space, time and matter
TRANSCRIPT
-
TIME AND MATTERSPACE
8897hc_9789814522830_tp.indd 1 21/2/14 5:22 pm
-
May 2, 2013 14:6 BC: 8831 - Probability and Statistical Theory PSTws
This page intentionally left blankThis page intentionally left blank
-
N E W J E R S E Y L O N D O N S I N G A P O R E BE IJ ING S H A N G H A I H O N G K O N G TA I P E I C H E N N A I
World Scientific
Dipak K SenUniversity of Toronto, Canada
TIME AND MATTERSPACE
8897hc_9789814522830_tp.indd 2 21/2/14 5:22 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 DataA catalogue record for this book is available from the British Library.
SPACE, TIME AND MATTER
Copyright 2014 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 978-981-4522-83-0
Typeset by Stallion PressEmail: [email protected]
Printed in Singapore
-
February 22, 2014 14:3 9in x 6in Space, Time and Matter b1716-fm
PREFACE
In 1918, Hermann Weyl wrote a book entitled Raum-Zeit-Materie.
This short monograph, with the same title, however, treats the same
subject matter in a somewhat unconventional way.
We present a novel, albeit equivalent, formalism of relativistic kine-
matics and general relativistic field dynamics in which time does not
have the same primary role as space as in conventional relativity.
Finally, we present a theory of formation of fundamental particles
where the fundamental constituents are left-handed and right-handed
2-component Weyl neutrinos.
The material presented here uses the technics of geometry of
manifolds extensively, in particular, that of vector fields on mani-
folds. Readers unfamiliar with this subject should start first with
Appendices A and B.
Dipak K. Sen
Toronto, Canada
December, 2013
v
-
May 2, 2013 14:6 BC: 8831 - Probability and Statistical Theory PSTws
This page intentionally left blankThis page intentionally left blank
-
February 22, 2014 14:3 9in x 6in Space, Time and Matter b1716-fm
CONTENTS
PREFACE v
1. SPACE AND TIME 1
1.0. Introduction . . . . . . . . . . . . . . . . . . . . . . . 1
1.1. The Hyperbolic Structure of the Space
of Relative Velocities . . . . . . . . . . . . . . . . . . 4
1.2. Relativistic Kinematics on 3-Manifolds . . . . . . . . 8
1.3. Class of Inertial Observers and Generalized
Lorentz Matrices . . . . . . . . . . . . . . . . . . . . . 11
1.4. Classical Relativistic Field Dynamics . . . . . . . . . 13
1.4.1. One-dimensional Heat equation . . . . . . . . . 13
1.4.2. One-dimensional Wave equation . . . . . . . . 17
1.4.3. GaussEinstein equations . . . . . . . . . . . . 18
1.5. Relativistic Field Dynamics on 3-Manifolds . . . . . . 21
1.5.1. Three-dimensional field equations
and relationship with Einstein equations . . . 21
1.6. Flat-Space Solutions . . . . . . . . . . . . . . . . . . . 24
1.7. The de Sitter Solution . . . . . . . . . . . . . . . . . . 26
1.8. A New Solution of the Vacuum Einstein
Field Equations . . . . . . . . . . . . . . . . . . . . . 27
vii
-
February 22, 2014 14:3 9in x 6in Space, Time and Matter b1716-fm
viii Space, Time and Matter
1.9. A Solution of Mathematical Interest . . . . . . . . . . 34
1.10. The Schwarzschild Solution . . . . . . . . . . . . . . . 38
1.10.1. The Schwarzschild metric . . . . . . . . . . . . 39
1.11. From Space-Time 4-Metric to 3-Metric
and 3-Vector Field . . . . . . . . . . . . . . . . . . . . 44
1.12. The Kerr Solution . . . . . . . . . . . . . . . . . . . . 49
1.13. The Maxwell Equations . . . . . . . . . . . . . . . . . 56
References . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
2. MATTER 61
2.0. Introduction . . . . . . . . . . . . . . . . . . . . . . . 61
2.1. The Photon and the Weyl Neutrinos . . . . . . . . . 63
2.2. A Neutrino Theory of Matter . . . . . . . . . . . . . 68
2.2.1. Composite LR system without
interaction . . . . . . . . . . . . . . . . . . . . 68
2.2.2. Composite LR system with interaction . . . 70
2.2.3. A reduced 1-dimensional model . . . . . . . . . 72
2.2.4. Conclusion . . . . . . . . . . . . . . . . . . . . 77
References . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
APPENDIX A. VECTOR FIELDS ON MANIFOLDS 79
A.1 Vector Fields on Manifolds . . . . . . . . . . . . . . . 79
A.2 Example of a Vector Field whose Integral
Curves are Knots . . . . . . . . . . . . . . . . . . . . 83
References . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
APPENDIX B. DYNAMICAL VECTOR FIELDS
OF CLASSICAL MECHANICS 87
B.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . 87
-
February 22, 2014 14:3 9in x 6in Space, Time and Matter b1716-fm
Contents ix
B.2 Configuration Spaces of Mechanical Systems
as Manifolds . . . . . . . . . . . . . . . . . . . . . . . 89
B.3 Dynamical Vector Fields
in Classical Mechanics . . . . . . . . . . . . . . . . . . 99
References . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
APPENDIX C. MAPLE PROGRAM 1 129
APPENDIX D. MAPLE PROGRAM 2 131
APPENDIX E. MAPLE PROGRAM 3 133
AUTHOR INDEX 137
SUBJECT INDEX 139
-
May 2, 2013 14:6 BC: 8831 - Probability and Statistical Theory PSTws
This page intentionally left blankThis page intentionally left blank
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
1. SPACE AND TIME
1.0. Introduction
It was Minkowski who paraphrased that, . . .Henceforth space by
itself, and time by itself, are doomed to fade away into mere shadows,
and only a kind of union of the two will preserve an independent
reality. And ever since relativists are accustomed to always think in
the framework of a 4-dimensional space-time, it has almost become
second nature to them.
Strictly speaking, however, the Minkowski signature ( + + +)makes a clear distinction as to which coordinates can be interpreted
as spatial and which ones as temporal, and recently, the usefulness
of 3 + 1 decomposition techniques has shown that sometimes the
separation of space from time provides a better insight to what is
going on rather than the traditional fusion of space with time.
Imagine a universe in which nothing moves. In such a universe, the
notion of time disappears. So, space and motion in space are really
the primary concepts and time is a secondary concept. It should thus
be possible to construct a theory of Kinematic Relativity based on
the primary notions of space and motion instead of space and time.
1
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
2 Space, Time and Matter
In this monograph, we present a new formulation of both relati-
vistic kinematics and eld dynamics in an entirely 3-dimensional
space. The basic idea is to represent local physical observers as
non-singular 3-dimensional local vector elds and the dynamics
of physical elds by the ows of the vector elds, which are
1-parameter groups of local smooth transformations of the space.
The ow parameter thus plays the role of local time for each physi-
cal observer, and Lie-derivative replaces the time derivative.
In the kinematics part, we rst introduce the basic notions and
postulates of our formalism which take place in a 3-dimensional Rie-
mannian space (M 3, g). A physical observer is dened to be a non-
singular vector eld X on M3 with 0
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
Space and Time 3
In the dynamics part, we rst show how one can describe the time
evolution of certain classical elds by subjecting the 3-dimensional
elds to a transformation by the ow of a fundamental 3-dimensional
vector eld. We illustrate this with some simple 1-dimensional cases
such as the 1-dimensional Heat and Wave equations. We next apply
our basic procedure to the GaussEinstein equations, that is, the
Einstein eld equations in Gaussian normal coordinates, and obtain
a set of equations for the 3-metric g and a 3-dimensional vector eld
X. Every solution (g,X) of these equations determine uniquely a
space-time 4-metric solution of the Einstein eld equations.
Our rst example is the de Sitter solution. A generalization of the
de Sitter case leads us to a complex metric of Kasner type, which
in turn, provides us with a new (real) solution of the vacuum Ein-
stein eld equations, containing ve real parameters. We then give
another solution of our 3-dimensional equations which is mathemat-
ically interesting and non-trivial, but however, corresponds to the
physically trivial at space-time metric. Nevertheless, this solution
clearly illustrates the basic principles of our 3-dimensional approach.
Finally, we apply our formalism to the Maxwell equations and
obtain a corresponding 3-dimensional set of equations, which turn
out to be, not only concise, but also elegant.
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
4 Space, Time and Matter
1.1. The Hyperbolic Structure of the Spaceof Relative Velocities
We rst consider the geometry of the space of relative velocities in
conventional Special Relativity. This will suggest the formalism of a
theory of Kinematic Relativity on 3-manifolds which reproduces the
essential features of conventional relativity.
In the special theory of relativity, the velocity transformation for-
mula relating the relative velocities of three equivalent inertial sys-
tems u, v, w is given by (in units c = 1)
w2 = (u v)2 (u v)2
(1 u v)2 . (1.1)
Equation (1.1) has the following geometrical signicance. Let us
choose some reference inertial system and consider the velocities u =
(u1, u2, u3) of all other equivalent inertial systems relative to this
reference inertial system. Since for physical inertial systems u =[|(ui)2|]1/2 < 1, the space of relative velocities is topologically a 3-dimensional open disk
D3 = {u R3|u < 1}.
Note: For tachyons, the appropriate space to consider would be the
closure of the complement of D3, that is, R3\D3.Now, we introduce the standard (positive denite) hyperbolic met-
ric G on D3 by
ds2 =i,k
Gik(u)duiduk
=[1(ui)2][(dui)2] + [uidui]2
[1(ui)2]2 , (1.2)
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
Space and Time 5
that is,
Gik(u) =[1(ui)2]ik + uiuk
[1(ui)2]2 .Then, H3 =(D3, G) is the standard hyperbolic 3-space of con-
stant curvature 1. This is the so-called Poincare disk model ofH3 ([1]). Alternatively, one can introduce homogeneous coordinates
= (0, 1, 2, 3) by i = 0ui and dene for = (0, 1, 2, 3), =
(0, 1, 2, 3), the inner product (, ) = 00
i ii. Then,
ds2 =(d, )(, d) (, )(d, d)
(, )2. (1.3)
H3 is geodesically complete and has a distance function d given by
d(u, v) = cosh1
[1
iuivi
][1
i(u)2
]1/2 [1
i(v)2
]1/2 (1.4)
or, in homogeneous coordinates,
d(, ) = cosh1[ |(, )|(, )1/2(, )1/2
]. (1.5)
A comparison of Eq. (1.1) with Eq. (1.5) suggests that the dis-
tance d(u, v) is somehow related to the relative velocity w between
two inertial systems whose velocities relative to the reference inertial
system are u and v. To see the actual relationship, we should note
that the reference system is at rest relative to itself. Now, from (1.4)
d(0, v) = cosh1[
1(1 v2)1/2
]= tanh1 v.
That is, v = tanh d(0, v). We can, therefore, reinterpret Eq. (1.1) asa relation between the relative velocity w and the hyperbolic distance
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
6 Space, Time and Matter
4
3
2
10
1
2
3
4
y
4 2 2 4x
Fig. 1. The hypersurfaces M,M.
d(u, v) by
w = tanh[d(u, v)]. (1.6)
The isometry group of H3 is intimately connected with the Lorentz
group, as can be seen in what follows.
First, we consider an abstract R4 with the Lorentz metric gR4 given
by ds2 = (dx1)2+(dx2)2+(dx3)2(dx4)2. Consider now the hypersur-face M of R4: M = {x = (x1, x2, x3, x4) R4|(x1)2+(x2)2+(x3)2(x4)2 = 1}. M has two components: M = {x M|x4 1} andM = {x M|x4 1} (see Fig. 1; here (x, y) represents (x1, x4)).
Both M and M are dieomorphic to D3 and the dieomorphism
map is given by
(x1, x2, x3, x4) (u1, u2, u3),ui = xi/x4, (i = 1, 2, 3)
with its inverse
(u1, u2, u3) (x1, x2, x3, x4),
xi = uix4, x4 = [
1(1(ui)2)
]1/2.
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
Space and Time 7
Now, let gM be the induced Riemannian metric on M obtained
from gR4 on R4. In coordinates (ui), gM is given exactly by (1.2).
Thus, H3 = (D3, G) is isometric to (M , gM |M ) and (M , gM |M ).
The isometry group ofH3 was shown by Poincare to be PSL(2,C),
the projective special linear group in two complex dimensions and is,
thus, isomorphic to L+, the orthochronous Lorentz group. It is clear
from (1.5) that the isometry group of H3 leaves the inner prod-
uct (, ), in homogeneous coordinates, invariant. Let its action be
given by
(0, 1, 2, 3) (0, 1, 2, 3),
where
= L , L
L+ (, = 0, 1, 2, 3).
Since ui = i0, L+ acts on H
3 as a fractional linear transforma-
tion:
ui ui = (Li0 + Likuk)/(L00 + L0kuk).
Thus, the fundamental aspect of the Special Theory of Relativity,
namely, Lorentz invariance, is contained in the hyperbolic structure
of the space of relative velocities.
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
8 Space, Time and Matter
1.2. Relativistic Kinematics on 3-Manifolds
Let (M3, g) be a 3-dimensional space with a (positive-denite) Rie-
mannian metric g. By an observer (or a particle path), we usually
mean a smooth curve in M3, i.e., a smooth map c : I M3 from areal interval I into M3.
Since, to every such curve c, one can associate a local vector eld C
such that c is an integral curve of C ([2]). We shall regard the set of
all non-singular vector elds on M3 as the set of all observers in M3.
Non-singularity is necessary to avoid the possibility that an
observer comes to an absolute rest relative to M3. To each vector
eld X corresponds a local ow t , i.e., a local 1-parameter group
of local transformations of M3. Here, the parameter t measures the
ow of time as perceived by the observer represented by X. In this
sense, time is a local concept since X need not be globally dened
and is not necessarily complete unless M3 is compact ([3]).
Let us now consider only those non-singular vector elds X such
that g(X,X) < 1. Such observers will be called physical observers.
(Corresponding to the fact that no such observer can attain the veloc-
ity of light.)
From now on, we consider only physical observers and denote by
(M3) the set of all physical observers.
We dene a map V , called the relative velocity map, from a pair
of physical observers to a smooth function on M3 as follows:
V : (M3)(M3) C(M3),(X,Y ) V (X,Y ),
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
Space and Time 9
where
V (X,Y ) =[1 1
[F (X,Y )]2
]1/2,
F (X,Y ) =1 g(X,Y )
[1 g(X,X)]1/2[1 g(Y, Y )]1/2 .(1.7)
Here, C(M3) denotes the set of all smooth functions on M3.
Let us now dene an equivalence relation in (M3) by:
X Y if and only ifV (X,Y ) = constant function on M3.
(1.8)
In other words, the observers X and Y are said to be (inertially)
equivalent if the relative velocity function is constant on M3.
We shall now give theoretical denitions of space and time inter-
vals of a physical particle path relative to any physical observer X.
Let c : I M3, c() be a particle path and C any represen-tative corresponding vector eld associated with it such that c is an
integral curve of C. For a physical particle path, 0 < g(C,C) < 1
on c(I).
Definition. The time interval of c relative to an observer X, denoted
by tX , is given by the following integral over I:
tX =IF (X,C)d. (1.9)
Note that we can reparametrize c by tX , where dtX/d = F (X,C).
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
10 Space, Time and Matter
Definition. The space interval of c relative to the observer X,
denoted by dX , is given by
dX =IV (X,C)dtX =
IV (X,C)F (X,C)d. (1.10)
If (tY ,dY ) are time and space intervals of c relative to another
observer Y , then (tX ,dX) (tY ,dY ) provides a space-timetransformation from observer X to observer Y .
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
Space and Time 11
1.3. Class of Inertial Observers and GeneralizedLorentz Matrices
We now consider the problem of constructing a class of inertial
observers starting from a representative physical observer. In other
words, given a physical observer u, the problem of constructing an
observer w such that u and w are inertially equivalent, that is, the
relative velocity between u and w, V (u,w), as given by (1.7), is a
constant function on M 3.
Since M3 is 3-dimensional, it is parallelizable, that is, there exists a
global framing of M3 by means of three linearly independent vector
elds {e(i)}, i = 1, 2, 3, so that any vector eld u can be expressedas u = ui(x)e(i), and g(e(i), e(j)) = gij(x), where ui(x), gij(x) are
functions on M3. One can, of course, also consider everything locally
in some local coordinate system (xi), so that ui(x), gij(x) are func-
tions of the coordinate (xi). Since u is a physical observer, we have
g(u, u) = gikuiuk = uiui < 1, where ui = gikuk.
Let
u = [1 g(u, u)]1/2Au = (u 1)/g(u, u)
}(1.11)
and let us dene the 4 4 matrix functions L(u), , = 0, 1, 2, 3 asfollowsa:
L00(u) = u, L0k(u) = uuk
Lk0(u) = uuk, Lkj (u) =
kj + Auu
kuj.
(1.12)Note that in (1.8) uj dier from uk by the lowering of indices by
gjk and also that L are functions on M . These generalized Lorentz
aIn what follows Latin indices i, j, k, . . . = 1, 2, 3 and the Greek indices
, , , . . . = 0, 1, 2, 3.
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
12 Space, Time and Matter
matrices L(u) are easily seen to satisfy the following generalized
properties:
gikLi0L
k0 = (L
00)
2 1gmnL
m0 L
ni = L
00L
0i
gmnLmi L
nj = L
0i L
0j + gij
(1.13)and when (M3, g) = (R3, ), that is, when (M3, g) is the Euclidean
space R3 with the Euclidean metric gik = ik, L(u) reduces to
the usual Lorentz matrix L(u) with the pure boost given by u =
(u1, u2, u3) and satisfy the usual properties:
Li0Li0 = (L
00)
2 1Lm0 L
mi = L
00L
0i
Lmi Lmj = L
0i L
0j + ij .
(1.14)Let v = vi(x)e(i) be another vector eld where g(v, v) < 1.
We now dene a vector eld w = wi(x)e(i) by
wi =Li0(u) + L
ik(u)v
k
L00(u) + L0k(u)vk
. (1.15)
Then it follows that g(w,w) < 1, so that w also represents a physical
observer.
Using the generalized properties (1.9) one now shows that the
relative velocity function V (u,w) as given by (1.1) satises:
[V (u,w)]2 = g(v, v).
So, if v is chosen so that g(v, v) = const., the observers u and w
are inertially equivalent and, in this way, starting from u, we can
construct a class of inertially equivalent observers.
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
Space and Time 13
1.4. Classical Relativistic Field Dynamics
In classical eld theory, the dynamics of a eld is usually described
by a set of evolution equations (together possibly with some con-
straint equations) in some space-time coordinates. We shall now
demonstrate that in many cases the dynamics can also be described
in a purely spatial (i.e., 3-dimensional) setting without the neces-
sity of introducing explicitly a time-coordinate. Instead of the time-
coordinate and time-dependent quantities, the temporal evolution is
provided by the ow of a 3-dimensional vector eld representing a
physical observer.
Before considering the Einstein eld equations in some detail, we
shall rst illustrate our basic approach with some simple examples
from classical eld theory, such as the Heat and Wave equations. For
simplicity, we restrict our analysis to the 1-dimensional case only.
The generalization to three spatial dimensions is obvious.
1.4.1. One-dimensional Heat equation
Consider the 1-dimensional Heat equation with some initial data:
t = xx(x, 0) = (x).
}(1.16)
We look for solutions (x, t) of (1.16) which come from the initial
data (x) through a transformation by the ow of a vector eld
X depending on the spatial coordinate only, as follows. Let X be
given by
X = (x)d
dx. (1.17)
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
14 Space, Time and Matter
Consider its integral curves given by the dynamical system
dx
dt= (x). (1.18)
That is,dt =
dx
(x)= (x), say. So that x(t) = 1(t + c), c =
const., and x(0) = 1(c) or c = (x(0)). The ow t of X transforms
the initial point x(0) onto the point x(t), that is,
t : x t(x) = 1(t + (x)). (1.19)
Now letb
(x, t) = (t)(x)
= (t(x))
= (1(t + (x)))
(1.20)so that
(x, 0) = (x).
Then (x, t), which now also depends on the ow parameter t, is
the transformed initial data (x), which of course depends only on
the space variable x. Let us now substitute (x, t) in (1.16).
Put t + (x) = z, and (x) = y. Then, 1(y) = 1/(x) = (x);
1(y) = (x)(x). And
t(x, t) = (1(z)
)1(z)
t(x, 0) = (x)(x) = LX
}(1.21)
x(x, t) = (1(z)
)1(z) (x)
x(x, 0) = (x)1(y) (x) = (x)
}(1.22)
bGeometrical objects (e.g., scalar, vector, tensor fields and densities) can be
transformed by the flow of a vector field ([3]). We could have also transformed (x)
by t1 = t , by reversing the flow, that is, by letting (x, t) = (t
1(x)).This would have preserved the scalar character of under the flow.
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
Space and Time 15
xx(x, t) = (1(z)
)[1(z)
]2[ (x)
]2+(1(z)
)1(z)[ (x)]2
+(1(z))1(z) (x). (1.23)
xx(x, 0) = (x)[1(y)
]2[ (x)
]2 + (x)1(y)[ (x)]2+(x)1(y) (x)
= (x)[(x)
]2[(x)
]2+(x)(x)(x)
[(x)
]2(x)(x)(x)[(x)]2 = (x). (1.24)
If our (x, t) is to satisfy (1.16), we must have
(1(z)
)1(z) =
(1(z)
)[1(z)
]2[ (x)
]2+
(1(z)
)1(z)
[ (x)
]2+
(1(z)
)1(z) (x). (1.25)
Let us put 1(z) = u. So that again, 1(z) = 1(u) = (u),
1(z) = (u)(u). We can write (1.25) as
(u)(u) =[(u)
[(u)
]2 + (u)(u)(u)][ (x)]2+(u)(u) (x).
The variables x and u can thus be separated to give
1 (x)[ (x)]2
=(u)(u)
(u) + (u)
or
[(x)
]2 + (x) = (u)(u)
(u) + (u) = (const.) (1.26)
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
16 Space, Time and Matter
which must be satised for all x and u. In particular, for u = x, we
must have from (1.26)
(x)(x) = (x)
or LX = .
}(1.27)
Note that (1.27) also follows directly from (1.16) and the initial val-
ues in (1.21), (1.24), and that (1.27) can be obtained from (1.16)
by replacing (x, t) with (x) and the time-derivative t with the
Lie-derivative LX with respect to X. Any solution (,X) of (1.24)provides a solution of (1.16), because from (1.27), ([3]),
t =
t(t) = t(LX) = t() = (t) = . (1.28)
Since (1.26) or (1.27) are equations for both X and , the initial
data cannot be arbitrary and, therefore, our procedure provides only
special solutions of (1.16). We shall give two examples of such solu-
tions ([4]).
Example 1. (x) = ex, (x) = = const., X = ddx . Then (1.27)
is satised, and (x) = x , 1(y) = y, 1
(t + (x)
)= t + x. So
(x, t) = (1(t + (x))
)= (t + x) = e
2t+x. From (1.26), we
get 2 = . If we also allow to be negative, we also obtain complex
solutions (x, t) = e2tix with = i.
Example 2. (x) = x2, (x) = 1x , X =1x
ddx . Again, (1.27) is
satised and (x) = x2
2 , 1(y) = 2y, 1(t+(x)) = 2t + x2.
So (x, t) = (1(t + (x))
)= (2t + x2) = 2t + x2. This is a
well-known polynomial solution of (1.16).
In fact, it is easily possible to obtain the general solution of (1.26)
or (1.27).
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
Space and Time 17
1.4.2. One-dimensional Wave equation
Consider now the 1-dimensional Wave equation with some initial
data:
tt = xx(x, 0) = (x)
t(x, 0) = (x).
(1.29)Again let X = (x) ddx and its ow t be given by (1.16), where
(x) = 1(x)
, and (x, t) = (1(t + (x))
). Now since t(x, 0) =
LX, we must have
LX = . (1.30)
From (1.24)
tt(x, t) = (1(z)
)[1(z)
]2 + (1(z))1(z)tt(x, 0) = (x)
[(x)
]2 + (x)(x)(x).}
(1.31)
It follows from (1.24), (1.29), (1.31) that
(x)[(x)
]2 + (x)(x)(x) = (x)or LXLX = .
}(1.32)
Again (1.32) is obtained from (1.29) by replacing (x, t) by (x)
and the time-derivative t by the Lie-derivative LX . Any solution of(1.30), (1.32) provides a solution of (1.29). Note that the initial data
and now have to be related by the constraint equation (1.30)
and (,X) must satisfy (1.32). We give two special solutions of (1.30)
and (1.32), and hence of (1.29).
(i) (x) = +1, X = ddx , (x) = x, 1(t + (x)) = t + x, (x, t) =
(x + t) with (x) = (x),
(ii) (x) = 1, X = ddx , (x) = x, 1(t + (x)
)= t + x,
(x, t) = (x t) with (x) = (x).
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
18 Space, Time and Matter
It is also easily possible to obtain the general solution of (1.30)
and (1.29).
1.4.3. GaussEinstein equations
In conventional General Relativity the space-time metric g satises
the Einstein eld equations, which are basically evolution equations
with certain constraints, and it is well known ([5]) that the vacuum
Einstein eld equations: R = 0 equivalent to
G0 = 0 (constraints equations)
Rik = 0 (evolution equations)
}(1.33)
provided that g00 = 0; for example, on a non-null hypersurface. HereG R 12gR is the Einstein tensor. A slight modicationis necessary if one wishes to include a non-zero cosmological con-
stant in the eld equations. The vacuum Einstein eld equations
with a cosmological constant: R = g also split up into a set of
constraint and a set of evolution equations and they take a particu-
larly simple form if one uses a Gaussian normal coordinate system in
which g00 = g00 = 1 and g0i = g0i = 0. Then the eld equations:
R0i = 0
R00 = Rik = gik
(1.34)are equivalent to:
G0i = 0
G00 =
}(constraint equations)
Rik = gik (evolution equations)
(1.35)
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
Space and Time 19
because (1.34) implies (1.35), since G0i = g00Ri0 + g0kRik = 0 and
G00 =12g
00R00 12gikRik = 12(g00g00 gikgik) = .Conversely, (1.35) implies (1.34), since if G0i = g
00Ri0+g0kRik = 0,
then from (1.35) Ri0 = 0; and if G00 =12g
00R00 12gikRik = , thenfrom (1.35), 12g
00R00 32 = , that is, R00 = .The corresponding matter eld equations with an energy momen-
tum tensor T : G + g + T = 0 or, equivalently, R =12Tg T + g take the following form (in the case of dustlikematter with density , pressure p = 0, and also in a Gaussian normal
coordinate system):
R0i = 0
R00 = 12
Rik =(1
2 +
)gik.
(1.36)
Equation (1.36) is then equivalent to
G0i = 0
G00 =
}(constraint equations)
Rik =(1
2 +
)gik (evolution equations).
(1.37)
Gaussian normal coordinates can be introduced in general always,
and with x = (x0 = t, x1, x2, x3), the metric takes the form: ds2 =
dt2 + gik(x)dxidxk. So that g00 = g00 = 1, g0i = g0i = 0. The4-metric g thus determines a 3-metric gik, with its inverse gik,
which is positive denite on the hypersurfaces t = const. The eld
equations (1.34) are then explicitly, in a Gaussian normal coordinate
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
20 Space, Time and Matter
system ([6]):
G0i 12(gi,0); 12(g
kgk,0),i = 0, (1.38)
G00 12R +
18gmgm,0g
ikgik,0
18gigkmgik,0gm,0 = , (1.39)
Rik Rik 12gik,00 14gmgm,0gik,0 +
12gmgi,0gkm,0
=(1
2 +
)gik (1.40)
where
gik,0 gik/x0 = gik/t,gik,00 2gik/t2,Rik Ricci-tensor of the 3-metric gik,R Ricci-scalar of the 3-metric gik,
A; covariant derivative of A.We shall refer to Eqs. (1.38)(1.40) as the GaussEinstein eld equa-
tions.c
cSolutions of these equations have been considered by Marsden and Fischer
([7]) in an infinite dimensional setting. Our approach to the problem is, however,
entirely finite dimensional.
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
Space and Time 21
1.5. Relativistic Field Dynamics on 3-Manifolds
1.5.1. Three-dimensional field equations andrelationship with Einstein equations
As we have seen that in classical eld theory, the dynamics of a eld
is usually described by a set of evolution equations (together pos-
sibly with some constraint equations) in some space-time manifold.
We shall now demonstrate that in many cases the dynamics can also
be described in a purely 3-dimensional setting without the neces-
sity of explicitly introducing a time-coordinate. Instead of the time-
coordinate and time-dependent quantities, the temporal evolution is
provided by the ow of a 3-dimensional vector eld representing a
fundamental observer.
The starting point of our dynamical theory would be a set of dif-
ferential equations for a 3-metric g and a fundamental 3-vector eld
X on a M3. Let X = Xi(x)/xi be a vector eld on M3 (in some
local coordinate system (xi)), and h = LXg, the Lie-derivative of gwith respect to X.
So that
h = hikdxi dxk where hik = (LXg)ik .
Let
f,i, f;i partial and covariant derivative of f , respectively,Rik Ricci-tensor of the 3-metric gik,R Ricci-scalar of the 3-metric gik.
Then our basic equations are:
hi; (gkhk),i = 0, (1.41)
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
22 Space, Time and Matter
12R +
18(gmhm)(gikhik)
18(gihik)(gkmhm) = 0, (1.42)
Rik 12(LXh)ik 14(gmhm)hik
+12(gmhi)hkm = 0. (1.43)
We regard Eqs. (1.41)(1.43) as a set of dierential equations for
both the 3-metric g and the 3-dimensional vector eld X on M3.
Every solution (g,X) of (1.41)(1.43) determines uniquely a solution
of the vacuum Einstein eld equations as follows:
The integral curves of X determine a ow , which are local
1-parameter group of local dieomorphisms ([5]) of M3. Thus, for
each value of the ow parameter , denes a point transformation:
x x = (x) with 0(x) = x. The ow transformed metric g = (g) is given by gik(, x) = x
xixm
xk gm(x), and is thus -dependent.
Theorem 1. Let (gik;Xi) be a solution of (1.9)(1.11) andgik(, x) the -dependent metric transformed by the ow of X.
Then, the 4-dimensional metric,
ds2 = d2 + gik(, x)dxidxk (1.44)
is a solution of the vacuum Einstein equations in a Gaussian normal
coordinate system (x0 = , xi).
Proof. Let
K any tensor eld, (K) the ow transformed tensor eld K,LXK the Lie-derivative of K with respect to X.
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
Space and Time 23
Then one has ([5])
(LXK) =
( (K)). (1.45)
Therefore, by applying to Eqs. (1.41)(1.43), one can replace the
Lie-derivatives of gik in (1.41)(1.43) by the derivatives of gik(, xi)
with respect to , and obtain
(gkgik,0); (gkgk,0),i = 0, (1.46)
12R +
18gmgm,0g
ikgik,0 18 gigkmgik,0gm,0 = 0, (1.47)
Rik 12 gik,00 14gmgm,0gik,0 +
12gmgi,0gkm,0 = 0. (1.48)
(Here, Rik, R and the covariant derivatives are all relative to the
3-metric gik(, xi), and gik,0, gik,00 the rst and second partial deriva-
tives of gik with respect to .) These are precisely the vacuum
Einstein eld equations for the metric (1.44) in the Gaussian nor-
mal coordinate system (x0 = , xi) ([6]).
Note that the signature of the 4-metric (1.44) would depend on the
signature of the 3-metric gik. Thus, may or may not be the physical
time-coordinate.
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
24 Space, Time and Matter
1.6. Flat-Space Solutions
We now consider some special solutions of (1.41)(1.43). First, note
that if, either X = 0 or X is a Killing vector eld of g, then h =
LXg = 0. So that (1.41) is identically satised. Furthermore, if = = 0, then (1.43) implies that Rik = 0 which in turn implies that
Rijk = 0 (since M3 is 3-dimensional), that is, M3 is at with the
metric g = (i.e., gik = ik in some coordinate system). So, {X = 0,g = } and {X is Killing, g = } are both solutions of (1.41)(1.43)with = = 0. These are trivial at-space solutions.
For non-trivial at-space solutions, we put gik = gik = ik, so that
Rik = 0, R = 0, and
hik = X i,k + Xk,i,
(LXh)ik = hik,X + hiX,k + hkX,i,
hi; = hi,; (gkhk),i =
(
h
),i
, (1.49)
18(gihik)(gkmgm) =
18(h211 + h
222 + h
233)
+14(h212 + h
213 + h
223).
Therefore, (1.41)(1.43) becomes (together with (1.49)):
12(hi1,1 + hi2,2 + hi3,3)
12(h11,i + h22,i + h33,i) = 0, (1.50)
14(h11h22 + h11h33 + h22h33)
14(h212 + h
213 + h
223) = , (1.51)
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
Space and Time 25
12(hik,X + hiX,k + hkX
,i) +
12(hi1hk1 + hi2hk2 + hi3hk3)
14(h11 + h22 + h33)hik =
(1
2 +
)ik. (1.52)
In terms of Xi and its derivatives, Eqs. (1.50)(1.52) are
12(Xi, + X
,i), (X,),i = 0, (1.53)
12(Xi,i)
2 18(Xi,m + X
m,i )(X
m,i +X
i,m) = , (1.54)
12(Xi,k + X
k,i),X
12(Xi, + X
,i)X
,k
12(X,k + X
k,)X
,i
12X,(X
i,k + X
k,i) +
12(Xi,m + X
m,i )(X
m,k +X
k,m)
=(1
2 +
)ik. (1.55)
In spite of the complexity of the above equations, it is possible to
nd some non-trivial solutions as we shall see next.
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
26 Space, Time and Matter
1.7. The de Sitter Solution
As a non-trivial solution of (1.53)(1.55) or (1.49)(1.52), let = 0
and Xi = xi, where is a constant to be determined. Then by(1.49) (LXg)ik = hik = 2ik and
12(LXh)ik = 12[hix
,k hkx,i] = 22ik,
14(gmhm)hik = 14(
mhm)hik = 32ik,12(gmhi)hkm =
12(mhi)hkm = 22ik.
So (1.50) is identically satised and (1.51)(1.52) or (1.42)(1.43),
become
32 = , (1.56)
22ik + 22ik 32ik = ik. (1.57)
These equations are satised by taking =
3 . So that {g = ,
X =
3 x
i xi }, = 0 is a solution of (1.49)(1.52), that is, a
at-space solution of (1.41)(1.43).
The ow generated by the vector eld X = xi xi is simplyxi xi = etxi, so that xixk = etik and the transformed time-dependent 3-metric is therefore gik(x, t) = e2tik. The corresponding
4-metric (in coordinates (t, xi)):
ds2 = dt2 + gikdxidxk = dt2 + e2tikdxidxk
with =
3 is the well-known de Sitter solution of the Gauss
Einstein equations (with = 0).
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
Space and Time 27
1.8. A New Solution of the Vacuum EinsteinField Equations
As our next non-trivial solution of (1.49)(1.52), we make the ansatz:
= = 0 and X1 = 1x1, X2 = 2x2, X3 = 3x3, i.e., Xi =ixi (no summation), where is are constants to be determined.Then hik = (i + k)ik (no summation), i.e., h11 = 21, h22 =22, h33 = 23, hik = 0 if i = k.
Again, (1.50) is identically satised, and (1.51) implies
12 + 13 + 23 = 0. (1.58)
Now 12(LXh)ik = 12(hiX,k + hkX,i). If i = k, for example,i = 1, k = 2, 12(LXh)12 = 0, 12 (h11h21 + h12h22 + h13h23) = 0 and14(h11 + h22 + h33)h12 = 0. So that (1.52) is identically satisedif i = k. If i = k, for example, i = 1, k = 1, 12(LXh)11 = 221,12(h11h11 + h12h12 + h13h13) = 2
21 and 14(h11 + h22 + h33)h11 =
1(1 + 2 + 3). So that (1.52) implies1(1 + 2 + 3) = 02(1 + 2 + 3) = 03(1 + 2 + 3) = 0.
So, we would have a solution of the form X i = ixi (no summation)provided
1 + 2 + 3 = 0
and
12 + 13 + 23 = 0
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
28 Space, Time and Matter
which is equivalent to:
1 + 2 + 3 = 0
21 + 22 +
23 = 0
(1.59)which is possible only if, either all i are zero or some of the iare complex. Let us entertain the possibility that some of the i are
complex.
The ow generated by the vector eld X = (1x1 x1 +2x2 x2 +3x
3 x3
)is now xi xi = eitxi (no summation), so that xixk =
eitik (no summation) and the transformed time-dependent 3-metric
is gik(x, t) = e2itik (no summation). The corresponding 4-metric (in
coordinates (t, xi), after removing the tilde)
ds2 = dt2 + e21t(dx1)2 + e22t(dx2)2 + e23t(dx3)2 (1.60)
with (1, 2, 3) satisfying (1.59), is therefore a solution of the
vacuum GaussEinstein equations (with =0), that is, R =0.
This can also easily be checked by direct computation. That this,
the solution is non-trivial can be seen by computing, for exam-
ple, the 1212-component of the Riemann curvature tensor. We get
R1212 = 12e2(1+2)t = 0 (if 1, 2 = 0).The complex solution (1.53) is related to the so-called (real) Kas-
ner ([8]) solutions in the following manner. Recall that the Kasner
solutions for the vacuum Einstein equations R = 0 are given by:
ds2 = t21(dx1)2 + t22(dx2)2 + t23(dx3)2 t20dt2 (1.61)
where the satisfy the two equations
1 + 2 + 3 = 0 + 1
21 + 22 +
23 = (0 + 1)
2.
}(1.62)
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
Space and Time 29
If one takes 0 = 0, one gets the standard Kasner solution
ds2 = dt2 + t21(dx1)2 + t22(dx2)2 + t23(dx3)2 (1.63)
with
1 + 2 + 3 = 1
21 + 22 +
23 = 1.
}(1.64)
A special solution of (1.64) is 1 = 23 , 2 =23 , 3 = 13 . The other
obvious special solution 1 = 1, 2 = 3 = 0, i.e.,
ds2 = dt2 + t2(dx1)2 + (dx2)2 + (dx3)2 (1.65)
is actually a at metric as can be seen by the following transformation
of coordinates: t = t coshx1, x = t sinhx1, x2 = x2, x3 = x3, which
transforms (1.65) into
ds2 = dt 2 + (dx1)2 + (dx2)2 + (dx3)2.
If we replace t by t = et in (1.61), we get
ds2 = e21 t(dx1)2 + e22 t(dx2)2 + e23 t(dx3)2 e2(0+1)t(dt )2. (1.66)
If we now take 0 = 1 in (1.62) and (1.66), we get precisely(1.60), with t instead of t,
ds2 = e21 t(dx1)2 + e22 t(dx2)2 + e23 t(dx3)2 (dt )2
with
1 + 2 + 3 = 0
21 + 22 +
23 = 0.
}(1.59)
Consider now the complex solutions of (1.59). They are
1 =12(1 + i
3)3
2 =12(1 i
3)3
(1.67)
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
30 Space, Time and Matter
where 3 can be real or complex. We shall choose 3 = r, a real
number. Then 1 = p + iq, and its complex conjugate 2 = p iq,with p = r2 , q = r
3
2 .
So that the non-zero (complex) metric tensor components for
(1.60) are
g11 = e2(p+iq)t
g22 = e2(piq)t
g33 = e2rt
g00 = 1
with p = r2, q = r
32
.
(1.68)
We now make a complex coordinate transformation: x x toobtain a real metric as follows:
x1 = (a + ib)12 x1 + (c + id)
12 x2
x2 = (a ib) 12 x1 + (c id) 12 x2
x3 = x3
t = t.
(1.69)
Here a, b, c, d are any real numbers, such that det(x
x ) = 0. Thetransformation matrix for (1.69) and its determinant are
(x
x
)=
(a + ib)
12 (c + id)
12 0 0
(a ib) 12 (c id) 12 0 00 0 1 0
0 0 0 1
, (1.70)
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
Space and Time 31
det(x
x
)=[(a + ib)(c id)] 12 [(a ib)(c + id)] 12
= i2D
where
D = Im[(a + ib)12 (c id) 12 ]. (1.71)
So the condition on a, b, c, d is that D = 0.The transformed non-zero (real) metric tensor components are
then
g11 =xi
x1xk
x1gik
=(
x1
x1
)2g11 +
(x2
x1
)2g22
= (a + ib)e2(p+iq)t + (a ib)e2(piq)t
= 2e2pt(a cos 2qt b sin 2qt).
Similarly, g22 = 2e2pt(c cos 2qt d sin 2qt) and
g12 =xi
x1xk
x2gik =
x1
x1x1
x2g11 +
x2
x1x2
x2g22
= (a + ib)12 (c + id)
12 e2(p+iq)t
+(a ib) 12 (c id) 12 e2(piq)t
= 2e2pt(A cos 2qtB sin 2qt)
where
A = Re[(a + ib)12 (c + id)
12 ] (1.72)
B = Im[(a + ib)12 (c + id)
12 ] (1.73)
and
g33 = g33 = e2rt, g00 = g00 = 1.
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
32 Space, Time and Matter
We thus now have a real 4-metric g containing ve real param-
eters a, b, c, d, r (in coordinates (t, xi), after removing the bar):
g11 = 2ert(a cos(r
3t) b sin(r3t))
g22 = 2ert(c cos(r
3t) d sin(r3t))
g12 = 2ert(A cos(r
3t)B sin(r3t))
g33 = e2rt
g00 = 1.
(1.74)
This is a solution of the vacuum Einstein equations R = 0, as
can also be checked by direct computation.
The corresponding real vector eld X is then given by X =
Xi(x) xi , where X
i(x) = xi
xk Xk(x). From (1.70), we have
xi
xk=
(c id)1/2
i2D(c + id)
1/2
i2D0
(a ib)1/2
i2D(a + ib)1/2
i2D0
0 0 1
,
so that,
X1(x) =
(c id)1/2i2D
(1x1) (c + id)1/2
i2D(2x2)
= (p iq)(c id)1/2
i2D[(a + ib)1/2x1 + (c + id)1/2x2
]+
(p + iq)(c + id)1/2
i2D[(a ib)1/2x1 + (c id)1/2x2]
=Im[(p iq)(c + id)1/2(a ib)1/2]
Im[(a + ib)1/2(c id)1/2] x1 q(c
2 + d2)1/2
Dx2
= qx1 q(c2 + d2)1/2
Dx2. (1.75)
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
Space and Time 33
Similarly, X2(x) = q(a2+b2)1/2
D x1 + qx2 and X3(x) = rx3. After
removing the bar again, the real vector eld X is given by
X =
(qx1 q(c
2 + d2)1/2
Dx2
)
x1
+
(q(a2 + b2)1/2
Dx1 + qx2
)
x2 rx3
x3. (1.76)
We, thus, have a real solution (g,X) of (1.41)(1.43) with g =
and X given by (1.76).
A special case of (1.74) is given by a = 1, b = c = 0, d = 1 with
A = B = 1/2,D = 1/2,g11 = 2er cos(r
3)
g22 = 2er sin(r3)
g12 =2er
(cos(r
3) sin(r3))
g33 = e2r
g00 = 1.
(1.77)
This solution can also be checked by direct calculation.d
Notes on signature: The signature of the metric (1.77) is non-
Lorentzian and the vector eld X is real. A Lorentzian metric is
obtained by changing the sign of g33, i.e., g33 = e2r . This can beachieved by taking x3 = ix3 in (1.69). However, then, the correspond-
ing vector eld has to be complex. In both cases, the corresponding
3-metrics must have indenite signatures.
dThis was done on MAPLE, the symbolic computation language. See
Appendix C.
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
34 Space, Time and Matter
1.9. A Solution of Mathematical Interest
We present now another solution of our equations which is mathe-
matically interesting but turns out to be physically trivial. However,
this solution illustrates the basic principle of our approach.
Again, suppose = = 0 and now we assume
hik = ikf (1.78)
where i are constants and f is a function of (xi) to be determined.
Then (1.51) is identically satised, whereas (1.50) implies
12(i1f,1 + i2f,2 + i3f,3) 12(
21 +
22 +
23)f,i = 0. (1.79)
We can satisfy (1.71) by assuming
if,k = kf,i (1.80)
which implies that
f(x) = F (u), where u = ixi (1.81)
and F is a function of the single variable u. Now, consider Eq. (1.52)
in view of (1.78) and (1.80). It becomes
12(LXh)ik = 14ikf
2j
2j
or
ikf,X + ifX,k + kfX
,i =
12ikf
2j
2j . (1.82)
Let us put
Xi =12iG(u). (1.83)
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
Space and Time 35
Then
hik = Xi,k + Xk,i = ikG
,
so that
F = G (1.84)
and (1.82) becomes
(F G + 2FG)j
j = F 2j
2j ,
or[GG+ 2(G)2
]j
j = (G)2j
2j .(1.85)
We can satisfy (1.85) by taking, say,j
j =j
2j = 1 (1.86)
and by making sure that G satises the following equation
GG + (G)2 = 0, (1.87)
the general solution of which is: G(u) = (au + b) 12 , where a, b arearbitrary constants.
A solution of (1.49)(1.52) is thus provided by (g,X) where g =
and X = Xi(x)/xi with Xi(x) = 12iu1/2, where i satisfy (1.86)
and u = ixi. Next, we shall derive the corresponding 4-metric solu-
tion of the vacuum eld equations (1.38)(1.40) by transforming the
3-metric g = by the ow of X. The ow of X is given by the
solution of the dynamical system:
dxi
dt=
12iu
12 = i(1x1 + 2x2 + 3x3)
12
xi(0) = xi0
(1.88)
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
36 Space, Time and Matter
together with (1.86). To solve (1.88), note that, according to (1.86)
and (1.88) dudt = idxi
dt =12(i
2i )u
12 = 12u
12 and u(0) = u0 = ixi0.
This implies that u12 = t4 + u
120 , and so
x2(t) =
122
(t
4+ u
120
)dt = x20 +
122u
120 t +
116
2t2,
x3(t) =
123
(t
4+ u
120
)dt = x30 +
123u
120 t +
116
3t2,
x1(t) =u
1 2
1x2(t) 3
1x3(t)
=11
(t
4+ u
120
)2 2
1
(x20 +
122u
120 t +
116
2t2
) 3
1
(x30 +
123u
120 t +
116
3t2
)= x10 +
121u
120 +
116
1t2.
Under the ow of X, the point (xi0) is mapped into the
point (xi(t)). The ow generated by X is thus t : xi xi = xi+
12iu
12 t + 116it
2. The inverse transformation is: xi xi = xi 12i(u
1/2 t4)t 116it2, where u = ixi. In view of (1.86), the Jacobianmatrix of the transformation turns out to be
xi
xk= ik ik; = t/4(ixi) 12
with det(
xi
xk
)= 1 w.
(1.89)
Under the ow of X, the at 3-metric gik(x) = ik is transformed
into gik(x, t), according to:
gik(x, t) =x
xixm
xkm.
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
Space and Time 37
Thus,
gik(x, t) = ik + ik2 2ik. (1.90)
And, we have a 4-metric solution g of the vacuum eld equations
R = 0 (in coordinates (t, xi), after removing the tilde in (1.90))
g =
(ik + ik2 2ik 0
0 1
)
= t/4(ixi)12 ;
i
i =
i
2i = 1
(1.91)
with det(g) = (1 )2. However, not only R = 0, but alsoR = 0. The solution thus represents a at space-time!
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
38 Space, Time and Matter
1.10. The Schwarzschild Solution
As the rst example of a non-trivial at 3-space solution of (1.41)
(1.43), consider a general spherically symmetric 3-metric g and a
spherically symmetric vector eld X in coordinates (x1 = , x2 =
, x3 = ):
g = gikdxidxk = f()2d2 + 2(d2 + sin2 d2),
X = Xi
xi= a()
.
(1.92)
Substituting (1.92) in (1.41)(1.43), we obtain the following
ordinary dierential equations for the unknown functions f()
and a()
4
(d
df()
)a()
f()= 0, (1.93)
42(
d
df()
) + (f())3 f() + 2
(d
df()
)(a())2(f())2
2(f())3
+42 (f())3
(d
da()
)a() + (a())2(f())3
2(f())3= 0, (1.94)
(2
d
df() + (f())2
(d2
d2f()
)(a())2
+3(
d
df()
)a() (f())2
(d
da()
)
) (f())1
a()(f())3
(d2
d2a()
) + (f())3
(d
da()
)2
f()
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
Space and Time 39
2(
d
df()
)(a())2(f())2 + 2(f())3
(d
da()
)a()
f()= 0,
(1.95)
(d
df()
) + (f())3 f() + (a())2(f())3
(f())3
2 (f())3
(d
da()
)a() +
(d
df()
)(a())2(f())2
(f())3= 0.
(1.96)
[Eq. (1.96)] (sin())2. (1.97)
Equation (1.93) implies that, for non-trivial a(), f() = const.
(which we set equal to one). Equation (1.94) then becomes
2(
d
da()
)a()
+a()2
2= 0 (1.98)
whose solution is a() = k1/2 , k = const. The remaining Eqs. (1.95)
(1.97) are then identically satised.
Thus the at 3-metric
g = d2 + 2(d2 + sin2 d2)
and the vector eld X = k12
(1.99)is the only solution of Eqs. (1.93)(1.97).
1.10.1. The Schwarzschild metric
We now wish to derive the space-time 4-metric which corresponds
to our solution (1.99), according to Theorem 1, by transforming the
3-metric g by the ow of the vector eld X. In order to calculate
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
40 Space, Time and Matter
the ow of X, it is convenient to transform X rst to a simpler form
by a simple change of coordinates. For example, the transformation
(, , ) (R, , ), where = (32kR)23 transforms the metric as
well as the vector eld in (1.99), into
g = k2(32kR
) 23
dR2 +(32kR
) 43
(d2 + sin2 d2)
X =
R.
(1.100)
It is clear from the invariant form of Eqs. (1.41)(1.43), that a
change of coordinates provides another solution of the metric and
the vector eld. It can also be checked directly that (1.100) is indeed
a solution of (1.41)(1.43). And, of course, the 3-metric in (1.100) is
still at.
The ow of X is given by: : R R = R + , = , = . Therefore, according to Theorem 1, (1.25) corresponds tothe space-time 4-metric (in Gaussian normal coordinates (, R, , )):
ds2 = d2 + k2[32k(R )
] 23
dR2
+[32k(R )
] 43
(d2 + sin2 d2). (1.101)
That (1.101) is a solution of the vacuum Einstein eld equations
can also be checked directly. In fact, (1.101) is the Schwarzschild
solution in the so-called Lemaitre coordinates ([911]) if we take
k = (2m)12 .
This can be seen by considering the following transformation:
(, R, , ) (t, r, , ),
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
Space and Time 41
where
= 2( r2m
) 12 + 2m log
r 2mr +
2m
t = (r, t)R =
23r32 (2m)
12 + = R(r, t)
= , =
(1.102)
with the inverse transformation:
(t, r, , ) (, R, , ),
where
r = (2m)13
[32(R )
] 23
= r(R, )
t = 2( r2m
) 12 + 2m log
r 2mr +
2m
= t(R, ) = , = .
(1.103)
(1.102) or (1.103) transforms (1.101) into the Schwarzschild metric:
ds2 = (1 2m
r
)dt2 +
(1 2m
r
)1dr2 + r2(d2 + sin2 d2).
(1.104)
Incidentally, we have thus proved a version of Birkos Theorem in
our formalism, namely, that the only spherically symmetric solution
of (1.41)(1.43) leads to the Schwarzschild space-time.
One might think that if (X, g) is a solution of (1.41)(1.43), then
adding a Killing vector eld of g to X would again give us another
solution. In fact, one has the following proposition.
Proposition 1. Let (X, g) be a solution and X0 a Killing vector
eld of g. Then, (X +X0, g) is also a solution provided [X,X0] = 0.
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
42 Space, Time and Matter
Proof. Let X = X + X0. Then, LXg = LXg and LX(LXg) =LX(LXg)+LX0(LXg). Now, [LX0,LX ] = L[X0,X]. Hence, [X0,X] = 0implies that LX0(LXg) = LX(LX0g) = 0, and therefore, we have alsoLX(LXg) = LX(LXg).
However, this does not generate a new space-time 4-metric in view
of the following Proposition.
Proposition 2. (X, g) and (X +X0, g), where [X,X0] = 0, gener-
ate equivalent space-time 4-metrics.
Proof. Since X commutes with X0, the ow of X + X0 is a com-
position (as maps) of the ow of X and the ow of X0. Since X0 is
Killing, the ow of X0 has no eect on the metric g. Therefore, the
eect of the ow of X +X0 on g is the same as that of X.
For example, consider the solution (1.99) together with the two
Killing vectors X0 and X1 of the 3-metric:
g: ds2 = d2 + 2(d2 + sin2 d2),
X = (2m/)12
,
X0 = sin
+ cot cos
,
X1 = sin cos
+ (cos cos/)
(cosec sin/)
.
Here, the Killing vector eld X0 corresponds to rotational isometry,
whereas X1 corresponds to translational isometry. [X,X0] = 0, but
[X,X1] = 0. (X + X0, g) is again a solution, but (X + X1, g) is not.The 4-metrics corresponding to (X, g) and (X+X0, g) are equivalent.
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
Space and Time 43
Our denition of a physical observer was that of a vector eld X on
(M3, g) such that 0 < g(X,X) < 1. Note that the vector eld X =
[2m ]12
in (1.99), therefore, ceases to be physical when 2m or
when R 43m, even though X has a mathematical singularity only at = 0. = 2m is, of course, the Schwarzschild horizon corresponding
to r = 2m in Schwarzschild coordinates (t, r, , ).
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
44 Space, Time and Matter
1.11. From Space-Time 4-Metric to 3-Metricand 3-Vector Field
There exists a converse to Theorem 1.
Theorem 2. Suppose a space-time 4-metric solution of the vacuum
Einstein eld equations has the form
ds2 = d2 + gik(, xi)dxidxk
in the Gaussian normal coordinates (, xi), where gik(, xi) is the
transformed -dependent metric by the ow of some 3-vector eld
X acting on a 3-metric gik(xi). Then (gik,X) is a solution of our
Eqs. (1.41)(1.43).Proof. By assumption, g = (g) satises Eqs. (1.46)(1.48).
Again, from (1.45), one has
(LXK)|=0 =
( (K))=0
.
Since 0 = Identity, (1.46)(1.48) implies (1.41)(1.43).
Corollary. In particular, if a space-time 4-metric solution of the
vacuum Einstein equations has the form:
ds2 = d2 + gik(x1 , x2, x3)dxidxk (1.105)
in the Gaussian normal coordinates (, xi), then (gik = gik(x1, x2,
x3),X = 1 x1 ) is a solution of (1.41)(1.43).
This is because the ow of X = 1 x1 is simply : x1 x1 =
x1 + , x2 x2 = x2, x3 x3 = x3.As an example of Theorem 2, we shall now illustrate how to obtain
the equivalent 3-metric g and 3-vector eld X from a space-time 4-
metric which satises the vacuum Einstein eld equations and which
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
Space and Time 45
has the specic form given by (1.105). The rst example will be the
Schwarzschild black hole metric and the procedure below will be help-
ful when we consider next the Kerr black hole in our 3-dimensional
formalism.
Consider the Schwarzschild metric gx in coordinates x = [x1, x2,
x3, x4]
gx =
x12
x12 2mx1 0 0 00 x12 0 0
0 0 x12 sin2 x2 0
0 0 0 1 + 2mx1
(1.106)
and, in anticipation of (1.103), make a coordinate transformation of
the form
[x1, x2, x3, x4] [y1, y2, y3, y4]x1 = F (y1 y4), x2 = y2,
x3 = y3, x4 = H(F (y1 y4)) y4,(1.107)
where the functions F and H would be determined by imposing suit-
able conditions in order to bring the metric into the appropriate form.
Then, the transformed 4-metric components gy (, = 1, 2, 3, 4)
are (see Appendix D)
gy11 = (D(F )(y1 y4))2[(F (y1 y4))2
+(D(H)(F (y1 y4)))2(F (y1 y4))2
4(D(H)(F (y1 y4)))2mF (y1 y4)
+ 4(D(H)(F (y1 y4)))2m2]/F (y1 y4)(F (y1 y4) 2m),(1.108)
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
46 Space, Time and Matter
gy12 = 0, (1.109)
gy13 = 0, (1.110)
gy22 = F (y1 y4)2, (1.111)gy23 = 0, (1.112)
gy33 = (F (y1 y4))2(sin(y2))2, (1.113)gy14 = D(F )(y1 y4)[D(F )(y1 y4)(F (y1 y4))2
+(D(H)(F (y1 y4)))2D(F )(y1 y4)(F (y1 y4))2
4 (D(H)(F (y1 y4 )))2D(F )(y1 y4)mF (y1 y4)+ 4 (D(H)(F (y1 y4)))2D(F )(y1 y4)m2
+D(H)(F (y1 y4))(F (y1 y4))2
4D(H)(F (y1 y4))mF (y1 y4)+ 4D(H)(F (y1 y4))m2]/F (y1 y4)(F (y1 y4) 2m),
(1.114)
gy24 = 0, (1.115)
gy34 = 0, (1.116)
gy44 = [(D(F )(y1 y4))2(F (y1 y4))2
+(D(H)(F (y1 y4)))2(D(F )(y1 y4))2(F (y1 y4))2
4(D(H)(F (y1 y4)))2(D(F )(y1 y4))2mF (y1 y4)+ 4(D(H)(F (y1 y4)))2(D(F )(y1 y4))2m2
+2D(H)(F (y1 y4))D(F )(y1 y4)(F (y1 y4))2
8D(H)(F (y1 y4))D(F )(y1 y4)mF (y1 y4)+ 8D(H)(F (y1 y4))D(F )(y1 y4)m2 + (F (y1 y4))2
4mF (y1 y4) + 4m2]/F (y1 y4)(F (y1 y4) 2m),(1.117)
where D is the derivative operator.
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
Space and Time 47
To reduce it to the appropriate Gaussian normal form we need to
impose the conditions
gy14 = 0, (1.118)
gy44 = 1. (1.119)
One can solve (1.118)(1.119) for the derivatives D(F ), D(H) in
terms of F (y1 y4) to obtain
D(F )(y1 y4) =2
m
F (y1 y4) , (1.120)
D(H)(F (y1 y4)) =2
m
F (y1 y4)(1 2 m
F (y1 y4))1
.
(1.121)
Substituting (1.120)(1.121) back into (1.108)(1.117) we obtain the
desired Gaussian normal form for the Schwarzschild metric (provided
F and H satisfy (1.120)(1.121))
gy14 = gy24 = gy34 = 0, gy44 = 1gy11 = 2
m
F (y1 y4)gy12 = gy13 = gy23 = 0
gy22 = (F (y1 y4))2
gy33 = (F (y1 y4))2(sin(y2))2.
(1.122)
According to the Corollary of Theorem 2 the above 4-metric is
generated by the 3-metric gz in coordinates z = [z1, z2, z3] (one
simply replaces y1 y4 by z1, y2 by z2 and y3 by z3)
gz =
2m/F (z1) 0 0
0 F (z1)2 0
0 0 F (z1)2(sin(z2))2
(1.123)
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
48 Space, Time and Matter
and the 3-vector eld
Xz = 1
z1. (1.124)
One can now easily solve (1.120) to nd that F (z1) = ((3/2)
(2m)1/2z1)2/3 (apart from a constant), and thus recover essentially
(1.100).
However, we do not need to solve (1.120) explicitly for F (z1) in
order to obtain (1.99). We simply make a sort of inverse transforma-
tion of the coordinates [z1, z2, z3] back to the original coordinates.
In other words, a transformation
[z1, z2, z3] [w1, w2, w3]z1 = F1(w1), z2 = w2, z3 = w3
}(1.125)
where z1 = F1(w1) is the inverse function of w1 = F (z1), so
that F (F1(w1)) = w1 and dF1
dw1 = 1/(dFdz1 ) = 1/(2m/w1)
1/2 from
(1.120). Under (1.125), (1.123)(1.124) is, therefore, transformed into
gw11 = 2m(
dF1
dw1
)2 /F (F1(w1)) = 1
gw12 = gw13 = gw23 = 0
gw22 = F (F1(w1))2 = w12
gw33 = F (F1(w1))2(sin(w2))2 = w12(sin(w2))2,
(1.126)
Xw =(1/(
dF1
dw1
))
w1= (2m/w1)1/2
w1(1.127)
which is essentially (1.99).
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
Space and Time 49
1.12. The Kerr Solution
As for the second example of Theorem 2, we shall now demonstrate
that the axially symmetric stationary Kerr solution [12] of the vac-
uum Einstein equations can also be brought into the appropriate
Gaussian normal form by a procedure similar to the one outlined in
the previous section, and thus can also be formulated in terms of
a 3-metric and a 3-vector eld on a 3-manifold. Consider the Kerr
metric in BoyerLindquist coordinates [x1, x2, x3, x4]
gx11 =x12 + a2(cos(x2))2
x12 2mx1 + a2 ,
gx12 = gx13 = gx23 = gx14 = gx24 = 0,
gx22 = x12 + a2(cos(x2))2,
gx33 =((x12 + a2)2 (x12 2mx1 + a2)a2(sin(x2))2)(sin(x2))2
x12 + a2(cos(x2))2,
gx34 = 2 amx1 (sin(x2))2
x12 + a2(cos(x2))2,
gx44 = 1 + 2 mx1x12 + a2(cos(x2))2
,
(1.128)
and make a coordinate transformation of the form:
[x1, x2, x3, x4] [y1, y2, y3, y4],x1 = F (y1 y4, y2), x2 = y2, x3 = K(y1 y4, y2) + y3,x4 = H(F (y1 y4, y2)) y4.
(1.129)
We choose the unknown function F,K,H in such a way that the
following conditions are satised for the transformed metric compo-
nents. (The expressions for the transformed metric components are
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
50 Space, Time and Matter
too long to reproduce here. See Appendix D.)
gy14 = gy24 = gy34 = 0, gy44 = 1. (1.130)
These equations can now be solved for the following derivatives of
F,K,H, in terms of F (y1 y4) to give
D[1](F )(y1 y4, y2)
=2m
F (y1 y4, y2)a2 + (F (y1 y4, y2))3(F (y1 y4, y2))2 + a2(cos(y2))2 , (1.131)
D[1](K)(y1 y4, y2)
= 2amF (y1 y4, y2)/[((F (y1 y4, y2))2 + a2(cos(y2))2)
((F (y1 y4, y2))2 2mF (y1 y4, y2) + a2)], (1.132)
D(H)(F (y1 y4, y2))
=2m
F (y1 y4, y2)a2 + (F (y1 y4, y2))3(F (y1 y4, y2))2 2mF (y1 y4, y2) + a2 . (1.133)
Here D[1](F ), D[1](K) are the rst partial derivatives of F,K with
respect to the rst argument y1 y4, respectively. Note that thereare no conditions on the rst partial derivatives of F,K with respect
to the second argument y2.
The above conditions (1.131)(1.133) are not only necessary but
also sucient for (1.130) to hold. In principle, therefore, one can solve
(1.131) for F (y1 y4, y2) and, then, (1.132)(1.133) to determineK(y1 y4, y2), H(F (y1 y4, y2)). Unfortunately, (1.131) cannot beintegrated explicitly as it involves elliptic integrals.
Let us assume the above conditions (1.131)(1.133). The compo-
nents of gy then have the appropriate Gaussian normal form given by
(1.105). They depend on functions of (y1y4, y2) and y2 only. There-fore, according to the Corollary of Theorem 2, gy is generated by the
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
Space and Time 51
3-metric gz in coordinates [z1, z2, z3] (where one replaces y1 y4 byz1, y2 by z2 and y3 by z3 in gyik (i, k = 1, 2, 3) to get gzik) and the
3-vector eld
Xz = 1
z1(1.134)
gzik (i, k = 1, 2, 3) contain only F (z1, z2), K(z1, z2) and their rst
partial derivatives.
To obtain an explicit form of the 3-metric we make an inverse
transformation of the coordinates [z1, z2, z3] (analogous to the
Schwarzschild case) back to the original coordinates as follows:
[z1, z2, z3] [w1, w2, w3]z1 = F1(w1, w2), z2 = w2,
z3 = w3K(F1(w1, w2), w2)(1.135)
where F1 is the inverse function of w1 = F (z1, z2) with respect to
the rst variable, that is, F (F1(w1, w2), w2) = w1.
Under (1.135) the vector eld Xz is transformed into
Xw =[1/
F1
w1(w1, w2)
]
w1
+ [D[1](K)(F1(w1, w2), w2)]
w3. (1.136)
The transformed components gwik contain F (F1(w1, w2), w2)
and only the following derivatives:
F1
w1(w1, w2),
F1
w2(w1, w2),
D[1](K)(F1(w1, w2), w2),
D[1](F )(F1(w1, w2), w2),
D[2](F )(F1(w1, w2), w2).
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
52 Space, Time and Matter
These can be expressed explicitly in terms of w1, w2 as follows. From
(1.131) and (1.132), we have
F (F1(w1, w2), w2) = w1, (1.137)
D[1](K)(F1(w1, w2), w2)
=K
z1= 2
amw1(w12 + a2(cos(w2))2)(w12 + a2 2mw1) ,
(1.138)
F1
w1(w1, w2)
= 1/
F
z1= 1/2
(w12 + a2(cos(w2))2)2
m
w1(w12 + a2). (1.139)
Now, from (1.131), by integrating once
w1z1
=2m
w1(w12 + a2)w12 + a2(cos(w2))2
,
we have
z1 =1
(2m)1/2
w12 + a2(cos(w2))2
w1(w12 + a2)dw1,
apart from a function of w2, which we set equal to zero. Therefore,
z1w2
= 2a2 sin(w2) cos(w2)
(2m)1/2
dw1
w1(w12 + a2)
or
F1
w2(w1, w2) = 2a
2 sin(w2) cos(w2)(2m)1/2
EL(w1), (1.140)
whereEL(w1) is a standard elliptic integral. And, again from (1.131),
D[1](F )(F1(w1, w2), w2)
=F
z1=2m
w1(w12 + a2)w12 + a2(cos(w2))2
. (1.141)
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
Space and Time 53
To calculate D[2](F )(F1(w1, w2), w2) = Fz2 , we dierentiate both
sides of (1.137) with respect to w2, to get
0 =w1w2
=F
z1F1
w2(w1, w2) +
F
z2z2w2
=
[2m
w1(w12 + a2)w12 + a2(cos(w2))2
]
[a
2 sin(w2) cos(w2)EL(w1)2
m
]+
F
z2 1
so that
D[2](F )(F1(w1, w2), w2)
= 2a2 sin(w2) cos(w2)EL(w1)
w1(w12 + a2)
w12 + a2(cos(w2))2. (1.142)
Finally, the explicit expressions for the 3-metric gw and the
3-vector eld Xw, which generate the Kerr solution, are:
gw11 =w12 + a2(cos(w2))2
w12 + a2 2mw1
2mw1(w12 + a2)(w12 + a2(cos(w2))2 2mw1)
(w12 + a2(cos(w2))2)(w12 + a2 2mw1)2(1.143)
gw12 = gw23 = 0, (1.144)
gw13 = 2 a(sin(w2))22m3/2w1
w13 + a2w1
(w12 + a2(cos(w2))2)(w12 + a2 2mw1) , (1.145)
gw22 = w12 + a2(cos(w2))2, (1.146)
gw33 =(w12 + a2 2mw1 + 2 mw1(w1
2 + a2)w12 + a2(cos(w2))2
)(sin(w2))2,
(1.147)
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
54 Space, Time and Matter
Xw =
[2mw13 + a2w1
w12 + a2(cos(w2))2
]
w1
+[2
amw1(w12 + a2(cos(w2))2)(w12 + a2 2mw1)
]
w3.
(1.148)
One can now verify (Appendix E) that the above (g,X) is indeed
a solution of our Eqs. (1.41)(1.43).
Note that, (1.143)(1.145) reduce to the Schwarzschild case
(1.126)(1.127) when a 0. When m 0, we obtain
gw =
w12 + a2(cos(w2))2
w12 + a20 0
0 w12 + a2 0
(cos(w2))20 0 (w12 + a2)
(sin(w2))2
,
(1.149)
Xw = 0. (1.150)
The above 3-metric (1.149) is at, and therefore (1.149)(1.150) is
equivalent to the at space-time.
If a = 0, the 3-metric (1.143)(1.145) is not at, in contrast to(1.126), the 3-metric in the Schwarzschild case. Its scalar curvature
is given by
Rw =2a2mw1(3(cos(w2))2 1)
w12 + a2(cos(w2))2(1.151)
and the 3-vector eld Xw in (1.148) has the length
g(X,X) = gw(Xw,Xw) =2mw1
w12 + a2(cos(w2))2. (1.152)
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
Space and Time 55
Thus X ceases to be physical when g(X,X) = 1, i.e.,
2mw1w12 + a2(cos(w2))2
= 1 (1.153)
which corresponds to the so-called stationary limit of the Kerr
black hole.
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
56 Space, Time and Matter
1.13. The Maxwell Equations
Our basic approach to evolution in a 3-dimensional manifold can also
be applied to the Maxwell equations.
We rst write the Maxwell equations on a space-time
curlE = Bt
curlH = 4j+Dt
divD = 4
divB = 0
(1.154)
for the electromagnetic variables E,H,B,D, j, (the usual elec-
tric, magnetic, elds and inductions, current and charges) in terms
of 3-dimensional time-dependent dierential forms on a 3-manifold
(M3, g) as follows. Dene the following (in some local coordinates
(xi) of (M3, g)):
1-forms, , 1(M3):
=
i
idxi, E = (1, 2, 3),
=
i
idxi, H = (1, 2, 3).
2-forms, , , 2(M3):
=i
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
Space and Time 57
3-form: 3(M3):
= dx1 dx2 dx3.
Note that these 3-dimensional forms are time-dependent with the
time t as a parameter. Using the exterior derivative operator d the
Maxwell equations (1.154) can then be written as:
d = t
d = 4 +
t
d = 4
d = 0.
(1.155)
As we did in the case of GaussEinstein equations, let us now
replace the time-derivative t in (1.155) by the Lie-derivative LXwith respect to some vector eld X on M3. Then (1.155) becomes
d = LX, (1.156)d = 4 + LX, (1.157)d = 4, (1.158)
d = 0. (1.159)
Here, , , , , , are to be regarded now as (time-independent)
purely 3-dimensional dierential forms on M3. Equations (1.156)
(1.159) are thus analogous to (1.38)(1.40), corresponding to the
GaussEinstein equations, and are to be considered as a set of equa-
tions for the forms , , , , , and the vector eld X. Every solu-
tion of (1.156)(1.159) determines uniquely a solution of (1.155) and,
thus, of (1.154).e One simply has to transform the time-independent
eAgain, the converse need not be true.
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
58 Space, Time and Matter
dierential form , for example, by the ow t of X to obtain a time-
dependent form: = t(), etc. The vector eld X thus provides the
temporal evolution of the electromagnetic elds.
The above equations take a surprisingly simple (and elegant) form
if one uses the relationship between the Lie derivative LX and theexterior derivative d and the contraction operator iX relative to X:
LX = iX d + d iX (1.160)
to obtain from (1.156) and (1.159)
d = (iX d + d iX) = d(iX)
or
d = 0 where = + iX. (1.161)
Similarly, from (1.87) and (1.88)
d = 4 + (iX d + d iX) = 4( + iX) + d(iX)
or
d = 4 where = iX, = + iX. (1.162)
Thus, Eqs. (1.156)(1.159) become
d = 0
d = 4
}(1.163)
where
= + iX
= iX = + iX.
(1.164)
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
Space and Time 59
In vacuum, = = = 0; thus, the corresponding vacuum
equations are
d = 0
d = 0.
}(1.165)
The continuity equation follows by taking the exterior derivative
of the second equation in (1.163), i.e., d = 0. From (1.164)
0 = d( + iX) = d + (LX iX d).
Since is a 3-form on a 3-manifold, d = 0. Thus
LX + d = 0 (1.166)
which is the continuity equation relative to the vector eld (observer)
X if we keep in mind that the Lie-derivative LX corresponds to thetime-derivative t and d is the divergence of the current j.
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch01
60 Space, Time and Matter
References
[1] G. D. Mostow, Strong Rigidity of Locally Symmetric Spaces, Ann.
Math. Series (Princeton Univ. Press, Princeton, 1973).
[2] S. Helgason, Dierential Geometry and Symmetric Spaces (Academic,
New York, 1972).
[3] S. Kobayashi and K. Nomizu, Foundations of Dierential Geometry,
Vol. 1 (Interscience, New York, 1983).
[4] D. V. Widder, The Heat Equation (Academic Press, New York, 1975).
[5] A. Lichnerowicz, Theories relativistes de la gravitation et de
lelectromagnetisme (Masson, Paris, 1955).
[6] J. S. Synge, Relativity, the General Theory (North-Holland, Amster-
dam, 1960).
[7] A. E. Fisher and J. E. Marsden, J. Math. Phys. 13 (1972) 546568.
[8] E. Kasner, American J. Math. 43 (1921) 217.
[9] G. Lemaitre, Lunivers en expansion, Ann. Soc. Sci. Bruxelles I A 53
(1933) 51.
[10] A. Z. Petrov, Einstein Spaces (Pergamon Press, Oxford, 1969).
[11] D. K. Sen, Class. Quant. Gravity 12 (1995) 553577.
[12] D. K. Sen, J. Math. Phys. 41 (2000) 75567572.
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch02
2. MATTER
2.0. Introduction
The basic constituents of all matter, namely, the elementary particles,
have the following salient features.
The massive ones, that is, those with non-zero rest mass, can be
classied into two basic categories: the lighter leptons and the heav-
ier baryons.. In each category, there is a stable particle, the lepton
electron and the baryon proton. The unstable leptons and baryons all
spontaneously decay eventually into the stable ones.
Then, there are the so-called zero rest mass particles, the photon
and the neutrino, which are essentially carriers of energy and vehicles
of interaction between the particles.
The so-called standard model has had considerable success as a
unied theory of all elementary particles. Together with the Higgs
boson and Higgs mechanism, we are able to explain the existence
and masses of several new bosons.
Nevertheless, it cannot be considered as a complete and fully sat-
isfactory theory of all elementary particles. It cannot, for example,
explain intrinsically why the proton is about 1836 times heavier than
the electron.
61
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch02
62 Space, Time and Matter
In the standard model, the phenomenon of neutrino oscillation
([1]) requires that neutrinos have non-zero mass.
In 1957, Heisenberg ([2, 3]) tried to formulate (without much
success) a unied theory of all elementary particles starting from a
nonlinear 4-component spinor equation with a built-in fundamental
constant.
Here, we ([4, 5]) suggest that massless 2-component Weyl neutri-
nos, instead of 4-component spinors, are probably more fundamental
than previously thought. We consider a composite system consist-
ing of a massless positively oriented 2-component Weyl neutrino and
a massless negatively oriented 2-component Weyl neutrino with a
certain specic symmetry-breaking interaction between the two.
We assume that the observable physical particles manifest as
energy states of the resulting 4-component system. A simple quan-
tum mechanical treatment shows that such a model should exhibit
2-fold branching and energy defects, which could then be interpreted
as formation of particles of non-zero rest mass.
Such a model can also provide a qualitative, alternative non-
standard explanation of the dierent avors of amassless 4-component
neutrino and thus, of neutrino oscillation without assuming a neutrino
mass.
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch02
Matter 63
2.1. The Photon and the Weyl Neutrinos
A long time ago, de Broglie ([6]) and Jordan ([7]) have tried to
construct a neutrino theory of light based on the formers idea of
fusion. Both these authors tried to explain the photon as a com-
bination of two 4-component neutrinos of essentially the same kind.
Pryce ([8]) and Barbour et al. ([9]) however showed that such a pho-
ton would be longitudinally polarized. For a historical review of the
neutrino theory of light, see [10].
We rst show that the photon is intimately connected not to the
4-component Dirac neutrino, but to the 2-component positively and
negatively oriented Weyl neutrinos.
The Maxwell equations in vacuum can be written as ([11, 12])
df = 0, d f = 0 (2.1)
where f is the electromagnetic dierential 2-form, d is the exterior
derivative operator and f is the Hodge dual of f . We shall use space-time coordinates x(= x, y, z, ict), = 1, 2, 3, 4, so that the metric
tensor in a Minkowski space-time is and consider only proper
Lorentz transformations. Then, in components, the Hodge dual of f
is given by
f = 12f (2.2)
where is the completely anti-symmetric LeviCivita symbol.
In general, for a p-form w on an n-dimensional pseudo-Riemannian
manifold, one has
w = (1)p(np)w (2.3)
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch02
64 Space, Time and Matter
so that f = f for the electromagnetic 2-form f on the 4-dimensionalMinkowski space-time. It is thus possible to decompose f in an invari-
ant manner into a self-dual part f s and an anti-self-dual part fa, as
follows
f = f s + fa, f s =12(f + f), fa = 1
2(f f) (2.4)
where f s = f s, fa = fa. The Maxwell equations (2.1) are thenequivalent to
f = f s + fa, df s = 0, dfa = 0. (2.5)
For an anti-symmetric tensor, the self-duality condition (in compo-
nents) f s = f s implies that
f s12 = fs34, f
s23 = f
s14, f
s31 = f
s24 (2.6)
so that it has only three independent components. Now, a self-dual
(or an anti-self-dual) anti-symmetric tensor cannot by itself describe
a physical electromagnetic eld, because in that case there should
exists a 4-vector A = (A, i), such that f s = AA. No such4-vector with three real components and one imaginary component
can exist if f s were to be self-dual, because (2.6) would then imply
that
yAx xAy = (1/ic)tAz iz. (2.7)
Consider, however, a 2-component quantity =(
12
)and a 4-vector
s =
1
i1
2i2
(2.8)
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch02
Matter 65
and set f s = s s. Then, the self-duality condition (2.6)
gives basically two equations
f s12 = 21 i11 = 42 i32 = f s34, (2.9)
f s23 = i31 + 22 = 41 i12 = f s14. (2.10)
The third condition in (2.6) gives again (2.10). Multiplying (2.9) by
i and (2.10) by i, we get two equations for (1, 2):
11 + i21 32 + i42 = 012 i22 + 31 + i41 = 0
(2.11)which by introducing the Pauli spin matrices
1 =
(0 1
1 0
), 2 =
(0 ii 0
), 3 =
(1 0
0 1
)(2.12)
can be written as a single equation (k = 1, 2, 3)
(kk + i4) = 0 (2.13)
which is nothing but the Weyl equation for the 2-component left-
handed neutrino L. It is easily seen that if s is to be a 4-vector,
must transform as a 2-component spinor ([4]).
Following a similar procedure for the anti-self-dual case, let
a =
2i1
1
i2
(2.14)
where =(
12
)is 2-component quantity and set fa =
aa.
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch02
66 Space, Time and Matter
Then, the anti-self-duality condition for fa implies that
fa24 = fa13, f
a34 = f
a21, f
a14 = f
a32 (2.15)
which in turn implies two equations for (1, 2).
11 i22 + 32 + i41 = 012 + i21 + 31 + i42 = 0.
(2.16)Introducing the matrices
1 =
(1 0
0 1
), 2 =
(0 ii 0
), 3 =
(0 1
1 0
), (2.17)
(2.16) can be written as
(kk + i4) = 0. (2.18)
Equation (2.18) diers from Eq. (2.13) only in the respect that the
set k diers from the Pauli matrices k only in the interchange of
the indices 1 and 3. As a result, while the Pauli matrices satisfy
12 = i3, 23 = i1, 31 = i2, (2.19)
the set k satisfy
12 = i3, 23 = i1, 31 = i2. (2.20)
Therefore, Eq. (2.18) describes a 2-component right-handed
neutrino R.
Conversely, if and satisfy (2.13) and (2.18), respectively, then
f = f s + fa with f
s =
s s and fa = a a,
where s and a are given by (2.8) and (2.14), would automatically
satisfy the Maxwell equations in vacuum ([8]). In other words, the
photon can be regarded in some sense as a fusion of a left-handed
and a right-handed neutrino L and R.
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch02
Matter 67
This suggests that the left-handed and right-handed neutrinos are
perhaps the fundamental constituents of all elementary particles and
that one should consider composite LR models with some fun-damental interaction between the two. Ideally, such a model should
be treated in the framework of quantum eld theory. However, as a
rst step, in this work, we shall consider a simple, heuristic quantum
mechanical model. This will give us an idea of what we can expect
in a rigorous quantum eld theoretical model.
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch02
68 Space, Time and Matter
2.2. A Neutrino Theory of Matter
2.2.1. Composite LR system withoutinteraction
Consider rst a composite LR system without interaction. TheHamiltonian of a left-handed (massless) neutrino L described by
(2.13) is given by HL = ic( ) with = (1, 2, 3) and itseigenfunctions EL satisfy
HLEL = ELEL
or ( )EL = (iEL/c)EL .
(2.21)From now on, we shall adopt the conventional units in which c =
= 1.
The solutions of (2.21) are well-known:
EL(x,p) = a(p)eixp
where
p2 p21 + p22 + p23 = E2L
and
a(p) =
(p1 ip2EL p3
). (2.22)
They describe a left-handed neutrino L with energy EL.
The spectrum of HL is not discrete and hence, the eigenfunctions
have to be normalized by the delta-function.
EL(x,p)|EL(x,p) = a(p)a(p)(p p). (2.23)
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch02
Matter 69
This may give rise to questions of rigor and technical diculties in
what follows. One can always enclose the neutrino in a cubical box
of length L, in which case the eigenfunctions would be
EL(x,n) = a(n)ei(2/L)xn,
where n = (n1, n2, n3), n21 + n22 + n
23 = E
2L, with integer ni, and
therefore, discrete energy values. We shall not follow this procedure
and proceed as if we are dealing with a discrete problem.
The eigenfunctions are also -fold degenerate, since p can takeany value on the energy shell. We shall remove this degeneracy by
integrating EL(x,p) over the energy shell S2EL : p21 + p
22 + p
23 = E
2L
which is a 2-sphere of radius EL. We get (with a slight abuse of
notation)
EL(x) =
S2EL
a(p)eixpdSp (2.24)
as a surface integral over S2EL (see later). Furthermore, we shall sup-
pose that EL(x) are normalized.
Similarly, the corresponding eigenfunctions for the right-handed
neutrino R with Hamiltonian HR = i( ) with energy ER aregiven by
ER(y,q) = b(q)eiyq
where
q21 + q22 + q
23 = E
2R
and
b(q) =
(q3 iq2ER q1
).
(2.25)
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch02
70 Space, Time and Matter
We use y for the coordinate of R and note that b(q) diers from
a(p) by an interchange of the indices 1 and 3. A similar integration
over the energy shell S2ER : q21 + q
22 + q
23 = E
2R gives
ER(y) =
S2ER
b(q)eiyqdSq. (2.26)
Let HL and HR be the respective Hilbert spaces for L and R.Consider now a composite LR system without interaction givenby the tensor product H = HL HR with the Hamiltonian Ho =HL + HR (or more precisely, HL I + I HR) ([13]). Since Ho =i( L) i( R), where L,R act on x,y, respectively, andthus does not depend on x and y explicitly, not only
1(x,y) = EL(x) ER(y) (2.27)
but also
2(x,y) = ER(x) EL(y) (2.28)
are both eigenfunctions of Ho with energy EL +ER. Thus, the com-
posite system LR behaves like a system of identical particles eventhough l is not identical to R. We remark that if the universe were
spatially non-orientable, L would be indistinguishable from R. But
in an orientable universe, they would be distinct particles. We thus
have a 2-fold degeneracy similar to the He-atom.
2.2.2. Composite LR system with interaction
Consider now an interaction V (x,y) between L and R of the form
V (x,y) = F (|x|)H(|x| |y|) + F (|y|)H(|y| |x|) (2.29)
-
February 22, 2014 14:2 9in x 6in Space, Time and Matter b1716-ch02
Matter 71
where F is a function to be specied and H is the Heaviside function.
Thus