Lecture Notes Schwarzschild Geometry

Download as pdf or txt
Download as pdf or txt
You are on page 1of 27

The Schwarzschild Solution

Sjors Heefer
June 5, 2023

These lecture notes contain the material for lectures 8 and 9 of the course General
Relativity (3ERX0), corresponding roughly to chapter 9 of the book [1].

Contents

1 Introduction 2

2 The Schwarzschild Solution 3


2.1 The general static spherically symmetric line-element . . . . . . . . . . . 3
2.1.1 Staticity: invariance under time translation and reversal . . . . . 3
2.1.2 Spherical symmetry . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.2 The exterior solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.3 Birkhoff’s theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

3 Particle Trajectories 9
3.1 Geodesic equations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
3.2 The effective potential . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
3.3 Circular orbits of massive bodies . . . . . . . . . . . . . . . . . . . . . . . 11
3.4 Photon trajectories . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
3.5 Gravitational Redshift . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

4 Some Exercises 17

5 Solutions to the Exercises 20

1
1 Introduction
When Einstein completed his general theory of relativity in 1915, the German physicist
and astronomer Karl Schwarzschild had been serving as a Lieutenant of the German
artillery in World War I. Due to an autoimmune skin disease, however, he was now stuck
in a hospital at the Russian front. In this period he studied Einstein’s results and only
about a month after Einstein had published his field equations Schwarzschild managed
to find the very first exact solution (other than flat Minkowski space), now called the
Schwarzschild solution. It describes the gravitational field surrounding a static, spherical
mass distribution, such as a planet, a star or even a black hole. Only a few months later
Schwarzschild died.

In Chapter 1 of these lecture notes we will derive, using the symmetries of the problem
and Einstein’s field equations, the form of the Schwarzschild metric.

In Chapter 2 we will discuss some of the phenomenological implications of the Schwarzschild


solution: the trajectories of freely falling objects and light rays (i.e. geodesics of the
Schwarzschild geometry), redshift, and gravitational time-dilatation.

2
2 The Schwarzschild Solution
2.1 The general static spherically symmetric line-element
In this chapter we will derive the form of the gravitational field surrounding a stationary
and spherically symmetric mass distribution like a star or a planet. It is to be expected
that the symmetries of the mass distribution carry over to the gravitational field, so
before we turn to the Einstein field equations (EFEs) the first thing we will do is make
use of these symmetries to reduce the degrees of freedom in the metric. In fact we will
find that the metric tensor will be completely determined by only 2 functions of a single
variable. Only then will we use the EFEs in order to determine the form of these two
functions. The relevant symmetries to impose are staticity, which is essentially time-
invariance, and spherical symmetry, which needs a particularly careful treatment in the
general relativistic setting.

2.1.1 Staticity: invariance under time translation and reversal


Given a stationary mass distribution, we expect the geometry of spacetime to be constant
in ‘time’. But given what we know about the relativity of (space and) time, we should
be very precise about what we actually mean by this. The proper statement to make is
that we require that there exists some coordinate system (x0 = t, x1 , x2 , x3 ) such that
the line-element is invariant under time translations t → t − a. As you might expect,
and as proven in Exercise 1, this is equivalent to the property that in these coordinates,
gµν does not depend on the time coordinate t. The line-element will thus in general be
of the form

ds2 = gtt (x)dt2 + 2gti (x)dtdxi + hij (x)dxi dxj , x = (x1 , x2 , x3 ).

A spacetime that satisfies this property of time-independence is said to be stationary.

We also expect the geometry to be invariant under time-reversal1 . In the line-element


above, all terms are time-reversal invariant except the dtdxi term, which acquires a mi-
nus sign under the transformation. Hence ds2 can only be invariant under t → −t if
gti (x)dtdxi = −gti (x)dtdxi , implying that gti (x)dtdxi = 0. Thus all dtdxi terms must
vanish. A spacetime that is time-reversal invariant as well as stationary is said to be
1
This has to do with the fact that our mass is assumed to be non-rotating. If it were rotating, a
change t → −t would flip the rotation direction of the mass and might therefore also flip the gravitational
field in some way. For a non-rotating body we don’t expect this to happen.

3
static. Our results so far can therefore be summerized as follows.

In a static spacetime the line-element must be of the following general form

ds2 = gtt (x)dt2 + hij (x)dxi dxj , x = (x1 , x2 , x3 ). (1)

2.1.2 Spherical symmetry


Spherical symmetry in the general relativistic context is a slippery concept because we
are working in an a priori arbitrarily curved space with arbitrary coordinates. What do
we even mean by spherical symmetry if we are not in Euclidean space? Below we give
an intuitive approach to the concept2 .
The first step is to view our space as foliated by spherical shells, or simply spheres. We
know that the line-element on a sphere is given by3

ds2 = −r2 dθ2 + sin2 θdϕ2 =: −r2 dΩ2 ,




where θ, ϕ are the usual angular coordinates. The coordinate r is usually interpreted as
the radius of the sphere, however, this interpretation is based on the conventional view
that the sphere is embedded in Euclidean three-dimensional space! In our present situ-
ation the spheres will be embedded in some general, as of yet unknown geometry, which
might (and will) turn out to be curved. So we should not interpret r as a conventional
radius as we usually would. Its sole purpose at this stage is to characterize all the differ-
ent possible spherical surfaces. We will come back to its actual physical meaning later.
This foliation of space into spheres then gives us a preferred set of spatial coordinates
r, θ and ϕ, in addition to the preferred time coordinate t that we already obtained from
staticity, so that Eq. (1) now reads

ds2 = gtt (r, θ, ϕ)dt2 + grr (r, θ, ϕ)dr2 + 2grθ (r, θ, ϕ)drdθ + 2grϕ (r, θ, ϕ)drdϕ + dΩ̃2 ,
2
The more mathematically oriented mind might be happy to know that a perfectly precise definition
of spherical symmetry can be formulated with the help of advanced Lie group theory and differential
geometry. Following [2], a (pseudo-)Riemannian manifold is said to be spherically symmetric if the
rotation group SO(3) acts on it as isometry (sub)group, in such a way that each orbit (i.e. the collection
of points that can be reached from a given starting point by applying rotations) is a spacelike, simply-
connected, complete two-dimensional surface. Given this definition a theorem can be proven rigorously,
stating that any spherically symmetric line-element must indeed be of the form (2) in certain local
coordinates. That is, the rigorous mathematical approach is completely consistent with ours. Details
can be found in [2, 3, 4].
3
The minus sign comes form the fact that we are using a (+, −, −, −) spacetime signature, leading to
a (−, −, −) spatial signature, whereas usually if we consider a sphere in R3 we use (+, +, +) signature.
This gives an overall minus sign.

4
where dΩ̃2 represents the terms involving dθ2 , dθdϕ and dϕ2 . Note that by setting t =
const and r = const (so that dt = dr = 0) we find that the spherical shell at the given
value of t and r, which we argued above should satisfy ds2 = −r2 dΩ2 , has the line-
element ds2 = dΩ̃2 , so that we can immediately deduce that dΩ̃2 = −r2 dΩ2 .
By spherical symmetry we now simply mean that the line-element should be invariant
under rotations4 . Consider first the drdθ and drdϕ terms. We can rewrite
grθ (r, θ, ϕ)drdθ + grϕ (r, θ, ϕ)drdϕ = dr (grθ (r, θ, ϕ)dθ + grϕ (r, θ, ϕ)dϕ) =: dr ω.
Here dr is spherically invariant, but ω is a covector (field). Just like a vector defines a
preferred direction, a covector defines a preferred hyperplane through the origin, which
breaks spherical invariance just as well as a preferred direction would. Therefore the
term drω cannot be spherically invariant and has to vanish. The metric thus reduces to
ds2 = gtt (r, θ, ϕ)dt2 + grr (r, θ, ϕ)dr2 − r2 dΩ2 .
Last but not least, spherical symmetry now requires that gtt and grr cannot depend on θ
and ϕ, because dt and dr do not change under rotations. In summary, we can conclude
the following:

In a static, spherically symmetric spacetime the metric must be of the form

ds2 = A(r)dt2 − B(r)dr2 − r2 dθ2 + sin2 θdϕ2 .



(2)

We have thus reduced our problem to the determination of just two functions, A(r) and
B(r), of the single coordinate r.

2.2 The exterior solution


We have shown above that the most general static spherically symmetric line-element is
of the form (2). To find A(r) and B(r) we turn to Einstein’s field equations. Note that
A(r) > 0 and B(r) > 0 because we want t to be a timelike coordinate5 and the signature
of the metric to be Lorentzian. Outside of the mass distribution the energy-momentum
tensor vanishes, so we can use the following form of the Einstein field equations in vacuum
Rµν = 0.
4
Rotations act on the preferred coordinates (t, r, θ, ϕ) in the usual way; in particular they transform
only the angles and leave r and t unchanged.
5
What we mean by this is that the 4-velocity of a worldline that goes only in the t-direction should
be timelike.

5
Given the line element (2) one can calculate the the Christoffel symbols, the Riemann
tensor and finally the Ricci tensor. We will not reproduce this straightforward but tedious
calculation here, but in the end one finds that the only non-trivial components of the
Ricci tensor, and hence the Einstein equations, are given by

A′′ A′ A′ B ′ A′
 
Rtt = − + + − = 0,
2B 4B A B rB
A′′ A′ A′ B ′ B′
 
Rrr = − + − = 0,
2A 4A A B rB
 ′
B′

1 r A
Rθθ = − 1 + − = 0,
B 2B A B
Rϕϕ = Rθθ sin2 θ = 0.

Note that the last equation is redundant as it is implied by the third. Note also that we
can write
1 A′ B ′
 
A 1 1
0 = Rtt + Rrr = − + =− (A′ B + AB ′ ) = − (AB)′ ,
B r A B rAB rAB

so that

AB = α

for some constant α. Substituting B = α/A in the θ-equation it reduces to


 ′
A rA A A rA′ 1
0 = Rθθ = − 1 + 2 = −1+ = (rA)′ − 1
α 2α A α α α

so that rA = α(r + k), with k another integration constant6 and we can conclude that
   −1
k k
A(r) = α 1 + , B(r) = 1+ .
r r

This leads to the line-element


   −1
2 k 2 k
dr2 − r2 dθ2 + sin2 θdϕ2 ,

ds = α 1 + dt − 1 + (3)
r r
6
Note that the actual integration constant is k̃ = αk, so that the notation with k is only possible in
general as long as α ̸= 0. It is justified here, however, because we will see that indeed α ̸= 0.

6
where the constants α and k are still to be determined.

Asymptotically, as r → ∞, this reduces to

ds2 = α dt2 − dr2 − r2 dθ2 + sin2 θdϕ2 .



(4)

The requirement that sufficiently far away from the source the gravitational ‘force’ is
negligible and hence that spacetime should reduce to flat Minkowski space in this limit
yields7 α = c2 . More generally, by expanding to first order in k/r we note that for large
values of r the metric becomes approximately equal to

   
2 2 k 2 k
dr2 − r2 dθ2 + sin2 θdϕ2 .

ds = c 1 + dt − 1 − (5)
r r

Recall (see Hobson Eq. 7.8) that in the weak field limit, in order to have the correct
Newtonian limit, we must have g00 = 1 + 2Φ/c2 = 1 − 2GM/c2 r. But note that presently
we have an additional factor c2 , as we are working with the coordinate t instead of the
coordinate x0 = ct. Hence gtt = c2 (1 − 2GM/c2 r). Requiring that in the weak field
limit general relativity reduces to Newtonian gravity we thus find that k = −2GM/c2 ,
and we conclude the following:

The gravitational field outside a static, spherical mass distribution is described by


the Schwarzschild metric:

   −1
2 22GM 2 2GM
dr2 − r2 dθ2 + sin2 θdϕ2 .

ds = c 1 − 2 dt − 1 − 2 (6)
cr cr

Notice that something peculiar happens at a coordinate radius r = rs := 2GM c2


. Here
8
the metric seems to blow up and everything seems to break down. For a typical star
like our sun, however, the surface is located at a radius much larger than this so-called
7
Another way of looking at it is that different values of α simply correspond to different choices of
units in which we measure time, since we can change α by scaling the t-coordinate.
8
However, as will be discussed later, by choosing a different set of coordinates, it is possible to
describe this part of spacetime perfectly well. That is, r = rs is only a so-called coordinate singularity,
not a true singularity.

7
Schwarzschild radius rs . Only for extremely compact objects – objects with an enormous
mass compared to their size – will this radius play a role. This would lead us into the
realm of black holes; these will be discussed later in the course. (See Hobson chapter
11).

2.3 Birkhoff’s theorem


Birkhoff’s theorem, which we will not prove here, states that the Schwarzschild solu-
tion is the only spherically symmetric solution to Einstein’s field equations in vacuum.
Note the absence of the word ‘static’. Indeed, staticity turns out to be a superfluous
assumption in our derivation above; it is automatically implied by spherical symmetry.
In particular, this implies that there exist no ‘time-dependent’ solutions. An important
consequence of the theorem is that a purely radially pulsating star cannot emit gravita-
tional radiation, because outside of this star such gravitational radiation would amount
to a time-dependent spherically symmetric spacetime geometry in (approximate) vac-
uum, which, according to the Birkhoff’s theorem, cannot be consistent with Einstein’s
field equations.

8
3 Particle Trajectories
3.1 Geodesic equations
Given the Schwarzschild line-element derived in the previous chapter, it is in principle
straightforward to calculate its Christoffel symbols and consequently the geodesic equa-
tions (or one may use the method with the Euler-Lagrange equations, which is not as
tedious, and it is in fact a very good exercise to do this yourself). We do not reproduce
the calculation in detail here, this can be found in (Hobson §9.5). The geodesic equations
for the Schwarzschild geometry read
 

1− ṫ = k, (7)
r
−1 −2
µc2 2
 
2µ 2µ µ 2 
2 2 2

1− r̈ + 2 ṫ − 1 − ṙ − r θ̇ + sin θϕ̇ = 0, (8)
r r r r2
2
θ̈ + ṙθ̇ − sin θ cos θϕ̇2 = 0, (9)
r
r2 sin2 θϕ̇ = h, (10)

where µ = GM/c2 , and k and h are constants. It turns out that all geodesics in the
Schwarzschild geometry are planar, i.e. lie in a 2D plane through the origin. Should you
be wondering why this is the case, Exercise 3 guides you through a proof. In what follows
we will take the result for granted. By spherical symmetry we may then simply chose
the orbital plane to be the θ = π/2 plane. Throughout the remainder of this section we
will therefore set θ = π/2, and, unless otherwise stated, in any exercise you may do so
as well. The geodesic equations then simplify to

 

1− ṫ = k, (11)
r
−1 −2
µc2 2
 
2µ 2µ µ 2
1− r̈ + 2 ṫ − 1 − ṙ − rϕ̇2 = 0, (12)
r r r r2
r2 ϕ̇ = h. (13)

The last ingredient needed to completely describe the movement of particles and light in
the Schwarzschild geometry is the correct normalization of geodesics, gµν ẋµ ẋν = c2 for
massive particles or gµν ẋµ ẋν = 0 for massless particles.

9
In summary, in the Schwarzschild geometry we have the following equations of motion:

Schwarzschild Equations of Motion

Schwarzschild geodesics:
 

1− ṫ = k, (14)
r
r2 ϕ̇ = h, (15)
−1 −2
µc2 2
 
2µ 2µ µ 2
1− r̈ + 2 ṫ − 1 − ṙ − rϕ̇2 = 0, (16)
r r r r2

Schwarzschild normalization equation:


   −1  2
2 2µ 2 2µ 2 2 2 c (massive particles)
c 1− ṫ − 1 − ṙ − r ϕ̇ = . (17)
r r 0 (massless particles)

The constants k and h represent a kind of energy and angular momentum, respectively,
as will be explored in Exercise 1.

As a remark we note that in a lot of cases, Eq. (16) is in fact redundant and the two
other geodesic equations (14),(15) together with the somewhat simpler normalization
equation, Eq. (17), suffice to completely describe the motion. (See Exercise 2 for a
proof.)

3.2 The effective potential


Let us do the following calculation for massive and massless particles simultaneously
by writing gµν ẋµ ẋν = εc2 with ε = 1 and ε = 0 for massive and massless particles,
respectively. Substituting Equations (14) and (15) in (17) and multiplying by (1 − 2µ/r)
yields

2µ h2
   
2 2 2 2µ
c k − ṙ − 1 − = 1− εc2 ,
r r2 r

10
or
2GM h2 2GM
 
2
ṙ + 1 − 2 2
− ε = c2 (k 2 − ε),
cr r r
which we recognize as being very similar to the equation governing a Newtonian particle
in a 1-dimensional potential V (r), namely we have

1 2 c2 (k 2 − ε) GM h2 GM h2
ṙ + V (r) = = const, V (r) = − ε+ 2 − 2 3 .
2 2 r 2r cr
(18)

This equation reveals an enormous amount of information about the possible orbits in
the Schwarzschild geometry. Just as in Newtonian mechanics, another derivative (in the
present case with respect to the affine parameter) reveals that9

r̈ = −V ′ (r). (19)

This means nothing else than that particles feel an effective ‘force’ towards the direction
of smaller effective potential. The shape of the potential is shown in Figure 3.2 for
different values of the parameters. These plots already give us a good qualitative idea of
the possible orbits. Something peculiar we can infer immediately from the local maxima
is that even for massless particles there is apparently a circular orbit! The union of
all such orbits forms a sphere, called the photon sphere. In principle then, if you could
get to this radius, it would be possible to see the back of your own head by looking
around the massive body. However, as shown in Exercise 7, the coordinate radius of the
photon sphere is given by r = 3µ, so for a typical star like our sun, for which 3µ ≪ R⊙ ,
this orbit is irrelevant, as it would lie underneath the sun’s surface. We will discuss
circular orbits of massive particles, and by (very good) approximation, massive bodies
like planets moving around a star, in the next section.

3.3 Circular orbits of massive bodies


For circular orbits we have r = const, so from the viewpoint of the effective potential we
are interested in the points where V ′ (r) = 0. For stable circular orbits we additionally
9
Actually, you might object that this argument only works when ṙ ̸= 0. And you would be right.
Nevertheless (19) always holds, also when ṙ = 0, which can be seen by substituting the ṫ and ϕ̇ equation
and (18), in the r̈ geodesic equation, eliminating ṫ, ϕ̇, and ṙ, respectively.

11
Figure 1: These plots, made by Peter Bobbert, show the shape of the effective potental
for different values of h/cµ and for massless as well as massive particles. In the massive
case the exact shape depends on the value of this parameter. For instance, if the value is
less than 12 there is no local minimum, and hence no circular orbits, whereas for larger
values, such orbits do exist. In the massless case, on the other hand, different values
of h/cµ do not lead to fundamentally different shapes. In fact the massless potential is
always a constant multiple of the one for h/cµ = 10 shown in the figure.
12
require that V ′′ (r) > 0, i.e. stable circular orbits occur precisely at the local minima of
the potential. The form of V (r) is such that it has at most one local minimum10 . The
equation V ′ (r) = 0 can be solved for h, giving

2 µc2 r2
h = .
r − 3µ
By positivity of the LHS this shows that all circular orbits must satisfy r > 3µ, and that
if indeed r > 3µ there is precisely 1 orbit (since a given value of r completely fixes ϕ̇
and ṫ, fixing the orbit, at least up to initial conditions.) Substituting the expression for
h into V ′′ (r) we find that

µc2 (r − 6µ)
V ′′ (r) = .
r3 (r − 3µ)
Consequently:

For r > 6µ circular orbits are stable and for 3µ < r ≤ 6µ circular orbits are not
stable. For r < 3µ no circular orbits exist and any massive free-falling particle
crossing this radius will keep falling to smaller and smaller radii.

Now let’s find the remaining components of the 4-velocity for circular orbits. Using
1 2
ṙ + V (r) = c2 (k 2 − 1)/2
2
we find that
1 − 2µ/r
k=p .
1 − 3µ/r
This fixes
 −1

ṫ = k 1 − .
r
Similarly, h, fixed by r, fixes
h
ϕ̇ = .
r2
10
To see this, e.g. write r = 1/u, then V is simply a third order polynomial. which has at most one
local minimum. Note that a local minimum in u corresponds to local minimum in r.

13
Hence the four-velocity of a massive particle in a circular orbit is given by
!
k h
uµ = , 0, 0, 2 .
1 − 2µ
r
r

You can check that we indeed have uµ uµ = c2 , as required for a massive object.

Apart from stability of orbits we can also talk about boundedness. An orbit is said to
be bounded if the particle in the orbit can never reach (spatial) infinity. Looking at the
effective potential equation, this is the case when ϵ = c2 (k 2 − 1)/2 < limr→∞ V (r) = 0.
That is, k < 1. Looking at the expression for k, this implies that circular bound orbits
are the ones for which r < 4µ. Hence a circular orbit with 4µ < r < 6µ is unstable and
not bound, so a particle in this orbit might, due to even the slightest perturbation in its
orbit, fly off to infinity.

3.4 Photon trajectories


As shown in Exercise 7 there is a single circular orbit for massless particles (in the
equatorial and hence any plane), forming the photon sphere. This unstable orbit is
located at r = 3µ. The radial trajectories of photons are treated in Exercise 6.

3.5 Gravitational Redshift


The last topic to be discussed in these notes is gravitational redshift, to be precise: red-
shift due to the gavitational field of a massive spherical body such as a star or a planet.
We consider two stationary observers, an emitter OE at radius rE and a receiver OR at
radius rR . and a photon traveling, along a geodesic, from emitter to receiver. We would
like to find out what, if any, is the difference between the photon energy (equivalently,
photon frequency or wavelength) the two observers measure.

In general the energy of a photon (or of any particle, really) as measured by an observer
is given by

E = gµν uµ pν , (20)

where uµ is the four-velocity of the observer and pµ is the four-velocity of the photon.
This follows from the equivalence principle: in a local intertial frame that is (at some
fixed point along the observer’s worldline) freely falling with the observer, the laws of

14
physics should reduce to those of special relativity. Hence the metric in these coordinates
is simply the Minkowski metric ηµν = diag(1, −1, −1, −1), and since the observer is at
rest in these coordinates (at least at the fixed point on his worldline, not necessarily
away from it because the observer itself need not be freely falling), we moreover have
uµ = (u0 , 0, 0, 0). Consequently the formula for the energy reduces in these coordinates
to E = η00 u0 p0 = cp0 , which is indeed exactly the familiar formula for the energy in
special relativity. So in the local inertial coordinates, Eq. (20) is correct. But since it is
a tensorial equation with no free indices on the LHS, the result is independent of the coor-
dinate system. And hence this formula for the energy is correct in any coordinate system.

Now let’s return to our two observers and let’s consider either of them. Since the observer
is stationary in the Schwarzschild geometry, we have uµ = (u0 , 0, 0, 0) (in our standard
−1/2
Schwarzschild coordinates). The normalization equation then tells us that u0 = gtt .
Hence
1/2
E = gtt p0 .

The photon momentum is given by pµ = dxµ /dσ in terms of some affine parameter11 σ,
dt
so that p0 = dt/dσ. Now recall that, by the geodesic equation (11), gtt dσ = k is constant
along the photon’s worldline. Substituting all this into the formula for the energy, we
find that
−1/2
E = gtt k

and we can compare the two energies the two observers measure as
s
−1/2
ER g (R)k 1 − 2µ/rE µ µ
= tt−1/2 = ≈1− + .
EE gtt (E)k 1 − 2µ/rR rE rR

Note that this result is independent of the exact photon worldline (it does not have to
be radial, for instance). The last approximate equality is valid for values of rE,R that
11
Recall that for any massive particle we can always parameterize the worldline using the proper
time τ , and thus the 4-velocity is given, by definition, by dxµ /dτ and the 4-momentum by mdxµ /dτ .
For a massless particle there is no such thing as proper time, so we cannot use these same definitions.
We can still define a photon 4-velocity by dxµ /dσ, where σ is any affine parameter, but note that
this depends on what affine parameter that we choose. Similarly, the 4-momentum can be defined as
pµ = αdxµ /dσ for some constant α, but by absorbing α into a new affine parameter λ = σ/α, we can
also write pµ = dxµ /dλ in terms of the new affine parameter. What this affine parameter looks like
cannot be inferred from the photon worldline alone, but is determined by the photon’s physical energy
and momentum.

15
are very large compared to µ. On Earth this is an excellent approximation, since the
Schwarzschild radius of the Earth is about a centimeter. If moreover the emitter is
relatively close to the receiver (again relative to µ), rE = rR + δr, we have approximately

ER µ
= 1 + 2 δr.
EE rE

Plugging in some values for the earth,

G = 6.674 · 10−11 m3 kg−1 s−2 ,


M = M⊕ = 5.972 · 1024 kg,
rE = r⊕ = 6378 km,
δr = 22.5 m,

we find that

ER /EE − 1 = 2.45 · 10−15 .

In other words

ER − EE = 2.45 · 10−15 EE .

This is an extremely small difference in energy. Still it is of relevance and can in fact
be measured! In the Pound-Rebka experiment in 1960 an emitter was placed at the
top of the Harvard physics building, sending out γ-photons through radioactive decay,
and a receiver at the bottom, made of atoms of 57 Fe that can essentially only absorb
photons of exactly the emitted frequency. No absorption was detected, as expected from
gravitational redshift. However, subsequently, Pound and Rebka gave the emitter a
velocity, so that the Doppler-shift would, in theory, exactly cancel out the gravitational
redshift. And indeed, now there was absorption, thus confirming the predictions of GR.

16
4 Some Exercises
Exercise 1. Symmetries = conserved quantities.
This exercise aims to illustrate how symmetries and conserved quantities are intimately
linked, leading to a deep physical interpretation of the (at first sight) mysterious con-
stants k and h appearing in the geodesic equations for the Schwarzschild geomety.

Consider a spacetime with metric gµν (t, xi ). By time-translation invariance we mean that
the line-element is invariant under time translations, that is, under coordinate transfor-
mations of the form t → t̃ = t − a, with a constant.
(a) Show that time-translation invariance is equivalent to the property that gµν =
gµν (xi ), i.e. gµν does not depend on t. Hint: keep carefully track of the argument of
gµν when applying the transformation!
(b) Consider a massive particle of mass m moving along a geodesic xµ (τ ). Show that
the covariant form of the 4−momentum pµ satisfies
d m dxα dxβ
pµ = (∂µ gαβ ) .
dτ 2 dτ dτ
(c) Combine (a) and (b) to show that if a spacetime is time-translation invariant then
the covariant energy12 E ≡ pt is conserved along geodesics.
(d) Generalize your findings in (c) to translation-invariance in an arbitrary coordinate,
i.e. not just the t coordinate.
(e) Apply your results to the Schwarzschild metric (6) and provide a physical inter-
pretation of the constants k and h appearing in the geodesic equations for the
Schwarzschild geometry.
Exercise 2. Show that the normalization equation (17) together with the two geodesic
equations (14), (15) imply that the remaining geodesic equation (16) is automatically
satisfied, except possibly when ṙ = 0. Hint: employ the effective force equation, Eq.
(19).
Exercise 3. Particle trajectories are planar. This exercise serves to show that
without loss of generality we may always choose to work in the equatorial plane θ = π/2
when discussing geodesics. (Hence in contrast to most other exercises we initially do not
set θ = π/2 here.)
12
More precisely, this is the energy as measured by a stationary observer at infinity, which would be
a good exercise to check from the general energy formula as well.

17
(a) Show that the quantity
p2ϕ
L2 = p2θ +
sin2 θ
is conserved along geodesics in the Schwarzschild metric, where pµ = gµν ẋν . (We
don’t include the mass m here in the definition of the momentum because we also
want to be able to describe massless particles) Hint: Show that the θ-geodesic equa-
tion, Eq. (9), implies dL2 /dτ = 0.
(b) Check that we may write L2 as
h2
L2 = r4 θ̇2 +
sin2 θ
and show that at θ = π/2, θ̇ = 0, this expression attains a local minimum L2 = h2 , in
the sense that for any other choice of values (close to these values), L2 will necessarily
become larger.
(c) Consider a freely falling particle with initial conditions θ = π/2, θ̇ = 0. Conclude
that this particle will remain forever within the equatorial plane θ = π/2.
(d) Use the previous results together with spherical symmetry to argue that in the
Schwarzschild geometry any freely-falling particle trajectory is planar,
i.e. lies in some plane through the origin.
Exercise 4. Hobson exercise 9.12. A particle is dropped from rest at a coordinate
radius r = R in the Schwarzschild geometry. Obtain an expression for the 4-velocity of
the particle in (t, r, θ, ϕ) coordinates when it passes coordinate radius r.
Exercise 5. Hobson exercise 9.11, slightly modified. A particle is dropped from rest at
infinity in the Schwarzschild geometry.
(a) Find the amount of proper time ∆τ it takes for the particle to travel from a coordinate
radius r0 (with r0 > rs ) to another radius r, where rs = 2µ is the Schwarzschild
radius. Hint: use the result from the previous exercise. You may use the standard
integral

x3/2 2√ x−1
Z
dx = x(x + 3) + ln √
x−1 3 x+1

(b) Find the amount of coordinate time ∆t it takes for the particle to travel from r0 to
r.

18
(c) Conclude that, from the point of view of any static observer, the particle will never
reach r = rs , while from the point of view of the particle itself, it will do so in a
finite amount of proper time.

Exercise 6. Show that radial photon trajectories satisfy

r
c∆t = r + 2µ ln − 1 + const, (outgoing photon)

r
c∆t = −r − 2µ ln − 1 + const. (ingoing photon)

Exercise 7. Use the effective potential to prove that there exists a unique circular
photon orbit (in the equatorial plane). Show that this orbit is located at r = 3GM/c2
and is unstable.

19
5 Solutions to the Exercises
Solution to Exercise 1.

(a) Applying a coordinate transformation t → t̃ = t − a (supplemented by x̃i = xi for


proper bookkeeping), the metric in the new coordinates is given by
∂xα ∂xβ
g̃µν (t̃, x̃i ) = gαβ (t̃, x̃i ) = gµν (t̃, x̃i ) = gµν (t − a, xi ).
∂ x̃µ ∂ x̃ν
Time-invariance is the statement that this transformed metric is the same as the
original one, i.e. gµν (t − a, xi ) = gµν (t, x). Since a can take on any value this just
means that gµν does not depend on t.
(b) Letting the overdot denote d/dτ we start by writing out
d d d
pµ = (gµν pν ) = m (gµν ẋν )
dτ dτ dτ
= m (ġµν ẋν + gµν ẍν )
= m (∂ρ gµν ẋρ ẋν + gµν ẍν ) .
xµ is a geodesic, so it satisfies the geodesic equation
ẍν = −Γναβ ẋα ẋβ .
Plugging this into our expression, and using also the definition of the Christoffel
symbols, yields
d
pµ = m ∂ρ gµν ẋρ ẋν − gµν Γναβ ẋα ẋβ

dτ  
ρ ν 1 νσ α β
= m ∂ρ gµν ẋ ẋ − gµν g (∂α gσβ + ∂β gσα − ∂σ gαβ ) ẋ ẋ
2
 
ρ ν 1 σ α β
= m ∂ρ gµν ẋ ẋ − δµ (∂α gσβ + ∂β gσα − ∂σ gαβ ) ẋ ẋ ,
2
where we have used the fact that gµν g νσ = δµσ . The Kronecker δ ensures that the
whole second part of the expression is non-vanishing only when the dummy index σ
is equal to µ, so we may set σ = µ, leading to
 
d ρ ν 1 α β
pµ = m ∂ρ gµν ẋ ẋ − (∂α gµβ + ∂β gµα − ∂µ gαβ ) ẋ ẋ
dτ 2
 
ρ ν 1 α β 1 α β 1 α β
= m ∂ρ gµν ẋ ẋ − ∂α gµβ ẋ ẋ − ∂β gµα ẋ ẋ + ∂µ gαβ ẋ ẋ .
2 2 2

20
After relabeling the dummy indices (ρ → α, ν → β in the first term and β ↔ α in
the third term) this reads
 
d α β 1 α β 1 α β 1 α β
pµ = m ∂α gµβ ẋ ẋ − ∂α gµβ ẋ ẋ − ∂α gµβ ẋ ẋ + ∂µ gαβ ẋ ẋ
dτ 2 2 2

and we observe that the first three terms cancel each other out, leaving us with the
desired formula
d m
pµ = ∂µ gαβ ẋα ẋβ .
dτ 2

(c) If the spacetime is time-translation invariant then, by (a), gµν does not depend on
t, hence ∂t gαβ = 0 and by setting µ = t in (b) we find therefore that dpt /dτ = 0.
Hence pt is conserved along geodesics.

(d) There is nothing special about the t coordinate. What your results in (a-c) show
is that if the line-element is invariant under translations of a certain coordinate,
then the corresponding component of the covariant four-momentum is a conserved
quantity.

(e) The Schwarzschild metric is invariant under t-translations, and under ϕ-translations,
so the covariant energy and ϕ-component of momentum, which we interpret as an
angular momentum, are conserved quantities. Using the Schwarzschild metric we
compute that
 rs 
E = pt = gtt pt = mgtt ṫ = mc2 1 − ṫ = kmc2 ,
r
pϕ = gϕϕ pϕ = −mr2 sin2 θϕ̇ = −mh.

Thus we conclude that the constants k and h are a measure of the particle’s energy
and ϕ-angular momentum, respectively. k and h are constants because, due to the
symmetries of the Schwarzschild metric, this energy and angular momentum are
conserved quantities along geodesics.

Solution to Exercise 2. As derived in the text, if we eliminate ṫ and ϕ̇ using the


corresponding geodesic equations, the normalization equation gµν ẋµ ẋν = εc2 leads to
(18), i.e. ṙ2 /2 + V (r) = const, another derivative of which yields (r̈ + V ′ (r))ṙ = 0.
Hence, we see that r̈ = −V ′ (r), which is Eq. (19). This argument only works when
ṙ = 0, and although the resulting equation is true for ṙ = 0 as well (as explained in
the text), the geodesic equation (16) that we are currently trying to derive is needed

21
to be able to infer this, so we cannot use this fact right now and hence we have to
assume for the moment that ṙ = 0. Substituting V ′ (r) for its explicit expression and
then eliminating ε using the normalization equation, Eq. (17), leads to

µc2 k 2 µṙ2 h2
0 = r̈ + − − A,
Ar2 Ar2 r3
where we have written A = (1 − 2µ/r). Eliminating the constants k and h using the two
geodesic equations (14), (15) yields

µc2 A 2 µ 2
0 = r̈ + ṫ − ṙ − Arϕ̇2 ,
r2 Ar2
which, after multiplying by A−1 yields the desired remaining geodesic equation (16).

Solution to Exercise 3.

(a) First note that

pθ = gθθ pθ = −r2 θ̇,


pϕ = gϕϕ pϕ = −r2 sin2 θϕ̇ = −h,
ṗθ = −2rṙθ̇ − r2 θ̈,
ṗϕ = 0.

Then we compute

p2ϕ 2p2ϕ
 
d 2 d 2pϕ ṗϕ
L = +p2θ = 2pθ ṗθ + − cos θθ̇
dτ dτ sin2 θ sin2 θ sin3 θ
 
4 2 2
= −2r θ̇ θ̈ + θ̇ṙ − sin θ cos θϕ̇ ,
r

where we recognize the term in parenthesis as one of the geodesic equations, Eq. (9).
As the particle travels on a geodesic this term must vanish, so that
d 2
L =0

and L2 is conserved along the worldline of the particle

22
(b) The rewriting of L2 is just substituting the expressions for pθ and pϕ . Now note that
both terms in in L2 are nonnegative. The first term is minimal for θ̇ = 0 and the
second term is minimal for θ = π/2 or θ = 3π/2. If θ̇ ̸= 0, the first term will grow;
and if θ ̸= π/2 but still close to π/2, the second term will grow. So for all other
values of θ and θ̇, close to the given values, L2 will necessarily become larger.

(c) The particle starts out with L2 attaining its minimal value. But we showed in (a)
that L2 is conserved, so L2 will remain at its minimal value. If the particle were to
move out of the equatorial plane, θ would have to divert from π/2. However, we
have argued in (b) that when this happens, L2 becomes larger, which cannot happen
because L2 is conserver. Hence the particle is constrained to the equatorial plane.

(d) Consider an arbitrary particle trajectory with arbitrary initial conditions. By per-
forming a rotation we may choose our coordinate system such that the initial con-
ditions of the particle become θ = π/2, θ̇ = 0. By (c) then, the particle stays in the
equatorial plane in these coordinates and the trajectory is planar.

Solution to Exercise 4. The two relevant equations here are the normalization equation
(17) and the t-geodesic equation (14),
   −1  −1
2 µ ν 2 2µ 2 2µ 2µ
c = gµν ẋ ẋ = c 1 − ṫ − 1 − ṙ2 , ṫ = k 1 − .
r r r

(Note that because of the statement ‘is dropped from rest’ we can immediately infer that
the particle must be massive, not massless.) At r = R we have ṙ = 0 and hence
 −1
2 2µ
ṫ = 1 − ,
R

which fixes the constant k as


   1/2
2µ 2µ
k = 1− ṫ = 1 − .
R R

Then at coordinate radius r we find that


 −1  1/2  −1
2µ 2µ 2µ
ṫ = k 1 − = 1− 1−
r R r

23
and
   2
2 2 2µ 2 2µ
ṙ = −c 1 − +c 1− ṫ2
r r
   
2 2µ 2 2µ
= −c 1 − +c 1−
r R
 
2 1 1
= 2µc − .
r R
Therefore
s  
1 1
ṙ = − 2µc2 − ,
r R

where the minus sign comes from the fact (due to h = ϕ̇/r2 = 0) that r̈ = −V ′ (r) =
−GM/r2 < 0, so that the particle feels an effective force ‘downward’. Hence the 4-
velocity at coordinate radius r reads
 1/2  −1 s   !
2µ 2µ 1 1
ẋµ = (ṫ, ṙ, 0, 0) = 1− 1− , − 2µc2 − , 0, 0 .
R r r R

Solution to Exercise 5. Taking the limit R → ∞ in the previous exercise we see that
the 4-velocity of the particle is given by
 −1 r !
µ 2µ 2µc2
ẋ = (ṫ, ṙ, 0, 0) = 1− ,− , 0, 0 .
r r

In particular,
r
dr 2µc2
ṙ = =−
Z r dτ Z r
r
− dr = dτ
2µc2
s s
3
2 r0 2 r3
− = ∆τ,
3 2µc2 3 2µc2

where we have integrated from r0 to r and ∆τ is the proper time elapsed during the fall
from r0 to r. Next, in order to find the expression t(r), note that the t component of the

24
4-velocity says that
 −1  
dt 2µ dτ 2µ
= 1− , = 1− .
dτ r dt r

Then
dt dt dτ
=
dr dτ dr
 −1 s 3
2 2µ d r
=− 1−
3 r dr 2µc2
 −1 r
2µ r
=− 1− .
r 2µc2

Hence, writing r = 2µx,


Z  −1 r
2µ r
∆t = − 1− dr
r 2µc2
−1

Z 
2µ 1
=− 1− xdx
c x
x3/2
Z

=− dx
c x−1

2µ 2 √
 
x−1
=− x(x + 3) + ln √
c 3 x+1
" r p #
1 2 r r/2µ − 1
=− (r + 6µ) + 2µ ln p
c 3 2µ r/2µ + 1
" s p #r=r
r3 r/2µ − 1
r
2 4µ r 2µ
=− + + ln p
3 2µc2 c 2µ c r/2µ + 1 r=r
0
s s 
r03
 
r3  4µ
r r
2 r0 r
=  2
− 2
+ −
3 2µc 2µc c 2µ 2µ
p p
2µ r0 /2µ − 1 r/2µ + 1
+ ln p p .
c r0 /2µ + 1 r/2µ − 1

25
Now notice that
s s
2 r03 2 rs3
lim ∆τ = − < ∞,
r→rs =2µ 3 2µc2 32µc2
p p
2µ r0 /2µ − 1 r/2µ + 1
lim ∆t = finite + lim ln p p = ∞.
r→rs =2µ r→2µ c r0 /2µ + 1 r/2µ − 1
Thus it takes a finite amount of proper time for the particle to travel from r0 to rs .
However, it takes an infinite amount of coordinate time. A static observer at a coordinate
distance r = R has ∆τ = (1 − 2µ/R) ∆t, so any static observer will say it takes an infinite
amount of time before the particle reaches r = rs , i.e., it will never reach this radius
from the point of view of a static observer.
Solution to Exercise 6. A null geodesic satisfies ds2 = 0, hence
   −1  −1
2µ 2 2 2µ 2 dt 2µ
1− c dt = 1 − dr ⇒ c =± 1− .
r r dr r
Integrating, we find
Z Z
1 r
c∆t = ± 2µ dr = ± dr
1− r r − 2µ
Z  

=± 1+ dr
r − 2µ
= ± (r + 2µ ln |r − 2µ|) + const
 
r
= ± r + 2µ ln − 1 + const,

where in the last step we have absorbed a term ±2µ ln |2µ| into the constant.
Solution to Exercise 7. The effective potential for a massless particle, Eq. (18), reads
h2 GM h2
V (r) = 2r 2 − c2 r 3 . For a circular orbit r is constant and hence r must be located at a

local extremum of this potential:


h2 3GM h2 h2
 
′ 3GM
0 = V (r) = − 3 + = 4 −r + 2 .
r c2 r 4 r c
Hence r = 3GM/c2 . Taking another derivative, we find that V ′′ (r = 3GM/c2 ) < 0,
hence at this value of r the effective potential does not have a local minimum (rather a
local maximum) and therefore the orbit is unstable.

26
References
[1] M. Hobson, G. Efstathiou, and A. Lasenby, General Relativity: An Introduction for
Physicists. Cambridge University Press, 2006.

[2] J. Ehlers, “Survey of general relativity theory,” in Relativity, Astrophysics and


Cosmology, W. Israel, ed., pp. 1–125. Dordrecht, 1973.

[3] B. Schmidt, “Isometry gropus with surface-orthogonal trajectories,” Z Naturforsch


Sect A 22 (1967) .

[4] H. P. Künzle and H. Bondi, “Construction of singularity-free spherically symmetric


space-time manifolds,” Proceedings of the Royal Society of London. Series A.
Mathematical and Physical Sciences 297 (1967) no. 1449, 244–268.

27

You might also like