M.

Pettini: Introduction to Cosmology — Lecture 2

NEWTONIAN COSMOLOGY

The equations that describe the time evolution of an expanding universe

which is homogeneous and isotropic can be deduced from Newtonian dy-
namics and gravitation. Although the derivation is not strictly self-consistent
in that it requires the use of a result from general relativity (GR)—Birkhoff’s
theorem—it nevertheless provides some intuitive insights and is a valuable
first step.

First we show that the Hubble expansion is a natural property of an ex-

panding universe that obeys the cosmological principle. Referring to Fig-
ure 2.1:

Figure 2.1: All observers see galaxies expanding with the same Hubble law.

vA = H0 · rA (2.1)
and
vB = H0 · rB (2.2)
where v and r are respectively the velocity and position vectors and the
subscript 0 is used to indicate the present time.

By the rule of vector addition, the recession velocity of galaxy B as seen

by an observer on galaxy A is

vBA = vB − vA = H0 rB − H0 rA = H0 (rB − rA ) (2.3)

1
So the observer on galaxy A sees all other galaxies in the universe receding
with velocities described by the same Hubble law as on Earth.

In a homogeneous universe every particle moving with the substratum has

a purely radial velocity proportional to its distance from the observer. As
eq. 2.3 can be written for any two particles, we can change to a more
convenient coordinate system, known as comoving coordinates. These
are coordinates that are carried along with the expansion, so that we can
express the distance r as a product of the comoving distance x and a term
a(t) which is a function of time only:

rBA = a(t) · xBA (2.4)

The original r coordinate system, which does not expand, is usually known
as physical coordinates.

The term a(t) is the scale factor of the universe, and it tells us how physical
separations grow with time, since the coordinate distances x are by defini-
tion fixed. Deriving an equation for the universal expansion thus reduces
to determining a function which describes a(t).

In Newtonian cosmology this is done by considering the forces acting on

masses A, B, C, D on a sphere of radius r centred at O, as in Figure 2.2.

Figure 2.2: Birkhoff’s theorem: the force acting on A, B, C, D—which are particles
located on the surface of a sphere of radius r—is the gravitational attraction from the
matter internal to r only, acting as a point mass at O.

2
Birkhoff’s theorem states that the net gravitational effect of a uniform
external medium on a spherical cavity is zero—in other words, the force
acting on A, B, C, D is the gravitational attraction from the matter M
internal to r only, which acts as a point mass at O. We can then write the
total energy of a particle of mass m at A, B, C, D as the usual sum of
kinetic and gravitational potential energy
1 GM m 1 2 4π
U = T + V = mṙ2 − = mṙ − Gρr2 m (2.5)
2 r 2 3
where the dot denotes differentiation with respect to time, ρ is the density
of matter within the sphere of radius r, and G is Newton’s gravitational
constant.

Substituting (2.4) into (2.5) we have:

1 4π
U = mȧ2 x2 − Gρa2 x2 m (2.6)
2 3
which can be re-arranged into the familiar form of the Friedmann equa-
tion
ȧ 2 8πG kc2
!
= ρ− 2 (2.7)
a 3 a
where
2U
kc2 = − 2 (2.8)
mx
The parameter k is interesting. Note that k must be independent of x,
since the other terms in the equation are. Thus U ∝ x2 ; homogeneity
requires that U , while constant for a given particle, does change if we look
at different comoving separations x. From eq. 2.8 we can also see that
k 6= f (t), since for a given particle the total energy U is conserved and
ẋ = 0 by definition. Thus k is just a constant, unchanging with either
space or time.

An expanding universe has a unique value of k which it maintains through-

out its evolution. The value of k determines the form of this evolution. It
can be appreciated immediately from eq. 2.8 that:

• A positive k implies negative U , so that V > T in eq. 2.5—the expan-

sion will at some time t halt and reverse itself
• A negative k implies positive U , so that V < T in eq. 2.5—the expan-
sion will continue forever

3
 • If k = 0, the total energy is also U = 0 and the expansion of the
universe will slow down, but only halt at t = ∞

In the definition of eq. 2.8 k has the units of an inverse area.

Combining eqs. 2.1 and 2.4, we can write:

|ṙ| ȧ
v= r= r (2.9)
|r| a
Comparison with (2.1) identifies the Hubble parameter as:
ȧ
H= (2.10)
a
and Friedmann equation can be re-written in a form that explicitly ex-
presses the time evolution of the Hubble parameter H(t):

28πG kc2
H = ρ− 2 (2.11)
3 a
which defines a critical density today:

3H02
ρc = (2.12)
8πG

Returning to eq. 2.7, it is clear that we cannot use this equation to de-
scribe the time evolution of the scale factor of the universe, a(t), without
an additional equation describing the time evolution of the density ρ of
material in the universe.

From thermodynamics we know that:

dE + pdV = T dS (2.13)

where p is the pressure. Applying it to an expanding volume V of unit

comoving radius1 , and using E = mc2 , the energy within the volume is:
4π 3 2
E= a ρc
3
1
Beware of two more symbol ambiguities: earlier we used p to indicate momentum and V potential
energy.

4
The change of energy in a time dt is:
dE da 4π dρ
= 4πa2 ρc2 + a3 c2
dt dt 3 dt

while the change in volume with time is:

dV da
= 4πa2
dt dt

Assuming a reversible expansion, i.e. dS = 0 in eq. 2.13, we obtain:

ȧ p
!
ρ̇ + 3 ρ + 2 = 0 (2.14)
a c

Equation 2.14 is known as the fluid equation. It tells us that there are
two terms contributing to the change in density as the universe expands.
The first term in the brackets corresponds to the dilution in the density
because the volume has increased—that’s straightforward. The second
term corresponds to the loss in energy because the pressure of the material
has done work as the universe’s volume increased.

What about conservation of energy? Of course, energy is always conserved—

in this case the energy lost from the fluid via the work done has gone into
gravitational potential energy.

The term ‘pressure’ here, does not mean a pressure gradient which sup-
plies the force driving the expansion—there are no such pressure forces
in a homogeneous universe because density and pressure are the same ev-
erywhere. In cosmology the assumption is usually made that there is a
unique pressure associated with each density, so that p ≡ p(ρ). Such a
relationship is known as the equation of state. The form of the equation
of state depends on the nature of the constituent of the universe we are
considering:

• Non-relativistic matter has negligible pressure, p = 0. Examples are

galaxies (their only interaction is through gravity) and atoms in gen-
eral once the universe has expanded and cooled. Cosmologists some-
times refer to this component as ‘dust’, precisely to indicate matter
with negligible pressure.

5
 • Light, or more generally any highly relativistic particle, has an asso-
ciated pressure (radiation pressure) p = ρc2 /3.

Armed with the fluid equation, we can now derive an expression for the
acceleration of the universe, ä, as follows:

Differentiating w.r.t. time the Friedmann equation (2.7), we obtain:

ȧ aä − ȧ2 8πG kc2 ȧ
2 = ρ̇ + 2 3 (2.15)
a a2 3 a

Substituting in ρ̇ from 2.14, leads to:

!2
ä ȧ p kc2
!
− = −4πG ρ + 2 + (2.16)
a a c a2

Using again 2.7, we arrive at the acceleration equation:

ä 4πG 3p
!
=− ρ+ 2 (2.17)
a 3 c

Note that pressure here acts to increase the gravitational force, and so
further decelerates the expansion. Note also that the constant k, which is
so important in Friedmann equation, does not appear anywhere in 2.17—it
cancelled out in the derivation.

It is worthwhile considering in passing the behaviour of the Friedmann

equation in the simplest case where k = 0, for the two types of constituents
of the universe we mentioned above: (a) pressureless dust and (b) radiation.

Dust. With p = 0, the fluid equation (2.14) becomes:

ȧ
ρ̇ + 3 ρ = 0
a
which can be re-written as:
1 d
3
(ρa3 ) = 0
a dt
which implies
d
(ρa3 ) = 0
dt
6
That is, ρa3 is a constant and therefore:
1
ρ∝ (2.18)
a3

This rather obvious result tells us that the density of matter falls off in
proportion to the volume of the expanding universe. If we choose as the
unit scale factor a(t) the scale factor today, that is a0 = 1, so that physical
and comoving coordinates coincide today (recall our definition of both at
2.4), we have:
ρ = ρ0 /a3 (2.19)
and subsituting 2.19 into 2.7 with k = 0 we obtain:
8πGρ0 1
ȧ2 = (2.20)
3 a
which has the solution a ∝ t2/3 . As we have fixed a0 = 1, the full solution
is therefore
t 2/3 ρ0 ρ0 t2
!
a(t) = ; ρ(t) = 3 = 2 0 (2.21)
t0 a t

In this solution, the universe expands forever, but an ever decreasing rate:
ȧ 2
H(t) ≡ = (2.22)
a 3t

This is one of the classic cosmological solutions and is referred to as an

‘Einstein-de Sitter’ cosmology.

Notice from 2.22 that:

2 1
t0 =
3 H0
where t0 is the current age of the universe and 1/H0 is the Hubble time.
For H0 = 70 km s−1 Mpc−1 , H0−1 = 14 Gyr and t0 = 9.3 Gyr.

Radiation. Substituting the radiation pressure p = ρc2 /3 into the fluid

equation (2.14) we have:
ȧ
ρ̇ + 4 ρ = 0
a
which, following the same steps as above for the dust-dominated universe,
leads to the conclusion that:
1
ρ∝ 4 (2.23)
a
7
and
ρ0 t20
!1/2
t ρ0
a(t) = ; ρ(t) = 4 = 2 (2.24)
t0 a t

This is the second classic cosmological solution.

Note that the universe expands more slowly if it is radiation-, rather than
dust-dominated, because of the additional deceleration which the pressure
supplies (see 2.17). However, in each case the density falls off as t2 . From
2.23, we see that the radiation density drops as the fourth power of the scale
factor. Three of these powers we have already identified as the increase
in volume, while the fourth power is due to the redshift of the light [by a
factor (1 + z) = 1/a], which decreases its associated energy E = hν.

The above relations also show that when the universe is radiation domi-
nated (i.e. earliest epochs, when matter is relativistic),
1 1 1
a(t) ∝ t1/2 ; ρrad ∝ ; ρdust ∝ ∝ (2.25)
t2 a3 t3/2

which is an unstable situation since the density of radiation is falling off

faster than the density of dust.

On the other hand, once the universe becomes dust-dominated,

1 1 1
a(t) ∝ t2/3 ; ρdust ∝ ; ρrad ∝ ∝ (2.26)
t2 a4 t8/3

which is a stable situation—the dust becomes increasingly dominant over

the radiation as time progresses.