(Mini) Introduction to

cosmological perturbation theory

Juan P. Beltrán Almeida
Departamento de Física
Facultad de Ciencias

VII Uniandes Particle Physics School

December 6th, 2022

Universidad de los Andes
 Cosmological perturbations
Summary

1. Inflation. Background evolution and basic definitions.

2. Scalar field inflation.
3. Inflationary perturbations and gauge fixing.
4.Stochastic properties and statistical approach.
5. Cosmological perturbations in large scale structure
formation.

Notation
Natural units: c = h = 1.
Signature: ημν = diag(−1, 1, 1, 1) .
 Cosmological Background
Einstein equations and Friedman-Lemaitre-Robertson-Walker
(FLRW) solution
Einstein equations dictates the dynamics of the universe:
1
Gμν = Rμν − gμν R = 8πGTμν .
2
1 αβ
R σμρν = ∂ρΓσμν − ∂νΓσμρ + ΓσαρΓαμν − ΓσανΓαμρ, Γαμν = g (∂μgνβ + ∂νgμβ − ∂αgμν) .
2
Rμν = R σμσν, R = g μν Rμν .
Cosmological principle: homogeneous and isotropic fluid at large
scales. L ∼ 1 - 100 Mpc. 1 Mpc = 3x1024 cm ∼ 3x106 - 3x108 light-years.
Homogeneous Isotropic
 Cosmological Background
Einstein equations and Friedman-Lemaitre-Robertson-Walker
(FLRW) solution
FLRW solution: A solution reflecting isotropy and homogeneity

ds 2 = − dt 2 + a 2(t) δijdx idx j .

Symmetric under translations and rotations:

x i → x i + d i, x i → R ij x j, R ij ∈ SO(3) .
Kinematic quantities:
Observeres (timelike) 4-velocity
μ
dx
uμ = , gμνu μu ν = − 1, u μ = (∂t)μ = (1, 0, 0, 0) .
ds
Induced spacial metric: uμ
μ hμν
hμν = gμν + uμuν, hμνu = 0.
 Cosmological Background
Einstein equations and Friedman-Lemaitre-Robertson-Walker
(FLRW) solution
Matter contents: Energy momentum tensor of a general fluid:
Tμν = ρuμuν + phμν + 2q(μuν) + Σμν .
μ ν Matter-Energy 1 μν Isotropic
ρ = Tμνu u → p = Tμνh →
density. 3 pressure.
Zero trace
qμ = − hμσTσρu ρ → Energy flux Σμν = h(μσhν)ρTσρ → anisotropic
stress tensor

Energy momentum tensor of a single, perfect fluid (qμ = 0, Σμν = 0):

Tμν = (ρ + p)uμuν + pgμν, Tμν = diag(ρ, p, p, p) .

Continuity equation
μ
∇ Tμν = 0.
 Cosmological Background
Einstein equations and Friedman-Lemaitre-Robertson-Walker
(FLRW) solution
1
Friedmann equations Gμν = Rμν − gμν R = 8πGTμν .
··
2 ·· ·
(a (a) )
2
a a a
R00 = − 3 , Rij = δij (2a + aa), R = g Rμν = 6
· 2 ·· μν
+ .
a

Hubble a· 2 8πG a·· · 2 4πG

H≡ , H = ρ, =H+H =− (ρ + 3p) .
rate a 3 a 3
Continuity μ dρ
∇ Tμν = 0 → + 3H(ρ + p) = 0.
equation dt
Equation of state
(barotropic fluid)
p = ωρ .

dρ da 1 da d ln ρ
+ 3ρ(1 + ω) = 0 → + 3(1 + ω) = 0 → ρ = ρ0a −3(1+ω) .
da dt a dt d ln a
 Cosmological Background
Causal structure at large scales
Evolution of the scale factor

−3(1+ω) 2 8πG 2
ρ = ρ0a , H = ρ → a = a0t 3(1 + ω) .
3
Particular cases of the equation of state

Ultra non-relativistic matter ω = 0, ρ = ρ0a −3, a = a0t 2/3 .

Radiation ω = 1/3, ρ = ρ0a −4, a = a0t 1/2 .
Vacuum, Λ ω = − 1, ρ = ρ0, a = a0e Ht .
Light geodesics and particle horizon
2 2 2 i jdt
Photons trajectory 0 = ds = − dt + a (t) δijdx dx → dr = .
a(t)
t Particle horizon
dt′
∫0 a(t′)
Physical distance that a
Maximum distance that an
photon travels RH(t) = a(t) .
observer can reach in a
causal way.
 Cosmological Background
Causal structure at large scales
Comoving and physical scales
Comoving distance x. Physical distance xp = ax,
Conformal time τ
ds 2 = − dt 2 + a 2(t) δijdx idx j = a 2(τ)[−dτ 2 + δijdx idx j],
Causally

dt t
dt′ τ connected

∫0 a(t′)
⇒ dτ = , τ= . with A
a(t) B
B′
Null geodesics Causally

x=
x = ± τ + const .

τ
disconnected

x=
−
from A

τ
Frontiers between regions in causal xi
A
contact with an observer, and
regions without causal contact.
 Cosmological Background
Causal structure at large scales

∫0 ( aH )
Hubble radius t t a
dt′ 1 dt′ 1
∫0 a(t′) ∫0 a(t′) da
τ= = da = d ln a .

Comoving particle horizon τ.

Hubble radius: H −1.
Comoving Hubble radius: (aH)−1.
Scales and Hubble radius

Physical wavelength λ = aλcomoving = a2π/k

Comparison of scales: (aH)−1 k
2π = .
λ aH
k
≪1 → λ outside the horizon .
aH
k
≫1 → λ inside the horizon .
aH
Number of e-folds dN = Hdt = d ln a .
 Cosmological Background
Inflation and the horizon problem (in few words)
Surface of last scattering and causally connected regions. The time when
photons decouple from matter, at temperature T ∼ 0.3 eV, at time t ∼
3x105 years.

∫0 ( aH )
a
1 1
τ= d ln a ∝ a 2 (1+3ω) .

For ω > 0, the horizon is growing monotonically, so, photons that we see
today, tracked back in the past, come from regions causally
disconnected. The problem with that is that those causally disconnected
regions share very similar features, for instance, the distribution of
temperature perturbations around decoupling time is homogeneous and
isotropic. This is, causally non-communicating regions share basically the
same features.
In terms o the Hubble radius. If the Hubble radius grows monotonically,
then, homogeneous regions that we see today were far outside the Hubble
radius in the past

(H)
1
RH =
 Cosmological Background
Inflation and the horizon problem (in few words)
Surface of last scattering and causally connected regions.

(H)
1
rH =

The evolution of Hubble radius vs the

evolution of a scale λ, the distance
between two photons coming from last
scattering surface

1
λ < rH = → λ inside the horizon .
H
1
λ > rH = → λ outside the horizon .
H

log a
Figure 2: The horizon scale (green line) and a physical scale λ (red line) as fu
 Physical
scale

Cosmological Background

Log (physical scale)

Physical
Hubble

Inflation and the horizon problem (in few words) length

Surface of last scattering and causally connected regions.

Log (time)
06 Slow-roll inflation
INFLATION Comoving
Physical Hubble
scale length

Log (comoving scale)

Log (physical scale)

Physical Comoving
Hubble
Scale
length

Log (time)

INFLATION Comoving
Hubble Log (time)
length
Fig. 18.1. Two views of the size of a comoving region within the observable Universe,
Log (comoving scale)

relative to the Hubble length (horizon scale). The comoving Hubble length 1/aH is de-
Hubble radius Comoving
creasing during inflation
Comoving Hubble
and increases afterwards radius
at least up to the present. (What happens

(H) ( aH )
Scale
in the future depends on the nature of the dark energy, as discussed in Section 23.5.) The
1 1
upper panel shows the physical size of the region, the lower one its comoving size. The
rH = R =
vertical axis covers many powers of 10 in scale. The region starts well inside the horizon,
then crosses outside some time beforeH the end of inflation, reentering long after inflation is
over.

the comoving region that will become the observable Universe actually becomes
smaller during inflation. This is illustrated in Figure 18.1.
 Cosmological Background
Inflation and the horizon problem (in few words)
Inflation and the horizon problem. In order to solve the problem with
causality posed by the horizon problem, we should have a shrinking
Hubble radius in the past, so:

dt ( aH )
d d 1
RH < 0 → < 0 → a·· > 0.
dt
From Friedmann equations:
a·· · 2 4πG 1
=H+H =− (ρ + 3p) ⇒ (ρ + 3p) < 0 ⇒ ω < − .
a 3 3

Inflation defined in terms of the evolution of the scale factor.

dt ( aH )
d 1
< 0 ⇔ a·· > 0.
6 The Physics of Inflation
Single field inflation
Inflation is a very unfamiliar physical phenomenon: within a fraction a second the universe grew
exponential at an accelerating rate. In Einstein gravity this requires a negative pressure source o
Inflationa nearly
equivalently driven by aenergy
constant single scalar
density. field
In this section we describe the physical condition
under which this can arise.
Mpl2
[ 2 ]
1 μ
∫
4
S = d x −g R + ∂ ϕ∂μϕ − V(ϕ)
6.1 Scalar Field Dynamics 2

reheating

Figure 10: Example Illustration

of an inflaton potential.
of an Acceleration occurs when the potential energy o
inflaton potential
the field, V ( ), dominates over its kinetic energy, 12 ˙ 2 . Inflation ends at end when the
kinetic energy has grown to become comparable to the potential energy, 12 ˙ 2 ⇡ V . CMB
fluctuations are created by quantum fluctuations about 60 e-folds before the end o
 Single field inflation
Inflaton dynamics
Mpl2
[ 2 ]
1 μ
∫
4
S = d x −g R + ∂ ϕ∂μϕ − V(ϕ)
2
Equation of motion for ϕ
δS 1
: ∂μ ( −g∂ μϕ) + ∂ϕV = 0.
δϕ −g
Energy-Momentum tensor

(2 )
2 δS (ϕ) 1
− : Tμν = ∂μϕ∂νϕ − gμν ∂αϕ∂αϕ + V(ϕ) .
−g δg μν

Gravitational field equation

1 (ϕ) 2 8πG · 2 4πG

Rμν − gμν R = 8πGTμν → H = ρϕ, H+H =− (ρϕ + 3pϕ) .
2 3 3
 Single field inflation
Inflaton dynamics
Homogeneous field ϕ. A homogeneous scalar field ϕ(x, t) = ϕ(t)
behaves like a perfect fluid and support the inflationary evolution
·· ·
ϕ + 3Hϕ + ∂ϕV = 0.
Pressure, energy density and continuity equation
(ϕ) 1 ·2
T00 = ρϕ = 2 ϕ + V(ϕ),
dρϕ
(ϕ) 1 ·2 + 3Hρϕ(1 + ωϕ) = 0.
Tii = pϕ = 2 ϕ − V(ϕ), dt
pϕ 1 ·2
2 ϕ − V(ϕ)
ωϕ = ρ = 1 · 2 .
ϕ 2 ϕ + V(ϕ)

Gravitational field equation/ Friedman equations

3 (2 )
2 8πG 1 · 2
H = ϕ + V(ϕ) ,

( )
· 2 4πG 1 ·2
H+H =− (1 + 3ωϕ) ϕ + V(ϕ) .
3 2
 Single field inflation
Slow roll inflation
Definition of slow roll parameter ϵ
·2 ·
a·· 1 2 3 1 ϕ H
= − H (1 + 3ωϕ) = H 2(1 − ε), ε ≡ (1 + ωϕ) = =− 2.
a 2 2 2H 2 H
·
H d ln H
ε≡− 2 =− .
H dN
Accelerated expansion in terms of slow roll parameter

a·· > 0 ⇔ ε < 1

de Sitter limit ωϕ → − 1, ε → 0.
1 ·2
pϕ 2ϕ − V(ϕ) ·2
ωϕ = ρϕ
= 1 ·2
→−1 → V(ϕ) ≫ ϕ .
2ϕ + V(ϕ)
··
Definition of second slow roll parameter η. ϕ is small enough to
sustain accelerated expansion
··
·· · ϕ 1 dε
| ϕ | ≪ {3Hϕ, ∂ϕV} . η ≡ · = ε −
Hϕ 2ε dN
 Single field inflation
Slow roll inflation
Potential slow roll parameters
Using Friedmann equations, slow roll parameters can also be seen as
conditions on the shape of the potential

Mpl2 Vϕ
2
Vϕϕ 1
2 (V)
Mpl2 ≡ .
ϵV ≡ , ηV ≡ Mpl2 . 8πG
V
Relations between both set of parameters ε ≈ ϵV , η = ηV − ϵV .
Exponential expansion
· Vϕ 8πG ·
a
ϕ≈− , H2 ≈ V(ϕ) ≈ const . → = H → a ≈ a0e Ht .
3H 3 a
End of inflation and number of e-folds

( a ) ∫t
tend ϕend
aend dϕ
∫ϕ
ε ≈ ϵV (ϕend) = 1 → N(ϕ) = ln = Hdt = H ·
ϕ
ϕend ϕend Enough inflation to
dϕ V 1 ϕ dϕ
∫ϕ ∫ϕ Vϕ Mpl ∫ϕend 2ϵV
N(ϕ) = H · ≈ − 8πG dϕ ≈ . solve IC problems
ϕ

( astart )
aend
Ntot = ln ≳ 60
 Inflationary perturbations
Scalar, vector and tensor degrees of freedom
Matter perturbations and metric perturbations
The Einstein equations couple matter perturbations to the metric
perturbations
(ϕ)
δGμν = 8πGδTμν .
δϕ → δTμν → δgμν → δϕ .

ϕ(x, t) = ϕ 0(t) + δϕ(x, t)

 Inflationary perturbations
Cosmological Perturbations
We can study perturbations around a homogeneous background:

(x, t) = (t) + (x, t), gµ⌫ (x, t) = gµ⌫ (t) + gµ⌫ (t)
ds 2 = (1 + 2 )dt 2 + 2aBi dx i dt + a2 [(1 2 ) ij + Eij ] dx i dx j

Scalar, vector, tensor decomposition:

Bi = @ i B Si , @ i Si = 0
Eij = 2@ij E + 2@(i Fj) + hij , @ i Fi = 0, @ i hij = 0.

Coordinate transformation: t ! t + ↵, xi ! xi + ij 0 Scalar metric

j
and matter transformations:

! ˙
↵, B !B +a 1
↵ a˙

E !E , ! + H↵
⇢! ⇢ ˙
⇢↵, p! p ṗ↵
Juan P. Beltrán Almeida (UAN) Inflation and the Origin of LSS December 9, 2014 13 / 20
 Inflationary perturbations
Gauge invariant variables
Gauge invariant variables.
Curvature perturbation on uniform-density hypersurfaces
H H
⇣⌘ + ⇢⇡ + (Slow-roll)
⇢˙ ˙

Comoving curvature perturbation:

H H
R⌘ q⇡ + (Slow-roll)
⇢+p ˙

⇣=R
for slow-roll and on super horizon scales k ⌧ aH.
We can calculate statistical properties in the form of correlation functions
(power spectrum, bispectrum, etc.) of these gauge variables!
 Inflationary perturbations
Correlation Functions. Statistical Properties of
Cosmological Perturbations.

Power spectrum.
k3
< Rk Rk 0 >= (2⇡)3 (k + k 0 )PR (k), 2
R ⌘ 2 PR (k)
2⇡
d ln 2R dns
ns 1⌘ , ↵s ⌘
d ln k d ln k

Power law spectrum

✓ ◆ns (k⇤ ) 1+ 12 ↵s (k⇤ ) ln(k/k⇤ )+...
2 k
R = As (k⇤ )
k⇤
 Inflationary perturbations
Scalar Perturbations.
Scalar action.
Z 
4 p 1 1
S= d x g R (r )2 V( )
2 2
Expanding up to 2nd order in R, (this is a long exercise of integration by
parts) we get:
Z ˙2 h i
1
S(2nd order) = d 4 x a3 Ṙ2 a 2
(@i R)2
2 H2

Mukhanov action
2 ˙2
Defining v ⌘ zR, z⌘ a H2 = 2a2 " and we get the action:
Z  00
1 z
S= d⌧ d 3 x (v 0 )2 + (@i v )2 + v 2 , 0
= @⌧
2 z
Juan P. Beltrán Almeida (UAN) Inflation and the Origin of LSS December 6, 2022 16 / 20
Gauge
reparameterizations we choose the following gauge for the dynamical fields
Inflationary perturbations
ADM formalism
0, gij = a [(1 2R) ij + hij ] ,
2
@i hij = hii = 0 . (A.145)
2 2 2 i j i
ds = − N dt + g (dx + N dt)(dx + N dt), N → lapse,
n field is unperturbed and all scalar degrees of freedom are parameterized
ij i j N → shift .
μ
n R(t, x).1 4 Geometrically, R measures the spatial curvature of u
constant-
r R/a 2. ∫An important (
(3) −1 ij 2
2 S = 2 d x −g NR − 2NV + N (E E
property of R ijis that it− E )+remains constant houtside
μν
s us in Lecture −1 2 to· restrict our computation to correlation functions at
+N (ϕ − N i∂iϕ)2 − Ng ij∂iϕ∂jϕ − 2V) .
1 ·
Eij ≡ (gij − ∇i Nj − ∇j Ni), E = Eii .
Equations 2
−1 Extrinsic curvature
Kij = N Efollowing
) implies ij .
the constraint equations for the Lagrange multipliers N
of a section

ri [N 1
(Eji i
j E)] = 0, (A.146)
Hamiltonian
2 ˙2
R(3) 2V N (Eij E ij
2
E2) N = 0. constraints (A.147)

int Equations) Derive the constraint equations (A.146) and (A.147) from
 ies r↵,i [N and
(Ej ÑjiE)] then
= 0 ,admit expansions in powers of (A.146)
R,
ToTosolve the
2V constraints,
R(3)solve N 2
(E ij E we
ij split
E 2 the shift
)
the constraints, we split N =⌘ .1 +
2 ˙ 2Nvector
0the N↵i into
shift irrotational
. vector (A.147) (scalar)
Ni into and incompres
irrotational (sc
ctor) parts
tities
vector)
int and
↵,parts
Equations) Inflationary
Ñi then
Derive
↵ = ↵1 +perturbations
admit equations
the constraint expansions
↵2 + . . . ,
in and
(A.146) powers (A.147) of from
R,
Ni ⌘ ,i + Ñi , where Ñi,i = 0 , (A
). ADM formalism = N 1⌘+ 2 + + Ñ. . ., , where Ñ = 0 ,
d define the lapse perturbation as↵ = i↵(1) 1 + ↵2 +
,i i
(2). . . ,
i,i
nts, First
we split order
the shift solution Ñof
vector Ni into =
the Ñ
i irrotational +
constraintÑ
(scalar) and + equations
. . . ,
incompressible
nd define the lapse perturbation = N as1⌘+1 +2↵+. . . . ,
i i (A
en
↵ = O(R
quantities n ).+The
Ni ⌘ ↵,,i Ñi , Ñconstraint
and where Ñi,i =equations (1) may
Ñ = Ñ + Ñ + . . .+
i then admit 0expansions
, in powers
N
(2)then
⌘ 1
of be set. to zero order-by-or
, ↵
(A.148)
R,
i i i
turbation as ↵ = admit
↵1 + ↵expansions
he quantities ↵, and Ñ then 2 + ... , in powers of R,
6 (First-Order
g. ↵n = O(R ).NThe
n Solution
⌘1+ of
constraint
↵. i Constraint Equations)
equations may then be Show that at first
set to zero order-by-or
(A.149)
= 1 + 2 + ... ,
Ñi then admit expansions in powers of R, ↵ = ↵
(1) (2) +, ↵2 + . . .Show ,
16 (First-Order SolutionÑiofṘ =Constraint
Ñi + Ñi + 1. . . Equations) that at (Ao
first
(1)
↵ = ↵1 + ↵2 + . . .↵, 1 = , @=Ñi 1=+0 . 2 + . . . ,
2
ere, e.g. ↵n = O(R ). The constraintH
n equations may then be set to zero order-by-order.
= 1 + 2 + ... ,
Ṙ Ñ = (1)
Ñ
(1)
+ Ñ
(2)(1)
+ . . . ,
propriate
ercise 16 Ñ choice of
(First-Order
i = Ñ
(1)
boundary
+
(2)
Solution
Ñ + ↵
. . .
1 ,=conditions
of , i
Constraint @ 2
one Ñ may =
Equations)
i i 0set
. Ñ
(A.150)
i Show
i ⌘that0.atShow thatEq
first order
i i
H
plies
6) implies
here,
The e.g.
constraint
↵ n =
equations
O(R n ). then
may The
appropriate choice of boundary conditions
beconstraint equations
set to zero order-by-order.
ṘR 2 one may set mayÑ then
(1)
⌘ beShow
0. set tothatzeroa
↵1 = , a@ Ñi =2 0 .
2 (1) i (A
der Solution
146) implies of Constraint Equations) 1 = HShow + that ✏atv first
@ order Ṙ , Eqn. (A.147)
xercise 16 (First-Order Solution H Hof2 Constraint (1) Equations) Show
th an appropriate choice of boundary conditions R aone may2 set Ñi ⌘ 0. Show that at first
s defined via Ṙ (@ 2) = . 1 = + ✏v @ Ṙ ,(A.151)
2 2
mplies 1 =
n. (A.146) implies
↵ @ , @ Ñ
(1)
i = 0 . H H
H 2 Ṙ
2 is defined via @ 2 (@ 2 ) = . R a 2 (1)
oice of boundary conditions one may set1 Ñ =i ⌘ ↵
(1)
0. =✏v @that
+1 Show 2
Ñi = 0 .
, Ṙat, first@ order (A
H
2 2 2
145H H
 Inflationary perturbations
PS of Scalar Perturbations.
We can go to Fourier space:
Z
d 3x i ~k·~x
v (⌧, x) = 3
vk (⌧ )e ,
(2⇡)
so, the e.o.m becomes:
✓ ◆
z 00
vk00 + k 2 vk = 0.
z
z 00 a00 2
In de Sitter space z = a = ⌧2

Solution and PS.

✓ ◆ ik⌧
✓ ◆
2 e i
vk00 + k 2
vk = 0, ) vk = p 1
⌧2 2k k⌧
The PS of the variable =a 1v is:

|v (⌧ ) 2| H 2
k
< k k0 >= (2⇡)3 (k + k 0 ) 2
= (2⇡) 3
(k + k 0
) 3
(1 + k 2 2
⌧ )
a 2k

Mukhanov action
Juan P. Beltrán Almeida (UAN) Inflation and the Origin of LSS December 6, 2022 17 / 20
 Inflationary perturbations
PS of R.
H
R= ˙ at the time of horizon crossing a(t⇤ )H(t⇤ ) = k:

H 2 H2 H 2 H2
< Rk Rk 0 >= (2⇡)3 (k + k 0 ) ⇤3 ⇤ , 2
R (k) = ⇤ ⇤
.
2k ˙ 2⇤ 2k ˙ ⇤
3 2

For slow-roll inflation:

1 V 1
2
s (k) ⇡ 2 2 ⇤
, ns 1 = 2⌘v⇤ 6✏⇤v .
24⇡ MPl ✏v

Nearly scale invariant spectrum!

We can do the same for tensor perturbations and obtain:
2 V
2
t (k) ⇡ 2 2 , nt = 2⌘v⇤ .
3⇡ MPl
2 (k)
Additionally, rt ⌘ t
2 (k) = 16✏⇤v = 8nt .
2

Juan P. Beltrán Almeida (UAN) Inflation and the Origin of LSS December 6, 2022 18 / 20
 Inflationary perturbations
Planck 2018 results for r and ns
 Final remarks
1. Inflation is a theoretical proposal that solves several
problems of the unusual properties of the early universe
2. Inflationary perturbations can be calculated at linear
regime for super horizon scales limit and its result gives us
information about the statistical distribution of observed
temperature fluctuations at the CMB.
3. There are several approaches and techniques to evaluate the
evolution of cosmological perturbations.
4. Cosmological perturbation theory is also used at different
scales from those involved in inflation, for instance at
large scales during cold dark matter dominated epoch.