Physical Cosmology 6/5/2016
Alessandro Melchiorri
alessandro.melchiorri@roma1.infn.it slides can be found here:
oberon.roma1.infn.it/alessandro/cosmo2016
Today lecture is also slightly based on:
Structure formation
We are moving away from an homogeneous and isotropic universe.
We have introduced the density contrast:
“Mean”density.
Average is over Volume
And we assumed it as very small (linear perturbation theory):
Structure formation
If we now consider a pressure less fluid (w=0), with a very qualitative treatment, we have found that in an
expanding universe:
Remember: the density contrast is in the matter (w=0) component, but the expansion of the universe can be dominated by a different energy component !
Structure Formation in Expanding Universe
A key point is that the matter component that is collapsing could be different from the dominant energy component.
Introducing the time dependent density parameter:
We can write:
Structure Formation in Expanding Universe
Let us first consider the radiation here. During this epoch:
The equation can be therefore written as:
with solution:
during radiation dominated epoch the perturbations in matter grow only logarithmically !!!
Structure Formation in Expanding Universe
If we move to matter dominated we have:
Structure Formation in Expanding Universe
The equation is therefore:
The solution is a power law:
During matter dominated, matter fluctuations grow with a power of 2/3 (less than in a static universe).
We discard the 1/t solution since we expect a more homogeneous
Universe in the past
Structure Formation in Expanding Universe
Finally, in an epoch dominated by a cosmological constant:
The equation is:
And the solution:
If Lambda dominates the expansion, then the fluctuations in the matter component remain constant.
They are, in practice, frozen.
First Summary
If we consider perturbations in a pressureless matter component (Jeans length always zero) their growth depend on which kind of energy component is
dominating the expansion.
We have substantial growth only
here !
Moving to Fourier Space
In order to move to a more physical description let us consider an expansion in Fourier modes of the density contrast field:
Each Fourier mode is given by:
Each Fourier component is a complex number, which can be written in the form:
It is possible to show that assuming a linear perturbation theory, i.e. :
And following a more formal approach, one gets, in the Newtonian regime, for a pressure less fluid:
Moving to Fourier Space
Moving to Fourier Space
This is the same equation we got for the density contrast of a sphere of radius R, but now it applies to each
Fourier mode of a generic density contrast !
The fact that we assumed linear perturbation theory implies that the evolution of each Fourier mode is
independent from the other, i.e. we don’t have
density contrasts of different modes in the equation and their time evolution does not mix.
Pressure term
The previous equation holds for a pressure-less fluid (w=0). If we consider pressure, it is possible to show that the equation modifies to:
New term due to fluid pressure.
k is in comoving coordinates, so at each wavenumber k corresponds a physical scale at time t of:
From the above equation we identify the Jeans wavelength:
Each Fourier mode will therefore evolve with time in a different way if the corresponding k is larger or
smaller than the Jeans wavelength.
If
We can neglect this.
and we have the growth as discussed for a fluid with w=0.
Pressure term
If, on the contrary, we are below the Jeans lenght:
We neglect the first term and the solution is given by a more complicate oscillation term damped in time.
Pressure term
Horizon Scale
Given a time t, the quantity:
provides a causal horizon, i.e. particles that are a distances larger than it are not causally connected.
What happens if I consider a perturbation on scales larger than the horizon scale ?
We can treat them only using general relativity.
As we will see, the solution is also gauge-dependent.
Horizon Scale Horizon Scale
Assuming a synchronous gauge, it is possible to show that perturbations on scales larger than the horizon, i.e. such that at a given time t have:
they always grow, as:
Summary (single fluid)
We have therefore two important scales for structure formation: the horizon scale and the Jeans scale.
For cold dark matter (w=0) what is important is the
horizon scale at equivalence. Perturbations that enter the horizon before the epoch of equivalence are damped
respect to perturbations that enter the horizon later.
For CDM the Jeans scale is always zero.
For baryons, the Jeans length is approximately the
horizon scale until decoupling. The crucial scale is the horizon scale at decoupling. After decoupling baryons have w=0 (approximately).
Perturbations that enter the horizon before decoupling (z=1100) are strongly damped respect to perturbations that enter the horizon later.
For example, we can consider two modes, one entering the horizon before the matter-radiation equivalence and another one entering after it.
log δ
a log
Perturbation in Red: enters the horizon AFTER the equivalence.
Perturbation in Blue: enters the horizon BEFORE equivalence.
aEQ
a2
k1
k2
) (
/ 2 1 2
2 a c H a
k = −
1
2
k
k >
a1
Evolution for a w=0 (no pressure)
component.
Perturbations with k larger
than
are damped respect to perturbations with k smaller Cold dark
matter (only)
For example, we can consider two modes, one entering the horizon before the decoupling and another one entering after it.
log δ
a log
Perturbation in Red: enters the horizon AFTER the decoupling.
Perturbation in Blue: enters the horizon BEFORE decoupling.
a2
k1
k2
) (
/ 2 1 2
2 a c H a
k = −
1
2
k
k >
a1
Evolution
for the baryon component.
Perturbations with k larger
than
are strongly damped respect to perturbations with k smaller Baryons
(only)
We ar
e below the Jens lenght.
Damping+Oscillations
log δ
a log
Baryon/CDM Perturbation in Red: enters the horizon AFTER the decoupling.
Baryon Perturbation in Blue: enters the horizon BEFORE equivalence.
CDM Perturbation in Green: enters the horizon BEFORE equivalence.
a2
k1
) (
/ 2 1 2
2 a c H a
k = −
a1
CDM
+Baryons
aEQ
The situation is different if we consider a CDM+Baryon case.
Baryons “feel" the CDM gravitational potential.
Baryons,
after decoupling fall in the CDM potential wells.
k2
k2
Cosmological «Circuit»
Generator of Perturbations (Inflation)
Amplifier (Gravity)
Low band pass filter.
Cosmological and
Astrophysical effects
Tend to erase small scale (large k) perturbations
Power Spectrum
Each Fourier component is a complex number, which can be written in the form
The mean square amplitude of the Fourier components defines the power spectrum :
where the average is taken over all possible orientations of the wavenumber. If δ(⃗r) is isotropic, then no information is lost,
statistically speaking, if we average the power spectrum over all angles and we get an isotropic power spectrum:
Correlation function
Let us consider the autocorrelation function of the density field (usually called the correlation function):
Where the brackets indicates an average over a volume V.
We can write:
and, performing the integral we have:
Correlation function
Since the correlation function is a real number, assuming an isotropic power spectrum we have:
If the density field is gaussian, we have that the value of δ at a randomly selected point is drawn from the Gaussian
probability distribution:
where the standard deviation σ can be computed from the power spectrum:
Summary
In practice, our theory cannot predict the exact value of
in a region of the sky. But if we assume that the initial perturbations are gaussian we can predict the correlation function, the variance of the fluctuations and its the power spectrum P(k).
These are things that we can measure using, for example, galaxy surveys and assuming that galaxies trace the
CDM distribution.
Cosmological «Circuit»
Generator of Perturbations (Inflation)
Amplifier (Gravity)
Low band pass filter.
Cosmological and
Astrophysical effects
Tend to erase small scale (large k) perturbations
Power spectrum for CDM
The analysis we have presented up to now is very qualitative and approximated (just to have an idea…)
The true power spectrum for CDM density fluctuations can be computed by integrating a system of differential equations.
We will see this in better detail in the next lectures.
In any case, we assume a power law as initial power spectrum as:
The motivation of using this type of primordial spectrum comes from inflation (we will discuss this also in a future lecture). We have two free parameters: the amplitude A and the spectral index ns.
Power spectrum for CDM
The first numerical results on the CDM power spectrum appeared around 1980.
Bardeen, Bond, Kaiser and Szalay (BBKS) in 1986 proposed the following fitting function:
where: This is correct fit for a pure CDM model (no baryons or massive neutrinos).
Note the dependence on
that defines the epoch of equivalence.
P(k) for LCDM (from numerical computations).
Spectral index is assumed
n=0.96
Note these oscillations in the
CDM P(k).
Gravitational feedback from
baryons.
The position of the peak is related to size of the
horizon at equivalence, i.e.
to the matter density since radiation is fixed.
Primor dial r
egime
Damping (scales that enter
ed horizon befor
e quality)
Effect of the Cosmological parameters
Cold dark matter
Effect of the Cosmological parameters
Baryon density
Effect of the Cosmological parameters
Spectral index
CDM vs data (1996)
If we assume a flat universe made just of matter with
we get too much power on small scales
to match observations.
Already in 1996 the best fit was for
i.e. suggesting a low matter density universe.
CDM vs data (2003)
The 2dfGRS provided
the following constraints:
(assuming just CDM, no massive neutrinos)
2 sigma detection of baryons:
Measurements from SDSS
Mass fluctuations
Given a theoretical model a quantity can be often easily compared with observations is the
variance of fluctuations on a sphere of R Mpc:
where:
Usually one assumes R=8 Mpc hˆ-1, where the linear approximation is valid.
Examples from CAMB
http://lambda.gsfc.nasa.gov/toolbox/tb_camb_form.cfm