Cosmology II

Hannu Kurki-Suonio
Fall 2021
Preface
In Cosmology I we discussed the universe in terms of a homogenous and isotropic approximation
to it. In Cosmology II we add the inhomogeneity and anisotropy (Chapters 8 and 9). The
mathematical background required includes Fourier analysis (taught in Fysiikan matemaattiset
menetelmät I) and spherical harmonic analysis (taught in Fysiikan matemaattiset menetelmät
II). We will take some results from Quantum Field Theory and Cosmological Perturbation
Theory, but students are not expected to have them as background – they are more advanced
courses. We begin with Inflation, but postpone the discussion of generation of perturbations
during it to after we have discussed inhomogeneity in general and its later evolution – the chapter
on Structure Formation. Thus in the Inflation chapter we still assume the homogeneous FRW
model. We end with the Cosmic Microwave Background Anisotropy, which forms an important
part of observational data in cosmology.
7 Inflation
7.1 Motivation
In Cosmology I we discussed how the universe began with a Hot Big Bang. This leaves open
the question of initial conditions – how did the Hot Big Bang begin, and why did it begin with
such a state of high density and temperature, rapid expansion, and a high level of isotropy and
homogeneity. Inflation is a scenario to address this question, at least to some extent. Inflation
is a period in the very early universe, before the events discussed in Cosmology I, when the
expansion of the universe was accelerating.
Inflation is not really a specific theory; rather it is a more general idea of a certain kind
of behavior (i.e., a “scenario”) for the universe. It is not known for sure whether inflation
occurred, but it makes a number of predictions that agree with observations. Inflation has been
more successful than competing ideas for the very early universe and it has become part of the
standard model for cosmology. The most important property of inflation is that it provides
a mechanism for generating the initial density fluctuations, the primordial perturbations, from
which the structure of the universe, stars and galaxies, grew. However, this property was
discovered later, and the original motivation for inflation was to explain the initial flatness and
homogeneity of the universe and the lack of certain relics that could have been produced at the
very high temperatures of the very early universe[1]1 . This chapter discusses inflation in the
homogeneous and isotropic approximation. Perturbations are discussed in Chapter 8.
Much of this chapter follows Chapter 3 of the book by Liddle&Lyth.[2]

7.1.1 Flatness problem

The Friedmann equation can be written as
K
Ω−1= . (1)
a2 H 2
If the universe has the critical density, Ω = 1, it stays that way (since K = 0). But if Ω "= 1, it
evolves in time. The difference Ωk = 1−Ω grows with time during both the radiation-dominated
and matter-dominated epochs. If Ωk is small, its time evolution takes the form
1
mat.dom a ∝ t2/3 , H ∝ t−1 ⇒ ∝ t1/3 ⇒ Ωk ∝ t2/3 (2)
aH
1
rad.dom a ∝ t1/2 , H ∝ t−1 ⇒ ∝ t1/2 ⇒ Ωk ∝ t . (3)
aH

Since today, and at the end of the matter-dominated epoch, Ω0 = O(1) – and it is not essential
here that Ω0 is very close to 1, it would be enough that, say, 0.1 < Ω0 < 10 – we can calculate
backwards in time to, e.g., Big Bang Nucleosynthesis (BBN) and we find that the density
parameter must have been extremely close to 1 then:

|Ω(tBBN ) − 1| = |Ωk (tBBN )| ! 10−16 (4)

Thus we get as an initial condition to Big Bang, that Ω must have been initially extremely close
to 1. The flatness problem is to explain why it was so. Otherwise, if we start the FRW universe
in a radiation-dominated state with some initial value of the density parameter Ωi not extremely
close to 1, one of two things happens:
• Ωi > 1 ⇒ the universe recollapses almost immediately
1
Guth[1] was not the first to propose a period of accelerating expansion in the early universe, but it was his
proposal that became widely known and made inflation popular.

1
7 INFLATION 2

Figure 1: The horizon problem: regions on the CMB sky separated by more than about 1◦ had not had
time to interact, yet their temperature is the same with an accuracy of ! 10−4 .

• Ωi < 1 ⇒ the universe expands very fast and cools to T < 3 K in a very short time.

Thus the flatness problem can also be called the oldness problem: why did it take so long, 14
billion years, for the universe to cool to T = 2.7 K.

Exercise: Oldness problem. Assume Ωk (T = 100 keV) = 0.1. (BBN takes place near T =
100 keV). Include just curvature and radiation (with g∗ = 3.384) in the Friedmann equation. How long
does it take for the universe to cool to T = 2.7 K? Why would inclusion of matter (with, say η = 6×10−10 ,
and ρm = 6ρb ) not change the answer?

7.1.2 Horizon problem

The horizon problem can also be called the homogeneity problem. The cosmic microwave back-
ground (CMB), which shows the universe at z = 1090 (age 370 000 years), is remarkably
isotropic, the relative temperature variations being only O(10−4 ). This implies that density
variations at that time must have been also very small, so the early universe was very homoge-
neous. Calculated according to the standard Hot Big Bang model, the horizon distance at that
time was much smaller than the part of the early universe we see in the CMB, corresponding to
only about 1◦ on the sky. Thus there could not have been any process to homogenize conditions
over scales larger than this. This implies that this level of homogeneity must have been an initial
condition.
Even the small CMB anisotropies show correlations at larger scales that 1◦ , a fact discovered
after inflation was proposed.

7.1.3 Unwanted relics

If the Hot Big Bang begins at very high T it may produce objects surviving to the present, that
are ruled out by observations.

• Gravitino. The supersymmetric partner of the graviton. m ∼ 100 GeV. They interact
very weakly (gravitational strength) ⇒ they decay late, after BBN, and ruin the success
of BBN.
7 INFLATION 3

• Magnetic monopoles. If the symmetry of a Grand Unified theory (GUT) is broken

in a spontaneous symmetry breaking phase transition, magnetic monopoles are produced.
These are point-like topological defects that are stable and very massive, m ∼ TGUT ∼
1014 GeV. Their expected number density is such that their contribution to the energy
density today ' the critical density.

• Other topological defects (cosmic strings, domain walls). These may also be produced
in a GUT phase transition, and may also be a problem, but this is model-dependent. On
the other hand, cosmic strings had been suggested as a possible explanation for the initial
density perturbations—but this scenario fell later in trouble with the observational data
(especially the anisotropy of the CMB).

These relics are produced very early, at extremely high temperatures, typically T " 1014 GeV.
From BBN, we only know that we should have standard Hot Big Bang for T ! 1 MeV.

7.1.4 What is needed

The word “problem” in the preceding is not to be taken to imply that the Hot Big Bang theory
for the early universe would be in trouble. The theory by itself just does not contain answers
to some questions one may pose about its initial conditions, for which we thus need additional
ideas. We are perfectly happy if we can produce as an “initial condition” for Big Bang a universe
with temperature 1 MeV < T < 1014 GeV, which is almost homogeneous and has Ω = 1 with
extremely high precision.

7.2 Inflation introduced

7.2.1 Accelerated expansion
Inflation is not a replacement for the Hot Big Bang, but an addition to it, occurring at very early
times (e.g., t ∼ 10−35 s), without disturbing any of its successes. Thus we have first inflation,
then Hot Big Bang; so that inflation produces the initial conditions for the Hot Big Bang.
The origin of the flatness problem is that |Ω − 1| = |K|/(aH)2 grows with time. Now
! " ! "
d d 1 d 1 −2|K|
|Ω − 1| = |K| = |K| = ä . (5)
dt dt a2 H 2 dt ȧ2 ȧ3
# $
For an expanding universe, aH = a ȧa = ȧ > 0. Thus ȧ3 > 0, and

d
|Ω − 1| > 0 ⇔ ä < 0 . (6)
dt
Thus the reason for the flatness problem is that the expansion of the universe is decelerating,
i.e., slowing down. If we had an early period in the history of the universe, where the expansion
was accelerating, it could make an initially arbitrary value of |Ω − 1| = |K|/(aH)2 very small.
Definition: Inflation = any epoch when the expansion is accelerating.

Inflation ⇔ ä > 0 (7)

Consider then the horizon problem. The horizon at photon decoupling, dphor (tdec ) is some-
where between the radiation-dominated and matter-dominated values, H −1 and 2H −1 . For
comparing sizes of regions at different times, we should use their comoving sizes, dc ≡ dp /a. We
have
1
dchor (tdec ) ∼ , (8)
adec Hdec
7 INFLATION 4

whereas the size of the observable universe today is of the order of the present Hubble length

dchor (t0 ) ∼ H0−1 . (9)

The horizon problem arises because the first is much smaller than the second,
dchor (tdec ) a0 H 0
c ∼ * 1. (10)
dhor (t0 ) adec Hdec

Thus the problem is that aH, whose inverse gives roughly the comoving size of the horizon,
decreases with time,
d d
(aH) = (ȧ) = ä < 0 . (11)
dt dt
Having a period with ä > 0 could solve the problem.
In the preceding we referred to the (comoving) horizon distance at some time t, defined as
the comoving distance light has traveled from the beginning of the universe until time t. If there
are no surprises at early times, we can calculate or estimate it; like in the preceding where we
assumed radiation-dominated or matter-dominated behavior (standard Big Bang). If we now
start adding other periods, like accelerating expansion at early times, the calculation of dhor will
depend on them. In principle, dchor (t0 ) > dchor (tdec ) always, since t0 > tdec , so (0, tdec ) ⊂ (0, t0 ).
But note that in the horizon problem, the relevant present horizon is how far we can see: the
observable universe is given just by the integrated comoving distance the photon has traveled
in the interval (tdec , t0 ), which is not affected by what happens before tdec . Thus the relevant
present horizon is still ∼ H0−1 .
What is the relation between dchor (t) and 1/(aH) for arbitrary expansion laws? Introduce
the comoving, or conformal, Hubble parameter,
1 da
H ≡ aH = ≡ ȧ , (12)
a dη
where η is the conformal time, defined by dη = dt/a. The Hubble length is
ȧ
lH ≡ H −1 , where H ≡ , (13)
a
and the comoving Hubble length is

c lH 1 1
lH ≡ = = = H−1 . (14)
a aH ȧ
Roughly speaking, H−1 gives the comoving distance light travels in a “cosmological timescale”,
i.e., the Hubble time. This statement cannot be exact, since both the comoving Hubble length
and the Hubble time change with time. However, if H−1 is increasing with time, the comoving
distances traveled at earlier “epochs” are shorter, and thus H−1 (t) is a good estimate for the to-
tal comoving distance light has traveled since the beginning of time (the horizon). On the other
hand, if H−1 is shrinking, then at earlier epochs light was traveling longer comoving distances,
and we expect the horizon at time t to be larger than H−1 . In any case

dchor (t) " H(t)−1 . (15)

Since the Hubble length is more easily “accessible” (less information needed to figure it out)
than the horizon distance it has become customary in cosmology to use the word “horizon” also
for the Hubble distance. We shall also adopt this practice. The Hubble length gives the distance
over which we have causal interaction in cosmological timescales. The comoving Hubble length
gives this distance in comoving units.
7 INFLATION 5

If aH is decreasing (Eq. 11) then H−1 increases, and vice versa.

∴ Inflation = any epoch when the comoving Hubble length is shrinking.
d −1
Inflation ⇔ H <0 (16)
dt
Thus the comoving distance over which we have causal connection is decreasing during in-
flation: causal contact to other parts of the Universe is being lost.
Inflation can be discussed either 1) in terms of physical distances or 2) in terms of comoving
distances.
1) In terms of physical distance, the distance between any two points in the Universe is in-
creasing, with an accelerating rate. The distance over which causal connection can be maintained
is increasing (much) more slowly.
2) In terms of comoving distance, i.e., viewed in comoving coordinates, the distance between
two points stays fixed; regions of the universe corresponding to present structures maintain fixed
size. From this viewpoint (the one normally adopted), the region causally connected to a given
location in the Universe is shrinking.
To connect with dynamics, look at the second Friedmann equation,
ä 4πG
=− (ρ + 3p) (17)
a 3
∴ Inflation ⇔ ρ + 3p < 0 ⇔ w < − 13 (18)
Thus inflation requires negative pressure, p < − 31 ρ (we assume ρ ≥ 0).
There is a huge class of models to realize the inflation scenario. These models rely on so-far-
unknown physics of very high energies. Some models are just “toy models”, with a hoped-for
resemblance to the actual physics of the early universe. Others are connected to proposed
extensions (like supersymmetry) to the standard model of particle physics.
The important point is that inflation makes many generic 2 predictions, i.e., predictions that
are independent of the particular model of inflation. Present observational data agrees with
these predictions. Thus it is widely believed—or considered probable—by cosmologists that
inflation indeed took place in the very early universe. There are also numerical predictions
of cosmological observables that differ from one model of inflation to another, allowing future
observations to rule out classes of such models. (Many inflation models are already ruled out.)

7.2.2 Solving the problems

Inflation can solve3 all the problems discussed in Sec. 7.1. The idea is that during inflation the
universe expands by a large factor (at least by a linear factor of something like ∼ e70 ∼ 1030
to solve the problems). This cools the universe to T ∼ 0 (if the concept of temperature is
applicable). When inflation ends, the universe is heated to a high temperature, and the usual
Hot Big Bang history follows. This heating at the end of inflation is called reheating, since
originally the thinking was that inflation started at an earlier hot epoch, but it is actually not
clear whether that was the case.
2
I was once in a conference where a speaker began his talk on inflation by promising not to use the words
“generic” or “scenario”. He failed in one but not the other.
3
This is not to be taken too rigorously. The problems are related to the question of initial conditions of
the universe at some very early time, whose physics we do not understand. Thus theorists are free to have
different views on what kind of initial conditions are “natural”. Inflation makes the flatness and horizon problems
“exponentially smaller” in some sense, but inflation still places requirements—on the level of homogeneity—for
the initial conditions before inflation, so that inflation can begin.
7 INFLATION 6

Figure 2: Solving the flatness problem. This figure is for a universe with no dark energy, where the
expansion keeps decelerating after inflation ended in the early universe. Present observational evidence
indicates that actually the expansion began accelerating again (supposedly due to the mysterious dark
energy) a few billion years ago. Thus the universe is, technically speaking, inflating again, and Ω is
again being driven towards 1. However, this current epoch of inflation is not enough to solve the flatness
problem, or the other problems, since the universe has only expanded by about a factor of 2 during it.

Solving the flatness problem: The flatness problem is solved, since during inflation

|K|
|1 − Ω| = is shrinking. (19)
H2
Thus inflation drives Ω → 1. Starting with an arbitrary Ω, inflation drives |1 − Ω| so small that,
although it has grown all the time from the end of inflation to the recent onset of dark energy
domination, it is still very small today. See Fig. 2. In fact, inflation predicts that Ω0 = 1 to high
accuracy, since it would be an unnatural coincidence for inflation to last just the right amount
so that Ω would begin to deviate from 1 just at the current epoch.4
Solving the horizon problem: The horizon problem is solved, since during inflation the
causally connected region is shrinking. It was very large before inflation; much larger than
the present horizon. Thus the present observable universe has evolved from a small patch of a
much larger causally connected region; and it is natural that the conditions were (or became)
homogeneous in that patch then. See Fig. 3.
Getting rid of relics: If unwanted relics are produced before inflation, they are diluted to
practically zero density by the huge expansion during inflation. We just have to take care they
are not produced after inflation, i.e., the reheating temperature has to be low enough. This is
an important constraint on models of inflation.
Did we really solve the problem of initial conditions? Actually solving the flatness
and horizon problems is more complicated. We discussed them in terms of a FRW universe,
which by assumption is already homogeneous. In fact, for inflation to get started, a sufficiently
large region which is not too inhomogeneous and not too curved, is needed. We shall not discuss
this in more detail, since the solution of these problems is not the most important aspect of
inflation.
If inflation happened, we expect that the early universe after inflation was very homogeneous
except for fluctuations generated during inflation and that Ω0 = 1. Thus inflation leads to
predictions that can be tested with observations. More important than flatness and homogeneity
4
Thus, if it were discovered by observations, that actually Ω0 != 1, this would be a blow to the credibility of
inflation. However, there is a version of inflation, called open inflation, for which it is natural that Ω0 < 1. The
existence of such models of inflation have led critics of inflation to complain that inflation is “unfalsifiable”—no
matter what the observation, there is a model of inflation that agrees with it. Nevertheless, most models of
inflation give the same “generic” predictions, including Ω0 = 1.
7 INFLATION 7

Figure 3: Evolution of the comoving Hubble radius (length, distance) during and after inflation
(schematic).

are the predictions inflation makes about primordial perturbations, the “seeds” for structure
formation, discussed in the next chapter.
Thus we assume that sufficient inflation has already taken place to make the universe (within
a horizon volume) flat and homogeneous, and follow the inflation in detail after that, working
in the flat FRW universe.
7 INFLATION 8

7.3 Quantum field theory for children

The theories (known and hypothetical) needed to describe the (very) early universe are quantum
field theories (QFT). The fundamental entities of these theories are fields, i.e., functions of space
and time. For each particle species, there is a corresponding field, having at least as many (real)
components ϕi as the particle has internal degrees of freedom. For example, for the photon, the
& which you are probably
corresponding field is the vector field Aµ = (A0 , A1 , A2 , A3 ) = (φ, A),
5
familiar with from electrodynamics. The photon has two internal degrees of freedom. The
larger number of components in Aµ is related to the gauge freedom of electrodynamics. Since
Aµ is a (Lorentz) vector field, it has the same number of components as there are spacetime
dimensions, but other types of fields do not have this correspondence.
In classical field theory the evolution of the field is governed by the field equation. From
the field equation one can identify a field potential, an expression in terms of the field, which
helps to understand the field dynamics. Quantizing a field theory gives a quantum field theory.
Particles are quanta of the oscillations of the field around the minimum of its potential. The
state where the field values are constant at the potential minimum is called the vacuum. Up to
now, we have described the events in the early universe in terms of the particle picture. However,
the particle picture is not fundamental, and can be used only when the fields are doing small
oscillations. For many possible events and objects in the early universe (inflation, topological
defects, spontaneous symmetry breaking phase transitions) the field behavior is different, and
we need to describe them in terms of field theory. In some of these topics classical field theory
is already sufficient for a reasonable and useful description.
In this section we discuss “low-temperature” field theory in Minkowski space, i.e., we forget
high-temperature effects and the curvature of spacetime.
The starting point in field theory is the Lagrangian density, a function of space and time,
which is a scalar quantity constructed from the fields and their derivatives:
L(ϕi , ∂µ ϕi ) . (20)
The Lagrangian density can be expressed as a sum of two parts, the kinetic term, which depends
on field derivatives (gradients), and the field potential V (ϕ1 , . . . , ϕN ) (for a theory with N fields),
which does not. This expression for the Lagrangian density as a function of the fields and their
derivatives defines the field theory, and one can derive the field equations (differential equations
governing the field evolution) and the energy-momentum tensor (energy density and pressure of
the fields) from the Lagrangian density. Usually the kinetic term has a simple form, called the
canonical kinetic term, and we assume that here. The remaining freedom in defining the field
theory is in defining the potential.
The simplest case is a theory with one scalar field ϕ, for which
L = − 21 ∂µ ϕ ∂ µ ϕ − V (ϕ) . (21)
(We use the Einstein summation convention, where a repeated index implies summation over
it, here µ = 0, 1, 2, 3. Also ∂µ ≡ ∂/∂xµ , where x0 = t. Here we are in Minkowski space and use
Cartesian coordinates, so that ∂ 0 = −∂0 and ∂ j = ∂j for j = 1, 2, 3.) We write
dV d2 V
V # (ϕ) ≡ and V ## (ϕ) ≡ . (22)
dϕ dϕ2
The field equation, which determines the classical evolution of the field, is obtained from the
Lagrangian by minimizing (or extremizing) the action
%
L d4 x , (23)

5 " are the electromagnetic scalar and vector potentials. This is a different use of the word “potential”
φ and A
" are fields.
than what we use here. In our terminology, Aµ = (φ, A)
7 INFLATION 9

which leads to the Euler–Lagrange equation

∂L ∂L
− ∂µ = 0. (24)
∂ϕi (x) ∂[∂µ ϕi (x)]

For the above scalar field we get the field equation

∂µ ∂ µ ϕ − V # (ϕ) = 0 . (25)

where ∂µ ∂ µ ϕ = −ϕ̈ + ∇2 ϕ, so that the field equation is

ϕ̈ − ∇2 ϕ = −V # (ϕ) . (26)

Here we use the overdot to denote partial derivative with respect to time: ˙ = ∂0 = ∂/∂t.
The Lagrangian also gives us the energy tensor
∂L
T µν = − ∂ ν ϕ + gµν L . (27)
∂(∂µ ϕ)

For the scalar field

&1 '
T µν = ∂ µ ϕ∂ ν ϕ − gµν ∂ρ ϕ∂ ρ ϕ + V (ϕ) . (28)
2
In particular, the energy density ρ = T 00 and pressure p = 13 (T 11 + T 22 + T 33 ) of a scalar field
are
1 2 1
ρ = ϕ̇ + (∇ϕ)2 + V (ϕ) (29)
2 2
1 1
p = ϕ̇2 − (∇ϕ)2 − V (ϕ) . (30)
2 6
(We are in Minkowski space, so that g µν = diag(−1, 1, 1, 1)). We see that the pressure due to a
scalar field may be negative. The minimum value of V (ϕ) is the vacuum energy. In principle it
could be negative, acting like a negative cosmological constant. Any other contribution to ρ is
positive. Since there is no evidence for a negative vacuum energy or cosmological constant, let
us assume that V (ϕ) ≥ 0.
If V (φ) ≡ const (typically assumed to be 0) the field equation becomes the wave equation

ϕ̈ = ∇2 ϕ , (31)

whose solutions are waves propagating at the speed of light.

For the corresponding quantum theory, the potential gives information about the masses
and interactions of the particles that are the quanta of the field oscillations. The particles
corresponding to scalar fields are spin-0 bosons. Spin- 21 particles correspond to spinor fields
and spin-1 particles to vector fields. The case V (ϕ) ≡ 0 corresponds to massless noninteracting
particles. If the potential has the form
1
V (ϕ) = m2 ϕ2 , (32)
2
the particle corresponding to the field ϕ will have mass m and it will have no interactions. In
general, the mass of the particle is given by m2 = V ## (ϕ).
Interactions between particles of two different species are due to terms in the Lagrangian
which involve both fields. For example, in the Lagrangian of quantum electrodynamics (QED)
the term
−ieψ † γ 0 γ µ Aµ ψ (33)
7 INFLATION 10

Figure 4: Potential giving rise to spontaneous symmetry breaking.

is responsible for the interaction between photons (Aµ ) and electrons (ψ). (The γ µ are Dirac
matrices). A graphical representation of this interaction is the Feynman diagram

A higher power (third or fourth) of a field, e.g.,

1 4
V (ϕ) = λϕ , (34)
4

represents self-interaction, i.e., ϕ particles interacting with each other directly (as opposed to,
e.g., electrons, who interact with each other indirectly, via photons). In QCD, gluons have this
property.
Some theories exhibit spontaneous symmetry breaking (SSB). For example, the potential
1 1
V (ϕ) = V0 − µ2 ϕ2 + λϕ4 (35)
2 4
√
has two minima, at ϕ = ±σ, where σ = µ/ λ. At low temperatures, the field is doing small
oscillations around one of these two minima (see Fig. 4). Thus the vacuum value of the field is
nonzero. If the Lagrangian has interaction terms, cϕψ 2 , with other fields ψ, these can now be
separated into a mass term, cσψ 2 , for ψ and an interaction term, by redefining the field ϕ as

ϕ = σ + ϕ̃ cϕψ 2 = cσψ 2 + cϕ̃ψ 2 .

⇒ (36)
√
Thus spontaneous symmetry breaking gives the ψ particles a mass 2cσ. This kind of a field ϕ
is called a Higgs field. In electroweak theory the fermion masses6 are due to a Higgs field.

6
The origin of neutrino masses is not clear.
7 INFLATION 11

7.4 Inflaton field

As we saw in Sec. 7.2, inflation requires negative pressure. In Chapter 4 we considered systems
of particles where interaction energies can be neglected (ideal gas approximation). For such
systems the pressure is always nonnegative. However, negative pressure is possible in systems
with attractive interactions. In the field picture, negative pressure comes from the potential
term. In many models of inflation, the inflation is caused by a scalar field. This scalar field (and
the corresponding spin-0 particle) is called the inflaton.
Historical note. The idea of scalar fields playing an important role in the very early universe was
very natural at the time inflation was proposed by Guth[1]. We already mentioned how a scalar field,
the Higgs field, is responsible for the electroweak phase transition at T ∼ 100 GeV. It is thought that
at a much higher temperature, T ∼ 1014 GeV, another spontaneous symmetry breaking phase transition
occurred, the GUT (Grand Unified Theory) phase transition, so that above this temperature the strong
and electroweak forces were unified. This GUT phase transition gives rise to the monopole problem.
Guth realized that the Higgs field associated with the GUT transition might lead the universe to “inflate”
(the term was coined by Guth), solving this monopole problem. It was soon found out, however, that
inflation based on the GUT Higgs field is not a viable inflation model, since in this model too strong
inhomogeneities were created. So the inflaton field must be some other scalar field. The supersymmetric
extensions of the standard model contain many inflaton field candidates.

During inflation the inflaton field is almost homogeneous.7 The energy density and pressure
of the inflaton are thus those of a homogeneous scalar field,
1 2
ρ= ϕ̇ + V (ϕ)
2 (37)
1
p = ϕ̇2 − V (ϕ) ,
2
where V (ϕ) ≥ 0. For the equation-of-state parameter w ≡ p/ρ we have

ϕ̇2 − 2V (ϕ) 1 − (2V /ϕ̇2 )

w= = , (38)
ϕ̇2 + 2V (ϕ̇) 1 + (2V /ϕ̇2 )
so that
−1 ≤ w ≤ 1 . (39)
If the kinetic term ϕ̇2 dominates, w ≈ 1; if the potential term V (ϕ) dominates, w ≈ −1.
For the present discussion, the potential V (ϕ) is some arbitrary non-negative function. Differ-
ent inflaton models correspond to different V (ϕ). From Eq. (37), we get the useful combinations

ρ + p = ϕ̇2
& ' (40)
ρ + 3p = 2 ϕ̇2 − V (ϕ) .

We already had the field equation for a scalar field in Minkowski space,

ϕ̈ − ∇2 ϕ = −V # (ϕ) . (41)

For the homogenous case it is just

ϕ̈ = −V # (ϕ) . (42)
We get a working mental picture of the evolution of a homogeneous field by comparing it to the
classical mechanics equation for a particle in a gravitational potential V (r) whose accelaration
7
Inflation makes the inflaton field homogenous. Again, a sufficient level of initial homogeneity of the field is
required to get inflation started. We start our discussion when a sufficient level of inflation has already taken
place to make the gradients negligible.
7 INFLATION 12

Figure 5: The inflaton and its potential.

is given by r̈ = −∇V (r). Thus we can think of the field “rolling down” its potential like a stone
rolling down a hillside, see Fig. 5; this motion is governed by Eq. (42).
We need to modify (42) for the expanding universe. We do not need to go to the GR
formulation of field theory, since the modification for the present case can be simply obtained
by sticking the ρ and p from Eq. (37) into the energy continuity equation
ρ̇ = −3H(ρ + p) . (43)
This gives
ϕ̇ϕ̈ + V # (ϕ)ϕ̇ = −3H ϕ̇2 ⇒ ϕ̈ + 3H ϕ̇ = −V # (ϕ) , (44)
the field equation for a homogeneous ϕ in an expanding (FRW) universe. We see that the effect
of expansion is to add the term 3H ϕ̇, which acts like a friction term, slowing down the evolution
of ϕ.
The condition for inflation, ρ + 3p = 2ϕ̇2 − 2V (ϕ̇) < 0, is satisfied, if
ϕ̇2 < V (ϕ) . (45)
The idea of inflation is that ϕ is initially far from the minimum of V (ϕ). The potential then
pulls ϕ towards the minimum. See Fig. 5. If the potential has a suitable (sufficiently flat) shape,
the friction term soon makes ϕ̇ small enough to satisfy Eq. (45), even if it was not satisfied
initially.
We shall also need the Friedmann equation for the flat universe,
8πG 1
H2 = ρ= 2 ρ. (46)
3 3MPl
where we have introduced the reduced Planck mass
1 1
MPl ≡ √ mPl ≡ √ = 2.4353 × 1018 GeV . (47)
8π 8πG
Inserting Eq. (37), this becomes
( )
2 1 1 2
H = 2 ϕ̇ + V (ϕ) . (48)
3MPl 2

We have ignored other components to energy density and pressure besides the inflaton.
During inflation, the inflaton ϕ moves slowly, so that the inflaton energy density, which is
dominated by V (ϕ) also changes slowly. If there are matter and radiation components to the
energy density, they decrease fast, ρ ∝ a−3 or ∝ a−4 and soon become negligible. Again, this
puts some initial conditions for inflation to get started, for the inflaton to become dominant.
But once inflation gets started, we can soon forget the other components to the universe besides
the inflaton.
7 INFLATION 13

7.5 Slow-roll inflation

The friction (expansion) term tends to slow down the evolution of ϕ, so that we may easily reach
a situation where:

ϕ̇2 * V (ϕ) (49)

|ϕ̈| * |3H ϕ̇| (50)

These are the slow-roll conditions.

If the slow-roll conditions are satisfied, we may approximate (the slow-roll approximation)
Eqs. (48) and (44) by the slow-roll equations:

V (ϕ)
H2 = 2 (51)
3MPl
3H ϕ̇ = −V # (ϕ) (52)

The shape of the potential V (ϕ) determines the slow-roll parameters:

! "
1 2 V# 2
ε(ϕ) ≡ M
2 Pl V
##
2 V
η(ϕ) ≡ MPl (53)
V

Exercise: Show that

ε * 1 and |η| * 1 ⇐ Eqs. (49) and (50) (54)

Note that the implication goes only in this direction. The conditions ε * 1 and |η| * 1
are necessary, but not sufficient for the slow-roll approximation to be valid (i.e., the slow-roll
conditions to be satisfied).
The conditions ε * 1 and |η| * 1 are just conditions on the shape of the potential, and
identify from the potential a slow-roll section, where the slow-roll approximation may be valid.
Since the initial field equation, Eq. (44) was second order, it accepts arbitrary ϕ and ϕ̇ as initial
conditions. Thus Eqs. (49) and (50) may not hold initially, even if ϕ is in the slow-roll section.
However, it turns out that the slow-roll solution, the solution of the slow-roll equations (51) and
(52), is an attractor of the full equations, (48) and (44). This means that the solution of the
full equations rapidly approaches it, starting from arbitrary initial conditions. Well, not fully
arbitrary, the initial conditions need to lie in the basin of attraction, from which they are then
attracted into the attractor. To be in the basin of attraction, means that ϕ must be in the
slow-roll section, and that if ϕ̇ is very large, ϕ needs to be deeper in the slow-roll section.
Once we have reached the attractor, where Eqs. (51) and (52) hold, ϕ̇ is determined by
ϕ (since we replaced the second-order differential equation with a first-order one). In fact
everything is determined by ϕ (assuming a known form of V (ϕ)). The value of ϕ is the single
parameter describing the state of the universe, and ϕ evolves down the potential V (ϕ) as specified
by the slow-roll equations.
This language of “attractor” and “basin of attraction” can be taken further. If the universe
(or a region of it) finds itself initially (or enters) the basin of attraction of slow-roll inflation,
meaning that: there is a sufficiently large region, where the curvature is sufficiently small, the
inflaton makes a sufficient contribution to the total energy density, the inflaton is sufficiently
homogeneous, and lies sufficiently deep in the slow-roll section, then this region begins inflating,
7 INFLATION 14

Figure 6: The potential V (ϕ) = 21 m2 ϕ2 and its two slow-roll sections.

it becomes rapidly very homogeneous and flat, all other contributions to the energy density
besides the inflaton become negligible, and the inflaton begins to follow the slow-roll solution.
Thus inflation erases all memory of the initial conditions, and we can predict the later history
of the universe just from the shape of V (ϕ) and the assumption that ϕ started out far enough
in the slow-roll part of it.
Example: The simplest model of inflation (see Fig. 6) is the one where
1 2 2
V (ϕ) = m ϕ ⇒ V $ (ϕ) = m2 ϕ , V $$ (ϕ) = m2 . (55)
2
The slow-roll parameters are
! "2 
1 2 2 
ε(ϕ) = MPl 
 ! "2
2 ϕ MPl
⇒ ε=η=2 (56)
2 2

 ϕ
η(ϕ) = MPl 2

ϕ
and the slow-roll section of the potential is given by the condition

ε, η * 1 ⇒ ϕ2 ' 2MPl
2
. (57)

7.5.1 Relation between inflation and slow roll

From the definition of the Hubble parameter,

ȧ ä ȧ2 ä
H= ⇒ Ḣ = − ⇒ = Ḣ + H 2 (58)
a a a2 a

Thus the condition for inflation is Ḣ + H 2 > 0. This would be satisfied, if Ḣ > 0, but this is
not possible here, since it would require p < −ρ, i.e., w ≡ p/ρ < −1, which is not allowed by
Eq. (37).8 Thus
Ḣ ≤ 0 (59)
and
Ḣ
Inflation ⇔ − <1 (60)
H2
8
From the Friedmann eqs.,
! "2 
ȧ 8πG K
= ρ− 2

ȧ2

a 3 a ä K

⇒ Ḣ = − 2 = −4πG(ρ + p) + 2
ä 4πG  a a a
=− (ρ + 3p)

a 3
In the above, we are assuming space is already flat, i.e., K = 0. Then Ḣ > 0 ⇒ ρ + p < 0.
7 INFLATION 15

If the slow-roll approximation is valid,

V V # ϕ̇ V # H ϕ̇ 3H ϕ̇=−V ! V #2
H2 = 2 ⇒ 2H Ḣ = 2 ⇒ H 2 Ḣ = 2 = − 2
3MPl 3MPl 6MPl 18MPl
4 ! "2
Ḣ V #2 9MPl 1 2 V#
⇒ − 2 = 2 2
= MPl =ε*1
H 18MPl V 2 V

Therefore, if the slow-roll approximation is valid, inflation is guaranteed. This is a sufficient, not
necessary condition. The above result for slow-roll inflation, −Ḣ/H 2 * 1 can also be written
as . .
. Ḣ . ȧ
. .
. .* . (61)
.H . a
During slow-roll inflation, the Hubble parameter H changes much more slowly than the scale
factor a. For a constant H, the universe expands exponentially, since
ȧ d ln a a
= = H = const ⇒ ln = H(t − t1 ) ⇒ a ∝ eHt . (62)
a dt a1
Thus, in slow-roll inflation, the universe expands “almost exponentially”.
Note that accelerated expansion, which is defined to mean that ä > 0, does not mean that the
expansion rate, as given by H, would increase. Even during inflation, Ḣ < 0, so the expansion
rate decreases. (There may be some ambiguity in what is meant by an increasing/decreasing
expansion rate. The Hubble parameter is a better quantity to be called the expansion rate than
ȧ, since the value of the latter depends on the normalization of a0 . With the normalization
a0 = 1, H = ȧ “today”.)
Sometimes it is carelessly said that inflation was a period of very rapid expansion. Rapid
compared to what? Certainly the expansion rate was larger than today, or indeed larger than
during any period after inflation (since Ḣ < 0 always). But note that in the original Hot Big
Bang picture H → ∞ (and also ȧ → ∞) as t → 0. When we replace the earliest part of Hot Big
Bang with inflation, we replace it with slower expansion, H almost constant (and ȧ becoming
smaller towards earlier times—this is what acceleration means), instead of H → ∞.
It is possible to have inflation without the slow-roll parameters being small (fast-roll in-
flation), but we will see that slow-roll inflation produces the observed primordial perturbation
spectrum naturally (unlike fast-roll inflation).

7.5.2 Models of inflation

A model of inflation9 consists of

1. a potential V (ϕ)

2. a way of ending inflation

There are two ways of ending inflation:

1. Slow-roll approximation is no more valid, as ϕ approaches the minimum of the potential

with V (ϕmin ) = 0 or very small. For a reasonable approximation we can assume inflation
ends, when ε(ϕ) = 1 or |η(ϕ)| = 1. Denote this value of the inflaton field by ϕend .

2. Extra physics intervenes to end inflation (e.g., hybrid inflation). In this case inflation may
end while the slow-roll approximation is valid.
9
There are also models of inflation which are not based on a scalar field.
7 INFLATION 16

Figure 7: Potential for small-field (a) and large-field (b) inflation. For a typical small-field model, the
entire range of ϕ shown is * MPl .

Inflation models can be divided into two classes:

1. small-field inflation, ∆ϕ < MPl in the slow-roll section
2. large-field inflation, ∆ϕ > MPl in the slow-roll section
Here ∆ϕ is the range in which ϕ varies during (the observationally relevant part of) inflation.
See Fig. 7 for typical shapes of potentials for large-field and small-field models.
Example of small-field inflation:
/ ! "4 0
λ ϕ
V (ϕ) = V0 1− + ... , (63)
4 MPl
where the omitted terms, responsible for keeping V ≥ 0 for larger ϕ are assumed negligible in the region
of interest. We assume further that the second term is small in the slow-roll section, so that we can
approximate V (ϕ) ≈ V0 except for its derivatives. The slow-roll parameters are then
! "6 ! "2
1 ϕ ϕ
ε = λ2 and η = −3λ , (64)
2 MPl MPl
so that ! "4
ε 1 ϕ
= λ * 1. (65)
|η| 6 MPl
Thus η < 0 and ε * |η|, which is typical for small-field inflation, and inflation ends when
MPl
|η| = 1 ⇒ ϕend = √ . (66)
3λ
The assumption that the second
√ term in the potential is still small at ϕend , requires that λ " 1, and thus
|η| * 1 requires ϕ * MPl / 3, so this is indeed a small-field model.
Example of large-field inflation: A simple monomial potential of the form
V (ϕ) = Aϕn (n > 1) . (67)
The slow-roll parameters are
! "2 ! "2
n2 MPl MPl
ε= and η = n(n − 1) , (68)
2 ϕ ϕ
so that η > 0 and ε and η are of similar size, typical for large-field inflation. This is a large-field model,
since ε * 1 requires ϕ2 ' 21 n2 MPl
2
.
√
For the special case of V (ϕ) = 21 m2 ϕ2 , ε = η, and inflation ends at ϕend = 2MPl . To get inflation
# $4 # $2
to end, e.g., at energy scale V (ϕend ) ≡ m2 MPl 2
= 1014 GeV , we need m = 1014 GeV /MPl ≈
4 × 109 GeV.
7 INFLATION 17

7.5.3 Exact solutions

Usually the slow-roll approximation is sufficient. It fails near the end of inflation, but this just
affects slightly our estimate of the total amount of inflation. It is much easier to solve the
slow-roll equations, (51) and (52), than the full equations, (44) and (48). However, it is useful
to have some exact solutions to the full equations, for comparison. For some special cases, exact
analytical solutions exist.
One such case is power-law inflation, where the potential is
! 1 "
2 ϕ
V (ϕ) = V0 exp − , p > 1, (69)
p MPl

where V0 and p are constants.

An exact solution for the full equations, (44) and (48), is (exercise)

a(t) ∝ tp (70)
34 5
2 V0 t
ϕ(t) = 2pMPl ln . (71)
p(3p − 1) MPl

The general solution approaches rapidly this particular solution (i.e., it is an attractor). You
can see that the expansion, a(t), is power-law, giving the model its name.
The slow-roll parameters for this model are
1 1
ε= η= , (72)
2 p
independent of ϕ. In this model inflation never ends, unless other physics intervenes.

7.6 Reheating
During inflation, practically all the energy in the universe is in the inflaton potential V (ϕ), since
the slow-roll condition says 12 ϕ̇2 * V (ϕ). When inflation ends, this energy is transferred in
the reheating process to a thermal bath of particles produced in the reheating. Thus reheating
creates, from V (ϕ), all the stuff there is in the later universe!
Note that reheating may be a misnomer, since we don’t know whether the universe was in a
thermodynamical equilibrium ever before.
In single-field models of inflation, reheating does not affect the primordial density pertur-
bations,10 except that it affects the relation of ϕk and k/H0 given in (85), i.e., how much
the distance scale of the perturbations is stretched between inflation and today (these will be
discussed later).
Reheating is important for the question of whether unwanted—or wanted—relics are pro-
duced after inflation. The reheating temperature must be high enough so that we get standard
Big Bang Nucleosynthesis (BBN) after reheating, but sufficiently low so that we do not produce
unwanted relics. The latter constraint depends on the extended theory, but it should at least
be below the GUT scale. Thus we can take that

1 MeV < Treh < 1014 GeV . (73)

10
In more complicated models of inflation, involving several fields, reheating may also change the nature of
primordial density perturbations.
7 INFLATION 18

Figure 8: After inflation, the inflaton field is left oscillating at the bottom.

Figure 9: The time evolution of ϕ as inflation ends.

7.6.1 Scalar field oscillations

After inflation, the inflaton field ϕ begins to oscillate at the bottom of the potential V (ϕ), see
Fig. 8. The inflaton field is still homogeneous, ϕ(t, &x) = ϕ(t), so it oscillates in the same phase
everywhere (we say the oscillation is coherent). The expansion time scale H −1 soon becomes
much longer than the oscillation period.
Assume the potential can be approximated as ∝ ϕ2 near the minimum of V (ϕ), so that we
have a harmonic oscillator. Write V (ϕ) = 21 m2 ϕ2 :
 
ϕ̈ + 3H ϕ̇ = −V # (ϕ)  ϕ̈ + 3H ϕ̇ = −m2 ϕ
1 become 1# 2 $
ρ = ϕ̇2 + V (ϕ)  ρ= ϕ̇ + m2 ϕ2
2 2
What is ρ(t)?
−3H ϕ̇ oscillates
:; $<
9# 1# 2 $ 3 9# :; $<
ρ̇ + 3Hρ = ϕ̇ ϕ̈ + m2 ϕ +3H · ϕ̇ + m2 ϕ2 = H m2 ϕ2 − ϕ̇2
2 2
The oscillating factor on the right hand side averages to zero over one oscillation period (in the
limit where the period is * H −1 ).
Averaging over the oscillations, we get that the long-time behavior of the energy density is

ρ̇ + 3Hρ = 0 ⇒ ρ ∝ a−3 , (74)

just like in a matter-dominated universe (we use this result in Sec. 7.7.2). The fall in the energy
density shows as a decrease of the oscillation amplitude, see Fig. 9.
7 INFLATION 19

Figure 10: Remaining number of e-foldings N (t) as a function of time.

7.6.2 Inflaton decays

Now that the inflaton field is doing small oscillations around the potential minimum, the particle
picture becomes appropriate, and we can consider the energy density ρϕ to be due to inflaton
particles. These inflatons decay into other particles, once the Hubble time (∼ the time after
inflation ended) reaches the inflaton decay time.
If the decay is slow (which is the case if the inflaton can only decay into fermions) the inflaton
energy density follows the equation

ρ̇ϕ + 3Hρϕ = −Γϕ ρϕ , (75)

where Γϕ = 1/τϕ , the decay width, is the inverse of the inflaton decay time τϕ , and the term
−Γϕ ρϕ represents energy transfer to other particles.
If the inflaton can decay into bosons, the decay may be very rapid, involving a mechanism
called parametric resonance. This kind of rapid decay is called preheating, since the bosons thus
created are far from thermal equilibrium (occupation numbers of states are huge—not possible
for fermions).

7.6.3 Thermalization
The particles produced from the inflatons will interact, create other particles through particle
reactions, and the resulting particle soup will eventually reach thermal equilibrium with some
temperature Treh . This reheating temperature is determined by the energy density ρreh at the
end of the reheating epoch:
π2 4
ρreh = g∗ (Treh )Treh . (76)
30
Necessarily ρreh < ρend (end = end of inflation). If reheating takes a long time, we may have
ρreh * ρend . After reheating, we enter the standard Hot Big Bang history of the universe.

7.7 Scales of inflation

7.7.1 Amount of inflation
During inflation, the scale factor a(t) grows by a huge factor. We define the number of e-foldings
from time t to end of inflation (tend ) by

a(tend )
N (t) ≡ ln (77)
a(t)

See Fig. 10.

As we saw in Sec. 7.5.1, a(t) changes much faster than H(t) (when the slow-roll approximation
is valid), so that the comoving Hubble length H−1 = 1/aH shrinks by almost as many e-foldings.
(a(t) grows fast, H(t) decreases slowly.)
7 INFLATION 20

We can calculate N (t) ≡ N (ϕ(t)) ≡ N (ϕ) from the shape of the potential V (ϕ) and the
value of ϕ at time t:
% tend % ϕend % ϕ
a(tend ) H slow roll 1 V
N (ϕ) ≡ ln = H(t)dt = dϕ ≈ 2 dϕ . (78)
a(t) t ϕ ϕ̇ MPl ϕend V#

where we used
da dϕ
d ln a = = Hdt = H . (79)
a ϕ̇

Example: For the simple inflation model V (ϕ) = 21 m2 ϕ2 ,

% ϕ % ϕ ! "2
1 V 1 ϕ 1 # 2 $ 1 ϕ 1
N (ϕ) = 2 dϕ = 2 = 2 ϕ − ϕend 2 = − . (80)
MPl ϕend V $ MPl ϕend 2 4MPl 4 MPl 2

The largest initial√value of ϕ we may contemplate is that which & gives the Planck
' density, V (ϕ) =
4
MPl ⇒ ϕ = 2MPl 2
/m. Starting from this value we get 12 (MPl /m)2 − 1 e-foldings of inflation.
With m = 4 × 109 GeV (see the earlier example with this model), this gives 1.85 × 1017 e-foldings, i.e.,
17 16
expansion by a factor e1.85×10 ∼ 108×10 = 1080 000 000 000 000 000 . That’s quite a lot!

7.7.2 Evolution of scales

When discussing (next chapter) evolution of density perturbations and formation of structure
in the universe, we will be interested in the history of each comoving distance scale, or each
comoving wave number k (from a Fourier expansion in comoving coordinates).

2π λ
k= , k−1 =
λ 2π
An important question is, whether a distance scale is larger or smaller than the Hubble length
at a given time.
We define a scale to be
# −1 $
• superhorizon, when k < H k > H−1

• at horizon (exiting or entering horizon), when k = H

# −1 $
• subhorizon, when k > H k < H−1

Note that large scales (large k−1 ) correspond to low k, and vice versa, although we often talk
about “scale k”. This can easily cause confusion, so watch for this, and be careful with wording!
To avoid confusion, use the words high/low instead of large/small for k. Notice also that we are
here using the word “horizon” to refer to the Hubble length.11 Recall that:

Inflation ⇒ H−1 shrinking

All other times ⇒ H−1 growing

See Fig. 11.

11
As discussed in Cosmology I, there are (at least) three different usages for the word “horizon”:
1. particle horizon
2. event horizon (not used in Cosmology I/II)
3. Hubble length
7 INFLATION 21

Figure 11: The evolution of the Hubble length, and two scales, k1−1 and k2−1 , seen in comoving coordinates
(left) and in terms of physical distance (right).

We shall later find that the amplitude of primordial density perturbations at a given comoving
scale is determined when this scale exits the horizon during inflation. The largest observable
scales, k ≈ H0 = H0 , are “at horizon” today. (Since the universe has recently began accelerating
again, these scales have just barely entered, and are actually now exiting again.)
To identify the distance scales during inflation with the corresponding distance scales in the
present universe, we need a complete history from inflation to the present. We divide it into the
following periods:

1. From the time the scale k of interest exits the horizon during inflation to the end of
inflation (tk to tend ).

2. From the end of inflation to reheating. We assume (as discussed in Sec. 7.6.1) that the
universe behaves as if matter-dominated, ρ ∝ a−3 , during this period (tend to treh ).

3. From reheating to the present time (treh to t0 ).

Consider now some scale k, which exits at t = tk , when a = ak and H = Hk

⇒ k = H k = ak H k .

To find out how large this scale is today, we relate it to the present “horizon”, i.e., the Hubble
scale:
! "1 ! "1 ! "1
k ak H k ak aend areh Hk −N (k) ρreh 3 ρr0 4 ρk 2
= = = e , (81)
H0 a0 H0 aend areh a0 H0 ρend ρreh ρcr0

where ρk ≈ V (ϕk ) ≡ Vk (since 12 ϕ̇2 * V (ϕ) during slow roll) is the energy density when scale k
exited and N (k) ≡ number of e-foldings of inflation after that. Eq. (78) allows us to relate ϕk to
N (k). The factor areh /a0 = areh is related to the change in energy density from treh → t0 . The
behavior of the total energy density as a function of a changed from the radiation-dominated
to the matter-dominated to the dark-energy-dominated era, but we can keep things simpler by
considering just the radiation component ρr , which was equal to the total energy density ρreh at
end of reheating and behaves after that as ρr ∝ a−4 . This is slightly inaccurate, since ρr ∝ a−4
7 INFLATION 22

does not take into account the change in g∗ . However, the ∝ a−4 approximation is good enough12
for us—we are making other comparable approximations also. From end of inflation to reheating
! "1
−3 aend ρreh 3
ρ∝a ⇒ = ,
areh ρend

and the ratio Hk /H0 we got from

1 1 ! "1
8πG 8πG Hk ρk 2
Hk = ρk , H0 = ρcr0 ⇒ = .
3 3 H0 ρcr0
Thus we get that
! "1/12 ! "1/4 1/4 1/4
k ρreh Vk Vk ρr0
= e−N (k) 1/2
.
H0 ρend ρend ρcr0

We can now relate N (k) to k/H0 as

1/4 1/4 1/4 1/4
k 1 ρ Vk Vk 1016 GeV · ρr0
N (k) = − ln − ln end + ln + ln + ln , (84)
H0 3 ρ1/4 ρ
1/4 1016 GeV ρ
1/2
reh end cr0

where 1016 GeV serves as a reference scale for Vk . This is roughly an upper limit to Vk due to
lack of observation of primordial gravitational waves (discussed in the next chapter). Sticking
1/4
in the known values of ρr0 = 2.4 × 10−13 GeV (assuming massless neutrinos; however, neutrino
1/4
masses will not change the result for N (ϕk )) and ρcr0 = 3.000 × 10−12 GeV · h1/2 , the last term
becomes 60.85 − ln h ≈ 61.
The final result is
1/4 1/4
k V 1 ρ 1016 GeV
N (ϕk ) = − ln + 61 + ln k1/4 − ln end − ln , (85)
H0 ρ 3 ρ1/4 V
1/4
end reh k

where the terms have been arranged so that they are all positive (when the sign in front of them
is not included). Since the potential Vk changes slowly during slow roll, the k-dependence is
dominated by the first term and the third term is small. The fourth term depends on how fast
the reheating was. If it was instantaneous, this term is zero. The last term can be large if the
inflation scale is much lower than 1016 GeV.
12
Accurately this would go as:
' (1
areh g∗s (T0 ) 3 T0
g∗s a3 T 3 = const. ⇒ = (82)
a0 g∗s (Treh ) Treh
Eq. (81) approximates this with
! "1 ' (1
ρr0 4 g∗ (T0 ) 4 T0
= (83)
ρreh g∗ (Treh ) Treh
Taking g∗s (Treh ) = g∗ (Treh ) ∼ 100, the ratio of these two becomes
1 1
(82) g∗s (T0 ) 3 3.909 3
= 1 1 ≈ 1 1 = 0.79 ∼ 1
(83) g∗ (T0 ) g∗ (Treh )
4 12 3.363 4 100 12
−1/4
Note that a ∝ ρr is a better approximation than a ∝ T −1 , since these two differ by
' (1 ! "1
g∗ (Treh ) 4 100 4
∼ ∼ 2.33 .
g∗ (T0 ) 3.363
7 INFLATION 23

For any given present scale, given as a fraction of the present Hubble distance,13 Eq. (85)
identifies the value ϕk the inflaton had, when this scale exited the horizon during inflation. The
last three terms give the dependence on the energy scales connected with inflation and reheating.
In typical inflation models, they are relatively small. Usually, the precise value of N is not that
important; we are more interested in the derivative dN/dk, or rather dϕk /dk. We can see that
typically (for high-energy-scale inflation) about 60 e-foldings of inflation occur after the largest
observable scales exit the horizon. The number of e-foldings before that can be very large (e.g.,
billions or much more), depending on the inflation model and how the inflation is assumed to
begin.

7.8 Initial conditions for inflation

Inflation provides the initial conditions for the Hot Big Bang. What about initial conditions
for inflation? As we discussed earlier, inflation erases all memory of these initial conditions,
removing this question from the reach of observational verification. However, a complete picture
of the history of the universe should also include some idea about the conditions before inflation.
To weigh how plausible inflation is as an explanation we may contemplate how easy it is for the
universe to begin inflating.
Although inflation differs radically from the other periods of the history of the universe we
have discussed, two qualitative features still hold true also during inflation: 1) the universe is
expanding and 2) the energy density is decreasing14 (although slowly during inflation).
Thus the energy density should be higher before inflation than during it or after it. Often
it is assumed that inflation begins right at the Planck scale, ρ ∼ MPl 4 , which is the limit to how

high energy densities we can extend our discussion, which is based on classical GR. Consider
one such scenario:
When ρ > MPl 4 , quantum gravitational effects should be important. We can imagine that the

universe at that time, the Planck era, is some kind of “spacetime foam”, where the spacetime
itself is subject to large quantum fluctuations. When the energy density of some region, larger
than H −1 , falls below MPl4 , spacetime in that region begins to behave in a classical manner. See

Fig. 12. The initial conditions, i.e, conditions at the time when “our universe” (referring to one
such region) emerges from the spacetime foam, are usually assumed chaotic (term due to Linde
[3], does not refer to chaos theory), i.e., ϕ takes different, random, values at different regions.
Since ρ ≥ ρϕ , and
1 1
ρϕ = ϕ̇2 + ∇ϕ2 + V (ϕ) , (86)
2 2
we must have
ϕ̇2 ! MPl
4
, ∇ϕ2 ! MPl 4
, V (ϕ) ! MPl 4
(87)
in a region for it to emerge from the spacetime foam. If the conditions are suitable such a region
may then begin to inflate. Thus inflation may begin at many different parts of the spacetime
foam. Our observable universe would be just a small part of one such region which has inflated
to a huge size.
It is also possible that during inflation, for some part of the potential, quantum fluctuations
of the inflaton (not of the spacetime) dominate over the classical evolution, pushing ϕ higher in
some regions. These regions will then expand faster, and dominate the volume. This gives rise
to eternal inflation, where, at any given time, most of the volume of the universe is inflating.
(This possibility depends on the shape of the potential.) But our observable universe would be a
13
For example, k/H0 = 10 means that we are talking about a scale corresponding to a wavelength λ, where
λ/2π is one tenth of the Hubble distance.
14
Not necessarily true in all cases, e.g., eternal inflation.
7 INFLATION 24

Figure 12: Spacetime foam and some regions emerging from it.

part of a region, where ϕ came down to a region of the potential where the quantum fluctuations
of ϕ were small and the classical behavior began to dominate and eventually inflation ended.
Thus we see that the very, very, very large scale structure of the universe may be very
complicated. But we will never discover this, since our entire observable universe is just a small
homogeneous part of a patch which inflated, and then the inflation ended in that patch. All the
observable features of the Universe can be explained in terms of what happened in this patch
during and after inflation.
These ideas of spacetime foam and eternal inflation are rather speculative and there are also
other suggestions for the initial stages of the universe.

7.9 Inflation and Big Bang

There is some confusion about what exactly is meant by the term “Big Bang”. Initially the
term was introduced to refer to the beginning of the Universe (specifically, Hoyle compared “the
hypothesis that all matter of the universe was created in one big bang at a particular time in
the remote past”[4] to his own steady-state model of the universe which required a continuous
creation of matter). Matter- and/or radiation-dominated FRW models have a beginning where
a = 0, and we can set this to correspond to the origin (t = 0) of the time coordinate, so it
became customary to refer to this t = 0 as the Big Bang. In classical GR this is a singularity
(the spacetime curvature becomes infinite) with infinite energy density. It is clear that classical
GR must break down near this singularity, so instead of singularity, there should be something
else, which is unknown. Moreover, we have well-tested physics theories only below the energy
scale of the electroweak transition. This fuzziness about the very beginning was not an obstacle
to calling it a Big Bang and thinking that it referred to a specific time, for as long as it was
thought that the expansion was slowing down, because that made the time scale of the earlier
events shorter, so that the “fuzzy part” was thought to be an extremely short time compared
to the times for which there was rigorous scientific discussion.
Inflation changed this. In inflation the expansion is accelerating and the earlier times are
no longer negligibly short compared to the time scales under discussion (e.g., of the late part of
inflation during which the observable scales exit the horizon). In fact, we have no upper limit
on how long inflation may have lasted, and no way to know what happened before inflation.
Also, the conditions during inflation are very different from the hot dense particle soup that
was thought to be created in the Big Bang. When Guth proposed inflation, he had in mind the
earlier idea of Big Bang, where at first the temperature and density were extremely high so that
cooling lead to a GUT transition giving rise to inflation. Thus he still had a “Big Bang” before
inflation. This order of events has persisted in many popular presentations of the early universe.
However, the modern view is that this is not correct. We have no knowledge of what happened
before inflation, no evidence of an initial singularity or a Planck epoch, and thus we cannot
place with confidence anything before inflation. If we go back to Hoyle’s original phrase about
REFERENCES 25

creation of all matter, then in the inflation scenario the “Big Bang” would be the reheating. On
the other hand, we cannot be certain that inflation really happened, it is just a favorite scenario,
not something proved beyond doubt. The clear evidence about “Big Bang” is the abundances of
light element isotopes, which tell us about big bang nucleosynthesis, and the cosmic microwave
background, which tell us about recombination and photon decoupling. When we say that we
know that there was a Big Bang in the early universe, we really mean that we know that these
events took place, in a way that we described in Cosmology I. Thus my personal favorite for
the modern meaning of “Big Bang” is that it refers, not to a specific instant of time, but to
this epoch in the early universe, ending at photon decoupling, when the universe was filled with
an almost homogeneous hot soup of interacting particles. Not all cosmologists use the term
this way. But in any case, within modern cosmology, “Big Bang” can sensibly refer only to
something after inflation, not before it.

References
[1] A.H. Guth, Inflationary universe: A possible solution to the horizon and flatness problems,
Phys. Rev. D 23, 347 (1981)

[2] A.R. Liddle and D.H. Lyth: Cosmological Inflation and Large-Scale Structure (Cambridge
University Press 2000)

[3] A.D. Linde, Chaotic Inflation, Phys. Lett. 129B, 177 (1983)

[4] H. Kragh, Big Bang: the etymology of a name, Astronomy & Geophysics, 54, Issue 2, p
2.28–30 (2013), https://doi.org/10.1093/astrogeo/att035
8 Structure Formation
Up to this point we have discussed the universe in terms of a homogeneous and isotropic model
(which we shall now refer to as the “unperturbed” or the “background” universe). Clearly the
universe is today rather inhomogeneous. By structure formation we mean the generation and
evolution of this inhomogeneity. We are here interested in distance scales from galaxy size to the
size of the whole observable universe. The structure is manifested in the existence of luminous
galaxies and in their uneven distribution, their clustering. This is the obvious inhomogeneity,
but we understand it reflects a density inhomogeneity also in other, nonluminous, components
of the universe, especially the cold dark matter. The structure has formed by gravitational
amplification of a small primordial inhomogeneity. There are thus two parts to the theory of
structure formation:

1) The generation of this primordial inhomogeneity, “the seeds of galaxies”. This is the more
speculative part of structure formation theory. We cannot claim that we know how this
primordial inhomogeneity came about, but we have a good candidate scenario, inflation,
whose predictions agree with the present observational data, and can be tested more
thoroughly by future observations. In inflation, the structure originates from quantum
fluctuations of the inflaton field ϕ near the time the scale in question exits the horizon.

2) The growth of this small inhomogeneity into the present observable structure of the uni-
verse. This part is less speculative, since we have a well established theory of gravity,
general relativity. However, there is uncertainty in this part too, since we do not know the
precise nature of the dominant components to the energy density of the universe, the dark
matter and the dark energy. The gravitational growth depends on the equations of state
and the streaming lengths (particle mean free path between interactions) of these density
components. Besides gravity, the growth is affected by pressure forces.

We shall do the second part first. But before that we discuss statistical measures of inho-
mogeneity: correlation functions and power spectra.

8.1 Inhomogeneity
We write all our inhomogeneous quantities as a sum of a homogeneous background value, and
a perturbation, the deviation from the background value. For example, for energy density and
pressure we write

ρ(t, x) = ρ̄(t) + δρ(t, x)

p(t, x) = p̄(t) + δp(t, x) , (1)

where ρ̄ and p̄ are the background density and pressure, x is the comoving 3D space coordinate,
and δρ and δp are the density and pressure perturbations. We further define the relative density
perturbation
δρ(t, x)
δ(t, x) ≡ . (2)
ρ̄(t)
Since ρ ≥ 0, necessarily δ ≥ −1. These quantities can be defined separately for different
components to the energy density, e.g., matter, radiation, and dark energy. Perturbations in
dark energy are expected to be small, and if it is just vacuum energy, it has no perturbations.
When we discuss the later history of the universe, the main interest is in the matter density
perturbation,
δρm (t, x)
δm (t, x) ≡ , (3)
ρ̄m (t)

26
8 STRUCTURE FORMATION 27

and then we will often write just δ for δm .

We do the split into the background and perturbation so that the background is equal to the
mean (volume average) of the full quantity. An important question is, whether the ρ̄(t) and p̄(t)
defined this way correspond to a (homogeneous and isotropic) solution of General Relativity,
i.e., an FRW universe. We expect the exact answer to be negative, since GR is a nonlinear
theory, so that perturbations affect the evolution of the mean. This effect is called backreaction.
However, if the perturbations are small, we can make an approximation, where we drop from
our equations all those terms which contain a product of two or more perturbations, as these
are “higher-order” small. The resulting approximate theory is called first-order perturbation
theory or linear perturbation theory. As the second name implies, the theory is now linear in
the perturbations, meaning that the effect of overdensities cancel the effect of underdensities on,
e.g., the average expansion rate. In this case the mean values evolve just like they would in the
absence of perturbations.
While the perturbations at large scales have remained small, during the later history of
the universe the perturbations have grown large at smaller scales. How big is the effect of
backreaction, is an open research question in cosmology, since the calculations are difficult, but
a common view is that the effect is small compared to the present accuracy of observations.
For this course, we adopt this view, and assume that the background universe simultaneously
represents a FRW universe (“the universe we would have if we did not have the perturbations”)
and the mean values of the quantities in the true universe at each time t.
Moreover, in Cosmology II we shall (mostly) assume that the background universe is flat
(K = 0).

8.1.1 Statistical homogeneity and isotropy

We assume that the origin of the perturbations is some random process in the early universe.
Thus over- (δ > 0) and underdensities (δ < 0) occur at randomly determined locations and we
cannot expect to theoretically predict the values of δ(t, x) for particular locations x. Instead,
we can expect theory to predict statistical properties of the inhomogeneity field δ(t, x). The
statistical properties are typically defined as averages of some quantities. We will deal with two
kind of averages: volume average and ensemble average; the ensemble average is a theoretical
concept, whereas the volume average is more observationally oriented.
We denote the volume average of some quantity f (x) with the overbar, f¯, and it is defined
as !
¯ 1
f≡ d3 xf (x) . (4)
V V
The integration volume V in question will depend on the situation.
For the ensemble average we assume that our universe is just one of an ensemble of an infinite
number of possible universes (realizations of the random process) that could have resulted from
the random process producing the initial perturbations. To know the random process means to
know the probability distribution Prob(γ) of the quantities γ produced by it. (At this stage we
use the abstract notation of γ to denote the infinite number of these quantities. They could be
the generated initial density perturbations at all locations, δ(x), or the corresponding Fourier
coefficients δk . We will be more explicit later.) The ensemble average of a quantity f depending
on these quantities γ as f (γ) is denoted by "f # and defined as the (possibly infinite-dimensional)
integral !
"f # ≡ dγProb(γ)f (γ) . (5)

Here f could be, e.g., the value of ρ(x) at some location x. The ensemble average is also called
the expectation value. Thus the ensemble represents a probability distribution of universes. A
8 STRUCTURE FORMATION 28

cosmological theory predicts such a probability distribution, but it does not predict in which
realization from this distribution we live in. Thus the theoretical properties of the universe
we will discuss (e.g., statistical homogeneity and isotropy, and ergodicity, see below) will be
properties of this ensemble.
We now make the assumption that, although the universe is inhomogeneous, it is statistically
homogeneous and isotropic. This is the second version of the Cosmological Principle. Statistical
homogeneity means that the expectation value !f (x)" must be the same at all x, and thus we
can write it as !f ". Statistical isotropy means that for quantities which involve a direction,
the statistical properties are independent of the direction. For example, for vector quantities
v, all directions must be equally probable. This implies that !v" = 0. The assumption of
statistical homogeneity and isotropy is justified by inflation: inflation makes the background
universe homogeneous and isotropic so that the external conditions for quantum fluctuations
are everywhere the same.
If the theoretical properties of the universe are those of an ensemble, and we can only
observe one universe from that ensemble, how can we compare theory and observation? It
seems reasonable that the statistics we get by comparing different parts of the universe should
be similar to the statistics of a given part of the universe over different realizations, i.e., that
they provide a fair sample of the probability distribution. This is called ergodicity. Fields f (x)
that satisfy
f¯ = !f " (6)
for an infinite volume V (for f¯) and an arbitrary location x (for !f ") are called ergodic. We
assume that cosmological perturbations are ergodic. The equality does not hold for a finite
volume V ; the difference is called sample variance or cosmic variance. The larger the volume,
the smaller is the difference. Since cosmological theory predicts !f ", whereas observations probe
f¯ for a limited volume, cosmic variance limits how accurately we can compare theory with
observations.1

8.1.2 Density autocorrelation function

From ergodicity,
!ρ" = ρ̄ ⇒ !δρ" = 0 and !δ" = 0 . (7)
Thus we cannot use !δ" as a measure of the inhomogeneity. Instead we can use the square of δ,
which is necessarily nonnegative everywhere, so it cannot average out like δ did. Its expectation
value
!δρ2 "
!δ2 " = (8)
ρ̄2
is the variance of the density perturbation, and the square root of the variance,
!
δrms ≡ !δ2 " (9)
the root-mean-square (rms) density perturbation, is a typical expected absolute value of δ at an
arbitrary location.2 It tells us about how strong the inhomogeneity is, but nothing about the
shapes or sizes of the inhomogeneities. To get more information, we introduce the correlation
function ξ.
We define the density 2-point autocorrelation function (often called just correlation function)
as
ξ(x1 , x2 ) ≡ !δ(x1 )δ(x2 )" . (10)
1
Another notation I will use for volume average is f!, for smaller volumes, e.g., the volume observed in a galaxy
survey. I try to reserve f¯ for situations where we can assume f¯ = !f ", whereas cosmic variance is the difference
between f! and !f ".
2
In other words, δrms is the standard deviation of ρ/ρ̄.
8 STRUCTURE FORMATION 29

Figure 1: The 2-point correlation function ξ(r) from galaxy surveys. Left: Small scales shown in a
log-log plot. The circles with error bars show the observational determination from the APM galaxy
survey [1]. The different lines are theoretical predictions by [2] (this is Fig. 9 from [2]). Right: Large
scales shown in a linear plot. Red circles with error bars show the observational determination from the
CMASS Data Release 9 (DR9) sample of the Baryonic Oscillation Spectroscopic Survey (BOSS). The
dashed line is a theoretical prediction from the ΛCDM model. The bump near 100 h−1Mpc is the baryon
acoustic oscillation (BAO) peak that will be discussed later. This is Fig. 2a from [3].

It is positive if the density perturbation is expected to have the same sign at both x1 and x2 ,
and negative for an overdensity at one and underdensity at the other. Thus it probes how
density perturbations at different locations are correlated with each other. Due to statistical
homogeneity, ξ(x1 , x2 ) can only depend on the separation r ≡ x2 − x1 , so we redefine ξ as

ξ(r) ≡ #δ(x)δ(x + r)$ . (11)

From statistical isotropy, ξ(r) is independent of direction, i.e., spherically symmetric (isotropic),

ξ(r) = ξ(r) . (12)

We will have use for both the 3D, ξ(r), and 1D, ξ(r), versions. The correlation function is large
and positive for r smaller than the size of a typical over- or underdense region, and becomes
small for larger separations.
The correlation function at zero separation gives the variance of the density perturbation,

#δ2 $ ≡ #δ(x)δ(x)$ ≡ ξ(0) . (13)

! for a single realization as a volume average,
We can also define a correlation function ξ(r)
"
! ≡ 1
ξ(r) d3 x δ(x)δ(x + r) . (14)
V
Integrating over r and assuming periodic boundary conditions3 we get the integral constraint
" " " "
! 1 1
d3 r ξ(r) = d3 rd3 x δ(x)δ(x + r) = d3 x δ(x) d3 r δ(x + r) = 0 , (16)
V V
3
The other option is not to use periodic boundary conditions, but to understand the integral in (14) to go over
only those x, for which both x and x + r ∈ V . This is what we have to do when V refers to an actual survey.
8 STRUCTURE FORMATION 30

!
since the latter integral is δ̄ = 0. Since ξ(r) = !ξ(r)" the integral constraint applies to it likewise.
Therefore ξ(r) must become negative at some point, so that at such a distance from an overdense
region we are more likely to find an underdense region. Going to ever larger separations, ξ as
a function of r may oscillate around zero, the oscillation becoming ever smaller in amplitude.
Most of the interest in ξ(r) is for the small r within the initial positive region.

8.1.3 Fourier space

The evolution of perturbations is best discussed in Fourier space. Fourier analysis is a method
for separating out different distance scales, so that the dependence of the physics on distance
scale becomes clear and easy to handle.
For mathematical convenience, we assume the observable part of the universe lies within
a fiducial cubic box, volume V = L3 , with periodic boundary conditions. This box may be
assumed to be much larger than the region of interest, so that these boundary conditions should
have no effect. Since the infinite universe is now assumed periodic, the volume average over
the infinite universe will be equal to the volume average over the fiducial box. Thus also the
ergodicity assumption requires the fiducial volume to be large, so that it can provide a fair
sample of the ensemble.
We can now expand any function of space f (x) as a Fourier series
"
f (x) = fk eik·x , (17)
k

where the wave vectors k = (k1 , k2 , k3 ) take values

2π
ki = ni , ni = 0, ±1, ±2, . . . (18)
L
The Fourier coefficients fk are obtained as
#
1
fk = f (x)e−ik·x d3 x . (19)
V V
If f (x) is a perturbation so that its mean value vanishes, then the term k = 0 does not occur.
The Fourier coefficients are complex numbers even though we are dealing with real quantities
f (x). From the reality of f (x) follows that

f−k = fk∗ . (20)

This means that for each pair of terms fk eik·x + f−k e−ik·x the imaginary parts cancel. The real
part of fk eik·x is $ %
# fk eik·x = #fk cos k · x − %fk sin k · x . (21)

To visualize a Fourier component fk eik·x one may thus visualize just this real part, which is a
sinusoidal plane wave in the k direction.
The double integral in (16) then goes over all pairs (x1 , x2 ) in V and can be written as
! !
1
d3 x1 δ(x1 ) d3 x2 δ(x2 ) = 0 · 0 . (15)
V V V

For an actual survey, V is not really large enough to make cosmic variance negligible. This has two effects: First,
the integral does not extend to large enough separations r to capture the full ξ(r), so that typically negative ξ(r)
values at large separations r are missed. This would tend to make the integral positive. However, for an actual
survey, also the mean density has to be estimated from the " survey, so that δ will actually refer to the deviation
from the mean density of the survey, which again forces V d3 x δ(x) = 0. Thus this forced integral constraint
makes the survey typically underestimate the true correlation function.
8 STRUCTURE FORMATION 31

The Fourier expansion works only if the background universe is flat, although it can be used
as an approximation in open and closed universes,4 if the region of interest is much smaller than
the curvature radius.
The separation of neighboring ki values is ∆ki = 2π/L, so we can write

! " #3 $
ik·x L 1
f (x) = fk e ∆k1 ∆k2 ∆k3 ≈ f (k)eik·x d3 k , (22)
2π (2π)3
k

where
f (k) ≡ L3 fk . (23)
replacing the Fourier series with the Fourier integral. The size of the Fourier coefficients depends
on the fiducial volume V – increasing V tends to make the fk smaller to compensate for the
denser sampling of k in Fourier space.
In the limit V → ∞, the approximation in (22) becomes exact, and we have the Fourier
transform pair
$
1
f (x) = f (k)eik·x d3 k
(2π)3
$
f (k) = f (x)e−ik·x d3 x . (24)

Note that this assumes that the integrals converge, which requires that f (x) → 0 for |x| → ∞.
Thus we use only the Fourier series for, e.g., δ(x), but for, e.g., the correlation function ξ(x) the
Fourier transform is appropriate.
Even with a finite V we can use the Fourier integral as an approximation. Often it is
conceptually simpler to work first with the Fourier series (so that one can, e.g., use the Kronecker
delta δkk! instead of the Dirac delta function δD (k − k" )), replacing it with the integral in the
end, when it needs to be calculated. The recipe for going from the series to the integral is
" #3 ! $
2π
→ d3 k
L
k
L3 f k → f (k) (25)
" #3
L 3
δkk! → δD (k − k" ) .
2π

so that, e.g., $
! 1
ik·x
fk e → f (k)eik·x d3 k . (26)
(2π)3
k

8.1.4 Power spectrum

We now expand the density perturbation as a Fourier series
!
δ(x) = δk eik·x , (27)
k

with $
1
δk = δ(x)e−ik·x d3 x (28)
V V
4
An exact treatment in open and closed universes requires expansion in terms of suitable other functions
instead of the plane waves eik·x .
8 STRUCTURE FORMATION 32

and δ−k = δk∗ . Note that

!δ(x)" = 0 ⇒ !δk " = 0 . (29)
In analogy with the correlation function ξ(x, x# ), we may ask what is the corresponding
correlation in Fourier space, !δk∗ δk! ". Note that due to the mathematics of complex numbers,
correlations of Fourier coefficients are defined with the complex conjugate ∗ . This way the
correlation of δk with itself, !δk∗ δk " = !|δk |2 " is a real (and nonnegative) quantity, the expectation
value of the absolute value (modulus) of δk squared, i.e., the variance of δk . Calculating
! !
∗ 1 3 ik·x ! !
!δk δk! " = 2
d xe d3 x# e−ik ·x !δ(x)δ(x# )"
V
! !
1 3 ik·x !
= 2
d xe d3 re−ik ·(x+r) !δ(x)δ(x + r)"
V
! !
1 3 −ik! ·r !
= 2
d re ξ(r) d3 xei(k−k )·x
V
!
1 1
= δkk! d3 re−ik·r ξ(r) ≡ δkk! P (k) , (30)
V V

where we used !δ(x)δ(x + r)" = ξ(r), i.e., independent of x, which results from statistical
homogeneity, and the orthogonality of plane waves
!
!
d3 xei(k−k )·x = V δkk! → (2π)3 δD
3
(k − k# ) . (31)

Note that here δkk! is the Kronecker delta, 1 for k = k# , 0 otherwise – nothing to do with the
density perturbation! In the limit V → ∞ we get the Dirac delta function δD 3 (k − k# ).

Written in terms of δ(k) = V δk , the result (30) reads as

!δ(k)∗ δ(k# )" = V δkk! P (k) → (2π)3 δD

3
(k − k# )P (k) , (32)

Thus, from statistical homogeneity follows that the Fourier coefficients δk are uncorrelated.
The quantity !
2
P (k) ≡ V !|δk | " = d3 r e−ik·r ξ(r) , (33)

which gives the variance of δk , is called the power spectrum of δ(x). Since the correlation
function → 0 for large separations, we can replace the integration volume V in (33) with an
infinite volume.5 We see that the power spectrum is the 3D Fourier transform of ξ(r), and
therefore also !
1
ξ(r) = d3 k eik·r P (k) . (34)
(2π)3
Unlike the correlation function, the power spectrum P (k) is positive everywhere. Perturbations
at large distance scales are more commonly discussed in terms of P (k) than ξ(r).
From statistical isotropy

ξ(r) = ξ(r) ⇒ P (k) = P (k) (35)

(the 3D Fourier transform of a spherically symmetric function is also spherically symmetric),

so that the variance of δk depends only on the magnitude k of the wave vector k, i.e., on
the corresponding distance scale. Using spherical coordinates and doing the angular integrals
5
We want to avoid discussing ξ(r) for r ≥ L, since the artificially assumed periodicity would cause artifacts in
the behavior of ξ(r) at such separations. Thus L is assumed so large that ξ is completely negligible at such huge
separations.
8 STRUCTURE FORMATION 33

we obtain (exercise) the relation between the 1D correlation function ξ(r) and the 1D power
spectrum P (k),
! ∞
sin kr
P (k) = ξ(r) 4πr 2 dr
0 kr
! ∞
1 sin kr
ξ(r) = P (k) 4πk2 dk , (36)
(2π)3 0 kr

For the density variance we get

! ∞ ! ∞ ! ∞
2 1 2 1 3
!δ " ≡ ξ(0) = P (k)4πk dk = k P (k)d ln k ≡ P(k)d ln k . (37)
(2π)3 0 2π 2 −∞ −∞

where we have defined

k3
P(k) ≡ P (k) . (38)
2π 2
Another common notation for P(k) is ∆2 (k). The word “power spectrum” is used to refer to
both P (k) and P(k). Of these two, P(k) has the more obvious physical meaning: it gives the
contribution of a logarithmic interval of scales, i.e., from k to ek, to the density variance. P(k)
is dimensionless, whereas P (k) has the dimension of Mpc3 (when discussing observed values, it
is usually given in units of h−3 Mpc3 as distance determinations are proportional to the Hubble
constant).

" as the volume average of δ(x)δ(x + r), i.e., integrate x over the box V with
Exercise: Define ξ(r)
periodic boundary conditions, and show that
!
" V
ξ(r) = d3 k |δk |2 eik·r , (39)
(2π)3

for a single realization. Note that here we do not need any statistical assumptions (like statistical
homogeneity or ergodicity). Contrast this result with (34).

8.1.5 Scales of interest and window functions

In (37) we integrated over all scales, from the infinitely large (k = 0 and ln k = −∞) to the
infinitely small (k = ∞ and ln k = ∞) to get the density variance.6 Perhaps this is not really
what we want. The average matter density today is 3 × 10−27 kg/m3 . The density of the Earth
is 5.5 × 103 kg/m3 and that of an atomic nucleus 2 × 1017 kg/m3 , corresponding to δ ≈ 2 × 1030
and δ ≈ 1044 . Probing the density of the universe at such small scales finds a huge variance in
it, but this is no longer the topic of cosmology – we are not interested here in planetary science
or nuclear physics.
Even the study of the structure of individual galaxies is not considered to belong to cos-
mology, so the smallest (comoving) scale of cosmological interest, at least when we discuss the
present universe,7 is that of a typical separation between neighboring galaxies, of the order of
1 Mpc.
To exclude scales smaller than R (r < R or k > R−1 ) we filter the density field with a
window function. This can be done in k-space or x-space.
6
Note that large scales correspond to small k and vice versa. To avoid confusion, it is better to use the words
low and high for k, so that large scales correspond to low k, and small scales correspond to high k.
7
In early universe cosmology we may study events, or possible events, related to also smaller comoving scales.
8 STRUCTURE FORMATION 34

Figure 2: The 3D window functions W (r), top-hat (green), Gaussian (red), and k (blue), for R = 1.

The filtering in x-space is done by convolution. We introduce a (usually spherically sym-

metric) window function W (r) such that W (r) is relatively large for |r| ≤ R and W ∼ 0 for
|r| # R. We use normalization !
d3 r W (r) = 1 (40)

and define the filtered density field

!
δ(x, R) ≡ (δ ∗ W )(x) ≡ d3 x! δ(x! )W (x! − x) . (41)

The simplest window function is the top-hat window function

" #
4π 3 −1
WT (r) ≡ R for |r| ≤ R (42)
3
and WT (r) = 0 elsewhere, i.e., δ(x) is filtered by replacing it with its mean value within the
distance R. Mathematically more convenient is the Gaussian window function
1 1 2 /R2
WG (r) ≡ e− 2 |r| . (43)
(2π)3/2 R3
By the convolution theorem, the filtering in Fourier space becomes just multiplication:
δ(k, R) = δ(k)W (k) , (44)
where W (k) is the Fourier transform of the window function. For WT and WG we have (exer-
cise)
3(sin kR − kR cos kR)
WT (k) =
(kR)3
1 2
WG (k) = e− 2 (kR) . (45)
8 STRUCTURE FORMATION 35

We can also define the k-space top-hat window function

Wk (k) ≡ 1 for k ≤ 1/R (46)

and Wk (k) = 0 elsewhere. In x-space this becomes (exercise)

1 sin y − y cos y
Wk (r) = , where y ≡ |r|/R . (47)
2π 2 R3 y3
The variance of the filtered density field (Exercise: derive the second equalities of both
expressions)
!
2 2 1
σ (R) ≡ $δ(x, R) % = d3 k P (k)|W (k)|2
(2π)3
! !
2 1 3 2 V
σ
" (R) ≡ d x δ(x, R) = d3 k |δk |2 |W (k)|2 . (48)
V (2π)3

is a measure of the inhomogeneity at scale R. For the k-space top-hat window this becomes
simply
! R−1 ! − ln R
2 1 2
σ (R) = 4πk P (k)dk = P(k)d ln k . (49)
(2π)3 0 −∞

One may also ask, whether scales larger than the observed universe (the lower limit k = 0
or ln k = −∞ in the k integrals) are relevant, since we cannot observe the inhomogeneity at
such scales. Due to such very-large-scale inhomogeneities, the average density in the observed
universe may deviate from the average density of the entire universe. Inhomogeneities at scales
somewhat larger than the observed universe could appear as an anisotropy in the observed
universe. The importance of such large scales depends on how strong the inhomogeneities at
these scales are, i.e., how the power spectrum behaves as k → 0. The present understanding,
supported by observations, is that the contribution of such large scales is small.8

8.1.6 Power-law spectra

We have observational information and theoretical predictions for ξ(r) and P (k) for a wide
range of scales. (We will discuss the theory in detail later.) For certain intervals, they can be
approximated by a power-law form,

ξ(r) ∝ r −γ or P (k) ∝ kn . (50)

When plotted on a log-log scale, such functions appear as straight lines with slope −γ and n.
The proportionality constant can be given in terms of a reference scale. For ξ(r) we usually
choose the scale r0 where ξ(r0 ) = 1, so that
# $−γ
r
ξ(r) = . (51)
r0

For P (k) we may write

# $n # $n+3
k k
P (k) = A2 or P(k) = A2 , (52)
kp kp
8
The inflation scenario predicts that the universe outside the current horizon is similar to the observable
universe up to distances very much larger than the current horizon distance. However at “very very far away”,
beyond the pre-inflation horizon, the universe may be quite different. For this section, we exclude such “very very
far” regions from the concept of “universe”. (They are “other universes” in a “multiverse”.)
8 STRUCTURE FORMATION 36

Figure 3: Top panel: The correlation function from the 2dFGRS galaxy survey in log-log scale. The
dashed line is the best-fit power law (r0 = 5.05 h−1 Mpc, γ = 1.67). The inset shows the same in linear
scale. Bottom panel: 2dFGRS data (solid circles with error bars) divided by the power-law fit. The solid
line is the result from the APM survey and the dotted line from an N-body simulation. This is Fig. 11
from [4].

!
where kp is called a pivot scale (whose choice depends on the application) and A ≡
! P (kp ) or
P(kp ) is the amplitude of the power spectrum at the pivot scale.
We define the spectral index n(k) as
d ln P
n(k) ≡ . (53)
d ln k
It gives the slope of P (k) on a log-log plot. For a power-law P (k), n(k) = const = n. We can
study power-law ξ(r) and P (k) as a playground to get a feeling what different values of the
spectral index mean, and, e.g., how γ and n are related.9
The Fourier transform of a power law is a power law. For the correlation function of (51) we
get (exercise)
4π (2 − γ)π
P (k) = 3
Γ(2 − γ) sin (kr0 )γ
k 2
2 (2 − γ)π
P(k) = Γ(2 − γ) sin (kr0 )γ (54)
π 2
for 1 < γ < 2 or 2 < γ < 3, and
2π 2
P (k) = (kr0 )2
k3
P(k) = (kr0 )2 (55)

for γ = 2. Thus
n = γ −3 for 1 < γ < 3, i.e., −2 < n < 0 . (56)
9
In reality the spectral index is very different at small scales than at large scales. Observationally, for small
scales, γ ∼ 1.8, and for large scales, n ∼ 1. We discuss this later.
8 STRUCTURE FORMATION 37

Figure 4: The ratio of σ2 (R) to P(R−1 ) in the case of a power-law spectrum P(k) ∝ k n+3 for the three
different window functions: top-hat (green), Gaussian (red), and k (blue). They all diverge in the limit
n → −3 due to the contributions of ever larger scales (ln k → −∞).

The variance
! ∞ ! ∞
2 dk 1 " n+3 #∞
%δ & = ξ(0) = P(k) ∝ kn+2 dk = k 0
for n '= −3 (57)
0 k 0 n+3
diverges at small scales (high k) for n ≥ −3 and at large scales (low k) for n ≤ −3. In practice
we encounter only the small-scale divergence.
We cure the small-scale divergence with filtering as discussed in Sec. 8.1.5, replacing (57)
with (see Eq. 48) ! ∞
2 2 dk
σ (R) ≡ %δ(x, R) & = P(k)|W (k)|2 . (58)
0 k
For the three window functions given in Sec. 8.1.5, power-law spectra give
9 nπ Γ(n − 1)
σT2 (R) = n
(n + 1) sin P(R−1 ) for −3 < n < 1
2 $ % 2 n − 3
2 1 n+3
σG (R) = Γ P(R−1 ) for n > −3
2 2
1
σk2 (R) = P(R−1 ) for n > −3. (59)
n+3
For n ≥ 1, the top-hat window is not able to cure the small-scale divergence, since its Fourier
transform does not die out fast enough at high k (this is related to the sharp boundary of the
window). For integers −3 < n < 1 the formula for σT2 (R) in (59) is not defined, since either
n + 1 or sin nπ/2 gives 0 and Γ(n − 1) gives infinity. For these cases
3π 9 3π
σT2 (R) = P(R−1 ) , P(R−1 ) , P(R−1 ) for n = −2, −1, 0 . (60)
5 4 2
2 (R) = 1 P(R−1 ).
For n = 1, σG 2
2 (R) and σ 2 (R). (σ 2 (R) is more difficult.)
Exercise: Derive these results for σG k T

8.1.7 Galaxy 2-point correlation function

The most obvious way to try to measure the cosmological density perturbations is to observe
the spatial distribution of galaxies. We treat individual galaxies as mathematical points, so
8 STRUCTURE FORMATION 38

that each galaxy has a comoving coordinate value x. We define the galaxy 2-point correlation
function ξg (r) as the excess probability of finding a galaxy at separation r from another galaxy:

dP ≡ n̄ [1 + ξg (r)] dV (61)

where n̄ is the mean galaxy number density, dV is a volume element that is a separation r away
from a chosen reference galaxy, and dP is the probability that there is a galaxy within dV . (Here
dV is assumed so small that there is at most one galaxy in it.)
If the galaxy number density n(x) faithfully traces the underlying matter density, so that

δn δρm
δg ≡ =δ≡ , (62)
n̄ ρ̄m
then ξg becomes equal to the matter density autocorrelation function ξ: The probability of
finding a galaxy in volume dV1 at a random location x is

dP1 = "n(x)#dV1 = "n̄ + δn(x)#dV1 = n̄dV1 . (63)

The probability of finding a galaxy pair at x and x + r is

dP12 = "n(x)n(x + r)#dV1 dV2 = n̄2 "[1 + δ(x)][1 + δ(x + r)]#dV1 dV2
= n̄2 [1 + "δ(x)# + "δ(x + r)# + "δ(x)δ(x + r)#] dV1 dV2
= n̄2 [1 + "δ(x)δ(x + r)#] dV1 dV2 , (64)

since "δ(x)# = "δ(x + r)# = 0. Dividing dP12 with dP1 we get the probability dP2 of finding the
second galaxy once we have found the first one

dP2 = n̄ [1 + "δ(x)δ(x + r)#] dV2 = n̄ [1 + ξ(r)] dV2 . (65)

Thus ξg = ξ.
It is probable that the galaxy number density does not trace the matter density faithfully,
since galaxy formation is likely to be more efficient in high-density regions. This is called bias.
Specifically the bias, or galaxy bias bg , is defined as the ratio

δg
bg ≡ ⇒ ξg = b2g ξ , (66)
δm
where the expectation is that bg > 1. In principle the bias could depend on the scale k, the time
t (or redshift z), and/or the strength of the density perturbation δm . The simplest treatment of
bias is to assume bg is a constant over the observationally relevant ranges of these quantities.
The bias will depend on the type of tracer (all galaxies, specific types of galaxies, galaxy
clusters) and is typically larger for more massive objects.
For the galaxy number density field, observationally σT (R) ≈ 1 at R = 8 h−1 Mpc. This has
motivated the definition (will come later) of the quantity σ8 as a cosmological parameter related
to the amplitude of large-scale structure.
8 STRUCTURE FORMATION 39

8.2 Newtonian perturbation theory

We shall now study the evolution of perturbations during the history of the universe. Initially
the perturbations were small and we restrict the quantitative treatment to that part of the
evolution when they remained small (for large scales, this extends to the present time and the
future). This allows us to use first-order perturbation theory, where we drop from our equations
all those terms which contain a product of two or more perturbations (as these products are even
smaller). The remaining equations will then contain only terms which are either zeroth order,
i.e., contain only background quantities, or first order, i.e., contain exactly one perturbation. If
we kept only the zeroth order parts, we would be back to the equations of the homogenous and
isotropic universe. Subtracting these from our equations we arrive at the perturbation equations
where every term is first-order in the perturbation quantities, i.e., it is a linear equation for
them. This makes the equations easy to handle, we can, e.g., Fourier transform them.
As we discovered in our discussion of inflation, the different cosmological distance scales
first exit the horizon during inflation, then enter the horizon during various epochs of the later
history. Matter perturbations at subhorizon scales, i.e., after horizon entry, can be treated with
Newtonian perturbation theory, but scales which are close to horizon size or superhorizon require
relativistic perturbation theory, which is based on general relativity.
The Newtonian equations for (perfect gas)10 fluid dynamics with gravity are

∂ρ
+ ∇r · (ρu) = 0 (67)
∂t!
∂u 1
!
+ (u · ∇r )u + ∇r p + ∇r Φ̃ = 0 (68)
∂t ρ
∇2r Φ̃ = 4πGρ (69)

Here ρ is the mass density, p is the pressure, and u is the flow velocity of the fluid. We write Φ̃
for the Newtonian gravitational potential, since we want to reserve Φ for its perturbation. The
subscript r in ∇r emphasizes that the space derivatives are taken with respect to the Newtonian
space coordinate r (instead of a comoving coordinate). Although the Newtonian time coordinate
t! is equal to the cosmic time coordinate t, we need to make a distinction between t! and t in
partial derivatives as will become clear soon.
The first equation is the law of mass conservation. The second equation is called the Euler
equation, and it is just “F = ma” for a fluid element, whose mass is ρdV . Here the acceleration
of a fluid element is not given by ∂u/∂t! which just tells how the velocity field changes at a given
position, but by du/dt! , where
d ∂
!
≡ ! + (u · ∇r ) (70)
dt ∂t
is the convective time derivative, which follows the fluid element as it moves. The two other
terms give the forces due to pressure gradient and gravitational field.
We can apply Newtonian physics if:

1) Distance scales considered are # the scale of curvature of spacetime (given by the Hubble
length in cosmology11 )

2) The fluid flow is nonrelativistic, u # c ≡ 1.

3) We are considering nonrelativistic matter, |p| # ρ

10
perfect gas = no internal friction ⇒ pressure is isotropic
11
As discussed in Chapter 3, the spacetime curvature has two distance scales, the Hubble length H −1 and
the curvature radius Rcurv ≡ a|K|−1/2 . From observations we know that the curvature radius is larger than the
Hubble length (at all times of interest), possibly infinite.
8 STRUCTURE FORMATION 40

The last condition corresponds to particle velocities being nonrelativistic, if the matter is made
out of particles. Although the pressure is small compared to mass density, the pressure gradient
can be important if the pressure varies at small scales.
Note: Energy density and mass density. In Newtonian gravity, the source of gravity is mass
density ρm , not energy density ρ. For nonrelativistic matter, the kinetic energies of particles are negligible
compared to their masses, and thus so is the energy density compared to mass density, if we don’t count
the rest energy in it. The Newtonian equations for mass density and energy density are
∂ρm
+ ∇r · (ρm u) = 0 (71)
∂t!
∂ρu
+ ∇r · (ρu u) + p∇r · u = 0, (72)
∂t!
where ∇r · u gives the rate of change of the volume of the fluid element and p∇r · u is the work done
by pressure. In Newtonian physics, rest energy (mass) is not included in the energy density. Eq. (72)
applies whether we include it or not. Define total energy density as
ρ ≡ ρm + ρu ,
where ρu is the Newtonian energy density and ρm is the mass density. Adding Eqs. (71) and (72) gives
∂ρ
+ ∇r · (ρu) + p∇r · u = 0 . (73)
∂t!
For nonrelativistic matter ρu # ρm and p # ρm . We can thus drop the last term in (73) and ignore the
distinction between mass density and total energy density.

A homogeneously expanding fluid,

ρ = ρ(t0 )a−3 (74)

ȧ
u = r (75)
a
2πG 2
Φ̃ = ρr (76)
3
is a solution to these equations (exercise), with a condition to the function a(t) giving the
expansion law. It is the Newtonian version of the matter-dominated Friedmann model. Writing
H(t) ≡ ȧ/a we find that the homogeneous solution satisfies

ρ̇ + 3Hρ = 0 , (77)

and the condition for a(t) (from the exercise) can be written as
ä 4πG
= Ḣ + H 2 = − ρ. (78)
a 3
You should recognize these equations as the energy-continuity equation and the second Fried-
mann equation for a matter-dominated FRW universe.12 The result for Φ̃, Eq. (76), has no
relativistic counterpart, the whole concept of gravitational potential does not exist in relativity
(except in special cases; like here in perturbation theory, where we introduce potentials related
to perturbations).
12
The freedom of choosing the initial value of the expansion rate leaves the connection between H and ρ open
up to a constant. This constant has the same effect on the time evolution of a(t) as the curvature constant K in
the first Friedmann equation, but of course in the Newtonian treatment it is not interpreted as curvature, and it
does not otherwise have the same physical effects. We shall (unless otherwise noted) choose this constant so that
the background solution matches the flat FRW universe. Then we have
8πG 3 2
H2 = ρ or 4πGρ = H . (79)
3 2
8 STRUCTURE FORMATION 41

8.2.1 Comoving coordinates

Introduce now a new (comoving) coordinate system (t, x) which is related to the Newtonian
coordinate system (t! , r) by
t! = t r = a(t)x . (80)
Thus the time coordinate is the same in both coordinate systems, but we need to distinguish
between the partial derivatives ∂/∂t and ∂/∂t! , since in the first x is kept constant and in the
second r is kept constant. Relate now the partial derivatives:
∂ ∂t! ∂ ! ∂ri ∂ ∂ ! ∂ ∂
= + = + ȧxi = + Hr · ∇r
∂t ∂t ∂t! ∂t ∂ri ∂t! ∂ri ∂t!
i
∂ ∂t! ∂ ! ∂rj ∂ ! ∂ ∂
= !
+ = δij a = a ⇒ ∇x = a∇r . (81)
∂xi ∂xi ∂t ∂xi ∂rj ∂rj ∂ri
j j

Thus
∂ ∂ 1
!
= − Hx · ∇x and ∇r = ∇x . (82)
∂t ∂t a
(Later we will work exclusively in the comoving coordinates and write just ∇ for ∇x . The
“original” coordinates r are just an artifact of the Newtonian approach and do not appear in
relativistic perturbation theory.)

8.2.2 The perturbation

Now, consider a small perturbation, so that

ρ(t! , r) = ρ̄(t! ) + δρ(t! , r) (83)

! ! !
p(t , r) = p̄(t ) + δp(t , r) (84)
! ! !
u(t , r) = H(t )r + v(t , r) (85)
2πG ! 2
Φ̃(t! , r) = ρ̄(t )r + Φ(t! , r) , (86)
3
where ρ̄, p̄, and H denote homogeneous background quantities (solutions of the background, or
zeroth-order, equations) and δρ, δp, v, Φ are small inhomogeneous perturbations.
Inserting these into the Eqs. (67,68,69) and subtracting the homogeneous equations (76,77,78)
we get (exercise) the perturbation equations
∂δρ
+ 3Hδρ + Hr · ∇r δρ + ρ̄∇r · v = 0 (87)
∂t!
∂v 1
+ Hv + Hr · ∇r v + ∇r δp + ∇r Φ = 0 (88)
∂t! ρ̄
∇2r Φ = 4πGδρ . (89)

In terms of the comoving coordinates these become (exercise):

∂δρ ρ̄
+ 3Hδρ + ∇x · v = 0 (90)
∂t a
∂v 1 1
+ Hv + ∇x δp + ∇x Φ = 0 (91)
∂t aρ̄ a
∇2x Φ = 4πGa2 δρ . (92)

In terms of the relative density perturbation δ ≡ δρ/ρ̄ we have δρ = ρ̄ · δ and

∂δρ ∂δ
= ρ̄˙ · δ + ρ̄ where ρ̄˙ · δ = −3H ρ̄ δ , (93)
∂t ∂t
8 STRUCTURE FORMATION 42

and we can write

∂v 1 ∂
+ Hv = (av) (94)
∂t a ∂t
so that the set of perturbation equations becomes
∂δ 1
+ ∇x · v = 0 (95)
∂t a
∂ 1
(av) + ∇x δp + ∇x Φ = 0 (96)
∂t ρ̄
∇2x Φ = 4πGa2 ρ̄δ (97)

Finally, we Fourier expand the perturbations,

!
δ(t, x) = δk (t)eik·x etc. (98)
k

In Fourier space the perturbation equations become

ik · vk
δ̇k + = 0 (99)
a
d δpk
(avk ) + ik + ikΦk = 0 (100)
dt ρ̄
" a #2
Φk = −4πG ρ̄δk . (101)
k
Solving the evolution of the perturbations is a two-step process:
1) Solve the background equations to obtain the functions a(t), H(t), and ρ̄(t). After this,
these are known functions in the perturbation equations.

2) Solve the perturbation equations.

8.2.3 Vector and scalar perturbations

We now divide the velocity perturbation field v(t, r) into its rotational (solenoidal, divergence-
free) and irrotational (curl-free) parts,

v = v⊥ + v" , (102)

where ∇ · v⊥ = 0 and ∇ × v" = 0. For Fourier components this simply means that k · v⊥k = 0
and k × v"k = 0. That is, we divide vk into the components perpendicular and parallel to the
wave vector k. The parallel part we can write in terms of a scalar function v, whose Fourier
components vk are given by
v"k ≡ vk k̂ , (103)
where k̂ denotes the unit vector in the k direction.
We can now take the perpendicular and parallel parts of Eq. (100),
d
(av⊥k ) = 0 (104)
dt
d δpk
(avk ) + ik + ikΦk = 0 . (105)
dt ρ̄
We see that the rotational part of the velocity perturbation has a simple time evolution,

v⊥ ∝ a−1 , (106)
8 STRUCTURE FORMATION 43

i.e., it decays from whatever initial value it had, inversely proportional to the scale factor.
The other perturbation equations involve only the irrotational part of the velocity perturba-
tion. Thus we can divide the total perturbation into two parts, commonly called the vector and
scalar perturbations, which evolve independent of each other:
1) The vector perturbation: v⊥ .

2) The scalar perturbation: δ, δp, v, Φ, which are all coupled to each other.
The vector perturbations are thus not related to the density perturbations, or the structure
of the universe. Also, any primordial vector perturbation should become rather small as the
universe expands, at least while first-order perturbation theory applies.13 They are thus not very
important, and we shall have no more to say about them. The rest of our discussion focuses on
the scalar perturbations.

8.2.4 The equations for scalar perturbations

We summarize here the Fourier space equations for scalar perturbations:
ikvk a
δ̇k + = 0 ⇒ vk = i δ̇k (107)
a k
d δpk
(avk ) + ik + ikΦk = 0 (108)
dt ρ̄
! a "2
Φk = −4πG ρ̄δk . (109)
k
Inserting vk from (107) and Φk from (109) into (108) we get

k2 δpk
δ̈k + 2H δ̇k = − + 4πGρ̄δk . (110)
a2 ρ̄

8.2.5 Adiabatic and entropy perturbations

Suppose the equation of state is barotropic,

p = p(ρ) (111)

i.e., pressure is uniquely determined by the energy density. Then the perturbations δp and δρ
are necessarily related by the derivative dp/dρ of this function p(ρ),
dp dp
p = p̄ + δp = p̄(ρ̄) + (ρ̄)δρ ⇒ δp = δρ .
dρ dρ
The time derivatives of the background quantities p̄ and ρ̄ are related by this same derivative,
dp̄ dp dρ̄ dp
p̄˙ = = (ρ̄) = ρ̄˙ .
dt dρ dt dρ
Assuming this derivative dp/dρ is nonnegative, we call its square root the speed of sound
#
dp
cs ≡ . (112)
dρ
13
Thus we end up with an irrotational velocity field. The rotational motion (e.g., rotation of galaxies) which
is common in the present universe at small scales has arisen from higher-order effects from the primordial scalar
perturbations, not from the primordial vector perturbations.
8 STRUCTURE FORMATION 44

Figure 5: For adiabatic perturbations, the conditions in the perturbed universe (right) at (t1 , x)
equal conditions in the (homogeneous) background universe (left) at some time t1 + δt(x).

(We shall indeed find that sound waves propagate at this speed.) We thus have the relation

δp p̄˙
= = c2s .
δρ ρ̄˙
In general, when p may depend on other variables besides ρ, the speed of sound in a fluid is
given by ! "
∂p
c2s = (113)
∂ρ S
where the subscript S indicates that the derivative is taken so that the entropy of the fluid
element is kept constant. Since the background universe expands adiabatically (meaning that
there is no entropy production), we have that
! "
p̄˙ ∂p
= = c2s . (114)
ρ̄˙ ∂ρ S

Perturbations with the property

δp p̄˙
= (115)
δρ ρ̄˙
are called adiabatic perturbations in cosmology.
If p = p(ρ), perturbations are necessarily adiabatic. In the general case the perturbations
may or may not be adiabatic. In the latter case, the perturbation can be divided into an
adiabatic component and an entropy perturbation. An entropy perturbation is a perturbation
in the entropy-per-particle ratio.
For adiabatic perturbations we thus have

p̄˙
δp = c2s δρ = δρ . (116)
ρ̄˙

Adiabatic perturbations have the property that the local state of matter (determined here by
the quantities p and ρ) at some spacetime point (t, x) of the perturbed universe is the same
as in the background universe at some slightly different time t + δt, this time difference being
different for different locations x. See Fig. 5.
Thus we can view adiabatic perturbations as some parts of the universe being “ahead” and
others “behind” in the evolution.
Adiabatic perturbations are the simplest kind of perturbations. Single-field inflation pro-
duces adiabatic perturbations, since perturbations in all quantities are proportional to a pertur-
bation δϕ in a single scalar quantity, the inflaton field.
8 STRUCTURE FORMATION 45

Adiabatic perturbations stay adiabatic while they are outside horizon, but may develop
entropy perturbations when they enter the horizon. This happens for many-component fluids
(discussed a little later).
Present observational data is consistent with the primordial (i.e., before horizon entry) per-
turbations being adiabatic.

8.2.6 Adiabatic perturbations in matter

Consider now adiabatic perturbations of a non-relativistic single-component fluid. The equation
for the density perturbation is now
! "
c2s k2
δ̈k + 2H δ̇k + − 4πGρ̄ δk = 0 . (117)
a2

I shall call this the Jeans equation14 (although Jeans considered a static, not an expanding fluid).
This is a second-order differential equation from which we can solve the time evolution of
the Fourier amplitudes δk (t) of the perturbation. Before solving this equation we need to first
find the background solution which gives the functions a(t), H(t) = ȧ/a, and ρ̄(t).
The nature of the solution to Eq. (117) depends on the sign of the factor in the brackets.
The first term in the brackets is due to pressure gradients. Pressure tries to resist compression,
so if this term dominates, we get an oscillating solution, standing density (sound) waves. The
second term in the brackets is due to gravity. If this term dominates, the perturbations grow.
The wavenumber for which the terms are equal,
√ #
a 4πGρ̄ 31
kJ = = H, (118)
cs 2 cs
is called the Jeans wave number, and the corresponding wavelength
#
2π 2 −1
λJ = = 2πcs H (119)
kJ 3
the Jeans length. Here H ≡ aH, the comoving Hubble parameter. In the latter equalities we
assumed that the background solution is the flat FRW universe, so that

4πGρ̄ = 32 H 2 . (120)

For nonrelativistic matter cs $ 1, so that the Jeans length is much smaller than the Hubble
length, kJ % H. Thus we can apply Newtonian theory for scales both larger and smaller than
the Jeans length.
For scales much smaller that the Jeans length, k % kJ , we can approximate the Jeans
equation by
c2 k2
δ̈k + 2H δ̇k + s 2 δk = 0 . (121)
a
The solutions are oscillating, i.e., we get sound waves. The exact solutions of (121) are Bessel
functions, but for small scales we can make a further approximation by first ignoring the middle
term (which is smaller than the other two) and the time-dependence of a and cs to get that
δk (t) ∼ e±iωt , where ω = cs k/a. These oscillations are damped by the 2H δ̇k term, so the
amplitude of the oscillations decreases with time. There is no growth of structure for sub-Jeans
scales.
14
In the literature, there is usually no name given to this equation, but the terms Jeans length etc. are standard.
8 STRUCTURE FORMATION 46

Exercise: Sound waves. For short-wavelength modes k ! kJ , density perturbations in the matter-
dominated universe satisfy (121). Switch to conformal time, dη = dt/a, and solve δk (η) for the Ωm = 1,
ΩΛ = 0 cosmology, assuming cs = const . How does the amplitude and frequency of the oscillations
change with time and scale factor? (Hint: The solutions are spherical Bessel functions.)

For scales much longer than the Jeans length (but still subhorizon), H " k " kJ , we
can approximate the Jeans equation by

δ̈k + 2H δ̇k − 4πGρ̄δk = 0 . (122)

We dropped the pressure gradient term, which means that this equation applies also to nona-
diabatic perturbations for scales where pressure gradients can be ignored. Note that Eq. (122)
is the same for all k, i.e., there is no k-dependence in the coefficients. This means that the
equation applies also in coordinate space, i.e. for δ(x), as long as we ignore contributions from
scales that do not satisfy H " k " kJ .
For a matter-dominated universe, the background solution is a ∝ t2/3 , so that
ȧ 2
H= = (123)
a 3t
and
8πG 4 1
ρ̄ = H 2 = 2 ⇒ ρ̄ = , (124)
3 9t 6πGt2
so the Jeans equation becomes
4 2
δ̈k + δ̇k − 2 δk = 0 . (125)
3t 3t
The general solution is
δk (t) = b1 t2/3 + b2 t−1 . (126)
The first term is the growing mode and the second term the decaying mode. After some time
the decaying mode has died out, and the perturbation grows

δ ∝ t2/3 ∝ a . (127)

Thus density perturbations in matter grow proportional to the scale factor.

From Eq. (101) we have that

Φ ∝ a2 ρ̄δ ∝ a2 a−3 a = const.

The gravitational potential perturbation is constant in time during the matter-dominated era.

8.2.7 Radiation
Since radiation is a relativistic form of energy, we cannot apply the preceding Newtonian dis-
cussion to perturbations in radiation. However, the qualitative results are similar.
The equation of state for radiation is p = ρ/3, and the speed of sound in a radiation fluid is
given by
dp 1
c2s = = .
dρ 3
Thus the Jeans length for radiation is comparable to the Hubble length, and the subhorizon
scales are also sub-Jeans scales for radiation. Thus for subhorizon radiation perturbations we
only get oscillatory solutions. During the radiation-dominated epoch they are not damped by
expansion, but the oscillation amplitude stays roughly constant.
8 STRUCTURE FORMATION 47

Since
Φ ∝ a2 ρ̄δ ∝ a−2 δ ,
the amplitude of the gravitational potential oscillation decays.
Relativistic perturbations in non-expanding space. While the full treatment of relativistic
perturbations is beyond the level of this course, we can obtain the limit where we ignore the effect of
expansion by combining special relativity and the Newtonian limit of general relativity.
Special relativistic fluid dynamics follows from the energy-momentum continuity equation
∂T µν
≡ ∂ν T µν ≡ T µν,ν = 0 . (128)
∂xν
For a perfect fluid
T µν = (ρ + p)uµ uν + pg µν , (129)
µν µ
where the metric is now that of Minkowski space, g = diag(−1, 1, 1, 1). The 4-velocity u is related to
the 3-velocity v = v i by
uµ = (γ, γv) , (130)
√
where γ = 1/ 1 − v 2 .
By contracting the energy tensor T µν with the 4-velocity uµ we obtain uν T µν
,ν = 0, which gives

(ρuµ ),µ + puµ,µ = 0 , (131)

the energy continuity equation. Subtracting uν times this from (128) we get the special relativistic Euler
equation
(ρ + p)uµ uν,µ + (g µν + uµ uν )p,µ = 0 , (132)
where
uµ uν,µ ≡ aν (133)
is the 4-acceleration.
For small velocities, v % 1, we can approximate γ ≈ 1, so that

uµ ≈ (1, v) (134)

and (131),(132) become

∂ρ
+ ∇ · (ρv) = −p∇ · v
! ∂t "
∂
(ρ + p) + v · ∇ %v = −∇p − v(v · ∇p) ≈ −∇p . (135)
∂t
In the Newtonian limit of general relativity, but without the assumption p % ρ, the passive gravita-
tional mass density is given by ρ + p, so that the gravitational force on a volume element of fluid is given
by −(ρ + p)∇Φ and the active gravitational mass density by ρ + 3p, so that the gravitational potential
is given by
∇2 Φ = 4πG(ρ + 3p) . (136)
Thus the Euler equation with gravity becomes
! "
∂
(ρ + p) + v · ∇ v = −∇p − (ρ + p)∇Φ . (137)
∂t
For several fluid components, not interacting with each other except gravitationally, the fluid equa-
tions become thus
∂ρi
+ ∇ · (ρi vi ) = −pi ∇ · vi
! ∂t "
∂
(ρi + pi ) + vi · ∇ vi = −∇pi − (ρi + pi )∇Φ (138)
∂t
#
∇2 Φ = 4πG (ρi + 3pi ) .
i
8 STRUCTURE FORMATION 48

For perturbations ρi = ρ̄i + δρi = ρ̄i (1 + δi ), pi = p̄i + δpi , where the background density and pressure
are now constant both in space and time, we get to first order in perturbations
∂δi
= −(1 + wi )∇ · vi
∂t
∂vi
(ρ̄i + p̄i ) = −∇δpi − (ρ̄i + p̄i )∇Φ (139)
∂t
!
∇2 Φ = 4πG (ρ̄i δi + 3δpi ) ,
i

where wi ≡ p̄i /ρ̄i . For Fourier components this becomes

δ̇ik = −ik(1 + wi )k · vik

(ρ̄i + p̄i )v̇ik = −ikδpik − ik(ρ̄i + p̄i )Φk (140)
−4πG !
Φk = (ρ̄i δik + 3δpik ) .
k2 i

For vector perturbations the second equation gives

v̇i⊥k = 0 ⇒ vi⊥k = const , (141)

and for scalar perturbations the first and second equations become

δ̇ik = −i(1 + wi )kvik

δpik
v̇ik = −ik − ikΦk . (142)
ρ̄i + p̄i
If wi = const, so that ẇi = 0, we get the Jeans equation
δpik
δ̈ik + k 2 + k 2 (1 + wi )Φk = 0 . (143)
ρ̄i

8.2.8 Many fluid components

Assume now that the “cosmic fluid” contains several components i (different types of matter
or energy) which do not interact with each other, except gravitationally. This means that
each component feels only its own pressure15 , and that the components can have different flow
velocities. Then the Newtonian equations for each component i are
∂ρi
+ ∇r · (ρi ui ) = 0 (144)
∂t!
∂ui 1
+ (ui · ∇r )ui + ∇r pi + ∇r Φ̃ = 0 (145)
∂t! ρi
∇2r Φ̃ = 4πGρ , (146)
"
where ρ = ρi . Note that there is only one gravitational potential Φ̃, due to the total density,
and this way the different components do interact gravitationally.
We again have the homogeneous solution, where now each component has to satisfy

ρ̇i + 3Hρi = 0 , (147)

15
In standard cosmology, we actually have just one component, the baryon-photon fluid, which feels its own
pressure, and the other components do not feel even their own pressure (neutrinos after decoupling) or do not
even have pressure (cold dark matter). But we shall first do this general treatment, and do the application to
standard cosmology later.
8 STRUCTURE FORMATION 49

and the expansion law

4πG
Ḣ + H 2 = − ρ, (148)
3
is determined by the total density.
We can now introduce the density, pressure, and velocity perturbations for each component
separately,
ρi (t! , r) = ρ̄i (t) + δρi (t! , r) (149)
! !
pi (t , r) = p̄i (t) + δpi (t , r) (150)
! !
ui (t , r) = H(t)r + vi (t , r) , (151)
but there is only one gravitational potential perturbation,
2πG 2
Φ̃(t! , r) = ρ̄r + Φ(t! , r) . (152)
3
Following the earlier procedure, we obtain the perturbation equations for the fluid compo-
nents,
∂
δρi + 3Hδρi + Hr · ∇r δρi + ρ̄i ∇ · vi = 0 (153)
∂t!
∂ 1
!
vi + Hvi + Hr · ∇r vi + ∇r δpi + ∇r Φ = 0 (154)
∂t ρ̄i
∇2r Φ = 4πGδρ (155)
in Newtonian coordinate space, and
ik · vik
δ̇ik + = 0 (156)
a
d δpik
(avik ) + ik + ikΦk = 0 (157)
dt ρ̄i
a2 !
Φk = −4πG ρ̄i δik (158)
k2
"
in comoving Fourier space. Here δρ = δρi and
δρi
δi ≡ . (159)
ρ̄i
Separating out the scalar perturbations we finally get
k2 δpik
δ̈ik + 2H δ̇ik = − + 4πGδρk , (160)
a2 ρ̄i
where !
δρk = ρ̄j δjk . (161)
j

8.2.9 Adiabatic and entropy perturbations again

The simplest inflation models predict that the primordial perturbations are adiabatic. This
means that locally the perturbed universe at some (t, x) looks like the background universe at
some time t + δt(x). See Sec. 8.2.5.

#  δpi p̄˙ i

 =
˙
δρi (x) = ρ̄i δt(x) δρi ρ̄˙ i
⇒ (162)
δpi (x) = p̄˙ i δt(x) 
 δρ ρ̄˙ δi ρ̄˙ i ρ̄j
 i = i ⇒ =
δρj ρ̄˙ j δj ρ̄i ρ̄˙ j
8 STRUCTURE FORMATION 50

If there is no energy transfer between the fluid components at the background level, the energy
continuity equation is satisfied by them separately,

ρ̄˙ i = −3H(ρ̄i + p̄i ) ≡ −3H(1 + wi )ρ̄i , (163)

where wi ≡ p̄i /ρ̄i is the equation-of-state parameter of fluid component i. Thus for adiabatic
perturbations,
δi δj
= (164)
1 + wi 1 + wj
(which is thus related to ρ̄i ∝ a−3(1+wi ) ). For matter components wi ≈ 0, and for radiation
components wi = 31 . Thus, for adiabatic perturbations, all matter components have the same
perturbation
δi = δm
and all radiation perturbations have likewise
4
δi = δr = δm .
3
We can define a relative entropy perturbation16 between two components
! "
δρi δρj δi δj
Sij ≡ −3H − = − (165)
˙ρ̄i ˙ρ̄j 1 + wi 1 + wj

to describe a deviation from the adiabatic case. The relative entropy perturbation is a pertur-
bation in the ratio of the number densities of the two species. For a nonrelativistic species
δρi δni
ρi = mi ni ⇒ δρi = mi δni and δi ≡ = , (166)
ρ̄i n̄i

whereas for an ultrarelativistic species (µ & T and m & T )

δTi
ρi ∝ Ti4 ⇒ δρi = ρ̄i · 4
Ti
δT i
ni ∝ Ti3 ⇒ δni = n̄i · 3
Ti
δρi 4 δni
⇒ δi ≡ = . (167)
ρ̄i 3 n̄i
For both cases
δni
δi = (1 + wi ) . (168)
n̄i
Thus
δni δnj δ(ni /nj )
Sij = − = . (169)
n̄i n̄j n̄i /n̄j
Even if perturbations are initially adiabatic, relative entropy perturbation may develop inside
the horizon. We shall encounter such a case in Sec. 8.3.4.
16
There is a connection to entropy/particle of the different components, but we need not concern ourselves with
it now. It is not central to this concept, and it is perhaps somewhat unfortunate that it has become customary,
for historical reasons, to use the word “entropy” for these perturbations.
8 STRUCTURE FORMATION 51

8.2.10 The effect of a homogeneous component

The energy density of the real universe consists of several components. In many cases it is
reasonable to ignore the perturbations in some components (since they are relatively small in
the scales of interest). We call such components smooth and we can add them together into a
single smooth component ρs = ρ̄s .
Consider the case where we have perturbations in a nonrelativistic (“matter”) component
ρm , and the other components are smooth. Then

ρ = ρm + ρs (170)

but
δρ = δρm ≡ ρ̄m δ . (171)
We write just δ for δρm /ρ̄m , since there is no other density perturbation, but note that now
δ "= δρ/ρ̄ (beware of this trap!).
Assuming adiabatic perturbations, we have then from Eq. (160) that
! 2 2 "
cs k
δ̈k + 2H δ̇k + − 4πGρ̄m δk = 0 . (172)
a2

The difference from Eq. (117) is that now the background energy density in the “gravity” term
still contains only the matter component ρ̄m , but the expansion law, a(t) and H(t) comes from
the full background energy density ρ̄ = ρ̄m + ρ̄s .
Newtonian perturbation theory can be applied even with the presence of relativistic energy
components, like radiation and dark energy, as long as they can be considered as smooth com-
ponents and their perturbations can be ignored. Then they contribute only to the background
solution. In this case we have to calculate the background solution using general relativity, i.e.,
the background solution is a FRW universe, but the perturbation equations are the Newtonian
perturbation equations. We can also consider a non-flat (open or closed) FRW universe, as long
as we only apply perturbation theory to scales much shorter than the curvature radius (and
the Hubble length). Thus the background quantities are to be solved from the Friedmann and
energy continuity equations
K 8πG
H2 + = ρ (173)
a2 3
4πG
Ḣ + H 2 = − (ρ + 3p) (174)
3
ρ̇ = −3H(ρ + p) . (175)

Example: Matter perturbations in flat vacuum-dominated universe. Consider the case

where ρs = ρvac $ ρm and matter is approximated as pressureless (we do not then have to make a
separate adiabaticity assumption, since the pressure term does not appear). Then the Jeans equation
becomes
δ̈k + 2H δ̇k − 4πGρ̄m δk = 0 . (176)
To estimate the relative order of magnitude of the three terms it is better to divide the equation by H 2 ,
4πGρ̄m
H −2 δ̈k + 2H −1 δ̇k − δk = 0 , (177)
H2
so that the Hubble time H −1 provides the time scale for the time derivatives. Now
8πG 8πG
H2 = ρcr ≈ ρvac = const (178)
3 3
8 STRUCTURE FORMATION 52

and in the last term δk is multiplied with 23 ρ̄m /ρvac ! 1, so that we can drop the last term and approxi-
mate the Jeans equation by
δ̈k + 2H δ̇k = 0 . (179)
We see immediately that δk = const is a solution. For the other solution, solve first δ̇k :

dδ̇k dδ̇k
= −2H δ̇k ⇒ = −2Hdt , (180)
dt δ̇k

whose solution is ln δ̇k = −2Ht + const or δ̇k = Ce−2Ht . Integrating this gives

δk = Ae−2Ht + B , (181)

with a constant term and an exponentially decaying term. Thus in a vacuum-dominated universe matter
perturbations stay constant (after the decaying term has died out); or to be more precise and referring
to the original equation (177), the relative change in δk in a Hubble time is of order ρ̄m /ρvac ! 1

We shall do this calculation more accurately later, including the transition from matter
domination to vacuum domination. The main lesson now is that the increased expansion rate
due to the presence of a smooth component slows down the growth of perturbations.
Exercise: Find the solution for the Jeans equation for pressureless matter perturbations when a)
the energy density is dominated by a smooth radiation component b) when there is no other energy
component, but the universe has the open geometry (K < 0) and is curvature dominated, considering
only scales ! curvature radius.

8.2.11 Meszaros equation

Consider now a flat universe with just cold dark matter (CDM) and radiation, and ignore pertur-
bations in radiation so that radiation can be taken as a smooth component. This approximation
may be motivated by noting that subhorizon radiation perturbations do not grow. The CDM
is pressureless, and thus the CDM sound speed is zero, and so is the CDM Jeans length. Thus,
for CDM, all scales are larger than the Jeans scale, and we don’t get an oscillatory behavior.
Instead, perturbations grow at all scales.
We get the equation for the CDM perturbation from Eq. (172) by setting cs = 0 (or rather,
δp = 0; we need not invoke the assumption of adiabaticity, since CDM is pressureless),

δ̈k + 2H δ̇k − 4πGρ̄m δk = 0 . (182)

Note that the equation is the same for all k and therefore it applies also in the coordinate space,
i.e., for δ(x). To simplify notation, we drop the subscript k .
The Friedmann equation is
! "2
2 ȧ 8πG
H = = ρ̄ ,
a 3
where ρ̄ = ρ̄m + ρ̄r and ρ̄m ∝ a−3 and ρ̄r ∝ a−4 .
A useful trick is to study this as a function of a instead of t or η. We define a new time
coordinate,
a ρ̄m
y≡ = . (183)
aeq ρ̄r
(y = 1 at t = teq .) Now
y 3 y
4πGρ̄m = 4πG ρ̄ = H2 (184)
y+1 2y+1
and Eq. (182) becomes
3 y
δ̈ + 2H δ̇ − H2 = 0 . (185)
2y+1
8 STRUCTURE FORMATION 53

Performing the change of variables from t to y (Exercise; you may need the 2nd Friedmann
equation), we arrive at the equation (where ! ≡ d/dy)

2 + 3y ! 3
δ!! + δ − δ = 0, (186)
2y(1 + y) 2y(1 + y)

known as the Meszaros equation.

It has two solutions, one growing, the other one decaying. The growing solution is
! " ! "
3y 3 a
δ = δprim 1+ = δprim 1+ . (187)
2 2 aeq

We see that the perturbation remains frozen to its primordial value, δ ≈ δprim , during the
radiation-dominated period. By t = teq , it has grown to δ = 52 δprim .
During the matter-dominated period, y $ 1, the CDM perturbation grows proportional to
the scale factor,
δ ∝ y ∝ a ∝ t2/3 . (188)
8 STRUCTURE FORMATION 54

8.3 Perturbations at subhorizon scales in the real universe

8.3.1 Horizon entry
Newtonian perturbation theory is valid only at subhorizon scales, k ! H, or k−1 " H−1 .
During “normal”, decelerating expansion, i.e., after inflation but before the recent onset of dark
energy domination, scales are entering the horizon. Short scales enter first, large scales enter
later. We have not yet studied what happens to perturbations outside the horizon (for that we
need (general) relativistic perturbation theory, to be discussed In Sec. 8.4). So, for the present
discussion, whatever values the perturbation amplitudes δk have soon after horizon entry, are
to be taken as an initial condition, the primordial perturbation17 . Observations actually suggest
that different scales enter the horizon with approximately equal perturbation amplitude, whose
magnitude is characterized by the number18 few × 10−5 .
The history of the different scales after horizon entry, and thus their present perturbation
amplitude, depends on at what epoch they enter. The scales which enter during transitions
between epochs are thus special scales which should characterize the present structure of the
universe. Such important scales are the scale (exercise)
−1
keq = (Heq )−1 ∼ 13.7 Ω−1
m h
−2 −1
Mpc ≡ 13.7 ωm Mpc , (189)

which enters at the time teq (1 + zeq = 23 902 ωm ) of matter-radiation equality, and the scale
! "−1/2
−1 −1 Ωr
kdec = (Hdec ) ∼ 91 Ω−1/2
m1+ (1 + zdec ) h−1 Mpc
Ωm
! "−1/2
−1/2 ωr
≡ 91 ωm 1+ (1 + zdec ) Mpc , (190)
ωm

which enters at the time tdec (zdec = 1090) of photon decoupling. Here ωr = 4.184 × 10−5
includes relativistic neutrinos, since the result above only requires them to be relativistic at tdec .
For ΩΛ = 0.7, Ωm = 0.3, h = 0.7, these scales are
−1
keq = 65 h−1 Mpc = 93 Mpc
−1
kdec = 145 h−1 Mpc = 207 Mpc . (191)

The smallest “cosmological” scale is that corresponding to a typical distance between galaxies,
about 1 Mpc.19 This scale entered during the radiation-dominated epoch (well after Big Bang
nucleosynthesis).
The scale corresponding to the present “horizon” (i.e. Hubble length) is

k0−1 = (H0 )−1 = 2998 h−1 Mpc ∼ 4300 Mpc . (192)

Because of the acceleration due to dark energy, this scale is actually exiting now, and there are
scales, somewhat larger than this, that have briefly entered, and then exited again in the recent
past. The horizon entry is not to be taken as an instantaneous process, so these scales were
17
We shall later redefine primordial perturbation to refer to the perturbation at the epoch when all cosmologically
interesting scales were well outside the horizon, which is the standard meaning of this concept in cosmology.
18
Although in coordinate space the relative density perturbation δ(x) is a dimensionless number, the Fourier
quantity δk is not. The size of δk is characterized by the dimensionless value P(k)1/2 .
19
In the present universe, structure at smaller scales has been messed up by galaxy formation, so that it bears
little relation to the primordial perturbations at these scales. However, observations of the high-redshift universe,
especially so-called Lyman-α observations (absorption spectra of high-z quasars, which reveal distant gas clouds
along the line of sight), can reveal these structures when they are closer to their primordial state. With such
observations, the “cosmological” range of scales can be extended down to ∼ 0.1 Mpc.
8 STRUCTURE FORMATION 55

never really subhorizon enough for the Newtonian theory to apply to them. Thus we shall just
consider scales k−1 < k0−1 . The largest observable scales, of the order of k0−1 , are essentially at
their “primordial” amplitude now.
We shall now discuss the evolution of the perturbations at these scales (k−1 < k0−1 ) after
horizon entry, using the Newtonian perturbation theory presented in the previous section.

8.3.2 Composition of the real universe

The present understanding is that there are five components to the energy density of the universe,

1. cold dark matter (c)

2. baryonic matter (b)

3. photons (γ)

4. neutrinos (ν)

5. dark energy (d)

(during the time of interest for this section, i.e., from some time after BBN until the present).
Thus
ρ = ρc + ρb + ργ + ρν +ρd . (193)
! "# $ ! "# $
ρm ρr

(Note that ρc here is the CDM density, not the critical density, for which we write ρcr .)
Baryons and photons interact with each other until t = tdec , so for t < tdec they have to be
discussed as a single component,
ρbγ = ρb + ργ . (194)
The other components do not interact with each other, except gravitationally, during the time
of interest. The fluid description of Sec. 8.2 can only be applied to components whose particle
mean free paths are shorter than the scales of interest. After decoupling, photons “free stream”
and cannot be discussed as a fluid. On the other hand, the photon component becomes then
rather homogeneous quite soon, so we can approximate it as a “smooth” component20 . The
same applies to neutrinos for the whole time since the BBN epoch, until the neutrinos become
nonrelativistic. This will happen to at least two of the three neutrino species, and then they
should be treated as matter (hot dark matter), not radiation. According to observations, the
neutrino masses are small enough, not to have a major impact on structure formation. (However,
for accurate work this must be taken into account and this effect on structure formation provides
the tightest cosmological limits to neutrino masses.) Thus we shall here approximate neutrinos
as a smooth radiation component. Dark energy is believed to be relatively smooth. If it is a
cosmological constant (vacuum energy) then it is perfectly homogeneous.
The discussion in Sec. 8.2 applies to the case, where ρ can be divided into two components,

ρ = ρm + ρs , (195)

where the perturbation is only in the matter component ρm and ρs = ρ̄s is homogeneous.
For perturbations in radiation components and dark energy the Newtonian treatment is not
20
As long as we are interested in density perturbations only. When we are interested in the CMB anisotropy,
the momentum distribution of these photons becomes the focus of our attention.
8 STRUCTURE FORMATION 56

enough. Unfortunately, we do not have quite this two-component case here. Based on the above
discussion, a reasonable approximation is given by a separation into three components:

t < tdec : ρ = ρc + ρbγ + ρs (ρs = ρν + ρd ) (196)

t > tdec : ρ = ρc + ρb + ρs (ρs = ργ + ρν + ρd ) . (197)

After decoupling, both ρc and ρb are matter-like (p ! ρ) and we’ll discuss in Sec. 8.3.4 how this
case is handled. Before decoupling, the situation is more difficult, since ρbγ is not matter-like, as
the pressure provided by photons is large. Here we shall be satisfied with a crude approximation
for this period.
The most difficult period is that close to decoupling, where the photon mean free path λγ
is growing rapidly. The fluid description, which we are here using for the perturbations, applies
only to scales " λγ , whereas the photons are smooth only for scales ! λγ . Thus this period
can be treated properly only with large numerical “Boltzmann” codes, such as CMBFAST or
CAMB.

8.3.3 CDM density perturbations

Cold dark matter is the dominant structure-forming component in the universe (dark energy
dominates the energy density at late times, but does not form structure, or, if it does, these
structures are very weak, not far from homogeneous). Observations indicate that ρb ∼ 0.2ρc .
Thus we get a first approximation to the behavior of the CDM perturbations by ignoring the
baryon component and equating
ρm ≈ ρ c .
The CDM is pressureless, and thus the CDM sound speed is zero, and so is the CDM Jeans
length. Thus, for CDM, all scales are larger than the Jeans scale, and we don’t get an oscillatory
behavior. Instead, perturbations grow at all scales. On the other hand, as we shall discuss in
Sec. 8.3.4, perturbations in ρbγ oscillate before decoupling. Therefore the perturbations in ρbγ
will be smaller than those in ρc , and we can make a (crude) approximation where we treat ρbγ
as a homogeneous component before decoupling. This is important, since although ρb ! ρc , this
is not true for ρbγ at earlier, radiation-dominated, times. At decoupling ρb < ργ < ρc . Before
matter-radiation equality, there is an epoch when ρc < ργ , but δρc > δρbγ . For simplicity, we
now approximate
ρ = ρm + ρr + ρd (198)
where ρm = ρc and ρr = ργ + ρν is a smooth component (ρν truly smooth, ργ truly smooth
after decoupling, and (crudely) approximated as smooth before decoupling). We have ignored
baryons, since they are a subdominant part of ρbγ before decoupling, and a subdominant matter
component after decoupling. Likewise, ρd is also smooth, and becomes important only close to
present times.
We can now study the growth of CDM perturbations even during the radiation-dominated
period, as the radiation-component is taken as smooth and affects only the expansion rate. We
can study it all the way from horizon entry to the present time, or until the perturbations
become nonlinear (δc = δρc /ρ̄c ∼ 1).
For the radiation- and matter-dominated epochs, including the transition in between, we
then have the case of Sec. 8.2.11, and the matter perturbation grows as (187),
! " ! "
3y 3 a
δ = δprim 1 + = δprim 1 + .
2 2 aeq
The perturbation remains frozen to its primordial value, δ ≈ δprim , during the radiation-
dominated period. By t = teq , it has grown to δ = 52 δprim . During the matter-dominated
8 STRUCTURE FORMATION 57

Figure 6: Growth of CDM perturbation during radiation-dominated epoch for the case of adiabatic
primordial perturbations (qualitative). The time axis represents conformal time.

period, the CDM perturbation grows proportional to the scale factor,

δ ∝ y ∝ a ∝ t2/3 .

However, in the case of adiabatic primordial perturbations, the above approximation misses
an important effect: an additional logarithmic growth factor ∼ ln(k/keq ) the CDM perturbations
get from the gravitational effect (ignored in the above) of the oscillating radiation perturbation
during the radiation-dominated epoch. To get this boost the CDM perturbation must initially
be in the same direction (positive or negative) as the radiation perturbation, which is the case
for adiabatic primordial perturbations.
For adiabatic primordial perturbations, the baryon, CDM, and radiation perturbations are
related at horizon entry as δc = δb = 34 δγ . Consider scales that enter during the radiation-
dominated epoch (t < teq < tdec ). The gravitational effect is dominated initially by the radiation
perturbation, which begins to oscillate after horizon entry; the baryon perturbation will oscillate
with it until tdec . CDM, on the other hand, does not feel the radiation pressure responsible for
the oscillation, it sees only the gravitational effect of the baryon-photon fluid. In the first
phase of the oscillation period δc is of the same sign as δbγ , so δbγ adds to the gravitational
pull to increase δc . Since at first δρbγ > δρc , this additional pull is larger than that of CDM
itself, leading to a much faster growth of δc (which otherwise would grow very little during the
radiation domination). The flow of CDM is accelerated toward CDM overdensities. In the next
phase of the oscillation, the sign of δbγ reverses, and now the pull of δρbγ on CDM is in the
opposite direction, and will slow down the flow of CDM toward overdensities. But this is not
enough to reverse the CDM flow before the sign of δbγ changes again and begins to accelerate
CDM again toward CDM overdensities.
Thus the effect of the radiation oscillations is to increase δc stepwise, one step for each
oscillation period. See Fig. 6. As the ρ̄γ /ρ̄c ratio decreases the relative increases per step
decrease; but this effect keeps adding steps until tdec . The smaller the scale (the higher the
k) the more steps there are between horizon entry (tk ) and tdec , and the larger the first steps.
An analytic calculation (too complicated for this course, but it is done in [9] and I do it in
Cosmological Perturbation Theory) of this effect, in the small-scale limit k # keq and still
ignoring baryons, so that the oscillating radiation perturbation is just photons, gives that it
8 STRUCTURE FORMATION 58

leads to a boost by a factor ∼ 7.5 ln(0.17k/keq ), so that (187) is modified to

! " ! "
3 a k
δc ≈ δprim 1 + 7.5 ln 0.17 for k # 6keq and t > tdec (199)
2 aeq keq

(for k ! 6keq the logarithm is negative; this approximate result does not apply for such large
scales). There is more discussion of this result in Sec. 8.4.4, where we compare this approximate
analytical result to a more accurate numerical result from CAMB.

8.3.4 Baryon density perturbations

Although CDM is the dominant matter component in the universe, we cannot directly see it.
The main method to observe the density perturbation today is to study the distribution of
galaxies. But the part of galaxies that we can see is baryonic. Thus to compare the theory
of structure formation to observations, we need to study how the perturbation in the baryonic
component evolves.

Baryon Jeans length and speed of sound. We define the baryon Jeans length as λJ = 2πkJ−1 ,
where
cs
kJ−1 = √ , (200)
a 4πGρ̄b
and cs is the speed of sound for baryons (i.e., in the baryon-photon fluid before decoupling, and in the
baryon fluid after decoupling). This definition compares the pressure felt by baryons to baryon gravity,
so it addresses the question whether the baryon density perturbation can grow under its own gravity.
This is not the question we face in reality, since at early times the gravity was dominated by the radiation
perturbation and later by the CDM perturbation. The baryon Jeans length can still be used for order-
of-magnitude estimates of at what scales the baryon perturbation can grow, and for the argument that
we cannot match observations without CDM.
In general, ! "
∂p
c2s = , (201)
∂ρ σ
where σ refers to constant entropy per baryon. Since in our case the entropy is completely dominated by
photons,
4π 2 3 2π 4
sbγ ∼ sγ = T = nγ , (202)
45 45ζ(3)
we have
sbγ sγ 2π 4 nγ 1
σ≡ ∼ = ≈ 3.6016 , (203)
nb nb 45ζ(3) nb η
where η is the baryon-to-photon ratio.
We find the speed of sound by varying ρbγ and pbγ adiabatically, (i.e., keeping σ, the entropy/baryon
constant), which in this case means keeping η constant. Now
2ζ(3) 3 δT
ρb = mnb = mηnγ = mη T ⇒ δρb = ρ̄b · 3
π2 T
π2 4 δT
ργ = T ⇒ δργ = ρ̄γ · 4
15 T
π2 4 δT 4 δT
pγ = T ⇒ δpγ = p̄γ · 4 = ρ̄γ · .
45 T 3 T
Since pb ' pγ ⇒ δpb ' δpγ , we get
4
δp δpγ 3 ρ̄γ 1 1
c2s = = = = . (204)
δρ δργ + δρb 4ρ̄γ + 3ρ̄b 3 1 + 43 ρ̄ρ̄γb

This was a calculation of the speed of sound, which one gets by varying the pressure and density
adiabatically. It is independent of whether the actual perturbations we study are adiabatic or not.
8 STRUCTURE FORMATION 59

This result, Eq. (204), applies before decoupling. As we go back in time, ρ̄b /ρ̄γ → 0 and c2s → 1/3.
As we approach decoupling, ρ̄b becomes comparable to (but still smaller than) ρ̄γ and the speed of sound
falls, but not by a large factor.
Newtonian perturbation theory applies only to subhorizon scales. The ratio of the (comoving) baryon
Jeans length
2πcs
λJ = √
a 4πGρ̄b
to the comoving Hubble length
1
H−1 = !
a 8πG3 ρ̄

is "
λJ 2ρ̄
= HλJ = 2π cs .
H−1 3ρ̄b
Thus we see that before decoupling the baryon Jeans length is comparable to the Hubble length, and
thus all scales for which our present discussion applies are sub-Jeans. Therefore, if baryon perturbations
are adiabatic21 , they oscillate before decoupling22 .
After decoupling, the baryon component sees just its own pressure. This component is now a gas of
hydrogen and helium. This gas is monatomic for the epoch we are now interested in. Hydrogen forms
molecules only later. For a non-relativistic monatomic gas,
5Tb
c2s = , (205)
3m
where we can take m ≈ 1 GeV, since hydrogen dominates. Down to z ∼ 100, residual free electrons
maintain enough interaction between the baryon and photon components to keep Tb ≈ Tγ . After that
the baryon temperature falls faster,

Tb ∝ (1 + z)2 whereas Tγ ∝ 1 + z (206)

(as shown in an exercise in Chapter 4). For example, at 1 + z = 1000, soon after decoupling, Tb = 2725 K
= 0.2348 eV and the speed of sound is cs = 5930 m/s. The baryon density is ρ̄b = Ωb (1 + z)3 ρcr =
ωb (1 + z)3 1.88 × 10−26 kg/m3 , and we get for the Jeans length
√
πcs
λJ = (1 + z) √ (207)
Gρ̄b

that soon after decoupling

−1/2
λJ (1 + z = 1000) = ωb 0.96 × 103 pc = η10 0.016 Mpc ∼ 0.095 Mpc , (208)

where η10 ≡ 1010 η = 274 ωb or ωb = 0.00365 η10, and the last number is for η10 ∼ 6.
We define the baryon Jeans mass
π
MJ ≡ ρ̄b0 λ3J (209)
6
as the mass of baryonic matter within a sphere whose diameter is λJ . Note that since λJ is defined as a
comoving distance, we must use here the present (mean) baryon density ρ̄b0 . At 1 + z = 1000, the baryon
−1/2 −1/2
Jeans mass is ωb 1.3 × 105 M" = η10 2.1 × 106 M" ∼ 9 × 105 M" for η10 ∼ 6. This corresponds to
21
If there is an initial baryon entropy perturbation, i.e., a perturbation in baryon density without an accom-
panying radiation perturbation, it will initially begin to grow in the same manner as a CDM perturbation, since
the pressure perturbation provided by the photons is missing. (Such a baryon entropy perturbation corresponds
to a perturbation in the baryon-photon ratio η.) But as the movement of baryons drags the photons with them,
a radiation perturbation is generated, and the baryon perturbation begins to oscillate around its initial value
(instead of oscillating around zero).
22
We have not calculated this exactly, since all our calculations have been idealized, i.e., we have used per-
turbation theory which applies only to matter-dominated perturbations, and here we have ignored the CDM
component. But this qualitative feature will hold also in the exact calculation, and this will be enough for us
now.
8 STRUCTURE FORMATION 60

the mass of a globular cluster and is much less than the mass of a galaxy. Thus, for our purposes, the
baryonic component is pressureless after decoupling, i.e., baryon pressure can be ignored in the evolution
of perturbations at cosmological scales (greater than ∼ 1 Mpc). (The pressure cannot be ignored for
smaller scale physics like the formation of individual galaxies.)

The baryon Jeans length after decoupling is " Mpc. It would be relevant if we were
interested in the process of the formation of individual galaxies, but here we are interested in
the larger scales reflected in perturbations in the galaxy number density. Thus for our purposes,
the baryonic component is pressureless after decoupling.
After decoupling, the evolution of the baryon density perturbation is governed by the grav-
itational effect of the dominant matter component, the CDM.
We now have the situation of Sec. 8.2.10, except that we have two matter components,

ρ = ρc + ρb + ρs , (210)

where we approximate ρs = ργ + ρν + ρd as homogeneous. With the help of Sec. 8.2.8, the

discussion is easy to generalize for the present case.
We can ignore the pressure of both ρb and ρc . Therefore their perturbation equations are

δ̈c + 2H δ̇c = 4πGρ̄m δ (211)

δ̈b + 2H δ̇b = 4πGρ̄m δ (212)

where ρ̄m = ρ̄c + ρ̄b is the total background matter density and
δρc + δρb
δ= (213)
ρ̄c + ρ̄b
is the total matter density perturbation.
We can now define the baryon-CDM entropy perturbation,

Scb ≡ δc − δb , (214)

which expresses how the perturbations in the two components deviate from each other. Sub-
tracting Eq. (212) from (211) we get an equation for this entropy perturbation,

S̈cb + 2H Ṡcb = 0 . (215)

We assume that the primordial perturbations were adiabatic, so that we had δb = δc , i.e,
Scb = 0 at horizon entry. For large scales, which enter the horizon after decoupling, an Scb never
develops, so the evolution of the baryon perturbations is the same as CDM perturbations.
But for scales which enter before decoupling, an Scb develops because the baryon perturba-
tion is then coupled to the photon perturbation, whereas the CDM perturbation is not. After
decoupling, δb " δc , since δc has been growing, while δb has been oscillating. The initial con-
dition for Eqs. (211,212,215) is then Scb ∼ δc (“initial” time here being the time of decoupling
tdec ). During the matter-dominated epoch, when a ∝ t2/3 , so that H = 2/3t, the solution for
Scb is
Scb = A + Bt−1/3 , (216)
whereas for δc it is, neglecting the effect of baryons on it, from Eq. (126),

δc = Ct2/3 + Dt−1 ∼ Ct2/3 . (217)

We call the first term the “growing” and the second term the “decaying” mode (although for
Scb the “growing” mode is actually just constant). For δc the growing and decaying modes have
been growing and decaying since horizon entry, so we can now drop the decaying part of δc .
8 STRUCTURE FORMATION 61

Figure 7: Evolution of the CDM and baryon density perturbations after horizon entry (at t = tk ). The
figure is just schematic; the upper part is to be understood as having a ∼ logarithmic scale; the difference
δc − δb stays roughly constant, but the fractional difference becomes negligible as both δc and δb grow by
a large factor.

To work out the precise initial conditions, we would need to work out the behavior of Scb
during decoupling. However, we really only need to assume that initially there is no strong
cancellation between the growing and decaying modes in (216), so that Scb = δc − δb either
shrinks or stays roughly constant near the initial value of δc . While δc grows by a large factor,
δb must follow it to keep the difference close to the initial small value of δc , so that δb /δc → 1.
Thus the baryon density contrast δb grows to match the CDM density contrast δc (see Fig. 7),
and we have eventually δb = δc = δ to high accuracy.
The baryon density perturbation begins to grow only after tdec . Before decoupling the radi-
ation pressure prevents it. Without CDM it would grow only as δb ∝ a ∝ t2/3 after decoupling
(during the matter-dominated period; the growth stops when the universe becomes dark energy
dominated). Thus it would have grown at most by the factor a0 /adec = 1 + zdec ∼ 1100 after de-
coupling. In the anisotropy of the CMB we observe the baryon density perturbations at t = tdec .
They are too small (about 10−4 ) for a growth factor of 1100 to give the present observed large
scale structure23 .
With CDM this problem was solved. The CDM perturbations begin to grow earlier, at
t ∼ teq , and by t = tdec they are much larger than the baryon perturbations. After decoupling
the baryons have lost the support from photon pressure and fall into the CDM gravitational
potential wells, catching up with the CDM perturbations.
This allows the baryon perturbations to be small at t = tdec and to grow after that by much
more than the factor 103 , matching observations. This is one of the reasons we are convinced
that CDM exists.24
The whole subhorizon evolution history of all the different cosmological scales of perturba-
tions is summarized by Fig. 8.
23
This assumes adiabatic primordial perturbations, since we are seeing δγ , not δb . For a time, primordial baryon
entropy perturbations Sbγ = δb − 34 δγ were considered a possible explanation, but more accurate observations
have ruled this model out.
24
Historically, the above situation became clear in the 1980’s when the upper limits to CMB anisotropy (which
was finally discovered by COBE in 1992) became tighter and tighter. By today we have accurate detailed
measurements of the structure of the CMB anisotropy which are compared to detailed calculations including the
CDM so the argument is raised to a different level—instead of comparing just two numbers we are now comparing
entire power spectra (to be discussed later).
8 STRUCTURE FORMATION 62

Figure 8: A figure summarizing the evolution of perturbations at different subhorizon scales. The
baryon Jeans length kJ−1 drops precipitously at decoupling so that all cosmological scales became super-
Jeans after decoupling, whereas all subhorizon scales were sub-Jeans before decoupling. The wavy lines
symbolize the oscillation of baryon perturbations before decoupling, and the opening pair of lines around
them symbolize the ∝ a growth of CDM perturbations after teq . There is also an additional weaker
(logarithmic) growth of CDM perturbations between horizon entry and teq .

8.3.5 Late-time growth in the ΛCDM model

At late times, dark energy begins to accelerate the expansion, which will slow down the growth
of the density perturbation. In the ΛCDM model dark energy is just a constant vacuum energy,
so it has no perturbations and thus affects just the background. The perturbations are in CDM
and baryons, and we can ignore the pressure term in the Jeans equation, since at such small
scales where baryon pressure gradients would be important, first-order perturbation theory is
not valid anyway at late times. Thus we are facing a similar calculation as we did in Sec. 8.2.11,
the solution of Eq. (182),
δ̈k + 2H δ̇k − 4πGρ̄m δk = 0 , (218)
where 4πGρ̄m = 23 Ωm H02 a−3 , with δb = δc = δ, but instead of radiation we have now vacuum
energy contributing to the background solution, which is the Concordance Model discussed in
Cosmology I (Chapter 3):
! "1/3 # $ %
Ωm
a(t) = sinh2/3 3
2 ΩΛ H0 t . (219)
ΩΛ

The Hubble parameter is given by

$
H = H0 Ωm a−3 + ΩΛ . (220)

Again, it is better to use the scale factor as time coordinate. The difference in the power
of a in the behavior of the two density components is now 3 instead of 1, which makes the
calculation more difficult. We follow here Dodelson[9]. After the change of variable from t to a,
8 STRUCTURE FORMATION 63

(218) becomes (exercise)

! " ! "2
H! 3 3Ωm H0
δ!! + + δ! − δ = 0, (221)
H a 2a5 H

where ! ≡ d/da. The decaying solution is

#
δ ∝ H ∝ Ωm a−3 + ΩΛ (222)

and the growing solution is

$ a # $ a
dx −3
x3/2 dx
δ ∝ H ∝ Ωm a + ΩΛ % &3/2 (223)
H 3 x3 ΩΛ 3
1+ Ωm x

The effect of changing the lower limit of integration can be incorporated in the decaying solution;
so we can set the lower limit to 0. (Equation (221) is valid in general for matter perturbations
with an additional smooth background component. The first forms of the solutions (222) and
(223) are valid when the smooth component is vacuum energy or negative curvature.)
In the limit a $ 1, or rather, ΩΛ $ Ωm a−3 , the decaying solution becomes

δ ∝ a−3/2 ∝ t−1 (224)

and the growing solution becomes

δ ∝ a ∝ t2/3 (225)
the familiar results for the matter-dominated universe from Sec. 8.2.6. We can ignore the
decaying mode, since it has become completely negligible when the vacuum energy begins to
have an effect.
To fix the proportionality coefficient in the growing mode, we write it as
$ a
' −3
(1/2 x3/2 dx
δ = A Ωm a + ΩΛ % &3/2 (226)
0 ΩΛ 3
1+ Ωm x

and note that in the limit ΩΛ $ Ωm a−3 it becomes

$ a
1/2 −3/2 2
δ ≈ AΩm a x3/2 dx = Ω1/2 Aa . (227)
0 5 m

At a = a0 = 1 this would give

2 1/2 5 −1/2
Ω A ≡ δ̃ ⇒ A= Ω δ̃ , (228)
5 m 2 m

where we have defined δ̃ as the value δ would have “now”25 if there were no vacuum energy, i.e.,
the universe had stayed matter dominated.
25
Note that we defined “now” as a = a0 = 1, or in more physical terms as T = T0 = 2.7255 K; not as t = t0 .
The comparison situation ( ˜ ) we have in mind is that the early universe (where vacuum energy has no effect)
is the same as in the ΛCDM model, but there is no vacuum energy to accelerate the expansion at late times,
so that by “now” the expansion rate, i.e., H0 , is smaller than we observe in reality. The present matter density
1/2
ρm0 = (3/8πG)Ωm H02 is the same as in the ΛCDM model, but Ω̃m = 1, so H̃0 = Ωm H0 . The age of the
−1 2 −1/2 −1
universe is t̃0 = 3 H̃0 = 3 Ωm H0 , which for h = 0.7 and Ωm = 0.3 gives t̃0 = 17.0 × 109 years, instead of the
2

t0 = 13.5 × 109 years of the ΛCDM model.

8 STRUCTURE FORMATION 64

Thus we write (226) as

! "1/2 # a
5 −3 ΩΛ x3/2 dx
δ = δ̃ a + $ %3/2 . (229)
2 Ωm 0 ΩΛ 3
1+ Ωm x

Unfortunately, the integral in (229) does not give an elementary function. (I think it is a so-
called hypergeometric function, which does not give much useful information compared to just
integrating (229) numerically.) We can see that at late (future) times, when a ! 1, there is very
little growth, since the factor outside the integral approaches a constant and for any a1 ! 1 and
a2 ! 1, the contribution to the integral,
# a2 ! " # ! "
x3/2 dx Ωm 3/2 a2 −3 1 Ωm 3/2 & −2 '
$ %3/2 ≈ x dx = a1 − a−22 (230)
a1 ΩΛ 3 ΩΛ a1 2 ΩΛ
1 + Ωm x

is very small.
It turns out that the integral can be done if we extend it to the infinite future (exercise) :
As a → ∞,
! " # ! "
5 −3 ΩΛ 1/2 ∞ x3/2 dx 5 Ωm 1/3
δ → δ(∞) ≡ 2 δ̃ a + $ %3/2 = 6 δ̃ Ω B( 56 , 23 ) , (231)
Ωm 0 ΩΛ 3 Λ
1 + Ωm x

where # 1
Γ(p)Γ(q)
B(p, q) ≡ tp−1 (1 − t)q−1 dt = (232)
0 Γ(p + q)
is the beta function and &5 '
2
B 6, 3 ≈ 1.725 . (233)
Thus the perturbations “freeze”, i.e., approach a final value
! "
Ωm 1/3
δ(∞) = 1.437 δ̃ . (234)
ΩΛ
which for Ωm = 0.3, ΩΛ = 0.7 gives

δ(∞) = 1.084 δ̃ , (235)

i.e., the perturbations will never become much stronger than what they in the matter-dominated
model would be already “now”. To get the present density perturbation δ(a = 1) one has to do
(229) numerically. This is done in Fig. 9, from which one can read that δ(a = 1) ≈ 0.78 δ̃.
For perturbations that entered horizon well before matter-radiation equality teq , we have
from (199) that ! " ! "
3 k
δ̃ ≈ δprim 1 + 7.5 ln 0.17 , (236)
2aeq keq
assuming that this is still ' 1 so that first-order perturbation theory remains valid.
For Ωm = 0.3, ΩΛ = 0.7, h = 0.7, we have keq −1 = 65 h−1 Mpc and a
eq = 1/3514. Equa-
−1 −1
tion (236) gives then for the scale k = 8 h Mpc,

δ̃ ≈ 13 000 δprim and δ(a = 1) ≈ 10 000 δprim . (237)

The contribution of the boost from radiation oscillation is a factor 7.5 ln(0.17k/keq ) ≈ 2.4 at
this scale (and for smaller scales it is more). Actually, this scale is still too large (too low
8 STRUCTURE FORMATION 65

k) for (236) to apply; in reality the factor is somewhat larger.26 We chose the reference scale
k−1 = 8 h−1 Mpc, since observationally, the variance σT2 of the top-hat-filtered density field of the
galaxy distribution today is ≈ 1 at this scale. Because of the galaxy bias bg , the corresponding
variance for the matter distribution is less by a factor b−2g , but still not far from 1, meaning
that the linear perturbation theory approximation is beginning to break. Because of nonlinear
effects, the perturbation today is somewhat larger than our prediction from linear theory. We
noted earlier that the perturbations entered the horizon with amplitude P(k)1/2 ≈ few × 10−5 .
Thus today the amplitude at k = 1/8 h−1 Mpc should be somewhat more than few × 10−1 . From
Fig. 4 we see that σT2 is typically a bit more that 2 times P at the same scale (depending on the
shape of P(k)). So indeed we get a prediction that it should be somewhat less than 1. We will
do this comparison more quantitatively later.

8.3.6 Growth function

Inside the horizon, after photon decoupling the linear growth of matter perturbations is inde-
pendent of scale (once the decaying mode has died out and ignoring the subcosmological scales
where pressure gradients have a role). Thus it can be described by a function that depends on
time (or scale factor, or redshift) only, called the growth function,
δ(a)
D(a) ≡ (238)
δref
where δ(a) is the density perturbation (δk or δ(x); D(a) is the same function for any k or x)
when scale factor is a and δref is it at some reference time. The choice of reference time fixes
the normalization of D. During matter domination, D(a) ∝ a and a common normalization is
to normalize so that D(a) = a during matter domination. This corresponds to setting δref = δ̃.
So that in the ΛCDM model
! " #
5 −3 ΩΛ 1/2 a x3/2 dx
D(a) = a + $ %3/2 , (239)
2 Ωm 0 ΩΛ 3
1 + Ωm x

(from the onset of matter domination).

We define the growth rate
d ln D d ln δ a dδ
f≡ = = , (240)
d ln a d ln a δ da
which is independent of this normalization.
For the ΛCDM model of Sec. 8.3.5, we get from (229) (exercise)
 
& '
1  3
5 δ̃ 3 1  a 
f (a) = ΩΛ 3 2aδ − 2 = ΩΛ 3  $ %1/2 + −  . (241)
1 + Ωm a 1 + Ωm a  a x3/2 dx 2
a−3 + ΩΛ Ωm 0 ! Ω
"3/2
1+ Ω Λ x3
m

It turns out that a good approximation to (241) is

f (a) ≈ Ωm (a)γ , where γ = 0.55 , (242)
where γ is called the growth index. (We have assumed General Relativity, and the measurement
of the growth index from galaxy surveys is a way of testing gravity theory.) We plot D, f , and
the approximation (242) for ΛCDM in Fig. 9.
26
One should instead use the BBKS transfer function (which still ignores effect of baryons) here, which gives
a larger factor. On the other hand, baryonic effects decrease the result somewhat; but the net effect is a larger
factor than 2.4. These corrections will be discussed in Sec. 8.4.4.
8 STRUCTURE FORMATION 66

Figure 9: The growth function D(a) (blue, with normalization δref = δ̃), matter density parameter Ωm (a)
(black), growth rate f (a) (red), and the approximation (242) (red, dashed) for ΛCDM with Ωm = 0.3.

8.4 Relativistic perturbation theory

For scales comparable to, or larger than the Hubble scale, Newtonian perturbation theory does
not apply, because we can no more ignore the curvature of spacetime. Therefore we need to
use (general) relativistic perturbation theory. Instead of the Newtonian equations of gravity
and fluid mechanics, the fundamental equation is now the Einstein equation of general relativity
(GR). We assume a background solution, which is homogeneous and isotropic, i.e., a solution of
the Friedmann equations, and study small perturbations around it. This particular choice of the
background solution means that we are doing a particular version of relativistic perturbation
theory, called cosmological perturbation theory.
The evolution of the perturbations while they are well outside the horizon is simple, but the
mathematical machinery needed for its description is complicated. This is due to the coordinate
freedom of general relativity. For the background solution we had a special coordinate system
(time slicing) of choice, the one where the t = const slices are homogeneous. The perturbed
universe is no more homogeneous, it is just ”close to homogeneous”, and therefore we no more
have a unique choice for the coordinate system. We should now choose a coordinate system
where the universe is close to homogeneous on the time slices, but there are many different
possibilities for such slicing. This freedom of choosing the coordinate system in the perturbed
universe is called gauge freedom, and a particular choice is called a gauge.27 The most important
part of the choice of gauge is the choice of the time coordinate, because it determines the slicing
of the spacetime into t = const slices, ”universe at time t”. Sometimes the term ‘gauge’ is used
to refer only to this slicing.
27
If you are familiar with gauge field theories, like electrodynamics, the concept of ‘gauge’ may look different
here. The mathematical similarity appears when the perturbation equations are developed. In relativistic per-
turbation theory gauge has this geometric origin (this is where the use of the word “gauge” comes from), unlike
in electrodynamics.
8 STRUCTURE FORMATION 67

Because the perturbations are defined in terms of the chosen coordinate system, they look
different in different gauges. We can, for example, choose the gauge so that the perturbation in
one scalar quantity, e.g., proper energy density, disappears, by choosing the ρ = const 3-surfaces
as the time slices (this is called ”the uniform energy density gauge”).
The true nature of gravitation is spacetime curvature, so perturbations should be described
in terms of curvature.
We leave the actual development of cosmological perturbation theory to a more advanced
course (Cosmological Perturbation Theory, lectured in spring 2020), and just summarize here
some basic concepts and results.
In the Newtonian theory gravity was represented by a single function, the gravitational
potential Φ. In GR, gravity is manifested in the geometry of spacetime, described in terms of
the metric. Thus in addition to the density, pressure, and velocity perturbations, we have a
perturbation in the metric. The perturbed metric tensor is

gµν = ḡµν + δgµν . (243)

For the background metric, ḡµν , we choose that of the flat Friedmann-Robertson-Walker
universe,
ds2 = ḡµν dxµ dxν = −dt2 + a(t)2 (dx2 + dy 2 + dz 2 ) (244)
The restriction to the flat case is an important simplification, because it allows us to Fourier
expand our perturbations in terms of plane waves.28 Fortunately the real universe appears to be
flat, or at least close to it. And earlier it was even flatter. Inflation predicts a very flat universe.
For the metric perturbation, we have now 10 functions δgµν (t, x). So there appears to be ten
degrees of freedom. Four of them are not physical degrees of freedom, since they just correspond
to our freedom in choosing the four coordinates. So there are 6 real degrees of freedom.
Two of these metric degrees of freedom couple to density and pressure perturbations and
the irrotational velocity perturbation. These are the scalar perturbations. Two couple to the
rotational velocity perturbation to make up the vector perturbations. The remaining two are not
coupled to the cosmic fluid at all29 , and are called tensor perturbations. They are gravitational
waves, which do not exist in Newtonian theory.
The vector perturbations decay in time, and are not produced by inflation, so they are the
least interesting.
Although the tensor perturbations also are not related to growth of structure, they are
produced in inflation and affect the cosmic microwave background (CMB) anisotropy and polar-
ization. Different inflation models produce tensor perturbations with different amplitudes and
spectral indices (to be explained later), so they are an important diagnostic of inflation. No
tensor perturbations have been detected in the CMB so far, but they could be detected in the
future with more sensitive instruments if their amplitude is large enough.30
The three kinds of perturbations evolve independently of each other in linear perturbation
theory, so they can be studied separately. We shall first concentrate on the scalar perturbations,
returning to the tensor perturbations later.
28
In the Newtonian case this restriction was not necessary, and we could apply it to any Friedmann model,
as there is no curvature of spacetime in the Newtonian view, and only the expansion law a(t) of the Friedmann
model is used. The Newtonian theory of course is only valid for small scales were the curvature can indeed be
ignored.
29
This is true in first-order perturbation theory in the perfect fluid approximation, but not in general.
30
Typically, large-field inflation models produce tensor perturbations with much larger amplitude than small-
field inflation models. In the latter case they are likely to be too small to be detectable.
8 STRUCTURE FORMATION 68

8.4.1 Gauges for scalar perturbations

Consider now scalar perturbations. The gauges discussed in the following assume scalar pertur-
bations.
Perturbations appear different in different gauges. When needed, we use superscripts to
indicate in which gauge the quantity is defined: C for the comoving gauge and N for the
Newtonian gauge. Some other gauges are the synchronous gauge (S), spatially flat gauge (Q),
and the uniform energy density gauge (U).
There are two common ways to specify a gauge, i.e., the choice of coordinate system in the
perturbed universe:

• A statement about the relation of the coordinate system to the fluid perturbation. This
will lead to some condition on the metric perturbation.

• A statement about the metric perturbation. This will then lead to some condition on the
coordinate system.

The two gauges (C and N) we shall refer to in the following, give an example of each.
The comoving gauge is defined so that the space coordinate lines x = const follow fluid flow
lines, and the time slice, the t = const hypersurface, is orthogonal to them. Thus the velocity
perturbation is zero in this gauge,
vC = 0 . (245)
The conformal-Newtonian gauge, also called the longitudinal gauge, or the zero-shear gauge,
and sometimes, for short, just the Newtonian gauge, is defined by requiring the metric to be of
the form
ds2 = −(1 + 2Φ)dt2 + a2 (1 − 2Ψ)(dx2 + dy 2 + dz 2 ) . (246)
This means that we require

δg0i = 0, δg11 = δg22 = δg33 , and gij = 0 for i "= j . (247)

(This is possible for scalar perturbations). The two metric perturbations, Φ(t, x) and Ψ(t, x) are
called Bardeen potentials.31 Φ is also called the Newtonian potential, since in the Newtonian limit
(k # H and p $ ρ), it becomes equal to the Newtonian gravitational potential perturbation.
Thus we can use the same symbol for it. Ψ is also called the Newtonian curvature perturbation,
because it determines the curvature of the 3-dimensional t = const subspaces, which are flat in
the unperturbed universe (since it is the flat FRW universe).
It turns out that the difference Φ − Ψ is caused only by anisotropic stress (or anisotropic
pressure). We shall here consider only the case of a perfect fluid. For a perfect fluid the pressure
(or stress) is necessary isotropic. Thus we have only a single metric perturbation32

Ψ=Φ (248)

The density perturbations in these two gauges become equal in the limit k # H (inside
horizon), and we can then identify them with the “usual” density perturbation δ of Newtonian
theory.
31
Warning: The sign conventions for Ψ differ, and many authors call them Ψ and Φ instead.
32
In reality, neutrinos develop anisotropic pressure after neutrino decoupling. Therefore the two Bardeen
potentials actually differ from each other by about 10 % between the times of neutrino decoupling and matter-
radiation equality. After the universe becomes matter-dominated, the neutrinos become unimportant, and Ψ and
Φ rapidly approach each other. The same happens to photons after photon decoupling, but the universe is then
already matter-dominated, so they do not cause a significant Ψ − Φ difference.
8 STRUCTURE FORMATION 69

8.4.2 Evolution at superhorizon scales

When the perturbations are outside the horizon (meaning that the wavelength of the Fourier
mode we are considering is much longer than the Hubble length), very little happens to them,
and we can find quantities which remain constant for superhorizon scales. Such a quantity is
the (comoving gauge) curvature perturbation R(x), which describes how curved is the t = const
slice in the comoving gauge.33 For adiabatic perturbations, the curvature perturbation R stays
constant in time outside the horizon.
Using gauge transformation equations R can be related to the metric in the Newtonian
gauge. The result is
5 + 3w 2
R = − Φ− H −1 Φ̇ , (251)
3 + 3w 3 + 3w
where w ≡ p̄/ρ̄.
Because Rk stays constant while k # H, it is a very useful quantity for “carrying” the
perturbations from their generation at horizon exit during inflation to horizon entry at later
times. We now define the primordial perturbation to refer to the perturbation at the epoch
when it is well outside the horizon. For adiabatic perturbations, the primordial perturbation
is completely characterized by the set of these constant values Rk . We shall later discuss how
the primordial perturbation is generated by inflation, and how these superhorizon values Rk are
determined by it.
However, we would like to describe the perturbation in more “familiar” terms, the gravita-
tional potential perturbation Φ and the density perturbation δ. While Rk remains constant this
turns out to be easy. Eq. (251) can be written as a differential equation for Φk ,
2 −1 5 + 3w
H Φ̇k + Φk = −(1 + w)Rk . (252)
3 3
During any period, when also w = const, the solution of this equation is
3 + 3w
Φk = − Rk + a decaying part . (253)
5 + 3w
Thus, after w has stayed constant for some time, the Bardeen potential has settled to the
constant value
3 + 3w
Φk = − Rk (w = const ) . (254)
5 + 3w
In particular, we have the relations
2
Φ k = − Rk (rad.dom, w = 31 ) (255)
3
3
Φ k = − Rk (mat.dom, w = 0) . (256)
5
33
Technically, R is defined in terms of the trace of the space part of the comoving gauge metric perturbation
(−Ψ is the corresponding quantity in the Newtonian gauge), and it is related to the scalar curvature (3) RC of the
comoving gauge time slice (the (3) reminds us that we are considering a 3-dimensional subspace, and the C refers
to the comoving gauge) so that
(3) C
R = −4a−2 ∇2 R . (249)
For Fourier components we have then that
1 ! a "2 (3) C
Rk ≡ Rk . (250)
4 k
Another similar quantity is the (uniform-density-gauge) curvature perturbation ζ that is defined the same way,
but for the uniform-density-gauge time slice. For superhorizon scales they are equal, R = ζ (in the limit k $ H).
8 STRUCTURE FORMATION 70

8.4.3 From outside to inside horizon

After the perturbation has entered horizon, we can use the Newtonian perturbation theory
result, Eq. (109), which gives the density perturbation as
! "2 ! " ! "
k Φk 2 k 2 2 k 2
δk = − = − Φk = − Φk , (257)
a 4πGρ̄ 3 aH 3 H

where we used the background relation

8πG 3
H2 = ρ̄ ⇒ 4πGρ̄ = H 2 . (258)
3 2

The problem is to get Φk from the superhorizon epoch where it is constant (as long as
w = const) through the horizon entry to the subhorizon epoch where it evolves according to
Newtonian theory. We do this for the two cases, large (k # keq ) and small (k $ keq ) scales,
below.
Large scales. For scales k which enter while the universe is matter dominated, this is easy,
since in this case Φk stays constant the whole time (until dark energy becomes important).
Thus we can relate these constant values of Φk , and the corresponding subhorizon density
perturbations δk during the matter-dominated epoch to the primordial perturbations Rk by
3
Φk = − Rk (mat.dom)
5
! " ! " (259)
2 k 2 2 k 2 1
δk = − Φk = Rk ∝ ∝ t2/3 ∝ a
3 H 5 H (aH)2

Note that by Rk we refer always to the constant primordial value, when we use it in equations,
like (259), that give other quantities at later times.
Small scales. For perturbations which enter during the radiation-dominated epoch, the
potential Φk does not stay constant. We learned earlier, that in this case the radiation density
perturbation oscillates with roughly constant amplitude, which means that the amplitude for the
potential Φ must decay ∝ a2 ρ̄ ∝ a−2 . This oscillation applies to the baryon-photon fluid, whereas
the CDM density perturbation grows slowly. After the universe becomes matter dominated, it
is these CDM perturbations that matter.
We shall now make a crude estimate how the amplitudes of these smaller-scale perturbations
during the matter-dominated epoch are related to the primordial perturbations (in particular,
we ignore the slow growth of the CDM perturbation during the radiation-dominated epoch).
These perturbations enter during the radiation-dominated epoch. Assume that the relation
Φk = − 23 Rk holds all the way to horizon entry (k = H). Assume then that the Newtonian
relation (257) holds already. Then
! "2
2 k 2 4
δk ≈ − Φk = − Φk ≈ Rk (260)
3 H 3 9

at horizon entry. The universe is now radiation-dominated, and therefore δrk = δk . We are
assuming primordial adiabatic perturbations and therefore the adiabatic relations δc = 34 δr ,
δγ = δr hold at superhorizon scales. Assume that these relations hold until horizon entry. After
that δγk begins to oscillate, whereas δck grows slowly. Thus we have that at horizon entry

3 1
δck ≈ δk ≈ Rk . (261)
4 3
8 STRUCTURE FORMATION 71

Ignoring the slow growth of δc we get that δck stays at this value until the universe becomes
matter-dominated at t = teq , after which we can approximate δk ≈ δck and δk begins to grow
according to the matter-dominated law, ∝ a ∝ 1/H2 .
Thus
1
δk (teq ) ≈ Rk (262)
3
and ! " ! "
1 Heq 2 1 keq 2
δk (t) ≈ Rk = Rk for t > teq , (263)
3 H 3 H
as long as the universe stays matter dominated.

8.4.4 Transfer function

For large scales (k # keq ) which enter the horizon during the matter-dominated epoch, we got
! "
2 k 2
δk (t) = Rk (k # keq ) , (264)
5 H
for as long as the universe stays matter dominated.
This is a simple result, and we use this as a reference for the more complicated result at
smaller scales. That is, we define a transfer function T (k, t) so that
! "
2 k 2
δk (t) = T (k, t)Rk (265)
5 H
where Rk refers to the primordial perturbation. Thus by definition T (k, t) = 1 for k # keq .34
Using the rough estimate from the previous subsection we get, comparing (263) to (259),
that ! "
5 keq 2
T (k, t) ≈ (266)
6 k
during the matter-dominated epoch, where we can drop the factor 65 , since this is anyway just
a rough estimate.
Once we are well into the matter-dominated era, perturbations at all scales grow ∝ a ∝
1/(aH)2 and the transfer function becomes independent of time,35
T (k) = 1 k # keq
! "
keq 2
T (k) ∼ k % keq (267)
k
A more accurate calculation, including the gravitational effect of radiation perturbation os-
cillations on the CDM perturbation (see Sec. 8.3.3), assuming adiabatic primordial perturbations
and still ignoring baryons, adds a logarithmic growth factor, the ratio between (199) and the
δ = δprim (a/aeq ) of (263), and gives36
! " ! "
3 keq 2 0.17k
T (k) ≈ × 7.5 ln k % keq , (269)
2 k keq
34
With the given definition for T (k, t), this holds for t ! t0 , i.e, before we entered the present dark-energy-
dominated epoch.
35
We shall later define other transfer functions, but this is the transfer function T (k) of structure formation
theory. It relates the perturbations inside the horizon during the matter-dominated epoch to the primordial
perturbations, and it is independent of time.
36
This calculation is presented in Dodelson[9] (Sections 7.3 and 7.4). Dodelson (7.69) gives the result as
2 ! "
12keq k
T (k) = ln . (268)
k2 8keq
For some reason I get the somewhat different numerical factors in (269) when I do the same calculation.
8 STRUCTURE FORMATION 72

Figure 10: Transfer function T (k) for CDM, adiabatic primordial fluctuations. The black curve is the
BBKS transfer function (270), the red curve is the small-scale approximate analytical result (269) (the
dotted red curve is the Dodelson version (268)), the two black dotted lines correspond to (267), and the
green vertical line gives k = keq . The k scale is for our reference model, Ωm = 0.3, h = 0.7, for which
keq = 0.0153 h/Mpc = 1/(65 h−1 Mpc).

where we approximated 1 + a/aeq ≈ a/aeq (appropriate for application at late times, a " aeq ).
Note that that logarithm is negative for k ! 6keq ; the equation is not supposed to apply yet for
this low k.
To include the intermediate scales, which enter close matter-radiation equality, requires
numerical computation. For ωb # ωc (i.e., still essentially ignoring baryons), Bardeen, Bond,
Kaiser, and Szalay [10] gave a fitting formula, the BBKS transfer function

ln(1 + 2.34q) 1
T (k) = , (270)
2.34q [1 + 3.89q + (16.1q)2 + (5.64q)3 + (6.71q)4 ]1/4

where q = 0.073(k/keq ), to such numerical results. See Fig. 10 for these results. The slope of
the BBKS transfer function is
d ln T 2.34q 1 3.89q + 2(16.1q)2 + 3(5.64q)3 + 4(6.71q)4
= − − 1 . (271)
d ln q (1 + 2.34q) ln(1 + 2.34q) 4 1 + 3.89q + (16.1q)2 + (5.64q)3 + (6.71q)4
For later reference, we note that for our reference model, Ωm = 0.3, h = 0.7, this gives
d ln T /d ln q = −1.184 at k = 1/(8 h−1 Mpc).
According to present understanding, the universe becomes dark energy dominated as we
approach the present time. The equation-of-state parameter w begins to decrease (becomes
negative) and therefore Φ begins to change again. The growth of the density perturbations is
slowed down as we saw in Secs. 8.3.5 and 8.3.6. Since this affect all scales the same way, we can
model this with the growth function D(a), and keep the transfer function T (k) unaffected.
We are still missing the effect of baryons. There are publicly available computer programs
(such as CMBFAST, CAMB37 , and CLASS38 ; you give your favorite values for the cosmological
37
https://camb.info/
38
http://class-code.net/
8 STRUCTURE FORMATION 73

Figure 11: Left: Transfer function T (k) calculated with CAMB (blue curve) for adiabatic primordial
fluctuations in the flat ΛCDM model with ωb = 0.023, ωc = 0.124, h = 0.7 (so that Ωm = 0.3), and
massless neutrinos (a neutrino mass 0.06 eV for one neutrino species changes the transfer function by less
than the width of the curve). The black curve is the BBKS transfer function (270), the two black dotted
lines correspond to (267), and the green vertical line gives k = keq , as in Fig. 10. The main difference
from BBKS is due to baryons. Right: The ratio (blue) of the T (k) from CAMB to the BBKS transfer
function.

parameters as input) that include it and other small physical effects we have ignored. They
represent the current state of the art. The exact result can be given in form of the transfer
function T (k) we defined above. We show in Fig. 11 a transfer function calculated with CAMB.
The effect of baryon acoustic oscillations (i.e., the oscillations of δbγ before decoupling, which
leave a trace in δb ) shows up as a small-amplitude wavy pattern in the k > keq part of the
transfer function, since different modes k were at a different phase of the oscillation when that
ended around tdec .
Everything has still been calculated using linear perturbation theory. Linear perturbation
theory breaks down when the perturbations become large, δ(x) ∼ 1. We say that the pertur-
bation becomes nonlinear. This has happened for the smaller scales, k−1 < 10 Mpc by now.
Nonlinear effects speed up the growth of density perturbations. They cannot be captured in
a transfer function and a growth function, since now the ratio between the present-day and
primordial power spectra depends on the primordial power spectrum. CAMB can also calculate
nonlinear effects but within a more restricted set of cosmological models, because this is based
on results from N -body simulations.
When the perturbation becomes sufficiently nonlinear, i.e., an overdense region becomes
significantly (a few times) denser as the average density of the universe (see Sec. 8.5.2), it
collapses and forms a gravitationally bound structure, e.g. a galaxy or a cluster of galaxies.
Further collapse is prevented by the angular momentum of the structure. Galaxies in a cluster
and stars (and CDM particles) in a galaxy orbit around the center of mass of the bound structure.
8 STRUCTURE FORMATION 74

8.4.5 Tensor perturbations

In addition to scalar and vector perturbations, in general relativistic perturbation theory we have
tensor perturbations. They have the nice property that we do not have to worry about different
gauges, since they are gauge invariant in the sense that, if we first do a gauge transformation and
then separate out the scalar, vector, and tensor parts, the tensor part has remained unchanged.
These are perturbations of the metric that for one Fourier mode take the form
! "
ds2 = −dt2 + a(t)2 (1 + h)dx2 + (1 − h)dy 2 + dz 2
! "
= a(η)2 −dη 2 + (1 + h)dx2 + (1 − h)dy 2 + dz 2 (272)

where
h = hk (t)eikz (273)
is the perturbation and η is conformal time. In (272) we have chosen the z axis in the direction
of the wave vector, so that k = k k̂ and k·x = kz. Since the metric is a real quantity, in (272) and
(278) h should be interpreted as the real part of h; like one should always do when one makes
physical interpretations for a single Fourier mode. Remember that when one sums over Fourier
components the imaginary parts of hk (t)eikz + h−k (t)e−ikz cancel since h−k = h∗k , and thus the
imaginary parts have no physical significance, they are just a mathematical convenience.
The effect of the tensor perturbation is to stretch space in one direction (here x if h is
positive) and compress it in the other direction (here y) orthogonal to the wave vector of the
Fourier mode. In (272) we also chose the orientation of the x and y axes so that they correspond
to these stretch/compress directions. But of course the perturbation could be oriented differently.
We get the other possibilities by rotating the pattern around the wave vector k by some angle
ϕ, which is mathematically equivalent to rotating the coordinate system by angle −ϕ.
In matrix form the metric is
 
−1
 1+h 
[gµν ] = a2 

 (274)
1−h 
1
After rotation by ϕ around the z axis it becomes
   
1 −1 1
 cos ϕ − sin ϕ   1+h  cos ϕ sin ϕ 
[gµν ] = a2 
 sin ϕ cos ϕ





 (275)
1−h − sin ϕ cos ϕ
1 1 1
√
Rotation by 45◦ , i.e., cos ϕ = sin ϕ = 1/ 2, gives
 
−1
 1 h 
[gµν ] = a2 


 (276)
h 1
1

We call (274) the + mode and (276) the × mode. An arbitrary orientation of the stretch/compress
pattern can be obtained as a linear combination of these two modes, so that the general form
of the tensor perturbation is
 
−1
 1 + h+ h× 
[gµν ] = a2 

 (277)
h× 1 − h+ 
1
8 STRUCTURE FORMATION 75

or
! "
ds2 = −dt2 + a(t)2 (1 + h+ )dx2 + 2h× dxdy + (1 − h+ )dy 2 + dz 2
! "
= a(η)2 −dη 2 + (1 + h+ )dx2 + 2h× dxdy + (1 − h+ )dy 2 + dz 2 (278)

for a Fourier mode in the z direction. Thus we have two Fourier amplitudes h+k (t) and h×k (t)
for each wave vector k. In the following we mostly write just h(t) to represent an arbitrary such
mode.
The evolution equation for h(t),
# $2
k
ḧ + 3H ḣ + h=0 ⇔ H −2 ḧ + 3H −1 ḣ + (k/H)2 h = 0 , (279)
a

can be obtained from the Einstein equation. This derivation is beyond the level of this course,
but the equation has a simple and plausible form: it is the wave equation with a damping term
3H ḣ; the wave velocity is the speed of light = 1.
For superhorizon scales we can ignore the last term, and we get h = const as a solution
and another solution where ḣ ≡ dh/dt ∝ a−3 so it also approaches a constant. Thus tensor
perturbations remain essentially constant outside the horizon.
For evolution inside the horizon we get oscillatory solutions and then it is better to work
with conformal time. The h(η) evolution equation is

h## + 2Hh# + k2 h = 0 ⇔ H−2 h## + 2H−1 h# + (k/H)2 h = 0 , (280)

where # ≡ d/dη. If we first ignore the middle term, we get solutions of the form h ∝ e±ikη , where
− represents a wave moving in the k direction and + in the −k direction. These are gravitational
waves. They propagate at the speed of light and they are transverse waves. During one half-
period of the wave oscillation, space is stretched in one direction orthogonal to the direction of
propagation, and compressed in the other orthogonal direction. During the next half-period the
opposite happens. The amplitude of the stretching is given by h, meaning that the maximum
stretching is by factor 1 + |h| and the maximum compression is by factor 1 − |h|.
The middle term in (280) represents the damping of gravitational waves due to the expansion
of the universe. Write
h(η) = A(η)e−ikη (281)
and insert this into (280) to get

A## + 2HA# − 2ik(A# + HA) = 0 . (282)

For k % H, the part 2ik(A# + HA) dominates the left-hand side, and we get

a# 1
A# + HA = A# + A = (aA)# = 0 ⇒ aA = const ⇒ A ∝ a−1 . (283)
a a
Thus gravitational waves are damped inside the horizon as a−1 independent of the expansion
law.
For simple expansion laws one can also solve Eq. (280) exactly, covering also horizon en-
try/exit. These solutions are Bessel functions.
8 STRUCTURE FORMATION 76

8.5 Nonlinear growth

When δ grows the evolution becomes nonlinear, requiring a more complicated discussion. One
can get further with higher-order perturbation theory or something called the Zeldovich approx-
imation, but eventually one has to resort to numerical simulations. We shall not discuss these
in this course. The spherically symmetric special case can be done analytically by basing it on
solutions for FRW universes with different densities. We do it below for an overdensity in a flat
matter-dominated background universe.

8.5.1 Closed Friedmann model

In Cosmology I we derived the expansion law for the closed (Ω > 1) matter-dominated FRW
universe. It cannot be given in closed form as a(t), but can be given in terms of an auxiliary
variable, the development angle ψ, as

Ωi Ω(ψ)
a(ψ) = ai (1 − cos ψ) = a(ψ) (1 − cos ψ)
2(Ωi − 1) 2[Ω(ψ) − 1]
Ωi Ω(ψ)
t(ψ) = Hi−1 (ψ − sin ψ) = H(ψ)−1 (ψ − sin ψ) , (284)
2(Ωi − 1)3/2 2[Ω(ψ) − 1]3/2

where ai , Ωi , and Hi are the scale factor, density parameter, and Hubble parameter at some
reference time ti (usually chosen as the present time t0 , but below we will instead choose ti to
be some early time, when Ω is still very close to 1). In the second forms we took advantage of
the fact that we can choose ti to be any time during the development and replaced it with the
“current” time. See Fig. 12 for the shape of a(t). This curve is called a cycloid. (It is the path
made by a point at the rim of a wheel.) From (284) we solve
2
Ω(ψ) = . (285)
1 + cos ψ

Calculating da/dt = da/dψ × dψ/dt we find (exercise)

(Ωi − 1)3/2 sin ψ

H(ψ) = 2Hi . (286)
Ωi (1 − cos ψ)2

The matter density is given by

! "3
ai (Ωi − 1)3
ρ(ψ) = ρi = 8ρi . (287)
a(ψ) Ω3i (1 − cos ψ)3

The scale factor reaches a maximum ata (and the density a minimum) at the “turnaround”
time tta , when ψ = π, so that

Ωi π −1 Ωi (Ωi − 1)3
ata = ai , tta = Hi , and ρ(tta ) = ρi . (288)
Ωi − 1 2 (Ωi − 1)3/2 Ω3i

At this point H = 0 and then the universe begins to shrink. Since

3Ωi Hi2 3π
ρi = we have ρ(tta ) = . (289)
8πG 32Gt2ta

The universe collapses at tcoll = 2tta , when ψ = 2π and a = 0 again.

8 STRUCTURE FORMATION 77

Figure 12: The expansion law for the flat matter-dominated universe (blue) and for closed matter-
dominated universes with different initial values Ωi > 1 for the density parameter. Both axes are linear,
the units are arbitrary.

8.5.2 Spherical collapse

The expansion law (284) will hold also for a spherically symmetric overdense region within a
flat (Ω = 1) matter-dominated FRW universe. Denote the quantities for this flat background
universe by ā, H̄, ρ̄. (Time t is the same for both solutions and Ω̄ = 1, so we don’t need notations
for them.) The background universe has
! "2
2 8πG 2 1
H̄ = ρ̄ = ⇒ ρ̄ = (290)
3 3t 6πGt2

Thus we see that at tta , the density of the overdense region is

9π 2
ρ(tta ) = ρ̄(tta ) ≈ 5.5517ρ̄(tta ) , (291)
16
i.e., at the turnaround time the density contrast has the value

9π 2
δta = − 1 ≈ 4.5517 . (292)
16
Until then the overdense region has been expanding, although slower than the surrounding
background universe. At turnaround the overdense region begins to shrink (in terms of proper
distance).
The preceding applies both for an overdense region with homogeneous density and for one
with a spherically symmetric density profile. In the latter case, we have to apply it separately
for each spherical shell, and the density ρ refers, not to the density of the shell, but to the mean
density within the shell, as it is the total mass within the shell that is responsible for the gravity
affecting the expansion or contraction of the shell. To avoid shell crossing the density profile
8 STRUCTURE FORMATION 78

has to decrease outward, so that outer shells do not collapse before inner shells.39
In linear perturbation theory, which applies when δ ! 1, density perturbations in the flat
matter-dominated universe grow as
δlin ∝ a ∝ t2/3 . (293)
When the density contrast δ becomes large it begins to grow faster. Compare now the linear
growth law to the above result for δ at turnaround.
The initial density contrast δi is given by ρi = (1 + δi )ρ̄i . On the other hand
8πG 8πG
H̄i2 = ρ̄i and Hi2 = Ω i ρi (294)
3 3
so that
Hi2 H2
1 + δi = Ωi or at any time 1+δ =Ω . (295)
H̄i2 H̄ 2
Thus the density contrast is not simply given by Ω− Ω̄ = Ω−1, since also the Hubble parameters
are different for the two solutions. We can sort out the separate contributions from Ωi − 1 and
(Hi /H̄i )2 at an early time when Ω − 1 ! 1 and ψ ! 1, by expanding Ω, H and H̄ from (285),
(286) and (290&284) in terms of ψ (exercise) to get

Hi2
Ωi ≈ 1+ 41 ψi2 and 1 2
≈ 1− 10 ψ ⇒ 3 2
1+δi ≈ 1+ 20 ψ ⇒ δi ≈ 35 (Ωi −1) . (296)
H̄i2

We can now give the linear prediction for the density contrast at turnaround time40 :
! "2/3 ! "2/3 ! "2/3
lin āta tta 3π δi 3 3π
δta = δi = δi ≈ ≈ ≈ 1.0624 , (297)
āi ti 4 Ωi − 1 5 4

where we approximated
π −1 1
tta ≈ H̄i and ti = 23 H̄1−1 . (298)
2 (Ωi − 1)3/2

Thus we conclude that density perturbations begin to collapse when the linear prediction is
δ ∼ 1, at which time the true density perturbation is already over 4 times stronger.
The collapse is completed at tcoll = 2tta , when the linear prediction gives
lin
δcoll = 22/3 δta
lin
≈ 1.6865 . (299)

The above special case can be extended to the situation where the background universe is a
closed or open Friedmann model (i.e., a matter-dominated FRW universe), and to the ΛCDM
model, with more complicated math.

8.5.3 Without spherical symmetry

I suppose these idealized cases would lead to a supermassive black hole at the center of symmetry
(for perturbations at cosmological scales, for a smaller scale perturbation we might end up
with a star). In reality overdensities are never exactly spherically symmetric. The deviation
from spherical symmetry increases as the collapse progresses. For an ellipsoidal overdensity the
flattest direction collapses first leading first to a “Zeldovich pancake”, and the second flattest
39
More precisely, the density of an outer shell must not be more than the mean density inside it. We should
also include in our model an underdense region around our overdense region so that their combined mean density
equals that of the background universe, so as not to affect the evolution of the surroundings.
40
Note that Kolb&Turner[11], p. 328, misses the factor 3/5 .
8 STRUCTURE FORMATION 79

next leading then to an elongated structure. In the situation where the density refers to a number
density of galaxies instead of a smooth continuous density, the galaxies will pass the center point
at various distances (instead of colliding at the center as in the perfectly spherically symmetric
case), after which they will move away from the center and will be decelerated, eventually falling
back in and ending up orbiting the center, forming a cluster of galaxies.
For the real universe the different distance scales are in a different stage of the collapse. The
largest distance scales are still “falling in”, leading to flattened structures at the largest scales
and elongated structures, “filaments”, at somewhat smaller scales. These structures surround
rounder underdense regions, “voids”. Smaller scales have already collapsed into galaxy clusters.
8 STRUCTURE FORMATION 80

8.6 Perturbations during inflation

So far we have developed perturbation theory describing the substance filling the universe in
fluid terms, i.e., giving the perturbations in terms of δρ and δp. During inflation the universe is
dominated by a scalar field, the inflaton ϕ, so it is better to give the perturbation directly as a
perturbation in the inflaton field,
ϕ(t, x) = ϕ̄(t) + δϕ(t, x) . (300)

8.6.1 Evolution of inflaton perturbations

In Minkowski space the field equation for a scalar field is
ϕ̈ − ∇2 ϕ + V ! (ϕ) = 0 . (301)
In the flat Friedmann-Robertson-Walker universe (the background universe) the field equation
is
ϕ̈ + 3H ϕ̇ − a−2 ∇2 ϕ + V ! (ϕ) = 0 . (302)
(Here ∇ = ∇x , i.e., with respect to the comoving coordinates x, and therefore the factor 1/a
appears in front of it.)
We ignore for the moment the perturbation in the spacetime metric and just insert (300)
into Eq. (302),
(ϕ̄ + δϕ)¨ + 3H(ϕ̄ + δϕ)˙ − a−2 ∇2 (ϕ̄ + δϕ) + V ! (ϕ̄ + δϕ) = 0 . (303)
Here V ! (ϕ̄ + δϕ) = V ! (ϕ̄) + V !! (ϕ̄)δϕ and ϕ̄(t) is the homogeneous background solution from our
earlier discussion of inflation. Thus ∇2 ϕ̄ = 0, and ϕ̄ satisfies the background equation
ϕ̄¨ + 3H ϕ̄˙ + V ! (ϕ̄) = 0 . (304)
Subtracting the background equation from the full equation (303) we get the perturbation
equation
δϕ̈ + 3Hδϕ̇ − a−2 ∇2 δϕ + V !! (ϕ̄)δϕ = 0 (305)
In Fourier space we have
!" # $
k 2 2
δϕ̈k + 3Hδϕ̇k + + m (ϕ̄) δϕk = 0 , (306)
a
or !" #2 $
−2 −1 k m2
H δϕ̈k + 3H δϕ̇k + + 2 δϕk = 0 , (307)
aH H
where
m2 (ϕ̄) ≡ V !! (ϕ̄) . (308)
During inflation, H and m2 change slowly. Thus we make now an approximation where we
treat them as constants. If the slow-roll approximation is valid, m2 $ H 2 , since
m2 2 V
!!
= 3MPl = 3η $ 1 . (309)
H2 V
Thus we can ignore the m2 /H 2 in Eq. (307)41 . The general solution becomes then
δϕk (t) = Ak wk (t) + Bk wk∗ (t) , (312)
41
The general solution to (306), when H and m2 are constants, is
! " # " #$
k k
δϕk (t) = a−3/2 Ak J−ν + Bk Jν , (310)
aH aH
8 STRUCTURE FORMATION 81

where ! " ! "

k ik
wk (t) = i+ exp . (313)
aH aH
(Exercise: Show that this is a solution of (306) when H = const and m2 = 0.) The time
dependence of (312) is in
a = a(t) ∝ eHt . (314)
Well before horizon exit, k " aH, the argument of the exponent is large. As a(t) increases
the solution oscillates rapidly and its amplitude is damped. After horizon exit, k # aH, the
solution stops oscillating and approaches the constant value i(Ak − Bk ).
We have cheated by ignoring the metric perturbation. We should use GR and write the
curved-spacetime field equation using the perturbed metric. Perturbations in a scalar field
couple only to scalar perturbations, so we need to consider scalar perturbations only. For
example, in the conformal-Newtonian gauge the correct perturbation equation is
#! " $
k 2 % &
N N
δϕ̈k + 3Hδϕ̇k + + V (ϕ̄) δϕN
!!
k = −2Φ k V (ϕ̄) + Φ̇ k + 3Ψ̇ ˙.
k ϕ̄ (315)
a

That is, there are additional terms which are first order in the metric and zeroth order (back-
ground) in the scalar field ϕ.
Fortunately, it is possible to choose the gauge so that the terms with the metric perturbations
are negligible during inflation42 , and the previous calculation applies in such a gauge. The
comoving gauge is not such a gauge, so a gauge transformation is required to obtain the comoving
gauge curvature perturbation R. Gauge transformations are beyond the scope of these lectures,
but the result is
δϕ
R = −H . (316)
ϕ̄˙
Thus it is clear what we want from inflation. We want to find the inflaton perturbations δϕk
some time after horizon exit. We can use the constant value the solution (312) approaches after
horizon exit. Then Eq. (316) gives us Rk , which remains constant while the scale k is outside
the horizon, and is indeed the primordial Rk discussed in the previous section. And then we
can use the results of Sec. 8.4 to get δk .
We are still missing the initial conditions for the solution (312). These are determined by
quantum fluctuations, which we shall discuss in Sec. 8.6.3. Quantum fluctuations produce the
initial conditions in a random manner, so that we can predict only their statistical properties. It
turns out that the quantum fluctuations are a Gaussian process, a term which specifies certain
statistical properties, which we shall discuss next before returning to the application to inflaton
fluctuations.

8.6.2 Statistical properties of Gaussian perturbations

The statistical (Gaussian) nature of the inflaton perturbations δϕ(x) are inherited later by
other perturbations, which depend linearly on them. Let us therefore discuss a generic Gaussian
where Jν is the Bessel function of order ν and
!
9 m2
ν= − 2. (311)
4 H
With m2 = 0, ν = 32 . Bessel functions of half-integer order are spherical Bessel functions which can be expressed
in terms of trigonometric functions, or e±ikx .
42
One such gauge is the spatially flat gauge Q. For scalar perturbations it is possible to choose the time
coordinate so that the time slices have Euclidean geometry. This leads to the spatially flat gauge. (There are still
perturbations in the spacetime curvature; they show up when one considers the time direction).
8 STRUCTURE FORMATION 82

perturbation !
g(x) = gk eik·x , (317)
k

where the set of Fourier coefficients {gk } is a result of a statistically homogeneous and isotropic
Gaussian random process. We assume g(x) is real, so that g−k = gk∗ . We write gk in terms of
its real and imaginary part,
gk = αk + iβk . (318)
For Fourier analysis of statistically homogeneous and isotropic random perturbations, see
sections (8.1.1, 8.1.3, 8.1.4), where the probability distribution was treated as unknown. The
new ingredient (in addition to the assumption that the perturbations are small, allowing the
use of first-order perturbation theory, which we introduced in Sec. 8.2), is that the probability
distribution is known to be Gaussian. This means that
" #
1 1 |gk |2
Prob(gk ) = exp −
2πs2k 2 s2k
" # " # (319)
1 1 α2k 1 1 βk2
=√ exp − 2 × √ exp − 2 ,
2πsk 2 sk 2πsk 2 sk

i.e., the real and imaginary parts are independent Gaussian random variables43 with equal
variance s2k .
The expectation value of a quantity which depends on gk as f (gk ) is given by
$
$f (gk )% ≡ f (gk )Prob(gk )dαk dβk , (320)

where the integral is over the complex plane, i.e.,

$ ∞ $ ∞
dαk dβk .
−∞ −∞

We immediately get (exercise) the mean

$gk % = 0 (321)

and variance
$|gk |2 % = 2s2k = $α2k + βk2 % (322)
of gk .
The distribution has one free parameter, the real positive number sk which gives the width
(determines the variance) of the distribution. From statistical isotropy and homogeneity follows
that sk = s(k) and
$gk∗ gk! % = 0 for k '= k$ . (323)
We can combine Eqs. (322) and (323) into a single equation,

δkk! 2π 2 δkk!
$gk∗ gk! % = 2δkk! s2k = δkk! $|gk |2 % ≡ P (k) ≡ P(k) , (324)
V V k3
where " #3
L V 3
P(k) ≡ 4πk3 $|gk |2 % = k $|gk |2 % , (325)
2π 2π 2
43
gk is a complex Gaussian random variable and αk and βk are real Gaussian random variables.
8 STRUCTURE FORMATION 83

which gives the dependence of the variance of gk on the wave number k, is the power spectrum
of g.
Going back to coordinate space, we find
! #
" "
!g(x)" = gk eik·x = !gk "eik·x = 0 (326)
k k

The square of the perturbation can be written as

" " !
g(x)2 = gk∗ e−ik·x gk! eik ·x (327)
k k!

since g(x) is real. The typical amplitude of the perturbation is described by the variance, the
expectation value of this square,
" " " $ %3 "
2 ∗ i(k! −k)·x 2 2 2π 1
!g(x) " = !gk gk! "e = !|gk | " = 2 sk . = P(k)
L 4πk3
kk! k k k
& 3 & ∞ & ∞
1 d k dk
→ P(k) = P(k) = P(k)d ln k . (328)
4π k3 0 k −∞

Note that there is no x-dependence in the result, since this is an expectation value. g(x)2 of
course varies from place to place, but its expectation value from the random process is the same
everywhere—the perturbed universe is statistically homogeneous. Thus the power spectrum of
g gives the contribution of a logarithmic scale interval to the variance of g(x).
An alternative definition for the power spectrum is

P (k) ≡ V !|gk |2 " (329)

While this definition is simpler, the result for the variance of g(x) in terms of it and thus the
interpretation is more complicated. Because of the common use of this latter definition, we shall
make reference to both power spectra, and distinguish them by the different typeface. They are
related by
2π 2
P (k) = 3 P(k) . (330)
k
For Gaussian perturbations, the power spectrum gives a complete statistical description. All
statistical quantities can be calculated from it. In particular, (exercise)

!|gk |4 " = 2!|gk |2 "2 . (331)

Exercise: Show that, if α is a real Gaussian random variable, with !α" = 0, then

!|α|4 " = 3!|α|2 "2 ,

and that, if g is a complex Gaussian random variable (real and imaginary parts independent of each
other), with !g" = 0, then
!|g|4 " = 2!|g|2 "2 .

For a single realization, define P̂ (k) ≡ V |gk |2 . From (331)

!P̂ (k)2 " = 2P (k)2 . (332)

The typical deviation of P̂ (k) from its expectation value is given by the square root of the
variance
!|P̂ (k) − P (k)|2 " ≡ !P̂ (k)2 " − P (k)2 = P (k)2 , (333)
8 STRUCTURE FORMATION 84

which (the last equality) only holds for Gaussian perturbations. For a single mode k, this
variance is large, but we can define
V !
P̂ (k) ≡ |gk |2 , (334)
Nk
k

where the sum is over all k for which k − 21 ∆k < |k| ≤ k + 21 ∆k, where ∆k is the width of a
k-bin (a shell in k-space) over which we average, and Nk is the number of Fourier modes k in
the bin. We then get (exercise)
2
$|P̂ (k) − P (k)|2 % = P (k)2 . (335)
Nk

This is an example of the cosmic variance discussed in Sec. 8.1.1: the power spectrum P̂ (k)
measured from a finite volume V deviates from its expectation value P (k), and thus we can
measure P (k) only with finite accuracy. The estimate P̂ (k) is an average over Nk modes and its
variance is reduced by the factor Nk /2 (since g−k = gk∗ we have Nk /2 independent modes, i.e.,
Nk /2 independent complex random variables or Nk independent real random variables.) The
density of k modes in k-space goes as 1/V , so the larger the survey volume V , the more modes
we have in a k-bin (the larger os Nk ). Also, for higher k, there are more modes in a k-bin: a
given survey volume samples small scales better than large scales.

Exercise: Derive Eq. (335). Note that the gk are independent variables so that $|gk |2 |gk! |2 % =
$|gk |2 %$|gk! |2 % for k! &= ±k, but since g−k = gk∗ , $|gk |2 |gk! |2 % = $|gk |4 %.

It can be shown (under weak assumptions about the power spectrum), that statistically
homogeneous and isotropic Gaussian perturbations are ergodic, so that we do not need to make
a separate assumption of ergodicity.44

8.6.3 Generation of primordial perturbations from inflation

Subhorizon scales during inflation are microscopic45 and therefore quantum effects are important.
Thus we should study the inflaton field using quantum field theory.
This goes beyond the level of this course, so we have relegated the discussion into an ap-
pendix. The basic idea is that for scales that are inside horizon there are quantum fluctuations,
called vacuum fluctuations, in the inflaton field. For a homogeneous inflaton field, the Fourier
amplitudes δϕk of its perturbations would be identically zero, but analogous to a quantum har-
monic oscillator, it is not possible for them to stay there, but instead they fluctuate around this
value.
We saw in Sec. 8.6.1 that the classical solutions to the evolution of δϕk reach a constant
value after horizon exit (in the approximation H = const during horizon exit). The quantum
treatment gives that at this stage we can neglect further quantum fluctuations and treat δϕk
classically—the fluctuations “freeze”.
The final result is that well after horizon exit, k ' H, the Fourier amplitudes δϕk have
acquired a power spectrum
" #2
k3 2 H
Pϕ (k) ≡ V 2 $|δϕk | % = . (336)
2π 2π
44
Liddle & Lyth [6], in Sec. 4.3.3, make this claim but do not provide a proof.
45
We later give an upper limit to the inflation energy scale, i.e., V (ϕ) at the time cosmological scales exited
the horizon, V 1/4 < 1.9 × 1016 GeV. From H 2 = V (ϕ)/3MPl 2
we have H < 9 × 1013 GeV or for the Hubble length
−1 −30
H > 2.3 × 10 m. This is a lower limit to the horizon size, but it is not expected to be very many orders of
magnitude larger.
8 STRUCTURE FORMATION 85

After this we can ignore further quantum effects and treat the later evolution of the in-
flaton field, both the background and the perturbation, classically. The effect of the vacuum
fluctuations was to produce “out of nothing” the perturbations δϕk . We can’t predict their
individual values; their production from quantum fluctuations is a random process. We can
only calculate their statistical properties. Closer investigation reveals that this is a Gaussian
random process. All δϕk acquire their values as independent random variables (except for the
reality condition δϕ−k = δϕ∗k ) with a Gaussian probability distribution. Thus all statistical
information is contained in the power spectrum Pϕ (k).
The result (336) was obtained treating H as a constant. However, over long time scales, H
does change. The main purpose of the preceding discussion was to follow the inflaton pertur-
bations through the horizon exit. After the perturbation is well outside the horizon, we switch
to other variables, namely the curvature perturbation Rk , which, unlike δϕk , remains constant
outside the horizon, even though H changes. Therefore we have to use for each scale k a value of
H which is representative for the evolution of that particular scale through the horizon. That is,
we choose the value of H at horizon exit,46 so that aH = k. Thus we write our power spectrum
result as
! "2
k3 2 H
Pϕ (k) = V 2 !|δϕk | " = , (337)
2π 2π aH=k

to signify that the value of H for each k is to be taken at horizon exit of that particular scale.
Equation (337) is our main result from inflaton fluctuations.

8.6.4 Transfer functions

Since the inflaton fluctuations are assumed to be the origin of structure, all later perturbations
are related to the inflaton perturbations δϕk . As long as all inhomogeneities are small (“per-
turbations”), the relationship is linear. We can express these linear relationships as transfer
functions T (t, k), e.g.,
gk (t) = Tgϕ (t, k)δϕk (tk ) . (338)
The linearity implies several things:

1. The Fourier coefficient gk depends only on the Fourier coefficient of δϕ corresponding to

the same wave vector k, not on any other k# .

2. The relationship is linear, so that if δϕk were, e.g., twice as big, then so would gk be.

3. The perturbations of g inherit the Gaussian statistics of δϕ.

We could also define transfer functions relating perturbations at any two different times, t
and t# , and call them T (t, t# , k), but here we are referring to the inflaton perturbations at the time
of horizon exit, tk , which is different for different k. Actually, by δϕk (tk ) we mean the constant
value the perturbation approaches after horizon exit in the H = const = Hk approximation.
That the transfer function depends only on the magnitude k results from the fact that
physical laws are isotropic. The transfer function of Eq. (338) will then relate the power spectra
of {gk (t)} and {δϕk (tk )} as
Pg (t, k) = Tgϕ (t, k)2 Pϕ (k) . (339)
The transfer functions thus incorporate all the physics that determines how structure evolves.
46
One can do a more precise calculation, where one takes into account the evolution of H(t). The result is that
one gets a correction to the amplitude of PR (k), which is first order in slow-roll parameters, and a correction to
its spectral index n which is second order in the slow-roll parameters.
8 STRUCTURE FORMATION 86

For the largest scales, k−1 ! 10h−1 Mpc, the perturbations are still small today, and one
needs not go beyond the transfer function. For smaller scales, corresponding to galaxies and
galaxy clusters, the inhomogeneities have become large at late times, and the physics of structure
growth has become nonlinear. This nonlinear evolution is typically studied using large numerical
simulations. Fortunately, the relevant scales are small enough that Newtonian physics is usually
sufficient.
We are now in position to put together all the results we obtained. From Eq. (316)

δϕk
Rk = −H , (340)
ϕ̄˙
so that
Hk
TRϕ (k) = − (341)
˙ k)
ϕ̄(t
and ! "2 #! "! "$2
H H H
PR (k) = Pϕ (k) = , (342)
ϕ̄˙ ϕ̄˙ 2π H=k
where we used the result (337).
This primordial spectrum is the starting point for calculating structure formation (discussed
already) and the CMB anisotropy (Chapter 9). Thus CMB and large-scale structure observations
can be used to constrain PR together with other cosmological parameters.

8.6.5 Generation of primordial gravitational waves

The quantum fluctuations at subhorizon scales during inflation apply also to the spacetime itself.
We do not yet have a complete theory of quantum gravity, so we do not know how spacetime
behaves in the Planck era. At lower energy scales the spacetime fluctuations are smaller and
for small perturbations around a FRW universe we can use the linearized equations for metric
perturbations, for which quantization is straightforward. In fact, the proper treatment of the
generation of inflaton perturbations, where we include the scalar metric perturbations in the
inflaton perturbation equation (see Eq. 315), contains also the quantum treatment of scalar
metric perturbations.
Likewise, we have quantum fluctuations of tensor metric perturbations during inflation.
These do not couple to density perturbations, but they become classical gravitational waves
after horizon exit. These primordial gravitational waves have an effect on CMB anisotropy and
polarization. √
In the quantum treatment, (MPl / 2)h fluctuates like a scalar field, so that in inflation the
gravitational wave amplitudes h acquire a spectrum
! "2 ! "2
V 2 H 8 H
Ph (k) ≡ 4 2 k3 %|hk |2 & = 4 2 = 2 (343)
2π MPl 2π H=k MPl 2π H=k

(the factor 4 in this customary definition is related to the way h appears in several places in the
metric and to there being two modes for each k).
The tensor-to-scalar ratio is the ratio of the two primordial spectra (343) and (342),
! "2
Ph (k) 8 ϕ̄˙
r ≡ = 2 . (344)
PR (k) MPl H H=k
8 STRUCTURE FORMATION 87

8.7 The primordial spectrum

8.7.1 Primordial spectrum from slow-roll inflation
The final result of the previous section is thus that inflation generates primordial scalar pertur-
bations Rk with the power spectrum
!" #" #$2 " #2
H H 1 H2
PR (k) = = . (345)
ϕ̇ 2π H=ak 4π 2 ϕ̇ t=tk

and primordial tensor perturbations with the power spectrum

" #2
8 H
Ph (k) = 2 . (346)
MPl 2π t=tk

In this section ϕ and ϕ̇ refer to the background values.

Applying the slow-roll equations

V ϕ̇ !
2 V
H2 = 2 and 3H ϕ̇ = −V ! ⇒ = −MPl
3MPl H V

these become
1 1 V3 1 1 V
PR (k) = 2 6 !2
= 2 4 ε
12π MPl V 24π MPl
2 V
Ph (k) = 4 , (347)
3π 2 MPl

where ε is the slow-roll parameter. The tensor-to-scalar ratio is thus

Ph (k)
r ≡ = 16ε . (348)
PR (k)

According to present observational CMB and large-scale structure data, the amplitude of
the primordial power spectrum is about

PR (k)1/2 ≈ 5 × 10−5 (349)

at cosmological scales. This gives a constraint on inflation

" #1/4
V √ %
≈ 241/4 π 5 × 10−5 MPl ≈ 0.028MPl = 6.8 × 1016 GeV . (350)
ε

The best chance of detecting primordial gravitational waves is based on their effect on CMB.
They have not been observed so far and the present upper limit is about [8]

r < 0.07 ⇒ Ph (k)1/2 < 1.3 × 10−5 and ε < 0.004 . (351)

This implies an upper limit to the inflation energy scale47

V 1/4 ≈ ε1/4 0.028MPl < 0.007MPl = 1.7 × 1016 GeV . (352)

47
For the epoch when perturbations at observable cosmological scales were generated. During earlier phases of
inflation the energy scale was higher than in that epoch.
8 STRUCTURE FORMATION 88

Since during inflation, V and V ! change slowly while a wide range of scales k exit the horizon,
PR (k) and Ph (k) should be slowly varying functions of k. We define the spectral indices ns and
nt of the primordial spectra as
d ln PR
ns (k) − 1 ≡
d ln k
d ln Ph
nt (k) ≡ . (353)
d ln k
(The −1 is in the definition of ns for historical reasons, to match with the definition in terms of
density perturbations, see Sec. 8.7.2.) If the spectral index is independent of k, we say that the
spectrum is scale free. In this case the primordial spectra have the power-law form
! "ns −1 ! "n t
2 k 2 k
PR (k) = As and Ph (k) = At , (354)
kp kp
where kp is some chosen reference scale, “pivot scale”, and As and At are the amplitudes at this
pivot scale.
If the power spectrum is constant,

P = const. , (355)

corresponding to ns = 1 and nt = 0, we say that the spectrum is scale invariant. A scale-

invariant scalar spectrum is also called the Harrison–Zeldovich spectrum.
If ns #= 1 or nt #= 0, the spectrum is called tilted. A tilted spectrum is called red, if ns < 1
or nt < 0 (more structure at large scales), and blue if ns > 1 or nt > 0 (more structure at small
scales).
Using Eqs. (347) and (353) we can calculate the spectral index for slow-roll inflation.
Since P(k) is evaluated from Eqs. (345) and (346) or (347) when k = aH,

d ln k d ln(aH) ȧ Ḣ
= = + = (1 − ε)H ≈ H ,
dt dt a H
where we used Ḣ = −εH 2 (in the slow-roll approximation) in the last step. Thus

d 1 1 d 1 ϕ̇ d M2 V ! d !
2 V d
= = = − Pl ≈ −MPl . (356)
d ln k 1 − ε H dt 1 − ε H dϕ 1 − ε V dϕ V dϕ
Let us first calculate the scale dependence of the slow-roll parameters:
# $ #! " ! ! "2 !! $
! d M 2 ! ! "2 ! 4
dε 2 V V 4 V V V
= −MPl Pl
= MPl − = 4ε2 − 2εη (357)
d ln k V dϕ 2 V V V V

and, in a similar manner (exercise),

dη
= . . . = 2εη − ξ , (358)
d ln k
where we have defined a third slow-roll parameter

4 V ! !!!
ξ ≡ MPl V . (359)
V2
%
The parameter ξ is typically second-order small in the sense that |ξ| is of the same order of
magnitude as ε and η. (Therefore it is sometimes written as ξ 2 , although nothing forces it to be
positive.)
8 STRUCTURE FORMATION 89

We are now ready to calculate the spectral indices:

! "
1 dPR ε d V 1 dV 1 dε
ns − 1 = = = −
PR d ln k V d ln k ε V d ln k ε d ln k
!
2 V 1 dV
= −MPl · − 4ε + 2η = −6ε + 2η
V V dϕ
!
! ! "2
1 dPh 2 V 1 dV 2 V
nt = = −MPl = −MPl = −2ε . (360)
Ph d ln k V V dϕ V

Since ε > 0, the tensor spectrum is necessarily red. (This follows already from (346), since H
is decreasing, or from (347) since V is decreasing.) Slow-roll requires ε " 1 and |η| " 1, so
both spectra are close to scale invariant. For scalar perturbations this is verified by observation.
Based on CMB anisotropy data from the Planck satellite, the Planck Collaboration [8] finds

ns = 0.965 ± 0.004 . (361)

If one were able to measure all three values ns , r, and nt from observations, one could solve
from them the slow-roll parameters ε and η and moreover, check the consistency condition
r
nt = − (362)
8
for single-field slow-roll inflation. This consistency condition is the only truly quantitative
prediction of the inflation scenario (as opposed to some specific inflation model) – all the other
predictions (Ωk very small, ns close to 1 and nt close to 0, primordial perturbations Gaussian)
are of qualitative nature, not a specific number not equal to 0 or 1.
Unfortunately, the existing upper limit to r already means that it will be difficult to ever
determine the spectral index nt with sufficient accuracy to distinguish between nt = −r/8 and
nt = 0. The most sensitive probe to primordial gravitational waves is provided by polarization of
CMB on which they will imprint a characteristic pattern (discussed briefly in the next chapter).
The theoretical limit to detection is r ∼ 10−4 and there are proposals48 for future CMB satellite
missions that could reach r ∼ 10−3 . If r is significantly larger than these detection limits, after
detection one could still measure nt accurately enough to distinguish, say, nt ≈ −1, nt ≈ 0
(which includes the case nt = −r/8), and nt ≈ 1 from each other. There have been other
proposals (other than inflation) for very-early-universe physics, which predict primordial tensor
perturbations that deviate from scale invariance this much or more.
The Japanese space agency ISAS/JAXA selected in May 2019 the 3-year LiteBIRD mission49
to be launched in the late 2020s. LiteBIRD is expected to measure r with accuracy ∆r ∼ 10−3
(1 σ). A significant American and European participation in LiteBIRD is expected.
Detection of primordial gravitational waves, i.e., measurement of r, would be enough to
determine ε and η and thus the inflation energy scale from Eq. (350).
One can also calculate the scale-dependence of the spectral index (exercise):
dns
= 16εη − 24ε2 − 2ξ . (363)
d ln k
It is second order in slow-roll parameters, so it’s expected to be even smaller than the deviation
from scale invariance, ns − 1. Planck Collaboration finds it consistent with zero to accuracy
O(10−2 ), as expected.
Cosmologically observable scales have a range of about ∆ ln k ∼ 10. Planck measured the
CMB anisotropy over a range ∆ ln k ∼ 6 (missing the shortest scales, where the CMB is expected
48
See, e.g., http://www.core-mission.org/
49
http://litebird.jp/eng/
8 STRUCTURE FORMATION 90

to have negligible anisotropy). Some inflation models have |ns −1|, r, and |dns /d ln k| larger than
the Planck results, while others do not. These observations already ruled out many inflation
models.
Example: Consider the simple inflation model
1 2 2
V (ϕ) = m ϕ .
2
In Chapter 7 we already calculated the slow-roll parameters for this model:
2
MPl
ε=η=2
ϕ2
and we immediately see that ξ = 0. Thus
"2 !
MPl
ns = 1 − 6ε + 2η = 1 − 8
ϕ
! "4
dns MPl
= 16εη − 24ε2 − 2ξ = −32
d ln k ϕ
! "2
MPl
r = 16ε = 32
ϕ
! "2
MPl
nt = −2ε = −4
ϕ

To get the numbers out, we need the values of ϕ when the relevant cosmological scales exited the
horizon. The number of inflation e-foldings after that should be about N ∼ 50. We have
# ϕ #
1 V 1 ϕ 1 $ 2 2
%
N (ϕ) = 2 !
dϕ = 2 = 2 ϕ − ϕend ,
MPl ϕend V M Pl 2 4M Pl

2
√
and we estimate ϕend from ε(ϕend ) = 2MPl /ϕend 2 = 1 ⇒ ϕend = 2MPl to get

ϕ2 = ϕend 2 + 4MPl
2 2
N = 2MPl 2
+ 4MPl 2
N ≈ 4MPl N.

Thus ! "2
MPl 1
=
ϕ 4N
and
2
ns = 1− ≈ 0.96
N
dns 2
= − ≈ −0.0008
d ln k N2
8
r = ≈ 0.16
N
1
nt = − ≈ −0.02
N
We see that this model is ruled out by the observed upper limit r < 0.1.50

50
There was enormous excitement in early 2014, when the BICEP2 collaboration[12] claimed to have detected
the effect of primordial gravitational waves with r = 0.20+0.07
−0.05 in CMB polarization, consistent with this inflation
model. However, it turned out that their data was contaminated by polarized emission from dust in our own
galaxy.[13]
8 STRUCTURE FORMATION 91

8.7.2 Scale invariance of the primordial power spectrum

Inflation predicts and observations give evidence for an almost scale invariant primordial power
spectrum. Let us forget the “almost” for a moment and discuss what it means for the primordial
spectrum to be scale invariant.
The primordial spectrum is something we have at superhorizon scales, where we have dis-
cussed it in terms of the comoving curvature perturbation R, and we are calling it scale invariant,
when
PR (k) = A2s = const. (364)
We would like the spectrum in terms of more familiar concepts like the density perturbation,
but at superhorizon scales the density perturbation is gauge dependent.
For small scales the perturbation spectrum gets modified when the scales enter the horizon,
but for large scales k ! keq the spectrum maintains its primordial shape, at least as long as
the universe stays matter dominated. This allows the discussion of the primordial spectrum
at subhorizon scales, where we can talk about the density perturbations without specifying a
gauge.
From Eq. (259), the gravitational potential and density perturbation are related to the
curvature perturbation as
3
Φ k = − Rk (mat.dom)
5
! "2 ! " (365)
2 k 2 k 2
δk = − Φk = Rk ,
3 H 5 H
giving
9 9
PΦ (k) = PR (k) = A2s = const (366)
25 25
! "4 ! "4
4 k 4 k
Pδ (t, k) = PΦ (k) = PR (k)
9 H 25 H
! "4
4 k
= A2s ∝ t4/3 k4 (367)
25 H
Thus perturbations in the gravitational potential are scale invariant, but perturbations in density
are not. Instead the density perturbation spectrum is steeply rising, meaning that there is much
more structure at small scales than at large scales. Thus the scale invariance refers to the
gravitational aspect of perturbations, which in the Newtonian treatment is described by the
gravitational potential, and in the GR treatment by spacetime curvature.
The relation between density and gravitational potential perturbations reflects the nature of
gravity: A 1% overdense region 100 Mpc across generates a much deeper potential well than a
1% overdense region 10 Mpc across, since the former has 1000 times more mass. Therefore we
need much stronger density perturbations at smaller scales to have an equal contribution to Φ.
However, if we extrapolate Eq. (367) back to horizon entry, k = H, we get
! "2
2 4 2
δH (k) ≡ “Pδ (k, tk )” ≡ PR (k) = As = const (368)
25 5
Thus for scale-invariant primordial perturbations, density perturbations of all scales enter the
horizon with the same amplitude, δH = (2/5)As ∼ 2 × 10−5 . Since the density perturbation
at the horizon entry is actually a gauge-dependent quantity, and our extension of the above
Newtonian relation up to the horizon scale is not really allowed, this statement should be taken
just qualitatively (hence the quotation marks around the Pδ ). As such, it applies also to the
8 STRUCTURE FORMATION 92

smaller scales which enter during the radiation-dominated epoch, since the perturbations only
begin to evolve after horizon entry.
What is the deep reason that inflation generates (almost) scale invariant perturbations?
During inflation the universe is almost a de Sitter universe, which has the metric

ds2 = −dt2 + e2Ht (dx2 + dy 2 + dz 2 )

with H = const . In GR we learn that it is an example of a “maximally symmetric spacetime”.

In addition to being homogeneous (in the space directions), it also looks the same at all times.
Therefore, as different scales exit at different times they all obtain the same kind of perturbations.
In terms of the other definition of the power spectrum, P (k) ≡ (2π 2 /k3 )P(k), the relations
(367) for scale-invariant perturbations give

PR (k) ∝ k−3 PR ∝ k−3

(369)
Pδ (k) ∝ k−3 Pδ ∝ k4 PR ∝ kPR ∝ k

For PR (k) ∝ kn−1 we have Pδ (k) ∝ kn . This is the reason for the −1 in the definition of the
spectral index in terms of PR —it was originally defined in terms of Pδ .

8.8 The power spectrum today

8.8.1 Density perturbations
From Eq. (265), the density perturbation spectrum at late times is
! "4 # $
4 k D(a) 2
Pδ (k) = T (k, t)2 PR (k) (370)
25 H a

where, from Eq. (267)

T (k) = 1 for k $ keq

! "
keq 2
T (k) ∼ ln k for k & keq .
k

(For the more precise form of T (k) see Sec. 8.4.4, Eq. (270) and Fig. 11.) Thus the present-day
density power spectrum rises steeply ∝ k4 at large scales, but turns at ∼ keq to become less
steep (growing ∼ ln k) at small scales. This is because the growth of density perturbations was
inhibited while the perturbations were inside the horizon during the radiation-dominated epoch.
The ∼ ln k factor comes from the slow growth of CDM perturbations during this time.
−1 ∼
Thus the structure in the universe appears stronger at smaller scales (higher k), until keq
100 Mpc. The ∼ 100 Mpc scale is indeed quite prominent in large scale structure surveys, like
the 2dFGRS and SDSS galaxy distribution surveys. Towards smaller scales the structure keeps
getting stronger, but now more slowly. However, perturbations are now so large that first-order
−1
perturbation theory begins to fail, and that limit is crossed at around k−1 ∼ knl ∼ 10 Mpc.
Nonlinear effects cause the density power spectrum to rise more steeply than calculated by
perturbation theory at scales smaller than this.
The present-day density power spectrum Pδ (k) can be determined observationally from the
distribution of galaxies (Fig. 14). The quantity plotted is usually Pδ (k). It should go as

Pδ (k) ∝ kn for k $ keq

n−4
(371)
Pδ (k) ∝ k ln k for k & keq .

See Fig. 15.

8 STRUCTURE FORMATION 93

Figure 13: The whole picture of structure formation theory from quantum fluctuations during inflation
to the present-day power spectrum at t0 .

Example: Errors on P (k) estimation. The accuracy of power spectrum estimation from a galaxy
survey is affected by sample variance and shot noise. Sample (or cosmic) variance was discussed earlier.
Shot noise comes from the fact that instead of observing a continuous matter distribution, we are sampling
it with a finite number density n of galaxies. To reduce sample variance one should increase the survey
volume Vs . To reduce shot noise, one should increase n, by observing also smaller and fainter galaxies.
We skip the math for shot noise (discussed in Galaxy Survey Cosmology), and give just the result that
for Gaussian perturbations the combined effect of sample variance and shot noise is
! "2
2 2 1
!|P̂ (k) − P (k)| # = P (k) + , (372)
Nk n̄

where n̄ is the mean number density of galaxies in the survey. (This is Eq. (335) modified by adding the
effect of shot noise.)
Let us see how well this corresponds to the P (k) estimate obtained in [15] for the SDSS survey
(Fig. 15). Table I of [15] gives their estimate P̂ (k) and their estimated uncertainty ∆P (k) for each of
their 20 k-bins in numbers. We copy these numbers into our Table 1. The number of LRG galaxies in
the survey was N = 53 860. The effective sky area of the survey was Ωs = 4259 deg2 = 1.297 sr, and the
surveyed redshift range was z = 0.155–0.474. From Figs. 1 and 2 of [15] this corresponds to a comoving
distance range r = rmin – rmax = 450 – 1300 h−1Mpc, from which we get a survey volume and number
density
# 3 3
$
rmax rmin
Vs = Ωs − = 0.9107 h−3Gpc3 ,
3 3
N
n̄ = = 6.408 × 10−5 h3 Mpc−3 . (373)
Vs

If the survey volume were a cube, Vs = L3 , the side of the cube would be L = 969.3 h−1Mpc, which
means the components of k are integer multiples of kf = 2π/L = 0.006482 h/Mpc. The volume of a k-bin
8 STRUCTURE FORMATION 94

Figure 14: Distribution of galaxies according to the Sloan Digital Sky Survey (SDSS). This figure shows
galaxies that are within 2◦ of the equator and closer than 858 Mpc (assuming H0 = 71 km/s/Mpc).
Figure from astro-ph/0310571[14].
8 STRUCTURE FORMATION 95

10

1
2

0.1
k P(k) / 2π
3

0.01

0.001

0.0001

0.01 0.1 1
-1
k [Mpc ]

1e+05
P(k) [Mpc ]

10000
3

1000

100

0.01 0.1 1
-1
k [Mpc ]

Figure 15: The power spectrum from the SDSS obtained using luminous red galaxies (LRG) [15]. Note
that LRG are more strongly biased, b ≈ 1.9, than galaxies on average, so to get the P (k) of matter,
one should divide by b2 ≈ 3.6. The top figure shows Pδ (k) and the bottom figure Pδ (k). A Hubble
constant value H0 = 71.4 km/s/Mpc has been assumed for this figure. (These galaxy surveys only obtain
the scales up to the Hubble constant, and therefore the observed Pδ (k) is usually shown in units of
h−3 Mpc3 as a function of h Mpc−1 , so that no value for H0 need to be assumed.) The black bars are
the observations and the red curve is a theoretical fit, from linear perturbation theory, to the data. The
bend in P (k) at keq ∼ 0.01 Mpc−1 is clearly visible in the bottom figure. Linear perturbation theory
fails when P(k) ! 1, and therefore the data points do not follow the theoretical curve to the right of the
dashed line (representing an estimate on how far linear theory can be trusted). Figure by R. Keskitalo.
8 STRUCTURE FORMATION 96

is
4π ! 3 3
"
Vk = kmax − kmin . (374)
3
and the number of k modes in the bin is Nk = Vk /kf3 , which we have added as the third column in
Table 1.
Using the estimated P̂ (k) in place of P (k) on the rhs of Eq. (372), we obtain from it our estimate for
the relative uncertainty as
# # $ %
2 1
"|P̂ (k) − P (k)|2 # Nk P̂ (k) + n̄
≈ , (375)
P (k) P̂ (k)

which we have added as the last column of Table 1.

Compared to the actual SDSS analysis the above is extremely naive. The facts that the survey
volume is not a cube and that because of selection effects the mean number density of galaxies decreases
as a function of z required a more sophisticated analysis. Also the k-bins SDSS used where not sharp
[kmin , kmax ], but instead for each bin they used a smooth window function (of k) giving more weight to
the modes near the center of the bin and some weight even to modes outside [kmin , kmax ]. In addition,
nonlinear effects cause deviations from Gaussianity, which typically increase the sample variance, and
there are other effects contributing to the estimate uncertainty; so that we should expect (375) to be an
underestimate. Nevertheless, comparing the last column of Table 1 (our naive estimate) to the second-
to-last column (the SDSS uncertainty estimate) we see that we are on the right track.

SDSS LRG Power spectrum

kmin kmax Nk P̂ (k) ∆P (k) ∆P (k)/P̂ (k) Eq. (375)
0.008 0.017 68 124884 18775 0.150 0.193
0.013 0.018 56 118814 29400 0.247 0.214
0.016 0.022 101 134291 21638 0.161 0.157
0.018 0.025 151 58644 16647 0.284 0.146
0.021 0.028 195 105253 12736 0.121 0.116
0.025 0.033 312 77699 9666 0.124 0.096
0.029 0.037 404 57870 7264 0.126 0.089
0.033 0.043 670 56516 5466 0.097 0.070
0.037 0.051 1261 50125 3991 0.080 0.052
0.042 0.057 1708 45076 2956 0.066 0.046
0.050 0.066 2499 39339 2214 0.056 0.040
0.057 0.075 3640 39609 1679 0.042 0.033
0.066 0.086 5360 31566 1284 0.041 0.029
0.076 0.099 8171 24837 991 0.040 0.025
0.088 0.113 11710 21390 778 0.036 0.023
0.101 0.128 16407 17507 629 0.036 0.021
0.108 0.145 27511 15421 516 0.033 0.017
0.126 0.165 38320 12399 430 0.035 0.016
0.159 0.190 43665 11237 382 0.034 0.016
0.181 0.218 68135 9345 384 0.041 0.014

Table 1: The first and second column give roughly the extent of each k-bin used in units of h Mpc−1 .
(Note that the bins overlap.) Nk is the corresponding number of k modes in the bin. P̂ (k) and ∆P (k)
are the estimates from [15] in units of ( h−1 Mpc)3 , and in the last two columns we compare their ratio
to the uncertainty estimate we get from the survey characteristics.

8.8.2 σ8
Instead of As , the amplitude of the primordial power spectrum at a pivot scale, astronomers like
to use the parameter σ8 to describe the amplitude of cosmological perturbations of a cosmological
8 STRUCTURE FORMATION 97

model. σ8 is defined as the top-hat window matter density variance σT (R) at scale R = 8 h−1 Mpc
predicted by linear perturbation theory in the cosmological model.
We mentioned earlier (Sec. 8.1.7) that according to observations, σT,g (R = 8 h−1 Mpc) ≈ 1.
This means that nonlinear effects are becoming important at this scale. The difference between
σT,g (R = 8 h−1 Mpc) and σ8 is due to 1) galaxy bias and 2) nonlinear effects. Both should work
in the direction of making σT,g (R = 8 h−1 Mpc) larger than σ8 , so we expect that the correct
cosmological model has σ8 < 1, but not necessarily by very much.
To calculate σ8 in a cosmological model specified by, say, As , ns , Ωm , ΩΛ , ωm , and ωb ,
is not simple and requires numerical computation. This is done, e.g., by the CAMB code.
The computation involves the transfer function T (k), the growth function D(a), and doing the
integral (48a) for σT2 (R).
For the Planck 2018 best-fit ΛCDM model, σ8 = 0.8210.
Planck 2018 best-fit model σ8 . Let’s see how well the results of this chapter allow us to calculate
the Planck 2018 best-fit ΛCDM model σ8 . The model has parameter values (see Table I, Plik best-fit
column in [8])

ln(1010 A2s ) = 3.0448 ⇒ A2s = 21.006 × 10−10

ns = 0.96605
H0 = 0.8120
Ωm = 0.3158
ωb = 0.022383 , (376)
−1
where As is for the pivot scale kp = 0.05 Mpc .
The primordial power spectrum is
! "ns −1
k
PR (k) = A2s . (377)
kp
We want to find the present-day linear power spectrum Pδ (k). If the universe would have stayed matter
dominated until today, δk and Rk would be related by
! "2
2 k
δ̃k = T (k)Rk , (378)
5 H̃0
where˜denotes matter-dominated-model quantities, and “today” is defined by a = 1. As we noted in the
footnote in Sec. 8.3.5, the comparison is to a matter-dominated model with the same matter density today,
1/2
so it has a smaller total density and thus a smaller Hubble constant, H̃0 = Ωm H0 = 37.83 km/s/Mpc.
The true linear δk differs from (378) due to the different growth function. For the matter-dominated
model Dm (a) = a, so Dm (1) = 1, and
! "2
2 D(a) k
δk (t) = T (k)Rk
5 a H̃(a)
! "2
2 k
δk (t0 ) = D(1) T (k)Rk , (379)
5 H̃0
where D(a) is the growth function of the ΛCDM model, given by Eq. (239). Thus the power spectra are
related by
! "4
4 2 k
Pδ (k) = D(1) T (k)2 PR (k) . (380)
25 H̃0
We now calculate Pδ (k) for k = 1/(8 h−1Mpc). For this k,
k 1 h
= −1 = = 1.683
kp 0.05 Mpc 8 h−1 Mpc 0.4
k
= 666.9 , (381)
H̃0
8 STRUCTURE FORMATION 98

so that
Pδ (k) = 65.30D(1)2 T (k)2 . (382)
−1
The BBKS transfer function, using keq = 13.7Ω−1
m h
−2
Mpc = 64.44 h−1Mpc, gives for this k

TBBKS (k) = 0.1435 , with slope − 1.17933 , (383)

so taht ! "2
T (k)
Pδ (k) = 1.346D(1)2 . (384)
TBBKS (k)
We should now calculate D(1) for the Planck best-fit ΛCDM model, and run CAMB to get the true T (k)
for this model, but I am lazy here, and use the results for our reference model (Ωm = 0.3, h = 0.7) from
Sec. 8.3.5 and Fig. 11: D(1) = 0.78 and T (k)/TBBKS (k) ≈ 0.7 to give

Pδ (k) ≈ 0.40 . (385)

Finally we should calculate σT2 for R = 8 h−1 Mpc, which is an integral of Pδ (k) over k, but we try to
get away with just using the value and slope at k = 1/R, i.e., we approximate Pδ (k) with a power-law
function. The slope is n = 0.96605 − 2 × 1.17933 = −1.39261 (modifying the slope of PR (k) with the
slope of TBBKS (k)) and, from (59),

9 nπ Γ(n − 1)
σ82 ≈ n
(n + 1) sin Pδ (k) = 1.939 Pδ (k) ≈ 0.776 , (386)
2 2 n−3
giving
σ8 ≈ 0.881 > 0.8210 . (387)
Using the power-law approximation should give an overestimate, since the true Pδ (k) bends down com-
pared to it on both sides of the chosen k value, and indeed we overestimated σ8 , but not badly.

8.8.3 Primordial gravitational waves

We found that outside the horizon tensor perturbations remain constant,

hk (t) = hk,prim = const , (388)

whereas inside the horizon they become gravitational waves whose amplitude decays

|hk (t)| ∝ a−1 . (389)

Define the transfer function for gravitational waves

|hk (t0 )|
Th (k) ≡ , (390)
hk,prim

so that the present-day power spectrum of primordial gravitational waves is

Pgrav (k, t0 ) = Th (k)2 Ph (k) . (391)

Make the approximation that the transition from (388) to (389) is instantaneous at horizon
entry defined as
k = H = aH . (392)
Denote these values of a, H, and H by ak , Hk , and Hk . Then
ak
Th (k) = = ak . (393)
a0
REFERENCES 99

The shape of the transfer function is determined by the rate at which different comoving scales
k enter horizon as the universe expands. This is determined by the evolution of the comoving
Hubble distance H−1 .
In the matter-dominated universe
2
a ∝ t2/3 and H = ∝ a−3/2 ⇒ H ∝ a−1/2 . (394)
3t
Make first the approximation that the universe is still matter dominated. Then
! " ! "−2
ak Hk −2 k
Th (k) = = = (H0 < k < keq ) (395)
a0 H0 a0 H 0
for scales that entered during the matter-dominated epoch.
To correct this result for the effect of dark energy at late times, we note that because of dark
energy, the comoving Hubble distance H−1 = (aH)−1 stopped growing and began to shrink,
so that the scale k = H0 is actually exiting now, and it entered at an earlier time t1 when the
expansion was still (barely) matter dominated. Thus the above result for Th (k) should apply
(roughly) at that earlier time:
! "−2 ! "−2
k k
Th (t1 , k) = = (H0 < k < keq ) (396)
a1 H1 a0 H 0
While the scale k = H0 was inside the horizon, the universe expanded by about a factor of two,
so the correct transfer function is about half of (395).
Exercise: Extend the result (395) to scales k > keq . You can make the approximation where the
transition from radiation-dominated expansion law to matter-dominated expansion law is instantaneous
at teq . (This approximation actually underestimates Th (k > keq ) by a factor that roughly compensates
the overestimation in (395) from ignoring dark energy at late times.)
Gravitational waves were detected for the first time on September 14, 2015 at the LIGO observatory.
These were not primordial gravitational waves; they were caused by a collision of two black holes about 400
Mpc from here, and they were observed only for about 0.2 seconds. The peak amplitude was h ≈ 10−21 .
LIGO is sensitive to frequencies near 100 Hz, and with further refinements it is expected to reach a
sensitivity of h = 10−22 . Assume the primordial tensor perturbations had amplitude h = 10−5 (close to
the upper limit from CMB observations). What is their amplitude today at the 100 Hz frequency?
ESA is planning to launch a space gravitational wave observatory (LISA) in 2034. It would have
similar sensitivity as LIGO, but for frequencies lower by a factor 10−4 . What do you conclude about the
prospect for observing primordial gravitational waves this way?

References
[1] C.M. Baugh, The real-space correlation function measured from the APM Galaxy Survey,
Mon. Not. R. Astron. Soc. 280, 267 (1996), astro-ph/9512011

[2] A.J. Benson, R.G. Bower, C.S. Frenk, C.G. Lacey, C.M. Baugh, and S. Cole, What shapes
the luminosity function of galaxies?, Astrophys. J. 599, 38 (2003), astro-ph/0302450

[3] A.G. Sánchez et al., The clustering of galaxies in the SDSS-III Baryon Oscillation Spectro-
scopic Survey: cosmological implications of the large-scale two-point correlation function,
Mon. Not. R. Astron. Soc. 425, 415 (2012), arXiv:1203.6616

[4] E. Hawkins et al., The 2dF Galaxy Redshift Survey: correlation functions, peculiar velocities
and the matter density of the Universe, Mon. Not. R. Astron. Soc. 346, 78 (2003),astro-
ph/0212375
REFERENCES 100

[5] J.A. Peacock: Cosmological Physics (Cambridge University Press 1999), Chapter 16

[6] A.R. Liddle and D.H. Lyth: Cosmological Inflation and Large-Scale Structure (Cambridge
University Press 2000)

[7] Planck Collaboration, Astronomy & Astrophysics 594, A13 (2016), arXiv:1502.01589

[8] Planck Collaboration, Planck 2018 results. VI. Cosmological parameters, arXiv:1807.06209

[9] S. Dodelson: Modern Cosmology (Academic Press 2003), Chapter 7

[10] J.M. Bardeen, J.R. Bond, N. Kaiser, A.S. Szalay, The statistics of peaks in Gaussian ran-
domn fields, Astrophys. J. 304, 15 (1986), Appendix G

[11] E.W. Kolb and M.S. Turner: The Early Universe (Addison-Wesley 1990)

[12] P.A.R. Ade et al., Phys. Rev. Lett. 112, 241101 (2014), arXiv:1403.3985

[13] Planck Collaboration, Astronomy & Astrophysics 586, A133 (2016), arXiv:1409.5738

[14] J. Richard Gott III et al., A Map of the Universe, Astrophys. J. 624, 463 (2005), astro-
ph/0310571

[15] M. Tegmark et al., Cosmological Constraints from the SDSS Luminous Red Galaxies, Phys.
Rev. D74, 123507 (2006), astro-ph/0608632
9 Cosmic Microwave Background Anisotropy
9.1 Introduction
The cosmic microwave background (CMB) is isotropic to a high degree. This tells us that the
early universe was rather homogeneous at the time (t = tdec ≈ 370 000 years) the CMB was
formed. However, with precise measurements we can detect a low-level anisotropy in the CMB
(Fig. 1) which reflects the small perturbations in the early universe.
This anisotropy was first detected by the COBE (Cosmic Background Explorer) satellite in
1992, which mapped the whole sky in three microwave frequencies. The angular resolution of
COBE was rather poor, 7◦ , meaning that only features larger than this were detected. Mea-
surements with better resolution, but covering only small parts of the sky were then performed
using instruments carried by balloons to the upper atmosphere, and ground-based detectors lo-
cated at high altitudes. A significant improvement came with the WMAP (Wilkinson Microwave
Anisotropy Probe) satellite, which made observations for nine years, from 2001 to 2010.
The best CMB anisotropy data to date, covering the whole sky, has been provided by the
Planck satellite (Fig. 2). Planck was launched by the European Space Agency (ESA), on May
14th, 2009, to an orbit around the L2 point of the Sun-Earth system, 1.5 million kilometers from
the Earth in the anti-Sun direction. Planck made observations for over four years, from August
12th, 2009 until October 23rd, 2013. The first major release of Planck results was in 2013 [1]
and the second release in 2015 [2]. Final Planck results were released in 2018 and 2019.

Figure 1: Cosmic microwave background. The figure shows temperature variations from −400 µK to
+400 µK around the mean temperature (2.7255 K) over the whole sky, in galactic coordinates. The color
is chosen to mimic the true color of CMB at the time it was formed, when it was visible orange-red light,
but the brightness variation (the anisotropy) is hugely exaggerated by the choice of color scale. The fuzzy
regions, notable especially in the galactic plane, are regions of the sky where microwave radiation from
our own galaxy or nearby galaxies makes it difficult to separate out the CMB. (ESA/Planck data).

Planck observed the entire sky twice in a year. The satellite repeated these observations
year after year, and the results become gradually more accurate, since the effects of instrument
noise averaged out and various instrument-related systematic effects could be determined and
corrected better with repeated observations.

101
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 102

Figure 2: The Planck satellite and its microwave receivers. The larger horns are for receiving lower
frequencies and the smaller horns for higher frequencies.

Figure 3: Brightness of the sky in the nine Planck frequency bands. These sky maps are in galactic
coordinates so the Milky Way lies horizontally. From [2].
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 103

In addition to the CMB, there is microwave radiation from our own galaxy and other galaxies,
called foreground by those who study CMB. This radiation can be separated from the CMB
based on its different electromagnetic spectrum. To enable this component separation, Planck
observed at 9 different frequency bands; the lowest one centered at 30 GHz and the highest at
857 GHz (Fig. 3). There were two different instruments on Planck, using different technologies
to detect the variations in the microwave radiation. The Low Frequency Instrument (LFI) used
radiometers for the 30, 44, and 70 GHz bands. The High Frequency Instrument (HFI) used
bolometers for the bands from 100 GHz to 857 GHz. HFI is the barrel-shaped instrument at the
center in Fig. 2 right panel and LFI was wrapped around it. With the additional help of WMAP
and ground-based data 8 different foreground components could be distinguished (Fig. 4).

Figure 4: Result from Planck component separation. Also WMAP data and ground-based 408 MHz data
was used. The extracted nine different components of the microwave radiation from top left to bottom
right are: 1) CMB; 2) synchrotron radiation generated by relativistic cosmic-ray electrons accelerated
by the galactic magnetic field; 3) “free-free emission” (bremsstrahlung) from electron-ion collisions; 4)
emission from spinning galactic dust grains due to their electric dipole moment; 5) thermal emission
from galactic dust (the typical dust temperatures are of order 20 K, so the dust thermal spectrum is
peaked at much higher frequencies than CMB); 6) spectral line emission from HCN, CN, HCO, CS, and
other molecules; 7) spectral line emission from the CO (carbon monoxide) J = 1 → 0 transition; 8) CO
J = 2 → 1 line; 9) CO J = 3 → 2 line (these emission lines from transitions between the four lowest
rotation states of the CO molecule map the distribution of carbon monoxide in the Milky Way). From
[2].
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 104

Figures 5–7 show the observed variation δT in the temperature of the CMB on the sky (red
means hotter than average, blue means colder than average).

Figure 5: Cosmic microwave background: Fig. 1 reproduced in false color to bring out the patterns more
clearly. The color range corresponds to CMB temperature variations from −300 µK (blue) to +300 µK
(red) around the mean temperature. (ESA/Planck data).
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 105

Figure 6: The northern galactic hemisphere of the CMB sky (ESA/Planck data).
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 106

Figure 7: The southern galactic hemisphere of the CMB sky. The conspicuous cold region around
(−150◦ , −55◦) is called the Cold Spot. The yellow smooth spot at (−80◦ , −35◦) in galactic coordinates
is a region where the CMB is obscured by the Large Magellanic Cloud, and the light blue spot at
(−150◦ , −20◦) is due to the Orion Nebula. (ESA/Planck data).
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 107

Figure 8: The observed CMB temperature anisotropy gets a contribution from the last scattering surface,
(δT /T )intr = Θ(tdec , xls , n̂) and from along the photon’s journey to us, (δT /T )jour.

The photons we see as the CMB, have traveled to us from where our past light cone intersects
the hypersurface corresponding to the time t = tdec of photon decoupling. This intersection forms
a sphere which we shall call the last scattering surface.1 We are at the center of this sphere,
except that timewise the sphere is located in the past.
The observed temperature anisotropy is due to two contributions, an intrinsic temperature
variation at the surface of last scattering and a variation in the redshift the photons have suffered
during their “journey” to us,
! " ! " ! "
δT δT δT
= + . (1)
T obs T intr T jour

See Fig. 8. # $
The first term, δT T intr represents the temperature variation of the photon gas at t = tdec .
We also include in it the Doppler effect from the motion of this photon gas. At that time the
larger scales we see in the CMB sky were still outside the horizon, so we have to pay attention
to the gauge choice. In fact, the separation of δT /T into the two components in Eq. (1) is
gauge-dependent. If the time slice t = tdec dips further into the past in some location, it finds a
higher temperature, but the photons from there also have a longer way to go and suffer a larger
redshift, so that the two effects balance each other. We can calculate in any gauge we want,
getting different results for (δT /T )intr and (δT /T )jour depending on the gauge, but their sum
(δT /T )obs is gauge independent. It has to be, being an observed quantity.
One might think that (δT /T )intr should be equal to zero, since in our earlier discussion of
recombination and decoupling we identified decoupling with a particular temperature Tdec ∼
3000 K. This kind of thinking corresponds to a particular gauge choice where the t = tdec time
slice coincides with the T = Tdec hypersurface. In this gauge (δT /T )intr = 0, except for the
Doppler effect (we are not going to use this gauge). Anyway, it is not true that all photons have
their last scattering exactly when T = Tdec . Rather they occur during a rather large temperature
interval and time period. The zeroth-order (background) time evolution of the temperature of
the photon distribution is the same before and after last scattering, T ∝ a−1 , so it does not
matter how we draw the artificial separation line, the time slice t = tdec separating the fluid and
free particle treatments of the photons. See Fig. 9.
1
Or the last scattering sphere. “Last scattering surface” often refers to the entire t = tdec time slice.
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 108

Figure 9: Depending on the gauge, the T = Tdec surface may, or (usually) may not coincide with the
t = tdec time slice.

9.2 Multipole analysis

The CMB temperature anisotropy is a function over a sphere (the celestial sphere, or the unit
sphere of directions n̂). In analogy with the Fourier expansion in 3D space, we separate out the
contributions of different angular scales by doing a multipole expansion,
δT !
(θ, φ) = a!m Y!m (θ, φ) (2)
T0
where the sum runs over $ = 1, 2, . . . ∞ and m = −$, . . . , $, giving 2$ + 1 values of m for each $.
The functions Y!m (θ, φ) are the spherical harmonics (see Fig. 10), which form an orthonormal
set of functions over the sphere, so that we can calculate the multipole coefficients a!m from
"
∗ δT
a!m = Y!m (θ, φ) (θ, φ)dΩ . (3)
T0

Definition (2) gives dimensionless a!m . Often they are defined without the T0 = 2.7255 K in
Eq. (2), and then they have the dimension of temperature and are usually given in units of µK.
Here θ and φ are spherical coordinates, dΩ ≡ d cos θdφ, θ ranges from 0 to π and φ ranges from
0 to 2π.2
The sum begins at $ = 1, since Y00 = const. and therefore we must have a00 = 0 for a
quantity which represents a deviation from average. The dipole part, $ = 1, is dominated by the
Doppler effect due to the motion of the solar system with respect to the last scattering surface,
and we cannot separate out from it the cosmological dipole caused by large scale perturbations.
Therefore we are here interested only in the $ ≥ 2 part of the expansion.
Another notation for Y!m (θ, φ) is Y!m (n̂), where n̂ is a unit vector whose direction is specified
by the angles θ and φ.

9.2.1 Spherical harmonics

We list here some useful properties of the spherical harmonics.
They are orthonormal functions on the sphere, so that
"
dΩ Y!m (θ, φ)Y!∗! m! (θ, φ) = δ!!! δmm! . (4)

They are elementary complex functions and are related to the associated Legendre functions
P!m (x) by #
2$ + 1 ($ − m)! m
Y!m (θ, φ) = (−1)m P (cos θ)eimφ . (5)
4π ($ + m)! !
2
They can also be given in degrees, the colatitude θ ranging from 0◦ (North) to 180◦ (South) and the longitude
φ from 0◦ to 360◦ . There are a number of different astronomical coordinate systems (equatorial, ecliptic, galactic)
in use, with their own historical conventions for the coordinate names, symbols, and units. Typically they involve
the latitude 90◦ − θ instead of the colatitude, so that North is at +90◦ and South at −90◦ , and the longitude is
usually given between −180◦ and +180◦ , e.g., in Fig. 7.
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 109

Legendre polynomials
P0 (x) = 1
P1 (x) = x
P2 (x) = 21 (3x2 − 1)
P3 (x) = 12 (5x3 − 3x)
P4 (x) = 18 (35x4 − 30x2 + 3)

Associated Legendre functions P!m (x) = P!m (cos θ)

√
P11 (x) = 1 − x2 = sin θ
√
P21 (x) = 3x 1 − x2 = 3 cos θ sin θ
P22 (x) = 3(1 − x2 ) = 3 sin2 θ

Spherical harmonics
Y00 (θ, φ) = √1
4π
!
3
Y11 (θ, φ) = − 8π sin θeiφ
!
3
Y10 (θ, φ) = 4π cos θ
!
Y22 (θ, φ) = 96π5
3 sin2 θei2φ
!
5
Y21 (θ, φ) = − 24π 3 sin θ cos θeiφ
! " #
5 3 1
Y20 (θ, φ) = 4π 2
2 cos θ − 2

Spherical Bessel functions

sin x
j0 (x) =
x
sin x cos x
j1 (x) = 2 −
x x
$ %
3 1 3
j2 (x) = − sin x − 2 cos x
x3 x x

Table 1: Legendre functions, spherical harmonics, and spherical Bessel functions.

Thus the θ-dependence is in P!m (cos θ) and the φ-dependence is in eimφ . The functions P!m are
real and
Y!,−m = (−1)m Y!m
∗
, (6)
so that &
2# + 1
Y!0 = P! (cos θ) is real. (7)
4π
The functions P! ≡ P!0 are called Legendre polynomials. See Table 9.2.1 for examples of these
functions for # ≤ 2.
Summing over the m corresponding to the same multipole number # gives the addition
theorem
' 2# + 1
∗
Y!m (θ # , φ# )Y!m (θ, φ) = P! (cos ϑ) , (8)
m
4π
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 110

where ϑ is the angle between n̂ = (θ, φ) and n̂! = (θ ! , φ! ), i.e., n̂ · n̂! = cos ϑ. For n̂ = n̂! this
becomes
! 2$ + 1
|Y!m (θ, φ)|2 = (9)
m
4π

(since P! (1) = 1 always).

We shall also need the expansion of a plane wave in terms of spherical harmonics,
!
eik·x = 4π i! j! (kx)Y!m (x̂)Y!m
∗
(k̂) . (10)
!m

Here x̂ and k̂ are the unit vectors in the directions of x and k, and the j! are the spherical Bessel
functions.

9.2.2 Theoretical angular power spectrum

The CMB anisotropy is due to primordial perturbations, and therefore it reflects their Gaussian
nature. Because one gets the values of the a!m from the other perturbation quantities through
linear equations (in first-order perturbation theory), the a!m are also (complex) Gaussian random
variables. Since they represent a deviation from the average temperature, their expectation value
is zero,
!a!m " = 0 . (11)
From statistical isotropy follows that the a!m are independent random variables, except for the
reality condition (13), so that

!a!m a∗!! m! " = 0 if $ #= $! or m #= m! . (12)

Since δT /T0 is real,

a!,−m = (−1)m a∗!,m . (13)
Although thus a!,−m and a!m are not independent of each other, we still have !a!m a∗!,−m " = 0
(exercise), so that (12) is satisfied even in this case. For each $, there are 2$ + 1 independent
real random variables: a!0 (which is always real), and Re a!m and Im a!m for m = 1, . . . , $.
The quantity we want to calculate from theory is the variance !|a!m |2 " to get a prediction for
the typical size of the a!m . From statistical isotropy also follows that these expectation values
depend only on $ not m. (The $ are related to the angular size of the anisotropy pattern, whereas
the m are related to “orientation” or “pattern”. See Fig. 10.) Since !|a!m |2 " is independent of
m, we can define
1 !
C! ≡ !|a!m |2 " = !|a!m |2 " , (14)
2$ + 1 m
and altogether we have
!a!m a∗!! m! " = δ!!! δmm! C! . (15)
This function C! (of integers $ ≥ 2) is called the (theoretical) angular power spectrum. It is
analogous to the power spectrum P(k) of density perturbations. For Gaussian perturbations, the
C! contains all the statistical information about the CMB temperature anisotropy. And this is
all we can predict from theory. Thus the analysis of the CMB anisotropy consists of calculating
the angular power spectrum from the observed CMB (a map like Figure 5) and comparing it to
the C! predicted by theory.3
3
In addition to the temperature anisotropy, the CMB also has another property, its polarization. There are two
additional power spectra related to the polarization, C!EE and C!BB , and one related to the correlation between
temperature and polarization, C!T E .
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 111

Figure 10: The three lowest multipoles ! = 1, 2, 3 of spherical harmonics. Left column: Y10 , Re Y11 ,
Im Y11 . Middle column: Y20 , Re Y21 , Im Y21 , Re Y22 , Im Y22 . Right column: Y30 , Re Y31 , Im Y31 , Re Y32 ,
Im Y32 , Re Y33 , Im Y33 . Figure by Ville Heikkilä.
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 112

Just like the 3D density power spectrum P(k) gives the contribution of scale k to the density
variance !δ(x)2 ", the angular power spectrum C! is related to the contribution of multipole " to
the temperature variance,
!" # $ ! $
δT (θ, φ) 2 % %
= a!m Y!m (θ, φ) a∗!! m! Y!∗! m! (θ, φ)
T
!m !! m!
%%
= Y!m (θ, φ)Y!∗! m! (θ, φ)!a!m a∗!! m! "
!!! mm!
% % % 2" + 1
= C! |Y!m (θ, φ)|2 = C! , (16)
m
4π
! !

where we used (15) and (9).

Thus, if we plot (2" + 1)C! /4π on a linear " scale, or "(2" + 1)C! /4π on a logarithmic "
scale, the area under the curve gives the temperature variance, i.e., the expectation value for the
squared deviation from the average temperature. It has become customary to plot the angular
power spectrum as "(" + 1)C! /2π, which is neither of these, but for large " approximates the
second case. The reason for this custom is explained later.
Equation (16) represents the expectation value from theory and thus it is the same for all
directions θ,φ. The actual, “realized”, value of course varies from one direction θ,φ to another.
We can imagine an ensemble of universes, otherwise like our own, but representing a different
realization of the same random process of producing the primordial perturbations. Then ! "
represents the average over such an ensemble.
Equation(16) can be generalized to the angular correlation function (exercise)
& '
δT (n̂) δT (n̂" ) 1 %
C(ϑ) ≡ = (2" + 1)C! P! (cos ϑ), , (17)
T T 4π
!

where ϑ is the angle between n̂ and n̂" .

9.2.3 Observed angular power spectrum

Theory predicts expectation values !|a!m |2 " from the random process responsible for the CMB
anisotropy, but we can observe only one realization of this random process, the set {a!m } of our
CMB sky. We define the observed angular power spectrum as the average

(! = 1 %
C |a!m |2 (18)
2" + 1 m

of these observed values. ) *2

The variance of the observed temperature anisotropy is the average of δT (θ,φ) T over the
celestial sphere,
+ , - +
1 δT (θ, φ) 2 1 % %
dΩ = dΩ a!m Y!m (θ, φ) a∗!! m! Y!∗! m! (θ, φ)
4π T 4π
!m !! m!
+
1 %%
= a!m a∗!! m! Y!m (θ, φ)Y!∗! m! (θ, φ)dΩ
4π
!m !! m! . /0 1 (19)
δ!!! δmm!
1 %% % 2" + 1
(! .
= |a!m |2 = C
4π 4π
!
.m /0 1 !
!!
(2!+1)C
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 113

Figure 11: The angular power spectrum C!! as observed by Planck. The observational results are the
red data points with small error bars. The green curve is the theoretical C! from a best-fit model, and
the light green band around it represents the cosmic variance corresponding to this C! . The quantity
plotted is actually D! ≡ T02 [!(! + 1)/(2π)]C! . Note that the !-axis is logarithmic until 50 and linear after
that. (This is Fig. 21 of [1].)

Contrast this with (16), which gives the variance of δT /T at an arbitrary location on the sky
over different realizations of the random process which produced the primordial perturbations;
whereas (19) gives the variance of δT /T of our given sky over the celestial sphere.

9.2.4 Cosmic Variance

!! is equal to C! , the theoretical spectrum of
The expectation value of the observed spectrum C
Eq. (14), i.e.,
!! # = C! ⇒ "C
"C !! − C! # = 0 , (20)
but its actual, realized, value is not, although we expect it to be close. The expected squared
difference between C !! and C! is called the cosmic variance (of C! ). We can calculate it using
the properties of (complex) Gaussian random variables (exercise). The answer is

!! − C! )2 # = 2
"(C C2 . (21)
2! + 1 !
!! and C! is smaller for higher !. This
We see that the expected relative difference between C
is because we have a larger (size 2! + 1) statistical sample of a!m available for calculating the
C!! .
The cosmic variance limits the accuracy of comparison of CMB observations with theory,
especially for large scales (low !). See Fig. 11.
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 114

9.3 Multipoles and scales

9.3.1 Rough correspondence
The different multipole numbers ! correspond to different angular scales, low ! to large scales
and high ! to small scales. Examination of the functions Y!m (θ, φ) reveals that they have an
oscillatory pattern on the sphere, so that there are typically ! “wavelengths” of oscillation around
a full great circle of the sphere. See Figs. 10 and 12.

Figure 12: Randomly generated skies containing only a single multipole !. Starting from top left: ! = 1
(dipole only), 2 (quadrupole only), 3 (octupole only), 4, 5, 6, 7, 8, 9, 10, 11, 12. Figure by Ville Heikkilä.

Thus the angle corresponding to this wavelength is

2π 360◦
ϑλ = = . (22)
! !
See Fig. 13. The angle corresponding to a “half-wavelength”, i.e., the separation between a
neighboring minimum and maximum is then
π 180◦
ϑres = = . (23)
! !
This is the angular resolution required of the microwave detector for it to be able to resolve the
angular power spectrum up to this !.
For example, COBE had an angular resolution of 7◦ allowing a measurement up to ! =
180/7 = 26, WMAP had resolution 0.23◦ reaching to ! = 180/0.23 = 783, and Planck had
resolution 5" , allowing the measurement of C! up to ! = 2160.4
4
In reality, there is no sharp cut-off at a particular !, the observational error bars just blow up rapidly around
this value of !.
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 115

Figure 13: The rough correspondence between multipoles ! and angles.

Figure 14: The comoving angular diameter distance relates the comoving size of an object and the angle
in which we see it.

The angles on the sky are related to actual physical distances via the angular diameter
distance dA , defined as the ratio of the physical length (transverse to the line of sight) and the
angle it covers (see Chapter 3),
λphys
dA ≡ . (24)
ϑ
Likewise, we defined the comoving angular diameter distance dcA by

λc
dcA ≡ (25)
ϑ
where λc = a−1 λphys = (1 + z)λphys is the corresponding comoving length. Thus dcA = a−1 dA =
(1 + z)dA . See Fig. 14.
Consider now the Fourier modes of our earlier perturbation theory discussion. A mode with
comoving wavenumber k has comoving wavelength λc = 2π/k. Thus this mode should show up
as a pattern on the CMB sky with angular size
λc 2π 2π
ϑλ = c = c = . (26)
dA kdA !

For the last equality we used the relation (22). From it we get that the modes with wavenumber
k contribute mostly to multipoles around

! = kdcA . (27)
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 116

9.3.2 Exact treatment

The above matching of wavenumbers with multipoles was of course rather naive, for two reasons:
1. The description of a spherical harmonic Y!m having an “angular wavelength” of 2π/" is
just a crude characterization. See Fig. 12.

2. The modes k are not wrapped around the sphere of last scattering, but the wave vector
forms a different angle with the sphere at different places.
The following precise discussion applies only for the case of a flat universe (K = 0 FRW
universe as the background), where one can Fourier expand functions on a time slice. We start
from the expansion of the plane wave in terms of spherical harmonics, for which we have the
result, Eq. (10), !
eik·x = 4π i! j! (kx)Y!m (x̂)Y!m
∗
(k̂) , (28)
!m
where the j! are spherical Bessel functions.
Consider now some function !
f (x) = fk eik·x (29)
k
on the t = tdec time slice. We want the multipole expansion of the values of this function on the
last scattering sphere. See Fig. 15. These are the values f (xx̂), where x ≡ |x| has a constant
value, the (comoving) radius of this sphere. Thus
"
∗
a!m = dΩx Y!m (x̂)f (xx̂)
! "
= ∗
dΩx Y!m (x̂)fk eik·x
k
!!" !
= 4π ∗
dΩx fk Y!m (x̂)i! j!! (kx)Y!! m! (x̂)Y!∗! m! (k̂)
k !! m!
!
! ∗
= 4πi fk j! (kx)Y!m (k̂) , (30)
k

where we used the orthonormality of the spherical harmonics. The corresponding result for a
Fourier transform f (k) is
"
4πi!
a!m = d3 kf (k)j! (kx)Y!m
∗
(k̂) . (31)
(2π)3
The j! are oscillating functions with decreasing amplitude. For large values of " the position
of the first (and largest) maximum is near kx = " (see Fig. 16).
Thus the a!m pick a large contribution from those Fourier modes k where

kx ∼ " . (32)

In a flat universe the comoving distance x (from our location to the sphere of last scattering)
and the comoving angular diameter distance dcA are equal, so we can write this result as

kdcA ∼ " . (33)

The conclusion is that a given multipole " acquires a contribution from modes with a range
of wavenumbers, but most of the contribution comes from near the value given by Eq. (27). This
concentration is tighter for larger ".
We shall use Eq. (27) for qualitative purposes in the following discussion.
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 117

Figure 15: A plane wave intersecting the last scattering sphere.

0 20 40 60 80 100
0.3
0.3 j2(x)
0.2
j3(x)
0.1 j4(x)
0.2
0

-0.1
0.1

-0.1

0 2 4 6 8 10 12 14 16 18 20
0 200 400 600 800 1000

0.02 0.01 j200(x)

0.005 j201(x)
0.015 j202(x)
0

0.01 -0.005

-0.01
0.005

-0.005

-0.01
180 190 200 210 220 230 240 250

Figure 16: Spherical Bessel functions j! (x) for ! = 2, 3, 4, 200, 201, and 202. Note how the first and
largest peak is near x = ! (but to be precise, at a slightly larger value). Figure by R. Keskitalo.
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 118

9.4 Important distance scales on the last scattering surface

9.4.1 Angular diameter distance to last scattering
In Chapter 3 we derived the formula for the comoving distance to redshift z,
! 1
da
dc (z) = H0−1 " (34)
1
1+z
Ω0 (a − a ) − ΩΛ (a − a4 ) + a2
2

(where we have approximated Ω0 ≈ Ωm +ΩΛ ) and the corresponding comoving angular diameter
distance
dcA (z) = fK (dc (z)) , (35)
where 
 −1/2 sin(K 1/2 x) ,
K K>0
fK (x) ≡ x , K=0 (36)

 −1/2
|K| sinh(|K|1/2 x) , K < 0.
We also define 
sin x ,
 k=1
fk (x) ≡ x , k=0 (37)


sinh x , k = −1 .
For the flat universe (K = k = 0, Ω0 = 1), the comoving angular diameter distance is equal to
the comoving distance,
dcA (z) = dc (z) (K = 0) . (38)
For the open (K < 0, Ω0 < 1) and closed (K > 0, Ω0 > 1) cases we can write Eq. (35) as
'" (
c H0−1 |Ωk | c
dA (z) = " fk d (z)
|Ωk | H0−1
' ! 1 ( (39)
−1 1 " da
= H0 " fk |Ωk | " .
|Ωk | 1
1+z
Ω0 (a − a2 ) − ΩΛ (a − a4 ) + a2

Thus dcA (z) ∝ H0−1 , and has some more complicated dependence on Ω0 and ΩΛ (or on Ωm and
ΩΛ ).
We are now interested in the distance to the last scattering sphere, i.e., dcA (zdec ), where
zdec ≈ 1090.
For the simplest case, ΩΛ = 0, Ωm = 1, the integral gives
! 1 ) *
c −1 dx −1 1
dA (zdec ) = H0 √ = 2H0 1− √ = 1.94H0−1 ≈ 2H0−1 , (40)
1 x 1 + zdec
1+z

where the last approximation corresponds to ignoring the contribution from the lower limit.
We shall consider two more general cases, of which the above is a special case of both:
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 119

Horizon distance for =0

10
angular diameter distance
9 distance

-1
Comoving distance in H0
7

0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0

Figure 17: The comoving distance dc (z = ∞) (dashed) and the comoving angular diameter distance
dcA (z = ∞) (solid) to the horizon in matter-only open universe. The vertical axis is the distance in units
of Hubble distance H0−1 and the horizontal axis is the density parameter Ω0 = Ωm . The distances to last
scattering, dc (zdec ) and dcA (zdec ) are a few per cent less.

a) Open universe with no dark energy: ΩΛ = 0 and Ωm = Ω0 < 1. Now the integral gives
! # 1 $
H0−1 " dx
c
dA (zdec ) = √ sinh 1 − Ωm "
1 − Ωm 1
1+z
(1 − Ωm )x2 + Ωm x
 
−1 # 1
H dx
= √ 0 sinh  ' 
1 − Ωm 1
1+z
2 Ωm
x + 1−Ωm x
! * * $
H0−1 1 − Ωm 1 − Ωm 1
= √ sinh 2 arsinh − 2 arsinh
1 − Ωm Ωm Ωm 1 + zdec
! * $
H −1 1 − Ωm H −1
≈ √ 0 sinh 2 arsinh = 2 0 , (41)
1 − Ωm Ωm Ωm

where again the approximation ignores the contribution from the lower limit (i.e., it actu-
ally gives the angular diameter distance to the horizon, dcA (z = ∞), in a model where we
ignore the effect of other energy density"components besides matter). In the last step we
used sinh 2x = 2 sinh x cosh x = 2 sinh x 1 + sinh2 x. We show this result (together with
dc (z = ∞)) in Fig. 17.
b) Flat universe with vacuum energy, ΩΛ + Ωm = 1. Here the integral does not give an
elementary function, but a reasonable approximation, which we shall use in the following,
is
2
dcA (zdec ) = dc (zdec ) ≈ 0.4 H0−1 . (42)
Ωm
The comoving distance dc (zdec ) depends on the expansion history of the universe. The longer
it takes for the universe to cool from Tdec to T0 (i.e., to expand by the factor 1 + zdec ), the longer
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 120

Figure 18: The geometry effect in a closed (top) or an open (bottom) universe affects the angle at which
we see a structure of given size at the last scattering surface, and thus its angular diameter distance.

distance the photons have time to travel. When a larger part of this time is spent at small
values of the scale factor, this distance gets a bigger boost from converting it to a comoving
distance. For open/closed universes the angular diameter distance gets an additional effect from
the geometry of the universe (the fK ), which acts like a “lens” to make the distant CMB pattern
at the last scattering sphere to look smaller or larger (see Fig. 18).

9.4.2 Hubble scale and the matter-radiation equality scale

Subhorizon (k ! H) and superhorizon (k " H) scales behave differently. Thus we want to
know which of the structures we see on the last scattering surface are subhorizon and which
are superhorizon. For that we need to know the comoving Hubble scale H at tdec . This was
discussed in Sec. 8.3.1. At that time both matter and radiation are contributing to the energy
density and the Hubble parameter. The scale which is just entering at t = tdec is
! "−1/2
−1 −1 −1 −1/2 −1 −1/2 Ωr
kdec ≡ Hdec = (1 + zdec )Hdec = (1 + zdec ) H0 Ωm 1+ (1 + zdec )
Ωm
−1/2 −1 −1/2
= Ωm (1 + 0.046 ωm ) 91 h−1 Mpc (43)
−1 is ρ /ρ at t
(using zdec = 1090; here 0.046 ωm r m dec ) and the corresponding multipole number on
the last scattering sphere is

#H ≡ kdec dcA (44)

! "1/2 # $
−1
Ωr 2/Ωm = 66 Ω−0.5
m $ 1 + 0.046ωm (ΩΛ = 0)
= (1 + zdec )1/2 Ω−1/2
m 1+ (1 + zdec ) × −1
Ωm 2/Ω0.4
m ≈ 66 Ω0.1
m 1 + 0.046ωm (Ω0 = 1)

The angle subtended by a half-wavelength π/k of this mode on the last scattering sphere is
% &
π 180◦ −1 2.7◦ Ω0.5
m
ϑH ≡ = = 1 + 0.046 ωm × (45)
#H #H 2.7◦ Ω−0.1
m
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 121

For Ωm ∼ 0.3, ΩΛ ∼ 0.7, h ∼ 0.7, !H ≈ 67 and ϑH ≈ 3.5◦ (the angle subtended by k−1 is 1.12◦ ).
Another important scale is keq , the scale which enters at the time of matter-radiation equality
teq , since the transfer function T (k) is bent at that point. Perturbations for scales k # keq
maintain essentially their primordial spectrum, whereas scales k $ keq have lost relative power
between their horizon entry and teq . This scale is
−1 −1 −1 −1
keq = Heq ∼ 13.7 Ω−1
m h
−2
Mpc = 4.6 × 10−3 Ω−1
m h H0 (46)

and the corresponding multipole number of these scales seen on the last scattering sphere is
!
c 2/Ωm = 440 h (ΩΛ = 0)
leq = keq dA = 219 Ωm h × (47)
2/Ω0.4
m ≈ 440 h Ω 0.6 (Ω = 1)
m 0

Later we will introduce the sound horizon at photon decoupling, another important scale.
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 122

9.5 CMB anisotropy from perturbation theory

We began this chapter with the observation, Eq. (1), that the CMB temperature anisotropy is
a sum of two parts, ! " ! " ! "
δT δT δT
= + , (48)
T obs T intr T jour
and that this separation is gauge # δTdependent.
$ We shall consider this in the conformal-Newtonian
gauge, since the second part, T jour , the integrated redshift perturbation along the line of
sight, is easiest to calculate in this gauge. (However, we won’t do the calculation here.5 )
The result of this calculation is
! " % %
δT
= − dΦ + (Φ̇ + Ψ̇)dt + vobs · n̂
T jour
%
= Φ(tdec , xls ) − Φ(t0 , 0) + (Φ̇ + Ψ̇)dt + vobs · n̂
%
Ψ≈Φ
≈ Φ(tdec , xls ) − Φ(t0 , 0) + 2 Φ̇dt + vobs · n̂ (49)

where the integral is from (tdec , xls ) to (t0 , 0) along the path of the photon (a null geodesic), and
˙ ≡ ∂/∂t. The origin 0 is located where the observer is. The last term, vobs · n̂, is the Doppler
effect from observer motion (assumed nonrelativistic), vobs being the observer velocity and n̂ the
direction we are looking at. The ls in xls is just to remind us that x lies somewhere on the last
scattering sphere. In the matter-dominated universe the Newtonian potential remains constant
in time, Φ̇ = 0, so we get a contribution from the integral only from epochs when radiation
or dark energy contributions to the total energy density, or the effect of curvature, cannot be
ignored. We can understand the above result as follows. If the potential is constant in time,
the blueshift the photon acquires when falling into a potential well is canceled by the redshift
from climbing up the well. Thus the net redshift/blueshift caused by gravitational potential
perturbations is just the difference between the values of Φ at the beginning and in the end.
However, if the potential is changing while the photon is traversing the well, this cancelation is
not exact, and we get the integral term to account for this effect.
The value of the potential perturbation at the observing site, Φ(t0 , 0) is the same for photons
coming from all directions. Thus it does not contribute to the observed anisotropy. It just
produces an overall shift in the observed average temperature. This is included in the observed
value T0 = 2.7255± 0.0006 K, and so we just ignore this term, not attempting to separate it from
the “correct” unperturbed value.6 The observer motion vobs causes a dipole (# = 1) pattern
5
It is done in my course on Cosmological Perturbation Theory, Sec. 25.
6
I don’t think the local value of the gravitational potential, Φ(t0 , 0) is known very well. In a quick search I
couldn’t locate literature discussing it. We live within an overdensity (Solar System + Galaxy + Local Group
+ Local Supercluster) and thus in a potential well, so Φ(t0 , 0) is negative, contributing a blueshift, and the true
unperturbed value of T0 is slightly lower than the observed value. The exact calculation of ωγ from T0 would
require the use of this corrected value. While I don’t have a good number it appears likely that the correction
Φ(t0 , 0) to T0 is smaller than the uncertainty in the T0 measurement. The order of magnitude of the Galactic
contribution (which is much larger than the Solar contribution) is Φ ∼ v 2 , where v = 240 km/s = 8 × 10−4 [4]
is the local rotation velocity of the Galaxy. (For a spherically symmetric mass distribution, Φ = −v 2 , where v is
the circular orbit velocity, is exact outside the mass distribution; if there is mass outside the orbit, it does not
contribute to the gravitational field, but it does contribute to the potential. Most of the mass of the Galaxy is
further away from the center than the Solar System, and therefore the Galactic contribution to the local potential
Φ(t0 , 0) is larger in absolute value than −v 2 = −6.4 × 10−7 .) The large scale contribution could be estimated by
cosmic flow velocities using linear perturbation theory, where velocity is related to the potential gradient; for the
matter-dominated growing solution, v = − 32 H−1 ∇Φ. The motion of the Local Group in the CMB rest frame has
v = 620 km/s = 2.07 × 10−3 [4], but we would need the relevant scale over which Φ varies to get an estimate of
Φ from this. Anyway, this scale is much smaller than the Hubble scale, so the order of magnitude estimate of the
linear contribution to Φ(t0 , 0) is much less than this v.
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 123

in the CMB anisotropy, and likewise, we do not attempt to separate from it the cosmological
dipole on the last scattering sphere. Therefore the dipole is usually removed from the CMB map
before analyzing it for cosmological purposes. Accordingly, we shall ignore this term also, and
our final result is ! " #
δT
= Φ(tdec , xls ) + 2 Φ̇dt . (50)
T jour
$ %
The other part, δT T intr , comes from the local temperature perturbation at t = tdec and the
Doppler effect, −v · n̂, from the local (baryon+photon) fluid motion at that time. Since

π2 4
ργ = T , (51)
15
the local temperature perturbation is directly related to the relative perturbation in the photon
energy density, and ! "
δT 1
= δγ − v · n̂ . (52)
T intr 4
We can now write the observed temperature anisotropy as
! " #
δT 1
= δγN − vN · n̂ + Φ(tdec , xls ) + 2 Φ̇dt . (53)
T obs 4

(note that both the density perturbation δγ and the fluid velocity v are gauge dependent).
To make further progress we now

1. consider adiabatic primordial perturbations only (like we did in Chapter 8), and

2. make the (crude) approximation that the universe is already matter dominated at t = tdec .

For adiabatic perturbations

3
δb = δc ≡ δm = δγ . (54)
4
The perturbations stay adiabatic only at superhorizon scales. Once the perturbation has entered
horizon, different physics begins to act on different matter components, so that the adiabatic
relation between their density perturbations is broken. In particular, the baryon+photon pertur-
bation is affected by photon pressure, which will damp their growth and cause them to oscillate,
whereas the CDM perturbation is unaffected and keeps growing. Since the baryon and photon
components see the same pressure they still evolve together and maintain their adiabatic relation
until photon decoupling. Thus, after horizon entry, but before decoupling,
3
δc #= δb = δγ . (55)
4
At decoupling, the equality holds for scales larger than the photon mean free path at tdec .
After decoupling, this connection between the photons and baryons is broken, and the baryon
density perturbation begins to approach the CDM density perturbation,
3
δc ← δb #= δγ . (56)
4
We shall return to these issues as we discuss the shorter scales in Sections 9.7 and 9.8. But
let us first discuss the scales which are still superhorizon at tdec , so that Eq. (54) still applies.
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 124

9.6 Large scales: Sachs–Wolfe part of the spectrum

Consider now the scales k ! kdec , or ! ! !H , which are still superhorizon at decoupling. We
can now use the adiabatic condition (54), so that
1 1 1
δγ = δm ≈ δ , (57)
4 3 3
where the latter (approximate) equality comes from taking the universe to be matter dominated
at tdec , so that we can identify δ ≈ δm . For these scales the Doppler effect from fluid motion is
subdominant, and we can ignore it (the fluid is set into motion by gradients in the pressure and
gravitational potential, but the time scale of getting into motion is longer than the Hubble time
for superhorizon scale gradients).
Thus Eq. (53) becomes
! " #
δT 1 N
= δ + Φ(tdec , xls ) + 2 Φ̇dt . (58)
T obs 3

The Newtonian relation

! "2
1 2 1
δ = 2
∇2 Φ = ∇2 Φ
4πGρ̄a 3 aH

(here ∇ is with respect to the comoving coordinates, hence the a−2 ) or

! "2
2 k
δk = − Φk
3 H

does not hold at superhorizon scales (where δ is gauge dependent). A GR calculation using the
Newtonian gauge gives the result7
$ ! "2 %
2 k
δkN = − 2+ Φk (59)
3 H

for perturbations in a matter-dominated universe. Thus for superhorizon scales we can approx-
imate
δN ≈ −2Φ (60)
and Eq. (58) becomes
! " #
δT 2
= − Φ(tdec , xls ) + Φ(tdec , xls ) + 2 Φ̇dt
T obs 3
#
1
= Φ(tdec , xls ) + 2 Φ̇dt . (61)
3

This explains the “mysterious” factor 1/3 in this relation between the potential Φ and the
temperature perturbation.
This result is called the Sachs–Wolfe effect. The first part, 13 Φ(tdec , xls ), is called the ordinary
&
Sachs–Wolfe effect, and the second part, 2 Φ̇dt, the integrated Sachs-Wolfe effect (ISW), since
it involves integrating along the line of sight. Note that the approximation of matter domination
at t = tdec , making Φ̇ = 0, does not eliminate the ISW, since it only applies to the “early part”
of the integral. At times closer to t0 , dark energy becomes important, causing Φ to evolve again.
7
Cosmological Perturbation Theory, Sec. 13.
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 125

This ISW caused by dark energy (or curvature of the background universe, if k != 0) is called the
late Sachs–Wolfe effect (LSW) and it shows up as a rise in the smallest ! of the angular power
spectrum C! . Correspondingly, the contribution to the ISW from the evolution of Φ near tdec
due to the radiation contribution to the expansion law (which we ignored in our approximation)
is called the early Sachs–Wolfe effect (ESW). The ESW shows up as a rise in C! for larger !,
near !H .
We shall now forget for a while the ISW, which for ! " !H is expected to be smaller than
the ordinary Sachs–Wolfe effect.

9.6.1 Angular power spectrum from the ordinary Sachs–Wolfe effect

We now calculate the contribution from the ordinary Sachs–Wolfe effect,
! "
δT 1
= Φ(tdec , xls ) , (62)
T SW 3

to the angular power spectrum C! . This is the dominant effect for ! " !H .
Since Φ is evaluated at the last scattering sphere, we have, from Eq. (30),
#1
a!m = 4πi! ∗
Φk j! (kx)Y!m (k̂) , (63)
3
k

In the matter-dominated epoch,

3
Φ = − R, (64)
5
so that
4π ! # ∗
a!m = − i Rk j! (kx)Y!m (k̂) . (65)
5
k
The coefficient a!m is thus a linear combination of the independent random variables Rk ,
i.e., it is of the form #
bk Rk , (66)
k
For any such linear combination, the expectation value of its absolute value squared is
$% %2 &
%# % ##
% %
% bk Rk % = bk b∗k! $Rk R∗k! %
% %
k k k!
! "3 #
2π 1
= PR (k) |bk |2 , (67)
L 4πk3
k

where we used ! "3

2π 1
$Rk R∗k! % = δkk! PR (k) (68)
L 4πk3
(the independence of the random variables Rk and the definition of the power spectrum P(k)).
Thus
1 #
C! ≡ $|a!m |2 %
2! + 1 m
! " % %2
16π 2 1 # 2π 3 # 1 2% ∗ %
= PR (k)j! (kx) (
%Y!m %
k̂)
25 2! + 1 m L 4πk3
k
! "3 #
1 2π 1
= PR (k)j! (kx)2 . (69)
25 L k3
k
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 126

(Although all !|a!m |2 " are equal for the same !, we used the sum over m, so that we could use
Eq. (9).) Replacing the sum with an integral, we get
! 3
1 d k
C! = PR (k)j! (kx)2
25 k3
!
4π ∞ dk
= PR (k)j! (kx)2 , (70)
25 0 k

where x = dcA (zdec ), the final result for an arbitrary primordial power spectrum PR (k).
The integral can be done for a power-law power spectrum, PR (k) = A2s (k/kp )n−1 . In partic-
ular, for a scale-invariant (n = 1) primordial power spectrum,

PR (k) = const. = A2s , (71)

we have !
4π ∞
dk A2 2π
C! = A2s j! (kx)2 = s , (72)
25 0 k 25 !(! + 1)
since ! ∞
dk 1
j! (kx)2 = . (73)
0 k 2!(! + 1)
We can write this as
!(! + 1) A2
C! = s = const. (independent of !) (74)
2π 25
This is the reason why the angular power spectrum is customarily plotted as !(! + 1)C! /2π;
it makes the ordinary Sachs–Wolfe part of the C! flat for a scale-invariant primordial power
spectrum PR (k).
Observations are consistent with an almost scale-invariant primordial power spectrum (they
favor a small red tilt, n < 1). The constant As can be determined from the ordinary Sachs–Wolfe
part of the observed C "! . From Fig. 11 we see that at low !

!(! + 1) " 800 µK2

C! ∼ ∼ 10−10 (75)
2π (2.725 K)2

on the average. This gives the amplitude of the primordial power spectrum as
# $2
PR (k) = A2s ∼ 25 × 10−10 = 5 × 10−5 . (76)

We already used this result in Chapter 8 as a constraint on the energy scale of inflation.

Exercise: Find the C! of the ordinary Sachs-Wolfe effect due to a power-law power spectrum PR (k) =
A2s (k/kp )n−1 . Help:
! ∞
Γ(! + n2 − 21 )Γ(3 − n)
dx xn−2 j!2 (x) = 2n−4 π . (77)
0 Γ(! + 25 − n2 )Γ(2 − n2 )2
Take As = 4.58 × 10−5 , for a pivot scale kp = 0.05 Mpc−1 , and n = 0.965 (Planck 2018 central values).
−1
Give the numerical values for C2 and C20 . Use dcA (zdec ) = 2Ω−0.4
m H0 , with Ωm = 0.315 and H0 =
67.36 km/s/Mpc (Planck 2018 central values). Give also D2 and D20 , where D! ≡ T02 [!(! + 1)/(2π)]C! ,
and compare to Fig. 11. What explains the difference?
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 127

9.7 Acoustic oscillations

Consider now the scales k ! kdec , or ! ! !H , which are subhorizon at decoupling. The observed
temperature anisotropy is, from Eq. (53)
! " #
δT 1
= δγ (tdec , xls ) + Φ(tdec , xls ) − vγ · n̂(tdec , xls ) + 2 Φ̇dt . (78)
T obs 4
Since we are considering subhorizon scales, we dropped the reference to the Newtonian gauge.
We shall concentrate on the three first terms, which correspond to the situation at the point
(tdec , xls ) we are looking at on the last scattering sphere.
Before decoupling photons are coupled to baryons. Perturbations in the baryon-photon fluid
are oscillating, whereas CDM perturbations grow (slowly during the radiation-dominated epoch,
and then faster during the matter-dominated epoch). Therefore CDM perturbations begin to
dominate the total density perturbation δρ and thus also Φ already before the universe becomes
matter dominated and CDM begins to dominate the background energy density. Thus we make
the approximation that Φ is given by the CDM perturbation. The baryon-photon fluid oscillates
in these potential wells caused by the CDM. The potential Φ evolves at first but then becomes
constant as the universe becomes matter dominated.
We shall not attempt an exact calculation of the δbγ oscillations in the expanding universe.
One reason is that ρbγ is a relativistic fluid, and we have derived the perturbation equations
for a nonrelativistic fluid only. From Sec. 8.2.7 we have that the nonrelativistic perturbation
equation for a fluid component i is
! "
k2 δpki
δ̈ki + 2H δ̇ki = − 2 + Φk . (79)
a ρ̄i
The generalization of the (subhorizon) perturbation equations to the case of a relativistic
fluid is considerably easier if we ignore the expansion of the universe. Then Eq. (79) becomes
! "
2 δpki
δ̈ki + k + Φk = 0 . (80)
ρ̄i
According to GR, the density of “passive gravitational mass” is ρ + p = (1 + w)ρ, not just ρ
as in Newtonian gravity. Therefore the force on a fluid element of the fluid component i is
proportional to (ρi + pi )∇Φ = (1 + wi )ρi ∇Φ instead of just ρi ∇Φ, and Eq. (80) generalizes to
the case of a relativistic fluid as8
$ %
δpki
δ̈ki + k2 + (1 + wi )Φk = 0 . (81)
ρ̄i
In the present application the fluid component ρi is the baryon-photon fluid ρbγ and the
gravitational potential Φ is caused by the CDM. Before decoupling, the adiabatic relation δb =
3
4 δγ still holds between photons and baryons, and we have the adiabatic relation between pressure
and density perturbations,
δpbγ = c2s δρbγ . (82)
Thus we have & '
δ̈bγk + k2 c2s δbγk + (1 + wbγ )Φk = 0 . (83)
Here
δpbγ δpγ 1 δργ 1 ρ̄γ δγ 1 1 1 1
c2s = ≈ = = = ≡ (84)
δρbγ δρbγ 3 δργ + δρb 3 ρ̄γ δγ + ρ̄b δb 3 1 + 34 ρ̄ρ̄b 31+R
γ

8
Actually the derivation is more complicated, since also the density of “inertial mass” is ρi + pi and the energy
continuity equation is modified by a work-done-by-pressure term. The more detailed derivation of Eq. (81) was
given in Sec. 8.2.8.
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 128

gives the speed of sound cs of the baryon-photon fluid. We defined

3 ρ̄b
R≡ . (85)
4 ρ̄γ

The equation-of-state parameter for the baryon-photon fluid is

1
p̄bγ ρ̄γ 1 1
wbγ ≡ = 3 = , (86)
ρ̄bγ ρ̄γ + ρ̄b 3 1 + 43 R

so that
4
3 (1+ R)
1 + wbγ = (87)
1 + 43 R
and we can write Eq. (83) as
! "
4
2 1 1 3 (1 + R)
δ̈bγk + k δbγk + Φk = 0 . (88)
31+R 1 + 34 R

For the CMB anisotropy we are interested in9

1
Θ0 ≡ δγ , (89)
4
which gives the local temperature perturbation, not in δbγ . These two are related by

δρbγ δργ + δρb ρ̄γ δγ + ρ̄b δb 1+R

δbγ = = = = δγ . (90)
ρ̄bγ ρ̄γ + ρ̄b ρ̄γ + ρ̄b 1 + 34 R

Thus we can write Eq. (83) as (we are ignoring the expansion of the universe, so R is constant)
# $
2 1 1 4
δ̈γk + k δγk + Φk = 0 , (91)
31+R 3
or # $
2 1 1 1
Θ̈0k + k Θ0k + Φk = 0 , (92)
31+R 3
or
Θ̈0k + c2s k2 [Θ0k + (1 + R)Φk ] = 0 , (93)
If we now take R and Φk to be constant, this is the harmonic oscillator equation for the
quantity Θ0k + (1 + R)Φk with the general solution

Θ0k + (1 + R)Φk = Ak cos cs kt + Bk sin cs kt , (94)

or
Θ0k + Φk = −RΦk + Ak cos cs kt + Bk sin cs kt , (95)
or
Θ0k = −(1 + R)Φk + Ak cos cs kt + Bk sin cs kt . (96)
We are interested in the quantity Θ0 +Φ = 14 δγ +Φ, called the effective temperature perturbation,
since this combination appears in Eq. (78). It is the local temperature perturbation minus the
redshift photons suffer when climbing from the potential well of the perturbation (negative Φ
9
The subscript 0 refers to the monopole (! = 0) of the local photon distribution. Likewise, the dipole (! = 1)
of the local photon distribution corresponds to the velocity of the photon fluid, Θ1 ≡ vγ /3.
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 129

for a CDM overdensity). We see that this quantity oscillates in time, and the effect of baryons
(via R) is to shift the equilibrium point of the oscillation by −RΦk .
In the preceding we ignored the effect of the expansion of the universe. The expansion
affects the preceding in a number of ways. For example, cs , wbγ and R change with time. The
potential Φ also evolves, especially at the earlier times when radiation dominates the expansion
law. However, the qualitative result of an oscillation of Θ0 + Φ, and the shift of its equilibrium
point by baryons, remains. The time t in the solution (95) gets replaced by conformal time η,
and since cs changes with time, cs η is replaced by
! η ! t
cs (t)
rs (t) ≡ cs dη = dt . (97)
0 0 a(t)

We call this quantity rs (t) the sound horizon at time t, since it represents the comoving distance
sound has traveled by time t.
The relative weight of the cosine and sine solutions (i.e., the constants Ak and Bk in Eq. (94)
depends on the initial conditions. Since the perturbations are initially at superhorizon scales,
the initial conditions are determined there, and the present discussion does not really apply.
However, using the Newtonian gauge superhorizon initial conditions gives the correct qualitative
result for the phase of the oscillation.
We had that for adiabatic primordial perturbations, initially Φ = − 35 R and 14 δγN = − 32 Φ =
2 1 1
5 R, giving us an initial condition Θ0 + Φ = 3 Φ = − 5 R = const. (At these early times R # 1,
so we don’t write the 1 + R.) Thus adiabatic primordial perturbations correspond essentially to
the cosine solution.10 (There are effects at the horizon scale which affect the amplitude of the
oscillations—the main effect being the decay of Φ as it enters the horizon—so we can’t use the
preceding discussion to determine the amplitude, but we get the right result about the initial
phase of the Θ0 + Φ oscillations.)
Thus we have that, qualitatively, the effective temperature behaves at subhorizon scales as
Θ0k + (1 + R)Φk ∝ cos krs (t) , (98)
Consider a region which corresponds to a positive primordial curvature perturbation R. It
begins with an initial overdensity (of all components, photons, baryons, CDM and neutrinos),
and a negative gravitational potential Φ. For the scales of interest for CMB anisotropy, the
potential stays negative, since the CDM begins to dominate the potential early enough and the
CDM perturbations do not oscillate, they just grow. The effective temperature perturbation
Θ0 + Φ, which is the oscillating quantity, begins with a negative value. After half an oscillation
period it is at its positive extreme value. This increase of Θ0 +Φ corresponds to an increase in δγ ;
from its initial positive value it has grown to a larger positive value. Thus the oscillation begins
by the, already initially overdense, baryon-photon fluid falling deeper into the potential well,
and reaching its maximum compression after half a period. After this maximum compression
the photon pressure pushes the baryon-photon fluid out from the potential well, and after a full
period, the fluid reaches its maximum decompression in the potential well. Since the potential Φ
has meanwhile decayed (horizon entry and the resulting potential decay always happens during
the first oscillation period, since the sound horizon and the Hubble length are close to each
other, as the sound speed is close to the speed of light), the decompression does not bring the
δbγ back to its initial value (which was overdense), but the photon-baryon fluid actually becomes
underdense in the potential well (and overdense in the neighboring potential “hill”). And so the
oscillation goes on until photon decoupling.
These are standing waves and they are called acoustic oscillations. See Fig. 19. Because of
the potential decay at horizon entry, the amplitude of the oscillation is larger than Φ, and thus
also Θ0 changes sign in the oscillation.
10
The sine solution corresponds to what are called isocurvature primordial perturbations.
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 130

Figure 19: Acoustic oscillations. The top panel shows the time evolution of the Fourier amplitudes Θ0k ,
Φk , and the effective temperature Θ0k + Φk . The Fourier mode shown corresponds to the fourth acoustic
peak of the C! spectrum. The bottom panel shows δbγ (x) for one Fourier mode as a function of position
at various times (maximum compression, equilibrium level, and maximum decompression).

These oscillations end at photon decoupling, when the photons are liberated. The CMB
shows these standing waves as a snapshot11 at their final moment t = tdec .
At photon decoupling we have

Θ0k + (1 + R)Φk ∝ cos krs (tdec ) . (99)

At this moment oscillations for scales k which have

krs (tdec ) = mπ (100)

(m = 1, 2, 3, . . .) are at their extreme values (maximum compression or maximum decompres-

sion). Therefore we see strong structure in the CMB anisotropy at the multipoles

dcA (tdec )
# = kdcA (tdec ) = mπ ≡ m#A (101)
rs (tdec )

corresponding to these scales. Here

dcA (tdec ) π
#A ≡ π ≡ (102)
rs (tdec ) ϑs

is the acoustic scale in multipole space and

rs (tdec )
ϑs ≡ (103)
dcA (tdec )
11
Actually, photon decoupling takes quite a long time. Therefore this “snapshot” has a rather long “exposure
time” causing it to be “blurred”. This prevents us from seeing very small scales in the CMB anisotropy.
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 131

is the sound horizon angle, i.e., the angle at which we see the sound horizon on the last scattering
surface.
Because of these acoustic oscillations, the CMB angular power spectrum C! has a structure
of acoustic peaks at subhorizon scales. The centers of these peaks are located approximately at
!m ≈ m!A . An exact calculation shows that they will actually lie at somewhat smaller ! due to
a number of effects. The separation of neighboring peaks is closer to !A than the positions of
the peaks are to m!A .
These acoustic oscillations involve motion of the baryon-photon fluid. When the oscillation
of one Fourier mode is at its extreme, e.g., at the maximal compression in the potential well, the
fluid is momentarily at rest, but then it begins flowing out of the well until the other extreme, the
maximal decompression, is reached. Therefore those Fourier modes k which have the maximum
effect on the CMB anisotropy via the 41 δγ (tdec , xls ) + Φ(tdec , xls ) term (the effective temperature
effect) in Eq. (78) have the minimum effect via the −v · n̂(tdec , xls ) term (the Doppler effect) and
vice versa. Therefore the Doppler effect also contributes a peak structure to the C! spectrum,
but the peaks are in the locations where the effective temperature contribution has troughs.
The Doppler effect is subdominant to the effective temperature effect, and therefore the peak
positions in the C! spectrum are determined by the effective temperature effect, according to
Eq. (101). The Doppler effect just partially fills the troughs between the peaks, weakening the
peak structure of C! . See Fig. 22.
Fig. 20 shows the values of the effective temperature perturbation Θ0 + Φ (as well as Θ0 and
Φ separately) and the magnitude of the velocity perturbation (Θ1 ∼ v/3) at tdec as a function of
the scale k. This is a result of a numerical calculation which includes the effect of the expansion
of the universe, but not diffusion damping (Sec. 9.8).

9.8 Diffusion damping

For small enough scales the effect of photon diffusion and the finite thickness of the last scat-
tering surface (∼ the photon mean free path just before last scattering) smooth out the photon
distribution and the CMB anisotropy.
−1
This effect can be characterized by the damping scale kD ∼ photon diffusion length ∼
geometric mean of the Hubble time and photon mean free path λγ . Actually λγ is increasing
rapidly during recombination, so a calculation of the diffusion scale involves an integral over
time which includes this effect.
A calculation, that we shall not do here,12 gives that photon density and velocity perturba-
tions at scale k are damped at tdec by
2 2
e−k /kD , (104)
where the diffusion scale is !
−1 1 1 λγ (tdec )
kD ∼ . (105)
few a Hdec
Accordingly, the C! spectrum is also damped as
2 /!2
e−! D (106)

where
!D ∼ kD dcA (tdec ) . (107)
For typical values of cosmological parameters !D ∼ 1500. See Fig. 21 for a result of a numerical
calculation with and without diffusion damping.
12
See, e.g., Dodelson [9], Chapter 8.
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 132

0.5

Φ, Θ0
0

-0.5

0.5

(Θ0 + Φ)
0

-0.5

0.4

0.2

Θ1
0

-0.2

-0.4 ωm = 0.10
ωm = 0.20
0.6 ωm = 0.30
0.5 2
(Θ0 + Φ)
0.4
0.3
0.2
0.1
0

0.15
2

0.1
Θ1

0.05

0
0 200 400 600 800 1000
k/H0
Figure 20: Values of oscillating quantities (normalized to an initial value Rk = 1) at the time of
decoupling as a function of the scale k, for three different values of ωm , and for ωb = 0.01. Θ1 repre-
sents the velocity perturbation. The effect of diffusion damping is neglected. Figure and calculation by
R. Keskitalo.
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 133

0.7

ωm = 0.20
Undamped spectra
0.6 ωm = 0.25
ωm = 0.30
ωm = 0.35
0.5

0.4

0.3

0.2

0.1
Including damping

0
0 500 1000 1500 2000
Figure 21: The angular power spectrum C! , calculated both with and without the effect of diffusion
damping. The spectrum is given for four different values of ωm , with ωb = 0.01. (This is a rather low
value of ωb , so "D < 1500 and damping is quite strong.) Figure and calculation by R. Keskitalo.

Of the cosmological parameters, the damping scale is the most strongly dependent on ωb ,
since increasing the baryon density shortens the photon mean free path before decoupling. Thus
for larger ωb the damping moves to shorter scales, i.e., "D becomes higher (there is less damping).
(Of course, decoupling only happens as the photon mean free path becomes comparable to
the Hubble length, so one might think that λγ at tdec should be independent of ωb . However
there is a distinction here between whether a photon will not scatter again after a particular
scattering and what was the mean free path between the second-to-last and the last scattering.
And kD depends on an integral over the past history of the photon mean free path, not just the
last one. The factor 1/few in Eq. (105) comes from that integration, and actually depends on
ωb . For small ωb the λγ has already become quite large through the slow dilution of the baryon
density by the expansion of the universe, and relies less on the fast reduction of free electron
density due to recombination. Thus the time evolution of λγ before decoupling is different for
different ωb and we get a different diffusion scale.)
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 134

9.9 The complete C! spectrum

As we have discussed the CMB anisotropy has three contributions (see Eq. 78), the effective
temperature effect,
1
δγ (tdec , xls ) + Φ(tdec , xls ) , (108)
4
the Doppler effect,
−v · n̂(tdec , xls ) , (109)
and the integrated Sachs–Wolfe effect,
! t0
2 Φ̇(t, x(t))dt . (110)
tdec

Since the C! is a quadratic quantity, it also includes cross terms between these three effects.
The calculation of the full C! proceeds much as the calculation of just the ordinary Sachs–
Wolfe part (which the effective temperature effect becomes at superhorizon scales) in Sec. 9.6.1,
but now with the full δT /T . Since all perturbations are proportional to the primordial pertur-
bations, the C! spectrum is proportional to the primordial perturbation spectrum PR (k) (with
integrals over the spherical Bessel functions j! (kx), like in Eq. (70), to get from k to ").
The difference is that instead of the constant proportionality factor (δT /T )SW = −(1/5)R,
we have a k-dependent proportionality resulting from the evolution (including, e.g., the acoustic
oscillations) of the perturbations.
In Fig. 22 we show the full C! spectrum and the different contributions to it.
Because the Doppler effect and the effective temperature effect are almost completely off-
phase, their cross term gives a negligible contribution.
Since the ISW effect is relatively weak, it contributes more via its cross terms with the
Doppler effect and effective temperature than directly. The cosmological model used for Fig. 22
has ΩΛ = 0, so there is no late ISW effect (which would contribute at the very lowest "), and
the ISW effect shown is the early ISW effect due to radiation contribution to the expansion law.
This effect contributes mainly to the first peak and to the left of it, explaining why the first
peak is so much higher than the other peaks. It also shifts the first peak position slightly to the
left and changes its shape.
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 135

Full spectrum
0.3
Θ0 + Φ
Θ1
ISW
ISW cross terms
(Θ0+Φ)×Θ1
0.2
2l(l+1)Cl/2π

0.1

0 500 1000 1500 2000

l
0.2

0.15 0.001

(Θ0+Φ)×Θ1
Θ0+Φ

0.1 0

0.05 -0.001

0.08 ISW×(Θ0+Φ)
0.08
ISW×Θ1 0.06
0.06
0.04
Θ1

0.04 0.02
0.02 0
-0.02
0
0.02 ISW×(Θ0+Φ) + ISW×Θ1 0.08
0.01
0.01
0.06
0.04
ISW

0.001 0
0 500 1000 1500 0.02
0.0001 0
-0.02
1e-05
0 500 1000 1500 2000
l

Figure 22: The full C! spectrum calculated for the cosmological model Ω0 = 1, ΩΛ = 0, ωm = 0.2,
ωb = 0.03, As = 1, ns = 1, and the different contributions to it. (The calculation involves some
approximations which allow the description of C! as just a sum of these contributions and is not as
accurate as a CMBFAST or CAMB calculation.) Here Θ1 denotes the Doppler effect. Figure and
calculation by R. Keskitalo.
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 136

Figure 23: Spacetime diagram of the rescattering of CMB photons due to free electrons released in
reionization. The rescattering continues after reionization, but most of it happens relatively soon after
it, since the ne is diluted by expansion.

9.10 Reionization and optical depth

When radiation from the first stars reionizes the intergalactic gas, CMB photons may scatter
from the resulting free electrons. The optical depth τ due to reionization is the expectation
number of such scatterings per CMB photon. It is expected to be less than 0.1, i.e., most
CMB photons do not scatter at all. This rescattering causes additional polarization13 of the
CMB, and CMB polarization measurements are actually the best way to determine τ . Most of
this scattering happens relatively soon after the reionization, since the number density of free
electrons is diluted by the expansion of the Universe.
The optical depth is thus directly related to the reionization redshift zreion . A smaller τ
corresponds to later reionization and thus means that the first stars formed later.
Because of this scattering, not all the CMB photons come from the location on the last
scattering surface they seem to come from. The effect of the rescattered photons is to mix up
signals from different directions and therefore to reduce the CMB anisotropy. The reduction
factor on δT /T is e−τ and on the C" spectrum e−2τ . However, this does not affect the largest
scales, scales larger than the area from which the rescattered photons, reaching us from a certain
direction, originally came from. Such a large-scale anisotropy has affected all such photons the
same way, and thus is not lost in the mixing. See Fig. 23.

13
Due to time constraints, CMB polarization is not discussed in these lectures.
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 137

9.11 Cosmological parameters and CMB anisotropy

Let us finally consider the total effect of the various cosmological parameters on the C! spec-
trum. The C! provides the most important single observational data set for determining (or
constraining) cosmological parameters, since it has a rich structure which we can measure with
an accuracy that other cosmological observations cannot match, and because it depends on so
many different cosmological parameters in many ways. The latter is both a strength and weak-
ness: the number of cosmological parameters we can determine is large, but on the other hand,
some feature in C! may depend on more than one parameter, so that we may only be able to
constrain some combination of such parameters, not the parameters individually. We say that
such parameters are degenerate in the CMB data. Other cosmological observations are then
needed to break such degeneracies.
We shall consider 7 “standard parameters”:

• Ω0 total density parameter

• ΩΛ cosmological constant (or vacuum energy) density parameter

• As amplitude of primordial scalar perturbations (at some pivot scale kp )

• ns spectral index of primordial scalar perturbations

• τ optical depth due to reionization (discussed in Sec. 9.11.6)

• ωb ≡ Ωb h2 “physical” baryon density parameter

• ωm ≡ Ωm h2 “physical” matter density parameter

There are other possible cosmological parameters (“additional parameters”) which might
affect the C! spectrum, e.g.,

• mν i neutrino masses

• w dark energy equation-of-state parameter

dns
• d ln k scale dependence of the spectral index

• r, nT relative amplitude and spectral index of tensor perturbations

• B, niso amplitudes and spectral indices of primordial isocurvature perturbations

• Acor , ncor and their correlation with primordial curvature perturbations

We assume here that these additional parameters have no impact, i.e., they have the “standard”
values
dn !
r= = B = Acor = 0 , w = −1 , and mνi = 0.06 eV , (111)
d ln k
to the accuracy which matters for C! observations. This is both observationally and theoretically
reasonable. There is no sign in the present-day CMB data for deviations from these values. On
the other hand, significant deviations can be consistent with the current data, and may be
discovered by more accurate future observations. The primordial isocurvature perturbations
refer to the possibility that the primordial scalar perturbations are not adiabatic, and therefore
are not completely determined by the comoving curvature perturbation R.
The assumption that these additional parameters have no impact, leads to a determination
of the standard parameters with an accuracy that may be too optimistic, since the standard
parameters may have degeneracies with the additional parameters.
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 138

9.11.1 Independent vs. dependent parameters

The above is our choice of independent cosmological parameters. Ωm , Ωb and H0 (or h) are then
dependent (or “derived”) parameters, since they are determined by

Ω0 = Ωm + ΩΛ ⇒ Ωm = Ω0 − ΩΛ (112)
! !
ωm ωm
h= = (113)
Ωm Ω0 − ΩΛ
ωb ωb
Ωb = 2 = (Ω0 − ΩΛ ) (114)
h ωm

Note that the Hubble constant H0 ≡ h · 100 km/s/Mpc is now a dependent parameter! We
cannot vary it independently, but rather the varying of ωm , Ω0 , or ΩΛ also causes H0 to change.
Different choices of independent parameters are possible within our 7-dimensional parameter
space (e.g., we could have chosen H0 to be an independent parameter and let ΩΛ to be a
dependent parameter instead). They can be thought of as different coordinate systems14 in this
7D space. It is not meaningful to discuss the effect of one parameter without specifying what is
your set of independent parameters!
Some choices of independent parameters are better than others. The above choice represents
standard practice in cosmology today.15 The independent parameters have been chosen so
that they correspond as directly as possible to physics affecting the C! spectrum and thus to
observable features in it. We want the effects of our independent parameters on the observables
to be as different (“orthogonal”) as possible in order to avoid parameter degeneracy.
In particular,

• ωm (not Ωm ) determines zeq and keq , and thus, e.g., the magnitude of the early ISW effect
and which scales enter during matter- or radiation-dominated epochs.

• ωb (not Ωb ) determines the baryon/photon ratio and thus, e.g., the relative heights of the
odd and even peaks.

• ΩΛ (not ΩΛ h2 ) determines the late ISW effect.

There are many effects on the C! spectrum, and parameters act on them in different combi-
nations. Thus there is no perfectly “clean” way of choosing independent parameters. Especially
having the Hubble constant as a dependent parameter takes some getting used to.
In the following CAMB16 plots we see the effect of these parameters on C! by varying one
parameter at a time around a reference model, whose parameters have the following values.
Independent parameters:

Ω0 = 1 ΩΛ = 0.7
As = 1 ωm = 0.147
ns = 1 ωb = 0.022
τ = 0.1
14
The situation is analogous to the choice of independent thermodynamic variables in thermodynamics.
15
There are other choices in use, which are even more geared to minimizing parameter degeneracy. For example,
the sound horizon angle ϑs may be used instead of ΩΛ as an independent parameter, since it is directly determined
by the acoustic peak separation. However, since the determination of the dependent parameters from it is
complicated, such use is more directed towards technical data analysis than pedagogical discussion.
16
CAMB is a publicly available code for precise calculation of the C! spectrum. See http://camb.info
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 139

which give for the dependent parameters

Ωm = 0.3 h = 0.7
Ωc = 0.2551 ωc = 0.125
Ωb = 0.0449

The meaning of setting As = 1 is just that the resulting C! still need to be multiplied by the
true value of A2s . (In this model the true value should be about As = 5 × 10−5 to agree with
observations.) If we really had As = 1, perturbation theory of course would not be valid! This
is a relatively common practice, since the effect of changing As is so trivial that it makes not
much sense to plot C! separately for different values of As .

9.11.2 Sound horizon angle

The positions of the acoustic peaks of the C! spectrum provide us with a measurement of the
sound horizon angle
rs (tdec )
ϑs ≡ c
dA (tdec )
We can use this in the determination of the values of the cosmological parameters, once we
have calculated how this angle depends on those parameters. It is the ratio of two quantities,
the sound horizon at photon decoupling, rs (tdec ), and the angular diameter distance to the last
scattering, dcA (tdec ).

Angular diameter distance to last scattering

The angular diameter distance dcA (tdec ) to the last scattering surface we have already calcu-
lated and it is given by Eq. (39) as
" # 1 $
c −1 1 ! da
dA (tdec ) = H0 ! fk |Ω0 − 1| ! , (115)
|Ω0 − 1| 1+z
1 Ω0 (a − a2 ) − ΩΛ (a − a4 ) + a2
dec

from which we see that it depends on the three cosmological parameters H0 , Ω0 and ΩΛ . Here
Ω0 = Ωm +ΩΛ , so we could also say that it depends on H0 , Ωm , and ΩΛ , but it is easier to discuss
the effects of these different parameters if we keep Ω0 as an independent parameter, instead of
Ωm , since the “geometry effect” of the curvature of space, which determines the relation between
the comoving angular diameter distance dcA and the comoving distance dc , is determined by Ω0 .

1. The comoving angular diameter distance is inversely proportional to H0 (directly propor-

tional to the Hubble distance H0−1 ).

2. Increasing Ω0 decreases dcA (tdec ) in relation to dc (tdec ) because of the geometry effect.

3. With a fixed ΩΛ , increasing Ω0 decreases dc (tdec ), since it means increasing Ωm , which has a
decelerating effect on the expansion. With a fixed present expansion rate H0 , deceleration
means that expansion was faster earlier ⇒ universe is younger ⇒ there is less
time for photons to travel as the universe cools from Tdec to T0 ⇒ last scattering
surface is closer to us.

4. Increasing ΩΛ (with a fixed Ω0 ) increases dc (tdec ), since it means a larger part of the energy
density is in dark energy, which has an accelerating effect on the expansion. With fixed
H0 , this means that expansion was slower in the past ⇒ universe is older ⇒ more
time for photons ⇒ last scattering surface is further out. ∴ ΩΛ increases dcA (tdec ).
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 140

Here 2 and 3 work in the same direction: increasing Ω0 decreases dcA (tdec ), but the geometry effect
(2) is stronger. See Fig. 17 for the case ΩΛ = 0, where the dashed line (the comoving distance)
shows effect (3) and the solid line (the comoving angular diameter distance) the combined effect
(2) and (3).
However, now we have to take into account that, in our chosen parametrization, H0 is not
an independent parameter, but !
−1 Ω0 − ΩΛ
H0 ∝ ,
ωm
so that via H0−1 , Ω0 increases and ΩΛ decreases dcA (tdec ), which are the opposite effects to those
discussed above. For ΩΛ this opposite effect wins. See Fig. 26.

Sound horizon
To calculate the sound horizon,
" tdec " adec
cs (t) da
rs (tdec ) = dt = cs (a) , (116)
0 a(t) 0 a · (da/dt)
we need the speed of sound, from Eq. (84),
1 1 1 1
c2s (x) = ρ̄b = , (117)
3
3 1 + 4 ρ̄ 3 1 + 34 ωωb a
γ γ

where the upper limit of the integral is adec = 1/(1 + zdec ).

The other element in the integrand of Eq. (116) is the expansion law a(t) before decoupling.
From Chapter 3 we have that
da #
a = H0 Ωr + Ωm a + (1 − Ω0 )a2 + ΩΛ a4 . (118)
dt
In the integral (115) we dropped the Ωr , since it is important only at early times, and the
integral from adec to 1 is dominated by late times. Integral (116), on the other hand, includes
only early times, and now we can instead drop the ΩΛ and 1 − Ω0 terms (i.e., we can ignore the
effect of curvature and dark energy in the early universe, before photon decoupling), so that
# √
da √ ωm a + ωr
a ≈ H0 Ωm a + Ωr = H100 ωm a + ωr = , (119)
dt 2998 Mpc
where we have written
km/s h
H0 ≡ h · 100 ≡ h · H100 = . (120)
Mpc 2997.92 Mpc
Thus the sound horizon is given by
" a
cs (x)dx
rs (a) = 2998 Mpc √
0 ωm x + ωr
" a
1 dx (121)
= 2998 Mpc · √ !$ %$ %.
3ωr 0 ωm 3 ωb
1+ ωr x 1+ 4 ωγ x

Here

ωγ = 2.473 × 10−5 and (122)

& ' (4/3 )
7 4
ωr = 1 + Neff ωγ = 1.692 ωγ = 4.184 × 10−5 (123)
8 11
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 141

are accurately known from the CMB temperature T0 = 2.7255 K (and therefore we do not
consider them as cosmological parameters in the sense of something to be determined from the
C! spectrum).
Thus the sound horizon depends on the two cosmological parameters ωm and ωb ,

rs (tdec ) = rs (ωm , ωb )

From Eq. (121) we see that increasing either ωm or ωb makes the sound horizon at decoupling,
rs (adec ), shorter:
• ωb slows the sound down

• ωm speeds up the expansion at a given temperature, so the universe cools to Tdec in less
time.
The integral (121) can be done and it gives
√ √
2998 Mpc 2 1 + R∗ + R∗ + r∗ R∗
rs (tdec ) = √ √ ln √ , (124)
1 + zdec 3ωm R∗ 1 + r∗ R∗
where
! "
ρ̄r (tdec ) ωr 1 1 + zdec
r∗ ≡ = (1 + zdec ) = 0.0456 (125)
ρ̄m (tdec ) ωm ωm 1091
! "
3ρ̄b (tdec ) 3ωb 1 1091
R∗ ≡ = = 27.8 ωb . (126)
4ρ̄γ (tdec ) 4ωγ 1 + zdec 1 + zdec

For our reference values ωm = 0.147, ωb = 0.022, and 1 + zdec = 109117 we get r∗ = 0.310 and
R∗ = 0.614 and rs (tdec ) = 144 Mpc for the sound horizon at decoupling.

Summary
The angular diameter distance dcA (tdec ) is the most naturally discussed in terms of H0 , Ω0 ,
and ΩΛ , but since these are not the most convenient choice of independent parameters for other
purposes, we shall trade H0 for ωm according to Eq. (113). Thus we have that the sound horizon
angle depends on 4 parameters,
rs (ωm , ωb )
ϑs ≡ = ϑs (Ω0 , ΩΛ , ωm , ωb ) (127)
dcA (Ω0 , ΩΛ , ωm )

9.11.3 Acoustic peak heights

There are a number of effects affecting the heights of the acoustic peaks:
1. The early ISW effect. The early ISW effect raises the first peak. It is caused by the
evolution of Φ because of the effect of the radiation contribution on the expansion law
after tdec . This depends on the radiation-matter ratio at that time; decreasing ωm makes
the early ISW effect stronger.

2. Shift of oscillation equilibrium by baryons. (Baryon drag.) This makes the odd
peaks (which correspond to compression of the baryon-photon fluid in the potential wells,
decompression on potential hills) higher, and the even peaks (decompression at potential
wells, compression on top of potential hills) lower.
17
Photon decoupling temperature, and thus 1 + zdec , depends somewhat on ωb , but since this dependence is not
easy to calculate (recombination and photon decoupling were discussed in Chapter 4), we have mostly ignored
this dependence and used the fixed value 1 + zdec = 1091.
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 142

ωb = 0.01 ωb = 0.03

0.6 0.6
l(l+1)Cl/2π

0.4 0.4
ωm = 0.10
ωm = 0.20
0.2 0.2
ωm = 0.30
ωm = 0.40
0 0
0 500 1000 1500 2000 0 500 1000 1500 2000
l l

Figure 24: The effect of ωm . The angular power spectrum C! is here calculated without the effect of
diffusion damping, so that the other effects on peak heights could be seen more clearly. Notice how
reducing ωm raises all peaks, but the effect on the first few peaks is stronger in relative terms, as the
radiation driving effect is extended towards larger scales (smaller "). The first peak is raised mainly
because the ISW effect becomes stronger. Figure and calculation by R. Keskitalo.

0.6

0.5

0.4
2l(l+1)Cl/2π

0.3

0.2
ωb = 0.01
ωb = 0.02
ωb = 0.03
0.1
ωb = 0.04

0
0 500 1000 1500 2000
l

Figure 25: The effect of ωb . The angular power spectrum C! is here calculated without the effect of
diffusion damping, so that the other effects on peak heights could be seen more clearly. Notice how
increasing ωb raises odd peaks relative to the even peaks. Because of baryon damping there is a general
trend downwards with increasing ωb . This figure is for ωm = 0.20. Figure and calculation by R. Keskitalo.
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 143

3. Baryon damping. The time evolution of R ≡ 3ρ̄b /4ρ̄γ causes the amplitude of the
acoustic oscillations to be damped in time roughly as (1 + R)−1/4 . This reduces the
amplitudes of all peaks.

4. Radiation driving.18 This is an effect related to horizon scale physics that we have
not tried to properly calculate. For scales k which enter during the radiation-dominated
epoch, or near matter-radiation equality, the potential Φ decays around the time when
the scale enters. The potential keeps changing as long as the radiation contribution is
important, but the largest change in Φ is around horizon entry. Because the sound horizon
and Hubble length are comparable, horizon entry and the corresponding potential decay
always happen during the first oscillation period. This means that the baryon-photon
fluid is falling into a deep potential well, and therefore is compressed by gravity by a large
factor, before the resulting overpressure is able to push it out. Meanwhile the potential
has decayed, so it is less able to resist the decompression phase, and the overpressure is
able to kick the fluid further out of the well. This increases the amplitude of the acoustic
oscillations. The effect is stronger for the smaller scales which enter when the universe
is more radiation-dominated, and therefore raises the peaks with a larger peak number
m more. Reducing ωm makes the universe more radiation dominated, making this effect
stronger and extending it towards the peaks with lower peak number m.

5. Diffusion damping. Diffusion damping lowers the heights of the peaks. It acts in the
opposite direction than the radiation driving effect, lowering the peaks with a larger peak
number m more. Because the diffusion damping effect is exponential in #, it wins for large
#.

Effects 1 and 4 depend on ωm , effects 2, 3, and 5 on ωb . See Figs. 24 and 25 for the effects of
ωm and ωb on peak heights.

18
This is also called gravitational driving, which is perhaps more appropriate, since the effect is due to the
change in the gravitational potential.
 9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 144
0.4 0.4
0 = 1.1 = 0.80
0.35 0 = 1.0 0.35 = 0.70
0 = 0.9 = 0.60
0.3 0.3

0.25 0.25
( +1)C / 2

( +1)C / 2
0.2 0.2

0.15 0.15

0.1 0.1

0.05 0.05

0.0 0.0
0 200 400 600 800 1000 1200 1400 0 200 400 600 800 1000 1200 1400

0.4 0.4
0 = 1.1 = 0.80
0.35 0 = 1.0 0.35 = 0.70
0 = 0.9 = 0.60
0.3 0.3

0.25 0.25
( +1)C / 2

( +1)C / 2

0.2 0.2

0.15 0.15

0.1 0.1

0.05 0.05

0.0 1 2 3 0.0 1 2 3
2 5 10 2 5 10 2 5 10 2 2 5 10 2 5 10 2 5 10 2

Figure 26: The effect of changing Ω0 or ΩΛ from their reference values Ω0 = 1 and ΩΛ = 0.7. The
top panels shows the C! spectrum with a linear ! scale so that details at larger ! where cosmic variance
effects are smaller can be better seen. The bottom plot has a logarithmic ! scale so that the integrated
Sachs-Wolfe effect at small ! can be better seen. The logarithmic scale also makes clear that the effect of
the change in sound horizon angle is to stretch the spectrum by a constant factor in ! space.

9.11.4 Effect of Ω0 and ΩΛ

These two parameters have only two effects:

1. they affect the sound horizon angle and thus the positions of the acoustic peaks

2. they affect the late ISW effect

See Fig. 26. Since the late ISW effect is in the region of the C! spectrum where the cosmic
variance is large, it is difficult to detect. Thus we can in practice only use ϑs to determine Ω0
and ΩΛ . Since ωb and ωm can be determined quite accurately from C! acoustic peak heights,
peak separation, i.e., ϑs , can then indeed be used for the determination of Ω0 and ΩΛ . Since
one number cannot be used to determine two, the parameters Ω0 and ΩΛ are degenerate. CMB
observations alone cannot be used to determine them both. Other cosmological observations (like
the power spectrum Pδ (k) from large scale structure, or the SNIa redshift-distance relationship)
are needed to break this degeneracy.
A fixed ϑs together with fixed ωb and ωm determine a line on the (Ω0 , ΩΛ ) -plane. See
Fig. 27. Derived parameters, e.g., h, vary along that line. As you can see from Fig. 26, changing
Ω0 (around the reference model) affects ϑs much more strongly than changing ΩΛ . This means
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 145

Distance between successive acoustic peaks (Δ l)

1.5
no big ω = 0.022, ω = 0.147, h is derived parameter
bang b m 50

50
100
100
1
30 200
0
35 250
0
Λ
Ω

600

accel −−− decel 200

0.5 open −−− closed

250
40 30
0
50

0
0

0
0.2 0.4 0.6 0.8 1 1.2 1.4
Ωm

Figure 27: The lines of constant sound horizon angle ϑs on the (Ωm ,ΩΛ ) plane for fixed ωb and ωm .
The numbers on the lines refer to the corresponding acoustic scale #A ≡ π/ϑs (∼ peak separation) in
multipole space. Figure by J. Väliviita. See his PhD thesis[10], p.70, for an improved version including
the HST constraint on h.

that the orientation of the line is such that ΩΛ varies more rapidly along that line than Ω0 .
Therefore using additional constraints from other cosmological observations, e.g., the Hubble
Space Telescope determination of h based on the distance ladder, which select a short section
from that line, gives us a fairly good determination of Ω0 , leaving the allowed range for ΩΛ still
quite large.
Therefore it is often said that CMB measurements have determined that Ω0 ∼ 1. But as
explained above, this determination necessary requires the use of some auxiliary cosmological
data besides the CMB.
 9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 146

0.4 0.4
A = 1.1 n = 1.1
0.35 A=1 0.35 n=1
A = 0.9 n = 0.9
0.3 0.3

0.25 0.25
( +1)C / 2

( +1)C / 2
0.2 0.2

0.15 0.15

0.1 0.1

0.05 0.05

0.0 0.0
0 200 400 600 800 1000 1200 1400 0 200 400 600 800 1000 1200 1400

0.4 0.4
A = 1.1 n = 1.1
0.35 A=1 0.35 n=1
A = 0.9 n = 0.9
0.3 0.3

0.25 0.25
( +1)C / 2

( +1)C / 2

0.2 0.2

0.15 0.15

0.1 0.1

0.05 0.05

0.0 1 2 3 0.0 1 2 3
2 5 10 2 5 10 2 5 10 2 2 5 10 2 5 10 2 5 10 2

Figure 28: The effect of changing the primordial amplitude and spectral index from their reference
values As = 1 and ns = 1.

9.11.5 Effect of the primordial spectrum

The effect of the primordial spectrum is simple: raising the amplitude As raises the C! also, and
tilting the primordial spectrum tilts the C! also. See Fig. 28.
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 147
0.4
= 0.20
0.35 = 0.10
=0
0.3

0.25

( +1)C / 2
0.2

0.15

0.1

0.05

0.0
0 200 400 600 800 1000 1200 1400

0.4
= 0.20
0.35 = 0.10
=0
0.3

0.25
( +1)C / 2

0.2

0.15

0.1

0.05

0.0 1 2 3
2 5 10 2 5 10 2 5 10 2

Figure 29: The effect of changing the optical depth from its reference value τ = 0.1.

9.11.6 Optical depth due to reionization

The optical depth τ due to reionization was discussed in Sec. 9.10. See Fig. 29 for its effect on
C! .
 9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 148
0.4 0.4
b = 0.030 m = 0.200
0.35 b = 0.022 0.35 m = 0.147
b = 0.015 m = 0.100
0.3 0.3

0.25 0.25
( +1)C / 2

( +1)C / 2
0.2 0.2

0.15 0.15

0.1 0.1

0.05 0.05

0.0 0.0
0 200 400 600 800 1000 1200 1400 0 200 400 600 800 1000 1200 1400

0.4 0.4
b = 0.030 m = 0.200
0.35 b = 0.022 0.35 m = 0.147
b = 0.015 m = 0.100
0.3 0.3

0.25 0.25
( +1)C / 2

( +1)C / 2

0.2 0.2

0.15 0.15

0.1 0.1

0.05 0.05

0.0 1 2 3 0.0 1 2 3
2 5 10 2 5 10 2 5 10 2 2 5 10 2 5 10 2 5 10 2

Figure 30: The effect of changing the physical baryon density and matter density parameters from their
reference values ωb = 0.022 and ωm = 0.147.

9.11.7 Effect of ωb and ωm

These parameters affect both the positions of the acoustic peaks (through ϑs ) and the heights
of the different peaks. The latter effect is the more important, since both parameters have their
own signature on the peak heights, allowing an accurate determination of these parameters,
whereas the effect on ϑs is degenerate with Ω0 and ΩΛ .
Especially ωb has a characteristic effect on peak heights: Increasing ωb raises the odd peaks
and reduces the even peaks, because it shifts the balance of the acoustic oscillations (the −RΦ
effect). This shows the most clearly at the first and second peaks. Raising ωb also shortens the
−1
damping scale kD due to photon diffusion, moving the corresponding damping scale #D of the
C! spectrum towards larger #. This has the effect of raising C! at large #. See Fig. 30.
There is also an overall “baryon damping effect” on the acoustic oscillations which we have
not calculated. It is due to the time dependence of R ≡ 3ρ̄b /4ρ̄m , which reduces the amplitude
of the oscillation by about (1 + R)−1/4 . This explains why the third peak in Fig. 30 is no higher
for ωb = 0.030 than it is for ωb = 0.022.
Increasing ωm makes the universe more matter dominated at tdec and therefore it reduces
the early ISW effect, making the first peak lower. This also affects the shape of the first peak.
The “radiation driving” effect is most clear at the second to fourth peaks. Reducing ωm
makes these peaks higher by making the universe more radiation-dominated at the time the
corresponding scales enter, strengthening this radiation driving. The fifth and further peaks
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 149

Parameters for the ΛCDM model

Planck 2018 best fit
ωb 0.02237 ± 0.00015 0.022383
ωm 0.1424 ± 0.0012 0.14249
ΩΛ 0.685 ± 0.007 0.6841
τ 0.054 ± 0.007 0.0543
As 4.58 ± 0.04 × 10−5 4.5832 × 10−5
ns 0.965 ± 0.004 0.96605
H0 67.36 ± 0.54 km/s/Mpc 67.32 km/s/Mpc
ωc 0.1200 ± 0.0012 0.12011
Ωm 0.315 ± 0.007 0.3158
zeq 3402 ± 26
−1
keq 96.3 ± 0.8 Mpc
zdec 1089.92 ± 0.25
−1
kD 7.10 ± 0.02 Mpc
zreion 7.7 ± 0.7 7.68
ϑs 0.5965◦ ± 0.0002◦ 0.59651◦
t0 13.797 ± 0.023 × 109 a 13.7971 × 109 a

Table 2: These parameter values are based on the CMB temperature power spectrum C! , CMB po-
larization, and gravitational lensing of the CMB, as observed by the Planck satellite [5]. The first six
parameters, above the line, are independent parameters, and the parameters below the line are quantities
that can be derived from them in the ΛCDM model. The error estimates are 68% confidence limits. The
best-fit column gives a representative model that is an excellent fit to the data; nearby models in the
6-parameter space may! be practically equally good fits. Note that here Ωm includes the contribution
from neutrinos with mν = 0.06 meV (Ων = 0.0014) whereas ωm does not.

correspond to scales that have anyway essentially the full effect, and for the first peak this effect
is anyway weak. (We see instead the ISW effect in the first peak.) See Fig. 30.

9.12 Current best estimates for the cosmological parameters

9.12.1 Planck values for ΛCDM parameters
The most important data set for determining cosmological parameters is the Planck data [5]
on the CMB anisotropy. We give the parameter values determined by Planck for the ΛCDM
model in Table 2. Note that all independent parameters of the model are fit simultaneously to
the same data. The determination is based on the assumption that the model, here ΛCDM, is
correct. One can judge this assumption based on how well the model fits the data. In the case
of Planck and ΛCDM the fit is good; adding more parameters to the model does not improve
the fit significantly.
This model agrees reasonable well with most of the other available cosmological data, with
the exception of the distance-ladder determination of the Hubble constant, based on Cepheids
and Type Ia supernovae, which gives H0 = 73.5 ± 1.6 km/s/Mpc [11, 12]. This is called the local
measurement of H0 , since these measurements are from nearby parts of the Universe, in contrast
to the global determination from the CMB, where the CMB has traversed the entire observable
Universe. This discrepancy has been evident in the data for some time, but it has gradually
become more serious as the error bars on H0 from both CMB and local measurements have
become tighter without the central values changing much. One may suspect systematic errors
in the distance ladder data or that the ΛCDM model is a too simple model for the universe.
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 150

Constraints for extended models

ΛCDM Planck 2018 Planck+ext
Ω0 1.0 1.011 ± 0.013 0.9993 ± 0.0037
r 0 < 0.101 < 0.065
dns /d ln k 0 −0.005 ± 0.013 −0.004 ± 0.013
w
! -1 −1.6 ± 0.5 −1.04 ± 0.10
mν 0.06 eV < 0.241 eV < 0.120 eV
Neff 3.046 2.89 ± 0.38 2.99 ± 0.34

Table 3: Each row is a different model and we show limits only to the “additional” parameter. As
is customary with limits, the ranges are given as 95% confidence limits. Neff , the “effective number of
neutrino species”, refers to relativistic energy density (in addition to photons) near the time of photon
decoupling.

9.12.2 Extended models and external data

In the ΛCDM model the universe is flat, Ω0 = 1. We can also fit extended models, with additional
independent parameters. Such 7-parameter models, with one extra parameter in addition to the
ΛCDM parameters, are fit to Planck data in Table 3. Since the ΛCDM model is a good fit,
the estimates for these extra parameters are consistent with their values in the ΛCDM model.
Instead of the central value, we therefore concentrate on the estimated probable range, i.e., limits
to the deviation from the ΛCDM model. Note that in these extended models the ranges for the
6 ΛCDM parameters will be different from Table 2; they will be wider and the central values will
be slightly different. One could of course consider models with more independent parameters,
e.g., the 12-parameter model, where all the 6 parameters of Table 3 were added to ΛCDM. In
such a model the allowed ranges for all these parameters would be wider than in Tables 2 and 3.
The argument against such a model is Occam’s razor : if there are many models that fit the data,
one should prefer the simplest one; a corollary to this is that the models one should consider
next are those that are almost as simple. Of course, there is no guarantee against all these
parameters having a significant effect on the CMB. These one-parameter extensions to ΛCDM
do not relieve the tension with the local determination of H0 much, but by adding sufficiently
many additional parameters one can get rid of the tension.
Dark radiation. The parameter Neff corresponds to making ωr a free parameter. From the
discussion in Sec. 9.11.2 we see that we are constraining relativistic energy density at or before
tdec . Additional relativistic particle species, in addition to photons and neutrinos, would raise
Neff above the Particle Physics Standard Model value 3.046. The 95% confidence upper limit for
the contribution from such extra species is ∆Neff ≡ Neff − 3.046 < 0.3. This rules out any new
(currently unknown) particles that would decouple after QCD transition and stay relativistic
until photon decoupling, see Fig. 31.
External data. Because of degeneracies of cosmological parameters in the CMB data, most
importantly the geometrical degeneracy between parameters, like Ω0 , ΩΛ , and the dark energy
equation-of-state parameter w, whose main effect on CMB is via their effect on the angular
diameter distance to the last scattering sphere, some parameters of these extended models are
only weakly constrained by Planck data. To break these degeneracies, additional cosmological
data (BAO and BICEP2/Keck, see below) has been used in the fourth column of Table 3 (ext
= external to Planck). The impressive accuracy in this column is, however, mainly due to the
accuracy of Planck. The parameter values allowed by Planck form a narrow but long region
in the 7-parameter space, and the external data allows a region that is wider, but oriented
differently; the intersection of these regions is then a shorter segment of the region allowed by
Planck alone, see Fig. 32.
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 151

Figure 31: Upper panel: The effective number g∗ (T ) of degrees of freedom in the Standard Model of
particle physics. Note that the drop in g∗ (T ) due to the QCD transition is not sharp, since this is not a
phase transition (taking place at a fixed critical temperature Tc ), but is a cross-over transition (happening
gradually over a temperature range). Bottom panel: The colored curves show contributions to ∆Neff from
different types of light (relativistic at photon decoupling) thermal relics as a function of their decoupling
temperature. The darker yellow region is ruled out by the Planck upper limit ∆Neff < 0.3. (The lighter
yellow region corresponds to the 68% confidence upper limit ∆Neff < 0.13.) From [5].
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 152

Figure 32: Left: Constraints on ΩΛ and Ωm (or Ω0 = ΩΛ + Ωm ) in the ΛCDM+Ω0 model from Planck
2015 and BAO data. The colored dots represent parameter values that fit Planck temperature C! and
large scale polarization data, the color giving the value of H0 required for the fit. The black contours
(inner 68% and outer 95% confidence limits) give the models that remain allowed when Planck small-scale
polarization data is also used; and blue contours when Planck CMB lensing data is used instead. The red
contours show the effect of adding BAO data to break the Ω0 -ΩΛ degeneracy. From the colors one can
see that also independent H0 data could be used to break the degeneracy. The dashed line corresponds
to a flat universe. From [3]. Right: The same from Planck 2018 data, except shown for (Ωk ,Ωm ) instead
of (Ωm ,ΩΛ ). From [5].

Figure 33: The ! upper limit √ Ω0 < 1.003 means that if we live in a closed universe, its curvature radius
Rcurv = H −1 / |Ωk | > H −1 / 0.003 = 18.3H −1 = 5.9dc (z = 1090) is more than 5 times larger than the
distance we can see (to the last scattering sphere, corresponding to the red circle, which is actually drawn
too large here, since this figure was drawn when the limit was weaker).
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 153

Figure 34: Constraints on the dark energy equation-of-state parameters w0 and wa (see text) in the
8-parameter (w0 + wa )CDM model from Planck, BAO, and SNIa data. From [3].

Figure 35: Constraints on the parameters ns and r, which constrain inflation models, in the 7-parameter
ΛCDM+r model from Planck data. Gray contours are based on Planck data only; red and blue contours
include external data. Predictions from a selection of inflation models are marked on the plot. From [7].
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 154

Large scale structure surveys, i.e., the measurement of the 3-dimensional matter power spec-
trum Pδ (k) from the distribution of galaxies, mainly measure the combination Ωm h, since this
determines where Pδ (k) turns down. Actually it turns down at keq which is proportional to
ωm ≡ Ωm h2 , but since in these surveys the distances to galaxies are deduced from their redshifts
(these surveys are also called galaxy redshift surveys), which give the distances only up to the
Hubble constant H0 , these surveys determine h−1 keq instead of keq . This cancels one power of h.
Having Ωm h2 from CMB and Ωm h from the galaxy surveys, gives us both h and Ωm = Ω0 − ΩΛ ,
which breaks the Ω0 -ΩΛ degeneracy.
BAO. Measurements of Pδ (k) are now so accurate that the small residual effect from the
acoustic oscillations before photon decoupling can be seen as a weak wavy pattern [13]. This is
the same structure which we see in the C" but now much fainter, since now the baryons have
fallen into the CDM potential wells, and the CDM was only mildly affected by these oscillations
in the baryon-photon fluid. In this context these are called baryon acoustic oscillations (BAO).
The half-wavelength of this pattern, however, corresponds to the same sound horizon distance
rs (tdec ) in both cases.19 But now the angular scale on the sky is related to it by the angular
diameter distance dcA (z) to the much smaller redshifts z of the galaxy survey. This dcA (z) has
then a different relation to Ω0 , ΩΛ , and ωm . Comparing CMB data to galaxy surveys gives
us the ratio dcA (z)/dcA (tdec ), which gives us independent information on these parameters. The
large scale structure surveys used for the BAO measurements to supplement Planck 2018 data
were the 6dF Galaxy Survey (6dFGS) [14] and the Sloan Digital Sky Survey (SDSS) [15, 16].
Curvature. Because of the geometrical degeneracy, the CMB angular power spectra alone
are not good for constraining Ωk . The peak structure gives a precise measurement of the
angular diameter distance to last scattering dcA (tdec ). In the ΛCDM+Ωk model his translates
into a curve on the (Ωm , ΩΛ ) plane (see Fig. 27). The late ISW effect due to ΩΛ would break
this degeneracy, but since this affects only the lowest multipoles it is lost in the cosmic variance.
More significant is a higher-order (beyond linear perturbation theory) effect on the C" ; that of
gravitational lensing of the CMB due to large-scale structure. This smooths the acoustic peaks
and the effect is proportional to Ωm . Although the effect is small, it occurs at high " where
cosmic variance is small and provides some degeneracy breaking power between Ωm and ΩΛ ,
or equivalently, between Ωm and Ωk . The resulting constraint, from Planck C" only, on Ωk is
Ωk = −0.044+0.018
−0.015 (68% CL), which favors a closed universe at well over 2 σ. The best-fit such
models have Ωm > 0.45 and H0 < 55 km/s/Mpc, which are ruled out by other cosmological
data. The problematic feature in the data is that the acoustic peaks are slightly lower than
predicted by the ΛCDM model fit to the data, as if there was too much lensing (requiring higher
Ωm , which then leads to lower ΩΛ and negative Ωk as we move along the degeneracy line).
From the Planck data one can also measure the gravitational lensing of the CMB more
directly from the effect it has on higher-order correlations (higher than the 2-point correlation
measured by C" ) of the CMB. This measurement of CMB lensing agrees with the ΛCDM predic-
tion and thus with a flat universe, giving the constraint Ωk = −0.0106±0.00065 (when combined
with the Planck C" ). When one adds also BAO data to break the geometric degeneracy, one
arrives at the final result given in the Planck+ext column of Table 3,
Ωk = 0.0007 ± 0.0019 (68%CL) . (128)
(Table 3 gives 95% confidence limits, which are twice as wide.) The 95% upper limit Ω0 < 1.003,
or Ωk > −0.003 gives a minimum size to the Universe, see Fig. 33.
Supernovae. Another way to break the geometric degeneracy, is to use the redshift-distance
relationship from Supernova Type Ia (SNIa) surveys [17], or simply the distance-ladder determi-
nation of H0 , where Cepheids and Supernovae are the last two steps of the ladder. These were
19
To be accurate, the tdec value to represent the effect in Pδ (k), is not exactly the same as for C" , since photon
decoupling was not instantaneous, and in one we are looking at the effect on matter and in the other on photons.
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 155

not used in the Planck 2018 analysis of 6- and 7-parameter models, because of the discrepancy
with the local H0 determination20 , and since the SNIa data adds little statistical power to the
CMB+BAO combination, but the SNIa data was used for the following 8-parameter model.
Dark energy. To constrain properties of dark energy, the 7-parameter wCDM model is
probably too simplistic, since it assumes that the equation-of-state parameter w stays constant
during the epoch when dark energy has a significant effect on the expansion. To stay at a
phenomenological level, i.e., not assuming a particular dark-energy model, but just attempting
to constrain its equation of state, the next step is a two-parameter equation of state w(a) =
w0 + wa (1 − a), i.e., a first-order Taylor expansion with w0 the current value of w, and wa related
to its first derivative with respect to the scale factor, leading to an 8-parameter model. From
Fig. 34 you can see that the best fits are near the ΛCDM values w0 = −1, wa = 0, but that the
equation of state is poorly constrained.
Neutrino masses. Neutrino masses, i.e., the amount of hot dark matter, have a larger
effect on large-scale structure than CMB; the CMB data is mainly needed to determine the
other parameters after which the large-scale structure power spectrum Pδ (k) can be used to
determine the sum of the neutrino masses. The value 0.06 eV used for the ΛCDM model is the
minimum allowed by neutrino oscillation data.
Tensor perturbations. The polarization pattern of the CMB on the sky can be divided
into what are called E and B modes. This is analogous to the division of a vector field into irro-
tational (curl-free) and rotational (divergence-free) parts. To first (i.e. linear) order in perturba-
tion theory, only tensor perturbations produce B-mode polarization in the CMB. Only E-mode
polarization has so far been detected in the CMB. Upper limits to CMB B-mode polarization
provide upper limits to the tensor-scalar ratio r. Planck was not optimized for polarization mea-
surements, so its B-mode measurement is noisy and suffers from instrument systematics. Thus
the Planck upper limit to r from B modes is weak, r < 0.41, and the Planck constraint r < 0.101
comes from the effect of tensor perturbations on the CMB temperature C" . The ground-based
BICEP2/Keck Array [18] at the South Pole can measure polarization more accurately, but it
has limited sky coverage and needs to be combined with Planck data to separate the CMB from
the foreground. This combination leads to the B-mode upper limit r < 0.065.
Inflation models. Since inflation produces tensor perturbations, and many inflation models
predict that they should be strong enough to have an observable effect on the CMB, the simplest
way to constrain inflation is to fit the ΛCDM+r model to CMB data. From Fig. 35 you can see
that the V (ϕ) = 12 m2 ϕ2 inflation model, is ruled out by Planck data alone at a 95% confidence
level (assuming that ΛCDM+r is the correct model for the universe).

20
One should not combine discrepant data in parameter fitting. This would lead to artificially tight parameter
values with poor fits to both data sets.
9 COSMIC MICROWAVE BACKGROUND ANISOTROPY 156

Figure 36: CMB temperature anisotropy in ecliptic coordinates.

9.13 Issues with CMB data

While the agreement of the CMB observations with the predictions of the ΛCDM model is
impressive (see Fig. 11), it is not perfect. Also, while combining other cosmological data with
CMB data adds more support for the ΛCDM model, there are some discrepancies. There are at
least three issues:
Large scale anomalies. Comparing the northern (Fig. 6) and southern (Fig. 7) galac-
tic hemispheres, one may notice that the southern hemisphere has stronger large-scale CMB
anisotropies. The difference is more clear between the ecliptic hemispheres, see Fig. 36. This
is not what we would expect from statistical isotropy. Also the quadrupole and octupole have
planar shape (m = ! dominates, see the Y22 and Y33 in Fig. 10) and are aligned with each other
(and the quadrupole is rather weak). These large-scale anomalies are seen in both WMAP and
Planck data, so they are real. They may be just a statistical fluke, or a sign that the Universe
deviates from standard ΛCDM at the very largest observable scales. Because of cosmic variance
it is difficult to tell.
“Lensing smoothing”. While the scatter of data points around the theoretical prediction
(see Fig. 37) is mainly as expected (the error bars are 68% CL, so we expect 32% of the data
points deviate from the prediction by more than the error bar), there are some features. Data for
the low multipoles ! < 30 are mostly below prediction. This is due to the lack of large-scale power
in the northern ecliptic hemisphere, and thus related to the large-scale anomalies. In the range
! = 1100–2000 the data residuals seem to oscillate in opposite phase to the acoustic peak pattern,
i.e., the acoustic peaks in the data are slightly smoothed compared to theoretical prediction.
Gravitational lensing of the CMB by the large-scale structure causes such smoothing, and it
is as if this smoothing effect is some 10–20% larger than predicted by ΛCDM. These features
can be fit better with the ΛCDM+Ωk model with negative Ωk (closed universe), but the Planck
direct measurement of CMB lensing (from higher-order correlations) and other cosmological
data disagrees with such a model.
Local vs global Hubble constant. The Planck ΛCDM value for the Hubble constant,
H0 = 67.36 ± 0.54 km/s/Mpc disagrees with distance-ladder measurements, which give, e.g.,
H0 = 74.0 ± 1.4 km/s/Mpc [19]. One may think of many possible causes for this discrepancy. 1)
REFERENCES 157

Figure 37: The temperature C! from Planck 2018 data. Unlike in Fig. 11, here the cosmic variance
is included in the error bars, so the data can be more directly compared with theory. The blue curve
is the ΛCDM prediction, and the bottom panel shows the difference (data residuals) between data and
prediction. Note the different scale (on the right) for residuals at ! > 30, where also the horizontal axis
changes from logarithmic to linear. From [5].

There may be underestimated systematic errors in the distance-ladder measurements. 2) Note

that the Planck value is not a “measurement” of H0 . It is a result of a six-parameter fit to the
data, assuming the ΛCDM model, where H0 is one of the six parameters. So the discrepancy
could point to ΛCDM not being the correct model. One could alleviate the discrepancy by
adding extra parameters to the model. (However, e.g., in the ΛCDM+Ωk model the discrepancy
becomes worse.) 3) The distance-ladder measurements are from a “local” part of the universe,
z < 1, so it represents a local measurement of H0 ; whereas the result from the CMB is related
to the distance from here to the last scattering surface, so it corresponds to a value of H0 which
is representative of the entire observable Universe. The explanation of the discrepancy could
thus be an unexpectedly large inhomogeneity: we live in a large underdense region, which thus
expands faster than the Universe on average.
What should one make of such discrepancies? Similar issues are common in the progress of
science. More often than not, they go away with improved data; but sometimes they point to
something new, which improved data later confirms. For the large scale anomalies, “improved
data” will be difficult to get, since here we are limited by cosmic variance. More accurate data on
CMB polarization would help: are there similar large-scale anomalies in the polarization or not?
The Planck data on polarization was inconclusive on this question [6], since the Planck design
was not optimized for polarization measurements, and therefore there are residual systematic
effects in the polarization data at large scales, which limits the accuracy.

References
[1] Planck Collaboration, Planck 2013 results. I. Overview of products and scientific results,
arXiv:1303.5062, Astronomy & Astrophysics 571, A1 (2014)
REFERENCES 158

[2] Planck Collaboration, Planck 2015 results. I. Overview of products and results,
arXiv:1502.01582, Astronomy & Astrophysics 594, A1 (2016)

[3] Planck Collaboration, Planck 2015 results. XIII. Cosmological parameters,

arXiv:1502.01589, Astronomy & Astrophysics 594, A13 (2016)

[4] Planck Collaboration, Planck 2018 results. I. Overview, and the cosmological legacy of
Planck, arXiv:1807.06205v1 (2018)

[5] Planck Collaboration, Planck 2018 results. VI. Cosmological parameters,

arXiv:1807.06209v1 (2018)

[6] Planck Collaboration, Planck 2018 results. VII. Isotropy and Statistics of the CMB,
arXiv:1906.02552v1 (2019)

[7] Planck Collaboration, Planck 2018 results. X. Constraints on inflation, arXiv:1807.06211v1

(2018)

[8] A.R. Liddle and D.H. Lyth: Cosmological Inflation and Large-Scale Structure (Cambridge
University Press 2000)

[9] S. Dodelson: Modern Cosmology (Academic Press 2003)

[10] J. Väliviita, PhD thesis, University of Helsinki 2005

[11] A.G. Riess et al., New Parallaxes of Galactic Cepheids from Spatially Scanning the Hubble
Space Telescope: Implications for the Hubble Constant, arXiv:1801.01120

[12] A.G. Riess et al., Milky Way Cepheid Standards for Measuring Cosmic Distances and Ap-
plication to Gaia DR2: Implications for the Hubble Constant, arXiv:1804.10655

[13] W.J. Percival et al., Measuring the Baryon Acoustic Oscillation scale using the Sloan Digital
Sky Survey and Galaxy Redshift Survey

[14] F. Beutler et al., The 6dF Galaxy Survey: baryon acoustic oscillations and the local Hubble
constant, MNRAS 416, 3017–3032 (2011)

[15] A.J. Ross et al., The clustering of galaxies in the completed SDSS-III Baryon Oscillation
Spectroscopic Survey: cosmological analysis of the DR12 galaxy sample., arXiv:1607.03155,
MNRAS 470, 2617 (2017)

[16] S. Alam et al., The clustering of galaxies in the SDSS-III Baryon Oscillation Spectroscopic
Survey: baryon acoustic oscillations in the Data Releases 10 and 11 Galaxy samples, MN-
RAS 441, 24–62 (2014)

[17] M. Betoule et al., Improved cosmological constraints from a joint analysis of the SDSS-II
and SNLS supernova samples, Astronomy & Astrophysics 568, A22 (2014)

[18] Keck Array and BICEP2 Collaborations, Improved Constraints on Cosmology and Fore-
grounds from BICEP2 and Keck Array Cosmic Microwave Background Data with Inclusion
of 95 GHz Band, Phys. Rev. Lett. 116, 031302 (2016)

[19] A.G. Riess et al., Large Magellanic Cloud Cepheid Standards Provide a 1% Foundation
for the Determination of the Hubble Constant and Stronger Evidence for Physics beyond
ΛCDM, Astrophys. J. 876, 85 (2019)
B Quantum Fluctuations during Inflation
Subhorizon scales during inflation are microscopic and therefore quantum effects are important.
Thus we should study the behavior of the inflaton field using quantum field theory. To warm
up we first consider quantum field theory of a scalar field in Minkowski space.

B.1 Vacuum fluctuations in Minkowski space

The field equation for a massive free (i.e. V (ϕ) = 12 m2 ϕ2 ) real scalar field in Minkowski space is

ϕ̈ − ∇2 ϕ + m2 ϕ = 0 , (1)

or
ϕ̈k + Ek2 ϕk = 0 , (2)
where Ek2 = k2 + m2 , for Fourier components. We recognize Eq. (2) as the equation for a har-
monic oscillator. Thus each Fourier component of the field behaves as an independent harmonic
oscillator.
In the quantum mechanical treatment of the harmonic oscillator one introduces the creation
and annihilation operators, which raise and lower the energy state of the system. We can do
the same here.
Now we have a different pair of creation and annihilation operators â†k , âk for each Fourier
mode k. We denote the ground state of the system by |0#, and call it the vacuum. As discussed
earlier, particles are quanta of the oscillations of the field. The vacuum is a state with no
particles. Operating on the vacuum with the creation operator â†k , we add one quantum with
momentum k and energy Ek to the system, i.e., we create one particle. We denote this state
with one particle, whose momentum is k by |1k #. Thus

â†k |0# = |1k # . (3)

This particle has a well-defined momentum k, and therefore it is completely unlocalized (Heisen-
berg’s uncertainty principle). The annihilation operator acting on the vacuum gives zero, i.e.,
not the vacuum state but the zero element of Hilbert space (the space of all quantum states),

âk |0# = 0 . (4)

We denote the hermitian conjugate of the vacuum state by $0|. Thus

$0|âk = $1k | and $0|â†k = 0 . (5)

The commutation relations of the creation and annihilation operators are

[â†k , â†k! ] = [âk , âk! ] = 0, [âk , â†k! ] = δkk! . (6)

When going from classical physics to quantum physics, classical observables are replaced by
operators. One can then calculate expectation values for these observables using the operators.
Here the classical observable !
ϕ(t, x) = ϕk (t)eik·x (7)
is replaced by the field operator
!
ϕ̂(t, x) = ϕ̂k (t)eik·x (8)

159
B QUANTUM FLUCTUATIONS DURING INFLATION 160

where1
ϕ̂k (t) = wk (t)âk + wk∗ (t)â†−k (9)
and
1
wk (t) = V −1/2 √ e−iEk t (10)
2Ek
is the mode function, a normalized solution of the field equation (2). We are using the Heisenberg
picture, i.e. we have time-dependent operators; the quantum states are time-independent.
Classically the ground state would be one where ϕ = const. = 0, but we know from the
quantum mechanics of a harmonic oscillator that there are oscillations even in the ground state.
Likewise, there are fluctuations of the scalar field, vacuum fluctuations, even in the vacuum
state.
We shall now calculate the power spectrum of these vacuum fluctuations. The power spectrum
is defined as the expectation value

k3
Pϕ (k) = V "|ϕk |2 # (11)
2π 2
and it gives the variance of ϕ(x) as
∞
dk
!
2
"ϕ(x) # = Pϕ (k) . (12)
0 k

For the vacuum state |0# the expectation value of |ϕk |2 is

"0|ϕ̂k ϕ̂†k |0# = |wk |2 "0|âk â†k |0# + wk2 "0|âk â−k |0# + (wk∗ )2 "0|â†−k â†k |0# + |wk |2 "0|â†−k â−k |0#
= |wk |2 "1k |1k # = |wk |2 , (13)

since all but the first term give 0, and our states are normalized so that "1k |1k! # = δkk! . From
Eq. (10) we have |wk |2 = 1/(2V Ek ). Our main result is that

k3
Pϕ (k) = V |wk |2 (14)
2π 2
for vacuum fluctuations, which we shall now apply to inflation, where the mode functions wk (t)
are different.

B.2 Vacuum fluctuations during inflation

During inflation the field equation (for inflaton perturbations) is, from Sec. 8.6,
"# $ %
k 2 $$
δϕ̈k + 3Hδϕ̇k + + V (ϕ̄) δϕk = 0 . (15)
a

There are oscillations only in the perturbation δϕ, the background ϕ̄ is homogeneous and evolv-
ing slowly in time. For the particle point of view, the background solution represents the
vacuum,2 i.e., particles are quanta of oscillations around that value.
1
We skip the detailed derivation of the field operator, which belongs to a course of quantum field theory.
See, e.g., Peskin & Schroeder, section 2.3 (note different normalizations of operators and states, related to doing
Fourier integrals rather than sums and considerations of Lorentz invariance.)
2
This is not the vacuum state in the sense of being the ground state of the system. The true ground state has
ϕ̄ at the minimum of the potential. However there are no particles related to the background evolution ϕ̄(t).
B QUANTUM FLUCTUATIONS DURING INFLATION 161

After making the approximations H = const. and

V !!
= 3η ≈ 0 (16)
H2
we found that the two independent solutions for δϕk (t) are
! " ! "
H k ik
wk (t) = V −1/2 √ i+ exp (17)
2k3 aH aH

and its complex

√ conjugate wk∗ (t), where the time dependence is in a = a(t) ∝ eHt . The factor
V −1/2 H/ 2k3 is here for normalization purposes (V = L3 being the reference volume, not the
inflaton potential).
When the scale k is well inside the horizon, k $ aH, δϕk (t) oscillates rapidly compared to
the Hubble time H −1 . If we consider distance and time scales much smaller than the Hubble
scale, spacetime curvature does not matter and things should behave like in Minkowski space.
Considering Eq. (17) in this limit, one finds (exercise) that wk (t) indeed becomes equal to the
Minkowski space mode function, Eq. (10). (We cleverly chose the normalization in Eq. (17) so
that the normalizations would agree.) Therefore the wk (t) of Eq. (17) is our mode function. We
can use it to follow the evolution of the mode functions as the scale approaches and exits the
horizon.
The field operator for the inflaton perturbations is

δϕ̂k (t) = wk (t)âk + wk∗ (t)â†−k , (18)

and the power spectrum of inflaton fluctuations is

k3
Pϕ (k) = V |wk |2 . (19)
2π 2
Well before horizon exit, k $ aH, observed during timescales % H −1 , the field operator
δϕ̂k (t) becomes the Minkowski space field operator and we have standard vacuum fluctuations
in δϕ.
Well after horizon exit, k % aH, the mode function becomes a constant
iH
wk (t) → V −1/2 √ , (20)
2k3
the vacuum fluctuations “freeze”, and the power spectrum acquires the constant value
"2
k3
!
H
Pϕ (k) = V 2 |wk |2 = . (21)
2π 2π