CHAPTER 3

THE INFLATIONARY PARADIGM

Ubi materia, ibi geometria.

Johannes Kepler

3.1 The hot Big Bang paradise

In General Relativity, the Universe as a whole becomes a dynamical entity that can be
modeled and measured. The combination of Einstein’s theory of gravity with the Standard
Model of particle physics gives rise to the successful hot Big Bang (hBB) scenario, describing
the evolution of the Universe and its matter content from the first fraction of a second till
the present era. The expansion of the Universe, the relative abundance of light nuclei or the
discovery of the Cosmic Microwave Background (CMB) give confidence in the basic picture,
the expansion and cooling of a primordial soup. Many of the key cosmological parameters
describing the Universe have been accurately determined. This has led to the establishment
of a precision cosmological model known as Λ Cold Dark Matter (ΛCDM). At the same time,
these parameters provide useful information for particle physics. The stringent limits on the
sum of neutrino masses and on the variations of fundamental constants clearly illustrate the
entanglement between cosmology and high-energy physics.

3.1.1 Homogeneity and isotropy

In this course, we will not be interested in local objects such as galaxies or stars, but rather on
the dynamics of the Universe as a whole. In particular, we will average over local structures
and assume the Universe to be described by an approximately homoheneous and isotropic
“gas” of matter, whose “molecules” are, for example, galaxies. On physical grounds, homo-
geneity means that the physical conditions are the same at every point. Isotropy at every
point automatically implies homogeneity.

Exercise
Convince yourself that homogeneity does not imply isotropy. Provide some examples.
3.1 The hot Big Bang paradise 26

Figure 3.1: The Cosmic Microwave Background as seen by Planck. The fluctuations on top
of the average temperature T = 2.73K ' 0.235 meV are one part in 105 .

At first sight, the idealization of the Universe as an homogeneous and isotropic object might
seem a bit drastic. On the other hand, we know from hydrodynamics that a continuous
description of gases works very well even if these have a very discontinuous structure at
molecular scales. The homogoneous and isotropic approximation seems to be indeed in good
agreement with observations. Indeed. both the CMB and the galaxy distribution look rather
homogeneous when averaged on sufficiently large scales (cf. Figs. 3.1 and 3.2).
A given spacetime in General Relativity is specified by its metric tensor gµν . This quantity
defines the line element
ds2 = gµν dxµ dxν , (3.1)
where dxµ stands for infinitesimal displacements in the coordinates xµ . From a mathematical
point of view, an homogeneous and isotropic Universe must be equipped with a metric tensor
invariant under translations and rotations in the spatial components. The most general 4-
dimensional geometry consistent with these symmetries is the so called Friedmann-Lemaı̂tre-
Robertson-Walker (FLRW) spacetime,

dr2
 
2 2 2 2 2 2 2


ds = −dt + a (t) + r dθ + sin θdφ . (3.2)
1 − kr2

This equation represents a time-ordered slicing of spacetime with respect to a global time t
whose 3-dimensional spacial surfaces are maximally symmetric. Here r is a radial coordinate
and θ and φ are the usual angular coordinates on a two-sphere, ranging between 0 < θ < π
and 0 ≤ φ < 2π. The coordinates (r, θ, φ) are usually called comoving coordinates, since they
are decoupled from the effect of expansion.
The FLRW metric is invariant under the redefinition
k p a
k→ , r→r |k| , a→ p , (3.3)
|k| |k|

meaning that the the only relevant parameter is the sign of k. We can therefore distinguish
three types of spatial sections:

1. Flat: for k = 0 the spacial slices are flat and r ranges from zero to infinity, 0 < r < ∞.
3.1 The hot Big Bang paradise 27

Figure 3.2: The Sloan Digital Sky Survey map. Each dot is a galaxy. The empty regions are
just areas that the survey did not cover.

Figure 3.3: 2-dimensional projection of the 3-dimensional slices of the FLRW metric for
k = +1 (left) and k = −1 (right).

2. Open: for k = −1 the spacial slices are hyperbolic and again 0 < r < ∞.

3. Closed: for k = 1, the spacial slices are three-spheres and the radial coordinate r is
restricted to a compact range, 0 < r < 1.

The scale factor a(t) characterizes the relative size of the spacial sections at a given time. Its
temporal evolution depends on the matter content of the Universe via the Einstein equations
1
Rµν − R gµν + Λgµν = 8πG Tµν , (3.4)
2
with G the Newton’s constant, Rµν the Ricci tensor, R = g µν Rµν the Ricci scalar and Λ the
infamous cosmological constant. The energy-momentum tensor Tµν encodes the Universe’s
matter content and is locally conserved,

∇µ Tµν = 0 . (3.5)
3.1 The hot Big Bang paradise 28

The homogeneity and isotropy of the background metric restricts the form of the energy-
momentum tensor to the perfect fluid case

Tµν = p gµν + (ρ + p)uµ uν , (3.6)

with uµ the comoving four-velocity satisfying uµ uµ = −1 and ρ(t) and p(t) the local energy
density and pressure of the fluid. For an observer comoving with the fluid, uµ = (1, 0, 0, 0),
the energy-momentum tensor looks isotropic

T µν = diag(−ρ(t), p(t), p(t), p(t)) . (3.7)

Note that the trace is given by

T = T µ µ = −ρ + 3p . (3.8)

Exercise
Derive Eqs. (3.7) and (3.6).

3.1.2 Friedmann equations

Given the FLRW metric (3.2) metric we can compute the connection coefficients, the Ricci
tensor components and the Ricci scalar. The computation of these quantities is straightfor-
ward but quite tedious, so we will simply summarize the results:

i) Connection coefficients
aȧ
Γ011 = Γ022 = aȧr2 Γ033 = aȧr2 sin2 θ
1 − kr2
ȧ
Γ101 = Γ110 = Γ202 = Γ220 = Γ303 = Γ330 =
a
Γ122 = −r(1 − kr2 ) Γ133 = −r(1 − kr2 ) sin2 θ
1
Γ212 = Γ221 = Γ313 = Γ331 =
r
Γ233 = − sin θ cos θ Γ23 = Γ332 = cot θ .
3
(3.9)

ii) Ricci tensor components

ä
R00 = −3
a
aä + 2ȧ2 + 2k
R11 =
1 − kr2
R22 = r2 (aä + 2ȧ2 + 2k)
R33 = r2 (aä + 2ȧ2 + 2k) sin2 θ , (3.10)

iii) Ricci scalar

6
R= (aä + ȧ2 + k) . (3.11)
a2
3.1 The hot Big Bang paradise 29

Combining these expressions with the energy-momentum tensor (3.7) we can particularize
the Einstein’s equations (3.4) to the homogenous and isotropic case. We obtain the so-called
Friedmann equations
ρ Λ k
H2 = 2 + − 2, (3.12)
3MP 3 a
1 Λ
Ḣ + H 2 = − 2 (ρ + 3p) + , (3.13)
6MP 3
with the dots denoting derivatives with respect to the coordinate time t and

MP = (8πG)−1/2 = 2.436 × 1018 GeV (3.14)

the reduced Planck mass. The quantity

ȧ
H≡ (3.15)
a
is the so-called Hubble rate. This parameter is positive for an expanding Universe and negative
for a contracting one. The value of the Hubble parameter at the present epoch is the Hubble
constant
H0 = 100 h km s−1 Mpc , (3.16)
with h = 67.8 ± 0.9 and Mpc= 3 × 1024 cm standing for “megaparsec”. This value allows us
to estimate the present age and size of the Univere

H0−1 = 9.77 h−1 Gyr cH0−1 = 3000 h−1 Mpc . (3.17)

The Friedmann equations (3.12) and (3.13) can be combined to obtain the continuity equation

ρ̇ + 3H (ρ + p) = 0. (3.18)

Exercise
Derive Eq. (3.18) i) by combining Eqs. (3.12) and (3.13) and ii) from the covariant
energy-momentum conservation (3.5). Interpret the result by considering the adiabatic
dilution of energy due to the expansion and the work done by pressure.
Hint: Consider the second law of thermodynamics T dS = dU + pdV .

The cosmological evolution following from Eqs. (3.12), (3.13) and (3.18) can be determined
once a pressure to energy density relation p(ρ) is specified. We will restrict ourselves to
barotropic fluids for which the pressure is linearly proportional to the energy density,

p = wρ , (3.19)

with w the so-called equation-of-state parameter. This case covers the two main matter
components in the hot Big Bang scenario, namely (non-relativistic) matter (w = 0) and
radiation (w = 1/3).
3.2 Troubles in paradise 30

Exercise
Consider a macroscopic collection of structureless point particles interacting through
spatially localized collisions. On distances d much larger than the typical mean free
path, the number of particles is large and the statistical fluctuations about the mean
properties of the fluid are expected to be small. If the fluid is isotropic,a the mean
density and pressure observed by a comoving observer over the volume ∆V = d3 can
be written as
DX E 1 X D X i i (3) E
ρ= En δ (3) (x − xn ) , p= pn vn δ (x − xn ) , (3.20)
n
∆V 3 n
∆V
i
p
with En = p2n + m2n the energy of the individual particles. The index i is a space
index ranging from 1 to 3 and n selects the particle of mass mn and momentum pn . Use
these microscopic expressions to derive the equation of state for non-relativistic matter
and radiation.
a
i.e if the fluid is perfect.

In our our Universe, several species with different equations of state coexist. Their relative
contribution is traditionally parametrized by the dimensionless parameters
ρM ρR Λ k
ΩM ≡ , ΩR ≡ , ΩΛ ≡ , ΩK ≡ − , (3.21)
3MP2 H 2 3MP2 H 2 3MP2 H 2 (aH)2
with the subindices M, R, Λ and K standing for matter, radiation, cosmological constant and
curvature contributions. At present time, the radiation and curvature contributions are very
small (ΩR ' 5 × 10−5 , ΩK < 0.005) and
ΩM ' 0.3 , ΩΛ ' 0.7 , (3.22)
Our present Universe is therefore dominated by a cosmological constant or dark energy com-
ponent. Note however that is was dominated by matter and radiation in the past. This can
be easily seen by considering the scaling of non-relativistic matter and radiation. Integrating
the conservation equation (3.18) and using Eq. (3.12) for the zero curvature case (k = 0),
we get  2/3(1+w)
−3(1+w) t w 6= −1 ,
ρ∝a , a(t) ∝ Ht (3.23)
e w = −1 .
For non-relatistic matter (w = 0), the energy density dilutes with the volume ρM ∼ a−3 ,
reflecting mass conservation. For relativistic matter (w = 1/3), the energy density dilutes
as ρR ∼ a−4 , due to the additional redshift of energy (∝ a−1 ). Note that the radiation
domination period cannot be eternal to the past. When t → 0, the scale factor goes to zero
and the physical energy density ρ diverges.

3.2 Troubles in paradise

In spite of the success of the hot Big Bang for describing the observed Universe, it is not free
of problems.
3.2 Troubles in paradise 31

3.2.1 Flatness problem

Consider the dimensionless energy density parameter
ρ
Ω≡ , (3.24)
ρcrit
with
ρcrit ≡ 3MP2 H 2 (3.25)
the so-called critical energy density. In terms of this quantity, the Friedmann equation (3.12)
becomes
k
Ω−1= . (3.26)
(aH)2
The quantity Ω − 1 measures the curvature of the Universe. A Universe with flat spacial
sections (k = 0) corresponds to Ω = 1. For k 6= 0, the evolution of the curvature depends
on the evolution of the comoving Hubble radius (aH)−1 . If the Hubble radius (aH)−1 in-
creases/decreases with time, the curvature increases/decreases accordingly. In a Universe
dominated by a fluid with equation of state w, the comoving Hubble radius evolves as
1
(aH)−1 ∝ a 2 (1+3w) . (3.27)

For standard matter sources satisfying the strong energy condition 1 + 3w > 0, (aH)−1 grows
as the Universe expands. For instance, during matter (MD) and radiation domination (RD)
we have
(aH)−1 ∝ a1/2 , (MD) (aH)−1 ∝ a . (RD) (3.28)

Exercise
Derive Eq .(3.27).

The density parameter Ω at present time is very close to one. Specifically, the latest Planck
satellite data combined with baryon acoustic oscillations (BAO) give

Ω0 − 1 = 0.000 ± 0.005 , (3.29)

at the 95% C.L. Taking into account this value (Ω0 ∼ 10−3 ) together with the evolution
equations (3.28) for the comoving Hubble radius during matter and radiation domination,
we can compute the value of Ω − 1 at the time of matter-radiation equality (zeq = 3600)
 
Ω(zeq ) − 1 = Ω0 − 1 (1 + zeq )−1 ≈ 2.8 × 10−5 , (3.30)

and at Big Bang Nucleosynthesis (zBBN = 1010 )

1 + zeq 2
 
Ω(zBBN ) − 1 = (Ω(zeq ) − 1) ≈ 3.6 × 10−18 . (3.31)
1 + zBBN
A percent deviation from flatness in the present Universe translates into unnaturally small
deviations at early epochs. In others words, in order to recover the Universe we observe
3.2 Troubles in paradise 32

2.0

1.5

Ω 1.0

0.5

0.0

0 1 2 3 4 5 6
log a

Figure 3.4: Evolution of the energy density parameter Ω in standard cosmology. The point
Ω = 1, corresponding to flat curvature, is a repeller.

today the initial conditions must be terribly fine-tuned. Any deviation from these initial
conditions translates either into a closed and recollapsing Universe or into an open Universe
completely dominated by curvature. This extreme dependence on the initials conditions is
highly unsatisfactory.

Exercise
One could argue that naturalness is just a question of taste and that the most symmetric
initial conditions are somehow more natural. However, this is not very convincing
from the point of view of the self-consistency of the theory, specially if those initial
conditions are unstable. Show that the dimensionless energy density parameter satisfies
the differential equation
dΩ
= (1 + 3w)Ω(Ω − 1) . (3.32)
d log a
Note that for both matter (w = 0) and radiation (w = 1/3) we have 1 + 3w > 0,
meaning that a flat Universe with Ω = 1 is an unstable fixed point. If Ω > 1 at some
point of the evolution, it will keep on growing; and viceversa, if Ω < 1 at some point, it
will keep on decreasing. This behaviour is illustrated in Fig. 3.4.

3.2.2 Horizon problem

Our Universe seems to be extremely homogeneous on large scales. The CMB temperature
anisotropies arise only at the level of one part in 105 . However, if the Universe were only
radiation- or matter-dominated in the past, any two regions in the sky with angular separation
of a few arc degrees would not have been able to communicate between the singularity and
recombination to decide which the common temperature is supposed to be. This is the so-
called horizon problem. To discuss this problem let us start describing the causal structure
3.2 Troubles in paradise 33

⌧ hBB cosmology ⌧ Inflationary cosmology

⌧0 ⌧0

⌧CMB ⌧CMB

⌧i=0 R
⌧end R

Causal
⌧i = 1

Figure 3.5: Conformal diagram for standard hBB cosmology (left) and inflationary cosmology
(right). Inflation solves the horizon problem by extending the conformal diagram to negative
conformal times.

of the FLRW metric. Performing a change of coordinates


dr  sin R for k = 1 ,
dR = √ , r = Sk (R) = R for k = 0 , , (3.33)
1 − kr2 
sinh R for k = −1 ,

the FLRW metric (3.2) can be written as

ds2 = −dt2 + a(t)2 dR2 + Sk2 (R)(dθ2 + sin2 θdφ2 ) .

 
(3.34)

Defining now the conformal time

dt
Z
τ= , (3.35)
a(t)
we can recast Eq. (3.34) in the conformal form

ds2 = a(τ )2 −dτ 2 + dR2 + Sk2 (R)(dθ2 + sin2 θdφ2 ) .

 
(3.36)

The causal structure of the FLRW metric is determined by the way in which light propagates
on null geodesics with ds2 = 0. Since the space is isotropic, we can freely set the coordinates
θ and φ to a constant value. In this case, the condition ds2 = 0 implies

R = ±τ + constant . (3.37)

The fact that geodesics are 45◦ lines in the {τ, R} plane is related to the fact that the the
FLRW metric (3.34) is conformally flat. If the Universe started at some initial time ti , then
there is a maximum amount of time for light to have travelled. The (comoving) particle
3.3 Inflationary paradigm 34

horizon is the largest distance that a photon can travel between ti and a later time t > ti
(recall that c ≡ 1)
t a ln a
dt da
Z Z Z
dH = τ − τi = = = (aH)−1 d ln a . (3.38)
ti a ai aȧ ln ai

According to this expression, the evolution of dH depends also on the evolution of the co-
moving Hubble radius (aH)−1 (cf. Eq. (3.27)).

Exercise
Show this.

For standard matter sources 1 + 3w > 0 and (aH)−1 grows as the Universe expands. When
that happens. the integral in Eq. (3.38) becomes dominated by its upper limit
2 1 2
dH ∝ a 2 (1+3w) = (aH)−1 . (3.39)
(1 + 3w) (1 + 3w)

Note that due to the presence of the singularity at ai = 0 (or equivalently at τi = 0) this
quantity is finite. At each instant of time, regions that were never in causal contact before
get into contact for the first time. The fact that two of these regions share approximately
the same temperature cannot by a consequence of thermal equilibrium. On general grounds,
these regions should be expected to look very different from each other. This applies also to
the CMB (see the left pannel of Fig. 3.5). The observed homogeneity of the CMB map is not
only remarkable but also strange and unexpected! Most points in that map share roughly
the same temperature even if the naive horizon scale at decoupling is just a few arc degrees.
How is this possible?

3.3 Inflationary paradigm

The horizon and flatness problems are intimately related to the fact that the comoving Hubble
radius (aH)−1 grows within the standard hot Big Bang scenario. A simple solution to these
problems is to postulate a decrease of (aH)−1 at early times

d
(aH)−1 < 0 . (3.40)
dt
or equivalently a violation the strong energy condition 1 + 3w > 0 (cf. Eq (3.27)). This
additional phase in the history of the Universe is called inflation. The name can be easily
understood by noticing that Eq. (3.40) implies accelerated expansion

d d ä
(aH)−1 = (ȧ)−1 = − 2 < 0 ⇒ ä > 0. (3.41)
dt dt (ȧ)
3.3 Inflationary paradigm 35

No weak energy condition violation

Note that the condition ä > 0 is very different from Ḣ > 0 with
ȧ 1
Ḣ = − H2 = − (ρ + p) . (3.42)
a 2MP2

In an expanding Universe the energy density is always decreasing or at most constant.

In order to have Ḣ > 0, the null energy condition ρ + p ≥ 0 should be violated.

If the inflationary stage lasts long enough, the hot Big Bang problems are automatically
solved. In particular, if the comoving Hubble radius (aH)−1 in Eq. (3.26) decreases, the
curvature is driven towards zero (the unstable point Ω = 1 in Eq. (3.32) becomes now an
attractor). This solves the flatness problem. On the other hand, if 1 + 3w < 0 the integral
in Eq. (3.38) becomes dominated by its lower limit and the singularity is pushed towards
negative conformal times,
2 1
(1+3w)
τi ∝ ai2 = −∞ . (3.43)
(1 + 3w)
The extension of the conformal diagram to negative conformal times allows the light cones
of widely separated CMB points to intersect (see the right pannel of Fig. 3.5). This solves
the horizon problem.

3.3.1 Minimal duration of inflation

Due to the Hubble shrinking, typical scales that were inside the horizon at the onset of
inflation, leave the radius of causal contact as inflation proceeds. When inflation ends, the
comoving Hubble radius (aH)−1 starts increasing and the scales reenter the horizon. This is
illustrated in Fig. 3.6. A simple inspection of this figure reveals that in order to solve the hot
Big Bang problems we must require

(ai Hi )−1 ≥ (a0 H0 )−1 , (3.44)

with (ai Hi )−1 the comoving Hubble radius at the onset of the inflationary regime and
(a0 H0 )−1 its value today. Let us assume for simplicity that the radiation-dominated epoch
starts immediately after the end of inflation and neglect the comparatively shorter matter
and dark-energy dominated epochs. Under these assumptions, we have

(ai Hi )−1 (ai Hi )−1 (a0 H0 )−1 a0

−1
= −1 −1
≥ , (3.45)
(aend Hend ) (a0 H0 ) (aend Hend ) aend
where in the last step we have made use of the condition (3.44) together with the radiation
domination scaling H ∝ a−2 . Taking into account that a0 /aend ∼ Tend /T0 and assuming
inflation to finish at an energy scale1 Tend ∼ 1014 GeV ∼ 1026 T0 , with T0 ∼ 10−3 eV the
1
Although accurate enough for a large set of inflationary models, this value is taken for illustration purposes
only. On general grounds, the precise number of e-folds must be computed model by model taking into account
not only the energy scale at the end of inflation, but also the details of the reheating process and the effects
of any intermediate era between the end of inflation and the onset of radiation domination.
3.3 Inflationary paradigm 36

Scales

1
(ai Hi ) 1 (aH)
Horizon exit Horizon reentry
1
(a0 H0 )

Today
CMB
log a
Inflation Heating Radiation Matter

Figure 3.6: Scales of cosmological interest as a function of the number of e-folds. Due to the
Hubble shrinking, typical scales λ ≡ (a0 H0 )−1 that were inside the horizon at the onset of
inflation, leave the radius of causal contact as inflation proceeds. When inflation ends, the
comoving Hubble radius (aH)−1 starts increasing and the scales reenter the horizon.

temperature of the Universe today, we get

(ai Hi )−1
≥ 1026 ' e60 . (3.46)
(aend Hend )−1

For Hi ' Hend (see below), this conditon becomes

 
aend
N ≡ log & 60 . (3.47)
ai

Therefore, in order to solve the inflationary problems we need, at least, N = 60 e-folds of

inflation.

3.3.2 Hubble flow parameters

The conditions for inflation are traditionally formulated as conditions on the variation of the
Hubble rate. Defining the fractional change of the Hubble rate per e-fold2 N ,

Ḣ d ln H
≡− =− , dN ≡ Hdt = d ln a , (3.48)
H2 dN
we can rewrite Eq. (3.40) as
d 1
(aH)−1 = − (1 − ) . (3.49)
dt a
2
This quantity will play a central role in the effective field theory of inflation to be presented in Chapter 7.
3.3 Inflationary paradigm 37

For inflation to take place,  must be smaller than one. As argued in Section 3.3.1, the
solution of the flatness and horizon problems requires a rather long inflationary stage. In
order to achieve this, the fractional change of ,
d ln  ˙
η≡ = , (3.50)
dN H
must also be small, |η| < 1. Note that the η parameter are just the first two elements of a
full series of Hubble flow parameters
d ln i ˙i
i+1 ≡ = . (3.51)
dN Hi

3.3.3 de Sitter spacetime

The minimal value of the  parameter is zero. In this case the Hubble rate H is constant and
the Universe expands exponentially fast, a(t) = eHt . This limit motivates the study of the
so-called de Sitter spacetime.
The de Sitter spacetime dS4 can be represented as a 4-dimensional hyperboloid extrinsically
embedded in a d=5 Minkowski spacetime ηAB = diag(−1, 1, 1, 1, 1) with coordinates z A

ηAB z A z B = −z02 + z12 + z22 + z32 + z42 = l2 . (3.52)

The quantity l ≡ 1/H is the so-called de Sitter radius.3 This representation makes explicit
the symmetries of the de Sitter space: rotations and Lorentz transformations in the 10 planes
formed by pairs of the five coordinates z A . This ten parameter SO(4, 1) group plays the same
instrumental role than the Poincare group in Minkowski spacetime. In particular, it greatly
facilitates computations as far as quantum field theory is concerned.
The dS4 spacetime can be also described in an intrinsic way. Consider the transformation
1 H
z0 = sinh(Ht) + eHt δij xi xj , (3.53)
H 2
Ht
zi = x i e , (3.54)
1 H
z4 = cosh(Ht) − eHt δij xi xj , (3.55)
H 2
with i = 1, 2, 3 and −∞ < t < ∞ and −∞ < xi < ∞. In this coordinate system the line
element (3.52) becomes a special case of the flat FLRW spacetime

ds2 = −dt2 + a2 (t)δij dxi dxj , (3.56)

with a(t) = eHt . Note however that Eq. (3.56) is not completely equivalent to (3.52), since
the coordinates {t, xi } cover only half of de Sitter manifold. This can be easily seen by adding
the z0 and z4 coordinates (see also Fig. (3.7)),
1 Ht
z 0 + z4 = e ≥ 0. (3.57)
H

3
The choice of notation will become clear soon.
3.3 Inflationary paradigm 38

Figure 3.7: The embedding of de Sitter space into a five dimensional flat space-time with
two spatial coordinates suppressed. The flat coordinates in (3.53)-(3.55) cover only half of
de Sitter manifold. The surfaces (lines) of constant t and constant x are also indicated.

Exercise
1. Derive Eq. (3.56) from the 5-dimensional embedding (3.52).

2. Other choices of coordinates leading to FLRW metrics with open and closed spatial
sections can be also considered. Find these sets of coordinates.
It is interesting to recast (3.56) in terms of the conformal time (3.35). Taking into account
that
1 1
τ = − e−Ht =⇒ a(τ ) = − , (3.58)
H Hτ
the line element takes the manifestly conformally flat form
1
ds2 = −dτ 2 + δij dxi dxj ,


(3.59)
H 2τ 2
with η ranging between −∞ and 0. Note that Eq. (3.59) is manifestly invariant under the
rescaling
τ → λτ , xi → λxi . (3.60)
As we will see in Section 6.1.3, this symmetry plays a central role in the properties of the
primordial perturbations generated during inflation. But let not anticipate things and focus
on the background evolution for the time being. What seems clear is that in order to recover
the hot Big Bang scenario the de Sitter phase cannot be eternal. In other words, we need to
equip the de Sitter space with a clock.