that both Eq. (5.120) and the Maxwell field equation (1.96) are
satisfied in all the four regions (Regions I–IV) as well as on the separation
null hypersurfaces Σu and Σv . Substituting Eq. (5.122) into Eq. (5.118), we
find
 1 2
Fμλ = 2 g  − g  B  H(v)δ[μ δν] . (5.123)
On the other hand, if we set Ψ = 0 in Eq. (1.102), we can show that the
only non-vanishing Ricci scalar for a neutrino field is Φ22 [see Eq. (1.105)].
Introducing the function Φ(0) by
Φ = A1/2 Φ(0) , (5.124)
where A is the function introduced in the null tetrad (2.23) related to
the function M via Eq. (2.25), we find that the Weyl neutrino equa-
tions (1.104a) and (1.104b) reduce to a single equation,
1
Φ(0)
U,v Φ(0) ,
,v = (5.125)
2
and that the corresponding energy–stress tensor takes the form,
Tμν
n
= i2[Φ(0) Φ(0) ,u − Φ(0) Φ(0) ,u ]l̂μ l̂ν . (5.126)
The solution of Eq. (5.125) can be written in the form
Φ(0) = eU/2 ρ(u)eiθ(u) . (5.127)
A particular solution of the functions ρ and θ is given by
1 
ρ(u) = (f − f  A )1/2 H(u), θ(u) = u. (5.128)
2
The combination of Eqs. (5.127), (5.128) and (5.126) yields,
Tμν
n
= eU (f  − f  A )H(u)l̂μ ˆlν . (5.129)
This expression now can be compared with the first term on the right-
hand side of Eq. (5.116), from which we conclude that this term can
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch05 page 158

158 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

be considered as representing a neutrino plane wave, which, according to

Eqs. (1.102), (5.124), (5.127) and (5.128), is given by
1
[A(f  − f  A )eU ]1/2 H(u)eiu OA .
φA = (5.130)
2
Having completed the proof that the above extended solutions can be
interpreted as representing the collision of a pulse of electromagnetic radi-
ation incident from the left in Fig. 5.1 with a pulse of neutrino radiation
incident from the right, let us turn to consider some specific examples.
We first choose the functions f and g to be of the forms,
f (u) = u2n , g(u) = v 2m , (5.131)
where n and m satisfy conditions (5.78). Then, we choose the functions A
and B to be of the forms,
A = p ln(1 − un ) + q ln(1 + un ),
(5.132)
B = r ln(1 − v m ) + s ln(1 + v m ),
with the tetrad of constants (p, q, r, s) being chosen so that
1
,
p = q ≥ 0, when n =
2
(5.133)
1
r = s ≥ 0, when m = .
2
Otherwise, these four constants have to be chosen so that the following
conditions always hold for any x ∈ (0, 1):
F (x) = [n(p + q) − (2n − 1)]x2n + n(p − q)xn + 2n − 1 ≥ 0,
(5.134)
G(x) = [m(r + s) − (2m − 1)]x2m + m(r − s)xm + 2m − 1 ≥ 0.
Then, Eqs. (5.132)–(5.134) define a six-parameter family of solutions
corresponding to sources described by the energy–momentum tensor,
2mv 2m−1 δ(v) 2nu2n−1 δ(u) ˆ
Tμν = Tμν
e
+ Tμν
n
+ 2n
n̂μ n̂ν + l̂μ lν , (5.135)
1 − u H(u) 1 − v 2m H(v)
where
2mv 2m−2 G(v)H(v)
Tμν
e
= n
μ n̂ν , (5.136)
(1 − v 2m )[1 − u2n H(u) − v 2m ]
2nu2n−2 F (u)H(u)
Tμν
n
= l̂μ l̂ν , (5.137)
(1 − u2n )[1 − u2n − v 2m H(v)]
with G(v) and F (u) being given by Eq. (5.134). The non-impulsive parts
of the above sources can be attributed to a pulse of electromagnetic wave
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch05 page 159

Collisions of Gravitational Waves with Matter Fields 159

and a pulse of neutrino wave given, respectively, with the field strengths,
 1/2
2mG(v) 2
Fμν = 2v m−1 H(v)n[μ δν] , (5.138)
1 − v 2m
 1/2
un−1 2nAF (u)eU
φA = H(u)eiu OA . (5.139)
2 1 − u2n
The above class of solutions covers a large variety of physically inter-
esting cases, which are obtained by choosing the corresponding parameters
appropriately. When the tetrad (p, q, r, s) vanishes and n = m = 1, for
example, we recover Griffiths’ solution (Griffiths, 1976b), which represents
the collision of a pair of constant profile shock waves made of photons
and neutrinos, respectively. When n = 2 and m = 1, the corresponding
solutions represent the collision of an electromagnetic shock wave and a
neutrino wave with smooth wavefront, etc.
Without going to more detailed analyses for each of specific models, we
turn to the issues, regarding to the ambiguity of the outcome of collisions
of matter fields, raised in Section 5.2. Choosing the parameters p, q, r, s, n
and m as

p = (2a + δ+ )2 , q = (2a − δ+ )2 , r = (2a + δ− )2 , s = (2a − δ− )2 ,

n = [2(1 − δ+ )2 ]−1 , m = [2(1 − δ− )2 ]−1 , (5.140)

where a and δA will be chosen as the same constants that specify the
family of scalar wave solutions analyzed in the last section, we obtain a
three-parameter family of the solutions given by Eqs. (5.131)–(5.134). It
is then easy to verify that in the pre-collision regions (Regions I–III), the
metric resulting from the choice of Eq. (5.140) is identical to the one cor-
responding to the scalar wave models studied in Section 5.2, even though
the solutions differ from each other in the interaction region (Region IV).
Thus, Taub’s results about the ambiguity from planar colliding null dust
clouds with impulsive waves now extend to solutions that are free from
such impulsive components. That is, given the metric (solutions of the
Einstein field equations) in Regions I–III, one would not be able to pred-
icate what is the outcome of the collision. The colliding process becomes
determinative, however, once the energy–stress tensor or equivalently, the
type of interaction is pre-assigned in Region IV. In the present case, the
resolution of the ambiguity was resolved by constructing the corresponding
energy–stress distributions from the well-defined matter fields, neutrino and
electromagnetic.
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch05 page 160

160 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

5.4. Collisions of Two Electromagnetic Plane Waves

The studies of collisions of two electromagnetic plane waves can be traced
back to the seminal work of Bell and Szekeres (1974), in which they
presented a solution that represents the collision of two pure plane elec-
tromagnetic shock waves, and the metric in the interaction region takes the
form,

cos(au − bv)
ds2IV = 2dudv − [1 − sin2 (au) − sin2 (bv)] (dx2 )2
cos(au + bv)

cos(au + bv)
+ (dx3 )2 , (5.141)
cos(au − bv)

where a and b are two real constants. It is interesting to note that in the cur-
rent case the focusing surface, 1−sin2 (au)−sin2 (bv) = 0 (or au + bv = π/2),
is not singular, but a Cauchy horizon (Bell and Szekeres, 1974). Clarke and
Hayward (1989) showed that extensions beyond this surface are not unique,
and similar to the vacuum case (Yurtsever, 1987), it was argued that such
a horizon is also not stable (Konkowski and Helliwell, 1991; Gürses and
Halilsoy, 1982; Gurtug and Halilsoy, 2000). Another interesting aspect of
the solution is that metric (5.141) is conformally flat,

q2 2
ds2IV = [dt − dr2 − r2 (dθ2 + sin2 θdφ2 )], (5.142)
r2
where 2ab = q −2 , and the mapping between the two sets of coordinates is
given explicitly by (Griffiths, 1991),
 
1 y
t + r = coth sech−1 cos(au + bv) − ,
2 2q
 
1 −1 y
t − r = −tanh sech cos(au + bv) + , (5.143)
2 2q
π 1
θ = au − bv + , φ= x.
2 2
Using the Khan–Penrose substitutions, one can easily find the metric in
Regions I–III,
⎧
⎨ 2dudv − cos2 (au)dΣ20 , Region III,
2
ds = 2dudv − cos2 (bv)dΣ20 , Region II, (5.144)
⎩
2dudv − dΣ20 , Region I,
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch05 page 161

Collisions of Gravitational Waves with Matter Fields 161

where dΣ20 ≡ (dx2 )2 + (dx3 )2 . Clearly, Region I is flat, while in each of the
two regions (Regions II and III) an electromagnetic shock wave is present,
Φ0 = bH(v), Φ2 = aH(u), (5.145)
while all the Weyl scalars ΨA (A = 0, 2, 4) vanish in these regions as well as
across their boundaries u = 0 and v = 0. Thus, the Bell–Szekeres solution
represents the collision of two pure electromagnetic plane waves. However,
as in the mixed cases presented in the last section, two impulsive gravita-
tional waves are created due to the collision, one moves along the hyper-
surface u = 0 and the other along the one v = 0 (Bell and Szekeres, 1974),
Ψ0 = −b tan(au)H(u)δ(v), Ψ4 = −a tan(bv)H(v)δ(u). (5.146)
In the general case, after Chandrasekhar and Xanthopoulos (1985a)
laid down the foundation of the problem, various solutions were obtained
subsequently (Griffiths, 1985, 1990; Halilsoy, 1988a, 1988b, 1989a, 1989b;
Garcı́a-Dı́az, 1988, 1989; Papacostas and Xanthopoulos, 1989; Li and Ernst,
1989; Bretón et al., 1998; Hogan, Barrabés and Bressange, 1998; Hogan
and Walsh, 2003). In particular, Chandrasekhar and Xanthopoulos (1985a)
showed explicitly how one can generalize the Khan–Penrose and Nutku–
Halii solutions to include electromagnetic plane shock waves, in which a
spacetime curvature singularity is finally formed on the focusing surface.
They also provided the generalization of the Bell–Szekeres solution to the
case with two non-collinearly polarized electromagnetic plane shock waves
(Chandrasekhar and Xanthopoulos, 1987b), and showed that a Cauchy hori-
zon is finally formed on the focusing surface.

5.5. Other Colliding Plane Waves

In the last three decades or so, colliding plane waves have been studied
extensively in various contents and theories of gravity (Griffiths, 1991;
Stephani et al., 2009; Griffiths and Podolsky, 2009). These include the dila-
ton gravity (Schwarz, 1997; Gürses and Sermutlu, 1995; Bretón, Matos
and Garcı́a, 1996; Gurtug, Halilsoy and Sakalli, 2003; Halilsoy and Sakalli,
2003), and high-dimensional spacetimes in the framework of string theory
(Mizoguchi, 1996; Feistein, 2002; Gutperle and Pioline, 2003; Chen, 2004;
Chen et al., 2004; Chen and Zhang, 2005; Tziolas, Wang and Wu, 2009),
as well as in the brane-worlds (de Leon, 2004; Tziolas and Wang, 2008).
Many interesting results were obtained, including the possibility of disap-
pearing of spacetime curvature singularities when left to higher dimensional
spacetimes (Tziolas, Wang and Wu, 2009).
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch05 page 162

162 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

In the framework of string/M -theory, the AdS/CFT correspondence

was investigated in the spacetimes of colliding gravitational shock waves
(Rosenbaum et al., 1986; Grumiller and Romatschke, 2008; Álvarez-Gaumé,
et al., 2009; Dueñas-Vidal and Vázquez-Mozo, 2010, 2012; Chesler and
Yaffe, 2011; Chesler, 2015).
In addition, the collision of plane-fronted gravitational waves was also
studied in curved backgrounds rather than the Minkowski. These include
an expanding universe (Centrella and Matzner, 1982; Alekseev, 2016) and
asymptotically flat spacetimes (Pretorius and East, 2018). Among other
properties, it was found that a black hole can be formed due to the mutual
focus of these plane-fronted waves (Pretorius and East, 2018), despite the
fact that the total energy of the incoming waves now can be finite, in
contrast to the uniform plane-fronted case, in which the total energy of the
incoming waves is always infinitely large.
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch06 page 163

Chapter 6

Isometries to Interiors of Black Holes

In the past three decades or so, the collisions of gravitational plane waves
coupled with various matter fields have been extensively studied, and very
important insight about the nonlinearity of the Einstein field equations has
been obtained.
One of the remarkable features is that the internal spacetimes of all
the known black holes (Stephani et al., 2009) are isometric to the interact-
ing regions of two colliding plane gravitational waves with or without the
presence of matter fields. The fundamental reason is that in both cases the
Einstein field equations can be written in terms of the Ernst equation (3.44)
(Chandrasekhar and Ferrari, 1984).1 So, in this chapter we shall present a
brief review on these fascinating phenomena.

6.1. The Schwarzschild Black Hole

The internal spacetime of the Schwarzschild black hole is locally isometric
to the interacting region of two pure plane gravitational impulsive + shock
waves (Ferrari and Ibañez, 1987a, 1987b; Yurtsever, 1988a, 1988b; Tsoubelis
and Wang, 1989; Griffiths, 1991). In fact, setting
1
a=− , δ1 = 1, δ2 = 0, (6.1)
2

1 In the cylindrically symmetric case, when the Abelian G2 group acts orthogonally tran-
sitively, the metric can be written in the form of Eq. (3.41), but with the x3 -coordinate
having closed orbits (Bronnikov, Santos and Wang, 2019). Then, the Einstein vacuum
field equations reduce to the “cylindrical” Ernst equation (Chandrasekhar, 1986). There-
fore, such isometries should exist between cylindrical spacetimes (Wang, 2003) and the
interiors of black holes, too.

163
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch06 page 164

164 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

the solutions (4.23) reduce to

   
dη 2 dμ2 1−η
ds2 = C0 (1 + η)2 2
− − (dx2 )2
1−η 1 − μ2 1+η
− (1 − μ2 )(1 + η)2 (dx3 )2 . (6.2)

Introducing the new coordinates (Chandrasekhar and Xanthopoulos, 1986a,

1986b; Ferrari and Ibañez, 1988),

r−m
η≡ , μ ≡ cos θ, x2 ≡ t, x3 ≡ mφ, C0 = m2 , (6.3)
m
we find that the metric (6.2) takes the form,
   −1
2 2m 2 2m
ds = 1 − dt − 1 − dr2 − r2 (dθ2 + sin θ2 dφ2 ), (6.4)
r r

which is precisely the form of the Schwarzschild black hole solution written
in the spherically symmetric coordinates (t, r, θ, φ). Note that the focusing
hypersurface η = 1 or t = 0 corresponds to the Killing-horizon r = 2m, in
other words, now the focusing hypersurface is not singular, and instead it
represents a horizon. This is consistent with the analysis presented between
Eqs. (4.45) and (4.47). In particular, the choice of Eq. (6.1) satisfies the
non-singular condition (4.47).
It should be noted that the above equivalence holds only locally, as one
can see from Eq. (6.3), which tells us that it is the x2 -coordinate that is
identical to the t-coordinate, while the space-like coordinate μ is identical
to the angular coordinate φ, and the time-like coordinate η to r, as inside
the black hole, r becomes time-like, and t is space-like.
On the other hand, from Eq. (4.28) we find that the choice of
Eq. (6.1) yields,

m = n = 1, (6.5)

for which the incoming gravitational waves consist of two parts, the impul-
sive and shock parts, as one can see from the analysis given between
Eqs. (4.34) and (4.42). In particular, along the wavefronts u = 0 and
v = 0, the two wave components Ψ0 and Ψ4 takes the forms,

Ψ0 = −3H(v) + δ(v), (across v = 0),

(6.6)
Ψ4 = −3H(u) + δ(u), (across u = 0).
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch06 page 165

Isometries to Interiors of Black Holes 165

Inside Region II (III), a right-hand (left-hand) moving gravitational shock

wave is present, given, respectively, by

3
ΨII
0 (v) = , (Region II),
(1 + v)2 (1 − v)
(6.7)
3
ΨIII
4 (u) = , (Region III).
(1 + u)2 (1 − u)

It is equally remarkable that the following specification of the free

parameters also leads to the internal solution of the Schwarzschild black
hole. In fact, setting (Ferrari and Ibañez, 1988),

1
a=+ , δ1 = 1, δ2 = 0, (6.8)
2

in the solutions given by Eq. (4.23), we find that

   
2 2 dη 2 dμ2 1+η
ds = C0 (1 − η) − − (dx2 )2
1 − η2 1 − μ2 1−η

− (1 − μ2 )(1 − η)2 (dx3 )2 , (6.9)

which can be obtained from the metric (6.2) by the replacement η → −η.
However, now the coordinate transformations (6.3) to the internal region
of the Schwarzschild black hole solution (6.4) become,

m−r
η≡ , μ ≡ cos θ, x2 ≡ t, x3 ≡ mφ, C0 = m2 , (6.10)
m

that is, all of them are the same as these given by Eq. (6.3) except for
η → −η. It is this difference that now the focusing hypersurface t = 0 or
η = 1 corresponds to r = 0, instead of r = 2m, as given in the last case.
Therefore, in the present case the hypersurface t = 0 (η = 1) becomes
a spacetime curvature singularity. Note that this is also consistent with
the analysis given between Eqs. (4.45) and (4.47), and in particular the
non-singular condition (4.47) is not satisfied.
From Eq. (4.28) we find that the choice of Eq. (6.8) also yields m = 1 =
n, and along the wavefronts u = 0 and v = 0, the two gravitational wave
components Ψ0 and Ψ4 take the forms,

Ψ0 = 3H(v) + δ(v), (across v = 0),

(6.11)
Ψ4 = 3H(u) + δ(u), (across u = 0).
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch06 page 166

166 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

Inside Region II (III), a right-hand (left-hand) moving gravitational shock

wave is present, but now given, respectively, by
3
ΨII
0 (v) = , (Region II),
(1 − v)2 (1 + v)
(6.12)
3
ΨIII
4 (u) = , (Region III).
(1 − u)2 (1 + u)
Comparing Eq. (6.6) with Eq. (6.11), we can see that the shock wave
part changes the signs, while their impulsive parts remain the same. On
the other hand, Eqs. (6.7) and (6.12) show that the shock waves along
the hypersurfaces u = 1 and v = 1 become more singular in the latter,
proportional to, respectively, (1 − u)−2 and (1 − v)−2 , while in the former,
they are proportional to, respectively, (1 − u)−1 and (1 − v)−1 .

6.2. The Kerr–NUT Black Hole

Chandrasekhar and Ferrari (1984) first showed that the Nutku–Halil solu-
tion (Nutku and Halil, 1977) is given by the Ernst potential [see Eq. (4.141)],

E = pη + iqμ, (p2 + q 2 = 1), (6.13)

where E is directly related to the metric coefficients V and W via the

relations [see Eq. (3.44)],

Z −1 eV sinh W eV
Z ≡ χ + iq2 , E≡ , χ≡ , q2 ≡ . (6.14)
Z +1 cosh W cosh W
The Nutku–Halil solution represents the collision of two pure gravitational
impulsive waves with uncorrelated (non-collinear) polarizations. It is a
direct generalization of the Khan–Penrose solution (Khan and Penrose,
1971), which represents the collision of two pure gravitational impulsive
waves with correlated (collinear) polarizations, and the corresponding Ernst
potential is given by Eq. (6.13) with q = 0.2
As first noticed by Chandrasekhar and Ferrari (1984), the Kerr solution
(Kerr, 1963) also follows from the same Ernst potential (6.13) (Chan-
drasekhar, 1983). But, in the stationary case the Ernst potential is related
to the metric coefficients through the function Φ̃ and Ψ̃ (Chandrasekhar,

2 Barrabés, Bressange and Hogan showed that the Khan–Penrose and Nutku–Halil solu-

tions can be also obtained from Einstein’s vacuum field equations as an initial value
problem (Barrabés, Bressange and Hogan, 1999).
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch06 page 167

Isometries to Interiors of Black Holes 167

1983, Section 54),

Z = Ψ̃ + iΦ̃, (6.15)
where Φ̃ is the potential for q2 , defined similar to Q given by Eq. (4.8), and
Ψ̃ similar to P defined by Eq. (4.10).
Therefore, to get something similar to the Kerr solution, we need first
to solve Eq. (4.8) for q2 , and then read off χ from Eq. (4.10) with
1 + E†
P + iQ = , (6.16)
1 − E†
where E † = pη+iqμ. In doing so, Chandrasekhar and Xanthopoulos (1986a,
1986b) found that
(1 − pη)2 + q 2 μ2
χ = (1 − η 2 )1/2 (1 − μ2 )1/2 ,
1 − p2 η 2 − q 2 μ2
2q (1 − η)(pμ2 + pη + μ2 − 1)
q2 = , (6.17)
1+p 1 − p2 η 2 − q 2 μ2
(1 − pη)2 + q 2 μ2
f = .
η 2 − μ2
To extend the above solution to Regions I–III, one can use the Khan–
Penrose substitutions, and then it can be shown that the solution rep-
resents the collision of two pure gravitational impulsive waves, each of
which is accompanied by a gravitational shock wave (Chandrasekhar and
Xanthopoulos, 1986a, 1986b).
In the interacting region (Region IV), setting (Chandrasekhar and
Xanthopoulos, 1986a, 1986b)
M −r
η = ∓√ , μ = cos θ,
M2 − J2
 
2q M
t = M x2 − x3 , φ = √ x3 , (6.18)
p(1 + p) M2 − J2
√
M2 − J2 J
p=∓ , q=± ,
M M
the solution can be cast in the form,
 2
ρ2 − 2M r 2JM r sin2 θ ρ2
ds2 = 2
dt + 2
dφ − [dr2 + Δdθ2 ]
ρ ρ − 2M r Δ
Δρ2 sin2 θ 2
− dφ , (6.19)
ρ2 − 2M r
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch06 page 168

168 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

where

Δ ≡ r2 − 2M r + J 2 , ρ≡ r2 + J 2 cos2 θ. (6.20)
The above solution is exactly the Kerr solution written in the Boyer–
Lindquist coordinates with M and J being the mass and angular momen-
tum parameters (Chandrasekhar, 1983).
Since −1 ≤ η ≤ +1, from Eq. (6.18) we find that this implies

r− ≤ r ≤ r+ , (r± = M ± M 2 − J 2 ), (6.21)
that is, the interacting region of the two pure colliding plane gravitational
waves is isometric to the ergo-sphere of the Kerr black hole, in which the
time-like Killing vector ∂t becomes space-like (in addition to the space-like
Killing vector ∂φ ). As a result, in this region there are two space-like Killing
vectors, which are consistent with the requirement of spacetimes with plane
symmetry (Stephani et al., 2009), although in the Kerr spacetime, one of
them has a closed orbit [cf. Eq. (6.18)]. Therefore, such an equivalence can
be only local.
On the other hand, setting η = 1, which corresponds to the focusing
hypersurface t = 0, from Eq. (6.18) we find that

r(t = 0) = M ± M 2 − J 2 . (6.22)
Thus, if the “−” sign in Eq. (6.18) is chosen, the focusing surface corre-
sponds to the null surface r = r+ , which is the location of the event horizon
of the Kerr black hole, and when the “+” sign in Eq. (6.18) is chosen, the
focusing surface corresponds to the null surface r = r− , which is the loca-
tion of the Cauchy horizon of the Kerr black hole. In each of the two cases,
the spacetime is not singular at t = 0, and extensions beyond these surfaces
are needed. Clearly, if we require the extension is analytical, then we shall
get a global structure quite similar to that of the Kerr black hole. However,
in the case r(t = 0) = r− the focusing hypersurface is a Cauchy horizon,
which is not stable against small perturbations (Yurtsever, 1987, 1988a),
so such extensions might not be needed in this case.
It is interesting to note that the solution (6.17) is precisely the two-
soliton solution with
1
a = , δ1 = δ2 = 0, ⇒ n = m = 1, (6.23)
2
given by Eqs. (4.142) and (4.143) by setting the NUT parameter l to zero.
When l = 0, the interacting region of the colliding plane gravitational
wave is isometric to the interior of the Kerr–NUT solution. When q = 0,
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch06 page 169

Isometries to Interiors of Black Holes 169

the solution reduces to that given by Eq. (4.145), in which the interacting
region is isometric to the interior of the Taub–NUT solution, first studied
by Ferrari and Ibañez (1988).
Finally, we note that in the current case we have α ≡ a + δ1 = 1/2,
which satisfies one of the conditions given in Eq. (4.134), and as a result
the corresponding solution is not singular at the focusing surface t = 0.
This is well consistent with the above analysis.

6.3. Other Black Holes

From the analysis given in the last two sections, we can see that, for any
given stationary spacetime, M , if it contains at least one more space-like
Killing vector, say, ∂x , then this Killing vector will form a G2 group together
with the time-like Killing vector ∂t . In such a space-time, if we further
assume that a black hole exists, then by definition this time-like Killing
vector ∂t will become space-like inside the black hole (or at least inside
a region of the black hole, such as that in the case of the Kerr solution
considered in the last section). Then, in this region the two space-like Killing
vectors can be (locally) identified with the two space-like Killing vectors ∂x2
and ∂x3 for the colliding plane wave space-times. Restricting oneself to this
internal region, and then using the Khan–Penrose substitutions
u → uH(u), v → vH(v), (6.24)
one can obtain the spacetime in Regions I–III. Then, such resulted space-
time will represent the collision of gravitational plane waves (possibly cou-
pled with matter), provided that certain physical conditions are satisfied,
such as the energy conditions (Hawking and Ellis, 1973), and satisfying
matter field equations. Clearly, such an obtained solution is isometric to
the interior of the black hole.
Along this vein, it can be shown that the interior of the Kerr–Newman
charged black hole is isometric to the interacting region of two colliding
gravitational plane waves, each of which is coupled with an electromag-
netic wave (Chandrasekhar and Xanthopoulos, 1987a, 1987b). Along the
same line, it was found that the interior of the Kerr–Newman–NUT solution
(Sephani et al., 2009) is also isometric to an interacting region of two col-
liding gravitational plane waves coupled with electromagnetic plane waves
(Gurtug and Halilsoy, 2009).
Similarly, Papacostas and Xanthopoulos (1989) found a five-parameter
class of solutions, which describe the collision of plane-fronted impul-
sive gravitational and shock electromagnetic waves. Again, the interaction
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch06 page 170

170 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

region of the two colliding waves is a locally known solution of the Einstein–
Maxwell electrovacuum equations of Petrov type D, and the collision results
in the formation of a Cauchy horizon.
In addition, Griffiths and Halburd (2007) showed that part of the
C-metric spacetime inside the black hole can be also interpreted as the
interaction region of two colliding plane waves with aligned linear (collinear)
polarizations. The focusing surface t = 0 is not singular, and instead repre-
sents a Cauchy horizon. Note that the C-metric was first found by Weyl in
1917 (Weyl, 1917), and subsequently rediscovered by many authors. But,
it was Kinnersley and Walker (1970) and Bonnor (1983) who first showed
that its analytic extension represents a pair of black holes that accelerate
away from each other due to the presence of a strut along the symmetry
axis.
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch07 page 171

Chapter 7

Concluding Remarks

In this book, we study the collision and nonlinear interaction of two pure
gravitational plane waves, or a gravitational plane wave with a matter wave,
or two matter waves, within the framework of Einstein’s theory of general
relativity, by using the distribution theory, first introduced by Taub to gen-
eral relativity in 1980 (Taub, 1980). The method is mathematically equal
to Israel’s junction conditions (Israel, 1966, 1967), when the singular hyper-
surface is time-like or space-like. When the surface is null, it is equivalent to
the analysis of Barrabés (see Barrabés, 1989; Barrabés and Hogan, 2003).
Thus, they can be considered as complementary to each other.
The advantage of the studies of spacetimes for colliding gravitational
and/or matter plane waves is that they can be investigated analytically,
thanks to the development of nonlinear differential equations in 1960s
(Gardner, Green, Kruskal and Miura, 1967; Whitham, 1974), and appli-
cations to Einstein’s general relativity in 1970s and 1980s (Belinsky and
Verdaguer, 2001; Griffiths and Podolský, 2009; Stephani et al., 2009),
including the soliton technique of Blelinsky and Zakharov (1978, 1979),
its generalization to Einstein–Maxwell equations (Alekseev, 1981), and
the Bäcklund transformations (Harrison, 1978, 1980; Neugebauer, 1979).
With these techniques, most of the well-known solutions were rediscov-
ered, including the Kerr–Newman–NUT black hole solutions (Belinsky and
Verdaguer, 2001).
In fact, such techniques can be applied to all the spacetimes with two
orthogonal Killing vectors, including stationary axially symmetric space-
times (Griffiths and Podolský, 2009; Stephani et al., 2009), cylindrical
spacetimes (Verdaguer, 1993; Bronnikov, Santos and Wang, 2019), and
spacetimes with plane symmetry, studied in this book.

171
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch07 page 172

172 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

With the studies of exact solutions that represent colliding plane wave
spacetimes, now it is clear that spacetime singularities are generically devel-
oped due to the nonlinear interaction of the two colliding plane waves,
as shown explicitly in Chapters 4 and 5. Killing–Cauchy horizons can be
formed in particular cases (Chandrasekhar and Xanthopoulos, 1986a), but
they are not stable against small perturbations (Yurtsever, 1987; Clarke
and Hayward, 1989; Griffiths, 2005), and are expected to be turned into
real spacetime singularities.
One might think that this is due to the high symmetry of the spacetimes,
and in particular because of the plane symmetry, the incoming waves always
have infinitely large amount of energy. However, recently it was showed
that, due to such nonlinear interactions, spacetime singularities can be still
formed in the head-on collision of axisymmetric distributions of null parti-
cles even with finite energy (Pretorius and East, 2018).
This result is extremely encouraging, and motivating various interest-
ing questions. One is regarding to the internal structure of black holes. As
shown in Chapter 6, the interiors of the most well-known black holes are iso-
metric to the interacting regions of two colliding plane waves. Is it possible
to shed lights on the internal structure of black holes by studying the head-
on collision of two gravitational and/or matter wave? What about cosmic
censorship and hoop conjectures (Penrose, 1969; Thorne, 1972)? Can we
also be able to say something about the critical phenomena of gravitational
collapse in the threshold of black hole formation (Choptuik, 1993; see also
the review articles, Wang, 2001; Gundlach and Martin-Garcia, 2007).
To be more realistic, one might like first to remove the assumption
of uniform plane-fronted waves, so one can deal with the case in which
incoming waves have finite energy, similar to what were done by Pretorius
and East (2018), and more recently by Baumgarte, Gundlach and Hilditch
(2019). Perturbative calculations of this kind already started in the early
of 1990s by D’Eath and Payne (1992a, 1992b, 1992c), and such obtained
resulted are quite consistent with these numerical ones (Pretorius and East,
2018). Recently, such studies have been also generalized to other cases,
including that the background is an expanding universe (Centrella and
Matzner, 1982; Alekseev, 2016).
The above works have been mainly related to some theoretical aspects
of Einstein’s theory. With entering the era of gravitational wave astron-
omy, one might like also to ask more observational ones, such as the detec-
tion of gravitational memory effects (Favata, 2010; Bieri, Garfinkle and
Yunes, 2017), just briefly mentioned in Section 2.5, in addition to the
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-ch07 page 173

Concluding Remarks 173

more theoretical works of soft-graviton theorems (Hawking, Perry and Stro-

minger, 2016, 2017; Strominger, 2017).
In Chapter 3, we find that, due to the nonlinear interaction, the polar-
ization of a colliding gravitational plane wave can be also get changed, a
precise analog of the electromagnetic Faraday rotation (Faraday, 1846a,
1846b), but now with the oppositely moving wave, either gravitational or
matter, as both the magnetic field and medium. Then, a natural question
raises: is it possible in the future to observe such phenomena? The answer
must be positive, but the real question is when?
With the detections of gravitational waves by LIGO/Virgo (Abbott
et al., 2019), which will be jointed soon by KAGRA (Akutsu et al., 2019),
and the direct detection of the supermassive M87 black hole by EHT
(Akiyama et al., 2019), two unprecedented new windows to study the strong
gravitational field regime have just opened, and indeed we are entering
the epoch of exploring the nonlinear nature of Einstein’s general relativ-
ity, the characteristics that bare stark contrast to Newtonian theory, not
only theoretically but also experimentally! All these just mark the begin-
ning of a new era, in which “the possibilities for discovery are as rich and
boundless as they have been with light-based astronomy.”
 b2530   International Strategic Relations and China’s National Security: World at the Crossroads

This page intentionally left blank

b2530_FM.indd 6 01-Sep-16 [Link] AM

March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-bibliography page 175

Bibliography

Abbott, B.P. et al. (LIGO Scientific Collaboration and Virgo Collaboration).

(2016). Observation of gravitational waves from a binary black hole merger,
Phys. Rev. Lett. 116, 061102.
Abbott, B.P. et al. (LIGO Scientific Collaboration and Virgo Collaboration).
(2017). GW170817: Observation of gravitational waves from a binary neutron
star inspiral, Phys. Rev. Lett. 119, 161101.
Abbott, B.P. et al. (LIGO Scientific Collaboration and Virgo Collaboration).
(2019). GWTC-1: A gravitational-wave transient catalog of compact binary
mergers observed by LIGO and Virgo during the first and second observing
runs, Phys. Rev. X 9, 031040.
Ade, P.A.R. et al. (2016). Planck 2015 results. XIII. Cosmological parameters,
Astron. Astrophys. 594, A13.
Aichelburg, P.C. and Sexl, R.U. (1971). On the gravitational field of a massless
particle, Gen. Relativ. Grav. 2, 303.
Akiyama, K. et al. (The Event Horizon Telescope Collaboration). (2019). First
M87 event horizon telescope results. I. The shadow of the supermassive black
hole, Astrophys. J. Lett. 875, L1.
Akutsu, T. et al. (KAGRA Collaboration). (2019). KAGRA: 2.5 generation inter-
ferometric gravitational wave detector, Nature Astronom. 3, 35.
Alekseev, G.A. (1981). N-soliton solutions of Einstein–Maxwell equations, JETP
Lett. 32, 277.
Alekseev, G.A. (2016). Collision of strong gravitational and electromagnetic waves
in the expanding universe, Phys. Rev. D 93, 061501(R).
Alekseev, G.A. and Griffiths, J.B. (2001). Solving the characteristic initial value
problem for colliding plane gravitational and electromagnetic waves, Phys.
Rev. Lett. 87, 221101.
Alekseev, G.A. and Griffiths, J.B. (2004). Collision of plane gravitational and elec-
tromagnetic waves in a Minkowski background: Solution of the characteristic
initial value problem, Class. Quantum Grav. 21, 5623.
Álvarez-Gaumé, L. et al. (2009). Critical formation of trapped surfaces
in the collision of gravitational shock waves, J. High Energy Phys.
02, 009.
Apostolatos, T.A. and Thorne, K.A. (1992). Rotation halts cylindrical, relativistic
gravitational collapse, Phys. Rev. D 46, 2435.

175
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-bibliography page 176

176 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

Arkani-Hamed, N. Dimopoulos, S. and Dvali, G. (1998). The hierarchy problem

and new dimensions at a millimeter, Phys. Lett. B 429, 263.
Arkani-Hamed, N. Dimopoulos, S. and Dvali, G. (1999). Phenomenology, astro-
physics, and cosmology of theories with submillimeter dimensions and TeV
scale quantum gravity, Phys. Rev. D 59, 086004.
Babala, D. (1987). Collision of a gravitational impulsive wave with a shell of null
dust, Class. Quantum Grav. 4, L89.
Baldwin, O.R. and Jeffery, G.B. (1926). The relativity theory of plane waves,
Proc. R. Soc. A 111, 95.
Barrabés, C. (1989). Singular hypersurfaces in general relativity: A unified
description, Class. Quantum Grav. 6, 581.
Barrabés, C. Bressange, G.F. and Hogan, P.A. (1999). Colliding plane impulsive
gravitational waves, Prog. Theor. Phys. 102, 1085.
Barrabés, C. and Hogan, P.A. (2003). Singular Null Hypersurfaces in General
Relativity: Light-like Signals from Violent Astrophysical Events (World Sci-
entific, Singapore).
Barrabés, C. and Hogan, P.A. (2014). Generating a cosmological constant with
gravitational waves, Gen. Relativ. Grav. 46, 1635.
Barrabés, C. and Hogan, P.A. (2015). Colliding impulsive gravitational waves and
a cosmological constant, Phys. Rev. D 92, 044032.
Barrabés, C. Gramain A. Lesigne E. and Letelier P.S. (1992). Geometric inequal-
ities and the hoop conjecture, Class. Quantum Grav. 9, L105.
Barrabés, C. Israel W. and Letelier P.S. (1991). Analytic models of nonspheri-
cal collapse, cosmic censorship and the hoop conjecture, Phys. Lett. A 160,
41.
Basov, N.G. (1979). Problems in General Theory of Relativity and Theory
of Group Representations. The Lebedev Physics Institute Series, Vol. 96
(Springa US) (Translated from Russian by A Mason, New York).
Baumgarte, T.W. Gundlach, C. and Hilditch, D. (2019). Critical phenomena in
the gravitational collapse of electromagnetic waves, Phys. Rev. Lett. 123,
171103.
Baumgarte, T.W. and Shapiro, S.L. (2010). Numerical Relativity: Solving
Einstein’s Equations on the Computer (Cambridge University Press,
Cambridge).
Belinsky, V.A. (1979). One-soliton cosmological waves, Sov. Phys. JETP 50,
623.
Belinsky, V.A. (1980). L–A pair of a system of coupled equations of the gravita-
tional and electromagnetic fields, JETP Lett. 30, 29.
Belinsky, V.A. and Francaviglia, M, (1982). Solitonic gravitational waves in
Bianchi II cosmologies. 1. The general framework, Gen. Relativ. Grav. 14,
213.
Belinsky, V.A. and Ruffini, R. (1980). On axially symmetric soliton solutions of
the coupled scalar-vector-tensor equations in general relativity, Phys. Lett. B
89, 195.
Belinsky, V.A. and Verdaguer, E. (2001). Gravitational Solitons (Cambridge
University Press, Cambridge).
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-bibliography page 177

Bibliography 177

Belinsky, V.A. and Zakharov, V.E. (1978). Integration of the Einstein equations
by means of the inverse scattering problem technique and construction of
exact soliton solutions, Sov. Phys. JETP 48, 985.
Belinsky, V.A. and Zakharov, V.E. (1979). Stationary gravitational solitons with
axial symmetry, Sov. Phys. JETP 50, 1.
Bell, P. and Szekeres, P. (1974). Interacting electromagnetic shock waves in gen-
eral relativity, Gen. Relativ. Grav. 5, 275.
Bennett, C.L. Banday, A. Gorski, K.M. Hinshaw, G. Jackson, P. Keegstra,
P. Kogut, A. Smoot, G.F. Wilkinson, D.T. and Wright, E.L. (1996). Four-
year COBE DMR cosmic microwave background observations: Maps and
basic results, Astrophys. J. 464, L1.
Berezin, V.A. Kuzmin, V.A. and Tkachev, I. (1987). Dynamics of bubbles in
general relativity, Phys. Rev. D 36, 2919.
Bieri, L. Garfinkle, D. and Yunes, N. (2017). Gravitational waves and their math-
ematics, AMS Notices 64, 07.
Blanchet, L. and Damour, T. (1992). Hereditary effects in gravitational radiation,
Phys. Rev. D 46, 4304.
Bondi, H. (1957). Plane gravitational waves in general relativity, Nature (London)
179, 1072.
Bondi, H. and Pirani, F.A.E. (1989). Gravitational waves in general relativity
XIII. caustic property of plane waves, Proc. R Soc. Lond. A 421, 395.
Bondi, H. Pirani, F.A.E. and Robinson, I. (1959). Gravitational waves in general
relativity III. Exact plane waves, Proc. R Soc. Lond. A 251, 519.
Bonnor, W.B. (1983). Physical interpretation of vacuum solutions of Einstein’s
equations. Part I. Time-independent solutions, Gen. Relativ. Grav. 15, 535.
Bonnor, W.B. and Vichers, P.A. ( 1981). Junction conditions in general relativity,
Gen. Relativ. Grav. 13, 29.
Boulware, D.G. (1973). Naked singularities, thin shells, and the Reissner-
Norstrom metric, Phys. Rev. D 8, 2363.
Braginsky, V.P. and Grishchuk, L.P. (1985). Kinematic resonance and memory
effect in free-mass gravitational antennas, Zh. Eksp. Teor. Fiz. 89, 744 [Sov.
Phys. JETP 62, 427].
Braginsky, V.P. and Thorne, K.P. (1987). Gravitational-wave bursts with memory
and experimental prospects, Nature 327, 123.
Brandenberger, R.H. (1985). Quantum field theory methods and inflationary uni-
verse models, Rev. Mod. Phys. 57, 1.
Braun, V. and Ovrut, B.A. (2006). Stabilizing moduli with a positive cosmological
constant in heterotic M-theory, J. High Energy Phys., 07, 035.
Bretón, N. Garcı́a, A. Macı́as, A. and Yáñez, G. (1998). Colliding plane waves in
terms of Jacobi functions, J. Math. Phys. 39, 6051.
Bretón, N. Matos, T. and Garcı́a, A. (1996). Colliding plane waves in Einstein–
Maxwell-dilaton fields, Phys. Rev. D 53, 1868.
Brinkman. H.W. (1923). On riemann spaces conformal to einstein spaces, Proc.
Natl. Acad. Sci. USA 9, 1.
Bronnikov, K.A. (1980). Gravitational and sound waves in stiff matter, J. Phys.
A: Math. Gen. 13, 3455.
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-bibliography page 178

178 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

Bronnikov, K.A. Santos, N.O. and Wang, A. (2019). Cylindrical systems in general
relativity, preprint, arXiv:1901.06561 [Class. Quantum Grav. in press (2020)].
Bruckman, W. (1986). Stationary axially symmetric exterior solutions in the five-
dimensional representation of the Brans–Dicke–Jordan theory of gravitation,
Phys. Rev. D 34, 2990.
Bruckman, W. (1987). Exact axially symmetric stationary solutions of the
Kaluza–Klein–Jordan–Thiry theory, Phys. Rev. D 36, 3674.
Campbell, S.J. and Wainwright, J. (1977). Algebraic computing and the Newman-
Penrose formalism in general relativity, Gen. Relativ. Grav. 8, 987.
Carot, J. and Verdaguer, E. (1989). Generalised soliton solutions of the Weyl
class, Class. Quantum Grav. 6, 845.
Carr, B.J. and Verdaguer, E. (1984). Soliton solutions and cosmological gravita-
tional waves, Phys. Rev. D 28, 2995.
Carroll, S. (2004). Spacetime and Geometry: An Introduction to General Relativity
(Addison-Wesley, New York).
Centrella, J. and Matzner, R.A. (1982). Colliding gravitational waves in expand-
ing cosmologies, Phys. Rev. D 25, 930.
Cespedes, J. and Verdaguer, E. (1987). Gravitational wave pulse and soliton wave
collision, Phys. Rev. D 36, 2259.
Challifour, J.L. (1972). Generalized Functions and Fourier Analysis (W. A.
Benjamin, Reading, MA).
Chan, R. da Silva, M.F.A. da Rocha, J.F. and Wang, A. (2011). Radiating gravas-
tars, J. Cosmol. Astropart. Phys. 10, 013.
Chandrasekhar, S. (1983). The Mathematical Theory of Black Holes (Clarendon
Press. Oxford).
Chandrasekhar, S. (1986). Cylindrical waves in general relativity, Proc. R. Soc.
Lond. A 408, 209.
Chandrasekhar, S. and Ferrari, V. (1984). On the Nutku–Halil solution for col-
liding impulsive gravitational waves, Proc. R. Soc. Lond. A 396, 55.
Chandrasekhar, S. and Xanthopoulos, B.C. (1985a). On colliding waves in the
Einstein-Maxwell theory, Proc. R. Soc. Lond. A 398, 223.
Chandrasekhar, S. and Xanthopoulos, B.C. (1985b). On the collision of impulsive
gravitational waves when coupled with fluid motions, Proc. R. Soc. Lond. A
402, 37.
Chandrasekhar, S. and Xanthopoulos, B.C. (1985c). Some exact solutions of grav-
itational waves when coupled with fluid motions, Proc. R. Soc. Lond. A 402,
205.
Chandrasekhar, S. and Xanthopoulos, B.C. (1986a). A new type of singularity
created by colliding gravitational waves, Proc. R. Soc. Lond. A 408, 175.
Chandrasekhar, S. and Xanthopoulos, B.C. (1986b). On the collision of impulsive
gravitational waves when coupled with null dust, Proc. R. Soc. Lond. A 403,
189.
Chandrasekhar, S. and Xanthopoulos, B.C. (1987a). The effect of sources on
horizons that may develop when plane gravitational waves collide, Proc. R.
Soc. Lond. A 414, 1.
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-bibliography page 179

Bibliography 179

Chandrasekhar, S. and Xanthopoulos, B.C. (1987b). On colliding waves that

develop time-like singularities: a new class of solutions of the Einstein -
Maxwell equations, Proc. R. Soc. Lond. A 410, 311.
Chandrasekhar, S. and Xanthopoulos, B.C. (1988). A perturbation analysis of
the Bell–Szekeres space-time, Proc. R. Soc. Lond. A 420, 93.
Chen, B. (2004). Colliding plane wave solutions in string theory revisited, J. High
Energy Phys. 12, 016.
Chen, B. Chu, C.S. Furuta, K. and Lin, F.L. (2004). Colliding plane waves string
theory, J. High Energy Phys. 02, 020.
Chen, B. Zhang, J.F. (2005). Two-flux colliding plane waves in string theory,
Commun. Theor. Phys. 44, 463.
Chernoff, D.F. Flanagan E.E. and Wardell, B. (2018). Gravitational back-reaction
on a cosmic string: Formalism, preprint, arXiv:1808.08631.
Chernoff, D.F. and Tye, S.H.H. (2018). Detection of low tension cosmic super-
strings, J. Cosmol. Astropart. Phys. 05, 002.
Chesler, P.M. (2015). Colliding shock waves and hydrodynamics in small systems,
Phys. Rev. Lett. 115, 241602.
Chesler, P.M. and Yaffe, L.G. (2011). Holography and colliding gravitational
shock waves in asymptotically AdS5 spacetime, Phys. Rev. Lett. 106, 021601.
Choptuik, M.W. (1993). Universality and scaling in gravitational collapse of a
massless scalar field, Phys. Rev. Lett. 70, 9.
Christodoulou, D. (1991). Nonlinear nature of gravitation and gravitational wave
experiments, Phys. Rev. Lett. 67, 1486.
Clarke, C.J.S. and Dray, T. (1987). Junction conditions for null hypersurfaces,
Class. Quantum Grav. 4, 265.
Clarke, C.J.S. and Hayward, S.A. (1989). The global structure of the
Bell–Szekeres solution, Class. Quantum Grav. 6, 615.
Connors, P.A. Piran, T. and Stark, R.F. (1980). Polarization features of X-ray
radiation emitted near block hole, Astrophys. J. 235, 224.
Connors, P.A. and Stark, R.F. (1977). Observable gravitational effects on
polarised radiation coming from near a block hole, Nature London 269, 128.
Copeland, E.J. Myers, R.C. and Polchinski, J. (2004). Cosmic F and D strings,
J. High Energy Phys. 06, 013.
Cosgrove, C.M. (1980). Relationships between the group-theoretic and soliton-
theoretic techniques for generating stationary axisymmetric gravitational
solutions, J. Math. Phys. 21, 2417.
Cosgrove, C.M. (1981). Backlund transformations in the Hauser–Ernst formalism
for stationary axisymmetric space-times, J. Math. Phys. 22, 2624.
Cosgrove, C.M. (1982). Relationship between the inverse scattering techniques of
Belinskii–Zakharov and Hauser–Ernst in general relativity, J. Math. Phys.
23, 615.
Darmois, G. (1927). Les équations de la gravitation einstetntenne, Mémori al des
Sciences, Mathématiques XXV (Gauthier-Villars. Paris).
Dautcourt, G. (1964). Wechselwirkungsfreie Gravitationsstoßwellen erster Ord-
nung, Math. Nachr. 27, 277.
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-bibliography page 180

180 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

Davies, T.M. ( 1976a). A simple application of the Newman-Penrose spin coeffi-

cients formalism (I), Int. J. Theor. Phys. 15, 315.
Davies, T.M. ( 1976b). A simple application of the Newman -Penrose spin coef-
ficient formalism (II), Int. J. Theor. Phys. 15, 319.
Da Silva, M.F.A. Herrera, L. Santos, N.O. and Wang, A. (2002). Rotating cylin-
drical shell source for Lewis spacetime, Class. Quantum Grav. 19, 3809.
D’Eath, P.D. and Payne, P.N. (1992a). Gravitational radiation in black-hole col-
lisions at the speed of light. I. Perturbation treatment of the axisymmetric
collision, Phys. Rev. D 46, 658.
D’Eath, P.D. and Payne, P.N. (1992b). Gravitational radiation in black-hole col-
lisions at the speed of light. II. Reduction to two independent variables and
calculation of the second-order news function, Phys. Rev. D 46, 675.
D’Eath, P.D. and Payne, P.N. (1992c). Gravitational radiation in black-hole col-
lisions at the speed of light. III. Results and conclusions, Phys. Rev. D 46,
694.
De Leon, J.P. (2004). Brane-world models emerging from collisions of plane waves
in 5D, Gen. Relativ. Grav. 36, 923.
Deser, S. and Ford, K.W. (1964). Lectures on General Relativity Brandeis Summer
Institute in Theoretical Physics, Vol. 1. (Prentice-Hall, Englewood Cliffs).
Devin, M. Ali, T. Cleaver, G. Wang, A. and Wu, Q. (2009). Branes in the
MD × Md+ × Md− Compactification of type II string on S 1 /Z2 and their
cosmological applications, J. High Energy Phys. 10, 095.
DeWitt, C. and DeWitt, B.S. (1964). Relativity, Groups, and Topology (Gordon
and Breach, New York).
DeWitt, C. Schatzman, E. and Veron, P. (1967). High Energy Astrophysics,
Vol. III (Gordon and Breach, New York).
DeWitt, C. and Wheeler, J.A. (1968). Battelle Rencontres: 1967 Lectures in Math-
ematics and Physics (W. A. Benjamin, New York).
Diaz, M. Gleiser, R.J. and Pullin, J.A. (1987). Solitonic perturbations of per-
fect fluid Friedmann-Robertson-Walker cosmological models, Class. Quan-
tum Grav. 4, 123.
Diaz, M. Gleiser, R.J. and Pullin, J.A. (1988). Solitonic solutions to the Kaluza-
Klein-Jordan formalism as cosmological models in general relativity, J. Math.
Phys. 29, 169.
Ding, X. Q. and Ding, Y. (2005). Hermite Expansion and Generalized Functions
(Huazhong Normal University Press, 2005).
D’Inverno, R. (2003). Introducing Einstein’s Relativity (Clarendon Press, Oxford).
Dray, T. and ’t Hooft, G. (1985a). The gravitational shock wave of a massless
particle, Nucl. Phys. B 253, 173.
Dray, T. and ’t Hooft, G. (1985b). The effect of spherical shells of matter on the
Schwarzschild black hole, Commun. Math. Phys. 99, 613.
Dray, T. and ’t Hooft, G. (1986). The gravitational effect of colliding planar shells
of matter, Class. Quantum Grav. 3, 825.
Dueñas-Vidal, A. and Vázquez-Mozo, M.A. (2010). Colliding AdS gravitational
shock waves in various dimensions and holography, J. High Energy Phys. 07,
021.
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-bibliography page 181

Bibliography 181

Dueñas-Vidal, A. and Vázquez-Mozo, M.A. (2012). A Note on the collision of

Reissner–Nordström gravitational shock waves in AdS, Phys. Lett. B 713,
500.
Dvali, G. and Vilenkin, A. (2004). Formation and evolution of cosmic D strings,
J. Cosmol. Astropart. Phys. 03, 010.
Echeverria F. (1993). Gravitational collapse of an infinite, cylindrical dust shell,
Phys. Rev. D 47, 2271.
Economou, A. (1988). Ph.D. dissertation, Physics Department, Ioannina Univer-
sity, Greece.
Economou, A. and Tsoubelis, D. (1988a). Rotating cosmic strings and gravita-
tional soliton waves, Phys. Rev. D 38, 498.
Economou, A. and Tsoubelis, D. (1988b). Interaction of cosmic strings with grav-
itational waves: A new class of exact solutions, Phys. Rev. Lett. 61, 2046.
Economou, A. and Tsoubelis, D. (1989). Multiple-soliton solutions of Einstein’s
equations, J. Math. Phys. 30, 1562.
Ehlers, J. and Kundt, W. (1962). Exact solutions of the gravitational field equa-
tions in Gravitation: An Introduction to Current Research, ed. Witten, L.
(John Wiley & Sons), pp. 49–101.
Einstein, A. (1905). Zur Elektrodynamik bewegter Körper, Ann. Phys. 17, 891.
Einstein, A. (1915a). Zur Allgemeinen Relativitätstheorie, Preuss. Akad. Wiss.
Berlin, Sitzber. 778.
Einstein, A. (1915b). Der Feldgleichungen der Gravitation, Preuss. Akad. Wiss.
Berlin, Sitzber. 844.
Einstein, A. (1916). Die Grundlage der allgemeinen Relativitätstheorie, Ann.
Phys. 49, 769.
Ellis, G.F.R. and Schmidt, B.G. (1977). Singular space-times, Gen. Relativ. Grav.
8, 915.
Ernst, F.J. (1968a). New formulation of the axially symmetric gravitational field
problem, Phys. Rev. 167, 1175.
Ernst, F.J. (1968b). New formulation of the axially symmetric gravitational field
problem. II, Phys. Rev, 168, 1415.
Ernst, F.J. Garcia, A.D. and Hauser, I. (1987a). Colliding gravitational plane
waves with non-collinear polarization. I, J. Math. Phys. 28, 2155.
Ernst, F.J. Garcia, A.D. and Hauser, I. (1987b). Colliding gravitational plane
waves with non-collinear polarization. II, J. Math. Phys. 28, 2951.
Ernst, F.J. Garcia, A.D. and Hauser, I. (1988). Colliding gravitational plane waves
with non-collinear polarization. III, J. Math. Phys. 29, 681.
Faraday, M. (1846a). I. Experimental researches in electricity — Nineteenth series.
Philos. Trans. 136.
Faraday, M. (1846b). Ann. Phys. 68, 105.
Farnsworth, D. Fink, J. Porter, J. and Thompson, A. (1972). Methods of local
and global differential geometry in general relativity, in Proceedings of the
Regional Conference on Relativity, University of Pittsburgh, Pittsburgh,
Pennsylvania, July 13–17, 1970 (Springer-Verlag).
Favata, M. (2010). The gravitational–wave memory effect, Class. Quantum Grav.
27, 084036.
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-bibliography page 182

182 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

Feinstein, A. (2002). Penrose limits, the colliding plane-wave problem and the
classical string backgrounds, Class. Quantum Grav. 19, 5353.
Feinstein, A. and Charach, C. (1986). Gravitational solitons and spatial flatness,
Class. Quantum Grav. 3, L5.
Feinstein, A. and Ibañez, J. (1989). Curvature-singularity-free solutions for col-
liding plane gravitational waves with broken u − v symmetry, Phys. Rev. D
39, 470.
Feinstein, A. MacCallum, M.A.H. and Senovilla, J.M.M. (1989). On the ambigu-
ous evolution and the production of matter in space-times with colliding
waves, Class. Quantum Grav. 6, L217.
Feinstein, A. and Senovilla, J.M.M. (1989). Collision between variably polarized
plane gravitational wave and a shell of null matter, Phys. Lett. A 138, 102.
Ferrari, V. (1988). Focusing process in the collision of gravitational plane waves,
Phys. Rev. D 37, 3061.
Ferrari, V. and Ibañez, J. (1987a). On the collision of gravitational plane waves:
a class of soliton solutions, Gen. Relativ. Grav. 19, 405.
Ferrari, V. and Ibañez, J. (1987b). A new exact solution for colliding gravitational
plane waves, Gen. Relativ. Grav. 19, 383.
Ferrari, V. and Ibañez, J. (1988). Type-D solutions describing the collision of
plane-fronted gravitational waves, Proc. R. Soc. Lond. A 417, 417.
Ferrari, V. and Ibañez, J. (1989a). On the gravitational interaction of null fields,
Class. Quantum Grav. 6, 1805.
Ferrari, V. and Ibañez, J. (1989b). Gravitational interaction of massless particles,
Phys. Lett. A 141, 233.
Ferrari, V. Ibañez, J. and Bruni, M. (1987a). Colliding gravitational waves with
non-collinear polarization: A class of soliton solutions, Phys. Lett. A 122,
459.
Ferrari, V. Ibañez, J. and Bruni, M. (1987b). Colliding plane gravitational waves:
A class of non-diagonal soliton solutions, Phys. Rev. D 36, 1053.
Frolov, V.P. (1979). The Newman–Penrose method in the theory of general
relativity, in Problems in the General Theory of Relativity and Theory of
Group Representations, ed. Basov, N. G. and translated from Russian by
Mason A. (Proceedings (Trudy) of the P. N. Lebedev Physics Institute,
Vol. 96).
Garcá-Dı́az, A. (1988). Colliding-wave generalizations of the Nutku–Halil metric
in the Einstein–Maxwell theory, Phys. Rev. Lett. 61, 507.
Garcá-Dı́az, A. (1989). Colliding wave generalization of the Ferrari–Ibañez solu-
tion in the Einstein–Maxwell theory, Phys. Lett. A 138, 370.
Gardner, C.S. Greene, J.M. Kruskal, M.D. and Miura, R.M. (1967).
Method for solving the Korteweg-deVries equation, Phys. Rev. Lett. 19,
1095.
Gelfand, I.M. and Shilov, G.E. (1964). Generalized Functions. Vol. I: Properties
and Operations (Academic Press, New York, London).
Geroch, R. (1971). A method for generating solutions of Einstein’s equations,
J. Math. Phys. 12, 918.
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-bibliography page 183

Bibliography 183

Geroch, R. (1972). A method for generating solutions of Einstein’s equations II,

J. Math. Phys. 13, 394.
Goldberger, W.D. and Wise, M.B. (1999). Modulus stabilization with bulk fields,
Phys. Rev. Lett. 83, 4922.
Gong, Y.-G. Wang, A. and Wu, Q. (2008). Cosmological constant and late tran-
sient acceleration of the universe in the Horava–Witten heterotic M-theory
on S 1 /Z2 , Phys. Lett. B 663, 147.
Gray, J. Lukas, A. and Ovrut, B. (2007). Flux, gaugino condensation, and
antibranes in heterotic M-theory, Phys. Rev. D 76, 126012.
Greenwald, J. Lenells, J. Satheeshkumar, V.H. and Wang, A. (2013). Gravita-
tional collapse in Horřava–Lifshitz theory, Phys. Rev. D 88, 024044.
Griffiths, J.B. (1976a). Colliding neutrino fields in general relativity, J. Phys. A:
Math. Gen. 9, 45.
Griffiths, J.B. (1976b). The collision of plane waves in general relativity, Ann.
Phys. 102, 388.
Griffiths, J.B. (1980). A problem with classical neutrino fields in general relativity,
Nukleonika, 25, 1415.
Griffiths, J.B. (1981). Neutrino fields in Einstein–Cartan theory, Gen. Relativ.
Grav. 13, 227.
Griffiths, J.B. (1985). On the Bell–Szekeres solution for colliding electromagnetic
waves, in Galaxies, Axisymmetric Systems and Relativity. Essays Presented
to W.B. Bonnor on his 65th Birthday, ed. MacCallum, M.A.H. (Cambridge
University Press, Cambridge), p. 199.
Griffiths, J.B. (1987). Colliding plane gravitational waves, Class. Quantum Grav.
4, 957.
Griffiths, J.B. (1990). Colliding plane gravito-electromagnetic waves, Int. J.
Theor. Phys. 29, 173.
Griffiths, J.B. (1991). Colliding Plane Waves in General Relativity (Clarendon
Press, Oxford). Republished by Dover Publications, Inc. (New York, 2016).
Griffiths, J.B. (2005). The stability of Killing–Cauchy horizons in colliding plane
wave space-times, Gen. Relativ. Grav. 37, 1119.
Griffiths, J.B. and Halburd, R.G. (2007). The C-metric as a colliding plane wave
spacetime, Class. Quantum. Grav. 24, 1049.
Griffiths, J.B. and Podolsky, J. (2009). Exact Space-Times in Einstein’s General
Relativity (Cambridge University Press, Cambridge).
Griffiths, J.B. and Santano-Roco, M. (2002). The characteristic initial value prob-
lem for colliding plane waves: The linear case, Class. Quantum Grav., 19,
4273.
Grishchuk, L.P. and Polnarev, A.G. (1989). Gravitational wave pulses with
velocity memory, Sov. Phys. JETP, 69 (1989) [Zh. Eksp. Teor. Fiz.
96, 1153].
Grumiller, D. and Romatschke, P. (2008). On the collision of two shock waves in
AdS5 , J. High Energy Phys. 08, 027.
Gundlach, C. and Martin-Garcia, J.M. (2007). Critical phenomena in gravita-
tional collapse, Living Rev. Relativ. 10, 5.
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-bibliography page 184

184 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

Gürses, M. and Halilsoy, M. (1982). Interacting superposed electromagnetic shock

plane waves in general relativity, Lett. Nuovo Cimento, 34, 588.
Gürses, M. Kahya, E.O. and Karasu, A. (2002). Higher dimensional metrics of
colliding gravitational plane waves, Phys. Rev. D 66, 024029.
Gürses, M. and Sermutlu, E. (1995). Colliding gravitational plane waves in dilaton
gravity, Phys. Rev. D 52, 809.
Gurtug, O. and Halilsoy, M. (2000). Horizon instability in the cross polarized
Bell–Szekeres spacetime, preprint, arXiv:gr-qc/000.6038.
Gurtug, O. and Halilsoy, M. (2009). Effect of NUT parameter on the analytic
extension of the Cauchy horizon that develop in colliding wave spacetimes,
Int. J. Mod. Phys. A 24, 3171.
Gurtug, O. Halilsoy, M. and Sakalli, I. (2003). New singular and nonsingu-
lar colliding–wave solutions in Einstein-Maxwell-scalar theory, Gen. Relativ.
Grav. 35, 2159.
Gutperle, M. and Pioline, B. (2003). Type-IIB Colliding plane waves, J. High
Energy Phys. 09, 061.
Halli, M. (1979). Colliding plane gravitational waves, J. Math. Phys. 20, 120.
Halilsoy, M. (1988a). Distinct family of colliding gravitational waves in general
relativity, Phys. Rev. D 38, 2979.
Halilsoy, M. (1988b). Colliding electromagnetic shock waves in general relativity,
Phys. Rev. D 37, 2121.
Halilsoy, M. (1989a). Colliding superposed waves in the Einstein–Maxwell theory,
Phys. Rev. D 39, 2172.
Halilsoy, M. (1989b). Colliding superposed waves in the Einstein–Maxwell theory,
Phys. Rev. D 39, 2172.
Halilsoy, M. Mazharimousavi, S.H. and Gurtug, O. (2014). Emergent cosmological
constant from colliding electromagnetic waves, J. Cosmol. Astropart. Phys.
11, 010.
Halilsory, M. and Sakalli, I. (2003). Collision of electromagnetic shock waves cou-
pled with axion waves: An example, Class. Quantum Grav. 20, 1417.
Harrison, B.K. (1978). Bäcklund transformation for the Ernst equation of general
relativity, Phys. Rev. Lett. 41, 1197.
Harrison, B.K. (1980). New large family of vacuum solutions of the equations of
general relativity, Phys. Rev. D 21, 1695.
Harte, A.I. (2013). Strong lensing, plane gravitational waves and transient flashes,
Class. Quantum Grav. 30, 075011.
Hassan, Z. Feinstein, A. and Manko, V. (1990). Asymmetric collision between
plane gravitational waves with variably polarization, Class. Quantum Grav.
7, L109.
Hauser, I. and Ernst, F.J. (1979a). Integral equation method for effecting
Kinnersley–Chitre transformations, Phys. Rev. D 20, 362.
Hauser, I. and Ernst, F.J. (1979b). Integral equation method for effecting
Kinnersley-Chitre transformations. II, Phys. Rev. D 20, 1783.
Hauser, I. and Ernst, F.J. (1980a). A homogeneous Hilbert problem for the
Kinnersely–Chitre transformations, J. Math. Phys. 21, 1126.
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-bibliography page 185

Bibliography 185

Hauser, I. and Ernst, F.J. (1980b). A homogeneous Hilbert problem for the
Kinnersely–Chitre transformations of electrovac space-times, J. Math. Phys.
21, 1418.
Hauser, I. and Ernst, F.J. (1981). Proof of a Geroch conjecture, J. Math. Phys.
22, 1051.
Hauser, I. and Ernst, F.J. (1989a). Initial value problem for colliding plane grav-
itational waves. I, J. Math. Phys. 30, 872.
Hauser, I. and Ernst, F.J. (1989b). Initial value problem for colliding plane grav-
itational waves. II, J. Math. Phys. 30, 2322.
Hauser, I. and Ernst, F.J. (1990). Initial value problem for colliding plane gravi-
tational waves. III, J. Math. Phys. 31, 871.
Hawking, S.W. (1975). Particles creation by black holes, Commun. Math. Phys.
43, 199.
Hawking, S.W. and Eillis, G.F.R. (1973). The Large Scale Structure of Space-
Time, Vol. 1 (Cambridge University Press, Cambridge).
Hawking, S. W. Perry, M.J. and Strominger, A. (2016). Soft hair on black holes,
Phys. Rev. Lett. 116, 231301.
Hawking, S. W. Perry, M. J. and Strominger, A. (2017). Superrotation
charge and supertranslation hair on Black Holes, J. High Energy Phys. 05,
161.
Helliwell, T.M. and Konkowski, D.A. (1990). Electromagnetic fields in Khan–
Penrose space-time, Phys. Rev. D 41, 2507.
Herrera, L. and Santos, N.O. (2005). Cylindrical collapse and gravitational waves,
Class. Quantum Grav. 22, 2407.
Hill, C.T. Schramm, D.N. and Fry, J.N. (1989). Cosmological structure formation
from soft topological defects, Comm. Nucl. Part. Phys. 19, 25.
Hirschmann, E.W. Wang, A. and Wu, Y. (2004). Collapse of a scalar field in 2 + 1
gravity, Class. Quantum Grav. 21, 1791.
Hoenselaers, C. and Dietz, W. (1984). Solutions of Einstein’s equations: tech-
niques and results, in Proceedings of the International Seminar on Exact
Solutions of Einstein’s Equations, in Retzbach, Germany, November 14–18,
1983 (Springer-Verlag).
Hoenselaers, C. and Ernst, F.J. (1990). Matching pp-waves to the Kerr metric,
J. Math. Phys. 31, 144.
Hoenselaers, C. Kinnersley, W. and Xanthopoulos, B.C. (1979a). Generation of
asymptotically flat space-times with any number of parameters, Phys. Rev.
Lett. 42, 481.
Hoenselaers, C. Kinnersley, W. and Xanthopoulos, B.C. (1979b). Symmetries
of the stationary Einstein-Maxwell equations VI-transformations which gen-
erate asymptotically flat space-times with arbitrary multiple moments,
J. Math. Phys. 20, 2530.
Hogan, P.A. Barrabés, C. and Bressange, G.F. (1998). Colliding plane waves in
Einstein–Maxwell theory, Lett. Math. Phys. 43, 263.
Hogan, P.A. and Walsh, D.M. (2003). Collision of high frequency plane gravita-
tional and electromagnetic waves, Int. J. Mod. Phys. 12, 1459.
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-bibliography page 186

186 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

Holvorcem, P.R. Letelier, P.S. and Wang, A. (1995). The interaction of outgoing
and ingoing spherically symmetric null fluids, J. Math. Phys. 36, 3663.
Hu, N. (1983). Proceedings of the Third Marcel Grossmann Meeting on General
Relativity, Shanghai, China (Science Press, Beijing).
Horava, H. and Witten, E. (1995). Heterotic and type I string dynamics from
eleven dimensions, Nucl. Phys. B 460, 506.
Horava, H. and Witten, E. (1996). Eleven-dimensional supergravity on a manifold
with boundary, Nucl. Phys. B 475, 94.
Ibañez, J. and Verdaguer, E. (1983). Soliton collision in general relativity, Phys.
Rev. Lett. 51, 1313.
Ibañez, J. and Verdaguer, E. (1986). Finite perturbations on Friedmann–
Robertson–Walker models, Astrophys. J. 306, 401.
Ipser, J. and Sikivie, P. (1984). Gravitational repulsive domain wall, Phys. Rev.
D 30, 712.
Ishihara, H. Takahashi, M. and Tomimatsu, A. (1988). Gravitational Faraday
rotation induced by a Kerr block hole, Phys. Rev. D 38, 472.
Islam, J.N. (1985). Rotating Fields in General Relativity (Cambridge University
Press, Cambridge).
Israel, W. (1966). Singular hypersurfaces and thin shells in general relativity,
Il Nuovo Cimento B 44, 1.
Israel, W. (1967). Singular hypersurfaces and thin shells in general relativity,
Il Nuovo Cimento B 48, 463(E).
Ivanov, B.V. (1998). Colliding axisymmetric pp waves, Phys. Rev. D 57, 3378.
Jantzen, R.T. (1980). Soliton solutions of the Einstein equations generated from
cosmological solutions with additional symmetry, Nuovo Cimento B 59, 287.
Jogia, S. and Griffiths, J.B. (1980). A Newman–Penrose formalism for space-tunes
with torsion, Gen. Relativ. Grav. 12, 597.
Jordan, P. Ehlers, J. and Kundt, W. (1960). Akad. Wiss. Matnz. Abn. Math.
-Nat. Kl., Jabrg 2.
Kachru, S. Schulz, M.B. and Trivedi, S. (2003). Moduli stabilization from fluxes
in a simple IIB orientifold, J. High Energy Phys. 10, 007.
Kerr, R.P. (1963). Gravitational field of a spinning mass as an example of alge-
braically special metric, Phys. Lett. 11, 237.
Khan, K.A. and Penrose, R. (1971). Scattering of two impulsive gravitational
plane waves, Nature (London) 229, 185.
Khorrami, M. and Mansouri, R. (1994). Cylindrically symmetric thin walls in
general relativity, J. Math. Phys. 35, 951.
Kibble, T.W.B. (1976). Topology of cosmic domains and strings, J. Phys. A:
Math. Gen. 9, 1387.
Kinnersley, W. (1969). Type D vacuum metrics, J. Math. Phys. 10,
1195.
Kinnersley, W. and Walker, M. (1970). Uniformly accelerating charged mass in
general relativity, Phys. Rev. D 2, 1359.
Kitchingham, D.W. (1984). The use of generating techniques for space-times with
two non-null commuting Killing vectors in vacuum and stiff perfect fluid
cosmological models, Class. Quantum Grav. 1, 677.
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-bibliography page 187

Bibliography 187

Kitchingham, D.W. (1986). The application of the homogeneous Hilbert problem

of Hauser and Ernst to cosmological models with spatial axes of symmetry,
Class. Quantum Grav. 3, 133.
Konkowski, D.A. and Helliwell, F.M. (1989). Singularities in colliding gravita-
tional plane wave space-times, Class. Quantum Grav. 6, 1847.
Konkowski, D.A. and Helliwell, F.M. (1991). Stability of the quasiregular singu-
larities in Bell–Szekeres spacetime, Phys. Rev. D 43, 609.
Kramer, D. and Neugebauer, G. (1980). The superposition of two Kerr solutions,
Phys. Lett. A 75, 259.
Kramer, D. and Neugebauer, G. (1984). Backlund transformations in general rela-
tivity, in Solutions of Einstein’s Equations: Techniques and Results, Proceed-
ings of the International Seminar on Exact Solutions of Einstein’s Equations,
Retzbach, Germany, November 14–18, 1983, Vol. 2, eds. Hoenselaers, C. and
Dietz, W. (Springer-Verlag).
Kramer, D. Stephani, H. Herlt, H. MacCallum, M. and Schmutzer, E. (1980).
Exact Solutions of Einstein’s Field Equations (Cambridge University Press,
Cambridge).
Kundt, W. and Trumper, M. (1962). Beitrage zur theorte der gravitationas
strahlungsfelder, Alcad. Wiss. Lit. Mainz. Abhandl. Math.-Nat. Kl. 12.
Laguna-Castillo, P. and Matzner, R.A. (1986). Inflation and bubbles in general
relativity, Phys. Rev. D 34, 2913.
Lanczos, K. (1922). Bemerkung zur de Sttterschen Welt, Phys. Z. 23, 539.
Lanczos, K. (1924). Flachenhafte verteliung der materie in der Einsteinschen grav-
itationstheorie, Ann. Phys. 74, 518.
Landau, L.D. and Lifshitz, E.M. (1972). The Classical Theory of Fields, Fourth
Revised English Edition, translated by M. Hamermesh (Pergamon Press).
Lasky, P.D. Thrane, E. Levin, Y. Blackman, J. and Chen, Y. (2016). Detecting
gravitational wave memory with LIGO: implications of GW150914, Phys.
Rev. Lett. 117, 061102.
Letelier, P.S. (1986). Soliton solutions to the vacuum Einstein equations obtained
from a non-diagonal seed solution, J. Math. Phys. 27, 564.
Letelier, P.S. (1989). On soliton solutions to the vacuum Einstein equations
obtained from a general seed solution, Class. Quantum Grav. 6, 875.
Letelier, P.S. and Wang, A. (1993a). Collisions of cosmic walls, Class. Quantum
Grav. 10, L29.
Letelier, P.S. and Wang, A. (1993b). Spherically-symmetric thin shells in the
Brans–Dicke theory of gravity, Phys. Rev. D 48, 631.
Letelier, P.S. and Wang, A. (1993c). On the interaction of null fluids in cosmology,
Phys. Lett. A 182, 220.
Letelier, P.S. and Wang, A. (1994). Singularities formed by the focusing of cylin-
drical null fluids, Phys. Rev. D 49, 5105.
Letelier, P.S. and Wang, A. (1995). Spacetime defects, J. Math. Phys. 36, 3023.
Li, W. (1989). New families of colliding plane gravitational waves with collinear
polarization, Class. Quantum Grav. 6, 477.
Li, W. and Ernst, F.J. (1989). A family of electrovac colliding wave solutions of
Einstein’s equations, J. Math. Phys. 30, 678.
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-bibliography page 188

188 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

Lichnerowicz, A. (1955). Theories relatlvistes de la gravitation et de

l’electromagnettsme (Masson, Paris).
Lind, R.W. (1974). Shear-free, twisting Einstein–Maxwell metrics in the Newman–
Penrose formalism, Gen. Relativ. Grav. 5, 25.
Linde, A.D. (1984). The inflation universe, Rep. Prog. Phys. 47, 925.
Lukas, A. Ovrut, B.A. Stelle, K.S. and Waldram, D. (1999). The universe as a
domain wall, Phys. Rev. D 59, 086001.
Lyth, D.H. and Liddle, A.D. (2009). The Primordial Density Perturbation: Cos-
mology, Inflation, and the Origin of Structure (Cambridge University Press,
Cambridge).
Maartens, R. (2004). Brane-World Gravity, Living Rev. Relativity 7, 7.
Maartens, R. and Koyama, K. (2010). Brane-World Gravity, Living Rev. Relativity
13, 5.
MacCallum, M.A.H. (1984). Exact solutions in cosmology, in Solutions of Ein-
stein’s Equations: Techniques and Results, Proceedings of the International
Seminar on Exact Solutions of Einstein’s Equations, Retzbach, Germany,
November 14–18, 1983, Vol. 2, eds. Hoenselaers, C. and Dietz, W. (Springer-
Verlag).
Maggiore, M. (2008). Gravitational Waves: Volume 1: Theory and Experiments
(Oxford University Press).
Matzner, R.A. and Tippler, F.J. (1984). Metaphysics of colliding self-gravitational
plane waves, Phys. Rev. D 29, 1575.
Misner, C.W. Thorne, K.S. and Wheeler, J.A. (1973). Gravitation (W.H. Freeman
and Company).
Mizoguchi, S. (1996). Colliding wave solutions, duality, and diagonal embedding
of general relativity in two-dimensional heterotic string theory, Nucl. Phys.
B 461, 155.
Nakao, K.-I. Harada, T. Kurita, Y. and Morisawa, Y. (2009). Relativistic grav-
itational collapse of a cylindrical shell of dust. II, Prog. Theor. Phys. 22,
521.
Nakao, K.-I. Kurita, Y. Morisawa, Y. and Harada, T. (2007). Relativistic
gravitational collapse of a cylindrical shell of dust, Prog. Theor. Phys. 117,
75.
Neugebauer, G. (1979). Backlund transformations of axially symmetric stationary
gravitational fields, J. Phys. A: Math. Gen. 12, L67.
Neugebauer, G. (1980a). A general integral of the axially symmetric stationary
Einstein equations, J. Phys. A: Math. Gen. 13, L19.
Neugebauer, G. (1980b). Recursive calculation of axially symmetric stationary
Einstein fields, J. Phys. A: Math. Gen. 13, 1737.
Newman, E. and Penrose, R. (1962). An approach to gravitational radiation by
a method of spin coefficients, J. Math. Phys. 3, 566.
Newman, E. and Penrose, R. (1963). An approach to gravitational radiation by
a method of spin coefficients, J. Math. Phys. 4, 998(E).
Nogales, J.A.C. and Wang, A. (1998). Instability of cosmological event horizons
of global non-static cosmic strings II: Perturbations of gravitational waves
and massless scalar field, Phys. Rev. D 57, 6089.
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-bibliography page 189

Bibliography 189

Nolan, B.C. (2000). Central singularity in spherical collapse, Phys. Rev. D 62,
044015.
Nutku, Y. (1981). Comment on “collision of plane gravitational waves without
singularities”, Phys. Rev. D 24, 1040.
Nutku, Y. and Halli, M. (1977). Colliding impulsive gravitational waves, Phys.
Rev. Lett. 39, 1379.
O’Brien, S. and Synge, J.E. (1952). Jump conditions at discontinuities in general
relativity, Comm. Dublin Inst. Adv. Stud. A9.
Oliver, G. and Verdaguer, E. (1989). A family of inhomogeneous cosmological
Einstein–Rosen metrics, J. Math. Phys. 30, 442.
O’Raifeartaigh, L. (1972). General Relativity, Papers in Honour of J.L. Synge
(Clarendon Press, Oxford).
Ori, A. (2000). Strength of curvature singularities, Phys. Rev. D 61, 064016.
Paiva, F.M. and Wang, A. (1995). Geometry of planar domain walls, Phys. Rev.
D 52, 1281.
Palenta, S. and Meinel, M. (2017). A continuous Riemann-Hilbert problem for
colliding plane gravitational waves, Class. Quantum Grav. 34, 195011.
Pantaleo, X. (1979). Relativity, Quanta and Cosmology, Einstein 1879–1979 (New
York, Johnson).
Papacostas, T. and Xanthopoulos, B.C. (1989). Collisions of gravitational and
electromagnetic waves that do not develop curvature singularities, J. Math.
Phys. 30, 97.
Papapetrou, A. and Hamoui, A. (1968). Couches simples de mattere en relativite
generale, Ann. Inst. Henri Poincare IX, 179.
Penrose, R. (1960). A spinor approach to general relativity, Ann. Phys. 10, 171.
Penrose, R. (l963). Asymptotic properties of fields and space-times, Phys. Rev.
Lett. 10, 66.
Penorse, R. (1964). Conformal treatment of infinity, in Relativity, Groups and
Topology, The 1963 Les Houches Lectures, eds. DeWitt, C. and DeWitt, B.
(Gordon and Breach, New York), pp. 565–584.
Penrose, R. (1965). A remarkable property of plane waves in general relativity,
Rev. Mod. Phys. 37, 215.
Penrose, R. (1968). Structure of space-time, in Battelle Rencontres, Lectures in
Mathematical Physics, eds. DeWitt, C. and Wheeler, J.A. (Benjamin, New
York).
Penrose, R. (1969). Gravitational collapse: the role of general relativity, Riv.
Nuovo Cimento, 1, 252.
Penrose, R. (1972). The geometry of impulsive gravitational waves, in General
Relativity, Papers in Honor of J.L Synge, ed. O’Raifeartaigh, L. (Oxford
University Press, Oxford), pp. 101–115.
Pereira, P.R.C.T. and Wang, A. (2000a). Dynamic cylindrical shells with rotation
in general relativity, Gen. Relativ. Grav. 32, 2189.
Pereira, P.R.C.T. and Wang, A. (2000b). Gravitational collapse of counter-
rotating cylindrical dust shells, Phys. Rev. D 62, 124001.
Pereira, P.R.C.T. and Wang, A. (2002). Gravitational collapse of counter-rotating
cylindrical dust shells, Int. J. Mod. Phys. D 11, 561.
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-bibliography page 190

190 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

Petrov, A.Z. (1955). New Methods in General Relativity (Nauka, Moscow) (in
Russian) [English edition of Petrov’s book: Einstein Spaces, Pergamon Press,
1969].
Pirani, F. (1964). Introduction to gravitational radiation theory, in Lectures in
General Relativity, Brandeis Summer Institute in Theoretical Physics, Vol. 1.
eds. Deser, S. and Ford, K. (Prentice-Hall, Englewood Cliffs).
Piran, T. and Safier, P.N. (1985). A gravitational analogue of Faraday rotation,
Nature (London) 318, 271.
Piran, T. Safier, P.N. and Katz, J. (1986). Cylindrical gravitational waves with
two degrees of freedom: an exact solution, Phys. Rev. D 34, 331.
Piran, T. Safier, P.N. and Stark, R.F. (1985). General numerical solution of cylin-
drical gravitational waves, Phys. Rev. D 32, 3101.
Pretorius, F. and East, W.E. (2018). Black hole formation from the collision of
plane-fronted gravitational waves, Phys. Rev. D 98, 084053.
Randall, L. and Sundrum, R. (1999a). Large mass hierarchy from a small extra
dimension, Phys. Rev. Lett. 83, 3370.
Randall, L. and Sundrum, R. (1999b). An alternative to compactification, Phys.
Rev. Lett. 83, 4690.
Ringeval, C. and Suyama, T. (2017). Stochastic gravitational waves from cosmic
string loops in scaling, J. Cosmol. Astropart. Phys. 12, 027.
Rocha, P. Chan, R. da Silva, M.F.A. and Wang, A. (2008a). Stable and “bounded
excursion” gravastars, and black holes in Einstein’s theory of gravity, J. Cos-
mol. Astropart. Phys. 11, 010.
Rocha, P. Miguelote, A.Y. Chan, R. da Silva, M.F.A. Santos, N.O. and Wang,
A. (2008b). Bounded excursion stable gravastars and black holes, J. Cosmol.
Astropart. Phys. 06, 025.
Rosen, N. (1937). Plane polarised waves in the general theory of relativity, Phys.
Z 12, 366.
Rosenbaum, M. Ryan, M. Urrutia, L.F. and Matzner, R.A. (1986). Colliding plane
waves in N = 1 classical supergravity, Phys. Rev. D 34, 409.
Roy, S. (2003). Accelerating cosmologies from M /string theory compactifications,
Phys. Lett. B 567, 322.
Ruffini, R. (1976). Proceedings of the First Marcel Grossmann Meeting on General
Relativity, 7–12 July 1975 (North-Holland).
Sachs, R.K. (1960). Propagation laws for null and type III gravitational waves,
Phys. Z 157, 462.
Sachs, R.K. (1961). Gravitational waves in general relativity, IV: the outgoing
radiation condition, Proc. R. Soc. Lond. A 264, 309.
Sachs, R.K. (1962). Gravitational waves in general relativity, VIII: waves in
asymptotically flat space-time, Proc. R. Soc. Lond. A 270, 103.
Sachs, R.K. (1964). Gravitational radiation, in Relativity, Groups and Topology,
eds. DeWitt, C. and DeWitt, B.S. (Gordon and Breach, New York).
Sanz, J.L. and Goicoechea, L.J. (1984). Observational and Theoretical Aspects
of Relativistic Astrophysics and Cosmology (World Scientific Publishing Co.).
Sharma, P. Tziolas, A. Wang, A. and Wu, Z.-C. (2011). Spacetime Singularities
in string and its low dimensional effective theory, Int. J. Mod. Phys. A 25,
273.
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-bibliography page 191

Bibliography 191

Schmidt, H.-J. and Wang, A. (1993). Plan domain walls when coupled with the
Brans-Dicke scalar field, Phys. Rev. D 47, 4425.
Schwarz, P. (1997). Colliding axion–dilaton plane waves from black holes, Phys.
Rev. D 56, 7833.
Siegel, H.P. (1981). Evolving dust shells, Phys. Rev. D 23, 2835.
Singh, P. and Griffiths, J.B. (1990). The application of spin coefficient techniques
in the vacuum quadratic Poincare gauge field theory, Gen. Relativ. Grav. 22,
269.
Souriau, J.-M. (1973). Ondes et radiations gravitationnelles, Colloq. Int. CNRS,
Paris 220, 243.
Spergel, D.N. et al. (WMAP) (2007). Wilkinson microwave anisotropy probe
(WMAP) three year results: implications for cosmology, Astrophys. J. Suppl.
170, 377.
Stark, R.F. and Connors, P.A. (1977), Observational test for the existence of a
rotating black hole in CygX-1, Nature (London) 266, 429.
Stephani, H. Kramer, D. MacCallum, M. Hoenselaers, C. and Herlt, E. (2009).
Exact Solutions of Einstein’s Field Equations, Cambridge Monographs on
Mathematical Physics (Cambridge University Press, Cambridge).
Stoyanov, Yu.G. (1979). Interaction of plane gravitational waves, Sov. Phys.
JETP. 44, 1051.
Strominger, A. (2017). Lectures on the Infrared structure of gravity and gauge
theory, preprint, arXiv:1703.05448.
Synge, J.L. (1965). Relativity: The General Theory (North-Holland).
Szekeres, P. (1965). The gravitational compass, J. Math. Phys. 6, 1387.
Szekeres, P. (1966). On the propagation of gravitational fields in matter, J. Math.
Phys. 7, 751.
Szekeres, P. (1970). Colliding gravitational waves, Nature (London) 228, 1183.
Szekeres, P. (1972). Colliding plane gravitational waves, J. Math. Phys. 13, 286.
Tabensky, R. and Taub, A.H. (1973). Plane symmetric self-gravitating fluids with
pressure equal to energy density, Commun. Math. Phys. 29, 61.
Talbot, C.J. (1969). Newman–Penrose approach to twisting degenerate metrics,
Commun. Math. Phys. 13, 45.
Taub, A.H. (1956). Isentropic hydrodynamics in plane symmetric space-times,
Phys. Rev. 103, 454.
Taub, A.H. (1975). Spatially homogenous universe, in Proceedings of the First
Marcel Grossmann Meeting on General Relativity, ed. Ruffini, R. (North-
Holland).
Taub, A.H. (1980). Space-times with distribution valued curvature tensors,
J. Math. Phys. 21, 1423.
Taub, A.H. (1983). General relativistic hydrodynamic wave propagation, in Pro-
ceedings of the Third Marcel Grossmann Meeting On General Relativity,
August 30–September 1, 1982, Shanghai, China, ed. Hu, N. (Science Press,
Beijing).
Taub, A.H. (1988a). On the collision of planar impulsive gravitational waves,
J. Math. Phys. 29, 690.
Taub, A.H. (1988b). Collision of impulsive gravitational waves following dust
clouds, J. Math. Phys. 29, 690.
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-bibliography page 192

192 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

Taub, A.H. (1990). Interaction of null dust clouds fronted by impulsive plane
gravitational waves, J. Math. Phys. 31, 664.
Thorne, K.S. (1967). The general relativistic theory of stellar structure and
dynamics, in High-Energy Astrophysics, vol. Ill, eds. DeWitt, C. Schatzman,
E. and Veron, P. (Gordon and Breach, New York), p. 414.
Thorne, K.S. (1972). In Magic without Magic: John Archibald Wheeler, ed.
J. Klauder (Freeman, San Francisco).
Thorne, K.S. (1992). Gravitational-wave bursts with memory: the Christodoulou
effect, Phys. Rev. D 45, 520.
Tippler, F.J. (1980). Singularities from colliding plane gravitational waves, Phys.
Rev. D 22, 2929.
Tomita, K. (1998). Plane-symmetric vacuum solutions with null singularities for
inhomogeneous models and colliding gravitational waves, preprint, arXiv:gr-
qc/9807033.
Townsend, P.K. and Wohlfarth, M.N.R. (2003). Accelerating cosmologies from
compactification, Phys. Rev. Lett. 91, 061302.
Tsoubelis, D. (1976). Homogeneous relativistic cosmological models of Bianchi
type IV, Ph.D. dissertation, Graduate Faculty in Physics Department, The
City University of New York.
Tsoubelis, D. (1989a). Cylindrical shells of cosmic strings, Class. Quantum Grav.
6, 101.
Tsoubelis, D. (1989b). Plane gravitational waves colliding with shells of null dust,
Class. Quantum Grav. 6, L117.
Tsoubelis, D. and Wang, A. (1989). Asymmetric collision of gravitational plane
waves: a new class of exact solutions, Gen. Relativ. Grav. 21, 807.
Tsoubelis, D. and Wang, A. (1990). Impulsive shells of null dust colliding with
gravitational plane waves, Gen. Relativ. Grav. 22, 1091.
Tsoubelis, D. and Wang, A. (1991). On the gravitational interaction of plane
symmetric null dust clouds, J. Math. Phys. 31, 1017.
Tsoubelis, D. and Wang, A. (1992). Head-on collision of gravitational plane waves
with non-collinear polarization: a new class of analytical models, J. Math.
Phys. 33, 1054.
Tziolas, A. and Wang, A. (2008). Colliding branes and formation of spacetime
singularities, Phys. Lett. B 661, 5.
Tziolas, A. Wang, A. and Wu, Z.-C. (2009). Colliding branes and formation of
spacetime singularities in string theory, J. High Energy Phys. 04, 038.
Verdaguer, E. (1984). Solitons and the generation of new cosmological solutions, in
Observational and Theoretical Aspects of Relativistic Astrophysics and Cos-
mology, eds. Sanz, J.L. and Goicoechea, L.J. (World Scientific, Singapore).
Verdaguer, E. (1987). Plane waves and soliton solutions, Il Nuovo. Cimento B
100, 787.
Verdaguer, E. (1993). Soliton solutions in spacetimes with two spacelike killing
fields, Phys. Rep. 229, 1.
Weyl, H. (1917). Zur gravitationstheorie, Ann. Phys. 54, 117.
Vladimirov, V.S. (1979). Generalized Functions in Mathematical Physics (MIR,
Moscow).
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-bibliography page 193

Bibliography 193

Vilenkin, A. (1981). Gravitational field of vacuum domain walls and strings, Phys.
Rev. D 23, 852.
Vilenkin, A. (1985). Cosmic strings and domain walls, Phys. Rep. 121, 263.
Vilenkin, A. and Shellard, E.P.S. (2000). Cosmic Strings and Other Topological
Defects (Cambridge University Press, Cambridge).
Wainwright, J. Ince, W.C.W. and Marshman, B.J. (1979). Spatially homogeneous
and inhomogeneous cosmologies with equation of state, Gen. Relativ. Grav.
10, 259.
Wald, R.M. (1984). General Relativity (University of Chicago, New York).
Wands, D. (2006). Brane-world cosmology, preprint, arXiv:gr-qc/0601078.
Wang, A. (1991a). The effect of polarization of colliding plane gravitational waves
on focusing singularities, J. Mod. Phys. A 6, 2273.
Wang, A. (1991b). Gravitational Faraday rotation induced from interacting grav-
itational plane waves, Phys. Rev. D 44, 1120.
Wang, A. (1991c). Plane walls interacting with gravitational and matter fields,
J. Math. Phys. 32, 274.
Wang, A. (1991d). Massless particle radiation of Vilenkin’s planar domain walls,
Phys. Lett. B 264, 2863.
Wang, A. (1991e). Planar domain walls emitting and absorbing electromagnetic
radiation, Phys. Rev. D 44, 1705.
Wang, A. (1991f). Interacting Gravitational, Electromagnetic, Neutrino and Other
Waves in the Context of Einstein’s General Theory of Relativity (Iaonnina
University, Greece).
Wang, A. (1992a). Gravitational interaction of plane gravitational waves and
matter shells, J. Math. Phys. 33, 1065.
Wang, A. (1992b). Planar domain walls coupled with pure radiation fields, Int.
J. Mod. Phys. A 7, 4521.
Wang, A. (1992c). Singularities of the space-time of a plane domain wall when
coupled with a massless scalar field, Phys. Lett. B 277, 49.
Wang, A. (1992d). Transparency of domain walls to electromagnetic and neutrino
waves: an exact solution, Mod. Phys. Lett. A 7, 835.
Wang, A. (1992e). Dynamics of plane-symmetric thin walls in general relativity,
Phys. Rev. D 45, 3534.
Wang, A. (1992f). On the interaction of bubbles with gravitational and matter
fields, Mod. Phys. Lett. A 7, 1779.
Wang, A. (1993). Non-trivial interaction of plane domain walls with scalar fields,
Phys. Rev. D 48, 2591.
Wang, A. (1994). Gravitational collapse of thick domain walls: an analytic model,
Mod. Phys. Lett. A 9, 3605.
Wang, A. (2000). On “Comments on collapsing null shells of matter”, J. Math.
Phys. 41, 8354.
Wang, A. (2001). Critical phenomena in gravitational collapse: The studies so far,
Braz. J. Phys. 31, 188.
Wang A. (2003). Critical collapse of cylindrically symmetric scalar field in four-
dimensional Einstein’s theory of gravity, Phys. Rev. D 68, 064006.
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-bibliography page 194

194 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

Wang, A. (2010). Orbifold branes in string/M-theory and their cosmological appli-

cations, preprint, arXiv:1003.4991.
Wang, A. and de Oliveira, H.P. (1997). Critical phenomena of collapsing massless
scalar wave packets, Phys. Rev. D 56, 753.
Wang, A. and Letelier, P.S. (1995a). Local and global structure of a plane domain
wall spacetime, Phys. Rev. D 51, R6612.
Wang, A. and Letelier, P.S. (1995b). Domain wall spacetime: instability of cos-
mological event and Cauchy horizons, Phys. Rev. D 52, 1800.
Wang, A. and Letelier, P.S. (1995c). Dynamical wormholes and energy conditions,
Prog. Theor. Phys. 94, L137.
Wang, A. and Nogales, J.A.C. (1997). Instability of cosmological event horizons
of global non-static cosmic strings, Phys. Rev. D 56, 6217.
Wang, A. and Santos, N.O. (1996). Gravitational and particle radiation from
cosmic strings, Class. Quantum Grav. 13, 715.
Wang, A. and Santos, N.O. (2008). The cosmological constant in the brane world
of string theory on S 1 /Z2 , Phys. Lett. B 669, 127.
Wang, A. and Santos, N.O. (2010). The hierarchy problem, radion mass, local-
ization of gravity and 4D effective Newtonian potential in string theory on
S1 /Z2 , Int. J. Mod. Phys. A 25, 1661.
Wang, T.-Z. Fier, J. Li, B. Lv, G.-L. Wang, Z.-J. Wu, Y. and Wang, A. (2018).
Singularities of plane gravitational waves and their memory effects, preprint,
arXiv:1807.09397 [Gen. Relativ. Grav. in press (2020)].
Whitham, G.B. (1974). Linear and Nonlinear Waves (Wiley, New York).
Weinberg, S. (1972). Gravitation and Cosmology, Principles and Applications of
the General Theory of Relativity (Wiley, New York).
Westenholz, C. (1981). Differential Forms in Mathematical Physics (North-
Holland, New York).
Weyl, H. (1917). Zur Gravitationstheorie, Ann. Phys. 54, 117.
Weyl, H. (1922). Space–Time–Matter, translated by Brose, H. L. (Methuen,
London).
Witten, L. (1962). Gravitation: An Introduction on Current Research (Wiley,
New York).
Wohlfarth, M.N.R. (2003). Accelerating cosmologies and a phase transition in
M -theory, Phys. Lett. B 563, 1.
Wu, Q. Gong, Y.-G. and Wang, A. (2009). Brane cosmology in the Horava–Witten
heterotic M-theory on S 1 /Z2 , J. Cosmol. Astropart. Phys. 06, 015.
Wu, Q. Vo, P. Santos, N.O. and Wang, A. (2008). Late transient acceleration of
the universe in string theory on S 1 /Z2 , J. Cosmol. Astropart. Phys. 09, 004.
Wu, Y. da Silva, M.F.A. Santos, N.O. and Wang, A. (2003). Topological charged
black holes in high dimensional spacetimes and their formation from gravi-
tational collapse of a type II fluid, Phys. Rev. D 68, 084012.
Xanthopoulos, B.C. (1979). Multiple moments in general relativity, J. Phys. A:
Math. Gen. 12, 1025.
Xanthopoulos, B.C. (1981). Exterior spacetimes for rotating stars, J. Math. Phys.
22, 1254.
Xanthopoulos, B.C. (1986a). Rotating cosmic string, Phys. Lett. B 178, 163.
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-bibliography page 195

Bibliography 195

Xanthopoulos, B.C. (1986b). Cylindrical waves and cosmic strings of Petrov type
D, Phys. Rev. D 34, 3608.
Xanthopoulos, B.C. (1986c). The initial value problem for colliding gravitational
and hydrodynamic waves, J. Math. Phys. 27, 2129.
Xanthopoulos, B.C. (1987). Cosmic strings coupled with gravitational and elec-
tromagnetic waves, Phys. Rev. D 35, 3713.
Yurtsever, U. (1987). Instability of Killing-Cauchy horizons in plane-symmetric
space-times, Phys. Rev. D 36, 1662.
Yurtsever, U. (1988a). Structure of singularities produced by colliding plane grav-
itational waves, Phys. Rev. D 38, 1706.
Yurtsever, U. (1988b). New family of exact solutions for colliding plane gravita-
tional waves, Phys. Rev. D 37, 2790.
Yurtsever, U. (1989). Singularities and horizons in the collisions of gravitational
waves, Phys. Rev. D 40, 329.
Zel’dovich, Ya.B. (1980). Cosmological fluctuations produced near a singularity,
Mon. Not. R. Astron. Soc. 192, 663.
Zel’dovich, Ya.B. Kobzarev, [Link]. and Okun, L.B. (1976). Cosmological conse-
quences of a spontaneous breakdown of a discrete symmetry, Sov. Phys.
JETP 40, 1.
Zeldovitch, Ya.B. and Polnarev, A.G. (1974). Radiation of gravitational waves by
a cluster of superdense stars, Sov. Astron. 18, 17.
Zemanian, A.H. (1987). Distribution Theory and Transform Analysis: An Intro-
duction to Generalized Functions, with Applications (Dover Publications,
New York).
Zhang, P.-M. Duval, C. Gibbons, G.W. and Horvathy, P.A. (2017). The memory
effect for plane gravitational waves, Phys. Lett. B 772, 743.
Zhang, P.-M. Duval, C. Gibbons, G.W. and Horvathy, P.A. (2018). Velocity mem-
ory effect for polarized gravitational waves, J. Cosmol. Astropart. Phys. 05,
030.
 b2530   International Strategic Relations and China’s National Security: World at the Crossroads

This page intentionally left blank

b2530_FM.indd 6 01-Sep-16 [Link] AM

March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-index page 197

Index

affine connection, 3 Brinkman, 42

Aichelburg and Sexl, 39, 43 Bronnikov, Santos and Wang, 29, 71,
Alekseev, 162, 171 130, 163, 171
Alekseev and Griffiths, 84
anastigmatic, 82, 102, 107, 108, C-metric, 170
120–122 Campbell and Wainwright, 12, 50
Apostolatos and Thorne, 30 Carr and Verdaguer, 42, 85
astigmatic, 82, 107, 108, 120–122 Carroll, 1, 5
Cauchy horizon, 81–83, 101, 108, 129,
Bäcklund transformations, 84, 171 132, 160, 161, 168, 170, 172
Babala, 135, 140, 143, 144 Centrella and Matzner, 81, 162, 172
Baldwin and Jeffery, 49 Cespedes and Verdaguer, 88
Barrabés, 171 Chandrasekhar, 5, 12, 163, 168
Barrabés and Hogan, 171 Chandrasekhar and Ferrari, 68, 70,
Belinsky, 42, 84 84, 85, 92, 93, 110, 112, 115, 124,
Belinsky and Francaviglia, 88 163, 166
Belinsky and Ruffini, 84 Chandrasekhar and Xanthopoulos,
Belinsky and Verdaguer, 42, 67, 84, 70, 77, 81, 82, 85, 94, 101, 102, 124,
85, 171 130, 145, 164, 167, 169, 172
Belinsky and Zakharov, 42, 67, 84, Chandrasekhar–Ferrari method, 84,
86, 103 85
Bell–Szekeres solution, 129, 161 characteristic, 48, 54, 101, 173
Bianchi identities, 3, 5, 11, 15, 20, 37, Chernoff and Tye, 29
53, 58, 60, 66, 73 Choptuik, 172
Bieri, Garfinkle and Yunes, 39, 48, Christodoulou, 48
52, 172 Clarke and Dray, 31
Blanchet and Damour, 48 Clarke and Hayward, 160, 172
Blelinsky and Zakharov, 171 collinearly, 42, 56, 73, 81, 84, 91, 102,
Bondi and Pirani, 42, 49 103, 105, 119, 127
Bonnor, 170 collinearly polarized, 102, 130
Bonnor and Vichers, 28 compact support, 32, 37, 61
Boyer–Lindquist, 168 complex poles, 87
brane world, 30, 60 Connors and Stark, 70

197
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-index page 198

198 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

constantly polarized, 48, 91, 111, 112, electromagnetic field, 25, 79, 80, 82,
119, 131, 134, 138–140, 142 83, 85, 156
continuity, 33 electromagnetic plane waves, 129,
coordinate singularities, 49, 50 160, 161, 169
coordinate transformation, 4, 44, 46, Ellis and Schmidt, 49
49, 165 energy conditions, 27, 146, 169
cosmic acceleration, 30 energy density, 5, 27–29, 77, 145, 151,
cosmic censorship, 172 155
cosmic strings, 29 energy–momentum tensor, 2, 18, 28,
cosmic superstrings, 29 158
cosmological constant, 2, 5, 17 equation of state, 27, 84, 145, 149,
Coulomb field, 66, 78 150, 153
critical phenomena, 172 Ernst equation, 85, 92, 93, 163
curvature singularities, 49, 50, 81, 83, Ernst potential, 68, 85, 92, 124, 127,
96, 101, 107, 108, 122, 130, 161 166
cylindrical Ernst equation, 163 event horizon, 168
cylindrical spacetimes, 163, 171 expansion, 24, 28, 41, 96, 102

D’Inverno, 1, 5, 43 Faraday effect, 70, 76

Darmois, 28 Faraday rotation, 53, 70, 71, 76, 173
Darmois’ junction conditions, 28 Feinstein and Ibañez, 67, 81, 94, 96
diagonal matrix, 88 Ferrari, 67
diagonal solutions, 91, 103, 107, 108, Ferrari and Ibañez, 67, 81, 95, 101,
116, 120, 127 102, 125, 145, 163–165, 169
dilaton gravity, 161 Ferrari–Ibañez solution, 95, 125
Dirac delta function, 31, 32 fluid, 27, 83–85, 145, 146, 149–153
discontinuity, 28, 30, 34, 55, 63, 131 focusing hypersurface, 82, 101, 102,
distortions, 39, 49–51 119, 164, 165, 168
distribution derivative, 33 Frolov, 1, 12
distribution theory, 31, 171
distribution-valued tensors, 1 generalized functions, 1
distributions, 3, 32, 34, 53, 60, 64, geodesic congruence, 22, 24, 28, 45,
159, 172 47, 48, 56
domain walls, 29, 60 geodesic deviation, 1, 40
dominant energy condition, 28 geodesic deviation equations, 8
Dray and ’t Hooft, 63, 130, 134 Geroch, 83
Dvali and Vilenkin, 29 Gong, Wang and Wu, 30
gravitational collapse, 172
Economou and Tsoubelis, 29, 88, 89, gravitational radiation, 6, 137, 138
124 gravitationally active, 66
Ehlers and Kundt, 40–42 gravitationally inert, 66
eigenvalue problem, 86 Griffiths, 1, 31, 44, 59, 67, 68, 74, 80,
Einstein field equations, 1–3, 5, 12, 81, 94, 101, 127, 132, 159, 161, 163,
14, 25, 28, 36, 53, 67, 68, 77, 83, 94, 172
129–131, 145, 146, 154, 159, 163 Griffiths and Podolský, 12, 171
Einstein tensor, 5 Grishchuk and Polnarev, 49
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-index page 199

Index 199

Harrison, 84, 171 Kerr–Newman charged black hole,

Hauser and Ernst, 54, 84 169
Hauser–Ernst formulation, 84 Kerr–Newman–NUT solution, 169
Hawking and Ellis, 2, 6, 49, 125, 146, Kerr–NUT solution, 124, 168
169 Khan and Penrose, 59, 81, 95, 102,
Hawking, Perry and Strominger, 39, 109, 130, 134, 166
48, 52, 173 Khan–Penrose solution, 95, 102, 130,
Heaviside function, 59 134, 166
Herrera and Santos, 30 Khan–Penrose substitutions, 96, 97,
hierarchy problem, 30 104, 119, 131, 149, 160, 167, 169
high-dimensional spacetimes, 161 Kibble, 29
Hirschmann, Wang and Wu, 50, 51 Killing vectors, 132, 168, 169, 171
Hoenselaers and Ernst, 44 Kinnersley, 12
Hoenselaers–Kinnersley– Kinnersley and Walker, 170
Xanthopoulos transformations, Klein–Gordon equation, 25, 154
84 Kramer and Neugebauer, 84
hoop conjectures, 172 Kramer et al., 1, 10, 12, 25, 83, 124
Horava and Witten, 30 Kretschmann scalar, 21
Kronecker delta, 3
Ibañez and Verdaguer, 42, 84
impulsive gravitational waves, 112, Lanczos, 28
130, 134, 140, 161 Landau and Lifshitz, 5
impulsive shells, 63, 77, 130, 131, 134, Legendre functions, 94
137, 142 Letelier, 88
index, 53, 56, 67, 68, 70, 76, 77, 83 Letelier and Wang, 29, 30, 60
multiple indexes, 81 Lichnerowicz, 5, 28, 29
initial conditions, 54 Linde, 29
initial data, 59, 81, 101 linearly polarized, 42, 44, 70, 71
instability, 81 longitudinal, 16
integrability conditions, 86, 89 Lorentz boost, 39, 43
internal spacetimes, 163
intrinsic derivatives, 15 Maartens, 30
inverse scattering method, 67, 84, 103 Maartens and Koyama, 30
Ipser and Sikivie, 29 MacCallum, 83
isometric, 163, 168, 169, 172 manifold, 2, 3, 6, 31, 127
Israel, 28, 130, 171 massless particles, 129, 145
Israel junction conditions, 171 massless scalar waves, 129, 146
Maxwell equations, 25, 26, 171
Jantzen, 88 memory effects, 39, 48, 49, 51, 52,
junction conditions, 28 172
Misner, Thorne and Wheeler, 5
Kasner axes, 82, 107, 108, 120, 121 monopoles, 29
Kasner exponent, 107, 108, 120, 121 mutual focusing, 91, 135
Kasner solution, 82, 102
Kerr black hole, 168 Neugebauer, 84, 171
Kerr solution, 85, 166–169 Neugebauer–Kramer involution, 84
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-index page 200

200 Interacting Gravitational, Electromagnetic, Neutrino and Other Waves

neutrino field, 25–27, 80, 157 Ricci scalars, 25, 26, 45, 48, 53, 63,
Newman and Penrose, 1, 12, 15, 25 70, 76–78, 80, 92, 146
non-collinearly, 56, 73, 91, 102, 103, Rosen, 49
105, 106, 119, 129, 161
nonlinear interaction, 76, 77, 79, 142, Sachs, 40
171–173 sandwich waves, 55
NP formalism, 1, 12, 14, 16 Schmidt and Wang, 30, 60
null field, 39–41 Schwarzschild black hole, 163–165
null shell, 36 second fundamental form, 28, 31
null tetrad, 44, 56, 157 seed solutions, 103, 107, 116, 119
NUT parameter, 123, 124, 168 Sephani et al., 169
Nutku and Halil, 66, 81, 166 shear, 40, 41, 56, 57
Nutku–Halil solution, 102, 109, 110, singular surface, 28, 30, 31, 49, 60
112, 115, 123, 124, 166 soft graviton theorems, 49, 51
soliton solutions, 103
O’Brien and Synge, 28, 29 soliton structure, 95, 96
one-soliton solutions, 125, 138 soliton technique, 67
optical scalars, 1, 24 space-like singularity, 82, 83
Ori, 50, 51 spin coefficients, 13–16, 21, 26, 27, 45,
56, 62, 80
Papacostas and Xanthopoulos, 161, spinor, 26, 30, 80
169 Stark and Connors, 70
parallel transport, 1, 24 Stephani et al., 1, 10, 12, 25, 50, 83,
parallelly transported, 6, 8, 47, 53, 161, 168, 171
71, 72, 75, 76, 115 stochastic, 29
Penrose, 12, 30, 42, 172 string network, 29
Pereira and Wang, 30, 31 Strominger, 48, 52
perturbations, 29, 81, 84, 101, 168, strong energy condition, 28
172 Szekeres, 44, 66, 77, 80, 81, 95, 130,
phase transitions, 29, 129, 154 134
Piran and Safier, 71 Szekeres family, 127, 134
Pirani, 5, 12 Szekeres solution, 95, 127
plane-fronted, 41, 43, 162, 169, 172
polarization angles, 73, 110, 115 Taub, 63, 145, 146, 171
polarization direction, 56, 72, 76, 116, Taub–NUT solution, 125, 169
145 tetrad components, 9, 15, 17
polarizations, 71, 76, 77, 81, 82, 91, tetrad frame, 14, 15
93, 109, 114, 142, 143, 166, 170 Thorne, 5, 48, 172
pressure, 27, 145, 151 tidal forces, 50
Pretorius and East, 162, 172 topological defects, 29
pure radiation field, 25, 39, 40, 77, 81 Tsoubelis, 63, 130, 134, 135
Tsoubelis and Wang, 26, 63, 67, 68,
radiation field, 79 73, 77–83, 89, 94, 95, 97, 102, 103,
Randall and Sundrum, 30 126, 130, 145, 146, 148, 153, 163
real poles, 87 two-soliton solutions, 91, 116, 126,
Ricci identity, 4, 14 127
March 7, 2020 9:54 Interacting Gravitational, Electromagnetic. . . – 9in x 6in b3813-index page 201

Index 201

Tziolas and Wang, 30, 60, 161 Wang and Letelier, 30

Tziolas, Wang and Wu, 161 Wang and Santos, 29, 30, 60
wavefront, 53–55, 73, 100, 106, 137,
unbounded wavefront, 99 159, 164, 165
uniqueness, 129, 145, 146, 154 weak energy condition, 28
Weyl scalars, 16, 39, 97, 98, 100, 109,
variably polarized, 112, 113, 116, 119, 110, 112, 114, 115, 132, 135,
138–142 140–144, 161
Verdaguer, 42, 85, 171 Weyl tensor, 11, 12, 16, 45
Vilenkin, 29 Wu, Gong and Wang, 30, 60
Vilenkin and Shellard, 29
Xanthopoulos, 29, 84
Wald, 1, 5
Wang, 12, 30, 44, 60, 66, 67, 94, 119, Yurtsever, 59, 67, 81, 101, 102, 132,
163 160, 163, 168, 172
Wang and de Oliveira, 30, 31