Theorem For Nonrotating Singularity-Free Universes
Theorem For Nonrotating Singularity-Free Universes
The singularity theorems [1] of Hawking and Penrose In any singularity free nonrotating universe, open in
led to a widely held belief that a time or null geodetic in- all directions, the space-time average of all stress energy
completeness is an essential feature of all relativistic cos- invariants including the energy density vanishes.
mological solutions. That the proof of the theorem rested In the above, nonrotating means that the world lines of
on a number of conditions was often overlooked. There the matter in the universe form a normal (i.e., hypersur-
were four notable conditions: (1) The causality condi- face orthogonal) congruence. The singularity free nature
tion requiring the nonexistence of closed timelike lines, requires, in particular, the scalars from the Riemann ten-
1
(2) the strong energy condition fTik 2 2 Tgik gy i y k $ 0, sor to have bounded values, and openness in all directions
(3) a generality condition on the Riemann-Christoffel ten- means that the space time has topology R 3 3 R.
sor, and (4) the existence of a trapped surface. About the For such a universe taking the x 0 axes along the world
last condition, Misner, Thorne, and Wheeler [2] remark, lines of matter, the metric may be written in the form
“All the conditions except the trapped surface seem emi- 2
nently reasonable for any physically realistic space time.” ds2 g00 dx 0 1 gab dx a dx b , (2)
It is interesting that the recently discovered singularity where the Greek indices run from 1 to 3. The domain of
free solutions of Senovilla et al. [3,4] violate precisely the all the coordinates is from 2` to 1`.
trapped surface condition, whereas the other three condi- Our assumption about the openness in all directions
tions hold good. True it is difficult to reconcile the Sen- means that the ratio of the volume of any three dimen-
ovilla solutions with the characteristics of the presently sional subspace to that of the entire space time vanishes,
observed universe; nevertheless, hopes have been raised i.e.,
that there may exist singularity free solutions which can RRRp
serve as faithful models of the observed universe. j3 gj dx i dx k dx l
A look at the simplest Senovilla solution [3] reveals RRRRp 0, (3)
jgj d 4 x
some interesting features. The space time is open in all
the four dimensions but the physical and kinematic scalars where the indices i, k, l are unequal and may refer to
all vanish so rapidly at spatial and temporal infinity that space or time coordinate, j3 gj is the appropriate coefficient
their space time averages taken over the entire space time to give the invariant volume for the three dimensional
vanish. The average of a quantity x over the entire space element. For the unit vector y i along the timelike
time is defined as follows: coordinate x 0 , we have the Raychaudhuri equation
" R1x0 R1x1 R1x2 R1x3 p 4
#
2x0 2x1 2x2 2x3 x jgj d x u;i y i 1 yÙ i;i 1 1
u 2 1 2s 2 1 kfTik 2 1
gik T gy i y k 0 .
kxl ; R1x0 R1x1 R1x2 R1x3 p . 3 2
4
2x0 2x1 2x2 2x3 jgj d x lim xo ,x1 ,x2 ,x3 !` (4)
(1) Taking the space time average of each term in the
Of course, the average is defined only if the limit exists. above equation, we get [here the space time averages are
For the Senovilla solution, over infinite space time in the sense defined in (1)],
3kpl krl 0 ,
Ù
2kyÙ ;ii l 2 kul 1
ku 2 l 1 2ks 2 l
3
ku 2 l 0 ,
Ù 0; kyÙ m l 0 . 1 kkfTik 2 1
gik Tgy i y k l . (5)
kul ;m
2
All the above scalars appear linearly in the Raychaudhuri With the strong energy condition fTik 2 12 gik T gy i y k $
equation, and we present below a proof of the following 0 all the terms of the right-hand side are positive definite.
general theorem: Hence to get a positive value of the average density, we
must have the left-hand side positive. The first term on ing of uÙ as x 0 ! 6` may not make the x 0 integral conver-
the left gives gent. Nevertheless, the vanishing of uÙ would reduce the
R ip
yÙ ;i 2g d 4 x order of divergence of the numerator integral compared
kyÙ ;i l ; R p
i
0. to the integral in the denominator and hence one again
2g d 4 x Ù 0. Consequently all the averages occurring in
has kul
The integral in the numerator can be converted to an Eq. (5) vanish.
integral of yÙ i over the three surface orthogonal to yÙ i at The generality of our treatment needs to be emphasized.
infinity. As yÙ i is orthogonal to y i , this three surface The solutions of Senovilla type were based on the existence
contains two spacelike Rand one timelike dimension. In of doubtful symmetries and an apparently ad hoc splitting
any case it is given by jyÙ i j jdSj where jy i j is the norm of metric tensor components into factors involving sepa-
of the vector yÙ i and jdSj is the proper volume of the rately the time and space coordinates. Our result is based
orthogonal three dimensional element. solely on the existence of a global time coordinate which
The velocity vector of matter y i appears in the expres- is hypersurface orthogonal—one is tempted to identify it
sion for Tik . Thus the equation with the absence of rotation in the universe. The implicit
Rki 2kfTki 2 1
T dki g idea in our discussion is that the gravitational collapse is
2
arrested by the action of acceleration and that again means
makes y i expressible as an algebraic expression of the existence of a nongravitational force. In such situ-
Ricci tensor components. In particular, if the matter ations, our theorem shows that one has to sacrifice the idea
is perfect fluid, y i is the unit timelike eigenvector of of a finite average density. As one feels that observations
Rik . Hence, quite generally the kinematic variables like rule out such an “empty” universe, the hope of a reason-
the acceleration yÙ i , expansion u will be determined by able realistic singularity free cosmological solution has to
the Ricci tensor and its covariant derivatives. So an be given up.
unbounded value of any kinematic scalar would mean The question that naturally arises is the relation between
scalars of the Riemann tensor blowing up and thus signal the present theorem and the trapped surface condition. We
Ù etc., to be
a singularity. We can hence take jyÙ i yÙ i j, u, u, have not addressed ourselves to this in the present discus-
bounded everywhere. sion. Our condition seems physically more transparent.
Consequently, Thanks are due to the participants of the Relativity
R
jyÙ i j jdSi j and Cosmology Seminar in Jadavpur University for their
kyÙ ;i l R p
i
0, (6) helpful comments.
2g d 4 x
where we have used (3).
Ù defined by
In evaluating the value of kul
R p
uÙ 2g d 4 x [1] S. W. Hawking and G. F. R. Ellis, The Large Scale
Ù Rp
kul 0, Structure of the Universe (Cambridge University Press,
2g d 4 x Cambridge, 1973), Chap. 8, pp. 261– 275.
we note that as x 0 ! 6`, uÙ vanishes sufficiently rapidly
[2] C. W. Misner, K. S. Thorne, and J. A. Wheeler, Gravita-
p tion (W. H. Freeman and Company, San Francisco, 1973),
so that u remains finite for x 0 ! 6`. If 2g is finite as p. 935.
x ! 6`, then obviously the integral over x 0 in the nu-
0
[3] J. M. M. Senovilla, Phys. Rev. Lett. 64, 2219 (1990).
merator will converge to a finite value and consequently [4] E. Ruiz and J. M. M. Senovilla, Phys. Rev. D 45, 1995
Ù will vanish. If, however, p 2g blows up, the vanish-
kul (1992).
655