Massachusetts Institute of Technology

Department of Physics

Physics 8.942 Fall 2001

Problem Set #1
Due in class Tuesday, September 18, 2001.

1. Conserved Momenta
Show the following: If ∂α gµν = 0 for all {µ, ν}, then Pα is conserved along a geodesic
xµ (λ), where Pα ≡ gαβ dxβ /dλ.

2. Robertson­Walker Metric
Consider the general Robertson­Walker metric, written in the form
� �
2 2 dr2 2
ds = −dt + a (t) + r2 (dθ2 + sin2 θ dφ2 ) . (i)
1 − kr2

Note that for k > 0 the complete spacetime has two copies of the domain 0√≤ r ≤ k −1/2 ,
just as a unit sphere has two copies of the cylindrical coordinate range 0 ≤ x2 + y 2 ≤ 1
(the northern and southern hemispheres).
Find coordinate transformations that will put the line element in the following forms:
� �
ds2 = a2 (τ ) −dτ 2 + dχ2 + r2 (χ)(dθ2 + sin2 θ dφ2 ) , (ii)
⎡ ⎤
2 2 2 2 2
⎢ dr̄ + r̄ (dθ + sin θ dφ ) ⎥
ds2 = a2 (τ ) ⎣−dτ 2 + � �2 ⎦ , (iii)
1 + 14 kr̄2
� �−2
2 2 2k� 2 � � �
ds = −dt + a (t) 1 + x + y2 + z2 dx2 + dy 2 + dz 2 . (iv)
4
For each case, indicate the full range of the variables. Give explicit formulae for r(χ)
and r̄(r). (Hint: Different forms may be required for k > 0, k < 0, and k = 0. Note also
that a(τ ) is not the same function of its argument as a(t).)

3. Cosmological and Doppler Redshifts

Consider an object with radial coordinate χe in a Robertson­Walker spacetime (using the
form ii given in Problem 2). The object emits a burst of nearly monochromatic radiation
at time τe with wavelength λe in its own rest frame. A fundamental (comoving) observer
is at χ = 0 with 4­velocity V�o = a−1�eτ . The observer detects the radiation at time τo
with wavelength λo . The redshift is defined as z ≡ (λo /λe ) − 1.

1
 a) Assume that the emitter is comoving (i.e. at fixed spatial coordinates) so
that its four­velocity is V�e = a−1
e �eτ where ae ≡ a(τe ). Evaluate the redshift
in terms of the expansion scale factor a(τ ).

b) Now suppose that the emitter is no longer comoving. Instead, it has a radial
“peculiar” velocity component vr , which is the radial three­velocity measured
by a comoving observer at χe . (In other words, vr is the radial velocity
component in an orthonormal basis fixed at χe .) What are the emitter’s
four­velocity components Veτ and Veχ in terms of vr and ae ? Show that your
result makes sense in the non­cosmological limit a(τ ) = constant. (Do not
assume vr2 � 1.)

c) Continuing part b), what is the object’s redshift as seen by the observer?
Show that if a(τ ) = constant, you recover the radial Doppler shift formula of
special relativity while if vr = 0 you recover part a).

d) Now suppose that the emitter also has a tangential velocity (relative to the
comoving frame) with orthonormal components vθ and vφ , i.e. the peculiar
velocity has arbitrary direction. Show that the redshift is given by
ao 1 + vr
1+z = √ .
ae 1 − v 2

e) (Bonus challenge): Suppose that the observer at χ = 0 is no longer comoving

but has a three­velocity v o relative to the comoving frame. How is 1 + z
modified from part d)?

4. An Empty Universe
For a k = −1 Robertson­Walker spacetime with ρ = p = 0 show from the Friedmann
and energy conservation equations that the line element becomes
� � ��
ds2 = −dt2 + t2 dχ2 + sinh2 χ dθ2 + sin2 θ dφ2 .

Find an explicit coordinate transformation to show that this metric describes Minkowski
spacetime.

2
 Massachusetts Institute of Technology
Department of Physics

Physics 8.942 Fall 2001

Problem Set #2
Due in class Tuesday, September 25, 2001.

1. Cosmological Redshift as a Doppler Shift

For nearby objects with recession speed Hr � c, the first-order Doppler shift gives
λo /λe = 1 + Hr/c. To get the redshift factor R ≡ λo /λe for a distant object, we can
break the light path into many small segments such that HΔr � c for each, where Δr is
the proper length of a segment. Multiplying together the redshift factors for all segments
gives the net redshift factor R = R1 R2 · · · Rn .

a) Using the first-order Doppler shift, show that for one segment, lasting in time
from t to t + dt, the redshift factor is
� �
1 da
Rt = exp dt . (1)
a dt

b) Multiply the results from all segments to obtain the usual cosmological red­
shift formula λo /λe = ao /ae (where usually one sets ao = 1).

2. Cosmic Time-Redshift-Distance Relation

Newspaper articles presenting cosmological discoveries of distant objects usually do not
quote redshifts, but instead give the time since the big bang. One such account reported
that a galaxy has been found to be fully formed a mere 1 billion years after the big bang.
Suppose that you are asked by an astronomer friend what redshift was measured for the
galaxy. Additionally, a non-astronomer friend asks you the distance to the galaxy in
light-years. After explaining that distance is ambiguous, your friend says alright, just
tell me the comoving distance in light-years.
Estimate the redshift and the comoving distance using the following three cosmolog­
ical models. In each case, assume that the Hubble constant is 72 km s−1 Mpc−1 .

a) SCDM: flat Ωcdm = 1, ΩΛ = 0;

b) OCDM: open Ωcdm = 0.35, ΩΛ = 0;

c) ΛCDM: flat Ωcdm = 0.35, ΩΛ = 0.65.

1
Note: you may have to perform a numerical integral. The 8.942 webpage has links to
information on using Mathematica, Matlab, Maple, as well as a function (C and F77) to
perform numerical integration.

3. Angular Size vs. Redshift

An object of linear size L (perpendicular to the line of sight) subtends a small angle
α. Suppose that the object and the observer are both comoving and that the redshift
of the object is z. For a fixed linear size the angle is a function of redshift: α = α(z).
For α2 � 1, α ∝ L. Find the constant of proportionality in terms of τe and τo (the
conformal times when light is emitted by the object and received by the observer). For
an Einstein-de Sitter universe (k = 0 and a ∝ t2/3 ), α(z)/L has a minimum at a certain
redshift. What is that redshift? If galaxies formed long ago and all have about the same
typical size, say 5 kpc, what is the minimum angular size of a galaxy in an Einstein-de
Sitter universe, in arcseconds? Assume a Hubble constant of 72 km s−1 Mpc−1 .

4. Number Counts
Suppose that there exists a population of luminous sources distributed homogeneously
throughout space. For simplicity, assume that all sources have identical proper bolomet­
ric luminosity L and that they emit isotropically. (If the sources have a distribution of
luminosities we could integrate over that later.) The comoving number density of these
sources is n(t), i.e., the proper number density at cosmic time t is (a0 /a)3 n(t) where
a(t) is the expansion scale factor and a = a0 today. A terrestrial astronomer surveys the
entire sky and finds N sources with measured bolometric flux greater than S. Assuming
that we live in a Robertson-Walker universe, find N (S). Evaluate it fully in the sim­
ple case n(t) = constant and an Einstein-de Sitter universe (k = 0 matter-dominated).
Show that as S → ∞, N (S) approximates the Euclidean result N = (4πn/3)(L/4πS) 3/2 .
Why? In the opposite limit, S → 0, what is the behavior of N (S), and why?

2
 Massachusetts Institute of Technology
Department of Physics

Physics 8.942 Fall 2001

Problem Set #3
Due in class Tuesday, October 2, 2001.

1. Horizons
The particle (or causal) horizon distance lH (t) is defined to be the coordinate distance
χ travelled by a photon on a radial geodesic since t = 0, using the Robertson­Walker
metric with line element in the form
� �
ds2 = −dt2 + a2 (t) dχ2 + r2 (χ)(dθ2 + sin2 θdφ2 ) . (1)

For simplicity consider the k = 0 case so that r(χ) = χ. Find lH (t) for the following
cases: (1) a(t) ∝ t2/3 (Einstein­de Sitter, matter dominated universe); (2) a(t) ∝ t1/2
(radiation dominated universe); and (3) a(t) ∝ exp(Ht) with H =constant (vacuum
energy dominated universe, known also as the de Sitter universe). Show that in cases (1)
and (2), a photon emitted at t = t0 can reach any coordinate distance given sufficient
time but not so in the de Sitter universe. This result shows that the de Sitter universe
has an event horizon. Draw a spacetime diagram to explain what is going on.

2. Angle subtended by causal horizon

Consider two comoving observers A and B, each at the same comoving distance χ from
you, the observer. At conformal time τe , A and B come into causal contact for the first
time: a photon emitted by A at τ = 0 reaches B (and vice versa) at τ = τe . At τ = τe
both A and B emit photons along radial paths toward you which you receive today at
τ = τ0 . For simplicity, ignore the gravitational effect of radiation in the universe; this is
a fair approximation after z = 104 .
a) Assuming that you live in an Einstein­de Sitter universe, what angular separa­
tion do you measure between A and B? (You may use small­angle formulae.)
b) Evaluate the angle in degrees if 1 + ze = 1100 so that A and B are on
the cosmic photosphere at recombination (when the cosmic plasma became
transparent). Without an earlier inflationary era, places on the photosphere
separated by more than this amount are causally disconnected.
c) Redo part b) for the OCDM and ΛCDM models of Problem Set 2. For the
same Ωm in both models, does the angle subtended by the causal horizon at
recombination depend more strongly on Ωk or ΩΛ ?

1
3. Power­Law Inflation: Peacock 11.1
Verify that the potential �� �
16π
V (φ) ∝ exp φ
pm2Pl
leads to inflation with a scale­factor dependence a(t) ∝ tp . (Here mPl = G−1/2 is the
Planck mass in units where c = h ¯ = 1.) Since this is not de Sitter space, why does the
resulting behavior still lead to zero comoving spatial curvature?

4. The Future of Inflation

a) Show that inflation can keep the observed universe homogeneous only for a
time t < tcrit where
t2
tcrit ∼ i exp 2H(tf − ti )
tf
where ti and tf are the times at which inflation started and ended and H is
the Hubble parameter during inflation.

b) Consider a k > 0 universe which went through an inflationary phase and never
inflates thereafter. Since such a model is expected to recollapse eventually,
observations will eventually reveal that Ω > 1 at, say, t > T . Estimate T .

2
 Massachusetts Institute of Technology
Department of Physics

Physics 8.942 Fall 2001

Problem Set #4
Due in class Tuesday, October 16, 2001.

1. Temperature vs. time

Assume that we live in a Robertson-Walker universe with matter, radiation and curva­
ture. The present mass density is ρm = 3Ω0 H02 /(8πG), where H0 = 100 h km s−1 Mpc−1
and Ω0 (1+ρr,0 /ρm,0 ) ≤ 1 (i.e., k ≤ 0). The present radiation temperature is T0 = 2.725 K.
Assume that only photons, with present temperature T0 = 2.725 K contribute to the ra­
diation; ignore neutrinos in this problem.
a) At what redshift did the radiation energy density equal the matter density?
How old was the universe then and what was the radiation temperature?
Your answers should scale appropriately with Ωm and h.

b) How old was the universe when the radiation temperature was 1 MeV? 1
GeV? 1014 GeV? (Hint: you need g∗ , the effective number of relativistic spin
states contributing to the energy density. At 1 GeV g∗ = 61.75 and at 1014
GeV, g∗ = 106.75 without supersymmetry or double this with SUSY.)

2. Massless and massive neutrinos

Besides photons, there are 3 flavors of neutrinos, each with 2 spin states (neglecting
right-handed neutrinos) which may be relativistic today. Assume that the numbers of
neutrinos and antineutrinos of each flavor are equal.
a) Assuming that all neutrinos are massless and that the photon temperature is
increased by a factor (11/4)1/3 as a result of e+ e− annihilation, derive zeq , the
redshift at which (nonrelativistic) matter and radiation (relativistic particles,
including photons and neutrinos), have equal energy densities (Peacock eq.
9.3.) What is the ratio of energy densities in neutrinos and photons?

b) Now suppose that one neutrino flavor (e.g., ντ and ν̄τ ) has nonzero rest mass.
Derive the neutrino mass that is needed to close the universe, assuming that
the contribution made by other massive particles (nucleons and cold dark
matter) is small (Peacock eq. 9.31). Note that the shape of the phase space
distribution is invariant after neutrino decoupling. (This calculation was first
made by Cowsik and McClelland 1972, Phys. Rev. Lett., 29, 669.)

1
3. Equilibrium recombination
In terms of the conserved baryon/photon ratio η (eq. 9.83 of Peacock), find the CMB
temperature and redshift at which recombination ended, as defined by the condition that
the photon mean-free scattering rate equals the expansion rate, ne σT c = H. Use the
Saha equation, eq. 9.45 of Peacock, assuming the parameters of the ΛCDM model of
Problem Sets 2 and 3 plus the present CMB temperature T0 = 2.725 K and the baryon
abundance ΩB h2 = 0.02. (See Burles et al. 2001, ApJ 552, L1 for conversion of ΩB h2
to η.) Compare your resulting electron fraction and redshift with the graphs presented
in Seager et al., ApJS 128, 407. Is your recombination redshift too high or too low
compared with an exact calculation? Why?

4. Nonstandard nucleosynthesis
Under which of the following suppositions would primordial nucleosynthesis have pro­
duced less 4 He than the standard model? Less 2 H (deuterium)?

a) Suppose that the baryond density in the universe today is larger than we
think.

b) Suppose that there is a fourth neutrino flavor with mass much less than 1
MeV. (It must be “sterile,” e.g. right-handed, not to have been detected in
Z 0 decay at LEP.)

c) Suppose that there are many more neutrinos than antineutrinos or photons
in the cosmic background today.

d) Suppose that there are many more antineutrinos than neutrinos or photons
in the cosmic background today.

e) Suppose that there is a significant contribution of gravitational radiation to

the total energy density of the universe, comparable in magnitude with the
energy density of the microwave background radiation.

2
 Massachusetts Institute of Technology
Department of Physics

Physics 8.942 Fall 2001

Problem Set #5
Due in class Tuesday, October 23, 2001.

1. Thomson optical depth

a) Peacock equation (18.38) gives the Thomson optical depth along the line of
sight from the observer out to redshift z for a matter­dominated model with
no cosmological constant. Derive the correct expression for a general model
including matter, vacuum energy, radiation, and curvature. Assume that
helium is neutral but hydrogen is partially ionized with ionization fraction
xe . Your answer should depend on z, xe , Ωm , ΩB , ΩΛ , aeq , h, and Y .

b) Numerically integrate the result of part a) for the flat ΛCDM model with
ΩΛ = 0.65, Ωm = 0.35, ΩB h2 = 0.02, and h = 0.72, for a � aeq . At what
redshift would the Thomson optical depth equal unity if recombination never
occurred, i.e. xe = 1? Compare your answer with that for the OCDM model
(Ωk = 0.65 instead of ΩΛ = 0.65) using Peacock’s equation (18.41).

2. Kompaneets Equation and Inverse Compton Cooling

a) Given Peacock equations (12.75)–(12.77), complete the steps to derive the

Kompaneets equation (Peacock eq. 12.79) for the modification of a pho­
ton spectrum by Compton scattering with a uniform gas of nonrelativistic
electrons.

b) Consider a gas of blackbody radiation at temperature Tγ which is scattered by

thermal electrons with temperature Te > Tγ . From the Kompaneets equation
derive the rate of energy transfer per unit volume between the electrons and
photons, ρ̇γ . Your result should be proportional to (Te − Tγ )ργ .

c) Energy conservation implies that the photon heating is accompanied by elec­

tron cooling. Suppose that helium is neutral and that particle collisions
maintain thermal equilibrium among electrons, protons, hydrogen atoms,
and helium atoms. One can then define an inverse Compton cooling time

1
 tIC = −ρB /ρ̇e where ρB is the thermal energy density of the baryons (not the
rest mass density) and ρ̇e = −ρ̇γ is the electron energy loss rate. Derive an
expression for tIC and evaluate it using the parameters of the CMB and the
ΛCDM model. If the gas is fully ionized, the inverse Compton cooling time
is shorter than the Hubble time H −1 for z > zIC . Determine zIC .

3. CMB Spectral Distortions

a) Starting from the Kompaneets equation, show that the equilibrium solution
for the photon occupation number n is a Bose­Einstein distribution with an
arbitrary nonzero chemical potential. (The equilibrium solution corresponds
to the limit of large Thomson optical depth.)

b) Under what conditions could the CMB have a nonzero µ? (Hint: Comp­
ton scattering conserves photon number but not photon energy, as energy is
exchanged with electrons.)

c) In the limit of small Thomson optical depth, fill in the steps Peacock omits
to derive his equation (12.82) for the Sunyaev­Zel’dovich effect. Check that
in the Rayleigh­Jeans regime (hν � kT ), the brightness temperature change
is ΔT /T = −2y for y � 1. Why is the photon temperature decreased by
scattering from hotter electrons?

d) In the limit of small chemical potential, µ � kT , linearize the Bose­Einstein

phase space density. What is the predicted ΔT /T in the Rayleigh­Jeans
regime? Referring to Figure 9.1 of Peacock, explain qualitatively how the
measured CMB brightness temperature versus frequency constrains models
of cosmological energy input (Wright et al. 1994, ApJ, 420, 450).

4. Gravitational perturbations on photons

Ignoring gravitational radiation and gravitomagnetism, the metric of a weakly­perturbed
Robertson­Walker spacetime may be written
� �
ds2 = a2 (τ ) −(1 + 2φ)dτ 2 + (1 − 2ψ)dl2 (1)

where φ(xi , τ ) and ψ(xi , τ ) are small gravitational potentials (φ2 � 1 and ψ 2 � 1) and
dl2 is the spatial line element of a Robertson­Walker model (e.g. dl2 = dχ2 + r2 dΩ2 ).
Throughout this problem we drop terms quadratic in φ and ψ.

a) Let E and pi = Eni be the proper energy and momentum of a photon mea­
sured by an observer at fixed spatial coordinates. In other words, the photon
4­momentum is P� = E�eˆ0 + pi�eˆi where {�eµˆ } is an orthonormal basis. From

2
 the geodesic equation, show that the photon energy changes with conformal
time along its trajectory according to

d ln(aE)
= −ni ∂i φ + ∂τ ψ . (2)
dτ
Identify the terms corresponding to the standard cosmological and static
gravitational redshifts. Show also that dni /dτ is independent of E and there­
fore the photon tra jectory is unaffected by its energy.

b) The CMB anisotropy is fully described (in the absence of polarization) by

the brightness temperature Tbr (xµ , E, ni ) for photons of proper energy E
traveling in direction ni at spacetime point xµ . Recall that the brightness
temperature is defined by Iν = Bν (Tbr ) where Bν (T ) is the Planck function.
Defining the anisotropy Δ by TB = a−1 (τ )T0 (1 + Δ), show that equation (2)
implies that gravitational perturbations multiply the energy of every photon
traveling along a given ray by the same factor and therefore Δ is independent
of E.

3
 Massachusetts Institute of Technology
Department of Physics

Physics 8.942 Fall 2001

Problem Set #6
Due in class Tuesday, October 30, 2001.

1. Sachs­Wolfe Angular Power Spectrum

The 8.942 notes Cosmic Microwave Background Anisotropy give give a formula for the
angular power spectrum Cl as an integral over the three­dimensional power spectrum of
gravitational potential fluctuations. In the Sachs­Wolfe approximation, the large­scale
(low l) anisotropy in observed direction n is Δ0 (n) = 31 φ(−χe n, τe ).

a) Complete the steps of the Sachs­Wolfe derivation in the notes to derive

the angular power spectrum for a primordial scale­invariant spectrum Pφ =
A(c/H0 )3 (ck/H0 )n−4 ,
� �
2l+n−1
Γ(3 − n) Γ 2
Cl ∝ �
4−n
� �
2l+5−n
� (1)
Γ2 2
Γ 2

and find the constant of proportionality in terms of A. (Hint: be sure to

include the factor 0.9 relating the potential between y � 1 and y � 1 for a
matter plus radiation universe as discussed in the notes.)

b) Verify Peacock equation (18.31) and find the relation between Qrms−PS (Pea­
cock eq. 18.33) and A.

c) Suppose that the primeval spectrum has a tilt with n = 0.8 over the range
of scales probed by the CMB. What is the predicted ratio between l(l + 1)Cl
at l = 20 and l = 2?

Hint: for part a) you will need

� �
� ∞ l+l� +ν+1
π2ν−2 Γ(1 − ν)Γ 2
xν jl (x)jl� (x) dx = � �
l−l −ν+2
� ��
l −l−ν+2
� � �
l+l −ν+3
� (2)
0 Γ 2
Γ 2
Γ 2

which is valid for −(l + l� + 1) < ν < 1.

2. Cosmic Variance of the CMB Quadrupole
As Peacock discusses on page 595, the Q2rms derived in Problem 1 is the mean of a
statistically fluctuating quantity. Because each spherical harmonic component of the

1
anisotropy has a Gaussian distribution with zero mean and with variance Cl ≡ �|alm |2 �,
the low degree harmonic coefficients are expected to show large statistical fluctuations.
(We can only measure one CMB sky; no matter how precise the measurement, we can
never get more than 2l + 1 measurements of degree l.) For the quadrupole the situation
is even worse because emission from the plane of our own galaxy practically eliminates
one component, so that the observational estimate of Qrms = 10.7 µK (Kogut et al
1996, ApJ, 464, L5) is drawn from a distribution that is effectively chi­squared with four
degrees of freedom. Given this fact, determine 95%­confidence upper and lower limits
for the ensemble mean Qrms−PS . Does the low observed quadrupole rule out Qrms−PS =
18 µK, the best­fit value based on fitting the entire angular power spectrum? (Hint:
The chi­square distribution with four degrees of freedom has one parameter, namely
Qrms−PS . Tune this parameter so that the measured value lies at the 5% and 95% values
of the cumulative distribution, which you can find in any elementary statistics book. See
Hinshaw et al 1996, ApJ, 464, L17 for discussion.)
3. CMBFAST
Download and build the numerical code CMBFAST (available at the 8.942 links page
— note you must have access to a f77 compiler, e.g. on MIT server machines). Before
running cmbfast you will need to run jlgen and/or ujlgen. Read the online documentation
carefully.
a) Run CMBFAST on the three standard models (SCDM, ΛCDM, OCDM) of
Problem Set 2. (Use the default parameters given by cmbfast in square
brackets whenever you are unsure.) Compare with Figure 2 of Miller et al
(1999, ApJ, 524, L1). This will require you to change the Cl to temperature
units.

b) Holding fixed ΩΛ = 0 and other parameters (e.g. H0 = 72, ΩB h2 = .02, and

n = 1), vary Ω over the range (0.1, 1). Make a graph of lmax (Ω), the location
of the first acoustic peak. Then hold fixed Ω = 1 (i.e. spatially flat models)
and vary ΩΛ over the same range. Superpose the lmax (Ωm = 1 − ΩΛ ). Are
the results consistent with Problem 2 of Problem Set 3?
c) Run CMBFAST with a tilt n = 0.8 and compare the ratio of l(l + 1)Cl at
l = 2 and l = 20 with the Sachs­Wolfe result from Problem 1.

4. Gamma Ray Burst Luminosity

Gamma ray burst 990123 occurred in a galaxy of redshift z = 1.60. The measured fluence
was 3.0 × 10−4 erg cm−2 for photon energies above 20 keV (Briggs et al 1999, ApJ, 524,
82). Convert this to a total emitted burst energy in units of M� c2 assuming isotropic
emission. You will need to calculate the luminosity distance to z = 1.60. Evaluate it for
the three models of Problem Set 2 with h = 0.72. You will have to perform a numerical
integral for the ΛCDM model. You may use whatever is convenient, e.g. Matlab, Maple,
or Mathematica on MIT servers — see the 8.942 links for documentation.

2
 Massachusetts Institute of Technology
Department of Physics

Physics 8.942 Fall 2001

Problem Set #7
Due in class Tuesday, November 6, 2001.

1. Galaxy Number Counts

The differential number counts of galaxies have been measured to faint magnitudes in
the Hubble Deep Field (Williams et al 1996, AJ 112, 1335). At B = 29, the HDF results
give dN/dm = 105.54 galaxies per magnitude and per square degree.

a) Assuming isotropy, convert the observed number counts to the total counts
over the whole sky −dN/d ln S where S is the flux in the HST blue band­
pass. (See Peacock Chapter 13 for the relation between flux and magnitude.)
Extrapolate this to get N (S), the total number of galaxies in the universe
visible to us with B ≤ 29. (You may approximate the source counts assuming
a constant logarithmic slope d ln N/d ln S = −β.)

b) Let us naively assume that the galaxy luminosity function does not evolve
and that there is no K­correction (Peacock p. 395). Suppose that the blue
luminosity function is a Schechter function with α = −1, φ∗ = 0.015h3
Mpc−3 , and M ∗ = −19.50 + 5 log10 h. At B = 29, what is the luminosity
distance (in h−1 Mpc) for a galaxy of luminosity L∗ ? Call this number d∗29 .

c) Show that the differential counts at B = 29 in a flat universe with α = −1

are � �
dN c 3 ∗ � H0 τ0 � �
− = 4π φ exp −(dL /d∗29 )2 x2 dx (1)
d ln S H0 0

where x = H0 χ/c and the luminosity distance is a function of χ(z) (dL ≈ χ

for x � 1).

d) Show that for an Einstein­de Sitter universe, H0 dL /c = 4x/(2 − x)2 . Then

numerically integrate equation (1) to get the predicted differential source
counts at B = 29. Compare with your result of part a) from the HDF
measurement.

2. Gunn­Peterson Effect
Neutral hydrogen at redshift z absorbs background quasar light at a wavelength 121.6
nm, creating absorption at a wavelength 121.6(1 + z) nm as measured today at redshift

1
zero. Treating the Lyman alpha line as being infinitely narrow, the absorption cross
section per neutral hydrogen atom is

πe2
σ(ν) = σα δ(ν − να ) , σα = fα (2)
me c
where ν = c/λ is frequency, να = c/(121.6 nm) is the Lyman alpha frequency, fα = 0.416
is the oscillator strength, and cgs units are used in writing σα (cf. Peacock eq. 12.36). In
this problem we neglect peculiar velocities. The original reference by Gunn & Peterson
(1965, ApJ, 142 1633) is a very readable guide to this problem (though beware it has
some typos).

a) Derive the absorption optical depth at observed frequency ν0 due to neutral

hydrogen at 1 + z = να /ν0 . Your result should depend on σα , c, να , nHI ,
and H (the Hubble parameter at z). Compare your result with Peacock eq.
(12.41) for a Friedmann universe.

b) The measured optical depth versus frequency is not uniform but varies (the
Lyman alpha forest) with nHI along the line of sight. The mean optical depth
is unity at z ≈ 3. Assuming ΩB h2 = 0.019, h = 0.65, Y = 0.24 (with helium
neutral), and the flat ΛCDM model with ΩΛ = 0.65, what is the mean neutral
fraction 1 − xe of hydrogen atoms at z = 3? Compare with Peacock p. 364.

c) In equilibrium, hydrogen atoms are photoionized and recombine at equal

rates, implying

(2) 1 − xe nH α(2) (T )
nHI β = ne np α (T ) or = (3)
x2e β

where nH = (1 − Y )ρB /mH and

� ∞
dν
β= 4πJν σi (ν) ≈ 3 × 10−12 J−21 s−1 ;
νL hν
� �
hνL
α(2) = 2.06 × 10−11 T −1/2 φ2 (T ) cm3 s−1 , φ2 (T ) = 0.448 ln 1 + . (4)
kT

Here, hνL = 13.6 eV is the ionization energy of hydrogen, σi is the ionization

cross­section, T is the temperature in Kelvin, and J−21 is the mean intensity
at frequencies just above νL (in units of 10−21 erg cm−2 s−1 Hz−1 ster−1 ). The
α(2) excludes recombinations directly to the ground state (Case B recombi­
nation). For the ΛCDM model, assuming the gas temperature is 104 K, what
J−21 is needed to ionize the gas to the level determined in part b)?

2
 Massachusetts Institute of Technology
Department of Physics

Physics 8.942 Fall 2001

Problem Set #8
Due in class Tuesday, November 13, 2001.

1. Power Spectrum and Window Functions

Note: this problem set uses the Fourier transform convention of equation (1) below.
� �
a) Starting from the two­point correlation function in k­space, δ(�k1 ) δ(�k2 ) =
3 �
P (k1 ) δD (k1 + �k2 ), show that the correlation function and power spectrum are
Fourier transform pairs:
�
�
ξ(r) = d3 k eik·x� P (k) (1)

where r = |�x |. (We assume that space is flat and use the usual notation for
3­vectors.) This is known as the Wiener­Khintchine theorem.

b) Suppose� that we filter δ(�x ) by convolving � (�

with a smoothing kernel W x ):
�
˜ 3 � � � � 3 �
δ(�x ) = d x W (�x − �x ) δ(�x ) where d x W (�x ) = 1. Show that

δ̃(�k ) = δ(�k )W (�k ) (2)

and give an expression for W (�k ), called the window function, in terms of
� (�
W x ). What is W (0)?

c) Suppose that W� (� � (�
x ) = 3/(4πR3 ) for r < R and W x ) = 0 for r > R. This is
called a “spherical tophat” in cosmology. What is the corresponding window
function? You should derive a closed­form expression as a function of kR.
� (�
d) Suppose that W x ) ∝ exp(−r2 /2R2 ) is a Gaussian of width R. What is the
corresponding window function?

2. Normalization of the power spectrum

Peacock gives an empirical fit to the “linear” power spectrum of mass fluctuations in
his equations (16.137) and (16.138) where Δ2 (k) ≡ 4πk 3 P (k). This problem requires
numerical integrations over the power spectrum. You may use Matlab, Maple, Mathe­
matica, rombint.c, etc. for this. Be sure to truncate the integrals at a sufficiently high k
so that you get numerical errors smaller than one percent.

1
 � �
1/2
a) Using Peacock’s spectrum, compute σ8 = δ̃ 2 where δ˜ is the density
perturbation filtered with a spherical tophat of radius 8 h−1 Mpc. Compare
with Peacock equation (16.136) for Ωm = 0.35.

b) Using the cosmological Poisson equation �2 φ = 4πGρ̄m δ for a = 1, determine

the power spectrum of the gravitational potential in terms of Peacock’s Δ2 (k).
Show that k 3 Pφ (k) → constant as k → 0. Assuming that the gravitational
potential has decayed by a factor g(Ωm ) < 1 since recombination, compute
Qrms−PS in the Sachs­Wolfe approximation (Problem 1 of Problem Set 6) and
compare with the COBE measurement Qrms−PS = 18 µK.

2
 Massachusetts Institute of Technology
Department of Physics

Physics 8.942 Fall 2001

Problem Set #9
Due in class Tuesday, November 20, 2001.

1. CDM density perturbation growth

For wavelengths much less than the Hubble length (kc � H), the Fourier transform of
the density perturbation for nonrelativistic matter, δ(�k, τ ), obeys the well­known linear
wave equation for small­amplitude, isentropic, longitudinal perturbations:
ȧ � �
δ¨ + δ̇ + k 2 c2s − 4πG¯
ρa2 δ = 0 (1)
a
where dots denote conformal time derivatives.
a) Consider a universe dominated by cold dark matter (CDM) with cs = 0, but
also including radiation. Ignoring the radiation perturbations — a useful
approximation on scales smaller than the Hubble distance — equation (1)
is valid for the CDM throughout the radiation­ and matter­dominated eras
(with ρ̄ on the right­hand side being the mean density of the CDM). Using
the solution of the Friedmann equation for a(τ ) in a flat matter+radiation
universe, a(τ ) = a1 τ + a2 τ 2 (be sure to express a1 and a2 in terms of H0 and
aeq = [1+zeq ]−1 ), show that the growing mode solution is δ ∝ 1+(3/2)(a/aeq ).
Interpret this result physically for a < aeq as well as a > aeq .
b) Rewrite the wave equation given above in terms of the gravitational potential
perturbation φ(�k, τ ). Show that φ̇ = 0 for growing­mode density perturba­
tions in a Ω = 1 CDM­dominated universe, implying that in real space,
φ = φ(�x). What happens to φ if Ω = � 1 or cs =� 0? Now consider an iso­
lated, gravitationally bound, relaxed system (e.g., the Milky Way galaxy),
whose mass and proper size do not change with time. Assuming that the
gravitational potential at the center is finite, does it change with time? Can
one, therefore, estimate the initial gravitational potential fluctuations from
the gravitational potential well depths of galaxies and clusters?

2. Hot Dark Matter

Suppose that the universe is closed by one flavor of massive neutrino with Ων = 1. The
neutrinos stream freely (without scattering) since decoupling at T ≈ 1 MeV. The proper
momentum of a typical neutrino is p = (1 + z) kB Tν0 /c where Tν0 = (4/11)1/3 T0 is the
present neutrino “temperature.”

1
 a) Show that the proper distance travelled by a typical neutrino (the “free­
streaming distance”) is
� τ0 � �−1/2 kB Tν0
Lfs = a0 c dτ α (1 + z) 1 + α2 (1 + z)2 where α ≡ (2)
0 mν c2
and z = z(τ ) through the usual relation 1 + z = a0 /a(τ ).
b) At what redshift does the massive neutrino become nonrelativistic? Assuming
that this occurs before zeq , compute zeq for an Ω = 1 universe with two flavors
of massless neutrinos and one flavor of massive nonrelativistic neutrino. (cf.
Problem 1 of Problem Set 4.)
c) Using the exact z(τ ) for a K = 0 matter+radiation universe with two flavors
of massless neutrinos and one flavor of massive nonrelativistic neutrino, and
with h = 0.72, compute the exact free­streaming distance in Mpc by numer­
ically integrating equation (2). How doe the results scale with mν and h if
α � aeq ? If massive neutrinos are the dark matter, free­streaming erases
primordial fluctuations for wavelengths up to about Lfs .

3. Spherical infall model

The nonlinear evolution of spherically symmetric perturbations of a Friedmann­Robertson­
Walker (zero pressure) universe is most easily described by integrating the trajectories
of spherical shells containing fixed enclosed mass M .
a) Show that the solution R(t) for the proper radius of a sphere enclosing fixed
mass is given parametrically by R = A(1 − cos η), t = B(η − sin η). What is
the value of A3 /B 2 ?
b) By comparing R(t) with the result for an unperturbed Ω = 1 Einstein­de
Sitter universe, obtain the exact nonlinear solution (in parametric form) for
¯ ≡ M/M
δ(t) ¯ (t) − 1, where M¯ = (4π/3)ρ̄R3 . (Hint: Follow a shell of fixed M
and ask how much mass it would enclose, M ¯ (t), if the density were uniform
and equal to the critical density ρ̄. Note that δ¯ is the volume­average of the
density perturbation ρ/ρ̄ − 1.)
c) If the enclosed mass remains constant and the initial energy is negative,
R → 0 implying δ → ∞. Show that this occurs for growing­mode initial
perturbations when linear perturbation theory would predict δ¯ = δc ≈ 1.69.
Find the exact expression for δc in terms of π and simple numbers.

4. Zel’dovich Approximation
The Zel’dovich approximation is to write the tra jectories of pressureless matter (dark
matter, or baryons on scales larger than the Jeans length) as
� (�q ) ,
�x (�q, τ ) = �q + D+ (τ )ψ (3)

2
 � (�q ) is determined by the initial conditions and D+ (τ ) is the zero­pressure linear
where ψ
growing mode growth factor.

a) Consider a spherically symmetric density perturbation of the form

�
δi > 0 , |�x | < q0
δ(�x, τi ) = (4)
0, |�x | > q0

Find the radial displacement field ψ � (�q ) corresponding to this density field
to lowest order in δi . Show that the Zel’dovich approximation gives δ(|�x | <
q0 , τ ) → ∞ at some finite τ = τc . What is the corresponding prediction for
δ(τc ) from linear perturbation theory? Does the Zel’dovich approximation
agree with the exact solution of the spherical infall model from Problem 3
above?

b) The Zel’dovich approximation is exact for plane­parallel perturbations for

tra jectories that have not intersected others. Show this by considering a
one­dimensional density field ρ(x, τ ) with corresponding displacement field
ψx (qx ) and gravitational potential φ(x, τ ) obeying the Poisson equation ∂x2 φ =
4πGa2 (ρ−ρ̄). Hint: substitute the Zel’dovich approximation trajectories into
the exact equation of motion d2�x/dτ 2 + (ȧ/a)d�x/dτ = −� � φ. Show that the
� φ implied by this equation agrees with the solution of the Poisson equation
�
assuming mass conservation.

3
 Massachusetts Institute of Technology
Department of Physics
Physics 8.942 Fall 2001

Problem Set #10

Due in class Thursday, December 4, 2001.

1. Zel’dovich Approximation
The Zel’dovich approximation is to write the trajectories of pressureless matter (dark
matter, or baryons on scales larger than the Jeans length) as
� (�q ) ,
�x (�q, τ ) = �q + D+ (τ )ψ (1)
� (�q ) is determined by the initial conditions and D+ (τ ) is the zero-pressure linear
where ψ
growing mode growth factor.
a) Consider a spherically symmetric density perturbation of the form
δi > 0 , |�x | < q0
�
δ(�x, τi ) = (2)
0, |�x | > q0
Find the radial displacement field ψ � (�q ) corresponding to this density field
to lowest order in δi . Show that the Zel’dovich approximation gives δ(|�x | <
q0 , τ ) → ∞ at some finite τ = τc . What is the corresponding prediction for
δ(τc ) from linear perturbation theory? Does the Zel’dovich approximation
agree with the exact solution of the spherical infall model from Problem 3
above?
b) The Zel’dovich approximation is exact for plane-parallel perturbations for
trajectories that have not intersected others. Show this by considering a
one-dimensional density field ρ(x, τ ) with corresponding displacement field
ψx (qx ) and gravitational potential φ(x, τ ) obeying the Poisson equation ∂x2 φ =
4πGa2 (ρ−ρ̄). Hint: substitute the Zel’dovich approximation trajectories into
the exact equation of motion d2�x/dτ 2 + (ȧ/a)d�x/dτ = −�φ. � Show that the
�
�φ implied by this equation agrees with the solution of the Poisson equation
assuming mass conservation.
2. Linear growing mode
Small-amplitude cosmological density fluctuations with comoving wavenumber obeying
H � k � kJ grow in amplitude according to the well-known damped, driven wave
equation
¨ + ȧ Ḋ = 4πGρ̄m a2 D
D (3)
a
where a dot denotes a conformal time derivative.

1
 a) Suppose that the universe contains nonrelativistic matter, vacuum energy,
and (possibly) curvature, but no other types of matter. By combining equa­
tion (3) with the Friedmann equation, show that one solution is given by
D− (τ ) = H(τ ).

b) Using this solution and the method of variation of parameters, find a quadra­
ture solution for the growing mode, D+ (τ ). (Hint: write D+ = D− (τ )f (τ )
and substitute into eq. 3. You should obtain f (τ ) in the form of an integral.)

c) Using the exact solution of a(τ ) for an OCDM (matter-only, open) universe
(a ∝ cosh η − 1 where η ∝ τ , from problem 2b of Problem Set 2), verify
equation (15.31) of Peacock. Also show that for the Einstein-de Sitter model
(flat Ωm = 1), D+ ∝ a.

d) Now consider a flat ΛCDM model. Write the quadrature for D+ (a) using
expansion factor as the integration variable so that the quadrature is easy to
evaluate numerically. Evaluate the growth suppression factor g ≡ D+ (a =
1, Ωm = 0.35)/D+ (a = 1, Ωm = 1) numerically for the ΛCDM model and
compare with equation (15.43) of Peacock. How accurate is Peacock’s ap­
proximation?

e) Suppose that the universe contains quintessence with equation of state p =

wρ, with w being (for simplicity here) a constant in the range −1 < w < 0.
Show that the Peacock’s equation (15.42) is invalid unless w = −1 or w = − 13 .

3. Simple model of nonlinear evolution

In hierarchical clustering models of cosmic structure formation, matter is clumped strongly
on small scales and is smooth on large scales, with a transition mass Mnl that grows in
time. One may define Mnl as the mean mass contained in a smoothing window large
enough so that the filtered linear density fluctuation field has unit variance:
�
4π 3
σ(Mnl ) = 1 where σ 2 (M ) ≡ d3 k P (k)W 2 (kR) , M = ρR
¯ . (4)
3
The distribution of clump masses is given approximately by the Press-Schechter formula,
Peacock equation (17.13). In this problem we use the gaussian window function W (x) =
exp(− 12 x2 ).
a) Suppose that the power spectrum of δρ/ρ̄ in the linear regime may be ap­
proximated (for at least a useful range of k) as a power-law in k, P (k, τ ) =
2
D+ (τ )Ak n . Determine σ(M ) in terms of the relevant constants. (Express
the needed integral as a gamma function.)

b) Check Peacock equation (17.14) for the low-mass slope of the Press-Schechter
mass distribution.

2
 c) Today, Rnl ≈ 2.6 h−1 Mpc (using the gaussian window function). What is
the corresponding Mnl in solar masses (in terms of Ωm and h)? Now suppose
that we take the characteristic mass of a galaxy to be 1011 h−1 M� . At what
redshift was σ = 1 for this mass scale, i.e. what is the predicted redshift
of galaxy formation? Evaluate your result for SCDM and ΛCDM using the
standard parameters for these models with n = 1. (For ΛCDM you’ll have
to evaluate D+ (a) by numerical integration.)

4. Gravitational radiation with a simple equation of state

Gravitational waves correspond to transverse-traceless metric perturbations h ij with
�i hi j = hi i = 0. The perturbations evolve according to the wave equation for a massless
spin-2 field,
ȧ
� �
∂τ2 + 2 ∂τ − �2 + 2K hij = 8πGa2 Σij,T (5)
a
where K is the spatial curvature constant and Σij,T is the transverse-traceless shear stress,
the source for gravitational radiation. It vanishes in vacuum. Part a) of this problem
is concerned with the evolution of the background spacetime; gravitational radiation is
considered in the remainder. Assume that Ω ≈ 1 (H 2 � a−2 |K|) as is appropriate at
high redshift even if Ω �= 1 today.

a) Suppose that the unperturbed equation of state of the universe is p = wρ with

w = constant (e.g., w = 0 for a matter-dominated universe, w = 1/3 for a
radiation-dominated universe, and w = −1 for a vacuum energy-dominated
universe). Solve the Friedmann and energy conservation equations for the
unperturbed Robertson-Walker spacetime to get a(τ ). Show that for w >
−1/3 the result is a power law of τ but for w < −1/3 the solution is a power
of τ∞ − τ . What is the physical interpretation of τ∞ ? How does this relate
to inflation?

b) Show that gravitational radiation has two polarizations. (Hint: use the infor­
mation given at the beginning of this problem.) Write down 3 × 3 matrices
�+ ×
ij and �ij corresponding to the two independent linear polarization states
of plane gravitational waves traveling in the x3 -direction. The general gravi­
tational wave may then be written h+ (k, τ )�+ ×
ij + h× (k, τ )�ij summed over all
different plane waves.

c) For plane gravitational waves of comoving wavenumber k, show that the

source-free gravitational wave equation has two solutions (for each polariza­
tion) given in terms of powers and Bessel functions of kτ (for w > −1/3)
or k(τ∞ − τ ) (for w < −1/3). Show that h(k, τ ) is constant on scales much
larger than the Hubble distance (neglecting the decaying mode). Thus, both
density fluctuations and gravitational waves created during inflation can be

3
 cosmologically important at late times when they reenter the Hubble dis­
tance.

d) Show that the amplitude of gravitational waves decays as a−1 on scales small
compared with the Hubble distance (kτ � 1). Interpret this result physically
in terms of the energy flux carried by gravitational waves. (A qualitative
argument will suffice.) The decay of h once waves cross the Hubble length
implies that tensor mode contributions to the CMB anisotropy are negligible
for l > 100.