6/8/16

Introduction to Cosmology
Part 4

Professor Barbara Ryden

Department of Astronomy
The Ohio State University

ICTP Summer School on Cosmology

2016 June 8
2dF Galaxy Redshift Survey:
~105 galaxies

There are a few unsatisfactory aspects of

the standard Hot Big Bang model; these
led to the concept of cosmic inflation.

Flatness problem: Space is nearly flat today,

and was even flatter in the past.

Horizon problem: The universe is nearly

homogeneous on scales that are not causally
connected in the standard Big Bang Model.

1  
 6/8/16  

Combining SNIa, CMB, and .8

baryon acoustic oscillations,

1− Ω 0 ≤ 0.005 .75
You are
Friedmann equation tells us here.

ΩΛ,0
2 .7
⎛ c / H (t) ⎞
1− Ω(t) = ⎜
⎝ a(t)R0 ⎟⎠ .65

When radiation & matter are

.6
dominant, |1-Ω| increases with time. .2 .25 .3 .35 .4
Ωm,0

deuterium Planck
1− Ω nuc ≤ 10 −15 1− Ω P ≤ 2 × 10 −62
synthesis: time:

Horizon problem: consider looking

out at the last scattering surface.

In the standard Hot Big Bang model

(no inflation), particle horizon distance
at the time of last scattering was:

dt tls
dhor (tls ) = a(tls )c ∫
0 a(t)

= 2.24ctls = 0.25 Mpc

2  
 6/8/16  

The angular-diameter distance to

the last scattering surface is
d A ≈ 10 Mpc

The observed angular size of a

patch dhor = 0.25 Mpc across is
dhor 0.25 Mpc
θ hor = ≈ ≈ 0.025 rad ≈ 1.4 !
dA 10 Mpc

Without inflation, points more than ~1.4° apart

were outside each others’ horizon.
But... we see they had the same temperature to within ~10-5.

Inflation: during the very early universe,

ti ~ tGUT ~
there was a temporary era when ä > 0.
10-36 sec ?

Toy model: Exponential expansion began at time ti,

with Hubble constant Hi ~ (Λi/3)1/2 ~ ti-1, and ended at tf >ti.
The energy density of Λi is then transfered to relativistic
particles in a “reheating” process.

(Physical mechanisms for inflation

will be discussed next week...)

3  
 6/8/16  

How does inflation solve the flatness problem?

2
⎛ c / H (t) ⎞
1− Ω(t) = ⎜
⎝ a(t)R0 ⎟⎠
During exponential inflation,
2
⎛ c / Hi ⎞
1− Ω(t) = ⎜ H i (t−ti ) ⎟ ∝ e−2 H it
⎝ ai e R0 ⎠

If the universe had |1-Ω|~1 before inflation,

and if inflation started at ti ~ 10-36 sec, then
N>60 e-foldings are needed to match today’s flatness.

How does inflation solve the horizon problem?

During exponential inflation, 20
on
f inflati
the particle horizon size 10
foldin
gs o
65 e-
log(dhor) [meters]

grows exponentially. n
atio
0
f l
in
dhor (ti ) ~ 2cti ~ 6 × 10 −28 m no
−10

−20

dhor (t f ) ~ e N 2cti [> 7 cm if N >60] −30

−40 −30 −20 −10 0

log(t) [seconds]
a(tls )
dhor (tls ) = dhor (t f ) [> 1Mpc if N >60]
a(t f )
N>63 e-foldings ensure the last scattering surface is isotropic.

4  
 6/8/16  

Inflation, by increasing the particle horizon size,

prevents the CMB from having large
temperature fluctuations (δT/T ~ 1).

Inflation, by inflating quantum perturbations to

macroscopic scales, also causes the observed
small temperature fluctuations (δT/T ~ 10-5).

Quantum perturbations in the “inflaton” field

è small variations δN in the e-foldings of inflation
è slight differences in the time of reheating
è small fluctuations in the post-inflation density ε

When dark matter decouples from other

components of the universe (t ~ 1 sec for WIMPs),
it has low-amplitude density fluctuations:
! !
ρ (r,t) = ρ (t)[1+ δ (r,t)]
spatially averaged density fluctuation:
initially |δ| << 1
We expect that the density field δ resulting from
inflation will be a Gaussian random field.
⎛ δ2 ⎞
p(δ ) ∝ exp ⎜ − 2 ⎟
⎝ 2σ ⎠

5  
 6/8/16  

Take the Fourier transform of δ:

1 ! !! comoving
δ k! = ∫ δ (r ) e d r
ik⋅r 3
box
V
V (t) ∝ a(t)3
Each Fourier component can be written as
iϕ k!
δ k! = δ k! e
If δ(r) is a Gaussian random field, then the phases φk
are uncorrelated, and all useful information is in the
power spectrum
2
P(k) = δ k!

Prediction: inflationary density perturbations

should have a power spectrum
P(k) ∝ k n
with n ≈ 1. (If n = 1 exactly, this is called
a Harrison-Zel’dovich spectrum.)

Observable consequences: spheres of mean mass M have

δ M / M ∝ M −(3+n)/6
80

60

δφ ∝ δ M / r ∝ M (1−n)/6
∆T [µK]

40

20

Planck: n = 0.97±0.01 0
1 10 100 1000
l

6  
 6/8/16  

The initial P ~ k0.97 spectrum is modified on small

scales during the era of radiation domination.
When the physical size of a perturbation, λ~a(t)2π/k,
is larger than the Hubble distance, c/H(t)~2ct,
its amplitude grows. Why?
Wave crests are out of contact with troughs:
crests act like a patch of an Ω>1 universe,
troughs act like a patch of an Ω<1 universe.

λ ∝ a(t) ∝ t 1/2 c / H (t) ∝ t

Eventually, λ is overtaken by c/H, and the
amplitude of the perturbation freezes out.

10000 10000
1000 1000 ∝ M −0.66
∝ k 0.97
P [arbitrary units]

100 100
10 10 CDM
δM/M

1 CDM 1
.1 .1 HDM
.01 .01
.001 HDM .001
.0001 .0001
.001 .01 .1 1 10 10 12 14 16 18 20
−1
k [Mpc ] log(M/Msun)

Suppression of power is δM/M is largest on small

greatest on small length scales length scales (“bottom up”
(large comoving wavenumber). structure formation).

7  
 6/8/16  

During the matter-dominated era, density fluctuations

in dark matter evolve by gravitational instability:
“The rich get richer, the poor get poorer.”

[Evolution of baryonic fluctuations is more complicated,

because of interactions with photons.]

Simplest case of gravitational instability:

low-amplitude density fluctuations in completely
pressureless, completely dark matter.

Applying linear perturbation theory (|δ|<<1) to the

acceleration equation, we find [for λ<c/H(t)]:
Applies even when
δ!! + 2H (t)δ! = 4π G ρ (t)δ (t) universe is not
matter-dominated.
Static universe (H = 0, constant mean density):

δ!! = 4π G ρδ (t)

Well-known solution: exponential growth (or decay)

−t /t dyn
δ (t) = A1e + A2 e
t /t dyn
,
where t dyn = (4π G ρ )−1/2

8  
 6/8/16  

Growth of density
perturbations: δ!! + 2H (t)δ! = 4π G ρ (t)δ (t)

Hubble expansion Self-gravity

Hubble expansion decreases Self-gravity increases
density on a time scale density on a time scale
1/H(t)~(Gε/c2)-1/2 tdyn~(Gρ)-1/2

If matter is not dominant, ρ<<ε/c2, tdyn>> 1/H,

and perturbations grow extremely slowly.
If matter is dominant, ρ≈ε/c2, tdyn~ 1/H, and perturbations
grow as a power law (not exponentially).

A flat, matter-dominated universe: Ωm=1, H(t) = (2/3)t-1

4 2
δ!! + δ! = 2 δ
3t 3t
Power-law solutions (verify by substitution!)
δ = C1t 2/3 + C2t −1
When only the growing mode remains,
δ ∝ t 2/3 ∝ a(t) ∝ (1+ z)−1
δ mΛ 1+ zrm 3440
Maximum growth: ≈ ≈ ≈ 2600
δ rm 1+ zmΛ 1.31

9  
 6/8/16  

A region of space with δrm < 1/2600 ~ 4×10-4 at

radiation-matter equality never reaches δ ~ 1,
the overdensity at which collapse begins.

We are entering the Λ-dominated era:

the biggest collapsing objects that
we see today (superclusters) are
the biggest there will ever be.

On small scales,
today’s universe is very
inhomogeneous.

δ~1028 δ~1022

Baryonic matter can reach

very high densities because
it can radiate away excess
thermal energy.
δ ~10 δ~0

10  
 6/8/16  

Baryonic matter today: Bound baryons

Diffuse intergalactic gas:

δbary≤0, T<105K, ionized

Warm-hot intergalactic gas:

δbary~30, T~106K, ionized

Although baryons can make very

dense objects, most baryons don’t.
The baryons are ionized again!
How did that happen?

The baryons are ionized again!

When did that happen?

Recombination occurred at trec ~ 0.25 Myr.

The time of reionization can be

deduced from looking at the CMB.

Ionized intergalactic gas provides a

foreground screen of free electrons
that can scatter CMB photons. Slightly
translucent

11  
 6/8/16  

Because of the translucent foreground

material, our view of the last scattering
surface is slightly blurred.

The most recent analysis of the Planck results

(Aghanim et al., arXiv:1605.02985) yields
τ=0.055±0.009

(That is, about 1 in 18 CMB photons has scattered.)

The rate at which a CMB photon scatters

from free electrons in reionized gas:
Γ(t) = ne (t)σ ec
If the baryonic gas is reionized starting at a time tR,
then the optical depth of the reionized gas is:
t0 t0
τ = ∫ Γ(t)dt = σ ec ∫ ne (t)dt
tR tR

Simplifying assumptions: pure hydrogen

undergoes instant total reionization at t=tR.

12  
 6/8/16  

With these assumptions, at t > tR we find

nbary,0
ne = n p =
a(t)3
The optical depth is then
t0 dt
τ = Γ0 ∫
tR a(t)3
where Γ 0 = σ ecnbary,0 = 1.6 × 10 −4 Gyr −1 ≈ 0.002H 0
There’s an analytic solution!
τ=
2 Γ0
( 1/2
⎡ Ω m,0 (1+ zR ) + Ω Λ,0 ⎤⎦ − 1
3Ω m,0 H 0 ⎣
3
)

Using H0 = 68 km s-1 Mpc-1, ΩΛ,0=0.69, Ωm,0=0.31, and

τ=0.055±0.009,

we find that reionization took place at a redshift

zR = 6.9 ± 0.8 (tR ≈ 0.65 Gyr)

The “era of neutrality” was a brief interlude in

the history of the universe: tR-trec~0.05t0

So... what happened around z~7 that

could have reionized the universe?

13  
 6/8/16  

Hint: the highest redshift galaxies

known are at z > 7.

z = 8.68; a(te) = 1/(1+z) = 0.103

Galaxies contain two sources of ionizing photons:

hot stars and active galactic nuclei (AGN).

Intergalactic gas was photoionized.

(comoving) number density (comoving) star

of quasars (= luminous AGN) formation rate
6
10
dρ*/dt [MsunMyr Mpc ]
−3
−1

5
10

4
10

0 2 4 6 8
z
Probably too few AGN Probably (maybe?)
at z>7 to do the job. enough hot stars.

14  
 6/8/16  

Reionization: much work remains to be done.

Cosmology in general: much work remains to be done.

15