Professional Documents
Culture Documents
Lesgourgues
Exercises 12
RWTH Aachen University 9.01.2019
To be handed back on the 16.01.2019 at 8:30 (by groups of 2 minimum, 3 maximum). Will be
corrected on the 23.01.2019. You are encouraged to do the optional questions if you are curious
but they do not bring points.
dr2
2
dl2 = a2 + r 2
(dθ 2
+ sin θ dφ2
) . (1)
1 − kr2
We have seen in the course that the cases k > 0, k = 0 and k < 0 correspond respectively to a
spherical, euclidian or hyperbolic geometry. Moreover, when k 6= 0, we p can always absorb |k| in a
redefinition of the scale factor a (after the change of variable r −→ r/ |k|). Hence we can assume
without loss of generality that k = ±1. We are going to study these two cases.
1.1 Show that this 3-sphere can be described with the following parametrisation:
where the three angles χ, θ, φ are the spatial coordinate of the 3-sphere.
(max. 4 lines of calculations)
1.2 Normally (χ, θ, φ) should all vary in the range [0, 2π]. Find some symmetries and justify that
the entire 3-sphere is covered by varying the coordinates in the ranges 0 ≤ χ ≤ π, 0 ≤ θ ≤ π,
0 ≤ φ < 2π. (It is always helpful to think first of what you know about spherical coordinates for
the 2-sphere).
(max. 6 lines of calculations)
1.4 Write the expression of an infinitesimal volume element on the 3-sphere (again, the analogy
with the 2-sphere might help you). Show that the total volume of the 3-sphere is given by 2π 2 R3 .
(max. 6 lines of calculations)
1
1.5 Find a change of variables such that the metric becomes identical to Eq. (1) with k = +1.
Express the curvature radius and the volume as a function of a.
(max. 4 lines of calculations)
Unlike the 3-sphere, the homogeneous 3-hyperboloid of the FLRW model cannot be embedded
in 4-dimensional Euclidian space. It can be embedded in another 4-dimensional space of metric
dl2 = −dw2 + dx2 + dy 2 + dz 2 (similar to Minkowski space, but here w is like an extra spa-
tial dimension; we are not saying that this dimension exists physically, we use this 4-dimensional
space just as a mathematical tool for defining the physical 3-hyperboloid). In that space, the 3-
hyperboloid is the hypersurface of equation w2 −x2 −y 2 −z 2 = R2 (where R is the curvature radius).
1.6 Find a parametrisation of the 3-hyperboloid in terms of 3 coordinates (χ, θ, φ), where χ is a
standard real coordinate and (θ, φ) are angles, such that
1.7 (Optional) Normally χ should take any value and (θ, φ) should vary in the range [0, 2π]. Find
some symmetries and justify that the entire 3-sphere is covered by varying the coordinates in the
ranges 0 ≤ χ < ∞, 0 ≤ θ ≤ π, 0 ≤ φ < 2π.
(max. 6 lines of calculations)
1.9 (Optional) Find a change of variables such that the metric becomes identical to Eq. (1) with
k = −1. Express the curvature radius as a function of a.
(max. 4 lines of calculations)
In the course, we built the FLRW metric in such a way that there exists a set of comoving
observers at rest with the corredinate system and seeing an isotropic universe. Now, let us take
another point of view: somebody gives you the expression of the FLRW metric, and you want
to prove that this metric is compatible with the existence of comoving observers, i.e. of objects
keeping zero coordinate velocities. For that, you must show that the worldline of a massive particle
at rest in the coordinate system is indeed a valid solution of the geodesic equation
d2 xµ dxα dxβ
2
+ Γµαβ =0, (8)
dt dt dt
where t is the proper time of the particle.
2
3. Rescalings in the FLRW metric
We assume a homogeneous universe with an FLRW metric
dr2
2 2 2 2 2 2 2
ds = −dt + a(t) + r dθ + sin θ dφ (9)
1 − kr2
with a curvature parameter k ∈ <. The sign of k determines whether the curvature is null (flat),
a(t)
positive (spherical) or negative (hyperbolic), while the physical radius of curvature is Rc (t) = √ .
|k|
3.1. Show that the function a(t) is defined up to an arbitrary rescaling. To do so, rescale a(t) by
an arbitrary factor α, and show that a redefinition of the radius r and of the curvature parameter
k allows to come back to the previous form of the metric.
a(t)
3.2. Check that the two physical curvature radii of the Friedmann model, Rc (t) = √ and
|k|
a(t)
RH (t) = ȧ(t) = H(t)−1 , are left invariant by this arbitrary rescaling.
3.3. Show that it is then always possible to redefine the coordinates in such a way that, instead of
having an arbitrary k ∈ <, one deals only with the three cases k = 0, or k = +1, or k = −1. In
the rest of this exercise, we will call this the “normalised k convention”. What is the value of Rc (t)
when one works with this convention?
3.4. The following sentences could be the beginning of a research article in cosmology:
“We work in a homogeneous universe with an FLRW metric given by eq. (9) with k = 0, +1
or −1. For simplicity, we fix the arbitrary normalisation of the scale factor in such a way that
a0 ≡ a(t0 ) = 1, where t0 is the time today. With this convention, at the present time, coordinate
distances coincide with physical distances.”
Show that there is something wrong in this text!
R r dr̃
3.5. (Optional) Section 3.3.4 gives the definition of the comoving distance, χ(r) ≡ 0 √1−kr̃ 2
(here r̃ is just a dummy variable). Show explicitely that with the “normalised k convention”, the
relation between χ and r is given by r = fk (χ), where we defined
sin(x) if k = 1,
fk (x) ≡ x if k = 0, (10)
sinh(x) if k = −1.
4.6. (Optional) Generalise the function fk (i.e., the expression of r as a function of χ) to the
case where k is not normalised, and can take an arbitrary real value.
4.7. (Optional) Section 3.3.5. gives the relation between the angular distance dA and the redshift
z of a given object: Z z
a0 dz̃
dA (z) = fk
1+z 0 a0 H(z̃)
(here r̃ is just a dummy variable). We do not assume here the “normalised k convention”: hence k
can take any value, and the normalisation of a(t) is arbitrary. Since dA (z) is a physical quantity,
we expect that it should not depend on the arbitrary normalisation of the scale factor today, i.e.
on the value of a0 . Show that dA (z) is obviously independent of a0 in the case k = 0. When k 6= 0,
by carefully remembering why we say that a0 is arbitrary, show that this conclusion is still true: a
different choice for the value of a0 gives the same physical angular distance dA (z) for an object of
given measured redshift z.