afh lecture 2 -...
TRANSCRIPT
ALAN HEAVENS I N S T I T U T E F O R A S T R O N O M Y U N I V E R S I T Y O F E D I N B U R G H
A F H @ R O E . A C . U K T H I R D T R R 3 3 W I N T E R S C H O O L
P A S S O D E L T O N A L E ( I T A L Y ) 6 - 1 1 D E C E M B E R 2009
Overview Lecture 2 Theory for Observers, Observations for
Theorists
Causality issue
Big Bang theory has a number of puzzles, including the uniformity of the microwave background
Temperature is uniform to 10-5, but opposite sides of sky (at z=1000) should not have been in causal contact (size of particle horizon ~2ct; t=300,000 years; DA=14 Mpc, angular size of horizon ~1 degree)
Inflation
Inflation solves the problem of causality, by having a period of exponential accelerated expansion in the Early Universe.
With exponential expansion, the comoving Hubble radius c (aH)-1 decreases with time.
This is roughly the comoving distance a photon travels as the scale factor increases by e:
Hubble Radius
Comoving description:
With exponential expansion early on, photons can travel large comoving distances, solving the causality problem
Needs >60 e-foldings A decreasing Hubble radius
requires acceleration: This early inflation (t~10-35 s) has vastly higher energy density than current vacuum energy.
Inflation by scalar fields
Scalar fields ϕ(x) If quanta are massive, then
with
gives the Klein-Gordon equation:
Neglecting spatial derivatives
Scalar fields in cosmology
Scalar field characterised by an action
with equations of motion determined by δS=0. Technical point: S must be a scalar. d4x is not. Extra
term involving g=det(gµν) makes d4x √(-g) a scalar; thus requiring L to be a scalar.
Euler-Lagrange equations:
To get the Klein-Gordon equation, we need a Lagrangian given by: L =
12∂µφ ∂µφ− m2c4
22φ2
=Newton’s law
S =
dt(T − V )
Lagrangian and conserved current
L does not depend explicitly on xµ. This symmetry leads to conserved quantity (Noether current)
Generalise V so that This gives Now compare with a perfect fluid:
Gives
Equation of state
Equation of state of a homogeneous scalar field is
So we get inflation if
Equations of motion: expanding coords
The Euler-Lagrange equations give
Approximations: homogeneous, and ‘slow-roll’:
These give
Friedmann’s equation gives
Inflation dynamics
Slow-roll conditions:
Inflation ends when ϕ is near bottom, and slow-roll conditions are not satisfied.
Universe reheats to high temperature
≡ mpl
16π
V
V
2
1; η ≡m2
pl
8π
V
V; |η| 1
Ending inflation
Time derivatives become more important as field approaches the minimum
Rather than slow-rolling, the field oscillates, damped by the friction term
Coupling with matter fields produces particles, reheating the Universe
3Hφ
Fluctuations from Inflation (heuristic)
ϕ fluctuates because of quantum effects, with an r.m.s. (from QFT: sea Peacock)
Parts of the Universe with different ϕ exit inflation at different times, differing by
Hence there is a fractional variation in energy densities on the Hubble radius scale of
While V ≈ constant, H ≈ constant, so horizon-scale perturbations are roughly constant. This gives scale-invariant fluctuations, δH
2 ≈ constant. Power spectrum of density fluctuations: P (k) ∝ kn; n 1
Spectrum of fluctuations
More carefully, variations in V give a tilt. Fluctuations of a certain wavenumber k cross the Hubble radius when k-1 = 1/(aH). H ≈ constant, so
writing
gives the tilt of the spectrum as
Typically n<1.
d ln(δ2H
)d ln k
= ad ln(δ2
H)
da=
1H
d ln(δ2H
)dt
=φ
H
d ln(δ2H
)dφ
Number of e-foldings
In order to account for the observed uniformity in the Universe, there need to be at least 60 e-foldings
Flat potentials will generally inflate. N depends on slow-roll parameters Leads to a consistency relation
between N, n and tensor modes r in CMB
N =
Hdt = − 8π
m2pl
φfinal
φinit
dφV
V
Komatsu et al 2006
Perturbations
Perturbed FRW metric, in Newtonian gauge (observers follow unperturbed paths)
η is conformal time. Φ is potential perturbation. Ψ is curvature perturbation. These are scalar perturbations (vector [vorticity] and tensor [gravitational waves] modes also possible).
For fluid-like behaviour (no anisotropic stress), Φ=Ψ If v << c, only Φ matters.
ds2 = a2(η)(1 + 2Φ)dη2 − (1− 2Ψ)dxidxi
η ≡
dt
a(t)|Φ| 1, |Ψ| 1
Newtonian perturbation theory
x= proper coordinate; r = comoving coordinate
Continuity equation
Change to comoving coordinates:
x = a(t)rx = ar + arx = ar + 2ar + ar
∂ρ
∂t
x
+ ρ∇x · x = 0
∇x =1a∇;
∂
∂t
x
=∂
∂t+ u ·∇; u ≡ r
Zero-pressure perturbations
Let the overdensity δ be defined by
Then continuity becomes
Zero-order terms give And linear terms (in δ or u) give Equation of motion gives Take divergence, add the Poisson equation
which gives Finally
ρ(r, t) = ρ0(t) [1 + δ(r, t)]
ρ0 ∝ a−3
δ +∇ · u = 0 . ≡ ∂/∂t
∇2xΦ = 4πGρ
x = −∇xΦ ar + 2au + au = −∇a
(Φ + δΦ)
δ + 2H δ − 4πGρ0δ = 0∇2(Φ + δΦ) = 4πGρ0a
2(1 + δ)
∂[ρ0(1 + δ)]∂t
+ u ·∇[ρ0(1 + δ)] + ρ0(1 + δ)∇ · (r + au) = 0
(GR)
Solutions
Flat, matter-dominated (Einstein-de Sitter) Universe: so H=2/3t, and
Trial solution δ α tp gives p=2/3 or -1. Growing solution is then
In general, in GR, (with a very small dependence on Λ)
a ∝ t2/3 8πGρ0/3 = H2
δ +23t
δ − 23t2
δ = 0
δ ∝ t2/3
d ln δ
d ln a≡ f Ω0.56
m
Complications
We cannot assume Ωm ≈ 1 and ignore other components. Growth is slower.
In the early Universe, radiation is important, and growth rate is δ α a2 on super-Hubble radius scales. On smaller scales, growth is suppressed (Meszaros effect: perturbations are driven by Dark Matter, which is subdominant; expansion timescale is set by radiation, and it is short, because ρrad is large)
Photon-electron coupling prevents sub-Hubble radius growth of baryons (sound speed ≈ c/√3)
- they oscillate.
Growth
Baryon and photon amplitude depends on phase at recombination, which depends on the wavenumber
Transfer Function
T(k) defined so the linear power spectrum is
Super-Hubble radius modes grow like a2 before equality.
Modes which enter Hubble radius before equality don’t grow, so are suppressed by T=(aentry/aeq)2. Since k=aH on entry and H α 1/a2 (since a α t1/2), we find
P (k) ∝ T 2(k)Pinitial(k)
T (k) ∝ k−2 k →∞
Net effect
Linear power spectrum
Neutrino streaming
Neutrino oscillation experiments give
So at least some neutrinos have non-zero mass. Since they interact only weakly/gravitationally, they can leave small perturbations and partially erase them.
kT = 1.7 x 10-4 eV now
∆m212 = 7.9+1.0
−0.8 × 10−5eV2
|∆m223| = 2.2+1.1
−0.8 × 10−3eV2
Error on sum with Planck and Euclid weak lensing: 0.037eV (de Bernardis et al 2009)
Neutrino free-streaming length
Consider flat Universe and radially-moving neutrino.
Hence Geodesic equation gives Substitution gives
When the neutrinos are relativistic and the Universe radiation-dominated, r α t1/2. At late times, when the neutrinos are non-relativistic dr/dt α t-4/3, so r converges. Free-streaming length is 112 (mν/eV)-1 Mpc (if masses equal)
ds2 = c2dτ2 = c2dt2 −R2(t)dr2
1 = t2 − R2
c2r2 . = d/dτ
R2r = Ac = constantdr
dt=
r
t=
Ac
R2
1 + R2r2
c2
=Ac
R2
1 + A2
R2
Neutrino effects
Suppression on small scales
Note:
is the contribution to the variance of the matter from unit ln(k)
∆2(k) ≡ 12π2
k3P (k)
Present-day power spectrum
Nonlinear effects add extra power on small scales
Smith et al. (2003) fitting formula
Observations
2dF, SDSS Baryon oscillations seen Turnover not yet
definitively seen This is the galaxy power
spectrum A bias is often assumed:
Quadratic measures such as the power spectrum scale as
δn
n= b
δρ
ρ
Pgal(k) = b2P (k)
Bias?
Galaxies are not necessarily where the mass is
On large scales, detailed statistical analysis shows galaxies and mass DO follow the same distribution (Verde et al 2002; Seljak et al 2005)
Peculiar velocities
Deviations from Hubble flow depend on matter fluctuations. With an independent distance estimator, can be measured
The variance in this, in patches of size R, is
where W is the window function for the patch. Since we measure Pgal (k), peculiar velocities usually
tell us about
u
u2 =
H20f
2
2π2
d3kP (k)W 2(kR)
β ≡ f(ΩM )b
Redshift Distortions
More effective is to detect peculiar velocities by their effect on the redshift of galaxies
At small distances,
Radial distortions distort the correlation function, or power spectrum, so that it is not isotropic in ‘redshift space’.
Measures β Can get bias-independent measure:
cz = Hx + au
Guzzo et al 2005
f2(Ωm)P (k) = β2Pgal(k)
Taylor et al 2001