3 Spacetime metrics

3.1 Introduction
The efforts to generalize physical laws under different coordinate transfor-
mations would probably not have been very successful without differential
calculus. Riemann (1826-66) introduced multidimensional spaces and differ-
ential geometry, a necessary instrument in the development of the Theory
of Relativity. In Euclidian space the distance between 2 points can be cal-
culated from Pythagoras’ theorem. Under Special Relativity this is not an
invariant, i.e. it changes its value when we measure it in different inertial sys-
tems. But it could be made invariant by adding time to space and work with
spacetime instead of space. Hermann Minkowski defined the 4-dimensional
spacetime in 1908.

3.2 Flat spacetime

In spacetime, the separation or interval s, defining the distance between two
events like the Pythagoras’ theorem is defined by the metric (the definition of
the separation and the sign convention varies. I will follow Rydens textbook):

∆s2 = −c2 ∆τ 2 = −c2 ∆t2 + ∆x2 + ∆y 2 + ∆z 2 (19)

t is the coordinate time, sometimes also called real time. xyz are the space
coordinates. This space is non-Euclidean but flat. It has a Lorentzian signa-
ture (- + + +) as opposed to Euclidean (++++). It is flat because a vector
parallell transported along a closed curve always returns to its original po-
sition (cmp. e.g. to what happens if you move a vector on the surface of a
sphere). τ is the proper time and may be regarded as your local, ’personal
time’. It is also called clock time and process time, because it measures the
distance in time between two events, e.g. the swing of a pendulum from one
position to another. ∆τ is what your clock considers to be the time interval
between two local events. ∆t is the time interval you read from your clock
between two events that take place ’elsewhere’. If e.g. these processes happen
in a coordinate system in motion relative to you or in a gravitational field
stronger than your local field they will run slower. Thus, a clock riding along
with a photon would stand still as viewed from an observer at rest (which is
of course always the case since the speed of light is the same in all coordinate
systems). Thus your clock would say that it would take an infinite amount
of time to register the time between e.g. second marking ticks on the moving
clock.

9
In the Minkowski spacetime we can define two regions related to our own po-
sition, one timelike and one spacelike. They are separated by the lightcone,
defined by the incoming and outgoing light rays. Time runs along the sym-
metry axis of the lightcone and space is perpendicular to time. A separation
between two events in spacetime is timelike if inside the cone and spacelike
if outside. At the intersection the separation is lightlike, since this is the tra-
jectory of a photon. Matter cannot travel along spacelike trajectories. Light
and massless particles travel along the lightcone.
The trajectory of a particle in spacetime is called a worldline. The trajectory
of a particle in free fall in spacetime is called a geodesic. The position along
the worldline is given by the proper time τ . Remember that the value of
this parameter is independent of the observer, invariant under coordinate
transformations.

For two events occuring simultaneously −∆τ 2 = (∆x2 + ∆y 2 + ∆z 2 )/c2 =

∆r2 /c2 , where ∆ |r| is the proper distance between the two events. From this
we can derive the Lorentz contraction formula.

The Minskowski metric can be written:

∆τ 2 = −∆s2 /c2 = ∆t2 − (∆x2 + ∆y 2 + ∆z 2 )/c2 = ηµν ∆xµ ∆xν

X
(20)
µ,ν

η is the metric tensor, describing the metric of the coordinate system. Tensors
are convenient to use in differential calculus because the equations can be
written in a more compact form. Here however, we will follow the textbook
and refrain from using tensors in the following.
By definition, the equivalence principle states that in a freely falling coordi-
nate system, SRT applies and we can use the Minkowski metric to describe
the spacetime. The metric can be used to derive the time dilation in SRT.
Since v 2 ∆t2 = ∆x2 + ∆y 2 + ∆z 2 and using eq. 20 we obtain
q
∆t = ∆τ / (1 − v 2 /c2 ) (21)
A person who watches a clock onboard a spaceship in motion will notice that
the time runs slower than on his own clock. In this way travelling salesmen
will grow older more slowly than their family members at home. A business
man using a few flight connections per week to visit his business partners
will typically gain 10−5 s during his lifetime.
The separation between 2 events A and B can be expressed as

10
 Z B
τAB = dτ (22)
A
A free fall trajectory, a geodesic, requires that the proper time τAB is min-
imized along a straight world line joining the two events. This statement is
independent on which coordinate system is in use and is said to be generally
covariant. To modify equations of physics such that they obeyed general
covariance was one of Einsteins major goals. He stated: ”The general laws of
nature are to be expressed by equations which hold good for all systems of co-
ordinates. that is, are co-variant with respect to any substitutions whatever
(generally covariant). ”

3.3 Curved spacetime

If we carry out a coordinate transformation from an inertial system in free fall
to an arbitrary coordinate system it can be shown that the time derivative
of the velocity along a world line contains a term, the affine connection, that
acts as an acceleration if the metric is not flat. Einsteins gravity can thus
be regarded a property of curved spacetime. Cmp. Newton, who regarded
the gravitational force as a result of ’action at a distance’. The tensor that
describes the curvature of spacetime is the Riemann-Christoffel (or simply
the Riemann) tensor Rα βγδ , corresponding to the Gaussian curvature, K, in
three dimensions.
In general relativity a relationship between the distribution and flow of mass-
energy in the universe and the geometrical properties of spacetime is estab-
lished on the basis of the cosmological principle. This assumes that we can
regard the universe as a perfect fluid. A perfect fluid is defined such that an
observer in motion sees the fluid around him as isotropic. This will be the
case if the mean free path between collisions is small compared with the scale
of lengths used by the observer. For instance, the sound wave will propagate
in air if its wavelength is large compared to the mean free path, but at very
short wavelengths viscosity becomes important and the air stops acting like
a perfect fluid.
The connection between the density and flow of energy and momentum, given
by the energy-momentum or stress-energy-momentum tensor, and the metric
of spacetime, given by the Einstein tensor, are described by 10 independent
non-linear partial differential equations called the Einstein field equations.

11
4 Cosmic kinematics
4.1 The geometrical framework
From the cosmological principle it follows that spacetime has a constant
curvature. This simplifies the modelling quite considerably of course. First
we will have a look at the metric of a 2-dimensional surface embedded in 3
spatial dimensions. Most of the time I will use the same variable names as
in the textbook (Ryden). In 2.2.1 we defined the parameter x≡Sk (r), which
is the radial coordinate of a small circle on this surface projected on a plane
and φ and θ are polar coordinates. Let us look at the metric of a 2-dim space
with constant curvature as a function of the two spatial coordinates x and φ
in the projected coordinate system instead of the r and φ grid on the surface
itself as derived in sect. 2.2:

ds2 = f (x)dx2 + x2 dφ2 (23)

f(x) is an unknown function of x. In a flat, Euclidean 2-dim. space f(x) ≡
1. In the case of a positively curved 2-dim space f(x) > 1 since for a circular
area πs2 > πx2 .
Introduce the polar coordinate θ (sweeping along the meridian of a sphere if
positively curved). Then the proper distance is
Z xq
x
dp = f (x)dx = Rθ = Rsin−1 (24)
0 R
Derivate with respect to x:
q 1 1
f (x) = R q (25)
1 − (x/R)2 R

1 1
f (x) = 2
= (26)
1 − (x/R) 1 − Kx2
K is the curvature constant. Thus we may write the metric of a 2-dim.
maximally symmetric surface as

dx2
ds2 = + x2 dφ2 (27)
1 − Kx2
The metric of a 3-dim space with constant curvature may be expressed:

12
 dx2
ds2 = + x2 dθ2 + x2 sin2 θdφ2 (28)
(1 − Kx2 )
Now introduce the scale-free radial distance σ, which in fact is ≡ the scale-free
angular distance discussed below:

σ = x/R (29)
Let us also express the scale dependent curvature K in a form which contains
the scale free curvature constant k as in eq. 18. k is independent of the
expansion or the change of scale of the universe but contains the information
about if the space is flat, negatively or positively curved:

k=1 P os
k


K(t) ≡ ⇒ k = 0 F lat (30)
R2 (t)
k = −1 N eg



Now eq. 28 will read

dσ 2
ds2 = R2 ( 2
+ σ 2 dθ2 + σ 2 sin2 θdφ2 ) (31)
(1 − kσ )
If we keep θ and φ constant we can express the proper distance, i.e. the
distance you measure with a rod (cmp. eq. 24), as
Z r Z σ
dx dσ
dp = √ =R √ (32)
0 1 − Kx 2 0 1 − kσ 2
Now let us switch to 4-dim spacetime and introduce the comoving coordinate
system. Imagine this to be defined by objects in free fall, i.e. following their
worldline, each of them carrying a clock. Then τ =t. One may think of a
mesh, or a coordinate grid, following the objects in the dynamical evolution
of the universe. In a 2-dim analogy of a positively curved expanding universe
the positions of galaxies would be represented by points on a ballon that is
inflated. But remember that in the general relativistic universe the expansion
is not affecting the small scales (your body is not partaking in the expansion).
The reason is that the Friedmann equations are based on the cosmological
principle, assuming homogeneity. On smaller scales, of the size of a large
cluster of galaxies, this is no longer correct.
If we set the clocks according to the cosmological principle we may express
the spacetime metric in this coordinate system:

dσ 2
ds2 = −c2 dτ 2 = −c2 dt2 + R2 (t)( + σ 2 dθ2 + σ 2 sin2 θdφ2 ) (33)
(1 − kσ 2 )

13
This is the Robertson-Walker metric or simply the R-W metric. ds is also
called the Robertson-Walker line element (1934). Remember that k has been
chosen such as to become constant. The parameters used in this equation
are frequently used in classic textbooks to express the metric. σ, Θ and φ
are all comoving coordinates, i.e. they are like the coordinates of a grid that
grows in size, following the expansion. We will now however switch to the
notation used in Rydens book.

Define

R(t) = a(t)R0 (34)

a(t) is called the cosmic scalefactor and is unitless. R is expressed in length
units. In a positively curved universe R may be called the radius of the
universe. But it is not physically existing in the 3-dimensional space we live
in since we cannot simultaneously measure the distance R or point ”in the
direction of R”. In order to do this we would have to live in a 5-dimensional
spacetime with 4 spatial coordinates.
The R-W metric in Rydens notation reads

dx2
ds2 = −c2 dτ 2 = −c2 dt2 + a2 (t)( + x2 dθ2 + x2 sin2 θdφ2 ) (35)
(1 − kx2 /R02 )
or in a more compact notation

dx2
ds2 = −c2 dτ 2 = −c2 dt2 + a2 (t)( + x2 dΩ2 ) (36)
(1 − kx2 /R02 )
where now
x = σ/R0
and
dΩ2 = dθ2 + sin2 θdφ2
The metric can also be expressed with the radial coordinate r instead of x

ds2 = −c2 dτ 2 = −c2 dt2 + a2 (t)(dr2 + Sk (r)2 dΩ2 ) (37)

The coordinates used here (x, r etc.) are comoving coordinates. The x and r
have dimensions of length while a is dimensionless and has the value a(t0 ) = 1
today.
The time dependent proper distance is
Z r
dp = a(t) dr = a(t)r (38)
0

14
We obtain (see eq. 17)

R0 sin−1 (x/R0 ) k = 1



dp = a(t)r(x) = a(t)  r k=0 (39)

R0 sinh−1 (x/R0 ) k = −1
The proper velocity vp ≡ d˙p can be expressed as

ȧ(t)
d˙p = ȧ(t)r = dp = H(t)dp (40)
a(t)
For small r, Hdp ≈ Hr. This relation is called Hubbles law (Hubble 1929).
H is the Hubble parameter which at the present moment is called the Hubble
constant ≡ H0 . A recent determination (Riess et al. 2011, ApJ 730, 119)
using the Hubble telescope, gives H0 =73.8±2.4 km s−1 Mpc−1 . Instead of
H, it is common to use the related dimensionless parameter h = H/100.

4.2 Cosmological redshift

One of the fundamental observational supports of the Big Bang universe is
the redshift of distant galaxies, manifested in the Hubble law. The most
popular interpretation is that the redshift is caused by the expansion of the
universe. This may be tested using the Tolman test. The prediction is that
the surface brightness of an object should scale as (1+z)−4 (see above). The
problem with performing the test is however that we have to use galaxies at
large distances, which are affected by evolutionary effects for which we have
limited information. Alternative interpretations of the redshift have been
discussed (e.g. as a ’tired light’ effect or the ’new physics’ interpretation of
the redshifts of certain QSOs by Halton Arp and colleagues).
The trajectory of a light ray satisfies the equation

c2 dτ 2 = 0 = c2 dt2 − a(t)2 dr2 (41)

Assume that the emission takes place at time te and reaches us at t0 If two
crests of the lightwave are emitted at te and te +∆te and are received at t0
and t0 +∆t0 then
Z t0 Z r
dt
c = dr (42)
te a(t) 0

Since the relative change in R during the time span between te and te +∆te
is negligible, approximately the same condition holds for the next crest:

15
 Z t0 +∆t0
dt
c (43)
te +∆te a(t)

c∆t0 c∆te ∆t0 a(t0 )

− ≈0⇒ = (44)
a(t0 ) a(te ) ∆te a(te )
Let the wavelength of the emitted radiation be λe and that of the received
λ0 . Then the cosmological redshift is obtained from

λ0 c∆t0 a(t0 ) 1
1 + zcosm = = = = (45)
λe c∆te a(te ) a(te )
In addition to this redshift we also have to take into account the local redshift
zlocal , due to the peculiar motion of the galaxy, typically of the order of a few
hundred km s−1 , and the gravitational redshift, zgrav , e.g. when observing
regions close to a black hole in the centre of active galactic nuclei:

(1 + zobs ) = (1 + zgrav )(1 + zlocal )(1 + zcosm ) (46)

If we expand a(t) around a(t0 ) (see Ryden, chap. 7) we find

1 1
a(t) = 1+(te −t0 )ȧ(t0 )+ (te −t0 )2 ä(t0 )+... = 1+H0 (te −t0 )− q0 H02 (te −t0 )2 +...
2 2
(47)
q0 is the decelerationparameter:
äa ä
q0 = −( 2
)t=t0 = −( )t=t0 (48)
ȧ aH 2
Inverting equation 47, we find a relation between the Hubble constant, the
redshift, the deceleration parameter and the timespan which we will use later:
1
z = H0 (t0 − te ) + (1 + q0 )H02 (t0 − te )2 + ... (49)
2
and
1 1
t0 − te = [z − (1 + q0 )z 2 + ...] (50)
H0 2

16
4.3 Distances
We have already discussed the coordinate distance x, the dimensionless co-
ordinate distance σ and the proper distance dp . Let us look at how different
ways of determining distances link together. We are used to being able to de-
termine distances using the apparent brightness of an object. The apparent
brightness is simply proportional to the distance−2 . In a curved spacetime
this is not necessarily so. The luminosity distance, dL , can be defined as
L 1/2
dL = ( ) (51)
4πf
where L is the total luminosity (e.g. in Watt) and f is the apparent flux
(e.g. in Watt/m2 ), as we astronomers call it. In a R-W universe the light is
spread out across a sphere of proper radius dp (t0 ) = r and a proper surface
area, Ap , of

Ap (t0 ) = 4πSk (r)2 (52)

The cosmological redshift will decrease the energy of each photon with a
factor (1+z). In addition, the proper distance between the photons, and
thus time between the reception of the photons, will also increase with a
factor of (1+z). These two effects will cause a decrease of the measured
luminosity with a factor of (1+z)−2 . Thus the observed flux will be
L
f= (53)
4πSk (r)2 (1 + z)2
and luminosity distance to an object will be
L 1/2 σe
dL = ( ) = Sk (r)(1 + z) = R0 (54)
4πf a(te )
Likewise we may define an angular diameter distance dA of an object of di-
ameter l and apparent diameter δΘ, which is equal to the coordinate distance
Sk (r) at the time of emission
l
dA = = Sk (re ) = Sk (r)/(1 + z) = R0 σe a(te ) (55)
δΘ
q
We can also define the parallax distance dP = R0 σe / 1 − kσe2 and the proper
motion distance dM = R0 σe
Notice the simple relationships dA /dL = (1 + z)−2 and dM /dL = (1 + z)−1
These distance measures are the same at small distances d ∼ r.

17
The surface brightness s of an object with apparent diameter δΘ can be
expressed as

4f L Ld2A
s= = = ∝ (1 + z)−4 (56)
πδΘ2 4πSk (r)2 (1 + z)2 πδΘ2 π 2 Sk (r)2 (1 + z)2
This is the basis of the Tolman test.
If we express σe in t0 and te and expand, using eq. 49 we obtain:

LH02
l= [1 + (q0 )z + ...] (57)
4πz 2

m − M = 25 − 5logH0 + 5logcz + 1.086(1 − q0 )z + ... (58)

This m-z-relation has been much used in cosmology to determine the param-
eters H0 and q0 under the assumption that M is idependent of z. H0 may
then be determined at low redshifts but sufficiently large so it reflects the
general cosmic expansion (z≥0.01). q0 can be obtained from the shape of the
relationship at high z. Thus small z ⇒ H0 and larger z ⇒ q0 .

4.4 Horizons
The speed of light limits our vision of the universe. Below we will define two
types of horizons related to this fact.

4.4.1 The particle horizon or object horizon

Let us find an expression for the proper distance to the so called particle
horizon, dP H , i.e. the distance to the most remote object in the universe we
can observe today. That object emitted its light  after the Big Bang at time
t=0. The trajectory of a light ray emitted from the object at this moment
fulfills the condition dτ =0 ⇒

cdt = a(t)dr (59)

⇓
Z t0
dt
dP H = c (60)
0 a(t)

18
4.4.2 The event horizon
Under certain circumstances the properties of the universe are such that there
are objects that are so distant that we will never be able to see them. They
are outside the event horizon. To find an expression which we can use to
calculate this distance we simply take tmin and tmax as integration limits in
eq. 60. If the universe expands forever, tmax =∞ and if the expansion stops
and the universe starts contracting, tmax is the moment of the ‘Big Crunch’
or the ’gnaB giB’.

19