Sei sulla pagina 1di 11

TOPICS OF RELATIVISTIC COSMOLOGY

Nikolai V. MITSKIEVICH, Guadalajara-New Delhi

1 The general form of FRW metric estab-


lished from the properties of spatial homo-
geneity and isotropy
First we deduce the most general form of the three-dimensional squared in-
terval from the postulates of homogeneity and isotropy of the universe taken
at fixed cosmological time. The isotropy means, in terms of 3-curvature, that
not only spherical symmetry is applied, but all three existing in this case di-
agonal components of the Ricci tensor have the same values. Moreover, the
homogeneity demands that this unique value should be simply a constant
in all three-dimensional space. Thus the squared three-dimensional interval
(with the Euclidean signature + + +) has to be

dl2 = e2α dr2 + r2 (dϑ2 + sin2 ϑdφ2 ). (1.1)

From (1.1) immediately follow all three (in this case) connection 1-forms

e−α (2) (1) e−α (3) (2) cot ϑ (3)


ω (1) (2) = − θ , ω (3) = − θ , ω (3) = − θ (1.2)
r r r
as well as the Riemann curvature components

e−2α 1
R(1) (2)(1)(2) = R(1) (3)(1)(3) = α0 , R(2) (3)(2)(3) = 2 1 − e−2α

(1.3)
r r
which yield the Ricci tensor components

e−2α e−2α 1
R(1)(1) = −2α0 , R(2)(2) = R(3)(3) = −α0 − 2 1 − e−2α .

(1.4)
r r r
Since these components should be mutually equal and constant,
−2α
0e 1 −2α

α = 1 − e = −C, (1.5)
r r2

1
from where e−2α = 1 + Cr2 . This means that
dr2
dl2 = + r2 (dϑ2 + sin2 ϑdφ2 ). (1.6)
1 + Cr2
Square root of the first term in this expression can be written as √ dr = dχ,
√ 1+Cr2
sinh( Cχ)
while r = √
C
, so that (1.6) is equivalent to
 √  2
sinh Cχ
dl2 = dχ2 +  √  (dϑ2 + sin2 ϑdφ2 ). (1.7)
C

In this expression only three values of C are used: 1, 0, and −1, then giving

 = sinh χ when C = +1,
r= =χ when C = 0, (1.8)
= sin χ when C = −1.

They correspond respectively to the open three-dimensional (hyperbolical)


space, open plane three-space, and to closed (hyperspherical) three-space. It
is clear that the new variable χ is dimensionless, thus the interval (1.7) should
be multiplied by a factor having dimensionality of squared length. This factor
will be of great importance in cosmology, moreover, it will be not constant,
but depend on time thus directly describing the expansion of universe (its
scale factor). This expansion will not contradict to the spatial homogeneity
of the universe since the universe will expand as a whole, with the same rate
at all its points, naturally taken at one and the same time. This time will
be a privileged one, the time T of the synchronous system of coordinates,
alternatively the “cosmological time” differing from the synchronous one only
insignificantly. It is also worth mentioning that in the total ds2 only the part
dl2 will contain functions of spatial coordinates, but it will not change its
form under transformations adapted to the spatial homogeneity of dl2 , with
coming to a new origin of three spatial coordinates at any new point (the
form-invariance).
Thus we now come to the second step, introducing the squared four-
dimensional interval, Einstein’s equations for it, and leaving the consideration
of perfect fluid to the next section. This will be, in fact, a discussion of the
energy-momentum tensor of matter which fills the universe (supposed to be
a perfect fluid as averaged description of its material contents).

2
Using the spacetime signature (+, −, −, −), we write down the 4-interval
in the synchronous coordinates (and another, slightly different, system) as
ds2 = dT 2 − a2 dl2 = a2 (dt2 − dl2 ). (1.9)
Here a(T ) is the aforementioned scale factor depending on time T (but not on
spatial coordinates!); further on the right-hand side of (1.9) the same interval
is written in terms of the so-called cosmological time t, while the scale factor
is now supposed to depend on this t: dT = a(t)dt mutually relates these
two time variables and the scale factor. The latter will be determined via
solving Einstein’s equations when we shall sufficiently know their right-hand
side, the sources of gravitational field. Prior to considering the description
of perfect fluids, we shall deduce now the left-hand side parts of Einstein’s
equations for the squared interval (1.9).
The tetrad covector basis is
θ(0) = adt, θ(1) √

= adχ, √ (1.10)
θ(2) = √aC sinh( Cχ)dϑ, θ(3) = √aC sinh( Cχ) sin ϑdφ.
The differentials of these vectors are not too complicated:

dθ(0) = 0, dθ(1) = aȧ2 θ(0) ∧ θ(1) ,
√ √ 

dθ(2) = aȧ2 θ(0) ∧ θ(2) + √aC coth( Cχ)θ(1) ∧ θ(2) ,
√ √
dθ(3) = aȧ2 θ(0) ∧ θ(3) + aC coth( Cχ)θ(1) ∧ θ(3) + C cot
√ ϑ θ (2) ∧ θ(3) . 

a sinh( Cχ)
(1.11)
From them the connection 1-forms follow,
√ √ )
ω (0) (i) = aȧ2 θ(i)

, ω (1) (Ξ) = − a
C
coth( Cχ)θ(Ξ) ,
C cot (1.12)
ω (2) (3) = − a sinh( √ ϑ θ (3)
Cχ)

(here, as usual, the Latin index i takes values from 1 to 3, and additionally
we employed the capital Greek letter Ξ as another index to take values 2 and
3. Farther calculations yield the following expressions for components of the
curvature:
.
R(0) (1)(0)(1) = R(0) (2)(0)(2) = R(0) (3)(0)(3) = a12 ȧa ,

2 , (1.13)
R(1) (2)(1)(2) = R(1) (3)(1)(3) = R(2) (3)(2)(3) = aȧ2 − aC2 .
so that the Ricci tensor components read
 .   . "  2 #
3 ȧ 1 ȧ ȧ C
R(0)(0) = 2 , R(1)(1) = R(2)(2) = R(3)(3) = − 2 −2 − 2
a a a a a2 a
(1.14)

3
and the scalar curvature,
"    #
. 2
6 ȧ ȧ
R= 2 + −C . (1.15)
a a a

Hence Einstein’s conservative tensor (the left-hand side of Einstein’s equa-


tions) has components
h  i 
3 ȧ 2
G(0)(0) = − a2 a − C , 
h . 2 i (1.16)
G(1)(1) = G(2)(2) = G(3)(3) = a12 2 ȧa + aȧ − C 

from which it is obvious that the admissible source of the FRW gravita-
tional field on the right-hand side of Einstein’s equations can be the energy-
momentum tensor of a perfect fluid only (having arbitrary state equation).
Thus we now turn to consideration of perfect fluids and their properties.

2 Perfect fluids and reference frames


It is clear that we have here not to delve into the nature of perfect fluids (the
main ideas and possible restrictions in the statistical deduction of their prop-
erties) though, of course, such considerations might be of interest concerning
the details of this description of the averaged picture of galaxies and radi-
ation in the universe at its different evolution stages. In approaches to the
Friedmann–Robertson–Walker cosmological models dominates the viewpoint
that the present state of universe should be well described by a non-coherent
dust (perfect fluid without pressure, p = 0), but at much earlier stages there
had to exist more or less large admixture of non-coherent (chaotic) radiation.
I do not know if such mixed systems were sufficiently deep investigated, and
if it was really proved that they could be well described by the same theory
of perfect fluids, merely with the use of some averaged state equations. We
shall apply here the perfect fluids’ theory rather formally and taking into ac-
count only simplest equations of state. However one has to keep in mind that
if the universe is really built mainly of exotic materials such as dark energy
and dark matter, the situation could occur to be much different from that we
intend to consider below following the standard cosmological models. There
exists also other approach to fluids, not less formal, but much more flexible,
in which I am seriously interested [in particular, because I am its author, see

4
Mitskievich (1999a, 1999b)]; this is the field-theoretical description of perfect
fluids, but it will not be touched here in any detail since to this end a too
large additional material should be included in these lectures.
Perfect fluids (also called Pascal’s fluids) are those which are described
only by their mass distribution (mass density) and pressure (which is isotro-
pic); no viscosity nor thermodynamics are involved. Their theory is extremely
simple, it concentrates around the fluid’s energy-momentum (perhaps, in this
case it would be better to say: stress-energy) tensor

T = (µ + p)u ⊗ u − pg. (2.1)

Here µ is the mass (energy) density, p is pressure, u, the fluid’s local four-
velocity, and g, the metric tensor. It is clear that we have written this
tensor in the abstract completely invariant manner which, however, does not
mean that any of the characteristics we used are invariants in all-embracing
sense: the four-velocity u is a vector (or covector), and it is only given in the
invariant representation (let us not identify the concepts of a vector and its
components, this would be nothing more than to follow a vulgar habit). The
mass density and pressure are given here only in the “proper” sense, i.e., as
the proper mass density and proper pressure, which are nearer to “invariant”
in its ordinary meaning, but should be by no means understood as properties
whose measured values are independent of the state of motion of the observer
with respect to fluid (or vice versa).
In fact, here we, for the first time, encounter a specific problem: how
to separate a general description from its further application to particular
cases without destroying their very important unity (mutual succession)? It
is somewhat like the Zeno paradox, in our time existing more psychologically
than as a real obstacle. The problem consists not in any objective contra-
diction, but in a conflict between our more or less vulgar professional habits
and adequate understanding of the concepts we are operating with. The an-
swer will automatically come from a deduction and physical interpretation of
one important example: how change the directly observable characteristics
of a perfect fluid with a change of the reference frame in which the fluid is
considered? To this end we have to discuss not only the subject of perfect
fluids, but also that of the reference frames.
The concept of reference frame does not need for its definition neither use
nor even introduction of coordinates. When we measure some characteristics
of concrete objects or phenomena occurring in or between them, this can

5
be done without coordinates which are completely artificial inventions, while
in such measurements both observer and object (or objects) have to really
exist. Our old and often misleading habit is to base our definition of reference
frame on coordinates. Moreover, we mix these concepts together, we mistake
coordinates for reference frames and reference frames for coordinates, we treat
components of a vector when it is measured, as really observable quantities,
and this is completely wrong. Only in very particular cases and very special
circumstances this may partially work, but not in the curved spacetime, in
any case. Well, what is a reference frame? Take the simplest idealization of
an object or an observer, a three-dimensional point. To consider its motion
we have to introduce the fourth dimension, the time, and in spacetime we
now speak about such an object as about a world line. A reference frame in
general is an idealization of a swarm, a cloud of observers and their measuring
devices, all being only test objects to create no disturbances in phenomena
under measurement and scientific investigation in general. The basic ideas in
this formalism of reference frames were independently discovered by J. Ehlers
in Germany and A. Zel’manov in the Soviet Union. It is remarkable that they
both came to these ideas from cosmology and the use there of perfect fluids.
Thus we have to work with a congruence of timelike world lines which is
equivalent to a timelike vector field. Since we do not build anything more
than the concept of a reference frame, we take as already existent the metric
tensor in order to calculate the square of this vector and finally to normalize
it to unity. Thus we admit a unit vector field (say, τ ) as a definition of
reference frame.1 A combination of it and the metric tensor gives a new
tensor perpendicular to τ :
b := g − τ ⊗ τ (2.2)
which simultaneously is the projector on the local subspace ⊥ τ (globally, for
the field of τ , ⊥ to this field, but now in general not forming a global subspace;
such a subspace is formed only in the case of non-rotating congruence) and
the three-dimensional metric tensor (with the signature 0, −, −, −) on this
subspace. “0” in this signature means that b is in fact a three-dimensional
object restricted to the local subspace ⊥ τ , though it is formally described
as a four-dimensional one: for example, if we introduce some system of co-
ordinates, the components of both τ µ and bµν will be denoted using the
four-dimensional indices, but the determinant of this b will identically vanish
1
Since τ µ ∂µ = τ 0µ ∂µ0 = τ (µ) X(µ) = etc, the symbolic representation of vectors (and
tensors in general) surely is coordinate- and tetrad-independent, thus invariant.

6
(det bµν ≡ 0). A combination of (2.1) and (2.2) gives
T = µu ⊗ u − pw (2.3)
where we have taken w = g − u ⊗ u in analogy to (2.2).
At the same time, the term τ ⊗ τ in (2.2) is another projector, now on
τ (more frequently a simple scalar multiplication by τ , with a contraction
with the index in whose sense the projection is to be obtained, is used as
such a second projector). This is the projector on the local physical time
of the reference frame described by τ . Thus g = b + τ ⊗ τ . Let us do
a substitution of this expression in ds2 (which can be symbolically written
without using coordinates as ds2 = g(dx, dx) where dx = dxµ ∂µ , this time
dxµ representing the components of the infinitesimal vector between points
xµ and xµ + dxµ taken with respect to the vector basis ∂µ , while the above-
written expression with parentheses (dx, dx) means scalar product with the
rank 2 tensor g in the sense of both its indices; we shall sometimes use such
notations, also with one of the valences of g non-occupied). This substitution
yields ds2 = −dl2 + dt2 (dl2 := −b(dx, dx) and dt := τ · dx, both differentials
not being total ones, similar to ds; minus in the first expression corresponds
to the signature
of b). In particular, for propagation of light (ds2 = 0) this
dl
gives dt = v = 1 in every reference frames (including non-inertial ones).
µ
Considering the four-velocity u = dx ∂ of some object, we can easily find
ds µ
its three-dimensional velocity in any reference frame τ as
b(dx, ·) dt
v= ⇒ u = (τ + v) (2.4)
dt ds
where dt is non-total differential of the physical time (taken with respect
to the reference frame τ ) introduced above. In fact, this was a special case
involving projection; let us, more generally, consider now an arbitrary four-
(co)vector q; its projection onto the monad is a (frame-non-invariant) scalar,
(τ )
q := q · τ, (2.5)
thus
dt (τ )
=u, (2.6)
ds
and the projection of q onto the three-space of the reference frame is a (for-
mally) four-dimensional (co)vector,
(3)
q := b(q, ·), (2.7)

7
(3)
which is by a definition orthogonal to the monad (therefore q is in a certain
sense three-dimensional) and not changes by a repeated b-projection. (But
note that v in (2.4), also orthogonal to τ , has other structure.) Hence,
(τ ) (3)
q = q τ+ q ; (2.8)

here the factor τ in the first right-hand side term shows that a more orthodox
form of the projector onto the τ -congruence is also better logically [see τ ⊗ τ
in (2.2) already discussed above].
In (2.1) and (2.3) we had expressions of the energy-momentum tensor
of a perfect fluid given in terms of the vector u, the four-velocity of fluid.
This unit vector field is tangent to the congruence of fluid’s particles world
lines forming a timelike congruence which, naturally, can be considered as
a monad. It is clear that this is monad of the frame co-moving with the
fluid, thus only in this frame observer’s measurements will give values of the
proper mass density and proper pressure of the fluid. In all other frames
even Pascal’s fluid pressure must become anisotropic. Let us see this in more
detail. Substitute the expression (2.4) of u in terms of τ and v in (2.1); you
see that in an arbitrary reference frame the energy-momentum tensor of a
perfect fluid takes the form
 2  2  
pf (τ ) (τ )
T = u µ + u −1 p τ ⊗ τ +

2
(τ )
u (µ + p)(τ ⊗ v + v ⊗ τ + v ⊗ v) − pb. (2.9)
Here the scalar coefficient completely written inside the square brackets in
2
(τ )
the first term, represents fluid’s energy (mass) density, the next coefficient u
(µ + p) (in the second term) expresses coincidence of density of fluid’s energy
flow and of its three-momentum density (when, respectively, multiplied by τ ⊗
v and v ⊗ τ ); the last term in the same parentheses describes the anisotropic
part of pressure in the direction of fluid’s motion, while the last term (−pb)
gives isotropic part of pressure, all this is now done in τ -reference frame.
It is interesting that scalar quantities, the energy density and the pressure,
related to the co-moving frame of the fluid, both contribute to the energy
density, its flow density, and anisotropic part of the pressure, taken with
respect to non-co-moving reference frame, while the fluid remains one and the

8
same Pascal fluid. By evaluating the involved quantities one has to take into
(τ )
 factor u is always not smaller than unity
account the fact that the relativistic

(τ )
u = dt/ds = (1 − v 2 )−1/2 ≥ 1 . Though in cosmological calculations one,
quite obviously, has to use the co-moving frame of fluid modeling contents
of the universe, these considerations of transformation from one reference
frame to another are important in understanding that resulting scalar (i.e.
invariant under transformations of coordinates) quantities (mass density and
pressure of the fluid) are not invariant under a change of reference frame.
The energy-momentum tensor describes inertial and energetical proper-
ties, in our case, of a fluid. The gravitating mass or mass density follows
from participation of this tensor in Einstein’s equations, and it is strictly
shown (Mitskievich 2006b) that the latter represents a combination of iner-
tial mass and pressure. However this is completely contained in the structure
of Einstein’s equations, so that the gravitational mass has to be used only in
post-Newtonian approximations of general relativity in celestial mechanics.
Thus we have only to use the energy-momentum tensor in Einstein’s equa-
tions in our cosmological calculations. The only thing which we still need,
is to include more one concrete property of the fluid, its equation of state,
relating pressure to mass density. Here we shall use the simplest equation of
state,
p = (2k − 1)µ. (2.10)
The “fluid” with k = 1/2 (p = 0) is called incoherent dust, with k = 2/3
(thus p = µ/3, T having zero trace) is incoherent radiation, and k = 1 (p = µ,
and sound in the fluid propagates with the velocity of light in a vacuum),
is stiff matter. Below we shall see that for any value of the constant k it is
easy to find an exact solution of Einstein’s equations, even when this k has
no physical meaning.

3 Solution of Einstein’s equations for arbi-


trary k
Let us assume that the metric (1.9) is written in coordinates corresponding to
the co-moving frame of the matter filling the universe, and that this matter
can be in the large-scale described as a fluid. Then, taking the energy-
momentum tensor on the right-hand side of Einstein’s equations as (2.1) and

9
fluid’s equation of state as (2.10), we can write these equations as
G(0)(0) = −κ µ, G(1)(1) = −(2k − 1)κ µ. (3.1)
The components of conservative tensor G were already given in (1.16). But
first we combine these two equations eliminating µ and thus leaving only one
unknown function, a(t) in G:
 . "  #
2
ȧ ȧ
+ (3k − 1) − C = 0. (3.2)
a a

Of course, we choose a new unknown function, say, X(t) = ȧa = (ln a)˙ ;
then the equation reads Ẋ + (3k − 1) (X 2 − C) = 0, i.e. XdX
2 −C = −(3k − 1)dt
  √
which is equivalent to 2√1C X− dX
√ − dX
C

X+ C
= −(3k −1)dt. Now, ln X+√C
X− C
=

2 C(3k − 1)t where a new integration constant was included into t as its
translation. Then
√ √
d ln a = Xdt = C √ coth[ C(3k − 1)t]dt =
(3.3)
(3k − 1)−1 d ln sinh[ C(3k − 1)t],
and after integration
√ −1
a(t) = ρ{sinh[ C(3k − 1)t]}(3k−1) ,
with one more integration constant (ρ). We choose this constant as ρ =
√ −(3k−1)−1
C , to make the properties of the scale factor a better when C
takes all its three standard values,
( √ )(3k−1)−1
sinh[ C(3k − 1)t]
a(t) = √ . (3.4)
C

Note that for all values of k with positive pressure (and even some with not
too strong negative one) in the simplest equation of state (2.10) this solution
describes universes having an initial singularity at t = 0.
A substitution of (3.4) into the the first of two Einstein’s equations in
(3.1) [see also (1.16)] gives time-dependence of the mass density as
" √ 6k
# 3k−1
3 C
µ= √ . (3.5)
κ sinh[ C(3k − 1)t]

10
You see that at this initial singularity (t = 0) spacetime still does not exist (all
metric coefficients are equal to zero) while mass density and pressure (if we
do not consider the incoherent dust) have infinite values (at this limit). It is
interesting that the most strong singularity occurs for dust, and the weakest,
6k
for stiff matter (the power 3k−1 takes in the cases of dust, incoherent radiation
and the stiff matter values 6, 4, and 3, respectively), though, of course, the
strongest singularity takes place when k = 1/3, and it occurs not only at
t = 0, but at all finite times too, thus giving the lower limit for k. However
if one would formally consider the values of k below 1/3 (disjoint from the
physically acceptable ones by that perpetual singularity), even non-singular
cases of the ‘cosmological’ bahavour may be found.

References
Mitskievich, N.V. (1999a) Int. J. of Theor. Phys. 38, 997.

Mitskievich, N.V. (1999b) Gen. Relat. and Grav. 31, 713.

Mitskievich, N.V. (2006a) Relativistic Physics in Arbitrary Peference


Frames (New York: Nova Science Publishers).

Mitskievich, N.V. (2006b) Relativistic generalization of the inertial and


gravitational masses equivalence principle, arXiv: gr-qc/0611017 (Talk
at the 11th Marcel Grossmann Meeting, Berlin, July 2006).

Narlikar, J.V. (2002) An Introduction to Cosmology, third edition (Cam-


bridge: CUP).

11

Potrebbero piacerti anche