M.

Pettini: Introduction to Cosmology — Lecture 3

RELATIVISTIC COSMOLOGY

3.1 The Robertson-Walker Metric

The appearance of objects at cosmological distances is affected by the

curvature of spacetime through which light travels on its way to Earth. The
most complete description of the geometrical properties of the Universe is
provided by Einstein’s general theory of relativity. In GR, the fundamental
quantity is the metric which describes the geometry of spacetime.

Let’s look at the definition of a metric: in 3-D space we measure the

distance along a curved path P between two points using the differential
distance formula, or metric:

(d`)2 = (dx)2 + (dy)2 + (dz)2 (3.1)

and integrating along the path P (a line integral) to calculate the total
distance: Z 2q Z 2q
∆` = 2
(d`) = (dx)2 + (dy)2 + (dz)2 (3.2)
1 1

Similarly, to measure the interval along a curved wordline, W, connecting

two events in spacetime with no mass present, we use the metric for flat
spacetime:
(ds)2 = (cdt)2 − (dx)2 − (dy)2 − (dz)2 (3.3)
Integrating ds gives the total interval along the worldline W:
Z Bq Z Bq
∆s = (ds)2 = (cdt)2 − (dx)2 − (dy)2 − (dz)2 (3.4)
A A

By definition, the distance measured between two events, A and B, in a

reference frame for which they occur simultaneously (tA = tB ) is the proper
distance: q
∆L = −(∆s)2 (3.5)

Our search for a metric that describes the spacetime of a matter-filled

universe, is made easier by the cosmological principle. In a homogeneous
and isotropic universe, although the curvature of space may change with

1
time, it must have the same value everywhere at a given time since the Big
Bang.

On the surface of a sphere, curvature is defined as K ≡ 1/R2 . But a more

general expression for curvature in a 2-D space is (see Figure 3.1)
3 2πD − Cmeas
K = lim (3.6)
π D→0 D3

Figure 3.1: The circumference of a circle is equal to the radius ×2π only in a Eucledian
geometry. (Reproduced from Carroll & Ostlie’s Modern Astrophysics).

The distance between two points, P1 and P2 on the surface of a sphere is

given by (see Figure 3.2):
(d`)2 = (dD)2 + (r dφ)2 = (R dθ)2 + (r dφ)2 (3.7)
But r = R sin θ, so dr = R cos θ dθ and
dr R dr dr
R dθ = =√ 2 = q (3.8)
cos θ R − r2 1 − r2 /R2
so that:  2
2 dr  + (r dφ)2
(d`) =  q (3.9)
1 − r2 /R2
in terms of plane polar coordinates r and φ. More generally, in terms of
the curvature K of a two-dimensional surface:
!2
2 dr
(d`) = √ 2
+ (r dφ)2 (3.10)
1 − Kr
2
 Figure 3.2: (Reproduced from Carroll & Ostlie’s Modern Astrophysics).

This can be extended to 3-D by changing from polar to spherical coordi-

nates,
!2
2 dr
(d`) = √ 2
+ (r dθ)2 + (r sin θ dφ)2 (3.11)
1 − Kr
where r is now the radial coordinate. Eq. 3.11 shows the effect of the
curvature of our three-dimensional Universe on spatial distances.

The final step towards the spacetime metric involves the inclusion of time.
By distance we mean the proper distance between two spacetime events
that occur simultaneously according to an observer (eq. 3.5). In an ex-
panding universe, the position of two points must be recorded at the same
time if their separation is to have any meaning. In an isotropic, homoge-
neous universe, there is no reason why time should pass at different rates at
different locations; thus the temporal term should just be c dt. The metric
then becomes:
!2
2 2 dr
(ds) = (c dt) − √ − (r dθ)2 − (r sin θ dφ)2 (3.12)
1 − Kr2
q
and the differential proper distance is just ∆L = −(∆s)2 with dt = 0.

3
We can now change our radial coordinate to a comoving coordinate ac-
cording to eq. 2.4:
r(t) = a(t) · x
Because the expansion of the Universe affects all of its geometric proper-
ties, including its curvature, it is also useful to define the time-dependent
curvature in terms of the scale factor and a time independent constant k:
k
K(t) = (3.13)
a2 (t)
With these substitutions for r and K, we finally arrive at the important
Robertson-Walker metric:
 !2 
dx
(ds)2 = (c dt)2 − a2 (t)  √ + (x dθ)2 + (x sin θ dφ)2  (3.14)
1 − kx2
which is more usually written in the form:

dr2
 
2 2 2 2 2 2 2 
 
(ds) = (c dt) − a (t)  + r dθ + sin θ dφ (3.15)
1 − kr2

where, through an annoying change of notation, r now indicates comoving

radial distance.

Robertson and Walker independently demonstrated in the mid-1930s that

this is the most general metric possible for describing an expanding, ho-
mogeneous and isotropic universe.

4
3.2 The Friedmann Equations

The metric evolves according to Einstein’s field equations for calculating

the geometry of spacetime produced by a given distribution of mass and
energy:
8πG
Gαβ = 4 Tαβ (3.16)
c
where Tαβ is the stress-energy tensor which evaluates the effect of a given
distribution of mass and energy on the curvature of spacetime, as described
mathematically by Einstein’s tensor:
1
Gαβ = Rαβ − gαβ R (3.17)
2
where Rαβ and R are the Ricci tensor and scalar respectively. The in-
dices α, β run over the time coordinate (labelled ‘0’) and the three spatial
coordinates.

Making use of the tensor notation, one can write metric equations quite
generally:
ds2 = gαβ dxα dxβ
where gαβ is the metric tensor and summation over α and β is implied.
Orthogonal coordinate systems have diagonal metric tensors and this is
all that we need to be concerned with—the metric tensor contains all the
information about the intrinsic geometry of spacetime. The components
of the Robertson-Walker metric can be written as a diagonal matrix with
non-vanishing elements:
a2
g00 = 1, g11 = − 2
, g22 = −a2 r2 , g33 = −a2 r2 sin2 θ
1 − kr
For a comoving observer the time-time component T00 and the space-space
component T11 of the stress-energy tensor Tαβ on the right-hand side of
eq 3.16 are:
2 pa2
T00 = ρc , T11 = (3.18)
1 − kr2
where ρ and p are the mass density and the pressure respectively.

On the left-hand side of Einstein’s field equations we need G00 and G11 to
equate to T00 and T11 respectively. The result of a rather lengthy derivation
are:
G00 = 3 (ca)−2 (ȧ2 + kc2 ) (3.19)

5
 G11 = −c−2 (2aä + ȧ2 + k) (1 − kr2 )−1 (3.20)

Substituting (3.18), (3.19) and (3.20) into (3.16), we obtain two distinct
dynamical relations for the time evolution of the cosmic scale factor a(t):

!2
ȧ kc2 8π
+ = Gρ (3.21)
a a2 3

!2
ä ȧ kc2 8π
2 + + 2 = − 2 Gp (3.22)
a a a c

These equations were derived in 1922 by the Russian physicist and math-
ematician Alexandr Friedmann, seven years before Hubble’s discovery of
the universal expansion, at a time when even Einstein did not believe in his
own equations because they did not allow the Universe to be static. How-
ever, they did not gain general recognition until after Friedmann’s death,
when they were confirmed by an independent derivation in 1927 by the
Belgian cleric Georges Lemaı̂tre.

The first Friedmann equation (3.21) shows that the rate of cosmic expan-
sion, ȧ, increases with the mass density ρ of the universe. Subtracting
it from the second Friedmann equation (3.22) we obtain the acceleration
equation:
ä 4π
= − 2 G(ρc2 + 3p) (3.23)
a 3c
which shows that the acceleration of the expansion decreases with increas-
ing pressure and energy density, whether mass or radiation energy.

At the present time t0 , defining Ω0 = ρ0 /ρc (where ρc is the critical density

which we defined in eq. 2.12), the Friedmann equation (3.21) takes the
form:
8π 2
ȧ20 = Ga0 ρ0 − kc2 = H02 a20 Ω0 − kc2 (3.24)
3

This reduces to the Newtonian relation (2.7) if we make the identification:

kc2 = H02 a20 (Ω0 − 1) (3.25)

6
Equation 3.25 shows explicitly the relation between the curvature param-
eter k in the Robertson-Walker metric (3.15) and the present-day density
parameter Ω0 : to the k values of +1, 0, and −1 correspond an overcriti-
cal density Ω0 > 1 a critical density Ω0 = 1 and an undercritical density
0 < Ω0 < 1, respectively.

So, the Friedmann equations are the same, whether derived under Newto-
nian dynamics or General Relativity. It is only the interpretation of the
constant k which is different: in the Newtonian Universe k is related to the
mechanical energy of an expanding mass shell by eq. (2.8). In Einstein’s
Universe it is the present value of the curvature of the Universe (eq. 3.13
with a = 1).

3.3 The Cosmological Constant

Before the discovery of the cosmic expansion by Hubble in 1929, the uni-
verse was thought to be static. This then implies that the scale factor
a 6= f (t) but is a constant a0 , so that ȧ = ä = 0 (and the age of the
universe is infinite).

The two Friedmann equations (3.21 and 3.22) then reduce to:
kc2 8π 8π
2
= Gρ0 = − 2 Gp0 (3.26)
a 3 c
Note that since ρ0 must be a positive number, k must be +1. Note also
that this leads to the surprising result that the pressure of matter p0 is
negative!

Einstein corrected for this in 1917 by introducing a constant Lorentz-

invariant term Λgαβ into his field equations 3.16:
8πG
Gαβ − Λgαβ = Tαβ (3.27)
c4
In contrast to the two terms making up the Einstein tensor Gαβ in eq. 3.17,
the Λgαβ term does not vanish in the limit of flat spacetime.

7
With this addition, Friedmann’s equations take the form:

!2
ȧ kc2 8π Λc2
+ 2 = Gρ + (3.28)
a a 3 3

!2
ä ȧ kc2 8π
2 + + 2 = − 2 Gp + Λc2 (3.29)
a a a c

When we write the Friedmann equations in this form, we can see that
Λc2 /8πG corresponds to an energy density, the vacuum energy density,
i.e.:
8πG
Λ = ρvac 2 . (3.30)
c

Similarly, Λc4 /8πG corresponds to a pressure term.1 A positive value of

Λ corresponds to a repulsive force counteracting the conventional attrac-
tive gravitation, as wanted by Einstein. A negative Λ corresponds to an
additional attractive force.

From the standpoint of Newtonian cosmology, the additional Λ term in

Friedmann’s equations would result if we added an additional potential
energy term:
1
VΛ ≡ − Λmc2 r2 (3.31)
6
to the right-hand side of eq. 2.5, which would then become:
1 4π 1
U = T + V + VΛ = mṙ2 − Gρr2 m − Λmc2 r2 . (3.32)
2 3 6

The force due to this new potential is:

∂VΛ 1
FΛ = − r̂ = Λmc2 r r̂ (3.33)
∂r 3
which is radially outwards for Λ > 0.

1
In some texts, a factor c2 is incorporated in the definition of Λ—sometimes this is indicated with the
lower case letter λ = Λc2 . As defined here, Λ has the units of an inverse area.