Anda di halaman 1dari 20

The Most Misunderstood Exact Solution In

Theoretical Physics
R. E. Salvino
9 Thomson Lane, 15-06 Sky@Eleven
Singapore 927726
March 24, 2015

Abstract
The case is made that the Hilbert or textbook Schwarzschild solution is an excellent choice for the most misunderstood exact solution in
theoretical physics. This is based on the manner in which the solution
is obtained, the underlying mathematical properties that surround the
solution, the behavior of the solution in terms of relevant variables,
the physical significance of the solution, and the misguided analyses
that tried to turn a solution of the time dependent spherically symmetric vacuum field equations into a soluton of the time independent
equations.
Keywords: Hilbert solution, Schwarzschild solution, Combridge-Janne
solution, Birkhoffs theorem, time dependent gravitational field equations

Introductory Remarks

For the past 100 years, the Hilbert or textbook Schwarzschild solution [1, 2] may be best characterized as a solution in search of a problem.
While the solution arose in response to the time independent vacuum gravitational field equations, it seemed to resist the characterization as a time
independent solution. A significant amount of molding, hammering, and
massaging was performed to cast the solution in a manifestly time independent form. It has become apparent, however, that the molding, hammering,
and massaging was done in a hurried manner and contain a number of flaws.
1

This does not mean that the solution is incorrect, but it does mean that the
solution is not what it has been made out to be. Thus, it should be emphasized that the claim is that the solution is misunderstood, not that it is
incorrect. Misunderstandings may arise in the manner by which the solution is obtained, the behavior of the solution in terms of relevant variables,
and the physical significance of the solution which entails various forms of
interpretation. Many people assume that all exact solutions are adequately
understood, others may feel that solutions in the quantum domain are misunderstood, but the Hilbert solution has been misunderstood for nearly 100
years for a single reason: it has not been recognized that it is a solution of
the time dependent spherically symmetric vacuum field equations.
The ingredients for supplying a complete understanding of the solution
had been in place all along, but the urgency for a rush to judgement and
to demonstrate that general relativity was a viable theory of gravitation
with no outstanding peculiarities prevented conventional issues to be addressed and explained in a patient and reasonable manner. The solution
has been codified and sanctified to the point now that no one is willing
to re-examine the basis of the solution with an open mind. In addition,
the improper conventional formulation of Birkhoffs theorem prevented investigations into the time dependent vacuum field equations. As a result,
it is difficult to approach the subject without appearing contentious, inflammatory, or frivolous when all that is needed is a careful, honest, and
unbiased appraisal. Unfortunately, the re-discovery and translation of the
original 1916 paper by Schwarzschild [35] has provoked a series of attempts
to demonstrate that the original Schwarzschild solution is the one and only
solution to the time independent spherically symmetric vacuum problem,
and that the Hilbert solution is not a valid solution of the field equations.
As will become apparent, this claim is really only a quarter-truth and
has served to polarize the general relativistic community which, in turn, has
prevented any serious attempts to completely understand the situation.

A Marriage Made In Heaven

The time independent spherically symmetric problem permits a number of simplifications based purely on time independence and symmetry
considerations. Time independence requires the metric functions to be independent of the timelike coordinate and the condition that the line element
also show no time dependence requires g0k = 0 for k = 1, 2, and 3. One
2

argues from spherical symmetry that the metric tensor g is reduced from
16 components to 4 nonzero components which are the diagonal entries for
the tensor: g00 , g11 , g22 , and g33 with g33 = g22 sin2 [1, 2].
Similarly, since the Ricci tensor depends on the metric tensor, or more
exactly, on various derivatives of the metric tensor, one could argue that
spherical symmetry produces the same structure for the Ricci tensor. However, if one doesnt wish to rely on symmetry arguments, direct calculation
shows that the Ricci tensor does indeed have 4 nonzero components: R00 ,
R11 , R22 , and R33 with R33 = R22 sin2 [36].
The vacuum equations are obtained by setting the Ricci tensor equal to
zero. Thus, the time independent spherically symmetric vacuum equations
consist of 3 equations in 3 unknown metric functions. In other areas of
physics, this would be viewed as a marriage made in heaven. The conventional relativists, however, want a divorce.

I Want A Divorce

The traditional approach, following Hilbert [7], claims to exploit the


choice of coordinates by choosing coordinates such that g22 = r2 where r
is the radial coordinate. 1 For example, one starts with the line element in
the form
ds2 = A(r)c2 dt2 B(r)dr2 C(r)r2 (d2 + sin2 d2 )

(3.1)

and then introduces the new radial coordinate r by

r=

C(r)r

(3.2)

This requires the functiona A(r) and B(r) to transform according to

A(r) = A(r(r)) = A(r)


1

(3.3)

As far as I know, the only textbook that discusses the basis and context of this choice
is the one by Adler, Bazin, and Schriffer [1] and is summarized here.

B(r)dr2 =



r dC 1 2
B
1+
dr = Bdr2
C
2C dr

(3.4)

The line element (3.1) then has the form


ds2 = A(r)c2 dt2 B(r)dr2 r2 (d2 + sin2 d2 )

(3.5)

However, the range of the new radial coordinate r is not explicitly addressed
and it is tacitly assumed that 0 r < applies to Eq. (3.2). By writing
the g22 function as C(r)r2 rather than simply C(r), the new coordinate r
superficially appears to vanish for r = 0. But for this to be true, C(0) must
be finite and this is not necessarily true. As we will see in Section 5, the
behavior of r(r) as r 0 is related to a missing boundary condition.
From a more formal point of view, a choice of coordinates is one thing,
but a choice of coordinates such that ..., where ... is a condition to be
imposed, is something else. First, there is no guarantee that the condition
represented by ... is even consistent with the equations one is attempting to
solve. For the present problem, the condition that g22 = r2 is, in fact, not
consistent with the time independent field equations: it produces a solution
of the time dependent field equations that is time independent outside of
the event horizon at r = 2mG . Of more general relevance, a choice of
coordinates never results in the actual implementation of the condition to
which it has been tied. At best, a suitable choice of coordinates will allow
a condition to be imposed that may be difficult or impossible to impose
in other coordinates, but it does not result in the actual implementation
of the condition. For example, Schwarzschilds original solution [35] was

devised to also satisfy the condition that g = 1 since he was working with
Einsteins first 1916 article [8] in which that condition was stated as a part

of the theory. It is not possible to satisfy g = 1 in standard spherical


coordinates without severely constraining the radial coordinate r and the
angular coordinate . Schwarzschild introduced a new set of coordinates that
removed these constraints and which permitted the condition to be satisfied
without imposing any conditions on the coordinates. However, writing the

metric in these coordinates did not automatically result in g = 1, that


condition still had to be imposed by hand. In similar fashion, the choice
of coordinates in the conventional approach does not result in g22 = r2 ,
since an additional, unstated, and very specific boundary condition is also

being imposed. As a result, the g22 metric function is chosen to be r2 by


fiat.
From a pedagogical point of view, removing the g22 function from the
list of unknowns is unwise unless one can simultaneously demonstrate that
one has reduced the number of equations as well. Otherwise, there are
now three equations in two unknown functions and one risks having created
an overdetermined system. However, it turns out that the system is not
overdetermined and the resulting solution does satisfy all field equations.
But it also turns out that a choice of coordinates has nothing at all to do
with the matter. So was this just a piece of luck that the solution worked
out or is there something more fundamental in play?

Combridge-Janne? Whos that?

In the context of a perturbative treatment of the time independent spherically symmetric inhomogeneous equations, W. de Sitter [9] found that there
were only two independent field equations, not three as is usually assumed.
In the source-free limit or vacuum conditions, this is an exact statement.
This suggests that starting with three metric functions to be determined, the
field equations will provide a solution for two of the metric functions in terms
of the third while the third function remains undetermined. This is precisely
the solution that was found by J.T. Combridge [10] and H. Janne [11] independently in 1923 and which I denote by the name Combridge-Janne (CJ)
solution [12]: it expresses g00 and g11 in terms of g22 while g22 is undetermined. These results do not appear to be well-known. But now its clear
why the conventional approach works: g22 could be anything and still provide a solution to the field equations! The g22 function has to be restricted
by other considerations.
By choosing standard spherical coordinates defined by x = r cos sin ,
y = r sin sin and z = r cos , one obtains the line element in the standard
spherically symmetric form. By defining the metric functions to be g00 = e ,
g11 = e and g22 = R2 where , , and R are functons of the radial
coordinate r, the CJ solution [12] is
2mG
R

(4.1)

e = (R0 )2 e

(4.2)

e = 1

R = undetermined

(4.3)

As noted, the function R(r) is undetermined so that any choice for the
function R provides a viable solution. The function R is the radius of a great
circle of circumference C = 2R and is the radius of a sphere of surface
area A = 4R2 . It is only in the asymptotic Lorentz region where these
equations must approach their flat space counterparts, and this is satisfied
by the requirement that R r as r . This is the only requirement that
must be imposed on the function R. The fact that the textbook solution
provides the flat space results C = 2r and A = 4r2 throughout spacetime
should be viewed as an unexpected feature, not as a matter of course. There
is no reason to expect flat space results or flat space descriptions to be valid
in curved spacetime, especially in regions of large curvature.
While the CJ solution provides the basis for understanding how there
could be two Schwarzschild solutions, it does not provide the basis for claiming that these solutions are the same. It should be clear that providing
different functions for R(r) does not change the coordinate basis, the coordinates remain (t, r, , ) for all solutions. Thus, solutions which differ due
to different R(r) functions are not the same solutions expressed in different
coordinates, they are different solutions expressed in the same coordinates.
If we re-examine the coordinate choice argument presented in Section 3,
we see that we should make the identifications
C(r)r2 = R2 (r)

A(r) =

2mG
R

(4.4)




2mG 1
B(r) = (R ) 1
R
0 2

(4.5)

(4.6)

Substituting for B and C in the equation for B shows that






B
r dC 1
2mG 1
1+
=B = 1
C
2C dr
R

(4.7)

From the form of B(r) and using dR = R0 dr in the radial part of the line
element leads to the same result. But now the question about the behaviror
of C(0) is converted into the question about the range of R(r). Since R r
as r and since 0 r < , then R(0) R < . The function R(r) is
now the coordinate transformation that takes the original radial coordinate
r into the new radial coordinate R, but the value of R(0) is unknown as long
as the function R(r) is unknown. And remembering that R(r) is the original
g22 metric function, the value R(0) is the boundary value for that function
and must be obtained from the missing second boundary condition.

We Dont Need No Stinkin Boundary Condition

I have twice mentioned a missing boundary condition. To what does


this refer? A long history of mathematical physics tells us that we need an
adequate number of boundary conditions to uniquely determine the solution
of a linear differential equation. The number of boundary conditions is equal
to the order of the differential equation. This is actually a mathematical
statement without physical content. In fields such as electrodynamics and
classical gravity, the ruling equations are linear second order differential
equations and these equations require two boundary conditions to obtain
the unique solution.
In the case of the classical gravitational problem in unbounded space,
one boundary condition requires the potential to vanish as r . The
other boundary condition specifies the potential as the source of the field is
approached: 1/r behavior if the source is a point mass, Dirichlet conditions
if the potential is specified on the surface of a finite-size spherical source,
or Neumann conditions if the gradient of the potential is specified on the
surface of a finite-size source. These are three physically distinct solutions
corresponding to three physically distinct sources and are specified by three
distinct boundary conditions. This is an illustration of the mantra of typical
graduate courses in mathematical methods of physics: the mathematics is
in the solution of the equations, the physics is in the boundary conditions.
The fact that classical potential theory produces a potential which is overlaid
upon a pre-existing spacetime scaffolding while general relativity produces
the spacetime scaffolding itself is not relevant. This is a mathematical issue
independent of the interpretation of the equations and solution of the equations. And it simply reflects the idea that general relativity should produce
different spacetime scaffolds for different sources of spacetime curvature.
7

The general relativistic field equations are a set of nonlinear second order
differential equations. But if these equations are examined in the weak field
limit, a set of linear second order differential equations are obtained. Since
these equations require two boundary conditions for the unique solution to
a problem, one should also expect the set of nonlinear equations to require
two boundary conditions. However, the conventional approach produces the
unique solution by imposing the single boundary condition in the asymptotic
spacetime and never addresses the issue of a second boundary condition. In
fact, the solution proceeds without ever mentioning the source of spacetime
curvature at all, in stark contrast to conventional potential theory. There
must be a second boundary condition somewhere, but where is it? The
introduction of the unknown quantity R(0) in Section 4 supplies the hint:
look at g22 = r2 .

General Schwarzschild, At Your Service

The troubling aspect of the Combridge-Janne solution is that the determination of R(r) is ambiguous. One must supply an auxiliary condition
to determine R(r), but there does not appear to be any way to identify a
unique condition that must be imposed nor any justification for claiming
that that condition is a necessary condition. Furthermore, there doesnt
appear to be any particular reason to prefer one condition over another.
For instance, one may follow Schwarzschild for historical reasons and

choose coordinates so that the condition g = 1 can be imposed. Do


ing so is equivalent to requiring g = r2 sin in the original spherical
coordinates [12]. This condition then has the convenient but unexpected
feature that the 3-space volume element is the same as the flat 3-space vol
ume element. Imposing the condition g = r2 sin then yields a simple
differential equation for R(r) [12]

R2

dR
= r2
dr

(6.1)

which has the solution


R(r) = (r3 + r03 )1/3

(6.2)

where r0 is a constant of integration and is specified by means of the second missing boundary condition. Schwarzschild was explicitly solving the
problem for a point mass source and, consequently, he required the metric functions to be well-behaved everywhere except possibly at the spatial
origin r = 0 which he identified as the location of the point mass source.
This boundary condition requires r0 = 2mG . One arrives at the Hilbert
solution by imposing the boundary condition that g22 = r2 everywhere.
This requires r0 = 0. It should be clear that these two solutions are obtained from the same function by imposing different boundary conditions
on that function. The solutons are distinct, they are tied to distinct sources
of spacetime curvature, and the sources are reflected in the boundary conditions that are imposed on the function R(r). This is fully in line with
conventional methods of mathematical physics.
In addition, one can impose the boundary condition that the metric
functions be well-behaved everywhere including r = 0. This requires r0 >
2mG . For this case, if one sets r0 = R with R > 2mG , then r = 0 can
be interpreted as the surface of a spherical object of radius R and this will
allow Dirichlet and Neumann conditions to be applied. Thus the solution
(6.2) provides the general relativistic analogues to the classical gravitational
solutions for a point mass, a spherical object with Dirichlet conditions, and
a spherical object with Neumann conditions. In addition, it includes the
solution for the distinctly general relativisitic object, the wormhole. In
fact, it provides a family of wormholes since the solutions for r0 < 2mG
all show the same behavior but differ in the size of the wormhole with the
Hilbert solution as the maximum wormhole. All of these wormhole solutions
share one important property: they are all solutions of the time dependent
field equations that are time independent outside of their respective event
horizons and are not strictly time independent solutions.This, at bottom, is
the reason there is no classical analogue of this general relativistic solution.
Classically, the point mass is the limit of a spherical mass as the radius
of the sphere goes to zero. General relativistically, the point mass is the
boundary between the wormhole and the spherical mass: it is the limit of
a spherical mass as the radius of the sphere goes to zero and also the limit
of a wormhole as the size of the wormhole goes to zero. The point mass does
have an event horizon, but it coincides with its location at r = 0 and so it
has no inside.
Both the conventional view [1,2] and the Antoci-Leibscher view [13] claim
that only one solution obtained from (6.2) is valid. Strictly speaking, the
conventional view claims that all solutions obtained from (6.2) are equivalent, that the boundary condition used to obtain the value corresponding
9

value for r0 is of no consequence, so one might as well choose r = 0. This


makes very little physical sense and runs counter to nearly 250 years of
mathematical physics. The Antoci-Leibscher view maintains that only the
original Schwarzschild solution is valid and the only allowable boundary condition is the point mass condition that requires r0 = 2mG . This also makes
very little physical sense and also runs counter to the same extensive history
of mathematical physics. Their proofs, with varying degrees of explicitness
and implicitness, assume the source object is a point mass so its hardly surprising that they find the solution must correspond to that of a point mass.
They are correct, however, in observing that the Hilbert solution is not a
solution of the time independent equations, although they do not appear to
recognize that it is a solution of the time dependent field equations.

Fools Rush In

The main difficulty with the Hilbert solution is that one set out to solve
a time independent problem and yet wound up with a solution that was time
independent outside the event horizon at r = 2mG but was time dependent
inside the event horizon. Since it was believed that this solution had to be
time independent, one looked for coordinates which would show that the
solution was indeed time independent. The initial introduction of isotropic
coordinates was believed to have succeeded in accomplishing this goal. As a
result, the Hilbert solution was declared time independent and the original
spherical coordinates were labeled as pathological Schwarzschild coordinates since the time dependence inside the event horizon was believed to
be an artifact of those coordinates.
However, it was shown that the tranformation to isotropic coordinates
was valid only for r > 2mG , outside the event horizon, and the corrected
transformation shows the event horizon remains intact and the metric is
time dependent inside the event horizon in the corrected isotropic coordinates [14]. Since the focus had been on writing the line element in terms of
the isotropic coordinates, the fact that the transformation to the isotropic
radial coordinate was valid only for r 2mG had been missed or ignored. It
is possible to continue to use this transformation inside the event horizon,
but the isotropic radial coordinate must then play the role of a timelike coordinate inside the event horizon. Then one simultaneiously must convert the
original timelike coordinate to a spacelike radial coordinate. This mimics
the role swapping that occurs in the original spherical coordinates.
10

The Kruskal-Szekeres transformation [1, 15] fares no better. If one examines how the transformation works both inside and outside of the event
horizon, one sees that the transformation makes the region inside the event
horizon time independent by identifying the interior region with a finite shell
just outside the event horizon. In other words, it inappropriately identifies
the timelike coordinate inside the event horizon with the spacelike radial
coordinate outside the event horizon. Correctly maintaining the timelike
and spacelike coordinates in the regions interior to and exterior to the event
horizon leads to the same result found for the proper isotropic coordinate
transformation: the event horizon remains intact and the metric is time dependent inside the event horizon while remaining time independent outside
the horizon.
There is nothing pathological about any of these coordinates, whether
the coordinates are spherical, isotropic, Kruskal-Szekeres, or any other set
of coordinates. The properties of the solution, in particular the time dependence of the solution inside the event horizon, may be undesirable from the
point of view of trying to establish a time independent solution, but they
are physically meaningful properties and the solution does make sense if it
is allowed to speak for itself. The only pathological aspect of the solution
was the attempt to make a piece-wise time dependent solution turn into a
completely time independent solution.

General Birkhoff And His Theorem: Time Dependent


Solutions

The original form of Birkhoffs theorem [2, 16] was a demonstration that
the solution of the time dependent vacuum field equations was identical
to the solution of the time independent vacuum field equations. As a by
product of this theorem, an expanding, contracting, or pulsating spherical
distribution of mass cannot emit gravitational waves since the solution exterior to the distribution of mass is necessarily time independent. However,
the proofs of the theorem make assumptions which should actually be consequences of the theorem, not assumptions upon which the theorem is based.
In particular, the proofs assume that the g22 function is both independent
of time and has the specific form g22 = r2 . Since g22 = r2 is part of
the time independent solution, it should not be assumed in the proof that
only g22 = r2 is an allowable solution. Similarly, restricting the g22 function to be time independent should not be assumed, the time independent
11

property should be the result of the proof. In addition, the metric must be
assumed to be diagonal: if the metric cannot be put in diagonal form, then
the nondiagonal metric provides an immediate counterexample to Birkhoffs
theorem.
By removing the assumptions of time independence and the specific conventional form for the g22 function, it has been shown that the solution of
the time dependent vacuum equations is identical to the solution of the time
independent vacuum equations if and only if the g22 metric function is independent of time [17]. In other words, a time independent g22 function is a
necessary and sufficient condition for the equivalence of the time dependent
and time independent solutions. And in that case, the time independent
solution is the Combridge-Janne solution and not the Hilbert solution [17].
The sufficient condition is an immediate consequence of the time independence of the g22 function. The proof of the necessary condition requires
a single example of a time dependent solution that was distinct from the
time independent solution. One example of a time dependent metric that
was provided in the proof was the spatially homogenous metric which was
independent of all spatial coordinates and dependent solely on the timelike
coordinate [17]. Another time dependent solution that was provided was
the solution of the linearized equations. Under certain conditions, this solution produces the spherical gravitational wave solution of the linearized
equations [17].
The interesting thing about the spatially homogeneous solution is that,
for certain values of the constants in the solution, the spatially homogeneous
solution evolves in time, possibly from a singularity at an initial time t0 = 0,
until at time tH = 2mG /c the metric abruptly becomes time independent
and reproduces the Hilbert solution outside the event horizon: the Hilbert
solution is a time dependent solution that becomes time independent after
the time tH = 2mG /c [17]. This provides an explicit demonstration that
the Hilbert solution develops from a purely time dependent and spatially
homogeneous metric 2 and is not a solution of the time independent field
equations.
2
Curiously, this spatially homogeneous metric shows similarities to Geftners definition
of nothing [18].

12

Ive Fallen And I Cant Get Up

The conventional infalling observer analysis attempts to demonstrate


that the observer not only reaches the event horizon in a finite amount
of proper time, but the observer also reaches the spatial origin in a finite
amount of proper time [1]. The event horizon is viewed as a one-way membrane so that the infalling observer, once the event horizon has been passed,
can not re-enter the spacetime region outside the event horizon. However,
the conventional analysis mixes a time calculation outside the event horizon with a length calcuation inside the event horizon since the timelike and
spacelike radial coordinates swap roles inside the event horizon. Whether
one uses these coordinates with swapped roles or just inspects the time dependent solution, one finds that the singularity inside the event horizon is
not a spatial singularity but a temporal singularity, it exists at one specific time but not for any other times [17]. The infall analysis terminates
at the event horizon, but since the solution is spatially homogeneous inside
the event horizon, reaching the event horizon can be viewed as the same as
reaching the spatial origin. The observer does not reach the temporal origin, and the singularity is not encountered. As noted, the infall trajectory
terminates at the event horizon, so in that sense the event horizon can be
identified as a singularity.
The conventional analysis does apply unchanged for r 2mG [1],

1
1
2m
G
t =
1
c
r
r 2 =

(9.1)

2mG
r

(9.2)

where the initial conditions have been chosen so that ct = s as r in the


asymptotically flat Lorentz spacetime as in special relativity. The solution
of the second of these equatons is


2
3/2
3/2
s s0 =
r r
,
3 2mG 0

r 2mG

(9.3)

where r0 is the radial position at the proper time s0 /c. The solution for the
trajectory in terms of coordinate time t is given by
13

c t t0




2
3/2
3/2
r r 6mG ( r r0 )
=
3 2mG 0

( r + 2mG )( r0 2mG )

+ 2mG ln
,
( r0 + 2mG )( r 2mG )

r 2mG (9.4)

Note that while t logarithmically as r 2mG , the proper time to


reach r = 2mG is finite


s2mG
s0
2
3/2
3/2
r (2mG )

=
c
c
3 2mG 0

(9.5)

However, inside the event horizon the timelike coordinate is t = r/c and
the spacelike radial coordinate is r = ct. So, inside the event horizon the
equations are

1
2m
G
r = 1
ct
2

c2 t =

(9.6)

2mG
ct

(9.7)

The solution of the second of these equations is




2 c3/2
3/2
3/2
s1 s =
t t1
3 2mG

r < 2mG

(9.8)

where t1 is the coordinate time correspondiing to s1 , while the solution of


the first of the equations is given by



p
p

2
3/2
3/2
r1 r =
c3/2 t c3/2 t1 + 6mG ( ct ct1 )
3 2mG
p

( ct + 2mG )( ct1 2mG )

2mG ln p
, r < 2mG (9.9)

( ct1 + 2mG )( ct 2mG )


14

where r1 is the radial coordinate value at the coordinate time t1 . However,


it is important to note that the condition r < 2mG can be expressed as

r < 2mG

t<

2mG
c

(9.10)

Thus, both t1 and t are required to be less than 2mG /c inside the event
horizon. But for times less than 2mG /c, the metric functions are completely
independent of all spatial coordinates. Consequently, the infall must have
initiated at a time t0 > 2mG /c and occurs for times t t0 > 2mG /c.
Therefore, as stated above, the infalling trajectory must terminate at the
event horizon r = 2mG but, since the metric is spatially uniform inside the
event horizon, reaching the event horizon is indistinguishable from reaching
the spatial origin r = 0. However, no singularity is encountered since the
singularity is a temporal singularity occuring at the single time t = 0 and
there is no spatial singularity at r = 0. The event horizon is a one-way
membrane, but it does not hide a spatial spacetime singularity.

10

Survey Says ...

The Hilbert solution is a solution of the time dependent, spherically symmetric, vacuum field equations and is not a solution of the time independent
equations. It is spatially homogeneous until the time tH = 2mG /c when the
solution abruptly transitions from a time dependent and translationally invariant solution to a time independent and rotationally invariant solution
for all times greater than tH . Thus, it is not surprising that the solution
has been misunderstood for so long since the time independent part of the
solution came about first in response to the time independent vacuum field
equations. The apparent time dependence of the solution inside the event
horizon was viewed as spurious and attempts had been made to explain this
time dependence as due to pathological coordinates. And since the conventional form of Birkhoffs theorem later precluded even the possibility of
a time dependent solution that could transition into the time independent
part of the Hilbert solution, attempts were made to hammer the solution
into a fully time independent mold. The proper formulation of Birkhoffs
theorem [17] demonstrates that the Hilbert solution is indeed a time dependent solution and there is no longer a need to try to force the solution to
play the role of an exclusively time independent solution.
15

The Hilbert solution never had an opportunity to speak for itself. The
formulation of the problem was incomplete, the appropriate boundary conditions were never fully enunciated, and the resulting time dependence was
wished away rather than given proper handling, even prior to the emergence
of the conventional Birkhoff theorem. Whats particularly disturbing is that
there is nothing arcane and obscure in this material, it simply requres patient and careful development of traditional ideas of mathematical physics.
And the ideas and techniques for such a development were fully in place
and a complete understanding of the problem and solution could easily have
been obtained at the time. But the sense of urgency to establish general
relativity as the correct extension of classical gravitational theory into the
relativistic domain got in the way.
As mentioned above, the solution was originally obtained from the time
independent field equations and, consequently, the view was that the solution must be time independent and a great deal of mental gymnastics
was expended to try to show that such is the case. This was the reason
for labeling the original spatial spherical coordinates as the pathological
Schwarzschild coordinates. The incomplete form of isotropic coordinates
provided the basis for the faulty claim that the Hilbert solution was indeed
time independent and had the appearance of a solution for a point mass in
these coordinates. Once the original isotropic coordinates permitted and solidified the misinterpretation and misunderstanding of the Hilbert solution,
the Kruskal-Szekere coordinates finalized that misunderstanding since there
was no need to worry about the mishandling of the coordinates inside
the Schwarzschild radius by mapping those coordinates into the coordinates
outside the Schwarzschild radius.
So, how should the Hilbert solution be viewed and presented? First, the
problem should be reformulated without any simplifications. To claim
that a solution for the g22 function can be obtained simply by means of a
choice of coordinates is disingenuous at best: no one would ever make a
similar claim in any other area of physics.
Second, the time independent spherically symmetric field equations consist of two independent field equations and not three as is usually maintained.
This must become part of the common knowledge base of general relativity.
Third, the Combridge-Janne solution must be recognized as the proper
solution to the time independent spherically symmetric vacuum field equations. And this requires that R(r) 2mG for all values of the radial coordinate r. The Combridge-Janne solution for which R(r) < 2mG for some
values of r must be recognized as a solution of the time dependent field equations. To obtain a solution for R(r), one must supply an auxiliary condition
16

for that purpose. But even if the function R(r) is used as the radial coordinate, the boundary value R(0) can not be discarded simply by choosing
R(0) = 0. This is, in fact, related to the next issue: the number of correct
boundary conditions needed to provide the unique solution to the problem.
Fourth, the field equations are second order differential equations and
require a second boundary condition to provide the unique solution. The
idea that there can be only one source for the general relativistic problem, or
that the source is irrelevant, or that all sources are equivalent, runs counter
to an extensive history of mathematical physics and potential theory. The
boundary value R(0) is reflective of the source object, and different values
for R(0) are appropriate for different source objects. The claim that they
are all the same is analogous to claiming that the electrostatic potential for a
point mass and a dielectric sphere and a conducting sphere are all the same.
Such claims make no physical sense. The objection that general relativity
provides spacetime structure while potential theory provides a function in a
given spacetime structure is irrelevant: different source objects provide different spacetime structures just as different source objects provide different
potentials. The correct number of boundary conditions is a mathematical
issue, the meaning of those boundary conditions is an issue for physics.
Fifth, a generalization of Schwarzschilds original solution provides a very
simple solution to the indeterminate nature of R(r) and explicitly illustrates
the issue of boundary conditions. However, Schwarzschilds condition on
the determinant of the metric is only one of many possible auxiliary conditions. But it also demonstrates that the Hilbert solution and the original
Schwarzschild solution are obtained from the same function R(r) with differences due to the different boundary conditions. Thus, the two solutions
are not the same solution in different sets of coordinates, they are different
solutions in the same set of coordinates. The original Schwarzschild solution
is the appropriate solution to the time independent spherically symmetric
vacuum problem for a point mass source. In this matter, Antoci and Leibscher are correct [13]. However, they are incorrect in maintaining that this
is the only solution for the time independent problem. They, too, implicitly
accept the idea that only one set of boundary conditions are allowable and
that that set is for a point mass source object; they, too, accept the idea
that there can be one and only one source object for the general relativistic
problem and that idea is embedded in all of their analyses. As a result,
they show that the original Schwarzschild solution is the only permissible
solution because they assume that that solution for a point mass is the only
permissible solution.
Sixth, an objective and impartial evaluation of the various coordinate
17

transformations shows something that should be expected: it is not possible to convert a time dependent solution into a time independent solution
by means of a coordinate transformation. Such a coordinate transformation necessarily involves interchanging timelike and spacelike characteristics
of the relevant coordinates and, in priniciple, would permit transforming
a timelike spacetime interval into a spacelike spacetime interval, and vice
versa. Such transformations are not permitted for the simple reason that
they violate the basic and fundamental physical distinctions between timelike and spacelike intervals.
Seventh, the proper formulation of Birkhoffs theorem demonstrates the
built-in bias due to simplifications. Since Birkhoffs theorem is used as a
proof that the g22 function must be given by g22 = r2 , this statement can
not be used in the body of the proof, it must be a consequence of the proof.
Properly formulated, Birkhoffs theorem demonstrates the central importance of the g22 metric function goes above and beyond a possible role as a
radial coordinate. And in the process, it provides the basic understanding
of the Hilbert solution as a time dependent solution that transitions to a
time independent solution at the time tH = 2mG /c. The time independent
aspect of the Hilbert solution is an outgrowth of its initial time dependence
and this can be correctly viewed only from the vantage point of the time
dependent field equations.
Eighth, the Hilbert solution is depicted as the solution for a wormhole.
However, the wormhole is not the object that everyone believes it to be.
An infalling observers trajectory terminates at the event horizon. In that
sense, the event horizon is located at a singluarity. It is not, however, a
rupture in spacetime in the sense that the the spatial location of a point
mass provides a point rupture: the measures of curvature are not unbounded
at the location of the event horizon. The spatial location r = 0 is also not
the location of a singularity inside the event horizon. A spacetime curvature
singularity existed at the time t = 0 but no longer exists for times t > 0.
An infalling observer takes an infinite amount of coordinate time to reach
the event horizon but reaches the event horizon in a finite amount of proper
time. The trajectory takes the observer explicitly to the location of the event
horizon, but since the metric is spatially uniform inside the event horizon,
that is equivalent to reaching the spatial origin r = 0. However, the observer
does not encounter the singularity since the singularity does not exist for
times t > 0. For the observer to encounter the singularity would require
time travel back to t = 0 and such travel is not part of the trajectory of the
infalling observer. The observer wanders in a homogeneous space inside the
event horizon with, presumably, no means of escape.
18

While the above eight issues form the basis for claiming that the Hilbert
solution is the most misunderstood exact solution in theoretical physics, the
key facet that leads to these misunderstandings is that the Hilbert solution is
a solution of the time dependent field equations and not the time independent
equations. At bottom, it was the attempt to transform away rather than
to understand the time dependence of the Hilbert solution that created all
the misunderstandings surrounding the solution. A proper understanding
of this time dependence and a proper handling of the other identified issues
can remove all of the misunderstandings and enrich the general relativistic
approach to spherically symmetric vacuum problems.
References
[1] R. Adler, M. Bazin, and M. Schiffer, Introduction to General Relativity,
Second Edition (McGraw-Hill, New York, 1975).
[2] C.W. Misner, K.S. Thorne, and J.A. Wheeler, Gravitation (Freeman,
San Francisco, 1973).

[3] K. Schwarzschild, Uber


das Gravitationsfeld eines Massenpunktes
nach der Einsteinschen Theorie, Sitzungsber. Preuss. Akad. Wiss.,
Phys. Math. Kl., 189, (1916).
[4] K. Schwarzschild, On the Gravitational Field of a Mass Point According to Einsteins Theory, trans. by S. Antoci and A. Loinger, Gen.
Relativ. Gravit. 35, 951 (2003). Also arXiv:physics/9905030v1 (1999).
[5] K. Schwarzschild On the Gravitational Field of a Point-Mass, According to Einsteins Theory, trans. by L. Borissova and D. Rabounski,
Abra. Zel. Jour. 1 (2008).
[6] R.E. Salvino and R.D. Puff, The Two Schwarzschild Solutions: A
Critical Assessment, Academia.edu (2013).
[7] D. Hilbert, Nachr. Ges. Wiss. Gottingen, Math. Phys. Kl., 53 (1917).
For an English translation, see S. Antoci, David Hilbert and the
Origin of the Schwarzschild Solution , Section 6: Appendix B,
arXiv:physics/0310104 v1 (2003).
[8] A. Einstein, Sitzsungber. Preusse. Akad. Wiss., Phys. Math. Kl., 778,
799 (1915).

19

[9] W. de Sitter, On Einsteins Theory of Gravitation, and Its Astronomical Consequences (First Paper), Month. Not. R. Astr. Soc., 76, 699
(1916). In particular, see Sections 10, pg. 711, and 11, pg. 714.
[10] J.T. Combridge, Phil. Mag., 45, 726 (1923).
[11] H. Janne, Bull. Acad. R. Belg., 9, 484 (1923).
[12] R.E. Salvino and R.D. Puff, Combridge-Janne Solution and the Gravitational g22 Metric Function, Academia.edu (2013).
[13] S. Antoci and D-E Liebscher, Reconsidering Schwarzschilds Original
Solution, Astron. Nachr., 322, 137 (2001). Also arXiv:gr-qc/0102084
v2.
[14] H.A. Buchdahl, Isotropic Coordinates and Schwarzschild Metric, Int.
J. Theor. Phys., 24, 731 (1985).
[15] M.D. Kruskal, Maximal Extension of Schwarzschild Metric, Phys.
Rev., 119, 1743 (1960).
[16] G.D. Birkhoff, Relativity and Modern Physics (Harvard University
Press, Cambridge MA, 1923).
[17] R.E. Salvino, Proper
Academia.edu (2015).

Formulation

of

Birkhoffs

Theorem,

[18] A. Gefter, Trespassing On Einsteins Lawn (New York, Bantam Books,


2014).

20

Anda mungkin juga menyukai