SOSE 2016 Advanced General Relativity Lectures 1-12 - July 4, 2016

Advanced General Relativity, Exercises

Lecturer: Prof. L. Rezzolla Tutorial: Dr. De Laurentis

Lecture I
Exercise 1
Using the antisymmetric property of the Riemann tensor we can simply write down
3
3Rα[βγδ] = (Rαβγδ − Rαβδγ + Rαγδβ − Rαγβδ + Rαδβγ − Rαδγβ )
3!
= Rαβγδ + Rαγδβ + Rαδβγ . (1)
Using the properties of the Riemann tensor we can rewrite the Bianchi identity as
∇λ] Rαβ[γν ≡ Rαβγν; ν + Rαβλγ; ν + Rαβνλ; µ . (2)
To construct the Einstein tensor from the Bianchi identities we apply the Ricci contraction
g αµ [Rαβµν; λ + Rαβλµ; ν + Rαβνλ; µ ] = 0. (3)
αβ
Given the property of the covariant derivative applied to the metric g;µ = 0 we can write down
µ
Rαν; λ + (−Rβλ; ν ) + Rβνλ; µ = 0. (4)
And contracting again on the indices ν and β
 
µ µ
R; λ − Rλ;µ + −Rλ; µ = 0. (5)

Using the definition of the Ricci scalar R

(2Rλµ − δλµ R); µ = 0, (6)

and introducing a symmetric tensor Gαβ ≡ Rαβ − 12 g αβ R, we can see that Eq. (6) is equivalent to
∇α Gαβ = 0. (7)

Exercise 2
Given a connecting 4-vector ξ between two neighbouring geodesics, we say that ξ is Lie propagated if its
derivative with respect to u vanishes
Lu ξ = uβ ∇β ξ α − ξ β ∇β uα = 0. (8)
since the Lie derivative vanishes, we can write
uβ ∇β ξ α = ξ β ∇β uα . (9)
In analogy with the Newtonian expression for the geodesic deviation derived during the lecture, the relative
acceleration between of the 4-vector ξ is given by
∇u ∇u ξ α = uβ ∇β (uγ ∇γ ξ α )
= uβ ∇β ξ γ ∇γ uα + uβ ξ γ ∇β ∇γ uα . (10)

1
We can use the expression ∇β ∇γ uα − ∇γ ∇β uα = Rαδβγ uδ (see lecture notes)

∇u ∇u ξ α = uβ ∇β ξ γ ∇γ uα + ξ γ uβ ∇γ ∇β uα + ξ γ uβ uδ Rαδβγ . (11)

And applying the Leibniz rule and some index relabelling we obtain

∇u ∇u ξ α = ξ β [∇β (uγ ∇γ uα )] + ξ γ uβ uδ Rαδβγ , (12)

where the term uγ ∇γ uα = 0 (tangent vector to the geodesic). After rearranging the indices we arrive at the
equation for the geodesic deviation

∇u ∇u ξ α = Rαβγδ uβ uγ ξ δ , (13)

the sign change in Eq. (5) on the exercise sheet is again a consequence of the properties of Riemann tensor.

Exercise 3
We start by contracting the Einstein’s field equations with a cosmological constant Λ using g αβ ,

R − 2R + 4Λ = 8πT =⇒ R = −8πT + 4Λ, (14)

We can use this result to rewrite the Einstein’s equations as

1
Rαβ = 8π(Tαβ − gαβ T ) + Λgαβ . (15)
2
If we plug back this expression into Eq. (6) of the exercise sheet, we obtain

R = κ2 T + 2R − 4κ1 =⇒ R = 4κ1 − κ2 T, (16)

similarly
1
Rαβ = κ2 (Tαβ − gαβ T ) + κ1 gαβ . (17)
2
In the weak-field limit the componentes of the stress energy-tensor T 00  T 0j  T ij , the 00 covariant
component of the previous expression becomes
1
R00 = κ2 (T00 − g00 T ) − g00 Λ0 (18)
2
where we absorbed κ1 in the definition of the cosmological constant. With a bit of manipulation we can
derive
1 T 00 ρc2
κ2 (T00 − g00 T ) = T00 ≈ η0β η0α = κ2 . (19)
2 2 2
Eq. (18) becomes
ρc2
R00 = κ2 + g00 Λ0 , (20)
2
the left-hand side can be computed in the weak field regime as
1 1
R00 = − hi00,i = − ∇2 h00 ; (21)
2 2
where hµν is a correction to the flat space-time metric, gµν = ηµν + hµν . We need to relate h00 to the
Newtonian potential φ and we can do so by using the geodesic equation for a freely falling particle
2 µ
x dxα dxβ
2
+ Γµαβ = 0, (22)
dτ dτ dτ

2
 ¨ = −∇φ.
and the equation of motion in the presence of a gravitational field ~x ~ Considering only the spatial
2
components and neglecting terms of O(h )

Γµαβ uα uβ ≈ Γµ0 00u u0 = Γµ00 , (23)

and by comparing with the Newtonian equation of motion we obtain

¨ = −Γµ = − 1 ∇2 h00 ;
~x (24)
00
2

from which we can easily derive h00 = −2 cφ2 . If we go back to the Einstein equations,

1 2φ ∇2 φ
R00 = − ∇2 (− 2 ) = 2 , (25)
2 c c
and putting all together with ∇2 φ = 4πGρ, we can compute the value of κ

∇2 φ κρc2
2
= + g00 Λ . (26)
c 2
The Poisson equation is

∇2 φ = 4πG(ρ + ρλ ) , (27)

so that
∇2 φ 4πG κρc2
= (ρ + ρλ ) = +Λ, (28)
c2 c2 2
where ρλ is the ”mass-density“ of vacuum. From this expression we can easily derive the expressions for κ
and Λ
8πG
κ= , (29)
c4
Λc2
ρλ = . (30)
4πG

3
Lecture II
Exercise 1
Starting from the following Lagrangian
1
gµν ẋµ ẋν ,
L= (31)
2
we can obtain the equations of motion most quickly from the above equation as
 
d ∂L ∂L
µ
− = 0, (32)
ds ∂ ẋ ∂xµ
dxµ (λ(s))
where we use the comma followed by an index as an abbreviation for a partial derivative as ẋµ = ds .
I follows immediately that
∂L ∂L
= 2gνµ ẋν and = gνκ,µ ẋµ ẋκ , (33)
∂ ẋµ ∂xµ
then the Euler-Lagrange equations become
1 1
gνµ ẍν + (gνµ,κ − gνκ,µ )ẋν ẋκ = gνµ ẍν + gνµ,κ ẋν ẋκ − gνκ,µ ẋν ẋκ = 0 . (34)
2 2
If we first write the second term in this equation in the form
1
gνµ,κ ẋν ẋκ = (gνµ,κ + gκµ,ν ) ẋν ẋκ (35)
2
after some manipulations with indices we can simply rewrite the previous equation in the following form

gνµ ẍν + (gνµ,κ + gκµ,ν − gνκ,µ )ẋν ẋκ = 0, (36)

and contracting with g να it reduces to well know geodesic equation

ẍα + Γα ν κ
νκ ẋ ẋ = 0. (37)

Exercise 2
In the lectures notes we have shown that the equations governing the geodesic space-time with the line
element
dxα dxβ
ds2 = gαβ (38)
dτ dτ
can be derived from the Lagrangian
1
L = gµν ẋα ẋβ , (39)
2
where the dot represents the derivative with respect the affine parameter τ along the geodesic. When the
particle is massive, τ may be identified with a proper time s of the particle describing the geodesic. For the
Schwarzschild space-time

r2
 
2M
2
ds = − 1 − dt2 + + r2 dθ2 + r2 sin2 θdφ, (40)
r 1 − 2M/r

4
then the Lagrangian is
ṙ2
  
1 2M 2 2
L= 1− ṫ − − r θ̇ − r sin θφ̇ = m2 ,
2 2 2
(41)
2 r 1 − 2M/r
We can calculate the respective canonical momenta as
∂L
pα = (42)
∂ ẋα
the in our case we have that the momenta assume the following forms:
 
∂L 2M
pt = = 1− ṫ, (43)
∂ ṫ r
∂L
pφ = − = (r2 sin2 θ)φ̇, (44)
∂ φ̇
 −1
∂L 2M
pr = − = 1− ṙ, (45)
∂ ṙ r
∂L
pθ = − = r2 θ̇. (46)
∂ θ̇
The resulting Hamiltonian is

H = pt ṫ − (pr ṙ + pθ θ̇ + pφ ṗhi) − L = L . (47)

The equalities of the Hamiltonian and Lagrangian signifies that there is no ”potential energy” in the problem:
the energy is derived solely from the ”kinetic energy” as is, indeed manifest from the (39) for the Lagrangian.
The constancy of the Hamiltonian and of the Lagrangian follows from the fact:

H = L = const. . (48)

By rescaling the affine parameter τ , we can rearrange that 2L is equal to m2 = +1. Further integrals of the
motions follow from the
dpt ∂L dpφ ∂L
= = 0, and = = 0. (49)
dτ ∂t dτ ∂φ
Thus,
 
2M dt
pt = 1− = const. = E , (50)
r dτ
and
dφ
(r2 sin2 θ) = const. . (51)
dτ
While from the equation of motion for the θ coordinate
 2
dpθ d 2 ∂L dφ
= (r θ̇) = − = r2 sin θ cos θ . (52)
dτ dτ ∂θ dτ
We are left with the radial geodesic
" −1 # "  −1 #
dpr d 2M ∂L M 2M
= 1− ṙ = − = r(θ̇2 + sin2 θφ̇2 ) − 2 ṫ2 − 1 − 2M ṙ2 . (53)
dτ dτ r ∂r r r

5
Exercise 3
We introduce an orthonormal tetrad eα̂
µ such that the scalar product of the basis vectors constitute the flat
metric ηα̂β̂

eα̂ · eβ̂ = ηα̂β̂ , (54)

α̂ α̂ α̂β̂
ω̃ · ω̃ = η . (55)

The components of the tetrads are

 −1/2
2M
et̂ = 1− et (56)
r

 1/2
2M
er̂ = 1− er (57)
r

1
eθ̂ = eθ (58)
r

1
eφ̂ = eφ . (59)
r sin θ
The energy of a particle of mass m measured by an observer with 4-velocity u is given by

Elocal − p · u = −pα uα
−pα et̂ =
 −1/2
2M
= −pα 1 − et
r
 −1/2
2M
= −pt 1 −
r
 −1/2
2M
= E 1− , (60)
r

where we have used the expression for the 4-momenta pt = − 1 − 2M

r ṫ = −E.
Similarly for the angular velocity measured by a locally static observer

pφ̂ pα eα
φ̂
v φ̂ = =
pt̂ pt eα
t̂
pφ /(r sin θ) l/(r sin θ)
= = , (61)
Elocal Elocal

and we can find the requested expression l = v φ̂ r sin θ Elocal .

Exercise 4
Remembering that
 
2M dt
pt = 1 − = const. = E , (62)
r dτ

6
 where E is the energy, and

dφ
(r2 sin2 θ) = const. = l . (63)
dτ
and l is the angular momentum.
To derive the circular orbits we use the above equations and the constancy of Schwarzschild Lagrangian,
that give us

E2 ṙ2 `2
− − 2 = 2L = 1 , (64)
1 − 2M/r 1 − 2M/r r
from the above equation we find the two integrals of motion, isolating the energy per unit mass
 2 
`2
 
dr 2M
+ 1− 1 + 2 = E2 , (65)
dτ r r
and
dφ l
= 2 (66)
dτ r
where for simplicity we put in the equatorial plane (i.e. θ = π/2).
The first equation (65) allow us to define an effective potential
 2
dr
= E 2 − Veff
2
, (67)
dτ
where
`2
  
2M
Veff =− 1− 1+ 2 (68)
r r
The first derivative of the potential give us the circular orbit
`2 `2 M `2 3`2 M
 
dVeff d M M
= − 2+ 3 =− 2 + 3 − , (69)
dr dr r 2r r r r r4
in order to find the circular orbits we compute the zeros of the previous function

M r2 − `2 r + 3`2 M = 0 , (70)

we find the minimum and the maximum of the radius

√ r !
`2 ± l4 − 12`2 M 2 `2 12M 2
r1,2 = = 1± 1− , (71)
2M 2M `2

which gives us the condition for the existence of a circular orbit `2 < 12M 2 .
At the minimum value of the potential we can find a stable circular orbit
r !
`2 12M 2
r1 = 1+ 1− , (72)
2M `2

and obvioussly the unstable ones has a radius of

r !
`2 12M 2
r2 = 1− 1− , (73)
2M `2

7
 √
for the minimum L/M < 2 3 and so rc = 6M is the smallest possible radius of a stable circular orbit in the
Schwarzschild metric. It is called the radius of marginal stability, or ISCO (innermost stable circular orbit).
There are also unbound orbits, which come in from infinity and turn around, and bound but noncircular
ones, which oscillate around the stable circular radius.
So Schwarzschild solution possesses stable circular orbits for

6M < rc (stable) < +∞ , (74)

(75)

and unstable circular orbits for

3M ≤ rc (unstable) ≤ 6M . (76)

8
Lecture III
Exercise 1
Given the Schwarzschild metric
r2
 
2M
ds2 = − 1 − dt2 + dr2 + r2 dθ2 + r2 sin2 θdφ, (77)
r 1 − 2M/r

we see that there are no terms that explicitly depend on the time t and angular variable φ. Recalling that
the Lagrangian is

ṙ2
  
1 2M 2 2 2 2 2
L= 1− ṫ − − r θ̇ − r sin θφ̇ , (78)
2 r 1 − 2M/r

where the dot indicate the derivative with respect the affine parameter τ . The conjugate momenta are
respectively
 
∂L 1 2M
−pt = − = 1− ṫ , (79)
∂ ṫ 2 r

∂L
pφ = − = r2 φ̇ sin2 θ. (80)
∂ φ̇
We can define therefore two killing vectors

ξ = (1, 0, 0, 0) , (81)

η = (0, 0, 0, 1). (82)

The related conserved quantities are found taking the dot product of each vector with the four velocity.
 
2M dt
e = ξ · u = gαβ ξ α uβ = − 1 − , (83)
r dτ

+dφ dφ
l = η · u = gαβ η α uβ = r2 sin θ = r2 . (84)
dτ dτ
given θ = π/2. For a massive particle
m2
L=− , (85)
2
which can be easily proved
∂L
pα = − = gαβ x˙β , (86)
∂ x˙α
pα = g αβ pβ = g αβ gβµ x˙µ = ẋα . (87)

by taking the dot product

pα ˙α ˙β α 2
α = gαβ x x = 2L, pα = −m , (88)
which gives us 2L = −m2 .

9
Exercise 2
A massive particle in a marginally bound orbit should satisfy the following equation

ut = −1 (89)

which is equivalent to u2t = 1 and can be written down as

M r2 /(r − 3M ) (r − 2M )2
  
2 2M
ut = 1 − 1+ = (90)
r r2 r(r − 3M )

which gives a radius for a marginally stable orbit of r = rmb = 4M .

Given the effective potential !

2M
 ˜l
Veff = 1 − 1+ 2 , (91)
r r
as we well know from the previous execises, the maximum of the potential can be found by
∂Veff
= r˜l2 − 2M r2 − 2M ˜l2 = 0 , (92)
∂r

which for rmb = 4M becomes ˜l = 4M .

Exercise 3
Given the following line element

ds2 = −e2ν dt2 + e2ψ (dφ − ωdt)2 + e2µ (dR2 + dz 2 ) , (93)

all the observers whose velocity has at each point the form (for circular, axisymmetric motion)

uα = ut (tα + ωφα ) = eν (tα + ωφα ) , (94)

represent a family of zero-angular-momentum-observers (ZAMOs).

A natural tetrad in Kerr is given by the frame of a ZAMO, the basis covectors have the following
expression

ω 0̂ = eν dt , (95)
ω 1̂ = eψ (dφ − ωdt) , (96)
2̂ µ
ω = e dR , (97)
ω 3̂ = eµ dz , (98)

and the contravariant basis vectors are

e0̂ = e−ν (∂t + ω∂φ ) , (99)

−ψ
e1̂ = e ∂φ , (100)
−µ
e2̂ = e ∂R , (101)
−µ
e3̂ = e ∂z . (102)

10
Lecture IV
Exercise 1
An important feature of the Kerr solution is the frame dragging which arises ultimately because the metric
contains off-diagonal components gtφ = gφt . We can compute this effect by considering the trajectory of a
particle falling with zero angular momentum, pφ . Since gαβ is independent of the coordinate φ, pφ is still a
conserved momenta.
pφ = g φα pα = g φφ pφ + g φt pt , (103)
and for the time component of the momenta

pt = g tα pt = g tt pt + g tφ pφ . (104)

For a massive particle we have

dt
pt = m , (105)
dτ
dφ
pφ = m , (106)
dτ
so that the trajectory can be easily derived
dφ pφ g φt
= t = tt = ω(r, θ) , (107)
dt p g
ω(r, θ) is the angular velocity of a zero angular momentum particle and it measures the frame dragging. To
compute it explicitly we need g αβ , since the metric is diagonal in r and θ
∆
g rr = 1/grr = , (108)
ρ2
1
g θθ = 1/gθθ = 2, (109)
ρ
and we are left with the components of the matrix which depends on φ and t
 −1
gtt gtφ
g −1 = , (110)
gφt gφφ

and given its determinant D = gtt gφφ − (gtφ )2 we are left with the inverse matrix
 
−1 1 gtt −gtφ
g = . (111)
D −gφt gφφ
With some algebras we obtain

D = −∆ sin2 θ , (112)
2 2 2 2 2
(r + a ) − a ∆ sin θ
g tt = , (113)
ρ2 ∆
2M r
g tφ = −a 2 , (114)
ρ ∆
∆ − a2 sin2 θ
g φφ = . (115)
ρ2 ∆ sin2 θ

11
Inserting this result back into Eq. (107) we obtain

g φt 2M ra
ω(r, θ) = = 2 . (116)
g tt (r + a2 )2 − a2 ∆ sin2 θ)

For a stationary observer  

dt
uµobs u0obs , 0, 0, 0)


= = , 0, 0, 0 , (117)
dτ
and writing the normalisation uobs · uobs = −1,

uobs · uobs = g00 (u0obs )2 = −1 . (118)

For the Kerr metric  2

r + a2 cos2 θ − 2M r
  
2M r
g00 = − 1 − 2 =− . (119)
ρ r2 + a2 cos2 θ
which is equal to zero if
r2 − 2M r + a2 cos2 θ = 0 . (120)
This quadratic equation for the radius r gives us
p
re (θ) = M + M 2 − a2 cos2 θ , (121)

if g00 = 0.
If we consider a photon on the equatorial plane (θ = π/2) moving a circular orbit (dθ = dr = 0), the line
element becomes
gtt dt2 + 2gtφ dtdφ + gφφ dφ2 = 0 , (122)
dφ
from which we can compute the angular velocity Ω = dt
q
2 − 4g g
−2gtφ ± 4gtφ
dφ φφ tt
= (123)
dt 2gφφ
s 2
gtφ gtφ gtt
=− ± − . (124)
gφφ gφφ gφφ

By setting gtt = 0 we obtain the minimum value of Ωmin = 0 while the maximum is
2gtφ
Ωmax = − > 0. (125)
gφφ

Exercise 2
Let’s recall the expression for the orbits of a photon in Schwarzschild

2M `2
   
dr 2
=E − 1− , (126)
dτ r r2

and the effective potential can be defined as

2M `2
 
2
V (r) = 1 − , (127)
r r2

and for a static observer

12
 For a photon coming from infinity with energy E, the allowed orbits are the ones for which V is smaller
than E. At the point where E 2 = V 2 we also have
 2
dr
= 0. (128)
dτ

We can evaluate
`2
  
d 2M
1− = 0, (129)
dr r r2
which is satisfied for r = 3M and the potential at this radius is
L
V (r = 3M ) = √ . (130)
3 3M
For a photon
 −1
2M
p0 = 1− E, (131)
r

dr
pr = , (132)
dλ

dφ L
pφ = = 2. (133)
dλ r
From Eq. (133)
dφ dφ dλ L 1
= =± 2q
dλ dλ dr r E 2 − 1 − 2M

`2
r r2
  −1/2
1 1 1 2M
=± 2 2 − 2 1− , (134)
r b r r

where b = L/E is the impact parameter. We compute now the cone of avoidance for an incoming photon
with L > 0. For convenience, we introduce the following variable u = r−1 and we rewrite Eq. (134) as the
following cubic equation
du 1
f (u) = = 2 − u2 + 2M u3 . (135)
dφ b
The equation f (u) = 0 allows for a negative real root while the two remaining ones may be real or complex
and either distinct or coincident (see the lecture notes for the unbound time-like geodesics). The condition
for two coincident roots both real can be obtained by

f 0 (u) = 6M u2 − 2u = 0 , (136)

where the prime indicate the derivative with respect u. The above equation has u = 1/3M as a root; and
u = 1/3M is a root of f (u) = 0 if
√
b2 = 27M 2 or b = (3 3)M .

The product of the roots of Eq. (136) is given by

1
u1 u2 u3 = − , (137)
2M b2

13
from which we can finally compute the three roots of the polynomial equation

1 1
u1 = − and u2 = u3 = . (138)
6M 3M
We can use the roots to factorise Eq. (134)
 2   2
du 1 1
= 2M u + u− . (139)
dφ 6M 3M
Thus if we consider now a bundle of light rays launched from a point at a distance r from the black hole,
they describe a cone with an angle ψ directed toward the black hole
1 dr̃
cot ψ = , (140)
r dφ
where dr̃ is the line element along the generator of the cone
 −1/2
2M
dr̃ = 1 − . (141)
r
If we substitute u = 1/r in Eq. (140)
1 du
cot ψ = − √ , (142)
u 1 − 2M u dφ
and using Eq. (139),
1  r  r 1/2
cot ψ = − p r
1− 1+ . (143)
2M −1 3M 6M
From this last equation it follows that for an ingoing photon

√
3 3
ψ∼ M as r → ∞, (144)
r
π
ψ= for r = 3M , (145)
2
ψ=π for r = 2M . (146)
It can be easily found similar conditions for an outgoing photon.

Exercise 3
Given a special observer (not moving along a geodesic)
uµ = ut , 0, ωA, 0 ,


(147)
for which we can compute the energy of a particle measure by this observer as
EZAMO = −p · u
= − pt ut + pφ uφ

= A(E − wL) , (148)

where A can be found by the normalisation condition u · u = −1
gφφ
A2 = . (149)
−∆ sin2 θ

14
Lecture V
Exercise 1
In this exercise we have to show that the following relations hold for of massive particles in a Kerr spacetime
√
r2 − 2 M r ± a M r
E= p √ , (150)
r r2 − 3 M r ± 2 a M r

√
 √ 
M r r2 ± 2 a M r + a2
L= p √ . (151)
r r2 − 3 M r ± 2 a M r
For a circular orbit the potential can be written as

−(a2 u2 − 2M u + 1) + E 2 + 2M x2 u3 − (x2 + 2aEx)u2 = 0 , (152)

where we have defined x = L − aE and u = 1/r. The first derivative respect to the radial coordinate,

V 0 (r) = −(a2 u − M ) + 3M x2 u2 − (x2 + 2aEx)u = 0 . (153)

The cubic polynomial in Eq. (152) has a double root. From Eq. (153) we obtain

(x2 + 2aEx)u = −(a2 u − M ) + 3M x2 u2 , (154)

and by solving Eq. (152) for E

E 2 = −2M u + 1 − 2M x2 u3 + 3M x2 u3 + M u
= M x2 u3 − M u + 1 . (155)

By eliminating E from the previous equations we obtain an equation fro x

x4 u2 (3M u − 1)2 − 4a2 M u3 − 2x3 u (3M u − 1)(a2 u − M ) − 2a2 u(M u − 1) + (a2 u − M )2 = 0 .

   
(156)

The discriminant of the last equation is 4a2 M ∆2µ u where ∆2µ = 12 u2 − 2M u + 1.

We can rewrite Eq. (156) in a particularly simple form by writing

(3M u − 1)2 − 4a2 M u3 = Q+ Q− , (157)

√
where Q± = 1 − 3M u ± 2a M u3 . Thus we find
Q± ∆µ − Q+ Q− 1
x2 u2 = = (∆µ − Q∓ ) . (158)
Q+ Q− Q∓
The term in brackets on the RHS of the previous equation can be written
√ √
∆µ − Q∓ = u(a u ± M )2 , (159)

and the solution of Eq. (156) takes the following from

√ √
a u± M
x=− p , (160)
uQ∓

15
where the orbit with the minus sign are retrograde while the one with the plus are direct ones. Inserting x
in Eq. (155) we find the expression for the energy
1 h √ i
E=p 1 − 2M u ∓ a M u3 , (161)
Q∓

and easily

√
M h 2 u √ i
L = aE + x = ∓ p a u + 1 ± 2a M u3 , (162)
uQ∓

as requested.

Exercise 2
We set the energy for a particle at spatial infinity (measured by a static ZAMO with four momentum
u = (1, 0, 0, 0)) to be equal to unity and calculate the position of the ISCO (depending on a). Bound circular
orbits computed in the previous exercises are not all stable. Stability requires that V 00 (r) ≤ 0, which yields
to the following conditions

2M
1 − E2 ≥ , (163)
√ 3r
r2 − 6M r ± 8a M r − 3a2 ≥ 0 , (164)

or equivalently to
r ≥ Rms (165)
where Rms is the radius of the marginally stable orbit, and it is given by the following expression
n o
1/2
Rms = M 3 + Z2 ∓ [(3 − Z1 )(3 + Z1 + 2Z2 )] (166)
 
 a 1/3  a 1/3 a 1/3
Z1 ≡ 1 + 1 − 2 1+ ) + (1 − (167)
M M M
 2 1/2
a
Z2 ≡ 3 2 + Z12 . (168)
M

For a = 0, Rms = 6M and for a = 1 Rms = M (direct) or 9M (retrograde). Inserting the expression for
the radius of the marginally stable circular orbit in Eq. (161) we obtain a pretty complicated expression for
the energy loss (see Matthias’ Maple notebook).
The smallest energy drop is for a value of the spin parameter a = 0, as expected, (Eloss = 0.057) while
the largest one is for a = 1 (Eloss = 0.421).

Exercise 3
Given the Lagrangian
1
L= gµν ẋµ ẋν , (169)
2
and the Euler-Lagrange equation for r
d ∂L ∂L
= , (170)
dλ ∂ ṙ ∂r

16
we can write it down as
d 1 ∂gµν µ ν
(grr ṙ) = ẋ ẋ . (171)
dλ 2 ∂r
For a circular orbit we obtain
∂gtt 2 ∂gφt ∂gφφ 2
ṫ + 2 ṫφ̇ + φ̇ = 0 , (172)
∂r ∂r ∂r

and given the angular velocity ω = φ̇/ṫ

∂gφφ 2 ∂gtφ ∂gtt

ω +2 ω+ =0. (173)
∂r ∂r ∂r
Recalling coefficients of the Kerr metric we obtain

r3 − M a2 ω 2 + 2M aω − M = 0 ,


(174)

which has as discriminant

M 2 a2 + M (r3 − M a2 ) = M r3 , (175)

and the solutions are

√
M
ω± = ± √ , (176)
r3/2 ±a M
which for a = 0 the last equation becomes
r
M
ω± = ± . (177)
r3

17
Lecture VI
Exercise 1
Recalling that the area of the horizon for a black hole is of considerable importance because of the area
theorem, which states that the horizon area of a classical black hole can never decrease in any physical
process.
• The Kerr horizon corresponds to a constant value of r = r+
• Since the metric is stationary, the horizon is also a surface of constant t.

• Setting dr = dt = 0 in the Kerr line element gives the line element for the 2-dimensional horizon,
 2
2M r+
dσ 2 = ρ2+ dθ2 + sin2 θdφ2 , (178)
ρ+
2
where ρ2+ is defined by ρ2+ ≡ r+ + a2 cos2 θ
• This is not the line interval of a 2-sphere

Thus the Kerr horizon has constant Boyer-Lindquist coordinate r = r+ but it is not spherical. The metric
tensor for the Kerr horizon is
ρ2+
!
0
g= 
2M r+
 2 . (179)
0 ρ+ sin2 θ

The area of the Kerr horizon is then

Z 2π Z π Z 2π π
√
Z  p 
A= dφ gdθ = 2M r+ dφ sin θdθ = 8πM r+ = 8πM M + M 2 − a2 . (180)
0 0 0 0

We know that Stephen Hawking discovered an important theorem in 1971. This, the area theorem stater
that in the interactions involving black holes, the total surface area of the event horizon of a black hole can
never decrease (in absence of quantum effects); then we have that δA > 0. it can, at best, remain unchanged
(if the conditions are stationary). Furthermore for Penrose process we have also that δM < 0. Now let us
this area theorem to estimate the energy extraction limits. As we shown in the lectures the Penrose process
decreases to zero only the ”reducible” mass leaving behind the ”irriducible” mass Mirr

2 A 1 2 p 4 
Mirr = = M + M − J2 , (181)
16π 2
where J = aM , clearly if J = 0 we have that M 2 = Mirr . If we suppose that te process takes place
adiabatically so that black hole evolves, then

δJ = δ(aM ) = mδa + aδM , (182)

and then the inequality

2r+ M δM ≥ aδJ (183)

gives

2M r+ − a2 δM = r+
2


δM ≤ M aδa (184)

18
Now using the definition of irreducible mass follows that

2 1
δMirr = √ (r+ δM − M aδa) (185)
2 M 2 − a2
Therefore the inequality (184) is equivalent to the restriction
2
δMirr ≥ 0. (186)

In other words, by no continuous process can the irreducible mass of a black hole be decreased, a result which
justifies the designation. Furthermore from the (181) we may also say that by no continuous process can the
surface area of a black hole be decreased. The more general assertion, that no interaction whatsoever among
black hole can result in a decrease of their total surface area is the area theorem of Hawking. Since the best
that can be achieved is to keep the irreducible mass unchanged, process in which it remains constant are
said to be reversible. It may be also be noted that by virtue of the definition (181) we have the relation

4πJ 2 J2
M 2 = Mirr
2
+ 2
= Mirr + 2 (187)
A 4Mirr
2
J
Since Mirr is irreducible, we may interpret the second terms 4M 2 as the contribution of the rotational
irr
kinetic energy to the square of the inertial mass of the black hole, and that it is this rotational energy that
is being extracted by the Penrose process. A further result is

M δM ≥ aδa or δ(M 2 − a2 ) ≥ 0 (188)

Exercise 2
Static and spherically symmetric non?rotating stars therefore generate a spacetime of the following form

ds2 = −e2Φ(r) dt2 + e2λ(r) dr2 + r2 dθ2 + sin2 θdφ2

(189)

The Tolman-Oppenheimer-Volkoff (TOV) equations can be derived in a standard manner by calculating the
Christoffel symbols in Schwarzschild coordinates and from there the Riemann and Ricci tensor. Let start to
α
consider the Einstein tensor, that we well know follows from the Ricci-tensor Rβγ = Rβγδ

1
Gαβ = Rαβ − ηαβ R (190)
2
For the Ricci tensor we find the following expressions
0 0 0
R00 = −R101 − R202 − R303 , (191)

0 1 1
R11 = −R101 − R212 − R313 , (192)

0 1 2
R22 = −R202 − R212 − R323 = R33 , (193)

and
0 0 0 1 1 2


R = 2 R101 + R202 + R303 + R212 + R313 + R323 (194)

19
and therefore for the Einstein tensor
1 1 2


G00 = − R212 + R313 + R323 , (195)

0 0 2


G11 = − R202 + R303 + R323 , (196)

0 0 1


G22 = − R101 + R303 + R313 , (197)

0 0 1


G33 = − R101 + R202 + R212 = G22 , (198)
Inserting the expressions for the curvature, we finally obtain the equations
2λ0
 
1 −2λ 1 1 d h  i
G00 = 2
− e 2
− = 2 r 1 − e(−2λ) , (199)
r r r r dr
2Φ0
  
1 1
G11 = − 2 − e−2λ − , (200)
r r2 r
Φ0 − λ0
 
G22 = G33 = e−2λ (Φ0 )2 − Φ0 λ0 + (201)
r
and all other Gαβ = 0 for α 6= β. This is also a consequence of the high symmetry. The equality of G22
and G33 is a consequence of the isotropy on the sphere, but the radial component G11 will in general differ
from these. As a consequence, Einstein?s equations provide three equations for the two functions Φ(r) and
λ(r). In fact, the third equation contains the hydrostatic equilibrium, since the equations of motion are
not independent in General Relativity. The matter in the interior of the star is described in terms of the
energy?momentum tensor T αβ that assumes the form of a perfect fluid
T µν = ρc2 + P uµ uν + P g µν ,


(202)
where g µν are the covariant components of the metric tensor. In the above equation uµ is the local fluid
4-velocity uµ = dxµ /dτ , where dτ = ds/c. It satisfies the normalisation uµ uµ = −1 (time?like vector field). ρ
is the total mass-energy-density, and P the corresponding pressure. Because the star is static, the 3-velocity
of the vector field vanishes, and due to the normalisation
1 1
u0 = √ = = e−Φ(r) (203)
−g00 α(r)
or u0 = −α(r) is a measure of the redshift factor. For static stars the source of the gravitational field has
the form
Tβα = diag −ρc2 , P, P, P .

(204)
This demonstrates that Einstein?s field equations
8πG
Gαβ = Tαβ , (205)
c4
can explicitly be written as
2λ0
 
1 −2λ 1 8πG
G00 = 2
−e 2
− = 4 ρ(r) , (206)
r r r c
2Φ0
  
1 1 8πG
G11 = − 2 − e−2λ 2
− = 4 P (r) , (207)
r r r c
Φ0 − λ0
 
−2λ 0 2 0 0 8πG
G22 = G33 = e (Φ ) − Φ λ + = 4 P (r) . (208)
r c

20
The first two equations provide us two independent equations for the functions Φ(r) and λ(r)

2λ0
 
1 −2λ 1 8πG
2
−e 2
− = 4 ρ, (209)
r r r c
2Φ0
 
1 1 8πG
2
− e−2λ 2
− =− 4 P, (210)
r r r c
The first equation is equivalent to
8πG 2
(re−2λ )0 = 1 − ρr , (211)
c2
which can be integrated with the asymptotic flatness condition to yield the 3-space metric
2GM (r)
e−2λ = 1 − , (212)
rc2
with the total mass inside radius r given as
Z r
M (r) ≡ 4π ρ(r0 )r02 dr0 . (213)
0

By subtracting the second equation from the first one, we obtain

4πG
e−2λ (Φ0 + λ0 ) = (ρ + P ) r . (214)
c2
This is equivalent to
 
1 GM (r) 4πG
Φ0 = + 4 rP (215)
1 − 2GM (r)/rc2 c2 r2 c
This demonstrates now, how the gravitational force is generalized in General Relativity.
In particular, pressure is a source of the gravitational field, and the Schwarzschild metric acts as a
modification in the denominator of the force-law.
We now solve the first equation for λ0 and set c = 1 = G

−2rλ0 = 1 − 8πr2 ρ e2λ − 1 ,

(216)

and also solve the second equation for Φ0

2rΦ0 = 1 − 8πr2 P e2λ − 1 ,

(217)

and take the derivative of this last equation and multiply by r

2rΦ0 + 2r2 Φ00 = 2rλ0 1 + 8πr2 P + 16πr2 P + 8πr3 P 0 e2λ




(218)

We solve this equation for Φ00 using once again the above two relations

2rΦ00 = 1 + 16πr2 P + 8πr3 P 0 e2λ − 1 + 8πr2 P 1 + 8πr2 ρ e4λ

(219)

We can also square the above relations for Φ0 to obtain

1
2 1
2r2 (Φ00 )2 = 1 + 8πr2 P e4λ − 1 + 8πr2 P e2λ + .


(220)
2 2
These equations provide relations for Φ0 , λ0 , Φ00 and Φ02 in terms of ρ , P , P 0 and e2λ , which can be expressed
entirely in terms of the included mass M (r).

21
 Therefore, all the metric functions can be eliminated from the third of Einstein?s equations by substitution
of the above results.
In this way, we obtain after some lengthy calculations the equation for the relativistic hydrostatic equi-
librium
−1
4πr3 P (r)
  
dP GM (r)ρ(r) P (r) 2GM (r)
=− 1 + 1 + 1 − (221)
dr r2 ρ(r)c2 M (r)c2 c2 r
r
This equation could also be obtained directly from the equations of motion Tr;r = 0. This equation is,
however, a mere consequence of Einstein?s equation and should not be considered as an independent equation.
This demonstrates once again that Φ(r) is the analogue of the Newtonian potential Φ(r)

P 0 = − ρc2 + P Φ0 .


(222)

With the expression for Φ0 this immediately gives the TOV-equation (221).
A further consequence of this derivation is a relation for the gravitational force

dΦ GM (r) 1 + 4πr3 P/M (r)c2

= 2 2 . (223)
dr c r 1 − 2GM (r)/c2 r

For a given equation of state P = P (ρ), the TOV equations can easily be integrated from the origin with
initial conditions M (r) = 0 and an arbitrary value for the central density ρc = ρ(0), until the pressure P (r)
will vanish at some radius R.
To each possible equation of state, there is a unique family of stars parametrized by the central density,
i.e. we obtain a sequence of stellar models M (r) = M (ρc ).

22
Lecture VII
Exercise 1
For an incompressible fluid we have

 = ρc2 , ρ = const. (224)

In such conditions, recalling that the equation for the mass M (r) is
Z
4π 2
M (r) = c2 r dr , (225)

and putting in
2G
e−λ
(ext) = 1 − M (r) , (226)
c2 r
we obtain
−λ(r) r2 3c2
e(int) = 1 − 2 , r02 = ; (227)
r0 8πGρ

from the first of these above equations follows that, since e−λ not must vanish, R2 must be less of r02 , i.e.
2GM (r) RS
= < 1. (228)
c2 R R
This equation tells us that the physical surface of the structure must be external to the pathological area,
i.e the so-called horizon events. The integration of the hydrostatic equilibrium equation
dp 1 dν
+ ( + p) = 0. (229)
dr 2 dr
leads to the result
c4
ρc2 + p = costante × e−ν/2 = De−ν/2 ,


(230)
8πG
where D is a constant. This partial result can be used in the following equation
8πG 0
G0 0 − G1 1 = T 0 − T 11 .


4
(231)
c
In fact, taking into account that
G00 = g 00 G00 ,
where g 00 = e−ν and G00 is given by
eν h
0 i
G00 = 2
1 − re−λ (232)
r
and from
G1 1 = g 11 G11 ,
where g 11 = −e−λ e G11 is given by

ν0 1 eλ
G11 = + 2− 2, (233)
r r r

23
we can rewrite the (231) as follows

8πG
e−λ (λ0 + ν 0 ) = r ρc2 + p .


4
(234)
c
Substituting, then, the (230) in the last equation and taking into account that e−λ is given by (227) and
0 −λ −λ


that λ e = −d/dr e , we obtain the following differential equations
r2 d  ν/2 
 
r 1
1− 2 e + 2 eν/2 = Dr, (235)
r0 dr r0 2
which the obvious solution is
"  2 #1/2
ν/2 r
e =A−B 1− , (236)
r0

where A = (1/2) Dr02 e B is a positive constant (since ν 0 > 0). The solution is now almost complete: missing
only the determination of the constants of integration A and B. For this purpose, we rewrite the (230) at
r = R, where p = 0, and obtain that
8πGρ ν(R)/2
D= e .
c2
Therefore, taking into account the relation between A and D, we obtained that
3
A = eν(R)/2 .
2
obtaining eν(R)/2 from the latter equation and replacing it in (236), determine the relation from A and B
"  2 #1/2
R
A = 3B 1 − (237)
r0
Substituting this equation in (236), we have
 " 
 2 #1/2 "  2 #1/2 
ν/2
 R r
e(int) = B 3 1 − − 1− . (238)
 r0 r0 

To determine B, we impose r = R between the internal solution (238) and the external solution of
Schwarzschild
 2
2GM (r) R
eνext = e−λ
ext = 1 − = 1 − (239)
c2 R r0
this implies
1
B= (240)
2
The final solution is
 " 
3  2 #1/2 "  2 #1/2 
ν/2 R 1 r
e(int) = 1− − 1− . (241)
2 r0 2 r0 

By (236), taking account of the relation between A and D, from (237) and (240), we obtain that
  2 1/2   2 1/2
r
1 − r0 − 1 − rR0
2
p(r) = ρc  (242)
 2 1/2   2 1/2
3 1 − rR0 − 1 − rr0

24
Exercise 2
Let us consider a static spherically symmetric anisotropic distribution of matter with an energy-momentum
represented by Tµν = diag (ρ, −Pr , −P⊥ , −P⊥ ), where, ρ is the energy density, Pr the radial pressure and
P⊥ the tangential pressure. We adopt standard Schwarzschild coordinates (t, r, θ, φ) where the line element
can be written as
ds2 = e2ν(r) dt2 − e2λ(r) dr2 − r2 dΩ2 , (243)
with dΩ2 ≡ dθ2 + sin2 θdφ2 , the solid angle. The resulting Einstein equations are:

e−2λ
 
1 1
8πρ = 2 + 2λ0 − , (244)
r r r

e−2λ
 
1 0 1
−8πPr = − 2ν + and (245)
r2 r r

λ0 ν0
 
−8πP⊥ = e−2λ − − ν 00 + ν 0 λ0 − (ν 0 )2 , (246)
r r

where primes denote differentiation with respect to r. Using equations (245) and (246), or equivalently the
conservation law Tµν ;µ = 0, we obtain the hydrostatic equilibrium equation for anisotropic fluids

2
Pr0 = − (ρ + Pr ) ν 0 + (P⊥ − Pr ) . (247)
r
Equation (244) can be formally integrated to give

m(r)
e−2λ = 1 − 2 , (248)
r
where a mass function m(r), has been defined by
Z r
m(r) = 4π ρ r̄2 dr̄ , (249)
0

and it corresponds to the mass inside a sphere of radius r as seen by a distant observer.
Finally, from (247), (248) and (245) the anisotropic Tolman-Oppenheimer-Volkov (TOV) equation can
be written as
m + 4πr3 Pr
 
d Pr 2
= − (ρ + Pr ) + (P⊥ − Pr ) . (250)
dr r (r − 2m) r
Obviously, in the isotropic case (P⊥ = Pr ) it becomes the usual TOV equation.

25
Lecture VIII
Exercise 1
Misner and Sharp (1964) developed a general formalism for spherically symmetric gravitational collapse
including pressure. The energy-momentum tensor is

T µν = (ρ + P ) uµ uν + P g µν (251)

and the line element in co-moving coordinates is

ds2 = e2ϕ dt2 − eλ dr2 − R2 (r, t)dΩ2 , (252)

where ϕ and λ are functions of r and t . The components of the 4-velocity are: u0 = e−ϕ , ui = 0 for
i = r , θ , ϕ. We can now introduce a function m(r, t) by defining:
 −1  2
λ 2 2m ∂R
e = 1 + Ṙ − , (253)
r ∂r
where a dot means differentiation with respect to t and multiplication by
 
˙ µ ∂f −ϕ ∂f
f =u =e (254)
∂xµ ∂t
This is the co-moving proper time derivative. Integrating the conservation equation T;νµν and solving Ein-
stein?s field equations, we can find the Misner-Sharp equations for spherically symmetric collapse:

ṁ = −4πR2 P Ṙ , (255)
!
1 + Ṙ2 − 2m/r m + 4πR3 P

∂P
R̈ = − (256)
ρ+P ∂R R3

∂m
= 4πR2 ρ . (257)
∂R
These equations, along with the equation of state relating P and ρ, determine the dynamical evolution of
the spherical homogeneous collapse of the star. If P = 0 the results of Oppenheimer and Snyder (1939) are
recovered. When P = 0 the solution requires numerical integration. Any solution demands the specification
of initial values for R(r, 0) , m(r, 0), and U (r, 0), with U = eϕ Ṙ. It is also required that at r = 0 the functions
R, m, and U all vanish. If rb defines the outer boundary of the distribution of matter, then m(rb , t) = M is
constant and the interior metric can be smoothly joined at the surface r = rb to an exterior Schwarzschild
metric of mass M .

Exercise 2
We start by comparing the line elements of the two metrics. We recall that the line element for Misner-Sharp
is
ds2 = −a2 (r, t)dt2 + b2 (r, t)dr2 + R(r, t)dΩ2 . (258)
We choose first a gauge in which a2 = 1, so called comoving observer gauge. We assume also that R is
separable in a time dependent part and a radial one, R(r, t) = S(t)R̃(r), which leads to
 2 
2 2 2 b 2 2 2
ds = −dt + S (t) dr + R̃ dΩ , (259)
S2

26
and by comparing with the FRW metric

S(t)
b(r, t) = √ . (260)
1 − kr2
The assumption of an homogeneous fluid with non-zero pressure gives us the following conditions to plug
into the MS equations

Dr p = 0 , (261)
p 6= 0 , (262)
e(r) = const. (263)

The MS equations become

 
M
∂t U = − + 4πRρ , (264)
R2
Dr R = Γ , (265)
2m
where Dt U = ∂t U give that a = 1 and Γ = 1 + U 2 − R .

27
Lecture IX
Exercise 1
The areal radius of the radially outgoing photons satisfies
am
rs = a sin χ = (1 + cos η) sin[χe ± (η − ηe )] , (266)
2
dA
The condition dη ≤ 0 is equivalent to
dr
≤ 0. (267)
dη η=ηe

Using the expression for rs

− sin ηe sin χe + cos χe + cos ηe cos χe ≤ 0 ⇐⇒ (268)

cos(ηe + χe ) + cos ηe ≤ 0 ⇐⇒ (269)
π − χe ≤ ηe + χe ., (270)

and finally ηe ≥ π − 2χe .

28
Lecture X
Exercise 1

γµi uµ δµi uµ + ni nµ uµ
vi = = (271)
−nµ uµ αut
ui + (−β i /α)(−α)ut 1
ui + β i ut .


= t
= t
(272)
αu αu

Exercise 2
We have that
ui
 
1
vi = + βi (273)
α u0

but
ui
 
i γij j
v = γij v = + βi (274)
α u0

and
 i  j
ui uj β i ui
   
1 u u 1
v i vi = + β i
γ ij + β j
= γ ij + 2 + β i
β j =
α2 u0 u0 α2 (u0 )2 u0
1
−1 + w2
= 2
−1 + (αu0 )2 = (275)
α w2
then
p
w2 (1 − v i vi ) = 1 , andw = 1 − v i vi ) (276)

In component form

γµi uµ uj
 
t i γij
v = 0, v = = + βj (277)
αut α ut

γi µuµ uj
 
i γij
vt = βi v , vi = t
= + βj (278)
αu α ut

or, using the Lorentz factor

ui βi ui
 
i 1
v = + = + βi
w α α ut
ui ui
vi = = (279)
w αut
It recall that in Special Relativity we have

ui dxi dt dxi
vi = = = . (280)
ut dt dτ dτ

29
Exercise 3
We well know that the Schwarzschild metric in quasi-isotropic coordinates is
   4
1 − M/2r M
ds2 = − dt2 + 1 + dr2 + r2 dΩ2 .


(281)
1 − M/2r 2r

By simply comparing the two metrics

M
1− 2r
α= M
(282)
1+ 2r

and the shift vector is β i = 0. The three metric γ

 4
M
diag 1, r2 , r2 sin2 θ


γij = 1+ (283)
2r

and

nµ = α−1 , −α−1 β i

nµ = (−α, 0, 0, 0) . (284)

30
Lecture XI
Exercise 1

Ln γµν = nα ∇α γµν + γµα ∇ν nα + γνα ∇µ nα

= nα ∇α (nµ nν ) + gµα ∇ν nα + gνα ∇µ nα
= n α n µ ∇α n ν + n α n ν ∇ α n µ + ∇ν n µ + ∇µ n ν
= γ αµ ∇α nν + γ αν ∇α nµ = −2Kµν . (285)

Inverting expression (285) and restricting to (nonzero) spatial indices, the third and last expression for the
extrinsic curvature is therefore given by
1
Kij = − Ln γij . (286)
2

Exercise 2
I start by providing this ancillary relation that we are going to use in the next exercises. We recall from the
lectures that

α :=[−(∇µ t)(∇µ t]−1/2 , (287)

µ µ
n = − α∇ t (288)
µ
aν =n ∇µ nν , (289)

if we recall that γαβ = 0 we can write

Dβ log α = α−1 Dβ α
n o
−1/2
= α−1 γβγ ∇γ [−(∇α t)(∇α t)]
= α−1 γβγ α3 (∇γ ∇δ t)(∇δ t)
= α2 γβγ ∇δ (α−1 nγ ) nδ
 

= γβγ ∇δ nc
= aβ . (290)

Exercise 3
Derivation of the Gauss-Codazzi equation. We introduce an arbitrary vector v γ tangent to the 3-dimensional
hyper surface Σ and we apply the Ricci identity

Dα Dβ v γ − Dβ Dα = Rγµαβ v µ . (291)

if we replace D with its definition we obtain

Dα Dβ v γ = Dα (Dβ v γ )
= γαµ γβν γ γρ ∇µ (γ σν γ ρλ ∇σ v λ ) . (292)

Using the definition of the orthogonal projector γβα = δβα + nα nβ , we write

∇µ γνσ = ∇µ (δνσ + nσ nν )
= ∇µ n σ n ν + n σ ∇µ n ν , (293)

31
We can replace this expression in Eq. (292) and a bit of algebra we obtain
Dα Dβ v γ = γαµ γβν γργ (nσ ∇µ nν γλρ ∇σ v λ + γνσ ∇µ nρ nλ ∇σ v λ + γνσ γλρ ∇µ ∇σ v λ ) , (294)

where the second term inside the parentheses can be replaced by using ∇σ (nλ v λ ) = 0. If we introduce the
definition of the extrinsic curvature Kβα = −γαµ γβν ∇µ nν

Dα Dβ v γ = −Kαβ γλγ nσ ∇σ v λ − Kαγ Kβλ v λ + γαµ γβσ γλγ ∇µ ∇σ v λ . (295)

By considering the symmetry of Kαβ , we can write
Dα Dβ v γ − Dβ Dα v γ = (Kαµ Kβγ − Kβµ Kαγ )v µ + γαρ γβσ γλγ (∇ρ ∇σ v λ − ∇σ ∇ρ v λ ) . (296)
The last term corresponds to the Ricci identity for the connection ∇ and therefore we can write
 
Dα Dβ v γ − Dβ Dα v γ = Kαµ Kβγ − Kβµ Kαγ v γ + γαρ γβσ γλγ Rλµρσ v µ . (297)

By substituting the LHS with Eq. (291) we find the requested expression
 
Kαµ K γβ − Kβµ K γα v µ + γ ρα γ σβ γ γλ Rλµρσ v µ = Rγµαβ v µ , (298)

where v is a vector in the tangent space to the 3-dimensional manifold.

Exercise 4
Similarly to the previous exercise we apply the Ricci identity although this time we project onto the normal
vector n to the hypersurface, in short γ · γ · γ · n (4) R. We project the Ricci identity onto the 3-dimensional
hypersurface
γ µα γ νβ γ γρ Rρσµν nσ = γ µα γ νβ γ γρ (∇µ ∇ν nρ − ∇ν ∇µ nρ ) . (299)

by using the expression derived in the previous exercise, ∇β nα = −Kαβ − aα nβ , we get

γ µα γ νβ γ γρ ∇µ ∇ν nρ = γ µα γ νβ γ γρ ∇µ (−K ρν − aρ nν )
= −γ µα γ νβ γ γρ (∇µ K ρν + ∇µ aρ nν + aρ ∇µ nν )
= −Dα K γβ + aγ Kαβ . (300)
Therefore, by considering the symmetry properties of Kαβ as in the previous exercise we can easily obtain
the Codazzi-Mainardi equations
γ γρ nσ γαµ γ νβ Rρσµν = Dβ K γα − Dα K γβ . (301)

Exercise 5
The projection onto the hypersurfaces of the Riemann tensor contracted twice with the normal vector results
in the Codazzi-Mainardi equations
γα nθ γβγ nη Rθγη = γα nθ γβγ ∇[ ∇θ] nγ
= −2γα nθ γβγ ∇[ (Kθ]γ + nθ] Dγ log α)
= −γα nθ γβγ ∇ Kθγ + γα nθ γβγ ∇θ Kγ + γα γβγ ∇ Dγ log α
+ γα nθ γβγ (∇θ n )(Dγ log α)
= −K θα Kθβ + nθ ∇θ Kαβ − nα Kβ nθ ∇θ n − nβ Kα nθ ∇θ n
1
− nα nβ nθ nγ Kγ ∇θ n + Dα Dβ α , (302)
α

32
where we used the relation, ∇α nβ = −Kαβ − nα Dβ log α. The last step is to rewrite the last expression in
terms of the Lie derivative,
Ln Kαβ = nγ ∇γ Kαβ + Kγα ∇β + Kγβ ∇α . (303)
Using nθ ∇θ n = a we can finally obtain
1
γ α nθ γ γβ nη Rθγη = Ln Kαβ + Dα Dβ α + Kαγ K γβ , (304)
α
as advertised.

33
Lecture XII
Exercise 1
We define a normal observer uµ = W (1, v i ) = W (nµ + v µ ), we can rewrite the stress energy tensor as

T µν = hρW 2 (nµ + v µ )(nν + v ν ) + p(γ µν − nµ nν )

= hρW 2 (nµ nν + v µ nν + nµ v ν + v µ v n u) + p(γ µν − nµ nν )
= hρW 2 − p nµ nν + hρW 2 (v µ nν + nµ v ν ) + hρW 2 v µ v ν + pγ µν

= Enµν + hρW 2 v µ nν + hρW 2 v ν nµ + S µν

= Enµν + S µ nν + S ν nµ + S µν , (305)

where we have used the definition of the total energy for the Eulerian observe, the momentum density and
the purely spatial energy-momentum tensor.

Exercise 2
We can derive the Hamiltonian constraint starting from the Einstein equations and the Gauss-Codazzi
equations derived in the previous section

γ σγ γ µβ γ νδ R σµγν = Rβδ + KKβδ − Kβγ K ρδ , (306)

we contract the indices γ and α

γ σγ γ µν R σµγν = R + K 2 − Kαβ K αβ , (307)

Contracting the Codazzi-Mainardi equation gives

γ σγ γ µβ nδ R σµγν = Dβ K − Dα K αβ ; , (308)

The LHS of Eq. (307) can be expanded

γ σγ γ µν R σµγν = (g σγ + nσ nγ )(g µν + nµ nν )R σµγν

= R + nσ nγ Rσγ + nµ nν Rµν + nσ nγ nµ nν Rσγνµ
= R + 2nσ nγ Rσγ , (309)

where we have used the symmetry properties of R σµγν . Comparing with the Einstein equations

2nσ nγ Gσγ = 2nσ nγ Rσγ − nσ nγ gσγ R

= 2nσ nγ Rσγ − nσ nγ (γσγ − nσ nγ )R
= 2nσ nγ Rσγ + R . (310)

Inserting this equation in Eq. (307) yields

2nσ nγ Gσγ = R + K 2 − Kαβ K αβ , (311)

and if we define the energy density measured by a normal observer, nα , as e := nα nβ T αβ we find

R + K 2 − Kαβ K αβ = 16πe . (312)

34
This is the Hamiltonian constraint. Similarly, we can use Eq. (308)
γ σγ γ µβ nδ R σµγν = γ µβ nδ Rµδ + γ µβ nσ nγ nδ R σµγν . (313)
The second term is again zero while the first term on the RHS as
1
γ µβ nδ Gµδ = γ µβ nδ Rµδ − γ µβ nδ gµδ R . (314)
2
Since γ µβ nδ gµδ = γβδ nδ = 0, Eq. (308) reduces to

γ µβ nδ Gµδ = Dβ K − Dα Kβα . (315)

The momentum density can be defined as jα := −γ βα nρ Tβρ , which leads finally to the momentum constraint
Dβ K βα − Dα K = 8πjα . (316)

Exercise 3
Given the scalar field φ
 
D̃i D̃j φ = D̃i (∂j φ) = ∂i ∂j φ − Γ̃kij ∂k φ , (317)

as advertised.
1 2
D̃i D̃j φ = Di Dj φ + γij ∂ k φ ∂k φ − ∂i φ ∂j φ . (318)
φ φ
Using the definition of covariant derivative
1 2
∂i ∂j φ − Γ̃kij ∂k φ = ∂i ∂j φ − Γkij + γij ∂ k φ ∂k φ − ∂i φ ∂j φ . (319)
φ φ
and rearranging the previous equation
  2 1
Γ̃kij − Γkij ∂k φ = ∂i φ ∂j φ − γij ∂ k φ∂k φ . (320)
φ φ
Now we compute explicitly the affine connection coefficients for the conformal metric
1 kl
Γ̃kij = γ̃ (∂i γ̃jl + ∂j γ̃li − ∂l γ̃ij )
2
1
= 2 γ kl ∂i φ2 γjl + ∂j φ2 γli − ∂l φ2 γij


2φ
1
= 2 γ kl φ2 (∂i γjl + ∂j γli − ∂l γij ) + 2φγil ∂i φ + 2φ γli ∂j φ − 2φ γij ∂l φ ]

2φ
1
= Γkij + γ kl (γjl ∂i φ + γli ∂j φ − γij ∂l φ)
φ
= Γkij + δjk ∂i ln φ + δik ∂j ln φ − γij γ kl ∂l ln φ , (321)
note that this expression differs from Eq. (7.100) in Rezzolla & Zanotti’s book. We can substitute this
expression in the LHS of Eq. (320) which leads to
1 k
δj ∂i φ + δik ∂j φ − γij γ kl ∂l φ ∂k φ =


φ
2 1
∂i φ ∂j φ − γij ∂ k φ ∂k φ , (322)
φ φ
which is exactly the RHS of Eq. (320) as requested.

35
Exercise 4 (courtesy of Ziri Younsi)
Consider a metric in a 3 + 1 split and the geodesic equation for a massless particle with 4-momentum p, such
that p · p = 0. Show that the geodesic equation can be written as
dxj
= γ ij pi − β j p0 ,
dλ
dpi 1
= −(p0 )2 α∂i α + p0 pj ∂i β j − pl pm ∂i γ lm , (323)
dλ 2
where λ is the affine parameter. First, consider the metric in 3+1 form
ds2 = gµν dxµ dxν (324)
2 i i j
= g00 dt + 2g0i dtdx + gij dx dx . (325)
The covariant components of the metric tensor are given by
 2
−α + βi β i βi

gµν = . (326)
βi γij
To complete the derivation we must first derive the relationship between the momentum and the shift.
Consider the following
dxi
= pi
dλ
= giµ pµ
= gi0 p0 + gij pj
= γij β j p0 + γij pj . (327)
Now multiply both sides of equation (4) by γ ki :
γ ki pi = γ ki γij β j p0 + γ ki γij pj .
Using the identity γ ki γij = δjk we obtain

γ ki pi = pk + β k p0 . (328)
From equation (5) we obtain the following important identities which will be needed later:
pi = γ ki pk − β i p0 , (329)
j lj j 0
p = γ pl − β p , (330)
which upon rearranging yield the following expressions for the shift
β i p0 = γ ki pk − pi , (331)
j 0
β p = γ lj pl − pj . (332)
Note that equation (7) is identical to equation (6) in Hughes et al. (1994) and may be rewritten as
dxj
= γ ij pi − β j p0 , (333)
dλ
where we have let l → i. We are now in a position to derive the expression for p0 . Consider the following
0 = pµ pµ
= gµν pµ pν

2
−α2 + βi β i p0 + 2βi pi p0 + γij pi pj .


= (334)

36
Bringing (αp0 )2 to the LHS and raising the index on covariant shift terms using γij we obtain

2
αp0 β i p0 β j p0 + 2pi β j p0 + pi pj




= γij
= γij β j p0 β i p0 + 2pi + pi pj .


(335)

Using equations (8)–(9), equation (12) becomes

2
αp0 = γij γ lj pl − pj γ ki pk + pi + pi pj




= γij γ lj γ ki pl pk − γ ki pj pk + γ lj pi pl
 

= γ lk pl pk − δjk pj pk + δil pi pl . (336)

The second and third terms in equation (13) cancel as k and l are dummy indices and we obtain, upon letting
l → i, k → j, the result p
0 γ ij pi pj
p = , (337)
α
which is identical to equation (4) in Hughes et al. (1994), as required. Now consider the Euler-Lagrange
equations of motion:  
∂L d ∂L
= , (338)
∂xk dλ ∂ ẋk
where the Lagrangian is defined as

2L = gµν pµ pν (339)

2
= g00 p0 + 2g0i p0 pi + gij pi pj . (340)

Using equation (16), the RHS of equation (15) may be calculated as

   
d ∂L d ∂L
=
dλ ∂ ẋk dλ ∂pk
ν
 
1 d µ ∂p
= 2gµν p
2 dλ ∂pk
d
= (gµν pµ δkν )
dλ
dpk
= ,
dλ
hence we obtain the following result:
∂L dpk
= . (341)
∂xk dλ
Finally, we must compute the LHS of equation (15) using the expression for the Lagrangian given in equation
(17). We obtain the following:
∂L 1h 0 2

i j

0 2

j i

0 i j
i
= −2αα,k p + γ ij β β p + 2 γ ij β p p + γ ij,k p p . (342)
∂xk 2 ,k ,k

Now we collect terms in γij and γij,k :

   
∂L 0 2

1 i j
0 2

j i

0 1 i j 0
2 j i 0 1 i j
= −αα,k p + γij β β ,k p + β p ,k p + γij,k ββ p +β p p + p p
∂xk 2 2 2
2 1 h 2 j 0 i
i 1 h 2
i
= −αα,k p0 + γij β i β j ,k p0 + 2β,k p p + γij,k β i β j p0 + 2β j pi p0 + pi pj





2 2
0 2


= −αα,k p + T1 + T2 , (343)

37
where
1 h
2 j 0 i
i
γij β i β j ,k p0 + 2β,k


T1 ≡ p p , (344)
2
1 h
2 i
T2 ≡ γij,k β i β j p0 + 2β j pi p0 + pi pj . (345)
2
First let us turn our attention to equation (21). We may expand and subsequently simplify as follows:
1 h
j
2 j 0 i
i
T1 = γij 2β i β,k p0 + 2β,k p p
2

j 0
= γij β i p0 + pi β,k p
j 0
= γij γ ki pk β,k p
j
= β,k pj p0 , (346)
as in term 2 on the RHS of equation (5) in Hughes et al. (1994). Secondly, we turn our attention to equation
(22). First we may rewrite the expression as follows:
1
T2 = γij,k β j p0 β i p0 + 2pi + pi pj .


(347)
2
Now we employ equations (8)–(9), yielding:
1
γij,k γ lj pl − pj γ mi pm + pi + pi pj



T2 =
2
1
γij,k γ mi γ lj pm pl + γ lj pi pl − γ mi pj pm .
 
= (348)
2
At this point we must consider transforming γij,k into γ ij,k . Consider the following identity:
γ lm = γ mi γ lj γij . (349)
k
Differentiating w.r.t. x and rearranging yields the following identity:
γij,k γ mi γ lj = γ lm,k − γij γ mi γ lj ,k



 
= γ lm,k − γij γ lj,k γ mi + γ lj γ mi,k

= γ lm,k − γ lj,k δjm − γ mi,k δil . (350)

Before proceeding further, notice how terms two and three in equation (25) cancel: symmetry via γij = γji
and the fact that l and m are dummy indices reveals this. As such we are now left with
1
T2 = γij,k γ mi γ lj pm pl
2
1  lm 
= γ ,k pm pl − γ lj,k δjm pm pl − γ mi,k δil pm pl
2
1  lm 
= γ ,k pm pl − γ lj,k pj pl − γ mi,k pm pi . (351)
2
Note that l, m, i and j are all dummy indices. The choice j → m, i → l along with symmetry of the metric
upon interchange of indices simplifies equation (28), yielding
1
T2 = − γ lm,k pm pl . (352)
2
Finally, combining equations (20), (23) and (29), remembering the identity in equation (18), and letting
k → i, j → k we obtain the final result
dpi
2 1
= −α α,i p0 + β,ik pk p0 − γ lm,i pl pm , (353)
dλ 2
which is precisely the expression in Hughes et al. (1994), as required.

38
39