Solution Manual Modern General Relativity by Guidry PDF

Instructor Solutions Manual: Modern General Relativity
Mike Guidry
This document gives the solutions for all problems at the ends of chapters
for the first edition of Modern General Relativity: Black Holes, Gravitational
Waves, and Cosmology by Mike Guidry (Cambridge University Press, 2019).
Unless otherwise indicated, literature references, equation numbers, figure ref-
erences, table references, and section numbers refer to the print version of that
book.
1 Introduction
1.1 From Eq. (1.2), the value of γ is infinite if v = c, so there is no Lorentz transformation
to an inertial frame corresponding to a rest frame for light.
1.2 Since E = mγ , for a 7 TeV proton,
E 7 × 1012 eV
γ= = = 7460.
m 938.3 × 106 eV
Then from the definition of γ ,
s
v 1
= 1− = 0.999999991.
c γ2
This is a speed that is only about 3 meters per second less than that of light.
1.3 This question is ambiguous, since it does not specify whether the curvature is that of
the surface itself (which is called intrinsic curvature) or whether it is the apparent curvature
of the surface seen embedded in a higher-dimensional euclidean space (which is called
the extrinsic curvature). In general relativity the curvature of interest is usually intrinsic
curvature. Then the sheet of paper can be laid out flat and is not curved, the cylinder is
also flat, with no intrinsic curvature, because one can imagine cutting it longitudinally
and rolling it out into a flat surface, but the sphere has finite intrinsic curvature because it
cannot be cut and rolled out flat without distortion. The reason that the cylinder seems to
be curved is because the 2D surface is being viewed embedded in 3D space, which gives
a non-zero extrinsic curvature, but if attention is confined only to the 2D surface it has no
intrinsic curvature. This is a rather qualitative discussion but in later chapters methods will
be developed to quantify the amount of intrinsic curvature for a surface.
1
Coordinate Systems and
2
Transformations
2.1 Utilizing Eq. (2.31) to integrate around the circumference of the circle,
s 2
Z +R
dy
I I
2 2 1/2
C = ds = (dx + dy ) = 2 dx 1 + ,
−R dx
subject to the constraint R2 = x2 + y2 , where the factor of two and the limits are because x
ranges from −R to +R over half a circle. The constraints yield dy/dx = −(R2 − x2 )−1/2 x,
which permits the integral to be written as
s
R2
Z R
C=2 dx .
−R R 2 − x2
Introducing a new integration variable a through a ≡ x/R then gives
Z +1
da
C = 2R √ = 2π R,
−1 1 − a2
−1
since the integral is sin a. In plane polar coordinates the line element is given by Eq.
(2.32) and proceeding as above the circumference is
I I
C= ds =(dr2 + r2 d ϕ 2 )1/2
s
Z 2π 2 Z 2π
dr
= d ϕ r2 + =R d ϕ = 2π R,
0 dϕ 0
where r = R has been used, implying that dr/d ϕ = 0.

2.2 Under a Galilean transformation x ′ = x − vt and t ′ = t it is clear that the acceleration
a and the separation vector r = ∆xx between two masses are unchanged. Thus the second
law F = maa and the gravitational law F = Gm1 m2 r̂/r2 are invariant under Galilean trans-
formations.
2.3 Our solution follows Example 1.2.1 of Foster and Nightingale [88]. The tangent and
dual basis vectors, and the products for gi j = g ji = e i ·ee j , were worked out in Example 2.3.
The elements for gi j = g ji = e i ·ee j can be obtained in a similar fashion. For example,
g12 = g21 = ( 21 i + 12 j )·( 12 i − 12 j ) = 41 − 14 = 0,
where the orthonormality of the cartesian basis vectors has been used. Summarizing the
results,
 2   1 
4v + 2 4uv 2v 2 0 −v
gi j =  4uv 4u2 + 2 2u gi j =  0 1
2 −u 
2v 2u 1 2
−v −u 2u + 2v + 1 2
2
3 Coordinate Systems and Transformations
y
a b φ
θ
x
t
Fig. 2.1 Figure for Problem 2.5.
By direct multiplication the product of these two matrices is the unit matrix, verifying Eq.
(2.26) explicitly for this case. Utilizing Eq. (2.29), the line element is
ds2 = gi j dui du j
= guu du2 + 2guvdudv + 2guwdudw + gvvdv2 + 2gvw dvdw + gwwdw2
= (4v2 + 2)du2 + 8uvdudv + 4vdudw + (4u2 + 2)dv2 + 4udvdw + dw2
where gi j = g ji has been used and no summation is implied by repeated indices.
2.4 Using the spherical coordinates
u1 = r u2 = θ u3 = ϕ
defined through Eq. (2.2) and the results of Example 2.2,
e1 ·ee1 = 1 e2 ·ee2 = r2 e3 ·ee3 = r2 sin2 θ ,
while all non-diagonal components vanish. Thus the metric tensor is

 
1 0 0
gi j =  0 r 2 0 .
2 2
0 0 r sin θ
The corresponding line element is
ds2 = dr2 + r2 d θ 2 + r2 sin2 θ d ϕ 2 ,
where Eq. (2.29) has been used.
2.5 This solution is based on Problem 1.2 in Ref. [88]. From the parameterization r =
xii + y j + zkk with
x = (a + b cos ϕ ) cos θ y = (a + b cos ϕ ) sin θ z = b sin ϕ ,
where the radius of the doughnut a and radius of the circle b are defined in Fig. 2.1 [this
document], the tangent basis vectors are
∂r
eθ = = − sin θ (a + b cos ϕ ) i + (a + b cos ϕ ) cos θ j
∂θ
∂r
eϕ = = −(b sin ϕ cos θ ) i − (b sin ϕ sin θ ) j + (b cos ϕ ) k .
∂ϕ
The corresponding elements of the metric tensor gi j = e i · e j are
gϕϕ = b2 gϕθ = gθ ϕ = 0 gθ θ = (a + b cos ϕ )2 .
2.6 The tangent basis vectors and metric tensor gi j were given in Example 2.4. Since gi j
is the matrix inverse of gi j , which is diagonal,

1 0 ij 1 0
gi j = −→ g =
0 r2 0 1/r2
Then the dual basis may be obtained by raising indices with the metric tensor: e i = gi j e j ,
giving
1
e 1 = g11 e 1 + g12e 2 = e 1 e 2 = g21 e 1 + g22e 2 = 2 e 2
r
for the elements of the dual basis.
2.7 For a constant displacement d in the x direction
x′ = x − d y′ = y z′ = z.
Since d is constant
dx′ = dx dy′ = dy dz′ = dz
and therefore ds′ 2 = ds2 . From Eq. (2.41), a rotation in the x − y plane may be written
x′ = x cos θ + y sin θ y′ = −x sin θ + y cos θ z′ = z,
which gives the transformed line element

2
ds′ = (dx′ )2 + (dy′ )2 + (dz′ )2
= (cos θ dx + sin θ dy)2 + (− sin θ dx + cos θ dy)2 + dz2
= (cos2 θ + sin2 θ )dx2 + (cos2 θ + sin2 θ )dy2 + dz2
= dx2 + dy2 + dz2
= ds2 .
Therefore the euclidean spatial line element is invariant under displacements by a constant
amount and under rotations.
2.8 Taking the scalar products using Eqs. (2.8), (2.9), and (2.20) gives
e i ·V
V = e i ·(V j e j ) = V j e i ·ee j = V j δ ji = V i ,
V = ei ·(V j e j ) = V j ei ·ee j = V j δij = Vi ,
ei ·V
which is Eq. (2.22).

2.9 Utilizing that the angle θ between the basis vectors is determined by cos θ = e 1 ·
e 2 /|ee1 ||ee2 |, the area of the parallelogram is
dA = |ee1 ||ee2 | sin θ dx1 dx2
= |ee1 ||ee2 |(1 − cos2 θ )1/2 dx1 dx2
1/2 1 2
= |ee1 |2 |ee2 |2 − (ee1 ·ee2 )2 dx dx .
The components of the metric tensor gi j are
e 1 ·ee2 = g12 = g21 |ee1 ||ee1 | = e 1 ·ee1 = g11 |ee2 ||ee2 | = e 2 ·ee2 = g22 ,
so the area of the parallelogram may be expressed as
1/2 1 2 p
dA = g11 g22 − g212 dx dx = det g dx1 dx2 ,
where det g is the determinant of the metric tensor. This is the 2D version of the invariant
4D volume element given in Eq. (3.48).
3 Tensors and Covariance
3.1 For the three cases

µν µ ν ∂ x′ µ ∂ x′ ν α β ∂ x′ µ ∂ x′ ν αβ
T′ = V′ V′ = V V = T
∂ xα ∂ xβ ∂ xα ∂ xβ
′ ∂ xα ∂ xβ ∂ xα ∂ xβ
Tµν = Vµ′ Vν′ = ′ µ ′ ν Vα Vβ = ′ µ ′ ν Tαβ
∂x ∂x ∂x ∂x
′ ′ν ∂ xα ∂ x′ ν ∂ xα ∂ x′ ν β
Tµ′ν = Vµ V = ′ µ Vα V β
= T .
∂ x ∂ xβ ∂ x′ µ ∂ xβ α
3.2 From Eqs. (3.50) and (3.51) with indices suitably relabeled
λ ∂ xβ ∂ xα ∂ 2 xα
A′µ ,ν − Γ′ µν A′ λ = Aα ,β ν µ + A α
∂ x′ ∂ x′ ∂ x′ ν ∂ x′ µ !
α β ′λ ∂ 2 xα ∂ x′ λ ∂ xγ
κ ∂x ∂x ∂x
− Γαβ ′ µ ′ ν + ′ µ ′ν Aγ
∂ x ∂ x ∂ xκ ∂ x ∂ x ∂ xα ∂ x′ λ
∂ xβ ∂ xα ∂ 2 xα
= Aα ,β + A α
∂ x′ ν ∂ x′ µ ∂ x′ ν ∂ x′ µ
λ
∂ x ∂ x ∂ x ∂ xγ
α β ′ ∂ 2 xα ∂ x′ λ ∂ xγ
− Γκαβ ′ µ ′ ν A γ − Aγ
∂ x ∂ x ∂ xκ ∂ x′ λ ∂ x′ µ ∂ x′ ν ∂ xα ∂ x′ λ
∂ xβ ∂ xα ∂ 2 xα
= Aα ,β ′ ν ′ µ + Aα ′ ν ′ µ
∂x ∂x ∂x ∂x
∂ x α ∂ xβ ∂ 2 xα
− Γκαβ ′ µ ′ ν Aκ − Aα ′ ν ′ µ
∂x ∂x ∂x ∂x
β
∂x ∂x α ∂ x ∂ xβ
α
= Aα ,β ′ ν ′ µ − Γκαβ ′ µ ′ ν Aκ
∂x ∂x ∂x ∂x
∂ xα ∂ xβ
= Aα ,β − Γκαβ Aκ ,
∂ x′ µ ∂ x′ ν
3.3 (a) Since δµν is a rank-2 tensor with the same components in all coordinate systems
(see Section 3.8), under a coordinate transformation gµα gαν = δµν becomes g′µα g′αν = δµν .
Since gµν is a tensor, if we assume g µν is also a tensor then
∂ xκ ∂ xη ∂ x′α ∂ x′ν ρσ
g′µα = gκη . g′αν = g .
∂ x ′ µ ∂ x ′α ∂ xρ ∂ xσ
Then evaluating g′µα g′αν ,
∂ x′α ∂ x′ν ρσ ∂ xκ ∂ xη ∂ x′α ∂ x′ν ρσ ∂ x σ ∂ x ′ν
g′µα g = g κη g = = δµν ,
∂ xρ ∂ xσ ∂ x ′ µ ∂ x ′α ∂ xρ ∂ xσ ∂ x′ µ ∂ xσ
6
where we have used

∂ xη ∂ x′α
= δρη gκρ gρσ = δκσ .
∂ x′α ∂ xρ
Comparing the result
∂ x′α ∂ x′ν ρσ
g′µα g = δµν
∂ xρ ∂ xσ
with g′µα g′αν = δµν requires that
∂ x′α ∂ x′ν ρσ
g′αν = g .
∂ xρ ∂ xσ
which is the transformation law for a rank-2 contravariant tensor. Note that this result is an
example of the quotient theorem described in Problem 3.13. Since gµα gαν = δµν and gµν
and δµν are known to be tensors, g µν must also be a tensor.
(b) From Eq. (3.44) an arbitrary rank-2 tensor can be decomposed into a symmetric and
antisymmetric part,
gµν = 21 (g µν + gν µ ) + 21 (g µν − gν µ ).
Inserting this in the line element gives
ds2 = g µν dxµ dxν
= 21 (g µν + gν µ )dxµ dxν + 12 (gµν − gν µ )dxµ dxν
= gµν + 12 (gν µ − gν µ ) dxµ dxν

= gµν dxµ dxν .

Thus only the symmetric part of gµν contributes to the line element.
3.4 Under the transformation x → x′ ,
′µ ∂ xα ∂ xβ ∂ x′µ ∂ x′α γδ
T ν = g′να T ′µα = g αβ T
∂ x ′ν ∂ x ′α ∂ xγ ∂ xσ
∂ xα ∂ x′ µ α
γ ∂x ∂x
′µ
= gαδ T γδ ′ν γ
= T α ′ν ,
∂x ∂x ∂ x ∂ xγ
where in going from the first line to the second line
∂ xβ ∂ x′α β
= δδ
∂ x ′α ∂ x δ
has been used. This is a tensor transformation law so it is valid in all frames. Proceeding in
similar fashion,
∂ xε ∂ xλ ∂ xγ ∂ xδ ∂ x′α ∂ x′β τθ
′
Tµν = g′µα gνβ
′
T ′αβ = g ελ g γδ T
∂ x ′ µ ∂ x ′α ∂ x ′ν ∂ x ′β ∂ xτ ∂ xθ
∂ xε ∂ xγ
= Tεγ ,
∂ x ′ µ ∂ x ′ν
where in the last step
∂ xλ ∂ x′α ∂ x δ ∂ x ′β
= δτλ = δθδ gελ gγδ T λ δ = Tεγ
∂ x ′α ∂ x τ ∂ x ′β ∂ x θ
have been used. This is a tensor transformation law so it is valid in all frames.
µνα
3.5 (a) For example, consider a rank-4 tensor Tβ . Its transformation law is
∂ x ′ µ ∂ x ′ν ∂ x ′α ∂ x η
′ µνα γδ ε
Tβ = T .
∂ x γ ∂ x δ ∂ x ε ∂ x ′β η
Now set α = β for this tensor (implying a sum on this index). The resulting quantity must
have two upper indices by the summation convention, so define it to be T µν :
β µνα µνα
T µν ≡ δα Tβ = Tα .
Is T µν a tensor? From the preceding equations, its transformation law is
µν ′ µνα β ′ µνα
T′ ≡ Tα = δα Tβ
∂ x ′ µ ∂ x ′ν ∂ x ′α ∂ x η
β γδ ε ∂ x ′ µ ∂ x ′ν ∂ x ′α ∂ x η γδ ε
= δα Tη = T
γ
∂x ∂x ∂x ∂x δ ε ′ β ∂ xγ ∂ xδ ∂ xε ∂ x′α η
∂ x′µ ∂ x′ν η γδ ε ∂ x′µ ∂ x′ν γδ η ∂ x′µ ∂ x′ν γδ
= δ T = T = T ,
∂ xγ ∂ xδ ε η ∂ xγ ∂ xδ η ∂ xγ ∂ xδ
which is the transformation law for a contravariant rank-2 tensor. Similar proofs can be
carried out for tensors of any order. Thus, setting an upper and lower index equal on a
rank-N tensor and summing yields a tensor of rank N − 2.
ν ν
(b) For example, consider the linear combination of two rank-2 tensors, Tµ = aAµ +
ν
bBµ . The transformation law is
∂ x ′ν ∂ x β α ∂ x′ν ∂ xβ α
Tµ′ ν = a A′µν + bB′µν = a A + b B
∂ xα ∂ x′ µ β ∂ xα ∂ x′ µ β
∂ x ′ν ∂ x β α α
∂ x ′ν ∂ x β
α
= α aAβ + bBβ = α T .
∂x ∂x ′ µ ∂ x ∂ x′ µ β
A similar proof holds for any such linear combination of tensors.
3.6 The line element is ds2 = −dt 2 + dr2 + r2 d θ 2 + r2 sin2 θ d ϕ 2 , so the non-zero compo-
nents of the metric are
g00 = gtt = −1 g11 = grr = 1 g22 = gθ θ = r2 g33 = gϕϕ = r2 sin2 θ
and det g µν = −r4 sin2 θ . Then from Eq. (3.48) the invariant volume element is
dV = (−detg µν )1/2 dr d θ d ϕ = r2 dr sin θ d θ d ϕ ,
which gives a volume
Z Z R Z π Z 2π
V= dV = r2 dr sin θ d θ d ϕ = 34 π R3 ,
0 0 0
as expected.
3.7 Since A · B = Aµ B µ is a scalar it is unchanged by a coordinate transformation. Thus
from the vector transformation law for Bµ
∂ xν µ ∂ xν

µ µ
A′µ B′ = Aµ Bµ = Aν ′ µ B′ −→ Aν ′ µ − A′µ B′ = 0.
∂x ∂x
But B′ µ is an arbitrary vector that does not generally vanish.

Thus the quantity in parenthe-
′ ν ′µ
ses must be equal to zero, implying that Aµ = ∂ x /∂ x Aν , which is the transformation
law for a dual vector.
3.8 This problem is adapted from an example in Ref. [88]. From the transformation
equations between spherical and cylindrical coordinates assuming u = (r, θ , ϕ ) and u′ =
(ρ , ϕ , z),
1
u′ = ρ = r sin θ = u1 sin u2
2
u′ = ϕ = u3
3
u′ = z = r cos θ = u1 cos u2
and the inverse transformations are

q q
u1 = r = ρ 2 + z2 = (u′ 1 )2 + (u′ 3 )2
!
′1

2 −1 ρ −1 u
u = θ = tan = tan
z u′ 3
2
u3 = ϕ = u′ .
From these the partial derivative entries in the matrices U and Û defined in Example 3.7
may be computed directly. For example,
∂ u′ 1 ∂
U21 = = 2 (u1 sin u2 ) = u1 cos u2 = r cos θ
∂ u2 ∂u " !#
∂ u2 ∂ ′1 u′ 3 cos θ
2 −1 u
Û1 = 1
= 1
tan 3
= 1 2 3 2
= .
∂u ′ ∂u ′ u ′ ′
(u ) + (u ) ′ r
Computing all the derivatives and assembling them gives

 

sin θ r cos θ 0
 sin θ 0 cos θ
 cos θ sin θ 
U = 0 0 1 Û =  ,
 
 r 0 −
r 
cos θ −r sin θ 0 0 1 0
and by explicit matrix multiplication, ÛU = I.
3.9 From Eqs. (3.45) and (3.46),

1
T[αβ ](γδ ) = Tαβ (γδ ) − Tβ α (γδ )
2
1 1
(Tαβ γδ + Tαβ δ γ ) − 12 (Tβ αγδ + Tβ αδ γ )

= 2
2
1
= Tαβ γδ + Tαβ δ γ − Tβ αγδ − Tβ αδ γ .
4
3.10 (a) Use the symmetry properties and relabeling of dummy indices to write,
Aµν Bµν = −Aν µ Bµν (Aµν is antisymmetic)

= −Aν µ Bν µ (Bµν is symmetric)
µν
= −A Bµν (Interchange dummy indices µ ↔ ν ).
But Aµν Bµν = −Aµν Bµν can be true only if Aµν Bµν = 0.
(b) For example, if Aµν is symmetric, Aµν = Aν µ , then
µν ∂ x′ µ ∂ x′ ν γδ ∂ x′ µ ∂ x′ ν δ γ νµ
A′ = A = A = A′ ,
∂ xγ ∂ xδ ∂ xγ ∂ xδ
with an analogous proof if Aµν = −Aν µ .
µ
3.11 Contracting δν with the components V ν of an arbitrary vector gives
µ
δν V ν = V µ = gµα Vα = gµα gαν V ν .
µ
But V is arbitrary so g µα gαν = δν .
3.12 Multiply both sides of Tµν = Uµν by ∂ xµ /∂ x′α and ∂ xν /∂ x′β and take the implied
sums to give
∂ x µ ∂ xν ∂ x µ ∂ xν
α Tµν = Uµν .
∂ x′ ∂ x′ β ∂ x′ α ∂ x′ β
′ = U′ .
But from Eq. (3.36) this is just Tµν µν
3.13 In the scalar product expression A·B = g µν Aµ Bν of Eq. (3.43) the left side is a scalar
and A and B on the right side are vectors. Since the quantities gµν contracted with tensors
on the right side yield a tensor on the left side, by the quotient theorem g µν must define the
components of a type (0, 2) tensor.
3.14 This solution is adapted from Example 1.8.1 in Ref. [88]. For an arbitrary contravari-
ant vector V γ the transformation law given in the problem is
∂ x′ α ∂ xε δ ϕ
T αβ γ V γ = T V ,
∂ xδ ∂ x′ β εϕ
indicating that T αβ γ V γ transforms as a (1, 1) tensor. By the quotient theorem then T αβ γ
must be a (1, 2) tensor. The proof follows from inserting V γ = (∂ x′ γ /∂ xϕ )V ϕ on the left
side of the above equation and rearranging to give
′γ ∂ x′ α ∂ xε δ

α ∂x
T βγ ϕ − δ T V ϕ = 0.
∂x ∂ x ∂ x′ β εϕ
ϕ
This must be valid for any V ϕ so choose Vϕ = δλ such that the quantity inside the paren-
theses is required to vanish, giving
∂ x′ γ ∂ x′ α ∂ xε δ
T αβ γ = T .
∂ xλ ∂ xδ ∂ x′ β ελ
Multiply both sides of this expression by ∂ xλ /∂ x′ µ to give

∂ x′ γ ∂ xλ ∂ x′ α ∂ xε ∂ xλ δ
T αβ γ = T .
∂ xλ ∂ x′ µ ∂ xδ ∂ x′ β ∂ x′ µ ελ
But on the left side
∂ x′ γ ∂ xλ γ
= δµ ,
∂ xλ ∂ x′ µ
giving finally
∂ x′ α ∂ xε ∂ xλ δ
T αβ µ = T ,
∂ xδ ∂ x′ β ∂ x′ µ ελ
which is the transformation law obeyed by a (1, 2) tensor.
3.15 (a) One may write
∂ x′ α ∂ x µ ∂ x′ α ∂ x µ ∂ x′ α
δµν = = = δβ′ α
∂ xν ∂ x′ β ∂ x µ ∂ x′ β ∂ x′ β
which is the transformation law for a mixed, rank-2 tensor.
(b) In some coordinate system let Kµν = δµν = diag (1, 1, 1, 1). Then under an arbitrary co-
ordinate transformation,
∂ x′ ν ∂ xβ α ∂ x′ ν ∂ xβ α ∂ x′ ν ∂ xα
Kµ′ ν = K = α ′ µ δβ = α ′ µ = δµν .
∂ xα ∂ x′ µ β ∂x ∂x ∂x ∂x
Thus Kµν = δµν is a (1, 1) tensor that has the same components (those of the unit matrix) in
any coordinate system.
3.16 This is a particular example of a scalar product, so it must transform as a scalar.
Explicitly,
2 ∂ xα ∂ xβ ∂ x′ µ γ ∂ x′ ν δ
ds′ = gαβ ds ds
∂ x′ µ ∂ x′ ν ∂ xγ ∂ xδ
∂ xα ∂ x′ µ ∂ xβ ∂ x′ ν
= gαβ dsγ dsδ ′ µ
∂ x ∂ xγ ∂ x′ ν ∂ xδ
∂ xα ∂ xβ β
= gαβ dsγ dsδ γ = gαβ dsγ dsδ δγα δδ
∂ x ∂ xδ
= gαβ dsα dsβ = ds2
where Eq. (3.35) has been used. The squared line element (3.39) is clearly a scalar invariant
and so it has the same value in all coordinate systems.
3.17 By the usual rank-2 tensor transformation law,
∂ x′ µ ∂ x′ ν αβ
T ′µν (x′ ) = T (x).
∂ xα ∂ xβ
Upon differentiating Eq. (3.66),
∂ x′ µ µ
= δα + (δ u) ∂α X µ (x),
∂ xα
which may be substituted into the first equation to give

µ
T ′µν (x′ ) = δα + (δ u)∂α X µ δβν + (δ u)∂β X ν T αβ (x)

µ µ
= δα δβν + δα (δ u)∂β X ν + δβν (δ u)∂α X µ + O δ u2 T αβ (x)
h i
= T µν (x) + ∂β X ν T µβ + ∂α X µ T αν δ u,
where only terms first-order in δ u have been retained.

3.18 The transformation law for dual vectors is given by Eq. (3.29). Using the expansion
(3.66) to evaluate the partial derivative gives
∂ xα ∂Xα

A′µ (x′ ) = ′ µ Aα (x) = δµα − ′ µ δ uAα (x) = Aµ (x) − ∂µ X α (δ u)Aα (x).
∂x ∂x
By analogy with Eq. (3.68) the Lie derivative is then
Aµ (x′ ) − A′µ (x′ )

LX A µ ≡ Lim = X α ∂α Aµ + Aα ∂µ X α ,
δ u→0 δu
where a Taylor expansion as in Eq. (3.69) was used to evaluate Aµ (x′ ).
3.19 Let Aµν = UuVν . Then by the Leibniz rule,
LX Aµν = LX (Uµ Vν ) = (LX Uµ )Vν + Uµ (LX Vν )
= X α (∂α Uµ )Vν + Uµ (∂α Vν ) + Uα (∂µ X α )Vν + Uµ Vα (∂ν X α )

= X α ∂α Aµν + Aαν ∂µ X α + Aµα ∂ν X α ,

where in the second line Eq. (3.73) was used and in the third line Aµν = UuVν and
∂α Aµν = ∂α (Uµ Vν ) = Uµ (∂α V ν ) + (∂α Uµ )Vν
were used.
3.20 Let C = [A, B] = AB − BA and operate on an arbitrary function f ,
C f = [A, B] f = AB f − BA f
= Aν ∂ν (Bµ ∂µ f ) − Bν ∂ν (Aµ ∂µ f )
= Aν ∂ν Bµ ∂µ f + Aν Bµ ∂ν ∂µ f − Bν ∂ν Aµ ∂µ f − Bν Aµ ∂ν ∂µ f
= (Aν ∂ν Bµ )∂µ f − (Bν ∂ν Aµ )∂µ f ,
where in the second line the vectors were expanded in the basis ∂ν and the third line results
from taking the partial derivative of the product. Since the function f is arbitrary, this
implies the operator relation
C = [A, B] = (Aν ∂ν Bµ − Bν ∂ν Aµ )∂µ ,
and since ∂µ is a vector basis, C is a vector with components
C µ = [A, B]µ = Aν ∂ν Bµ − Bν ∂ν Aµ ,
which defines the Lie bracket [A, B] = −[B, A] for the vectors A and B. Comparison with
Eq. (3.72) indicates that the Lie bracket is equivalent to a Lie derivative of a vector field:
[A, B]µ = LA Bµ . The Lie derivative of a tensor then may be viewed as a generalization of
the Lie bracket for vectors.
3.21 (a) From Eqs. (3.15)–(3.17) and Example 3.4,
µ
V (eµ ) = V ν eν (eµ ) = δν V ν = V µ
ω (eµ ) = ων eν (eµ ) = ων δµν = ωµ ,
(b) For vectors V = V α eα , by the chain rule under a coordinate transformation xµ → x′ µ
the basis vectors transform as
∂ xν
eα → eα′ = ′ α eν .
∂x
Thus, to keep V invariant under xµ → x′ µ its components must transform as
µ ∂ x′ µ ν
V′ = V ,
∂ xν
which is equivalent to (3.31), since then
µ ∂ x′ µ ν ∂ xα
V → V ′ = V ′ e′µ = V eα
∂ xν ∂ x′ µ
∂ x′ µ ∂ xα ν
= V eα
∂ xν ∂ x′ µ
= δνα V ν eα
= V α eα = V.
3.22 The first two examples are trivial. Since two successive partial derivative operations
commute,

∂ ∂ ∂ ∂
, =0 , =0
∂x ∂y ∂r ∂θ
and obviously these are coordinate bases. But for the third example
∂ 1 ∂ ∂ 1 ∂ 1 ∂ ∂

[ê1 , ê2 , ] = , = −
∂r r ∂θ ∂r r ∂θ r ∂θ ∂r
1 ∂ 1 ∂2 1 ∂2
=− + −
r2 ∂ θ r ∂ r∂ θ r ∂ r∂ θ
1 ∂ ê2
=− 2 = − 6= 0.
r ∂θ r
Thus ê1 and ê2 do not commute and they define a non-coordinate basis.
3.23 The Lie derivative for a vector is given by Eq. (3.72). Replacing the partial derivatives
with covariant derivatives in this expression gives
LX Aµ = X α ∂α Aµ − Aα ∂α X µ −→ X α ∇α Aµ − Aα ∇α X µ

µ µ
= X α ∂α Aµ + Γβ α Aβ − Aα ∂α X µ + Γβ α X β .
The terms involving the connection coefficients cancel exactly:

µ µ µ µ
Γβ α X α Aβ − Γβ α X β Aα = Γαβ X β Aα − Γβ α X β Aα = 0,
where the dummy summation indices α and β have been interchanged in the first term and
the symmetry of the connection coefficient in its lower indices has been invoked in the last
step. Therefore,
LX Aµ = X α ∂α Aµ − Aα ∂α X µ ↔ X α ∇α Aµ − Aα ∇α X µ .
It can be shown generally that for the Lie derivative of any tensor all partial derivatives
may be replaced by covariant derivatives and vice versa on a manifold with a torsion-free
connection because, as in the above example, the correction terms that convert a partial
derivative to a covariant derivative vanish identically in the Lie derivative if the connection
coefficient is symmetric in its lower indices.
3.24 The infinitesimal displacement ds must be invariant under coordinate transformation:
ds = ds′ . Expand both sides in the basis eµ to give
µ
ds = dxµ eµ = dx′ e′µ .
∂ xµ ′ ν
But dxµ = dx , so
∂ x′ ν
∂ xµ ′ ν µ
dx eµ = dx′ e′µ .
∂ x′ ν
∂ xν ′µ
This is true generally only if e′µ = e . By a similar proof, e ′ µ = ∂ x eν .
µ ν
∂ x′ ∂ xν
3.25 An arbitrary vector A can be expanded as
∂
A = Aµ eµ = Aµ = Aµ ∂µ ,
∂ xµ
using the basis (3.6). Require that a vector A be unchanged by a transformation to a primed
coordinate system, A′ = A, so
ν ∂ ν ∂
A′ ν =A .
∂x ′ ∂ xν
µ
Operate on x′ µ with both sides and invoke ∂ x′ µ /∂ x′ ν = δν to give
∂ x′ µ ν
µ
A′ = A ,
∂ xν
which is the transformation law (3.31) for a vector.
3.26 (a) The components may be evaluated by inserting basis dual vectors as arguments:
(U ⊗ V )(eµ , eν ) = U(eµ )V (eν ) = Uµ Vν ,
where Eq. (3.19) was used.
(b) Insert basis states {eµ , eν , eλ , eε } as arguments, giving
µν ε
(U ⊗ V ⊗ Ω ⊗ W)(eµ , eν , eλ eε ) = U(eµ )V (eν )Ω(eλ )W (eε ) = U µ V ν Ωλ W ε ≡ S λ ,

(c) Generalizing part (a), the tensor product is defined through
(T ⊗ V )(A, B,C)) = T (A, B)V (C).
Inserting basis states as the arguments gives for the mixed-tensor components
µ µ
T (eµ , eν )V (eγ ) = T ν Vγ ≡ S νγ .
(d) The tensor product gives

β
(eµ ⊗ eν )(eα , eβ ) = eµ (eα )eν (eβ ) = δµα δν ,
where Eqs. (3.19) and (3.17) were used. Hence
β
T µν (eµ ⊗ eν )(eα , eβ ) = T µν δµα δν = T αβ ,
which are the contravariant components of T . Therefore we can expand T as
T = T µν (eµ ⊗ eν ),
and we see that eµ ⊗ eν acts as a basis for T = U ⊗V . More generally, we can expand T in
any of the forms
ν µ
T = T µν (eµ ⊗ eν ) = Tµν (eµ ⊗ eν ) = Tµ (eµ ⊗ eν ) = T ν (eµ ⊗ eν )
by inserting different combinations of basis vectors or basis dual vectors in the preceding
derivation.
Lorentz Covariance and Special
4
Relativity
4.1 After the transformation given by Eq. (4.20) the line element is
ds′ 2 = −c2 (dt ′ )2 + (dx′ )2 + (dy′ )2 + (dz′ )2

= −(c cosh ξ dt + sinh ξ dx)2 + (c sinh ξ dt + cosh ξ dx)2 + dy2 + dz2
= −c2 dt 2 + dx2 + dy2 + dz2 = ds2 ,
so it is invariant under the transformation.
4.2 Consider the following diagram:
v∆t'
d d δ
Rest frame In motion with velocity v to the right
The period in the rest frame of the clock is ∆t = d/c but the light for the moving observer is
seen to travel a distance δ > d at a speed c in one tick. From the geometry of the diagram,
the elapsed time observed for one tick of the moving clock is
p
′ δ d 2 + (v∆t ′ )2
∆t = =
c c
Square both sides and solve for ∆t ′ to give
d d
∆t ′ = p = γ = γ ∆t,
c 1 − v2/c2 c
which is the special relativistic time dilation formula: the observer in motion with respect
to the clock sees the clock run more slowly an observer in the rest frame of the clock.
4.3 For infinitesimal displacements the Lorentz transformation (4.26) is
dt ′ = γ dt − (v/c2)dx dx′ = γ (dx − vdt) dy′ = dy dz′ = dz.

The velocity transformations are then obtained by evaluating derivatives of displacements

16
17 Lorentz Covariance and Special Relativity
with respect to time:

dx′ γ (dx − vdt) dx/dt − v ux − v
u′x = = = =
dt ′ γ (dt − vdx/c2) 1 − (v/c2)dx/dt 1 − vux/c2
dy′ dy dy/dt uy
u′y = ′ = = =
dt γ (dt − vdx/c2) γ (1 − (v/c2)dx/dt) γ (1 − (v/c2)ux )
dz′ dz dz/dt uz
u′z = ′ = = =
dt γ (dt − vdx/c2) γ (1 − (v/c2)dx/dt) γ (1 − (v/c2)ux )
For all motion along the x axis one obtains the standard velocity addition formula
u−v
u′ = .
1 − uv/c2
Setting u = c gives u′ = (c − v)/(1 − v/c) = c, independent of v, implying a constant speed
of light in all inertial frames, while for v << 0 one obtains u′ = u − v, as expected from
Galilean invariance [see Eq. (2.43)].
4.4 The Minkowski line interval is (∆s)2 = −(c∆t)2 + (∆x)2 , so the lengths of the sides in
units of the grid spacing are
AB = (62 )1/2 = 6 BC = (32 )1/2 = 3 AC = (−32 + 62)1/2 = (27)1/2 = 5.2.
Thus AB is the longest side and BC is the shortest side. The distance by the path A → C is
5.2 grid spacings and the distance by the path A → B → C is 6 + 3 = 9 grid spacings.
4.5 Two vectors Aµ and Aν are orthogonal if their inner (scalar) product vanishes, gµν Aµ Aν =
0. But for lightlike (null) vectors X, gµν X µ X ν = Xν X ν = 0. Therefore, a lightlike vector
must be orthogonal to itself.
4.6 The variable ξ is a relativistic velocity parameter so the inverse transformation corre-
sponds to ξ → −ξ . Since sinh(−x) = sinh x and cosh(−x) = cosh x, one has for the inverse
of the transformation defined by Eq. (4.20),
′
cdt cosh ξ − sinh ξ cdt
= .
dx′ − sinh ξ cosh ξ dx
Because i sinh x = sin(ix) and cosh x = cos(ix), the Lorentz boost of Eq. (4.20) may be
interpreted a a rotation through an imaginary angle.
4.7 A monochromatic source moves with respect to an observer with a radial velocity v.
Let dt be a time interval measured in the rest frame of the source and dt ′ be a corresponding
time interval measured by the observer. From Example 4.2,
dt ′ 1
dt = γ≡p .
γ 1 − v2/c2
In dt ′ the distance between source and observer increases by vdt ′ (as seen by the observer).
The time for light to travel this extra distance is (v/c)dt ′ . Thus the total time between
wavecrests seen by the observer is
v v ′ v 1 + v/c
∆t = dt ′ + dt ′ = 1 + dt = 1 + γ dt = p dt
c c c 1 − v2/c2
Since frequency is inversely related to time, the ratio of the rest-frame frequency ν0 and
the observed frequency ν will be given by
ν0 1/dt ∆t (1 + v/c)dt
= = =p .
ν 1/∆t dt 1 − v2/c2 dt
Thus the relativistic Doppler shift is given by
s
ν0 1 + v/c 1 + v/c
=p = .
ν 1 − v2/c2 1 − v/c
If v2 /c2 is small it can be ignored in the denominator and one recovers the usual nonrela-
tivistic Doppler formula ∆v/ν = v/c.
4.8 From the diagram
B' t
2
d v (t 2 - t 1 )
∆ϕ θ t1
A B
~d
(i) ∆ϕ ≃ vδ t sin θ /d. Observer A sees the light from B at t1′ and from B′ at t2′ , with
d + vδ t cos θ d
t1′ = t1 + t2′ = t2 + .
c c
The time measured at A for the source to move from B to B′ is
d d + vδ t cos θ
∆t = t2′ − t1′ = t2 + − t1 −
c c
vδ t cos θ
= t2 − t1 − = δ t(1 − β cos θ ),
c
where β ≡ v/c and δ t ≡ t2 − t1 . Then the apparent transverse velocity for the motion B to
B′ observed at A is
vT d ∆ϕ d vδ t sin θ /d β sin θ
βT ≡ = = = .
c c ∆t c δ t(1 − β cos θ ) 1 − β cos θ
(ii) The maximum for βT is found from the above formula by setting ∂ βT /∂ θ = 0. Taking
the derivative, setting it equal to zero, and using sin(cos−1 β ) = (1 − β 2 )1/2 , yields that the
maximum value of βT is
β
βTmax = ,
1−β2
where β is the actual velocity (in units of c) and βT is the apparent velocity. Thus, as β
approaches its physical maximum of unity, the apparent transverse velocity grows without
bound and it is possible to observe any transverse velocity, even those appearing to exceed
the speed of light.
(iii) Setting θ = 10◦ and β = 0.995, gives from the preceding formula βTmax = 8.6. Thus
the apparent transverse velocity is observed to be 8.6 times that of light (superluminal),
even though the actual transverse velocity is only v = 0.995c (subluminal).
µ β
4.9 From the general transformation law for a rank-2 tensor, ηνα = Λ ν Λ α η µβ , and
from the index raising and lowering properties of the metric
λ β
Λµ = ηµβ η λ α Λ α,
where we define the matrices

∂ x′ µ ∂ xν

µ ν
[Λ ν] ≡ [Λµ ] ≡ .
∂ xν ∂ x′ µ
This may be used to show that
µ λ µ λα β µ β λα
Λ ν Λµ =Λ ν η µβ η Λ α =Λ ν Λ α η µβ η = ηνα η λ α = δνλ ,
µ ν
implying that Λ ν and Λµ are matrix inverses of each other.
4.10 From Fig. 4.6(a), if the unprimed coordinate system is moving to the right with
velocity v along the x axis (with tan ϕ = v/c), the primed axes plotted in the unprimed
system are sketched in the left figure below. Then from the primed system the unprimed
system moves to the left with velocity −v along the x axis and the unprimed axes plotted
in the primed coordinate system are given by the right figure below.
ct ct' ct ct'
Primed coordinate
c
system moving to
=
φ φ
v
right in unprimed x' x

=
c
φ φ
x x'
Unprimed coordinate
system moving to left
in primed system
4.11 In the following diagram the unprimed axes are orthogonal, so their scalar product
x·t = 0. But if ϕ1 = ϕ2 , the primed axes represent a Lorentz transform from the unprimed
axes [see Fig. 4.6(b)]. Since the scalar product is invariant under Lorentz transformation,
the primed axes also have vanishing scalar product and thus are orthogonal.
ct
ct'
π/4 − φ2 Lightcone
φ2
x'
φ1
π/4 − φ1
x
For the special case that ϕ1 = ϕ2 = π /4 in the above diagram, the two vectors become
equivalent and lie on the lightcone. But by the above argument they must also be orthogonal
to each other. Thus, a lightlike vector is orthogonal to itself.
4.12 Since the mass is macroscopic and the force is small, Newtonian mechanics describes
the acceleration, but raising the temperature by ∆T adds an energy ∆E ∼ k∆T , where k is
Boltzmann’s constant. Thus, for equivalent forces of magnitude F the ratio of acceleration
magnitudes is
a1 F/m1 m2 m1 + k∆T k∆T
= = = = 1+ ,
a2 F/m2 m1 m1 m1
where mass 2 is assumed to be the hotter one. The effect is of course tiny in magnitude,
because k∆T /m1 is small.
4.13 In frame S the interval is (∆s)2 = (∆x)2 − (c∆t)2 = −c2 (∆t)2 , since ∆x = 0. But the
interval is invariant under Lorentz transformations so in the S′ frame
−c2 (∆t)2 = (∆x′ )2 − (c∆t ′ )2 ,

1/2
which may be solved to give ∆x′ = c (∆t ′ )2 − (∆t)2 .
4.14 Since ds2 is an invariant, in any inertial frame Eq. (4.5) gives the measured time
interval between the two events as
c2 dt 2 = −ds2 + dx2 + dy2 + dz2 ,
where −ds2 is positive, since the interval is timelike. By definition, in the inertial frame
where the spatial separation between two events is zero (dx2 + dy2 + dz2 = 0), the measured
time is the proper time,
ds2
dt 2 = − ≡ dτ 2.
c2
(Since the separation is timelike, there always is an inertial frame where the spatial sepa-
ration between events is zero; see Fig. 4.7(b).) Thus, since dx2 + dy2 + dz2 can never be
negative, in any other inertial frame the measured time interval between the events will be
larger than the proper time: for events with timelike separation, the minimum time interval
that can be measured by any inertial observer is the proper time.
4.15 (a) From Eq. (4.50) the field tensor is defined by F µν = ∂ µ Aν − ∂ ν Aµ . Under the
gauge transformation Aµ → Aµ − ∂ µ χ given by (4.46), this transforms as
µν
F′ = ∂ µ (Aν − ∂ ν χ ) − ∂ ν (Aµ − ∂ µ χ )
= ∂ µ Aν − ∂ ν Aµ = F µν ,
where ∂ µ ∂ ν = ∂ ν ∂ µ was used. Thus F µν is gauge invariant. It is also Lorentz invariant,
since it is by explicit construction a rank-2 Lorentz tensor [Eq. (4.50) is a tensor equation].
(b) Appeal to Eqs. (4.33) to construct the components of the electric and magnetic fields in
terms of the potentials. For example, writing some components of Eq. (4.33) out explicitly,
E 1 = ∂ 1 A0 − ∂ 0 A1 = F 10 = −F 01 B2 = ∂ 1 A3 − ∂ 3 A1 = F 13 = −F 31 .
Proceeding in this manner, one finds that the six independent components of E and B are
elements of the antisymmetric rank-2 electromagnetic field tensor
F µν = −F ν µ ≡ ∂ µ Aν − ∂ ν Aµ
given by Eqs. (4.50) and (4.51).
(c) The equivalence may be established by multiplying the terms of Eqs. (4.53) and (4.54)
out explicitly using Eqs. (4.51), (4.52), and (4.43). For example, setting ν = 0 in Eq. (4.53)
and using Eq. (4.51) gives
∂1 E 1 + ∂2 E 2 + ∂3E 3 = j0 ,
which is equivalent to Eq. (4.28).
5 Lorentz-invariant Dynamics
5.1 The variational principle δ ds = 0 with line element ds2 = ηµν ẋµ ẋν d τ 2 (where ẋµ ≡
R
dxµ /d τ ) implies the Euler–Lagrange equation (5.18),

d ∂L ∂L p
− µ
+ µ =0 L ≡ −ηµν ẋµ ẋν .
d τ ∂ (dx /d τ ) ∂x
But ∂ L/∂ xµ = 0 and the preceding equation becomes

d ∂ µ ν 1/2
(− η µν ẋ ẋ ) = 0.
d τ ∂ ẋµ
Using that ηµν ẋµ ẋν is independent of time, this reduces to ηµν ddτ ẋν = ηµν ẍν = 0, where
ẍν ≡ d 2 xν /d τ 2 . In matrix form this is
−1 0 0 0 t¨
  
 0 1 0 0 ẍ
 0 0 1 0 ÿ = 0,
  
0 0 0 1 z̈
implying that trajectories obey
d 2t d 2x d 2y d 2z
= 0 = 0 =0 = 0.
dτ 2 dτ 2 dτ 2 dτ 2
This corresponds to straight lines in Minkowski space.
5.2 If the inertial frame of the particle is chosen it is at rest and its 4-momentum is p =
(m, 0, 0, 0). Assuming motion of the observer along the x axis, the 4-velocity of the observer
is u = (γ , γ v) = (γ , vγ , 0, 0) ≡ e0̂ . From Eq. (5.25) the observed energy is then
E ≡ −p·u = −p·e0 = γ m.
This should not be surprising, since if the laboratory were at rest and the particle moving
this would be the expected result in special relativity.
5.3 (a) Since c = 1 = 3 × 108 m s−1 , one has 1 s = 3 × 108 m. Therefore,
1 m2
1 J = 1 kg = 1.1 × 10−17 kg.
(3 × 108 m)2
(b) Using 1 s = 3 × 108 m again,
1 m−1
1 atm = 105 kg = 1.1 × 10−12 kg m−3 .
(3 × 108 m)2
22
5.4 (a) The SI unit of acceleration is m s−2 . The required conversion factor is c2 since
1 m−1 × c2 = (1 m−1 )(3 × 108 m s−1 )2 = 9 × 1016 m s−2 .
(b) The energy density has units of J m−3 in the SI system. The required conversion factor
is c2 :
(2 kg m−3 )(3 × 108 m s−1 )2 = 1.8 × 1017 J m−3 .
5.5 If L does not depend on x1 the Euler–Lagrange equation (5.18) implies that

d ∂L
− = 0,
d σ ∂ (dx1 /d σ )
where
dxν dxµ

∂L 1
= − g 1ν (x) + g µ1 (x) .
∂ (dx1 /d σ ) 2L dσ dσ
But gµν is symmetric and µ and ν are dummy summation indices so the terms can be
combined to give
∂L 1 dxµ
= − g 1 µ .
∂ (dx1 /d σ ) L dσ
Then using Ld σ = d τ gives
∂L
= −gα µ K α uµ = −K ·u,
∂ (dx1 /d σ )
where K α = (0, 1, 0, 0) is the Killing vector associated with the absence of x1 dependence
in the metric. Insertion of this result in the first equation above gives Eq. (5.29).
5.6 For low velocity E = mγ = m(1 − v2 )−1/2 ≃ m(1 + 12 v 2 ). Restoring to normal units
by m → mc2 and v → v /c gives E ≃ mc2 + 12 mv2 , which is a sum of rest-mass and kinetic
energies.
5.7 (a) 1 dyne cm−2 = 1 g cm−1 s−2 . But from Box 5.1, 1 g = 7.4237 × 10−29 cm and 1 s =
2.9979 × 1010 cm. Inserting these gives 1 dyne cm−2 = 8.26 × 10−50 cm−2 .
(b) 1 MeV = 1.6022 × 10−6 erg. Since k = 8.617 × 10−5 eV K−1 , in k = 1 units a kelvin
degree is
K = 8.617 × 10−5 eV = 1.3807 × 10−16 erg.
Also, from Box 5.1, 1 erg = 8.2601 × 10−50 cm. Utilizing these conversions, it may be
deduced that
1 MeV = 1.602 × 10−6 erg = 1.160 × 1010 K = 1.323 × 10−55 cm = 1.783 × 10−27 g.
where results from part (a) were used.
5.8 By dimensional analysis: For period P, total mass M, and reduced mass µ , the lumi-
nosity L in geometrized units is
128 2/3 4/3 2 π 10/3
L= 4 M µ ,
5 P
Let the standard unit of length be L , the standard unit of mass be M , and the standard
unit of time be T . In standard units L has dimension M L 2 T −3 (e.g., erg s−1 ). Since µ
and M have units of M and P has units of T , the right side of the equation for L must
be multiplied by a factor of powers of G and c having the units of M −7/3 T 1/3 L 2 to be
dimensionally correct. Because of the mass dependence of the equation G can enter only
as the 7/3 power so the required overall factor is G7/3 /c5 and in standard units
128 2/3 G7/3 4/3 2 π 10/3

L= 4 M µ .
5 c5 P
Inserting G in units of M⊙ −1 cm3 s−2 and c in units of cm s−1 , the factor G7/3 /c5 can be
evaluated and units apportioned conveniently as
M 4/3 µ 2 1 s 10/3

L = 2.3 × 1045 erg s−1 .
M⊙ M⊙ P
More automatically: Table B.1 indicates that conversion from geometrized units to stan-
dard units requires the replacements M → GM/c2 , µ → Gµ /c2 , P → cP, and L → (G/c5 )L.
Making these replacements and rearranging gives the same equations as above.
5.9 As for the derivation in Box 5.2, extremizing the classical action is equivalent to solv-
ing the corresponding Euler-Lagrange equation (5.18). Assuming a Lagrangian L(x, ẋ) =
1 2
2 mẋ − V (x),

∂L d ∂L ∂L dV
= mẋ = mẍ =− .
∂ ẋ dt ∂ ẋ ∂x dx
dV
Inserting this into the Euler–Lagrange equation (5.18) gives mẍ = − , which is Newton’s
dx
second law.
5.10 From the parameterized equations
x2 − t 2 = a2 (cosh2 σ − sinh2 σ ) = a2 ,
so the equation is an hyperbola with t = ±(x2 − a2 )1/2 , which has the following plot for
a = 2.
4
t 0
-2
-4
0 2 4 6
x
The motion clearly is accelerating because of the curvature. From the Minkowski metric
d τ 2 = −ds2 = dt 2 − dx2 . Evaluating dt/d σ and dx/d σ from the original parameterization
gives
d τ 2 = (ad σ )2 (cosh2 σ − sinh2 σ ) = a2 d σ 2 .
Integrating and choosing the integration constant so that τ = 0 when σ = 0 gives τ = aσ ,
and thus
τ τ
x(τ ) = a cosh t(τ ) = a sinh .
a a
The 4-velocity is given by
τ τ
dt dx
u= , = cosh , sinh .
dτ dτ a a
From the Minkowski metric with g = diag (−1, 1),
τ τ
u · u = gi j ui u j = g00 u0 u0 + g11u1 u1 = sinh2 − cosh2 = −1,
a a
which is the normalization given by Eq. (5.6). The 3-velocity is
dx dx/d τ sinh(τ /a) τ
vx = = = = tanh ,
dt dt/d τ cosh(τ /a) a
which has the behavior
1
vx = c
v x (τ/ a )
v x = -c
-1
-4 -2 0 2 4
τ/ a
In these units vx = 1 is the speed of light. Thus the 3-velocity is bounded asymptotically
by c.
5.11 Substituting xµ → x′ µ = xµ + ε K µ into
∂ x′ α ∂ x′ β
gµν (x) = g (x′ )
∂ xµ ∂ xν αβ
(in the following terms of order ε 2 are discarded) leads to
∂ ∂
gµν (x) = µ (xα + ε K α ) ν xβ + ε K β gαβ (xγ + ε K γ )
∂x ∂x
α α
β
= δµ + ε∂µ K δν + ε∂ν K β gαβ (x′ ).
Now expand gαβ (x′ ) in a Taylor series around gαβ (x),

∂ gαβ

′
∆xγ = gαβ (x) + ε K γ ∂γ gαβ (x) + O ε 2 ,

gαβ (x ) = gαβ (x) + γ
∂x x
where ∆xγ = ε K γ . Combining the preceding two equations gives

β
gµν (x) = δµα + ε∂µ K α δν + ε∂ν K β gαβ (x) + ε K γ ∂γ gαβ (x)

h i
= gµν (x) + ε K γ ∂γ gµν (x) + ∂µ K β gβ ν + ∂ν K β gµβ + O ε 2
Subtracting gµν (x) from both sides and noting that ε is arbitrary, this can hold generally
only if the quantity in square brackets vanishes. This leads to
gµβ ∂ν K β + gνβ ∂µ K β + K γ ∂γ gµν = 0,
or equivalently,
∂ν Kµ + ∂µ Kν + K γ ∂γ gµν = 0,
where contraction with the metric tensor was used to lower indices.
5.12 From Problem 5.11 or Box 5.3
∂ν Kµ + ∂µ Kν + K γ ∂γ gµν = 0.
But from the formula for the Lie derivative given in Eq. (3.74)
LK gµν = K α ∂α gµν + gµα ∂ν K α + gνα ∂µ K α
= ∂ν Kµ + ∂µ Kν + K γ ∂γ gµν .
For the Lie derivative in a metric space, partial and covariant derivative operations are
interchangeable (see Section 3.13.5), so
LK gµν = ∂ν Kµ + ∂µ Kν + K γ ∂γ gµν
= ∇ν Kµ + ∇µ Kν + K γ ∇γ gµν = 0,
and since ∇γ gµν = 0 from Eq. (3.63),
∇ν Kµ + ∇µ Kν = ∂ν Kµ + ∂µ Kν = 0,
which is Killing’s equation.
5.13 Choose an inertial frame in which the star at rest and assume the emitted light to be
monochromatic with frequency ω0 . The wavevector for the photon is k µ = (ω0 , ω0 , 0, 0).
For photons E = h̄ω , so from Eq. (5.25)
E = h̄ω = −p · u = −h̄k · u,
implying that ω = −k · u, where u is the velocity of the observer in the inertial frame of the
star. Writing this scalar product out explicitly gives
ω = −k · u = −ηµν k µ uν = k0 u0 − k1 u1
= ω0 (cosh(τ /a) − sinh(τ /a)) = ω0 e−τ /a ,
where τ is the proper time and u = (cosh(τ /a), sinh(τ /a)) from the solution of Problem
5.10 was used.
6 The Principle of Equivalence
6.1 Consider the following figure, where the arc approximates the path of the photon.
∆x
A C
D 1
∆y = 2
gt 2
1
2
φ
1
2
φ B
E Center of circle
The flight time of the photon is t ≃ ∆x/c for a small arc. In the time t the elevator falls a
distance ∆y = 12 gt 2 , where g is the gravitational acceleration. From the geometry, triangles
EBD and ABC contain the same angles, so BC / AC = DB / DE. But from the figure one
obtains
∆x
rc ≃ DE AC = ∆x BC = ∆y = 12 gt 2 DB = ,
2 cos(ϕ /2)
where rc is the curvature radius and the last relation follows because AD = DB, implying
that 12 ∆x/DB = cos(ϕ /2). Therefore,
1 2
2 gt ∆x/[2 cos(ϕ /2)]
= .
∆x rc
Assuming a small deflection justifies the approximation cos(ϕ /2) ≃ 1 and
c2 c2 R 2
rc = = ,
g GM
where in the rightmost expression g has been evaluated at the surface of a sphere of mass
M and radius R. Quantities calculated for the Earth, a white dwarf, and a neutron star are
displayed in the following table.
Object R(km) M(kg) ρ (g cm−3 ) g(m s−2 ) rc (km) R/rc
Earth 6378 6 × 1024 5.6 9.8 9.2 × 1012 6.9 × 10−10

White dwarf 5500 2.1 × 1030 ∼ 106 4.6 × 106 1.9 × 107 2.8 × 10−4
Neutron star 10 2 × 1030 ∼ 1014 1.3 × 1012 67.5 0.15
28
The corresponding vertical deflection of the light is ∆y = 21 gt 2 , where t is the flight time
for the light. For an elevator width of 2 meters, t = 6.67 × 10−9 seconds and the vertical
deflection ∆y is 2.2 × 10−16 m for Earth, 1.0 × 10−10 m for the white dwarf, and 3 × 10−5
m for the neutron star.
6.2 The particle created at z2 has mass m = hν /c2 , where h is Planck’s constant and ν is
the frequency of the photon. Upon dropping to z1 in the gravitational field, the energy is
mc2 + mg(z2 − z1 ). Thus, the system creates spontaneously an energy mg(z2 − z1 ) in each
cycle, unless the photon loses an energy hν g(z2 − z1 )/c2 in moving from z1 to z2 .
6.3 Apply Kepler’s laws to the approximately circular orbit of period 12 hours, giving
r ≃ 2.7 × 107 m and v ≃ 3.9 km s−1 . Defining β = v/c, the special relativistic time dilation
factor for the satellite is γ = (1 − β 2 )−1/2 ≃ 1 + 12 β 2 , where the small effect of Earth’s ro-
tation has been neglected. The fractional change in frequency is determined by the second
term,
νs − ν0
= − 12 β 2 = −8.5 × 10−11,
ν0
where the negative sign is because the time is dilated (νs < ν0 ) for the satellite viewed from
Earth. For the general relativistic time dilation, integrating Eq. (6.7) gives
Z νs Z rs
dν GM
= dr.
ν0 ν R r 2 c2
Evaluating the integrals on both sides yields (see the solution of Problem 6.7)

νs GM 1 1 GM 1 1
= exp − 2 − ≃ 1− 2 − .
ν0 c rs R c rs R
Solving this for the fractional shift in frequency gives

νs − ν0 GM 1 1
= 2 − = 5.3 × 10−10.
ν0 c R rs
This is opposite in sign relative to the special relativistic effect and about six times larger.
Thus, for every second of elapsed time
1. Special relativistic time dilation slows the satellite clock relative to the ground clock by
about 8.5 × 10−11 × 1 second = 0.085 ns.
2. Gravitational time dilation (general relativity) slows the ground clock relative to the
satellite clock by about 5.3 × 10−10 × 1 second = 0.53 ns.
The net effect is that for every second the satellite clock gains about 0.53 − 0.085 =
0.445 ns relative to the ground clock because of relativistic corrections. Suppose that an
accuracy of two meters is desired from the GPS system for locations on the ground. Light
takes 6.7 ns to travel two meters. Thus, without the above corrections for special and gen-
eral relativistic time dilation an error in timing that begins to compromise two-meter reso-
lution will have accumulated after about 15 seconds.
6.4 From the gravitational redshift, ∆ν = gvd/c2 for motion over a vertical distance d.
The corresponding loss in energy for the redshifted light is ∆E = h∆v = −hgvd/c2. But a
particle of mass m would lose an energy ∆E = −gmd in the same circumstances. Compar-
ing the two expressions, one sees that the photon loses energy as if it had an effective mass
m = hv/c2 in the gravitational field. Of course a photon doesn’t have a mass, but it behaves
in some respects as if it did. Note that a similar argument is made in Problem 6.2 without
assuming photons to be massive.
6.5 For a difference in height h, the change in the length of the time intervals is approxi-
mated by ∆τ1 = ∆τ2 (1 − gh/c2), between points 1 and 2. Therefore, the fractional differ-
ence is
∆τ2 − ∆τ1 gh
= 2 = 1.77 × 10−13.
∆τ2 c
This implies a difference of 5.6 × 10−6 seconds per year between the two clocks, with the
one at higher elevation running faster.
6.6 (a) For a pendulum with a string of length ℓ and negligible mass,
d2θ
ℓminertial = −mgrav g sin θ .
dt 2
For small oscillations sin θ ∼ θ and the solution is a harmonic oscillator with period
s
ℓ minertial
P = 2π .
g mgrav
Thus, if the gravitational and inertial masses are not equivalent, identical pendulums made
from different materials having different ratios minertial /mgrav should have different periods.
(b) For a block sliding along an inclined plane that makes an angle θ with the horizontal,
the component of gravitational force along the inclined plane is mgrav g sin θ . Thus,
mgrav
minertial a = mgrav g sin θ −→ a = g sin θ .
minertial
Therefore, if mgrav 6= minertial one expects to observe material-dependent accelerations in
inclined-plane experiments.
6.7 Integrating the gravitational time dilation formula (6.7) gives
Z νs Z rs
dν GM
= dr.
ν0 ν R r 2 c2
Evaluating the two integrals gives

GM 1 1
ln νs − ln ν0 = ln(νs /ν0 ) = −
c2 R rs
and if both sides are exponentiated

νs GM 1 1 GM 1 1
= exp − ≃ 1 + − .
ν0 c2 R rs c2 R rs
Therefore,

νs νs − ν0 GM 1 1
−1 = = 2 − .
ν0 ν0 c R rs
6.8 (a) From Eq. (6.7),

∆ν gh
= 2 = 2.45 × 10−15,
ν c
where a height of h = 22.5 m was used.
(b) To compensate for the gravitational redshift a blue Doppler shift ∆νDoppler = v/c is
required. Setting
gh v
=
c2 c
and solving for v gives v = gh/c = 7.35 × 10−5 cm s−1 . Notice that this is just the blueshift
invoked in Section 6.5.1 to compensate for the gravitational redshift in the falling-elevator
thought experiment.
6.9 Consider two arbitrary masses 1 and 2 at points A and B, respectively, attracting each
other gravitationally. Taking A as the origin of the coordinate system, the forces acting on
the two masses are
mp (1)ma (2) mp (2)ma (1)
F1 = G r̂r F 2 = −G r̂r ,
r2 r2
where subscripts a and p denote active and passive masses, respectively. But by Newton’s
3rd law F 1 = −F
F 2 and
mp (1) mp (2)
= ≡ C,
ma (1) ma (2)
where C must be a universal constant for all gravitational masses, since the two masses
were arbitrary. Choosing units appropriately so that C = 1 gives mp = ma ≡ mg .
6.10 From Eq. (6.12) the gravitational redshift at the surface of a spherical gravitating
body of radius R and mass M is

M R⊙
z = 2.12 × 10−6 .
M⊙ R
The gravitational shift in spectral lines for the Sun is then only ∆λ /λ ∼ 2 × 10−6. If the
gravitational redshift z is parameterized as a velocity giving rise to an equivalent Doppler
shift through v = cz,

M R⊙
v = 0.636 km s−1 .
M⊙ R
For the Sun this gives only ∼ 0.6 km s−1 , which must be disentangled from much larger
kinematic Doppler shifts caused by motion of the Sun relative to the Earth and motion of
gas in the solar surface. The situation is similar for other main sequence stars since the
gravitational redshift is determined to lowest order by M/R, which is the same to within a
factor of 2–3 for main sequence stars.
Curved Spacetime and General
7
Covariance
7.1 This problem is patterned after an example in Cheng [64].

(a) If the space can be parameterized so that the metric is globally independent of the
coordinates, all derivatives in Eq. (7.2) vanish and K = 0.
(b) Taking plane polar coordinates (x1 , x2 ) = (r, θ ) gives the position-dependent metric
g11 = 1 g22 = r2 ,
but insertion into Eq. (7.2) again gives K = 0.
(c) Choosing the spherical coordinates (x1 , x2 ) = (S, ϕ ) defined in Fig. 7.1 gives
1
x
g11 = 1 g22 = R2 sin2 .
R
Insertion into Eq. (7.2) gives K = R−2 , which is constant.
(d) Choosing the cylindrical coordinates (x1 , x2 ) = (r, ϕ ) defined in Fig. 7.1 gives
R2
g11 = g22 = r2 ,
R2 − r 2
and insertion into Eq. (7.2) again yields K = R−2 .
7.2 Since gµν is a rank-2 covariant tensor its transformation law is given by Eq. (3.56) as
gµν ;λ = gµν ,λ − Γαµλ gαν − Γνλ
α
gµα .
In a local inertial frame space is locally flat and gµν ,λ vanishes. Likewise, since from Eq.
(7.30) the affine connection is proportional to the derivative of the metric tensor, it also
vanishes in the local inertial frame. Thus gµν ;λ = 0 for a local inertial frame. But this is a
tensor equation, so it is valid in all reference frames. Thus the covariant derivative of the
metric tensor vanishes in any reference frame. This may be verified by direct computation.
For example, it follows from Eqs. (7.30) and (3.56) and some algebra. More elegantly,
substituting Eq. (7.29) into Eq. (3.56) gives
gµν ;λ = gµν ,λ − Γαµλ gαν − Γνλ
α
gµα
ρ ρ
= Γλ µ gρν + Γλ ν gρ µ − Γαµλ gαν − Γνλ
α
gµα
= Γλα µ gαν + Γλα ν gα µ − Γλα µ gαν − Γλα ν gα µ = 0
where the last step used that ρ and α are dummy indices and that gµν and Γλα µ are sym-
metric in their lower indices.
7.3 In cylindrical polar coordinates (r, θ , z) the line element is ds2 = dr2 + r2 d θ 2 + dz2 ,
32
33 Curved Spacetime and General Covariance
corresponding to metric components gi j = diag (1, r2 , 1) and gi j = g−1 −2

i j = diag (1, r , 1),
2
and obviously det g = r . The geodesic equations are given by Eq. (7.21)
d 2 xλ dxµ dxν
2
+ Γλµν = 0,
dτ dτ dτ
where the connection coefficients can be determined using Eq. (7.30),
1 αλ ∂ gνα ∂ gµα ∂ gν µ

λ
Γµν = 2 g + − .
∂ xµ ∂ xν ∂ xα
Because the metric is diagonal and only one entry is not constant, there are only three
non-vanishing connection coefficients:

1 ∂ g22 1 ∂ g22 1
Γrθ θ ≡ Γ122 = g11 − = −r Γθrθ ≡ Γ212 = Γ221 = g22 = .
2 ∂r 2 ∂r r
Inserting these in Eq. (7.21) gives three equations of geodesic motion,
2
d 2r dθ d 2θ 2 dr d θ d 2z
− r = 0 + = 0 = 0,
dσ 2 dσ dσ 2 r dσ dσ dσ 2
where σ parameterizes the position on a path.
7.4 For 2-dimensional polar coordinates (r, θ ) the line element is ds2 = dr2 + r2 d θ 2 , cor-
responding to a metric gi j = diag (1, r2 ). From Eq. (5.16) without the minus sign the La-
grangian is
" 2 #1/2
dr 2

2 dθ
L= +r ,
dσ dσ
and from Eq. (5.18) the equations of motion are

2
d 2r d 2θ 2 dr d θ

dθ d 2 dθ
= r r = + = 0,
dτ 2 dτ dτ dτ dτ 2 r dτ dτ
where τ = 01 d σ L, with σ parameterizing a path in the space. For the current 2-dimensional
R
case the geodesic equation (7.21) reduces to

d 2r a
0 dx dx
b
d 2θ a
1 dx dx
b
= −Γab = −Γab
dτ 2 dτ dτ dτ 2 dτ dτ
Thus, by comparing
2
d 2r dθ d 2r a
0 dx dx
b
2
=r ←→ = −Γab
dτ dτ dτ 2 dτ dτ
term by term one deduces that
Γ011 = −r Γ000 = Γ010 = Γ001 = 0,
and a corresponding comparison of
d 2θ dxa dxb

d dθ
r2 =0 ←→ 2
= −Γ1ab
dτ dτ dτ dτ dτ
yields
1
Γ101 = Γ110 = Γ100 = Γ111 = 0.
r
It is easily checked that the same coefficients result from solution of Eq. (7.30). For exam-
ple,

0 1 00 ∂ g10 ∂ g10 ∂ g11
Γ11 = g + − = −r.
2 ∂θ ∂θ ∂r
7.5 The connection coefficients Γαµν may be constructed using either the method of com-
paring geodesic equations with the Euler–Lagrange equations of motion, as illustrated in
Problem 7.4, or by direct solution of Eq. (7.30). Let’s use the latter method. From the line
element the metric is diagonal gµν = diag −1, 1, ρ 2 + r2 , (ρ 2 + r2 ) sin2 θ , implying that
gµν = g−1 2 2 −1 2 2 −1 −2

µν = diag −1, 1, (ρ + r ) , (ρ + r ) sin θ .
From Eq. (7.30), for a diagonal metric the summation is restricted to a single term:
1 σ σ ∂ gµσ ∂ gλ σ ∂ gµλ

σ
Γλ µ = g + − (no sum).
2 ∂ xλ ∂ xµ ∂ xσ
By inspection of the metric, the only non-vanishing derivatives are
∂ g22 ∂ g33 ∂ g33
= 2r = 2r sin2 θ = 2(ρ 2 + r2 ) sin θ cos θ ,
∂ x1 ∂ x1 ∂ x2
implying that

∂ g22
Γ122 = 21 g11 − 1 = −r
∂x

∂ g33
Γ133 = 21 g11 − 1 = −r sin2 θ
∂x

2 2 1 22 ∂ g22 r
Γ12 = Γ21 = 2 g 1
= 2
∂x ρ + r2

∂ g33
Γ233 = 21 g22 − 2 = − sin θ cos θ
∂x

3 3 1 33 ∂ g33 r
Γ13 = Γ31 = 2 g 1
= 2
∂x ρ + r2

∂ g33
Γ332 = Γ323 = 12 g33 = cot θ .
∂ x2
are the non-vanishing connection coefficients.
7.6 This example is worked out in Section 3.5 of Carroll [63], using the method illustrated
in the first part of the Problem 7.4 solution above. The µ = 0 component of the Euler–
Lagrange equation is
d 2t dxi dx j
+ a ȧδi j = 0.
dτ 2 dτ dτ
Comparison with the geodesic equation (7.21) for x0 = t,

d 2x0 µ ν
0 dx dx
+ Γ µν = 0,
dτ 2 dτ dτ
then requires that
Γ000 = 0 Γ0i0 = Γ00i = 0 Γ0i j = aȧδi j .
For any of the spatial coordinates the Euler–Lagrange equation is

d 2xi ȧ dt dxi
+ 2 = 0,
dτ 2 a dτ dτ
and comparison with the corresponding geodesic equation
d 2xi dxµ dxν
2
+ Γiµν =0
dτ dτ dτ
requires that
ȧ
Γij0 = Γi0 j = δ ji Γi00 = Γijk = 0
a
for the remaining connection coefficients.
7.7 This solution is adapted from an example in Cheng [64]. Consider 3 great circles
BCB′C′ , BAB′ A′ , and ACA′C′ on the sphere (a) in the following figure:
C C C C
γ γ
B B B B
β β
α A' A' A' A'
A A A A
α
B' B' B' B'

C' C' C' C'
(a) (b) (c) (d)
The area bounded by two great circles is called a lune. By spherical symmetry the two
angles associated with a lune are equal. The area of a lune is given by A = 2R2 θ , where θ is
the angle between the great circles and R is the radius of the sphere. The three great circles
define a total of six lunes (three pairs) having angles α , β , and γ respectively. Consider the
three lunes marked by heavy lines and shading in figures (b), (c), and (d) above. The sum
of the areas of these three lunes covers the entire facing hemisphere of the sphere, plus the
area of spherical triangle ABC [since it is included in both the lune in (b) and the lune in
(c)], plus the area of spherical triangle A′ B′C′ [since it extends into the back hemisphere
for the lune in (d)]. But spherical triangles ABC and A′ B′C′ are congruent and the area of
the front hemisphere may be expressed as
2π R2 = 2R2 (α + β + γ ) − 2A∆,
where A∆ is the area of the spherical triangle ABC or A′ B′C′ . Solving the above equation
for the angular excess ε ≡ α + β + γ − π of the spherical triangle ABC gives
A∆
ε = α +β +γ −π = = KA∆ ,
R2
where K = 1/R2 is the Gaussian curvature for a 2-sphere. This specific proof for spherical
triangles on spheres may be generalized to arbitrary polygons on smooth curved surfaces,
basically by noting that any spherical polygon can be decomposed into spherical triangles
and any curved surface can be approximated locally by a spherical surface.
7.8 By the chain rule d/d σ = (dxν /d σ )∂ν , where ∂ν ≡ ∂ /∂ xν . Therefore,
d 2x µ dxν dxµ
= ∂ν .
dσ 2 dσ dσ
Substitute the covariant derivative (3.55) for the partial derivative ∂ν (with Aµ = dxµ /d σ )
to give
dxν dxµ d 2x µ ν
µ dx dx
α
∂ν −→ + Γνα ,
dσ dσ dσ 2 dσ dσ
which is the left side of the geodesic equation (7.21). Thus d 2 xµ /d σ 2 = 0 becomes the
geodesic equation if partial derivatives are replaced by covariant derivatives.
7.9 From the general expression (7.30) for the connection coefficient in terms of the metric
tensor and its derivatives,
Γσλ σ = 21 gνσ (gσ ν ,λ + gλ ν ,σ − gσ λ ,ν )

= 21 gνσ (gλ ν , σ − gσ λ ,ν ) + 12 gνσ gσ ν , λ ,
where the second line is just a rearrangement of the first. But the quantity on the second line
inside the parentheses is antisymmetric under the exchange σ ↔ ν , so when it is contracted
with the symmetric gνσ it must vanish. More formally, using the symmetry of the metric
tensor under exchange of indices and relabeling of dummy summation indices,
gνσ gλ ν ,σ = gσ ν gλ ν ,σ = gνσ gλ σ ,ν = gνσ gσ λ ,ν .
Thus, gνσ (gλ ν ,σ − gσ λ ,ν ) = 0 and Γλσ σ = 12 gνσ gνσ , λ .
7.10 (a) From the definition (3.57),
δβα;γ = ∂γ δβα + Γεγ

α ε
δβ − Γβε γ δεα = 0 + Γβα γ − Γβα γ = 0.
(b) Utilizing the result from part (a),
δβα;γ = 0 = (gαε gεβ );γ = gαε αε αε

;γ gεβ + g gεβ ;γ = g;γ gεβ ,
where we’ve used the product rule for covariant derivatives and that gεβ ;γ = 0 (metric
connection). Thus gαε
;γ gεβ = 0 and multiplying by g
β ν gives gβ ν gαε g αε ν
;γ εβ = g;γ δε = 0 and
thus that gαν
;γ = 0.
7.11 Using Eq. (3.65), the condition (7.17) for parallel transport on a path parameterized
by τ is
DAλ dAλ dxν
= + Γλµν Aµ = 0.
Dτ dτ dτ
Setting Aµ equal to the vector dxµ /d τ tangent to the path gives immediately
d 2xλ dxµ dxν

2
+ Γλµν = 0,
dτ dτ dτ
which is the geodesic equation (7.21). Therefore, a particle obeying the geodesic equa-
tion follows a path that is the straightest possible in spacetime, in that the path may be
constructed by parallel transport of its own tangent vector.
7.12 The inner product is gαβ Aα Bβ and its absolute derivative on a path parameterized by
u is
D(gαβ Aα Bβ ) D(gαβ ) α β DAα β DBβ
= A B + gαβ B + gαβ Aα ,
Du Du Du Du
since the absolute derivative obeys the usual Leibniz rule for derivatives of products. But
DAα /Du = DBβ /Du = 0 (definition of parallel transport of vectors) and D(gαβ )/Du =
0 (property of metric connection; see Section 7.8). Thus D(gαβ Aα Bβ )/Du = 0 on the
path and the inner product is unchanged by parallel transport if the connection is a metric
connection.
7.13 From the general expression (7.30),

∂ gµν ∂ g0ν ∂ gµ 0 1 νσ ∂ g0ν ∂ g0µ

1
Γσ0µ = gνσ + − ≃ g − ,
2 ∂ x0 ∂ xµ ∂ xν 2 ∂ xµ ∂ xν
where the first term has been neglected because the fields are assumed to vary slowly with
time. But gµν = ηµν + hµν (x), so that ∂ g µν /∂ xλ = ∂ hµν /∂ xλ and
∂ h 0ν ∂ h 0 µ

1
Γσ0µ = η νσ − ,
2 ∂ xµ ∂ xν
where terms have been retained only to lowest order in hµν . (We have been a a little cav-
alier here since we have assumed that since hµν is small its derivatives are also small.)
Restricting this expression to µ = 0 gives

σ 1 νσ ∂ h0ν ∂ h00 1 ∂ h00
Γ00 = η − ν ≃ − η νσ ν ,
2 ∂ x0 ∂x 2 ∂x
where the first term has been dropped because the fields are assumed to vary slowly.
7.14 By the equivalence principle the local inertial frame corresponds to a locally flat
space. From Eq. (7.30), the Christoffel symbols vanish because the first derivatives of the
metric are all equal to zero in the local inertial frame. However, the second derivatives of
the metric do not generally vanish so the first derivatives of (7.30) will not be zero.
7.15 The spherical and cartesian coordinates are related by x = r cos θ and y = r sin θ .
From the transformation law e′i = (∂ x j /∂ x′i )e j one obtains that the basis vectors in the
(r, θ ) coordinates are
e r = cos θ e x + sin θ e y e θ = −r sin θ e x + r cos θ e y .
Since
∂ ei
= Γkij e k ,
∂xj
the partial derivatives ∂ e i /∂ x j are required to evaluate the connection coefficients. From
the above equations, these are
∂ er
=0
∂r
∂ er 1
= − sin θ e x + cos θ e y = e θ
∂θ r
∂ eθ 1
= − sin θ e x + cos θ e y = e θ
∂r r
∂ eθ
= −r cos θ e x − r sin θ e y = −reer .
∂θ
Now the connection coefficients can be read off from the expansion. For example,
∂ er 1 1
= Γrrθ e r + Γθrθ e θ = e θ −→ Γrrθ = 0 Γθrθ = .
∂θ r r
Carrying this out for all eight index combinations yields
1
Γθrθ = Γθθ r = Γθr θ = −r Γrrr = Γθrr = Γrrθ = Γθr r = Γθθ θ = 0.
r
This may be checked using Eq. (7.30), which can be written for this 2D case as
1 mk ∂ g jm ∂ gim ∂ g ji

k
Γi j = g + − m ,
2 ∂ xi ∂xj ∂x
with i and j taking on the values r or θ . In polar coordinates the line element is ds2 =
dr2 + r2 d θ 2 , so the metric takes the form

1 0 ij −1 1 0
gi j = g = (gi j ) = .
0 r2 0 1/r2
Thus, grr = 1, gθ θ = r2 , grθ = gθ r = 0, and grr = 1, gθ θ = 1/r2 , gθ r = grθ = 0. Since the

metric is diagonal, only the m = k terms will survive in the above summation for connection
coefficient because of the factor gmk , so for example

θ 1 θ θ ∂ gθ θ ∂ gr θ ∂ gθ r 1
Γrθ = g + − = ,
2 ∂r ∂θ ∂θ r
which is the same result as obtained above. The other Γkik can be checked in the same way.
(See also the related Problem 7.4.)
7.16 An alert reader may observe that the substitution r = sin ϕ in the line element ds2 =
dr2 + r2 d θ 2 for flat 2D space expressed in polar coordinates gives the line element ds2 =
cos2 ϕ d ϕ 2 + sin2 ϕ d θ 2 given in the problem. Thus, the parameterization obscures that this
is just flat 2D space and the Gaussian curvature must be zero. Let’s check this conclusion
using Eq. (7.5) and ds2 = cos2 ϕ d ϕ 2 + sin2 ϕ d θ 2 . Consider a line drawn from the point
(ϕ , θ ) = (0, 0) to the point (λ , 0) at constant θ = 0. The length of the line is
Z √ Z λ
S= ds2 = cos ϕ d ϕ = sin λ .
0
The locus of points at constant S = sin λ for all values of θ then traces a circle of radius
S = sin λ . The circumference of this circle is given by the integral
Z Z 2π
C= ds = sin λ d θ = 2π sin λ .
0
Then from Eq. (7.5),

6 C 6 2π sin λ
K = Lim 2 1− = Lim 1− = 0,
λ →0 S 2π S λ →0 sin2 λ 2π sin λ
and the space indeed has zero intrinsic curvature.
7.17 In the coordinates (r, θ ), construct a line segment from the origin (0, 0) to the point
(λ , 0). From the metric ds2 = (1 + α r2 ) dr2 + r2 d θ 2 , the length of this segment is
Z √ Z λ Z λ
2 1/2
S= ds2 = (1 + α r ) dr ≃ (1 + 12 α r2 + . . .) dr = λ + 61 αλ 3 ,
0 0
where d θ 2 = 0 was used and the square root was expanded in a binomial series since our
interest is in the limit λ → 0. The points (r = λ , θ ) as θ ranges 0 to 2π define a circle of
radius S = λ + 16 αλ 3 , which has a circumference
Z √ Z 2π
3
C= ds2 = (λ + 1
6 αλ ) d θ = 2πλ + 13 παλ 3 .
0
Then from Eq. (7.5) the Gaussian curvature is

!
6 2πλ + 31 παλ 3
K = Lim 2 1− = −α ,
λ →0 λ 2πλ
where terms of order λ 3 have been ignored in denominators.
7.18 If the fluid is at rest u = (u0 , 0, 0, 0), and the normalization u · u = g µν uµ uν = −1

gives u0 u0 = −1, since u µ uν = 0 unless µ = ν = 0. Hence all non-diagonal elements of
µ
T ν vanish and
T 00 = (ε + P)u0 u0 + P = −(ε + P) + P = −ε T 11 = T 22 = T 33 = P
in the rest frame of a perfect fluid.
7.19 Suppose a vector field V (λ ) defined only along a curve xµ (λ ) parameterized by λ in

a manifold. Assuming that V (λ ) = V µ (λ )eµ (λ ), where eµ (λ ) is a coordinate basis vector
evaluated at the point on the curve labeled by λ , then

dV dV µ deµ
= eµ + V µ
dλ dλ dλ
dV µ ∂ eµ dxν
= eµ + V µ ν ,
dλ ∂ x dλ
where the chain rule was used in the last step. As in Section 2.3, expand the partial deriva-
tive factor in the vector basis,
∂ eµ
= Γαµν eα ,
∂ xν
which gives
dV dV µ dxν
= eµ + Γαµν V µ eα
dλ dλ dλ
dV µ µ dxν
= eµ + Γαν V α eµ
d λ µ d λ
ν
dV µ dx
= + Γαν V α eµ
dλ dλ
DV µ
≡ eµ ,
Dλ
where in line two dummy summation indices in the second term were interchanged. Thus
DV µ dV µ µ dxν
= + Γαν V α ,
Dλ dλ dλ
which is Eq. (3.65) for the absolute or intrinsic derivative.
7.20 (a) Differentiation of the 4-vector V gives
dV d dV µ deµ µ
= V µ eµ = eµ + V
dλ dλ dλ dλ
µ
∂ V dx ν ∂ eµ dxν µ
= e µ + V
∂ xν d λ ∂ xν d λ
= ∂ν V µ uν eµ + V µ uν ∂ν eµ ,
where we have defined

dxν ∂
uν ≡ ∂ν ≡ .
dλ ∂ xν
Expanding ∂ν eµ = Γαµν eα as in Section 2.3 gives
dV
= ∂ν V µ uν eµ + Γαµν V µ uν eα
dλ
µ
= ∂ν V µ + Γαν V α uν eµ ,

where dummy summation indices were switched in the second term. This is Eq. (7.14).
(b) From the first part
∂V µ dxν

dV µ
= + Γαν V α eµ .
dλ ∂ xν dλ
Multiply both sides by d λ /dxβ to give

µ
dV ∂V µ α
= + Γαβ V eµ .
dxβ ∂ xβ
Take the scalar product of both sides with eη to give
dV µ µ
≡ ∇ν V µ = ∂ν V µ + Γνα V α ,
dxν
where some indices have been renamed. Comparison with Eq. (3.55) indicates that this
defines the covariant derivative of the vector V .
8 The General Theory of Relativity
8.1 This problem was adapted from a similar one in Ref. [141]. From Eq. (7.30), the
connection coefficients are
∂ g jℓ ∂ giℓ ∂ gi j

Γkij = 12 gℓk + − ,
∂ xi ∂xj ∂ xℓ
where i, j, k, ℓ take the values 0 or 1. From the line element the metric is

−1 0 ij −1 0
gi j = g = .
0 f (t)2 0 f (t)−2
The formula may now be used to compute the Γkij . For example,

∂ g11 ∂ g01 ∂ g10
Γ101 = Γ110 = 21 g11 + −
∂ x0 ∂ x1 ∂ x1
∂ g11 1 f˙
= 21 g11 0 = 2 (2 f f˙) = ,
∂x 2f f
where f˙ ≡ ∂ f /∂ x0 . Determining the other possibilities in a similar way gives
Γ101 = Γ110 = f˙/ f Γ011 = f f˙ Γ000 = Γ100 = Γ111 = Γ001 = Γ010 = 0.
Then from Eq. (8.14)
∂ Γ011 ∂ ( f f˙) f˙
R0101 = 0
− Γ011 Γ110 = 0
− f f˙ = f (t) f¨(t).
∂x ∂x f
for the curvature tensor component R0101 . Since the space is two-dimensional, there is only
one independent curvature component (see Box 8.1) and all other components must be zero
or related to R0101 by the symmetries (8.15).
8.2 Take the line element in the form
ds2 = −eσ dt 2 + eλ dr2 + r2 (d θ 2 + sin2 θ d ϕ 2 ),
where σ = σ (r,t) and λ = λ (r,t). The corresponding metric is
g µν = diag (−eσ , eλ , r2 , r2 sin2 θ )

−σ −λ 1 1
gµν = g−1
µν = diag −e , e . ,
r2 r2 sin2 θ
42
From Eq. (7.30) the non-vanishing Christoffel symbols are
Γ000 = 12 σ̇ Γ001 = 12 σ ′ Γ011 = 21 eλ −σ λ̇ Γ100 = 21 eσ −λ σ ′ Γ101 = 12 λ̇

Γ111 = 12 λ ′ Γ122 = −re−λ Γ133 = −re−λ sin2 θ Γ212 = 1/r
Γ233 = − sin θ cos θ Γ313 = 1/r Γ323 = cot θ
where dots indicate time derivatives and primes indicate derivatives with respect to r.
The Riemann curvature tensor may then be calculated from Eq. (8.14); the non-zero
components are
R0101 = 12 eσ σ ′′ − 14 eλ λ̇ 2 + 14 eλ σ̇ λ̇ − 21 eλ λ̈ + 14 eσ (σ ′ )2 − 14 eσ σ ′ λ ′
R0202 = 21 eσ −λ σ ′ R0212 = 12 rλ̇ R0303 = 21 reσ −λ σ ′ sin2 θ
R0313 = 12 rλ̇ sin2 θ R1212 = 12 rλ ′
R1313 = 21 rλ ′ sin2 θ R2323 = r2 sin2 θ (1 − e−λ ).
The Ricci tensor Rµν then follows from Eq. (8.16), with non-vanishing components
R00 = − 21 eσ −λ σ ′′ − 14 eσ −λ (σ ′ )2 + 21 λ̈ + 14 λ̇ 2 − 14 σ̇ λ̇ + 41 eσ −λ σ ′ λ ′ − eσ −λ λ ′ /r
R11 = 12 σ ′′ + 41 (σ ′ )2 − 12 eλ −σ λ̈ − 41 eλ −σ λ̇ 2 + 14 eλ −σ σ̇ λ̇ − 14 σ ′ λ ′ − λ ′ /r
R01 = −λ̇ /r R22 = 12 re−λ σ ′ − 21 re−λ λ ′ + e−λ − 1 R33 = R22 sin2 θ .
The Ricci scalar R may then be constructed from Eq. (8.17),
e− λ e− λ ′ e− λ ′
R=2 2
+2 σ −2 λ + e−λ σ ′′ + 12 e−λ (σ ′ )2
r r r
2
− e−σ λ̈ − 21 e−σ λ̇ 2 − 2 + 12 e−σ σ̇ λ̇ − 12 e−λ σ ′ λ ′ .
r
and the Einstein tensor is
eσ eσ − λ eσ − λ ′ λ̇ eλ 1 σ′
G00 = 2
− 2 + λ G01 = − G11 = 2 − 2 −
r r r r r r r
G22 = 21 rλ ′ e−λ − 12 rσ ′ e−λ − 12 r2 e−λ σ ′′ − 14 r2 e−λ (σ ′ )2 + 12 r2 e−σ λ̈
+ 14 r2 e−σ (λ̇ )2 − 14 r2 e−σ σ̇ λ̇ + 14 r2 e−λ σ ′ λ ′
G33 = G22 sin2 θ ,
where Eq. (8.20) was used. These results for a general spherical metric are summarized in
Appendix C.
8.3 To calculate the commutator requires taking two successive covariant derivatives on
an arbitrary vector, and then subtract from that the result of those two successive covari-
ant differentiations taken in the opposite order. But in taking the covariant derivative and
then taking the covariant derivative of the result it must be remembered that the covariant
derivative of the vector yields a rank-2 tensor, so the second covariant derivative is not that
of a vector but of a rank-2 tensor. Utilizing Eq. (3.53) in the first step and Eq. (3.56) on the
result, and that the covariant derivative obeys the usual Leibniz rule for the derivative of
products, the result of two successive covariant differentiations with respect to ν and then
λ on an arbitrary vector Vµ is
Vµ ; νλ = ∂λ (Vµ ; ν ) − Γαµλ Vα ; ν − Γνλ

α
Vµ ; α
= ∂λ ∂ν Vµ − (∂λ Γσµν )Vσ − Γσµν ∂λ Vσ
− Γαµλ (∂ν Vα − Γαν
σ α
Vσ ) − Γνλ (∂α Vµ − Γσµα Vσ ),
where ∂β ≡ ∂ /∂ xβ . Form an analogous expression but with ν and λ interchanged and

subtract it from the preceding expression to form the commutator [∇ν , ∇λ ]Vµ ≡ Vµ ; νλ −
Vµ ; λ ν . In the resulting expression various terms cancel because (1) repeated indices are
dummy indices that can be replaced by arbitrary repeated indices, (2) ∂ν ∂λ = ∂λ ∂ν , and
(3) the Christoffel symbols are symmetric in their lower indices. The surviving terms give
[∇ν , ∇λ ]Vµ = Rσ µνλ Vσ ,
where the definition (8.14) of the Riemann curvature tensor Rσ µνλ was used. Since the
dual vector Vµ is arbitrary, this proves that the curvature tensor is the commutator of the
covariant derivatives.
8.4 The deviation is given by the second term in Eq. (8.12):

GM −28 M 1cm
h00 = 2 2 = 1.483 × 10
rc 1g r
Estimates for several objects are shown in the following table.
Object Mass (g) Radius (cm) h00
Sun 1.989 × 1033 6.96 × 1010 4.23 × 10−6

Earth 5.976 × 1027 6.378 × 108 1.4 × 10−9
Proton 1.67 × 10−24 10−13 2.5 × 10−39
Sirius B 2.09 × 1033 5.5 × 108 5.6 × 10−4
Neutron star ∼ 2 × 1033 106 0.30
Among these examples, only the surface of a neutron star exhibits substantial deviation
from the flat metric. Therefore, Newtonian gravity should be a good approximation for
all of these except the neutron star, for which deviations from Newtonian gravity may be
substantial.
8.5 For the µ = 0 part of Eq. (8.4) d 2 x0 /d τ 2 = 0. For the µ = i components, from (8.4)
and (8.8),
2
d 2xi 1 ∂ h00 dx0

− = 0.
d τ 2 2 ∂ xi dτ
But for the first term

d 2xi dxi dx0 dx0 d dxi

d
= =
dτ 2 dτ dx0 d τ d τ d τ dx0
0 2
dx0 d dx0 dxi
i
dx d dx
= =
d τ dx0 d τ dx0 dτ dx0 dx0
2 i 2 2 i
1 dx0 1 dx0

d dx d x
= 2 = 2
c dτ dt dt c dτ dt 2
where dx0 = cdt and dx0 /d τ = constant has been used. Inserting this result in the first
equation above yields the second equation of Eq. (8.9).
8.6 In the rest frame of a clock falling freely special relativity is valid (equivalence prin-
ciple). The special relativistic time dilation formula gives
1/2
v2 v2

dτ = 1 − 2 dx0 ≃ 1 − 2 dx0 ,
c 2c
where the last step follows from the assumption of low velocity. By energy conservation,
1 2 2
2 mv = −mϕ , where ϕ is the gravitational potential, so v = −2ϕ . Inserting this in the
preceding equation gives
ϕ
d τ = 1 + 2 dx0 .
c
But from Eq. (8.10) in the weak-field limit,
2ϕ 1/2

1/2 ϕ
(−g00 ) = 1+ 2 ≃ 1+ 2.
c c
Therefore, for objects falling slowly in weak gravitational fields the proper time τ and
coordinate time x0 are related by d τ = (−g00 )1/2 dx0 .
8.7 This solution is adapted from one given in Cheng [64]. The metric tensor can be
expressed as

g11 0 ij 1/g11 0
gi j = g =
0 g22 0 1/g22
The non-trivial independent element of the Riemann curvature tensor (8.14) in two dimen-
sions may be taken to be
R1212 = g1k Rk212 = g11 R1212

1
∂ Γ21 ∂ Γ122

1 1 1 2 1 1 1 2
= g11 − − Γ Γ
11 22 − Γ Γ
21 22 + Γ Γ
12 21 + Γ Γ
22 21 .
∂ x2 ∂ x1
The required connection coefficients are given by Eq. (7.30). For example:

1 ∂ g11 ∂ g11 ∂ g11 1 ∂ g11 1 ∂ g11
Γ111 = g11 + − = g11 1 = .
2 ∂ x1 ∂ x1 ∂ x1 2 ∂x 2g11 ∂ x1
In a similar manner
1 ∂ g11 1 ∂ g22
Γ112 = Γ121 = Γ122 = −
2g11 ∂ x2 2g11 ∂ x1
1 ∂ g22 1 ∂ g22
Γ212 = Γ221 = Γ222 = .
2g22 ∂ x1 2g22 ∂ x2
Inserting these connection coefficients in the expression for R1212 gives
( " #
1 ∂ 2 g22 ∂ 2 g11 ∂ g11 2

1 ∂ g11 ∂ g22
R1212 = + − +
2 ∂ (x1 )2 ∂ (x2 )2 2g11 ∂ x1 ∂ x1 ∂ x2
" #)
∂ g22 2

1 ∂ g11 ∂ g22
− + ,
2g22 ∂ x2 ∂ x2 ∂ x1
which is Eq. (7.2), up to a normalization factor −1/ detg.

8.8 From the symmetries (8.15) it may be shown that all contractions of one upper index
on the curvature tensor with one of its lower indices either gives ±Rµν defined in Eq.
(8.16), or zero. For example
σ
gλ σ Rµλ σ ν = Rµ σν ≡ R′µν .
But also since Rσ µνλ = −Rµσ νλ ,
gλ σ Rµλ σ ν = −gλ σ Rλ µσ ν = −Rσ µσ ν = −Rµν .
Thus, R′µν = −Rµν . As another example, multiply Rσ µνλ = −Rµσ νλ , by gσ µ to give

µ µ µ
gσ µ Rσ µνλ = −gσ µ Rµσ νλ −→ R µνλ = −Rσ σ νλ −→ R µνλ = −R µνλ
µ
where dummy (repeated) indices were relabeled. This can be true only if R µνλ = 0.
8.9 By symmetry Rβ µαν = Rανβ µ . Therefore,
β
Rµν = gαβ Rβ µαν = gαβ Rανβ µ = R νβ µ = Rν µ ,
so Rµν is symmetric. Alternatively, start from the cyclic identity in Eq. (8.15) and raise the
first index by contracting with gασ to give
Rα µνλ + Rα λ µν + Rα νλ µ = 0.
Now set α = λ to contract on that index:
Rα µνα + Rα α µν + Rα να µ = 0.
But in Problem 8.8 it was shown that Rα α µν = 0, and from Eq. (8.15) the first term in
the above equation changes sign if the last two indices are switched. Thus −Rα µαν + 0 +
Rα να µ = 0, which is equivalent to Rµν = Rν µ .
8.10 Parameterize a point on a circle of latitude for a sphere by spherical coordinates
s
x1 ≡ θ x2 ≡ ϕ = ,
R sin θ
where R is the radius of the sphere, θ is the usual polar angle (so latitude is π2 − θ ), s is the
arc length around a circle of latitude measured from ϕ = 0, and R sin θ is the radius of a
circle of latitude. The geodesic equation for this 2D space may then be written as
d 2 xi j
i dx dx
k
+ Γ jk =0 (i, j, k = 1 or 2).
ds2 ds ds
The derivatives required for the geodesic equation are
d 2 x1 dx1 dx2 1 d 2 x2
2
= =0 = = 0.
ds ds ds R sin θ ds2
Since all second derivatives vanish and the first derivative vanishes for x1 , only two geodesic
equations are non-trivial:
dx2 dx2 Γ1 dx2 dx2 Γ2

Γ122 = 2 222 = 0 Γ222 = 2 222 = 0
ds ds R sin θ ds ds R sin θ
From Example 8.2, all the connection coefficients are zero except for
Γ122 = − sin θ cos θ Γ212 = Γ221 = cot θ .
Thus the second geodesic equation is satisfied identically because Γ222 = 0 and the first
geodesic equation gives for a particular choice of θ ,
Γ122 − sin θ cos θ − cot θ

2
= 2
= = 0.
2
R sin θ 2
R sin θ R2
But cot θ = 0 is satisfied only for θ = π2 in the interval 0 − π , so the equator is the only
circle of latitude that is a geodesic on a sphere. Since the orientation of the spherical co-
ordinate system is arbitrary, this also proves the well-known result that the geodesics of a
sphere are the great circles that can be drawn on the sphere. This exercise also illustrates an
important conceptual point: the geodesic equation describes the motion of a free particle in
a gravitational field, so it may seem surprising from the usual perspective that it was used
here to deduce a purely geometrical property of the sphere. But from the general relativity
perspective gravity is geometry, so the geodesic equation is in fact a geometrical statement
about the corresponding space.
8.11 Under most conditions where Newtonian gravity is valid the pressure contribution
in the stress–energy tensor may be ignored and Eq. (7.9) may be written as Tµν ≃ ε uµ uν .
The metric is assumed to be of the form gµν = ηµν + hµν , where h µν is a small correction,
so the Minkowski metric ηµν = diag (−1, 1, 1, 1) may be used to raise and lower indices.
Contracting Tµν with η λ ν gives Tµλ = ε uµ uλ and thus
µ
Tr T = Tµ = ε uµ uµ = −ε = −ρ c2 ,
where we have assumed the fluid to be nearly at rest so that the 4-velocities have only
timelike components and uµ uµ ∼ −1 (see the solution of Problem 7.18). Furthermore,
Tµν = ε uµ uν = εηµν uµ uµ = −εηµν = −ρ c2 ηµν ,

so that T00 = −η00 ρ c2 = ρ c2 . Taking the 00 component of the Einstein equations (8.23)
and inserting these results for the stress–energy tensor gives
8π G 8π G 4π G
R00 = 4 T00 − 21 g00 Tλλ ≃ 4 T00 − 12 η00 Tλλ = 2 ρ .
c c c
In Eq. (8.16) one may assume the connection coefficients to be small in the weak-field
limit and ignore the terms quadratic in Γλµν to give a second expression for R00 ,
R00 ≃ ∂λ Γλ00 − ∂0 Γλ0λ ≃ ∂i Γi00 (i = 1, 2, 3),
where the fields were assumed to be varying slowly in time, justifying neglect of all time-
like derivatives. But from Eq. (8.8), Γi00 = − 12 ∂i h00 in the weak-field limit and thus
R00 = − 21 ∂i ∂i h00 = − 21 ∇2 h00 .
Equating the two expressions derived above for R00 gives

4π G
− 21 ∇ 2 h00 = ρ,
c2
and using Eq. (8.10) to eliminate h00 in favor of a scalar field ϕ ≡ − 21 h00 c2 gives
∇ 2 ϕ = 4 π Gρ .
This is the Poisson equation (8.1) that governs Newtonian gravitation.

8.12 Assume a general spherical metric with line element
where σ and λ are positive and independent of time. From Eqs. (7.10) and (3.57) the
equation to be solved is
µ µ µ µ
Tν ;µ = Tν ,µ + Γα µ Tνα − Γναµ Tα = 0.
The non-zero connection coefficients required are given in the solution to Problem 8.2 and
in Appendix C (but here all connection coefficients proportional to time derivatives are
neglected, since the metric is static). The preceding equation corresponds to one equation
for each of the four possible value of ν , with each equation involving implied sums over
the repeated indices µ and α . When written out there are many terms, but a large number
are identically zero because by inspection either the connection coefficient vanishes or the
µ
term is not diagonal in Tν . Collecting the terms that survive gives
P′ + Γ010 Tν1 + Γ111 Tν1 + Γ212 Tν1 + Γ313Tν1 + Γ323Tν2

− Γ0ν 0 T00 − Γν1 1 T11 − Γ2ν 2 T22 − Γν3 3 T33 = 0,
where a prime denotes a partial derivative with respect to r. This represents four separate
equations for the respective choices ν = 0, 1, 2, 3. The only non-trivial result corresponds
to setting ν = 1, which gives
σ′
P′ + (P + ρ ) = 0,
2
upon substituting the expressions for the connection coefficients and using T00 = −ρ and
T11 = T22 = T33 = P.
8.13 From the solution of Problem 8.3, the commutator of the covariant derivative is re-
lated to the curvature tensor by
[∇ν , ∇λ ]Vµ = Rσ µνλ Vσ .
The covariant differentiation of vectors on the left side of this equation means that the
left side is tensorial and thus the right side must also be a tensor. Hence, by the quotient
theorem of Problem 3.13 the components Rσ µνλ that contract with the dual vector on the
right side must be the elements of a tensor. This is a much faster proof than checking Rσ µνλ
explicitly against the rank-4 tensor transformation law.
8.14 In a local inertial frame the connection coefficients vanish but not their first deriva-
tives. Therefore, from Eq. (8.14) and Eq. (7.30), in this frame at an arbitrary point
Rσ µνλ = gσ κ Rκ µνλ = 21 (∂λ ∂σ gµν − ∂λ ∂µ gσ ν + ∂ν ∂µ gσ λ − ∂ν ∂σ gµλ ),
where ∂α = ∂ /∂ xα . The symmetries
Rσ µνλ = −Rµσ νλ = −Rσ µλ ν Rσ µνλ = Rνλ σ µ
of Eq. (8.15) follow immediately by exchanging indices and noting that the order of dif-
ferentiation doesn’t matter and the metric tensor is symmetric in its indices. These results
were obtained in a special frame but since the relations are tensor relations they are valid
in all frames.
8.15 From Eq. (8.18) the Bianchi identity is
∇λ Rµναβ + ∇β Rµνλ α + ∇α Rµνβ λ = 0.
Contract this with gµα , remembering that since ∇ µg
µν = 0, raising an index by contraction
commutes with covariant differentiation,
∇λ gµα Rµναβ + ∇β gµα Rµνλ α + ∇α gµα Rµνβ λ = 0
→ ∇λ gµα Rµναβ − ∇β gµα Rµναλ + ∇α gµα Rµνβ λ = 0
→ ∇λ Rνβ − ∇β Rνλ + ∇α gµα Rµνβ λ = 0,
where in the second line the last two indices were switched in the second term using Eq.
(8.15) so that the contraction is consistent with the definition (8.16) of the Ricci tensor [see
the footnote following Eq. (8.16) and Problem 8.8], and in the last line Eq. (8.16) was used.
Now contract with gνβ to give
∇λ gνβ Rνβ − ∇β gνβ Rνλ + ∇α gµα gνβ Rµνβ λ = 0
β
→ ∇λ R − ∇β R λ − ∇α gµα gνβ Rν µβ λ = 0
β
→ ∇λ R − ∇β R λ − ∇α gµα Rµλ = 0
β
→ ∇λ R − ∇β R λ − ∇α Rα λ = 0
→ ∇λ R − 2∇α Rα λ = 0,
where in the second line the first two indices on R in the last term were switched to make
the contraction compatible with (8.16) and the definition (8.17) was used, and in the last
line the dummy summation indices were switched to the same variable so the last two
terms could be added. Now contract with gµλ to raise the index on the last term,
∇λ Rgµλ − 2∇α gµλ Rα λ = 0
→ ∇λ Rgµλ − 2∇α Rµα = 0
→ ∇ν Rgµν − 2∇ν Rµν = 0,
where dummy summation indices have been switched. Finally, multiply both sides by − 12
to give
∇ν (Rµν − 12 Rgµν ) = 0,
which is ∇µ Gµν = 0 for the symmetric Einstein tensor Gµν ≡ Rµν − 12 Rgµν defined in Eq.
(8.20).
9 The Schwarzschild Spacetime
9.1 It is convenient to introduce an exponential parameterization B(r) ≡ eν (r) and A(r) ≡

eλ (r) , so that the metric is

−ν −λ 1 1

ν λ 2 2 2 µν
gµν = diag −e , e , r , r sin θ g = diag −e , e , 2 , 2 2 .
r r sin θ
The unknown functions ν (r) and λ (r) may then be determined by requiring that the metric
be consistent with the vacuum (vanishing stress–energy tensor) Einstein equation. How-
ever, for a vacuum solution it is not necessary to construct the full Einstein tensor Gµν
because it may be shown (see Problem 22.1) that solution of the vacuum Einstein equation
is equivalent to solving Rµν = 0, where Rµν is the Ricci tensor. The metric and Eq. (7.30)
γ
give the Christoffel symbols (affine connections) Γαβ , and from these and Eqs. (8.14) and
(8.16) the Ricci tensor may be constructed. Setting Rµν = 0 yields four equations, only
three of which are independent. In particular,
ν′

ν −λ 1 ′′ 1 ′ 2 1 ′ ′
−R00 = e 2 ν + 4 (ν ) − 4 ν λ + r =0
λ′
R11 = 21 ν ′′ + 14 (ν ′ )2 − 41 ν ′ λ ′ − =0
r
R22 = e−λ 21 (ν ′ − λ ′ )r + 1 − 1 = 0,

where primes indicate derivatives with respect to r. The first two equations imply that
ν ′ = −λ ′ and thus that ν = −λ + constant. A time-independent solution is sought so the
timescale may be shifted freely to make the constant zero and obtain ν = −λ . Inserting this
into the R22 equation gives reν ν ′ + eν = 1, the left side of which is equivalent to d(reν )/dr,
implying that eν = 1 − c/r, where c is a constant. Choosing c = 2M, where M is another
constant (that will be interpreted as the mass when compared with Newtonian gravity),
2M −1

ν 2M λ −ν
e = 1− e = e = 1− ,
r r
which gives the Schwarzschild metric (9.5) when inserted into the equation for g µν above.
9.2 Use Eq. (9.30) to set dVeff /dr = 0 and obtain

r
ℓ2 1 ℓ4
r± = ± − 12 ℓ2,
2M 2 M 2
from which the desired results follow.
9.3 The radius of the innermost stable circular orbit is RISCO = 6M. The second derivative
51
of the effective potential (9.30) is
d 2Veff 2M 36M 2 144M 3

2
=− 3 + 4 − ,
dr r r r5
which gives zero when evaluated at r = 6M. Thus the ISCO is at a point of inflection in the
potential and it is marginally stable.
9.4 From Eq. (9.43),

6π GM −27 M cm 1
δϕ = = 1.398 × 10 rad/orbit
ac2 (1 − e2) g a 1 − e2

M AU 1
= 1.858 × 10−7 rad/orbit,
M⊙ a 1 − e2
Some results calculated with this formula are given in the following table.
Object Central mass (M⊙ ) a (AU) e Period δ ϕ /∆t
Mercury 1 0.387 0.206 88 d 43′′ /century

Earth 1 1 0.017 1 yr 3.8′′ /century
Binary Pulsar∗ 2.828 1.3 × 10−2 0.617 7.75 hr 4.2◦ /yr
∗ Mass is total mass of binary; a is average separation
π
9.5 From Eq. (9.23), ℓ = r2 (d ϕ /d τ ) if θ = is assumed, and for classical orbital motion
2

2 dϕ 2π
p
r = 1−e 2 a2 .
dt P
Combining these equations assuming weak gravity gives
2 2
2 dϕ 2 dϕ 4π 2

2
ℓ = r ≃ r = 2 (1 − e2)a4 = (1 − e2)GMa,
dτ dt P
where in the last step P2 = (4π 2 /GM)a3 (Kepler III) was used. Inserting this result in Eq.
(9.42) then gives Eq. (9.43).
9.6 Differentiate the expression for Veff in Eq. (9.55) with respect to r, set to zero, and solve
to give r = 3M = 23 rs for the maximum of the potential, implying an unstable circular orbit
at this radius. Substitute this value of r to give
1
Veff (r = 3M) =
27M 2
for the height at the maximum.
9.7 For the following diagram,

Photon
r
d
φ
x
define polar coordinates in the θ = π /2 plane,

x = r cos ϕ y = r sin ϕ .
Then the parameter b is
r2 sin2 θ d ϕ /d λ

ℓ
b = =
ε (1 − 2M/r)dt/d λ
r2 dϕ
=
(1 − 2M/r) dt
r2 d ϕ dr
= ,
(1 − 2M/r) dr dt
where Eqs. (9.53)–(9.55) have been used. But for r >> 2M in the Schwarzschild metric,
2M d dr
1− →1 ϕ→ → −1.
r r dt
Therefore, d ϕ /dr ≃ −d/r2 and
r2

d ϕ dr −d
b= = r2 (−1) = d,
(1 − 2M/r) dr dt r2
so b may be interpreted as the impact parameter for the photon.
9.8 From the static Schwarzschild metric restricted to radial coordinates (dt = d θ = d ϕ =
0), the coordinate distance dr is related to the proper distance ds by dr = (1 − 2M/r)1/2 ds.
In geometrized units M = 2.95 km and R = 7 × 105 km for the Sun. Since the rod is very
short the gravitational field may be assumed to be uniform over its length, giving the esti-
mate
2M 1/2

∆r ≃ 1 − ∆s ≃ 0.9999958 cm
r
for the coordinate length of the rod at the surface of the Sun. A neutron star packs of order
a solar mass into a radius of roughly 10 km. Thus for a neutron star 2M/r ≃ 0.59 and
∆r ≃ 0.64 cm, indicating a significant distortion of the metric from that of flat space at its
surface.
9.9 Compare the line element
ds2 = −(1 − 2M/r)dt 2 + α (ρ )2 d ρ 2 + ρ 2 d θ 2 + ρ 2 sin2 θ d ϕ 2 ,

term by term with the standard Schwarzschild line element (9.4). Requiring the coefficients
of d θ 2 + sin2 θ d ϕ 2 to be equal in the two expressions gives r2 = α 2 ρ 2 and equating the

radial terms requires (1 − 2M/r)−1 dr2 = α 2 d ρ 2 . Combining these last two expressions
gives
dρ dr
= ±√ .
ρ 2
r − 2Mr
Choose the positive sign by assuming ρ → ∞ for r → ∞ and integrate to give
p
ln ρ = ln −M + r + r2 − 2Mr + ln b,
where ln b is an arbitrary integration constant. Choose the integration constant such that
ρ (r = 2M) = M/2 (see Problem 9.10 for a justification of this choice) to yield r = ρ (1 +
M/2ρ )2 and thus α 2 = (1 + M/2ρ )4 . Inserting these in the first expression above and a bit
of algebra gives the isotropic form of the Schwarzschild line element,
(1 − M/2ρ )2 2
ds2 = − dt + (1 + M/2ρ )4 d ρ 2 + ρ 2 (d θ 2 + sin2 θ d ϕ 2 ) .

(1 + M/2ρ )2
This form has the merit of making time slices t = constant look as much as possible like
euclidean space.
9.10 Solve the quadratic equation r = ρ (1 + M/2ρ )2 derived in Problem 9.9 for ρ to give
p
−(M − r) ± (M − r)2 − M 2
ρ= ,
2
from which we may conclude
1. The variable ρ is complex if 0 ≤ r < 2M, so the region inside the Schwarzschild radius
rs = 2M is not in the real domain of ρ .
2. If r > 2M, each r corresponds to two real positive values of ρ , so this region of r-space
is mapped out twice in ρ -space, once in the region M/2 ≤ ρ < ∞ and once in the region
0 ≤ ρ ≤ M/2.
This second point is the reason for the choice of boundary condition ρ (r = 2M) = M/2 in
Problem 9.9.
9.11 This problem is based on an example discussed in Invitation to Astrophysics Astro-
physics, T. Padmanabhan (World Scientific) 2006. The coordinate systems are related by
dt¯ = dT and d r̄r = drr − v dT , which when substituted into the metric ds2 = −dt¯2 + d r̄r 2
expressed in the freely-falling coordinates gives
r
2 2M 2 2M
ds = − 1 − dT − 2 dr dT + dr2 + r2 d θ 2 + r2 sin2 θ d ϕ 2 .
r r
Introducing a new time coordinate t through the linear transformation
p
2M/r
dT = dt − dr
1 − 2M/r
and a little algebra eliminates the non-diagonal term and yields the Schwarzschild line
element of Eq. (9.4).
9.12 This problem is adapted from an example in Taylor and Wheeler [229]. Integrating
the relationship (9.7) between the proper interval ds and the coordinate radial interval dr
gives
2M −1/2 r1/2 dr z2 dz
Z Z Z
∆s = 1− dr = √ =2 √
r r − 2M z2 − 2M
p p
= z z2 − 2M + 2M ln z + z2 − 2M ,
where the substitution r = z2 with dr = 2zdz has been used. Rewriting in terms of r and
evaluating between the two spherical shell radii gives finally
hp √ √ ir2
∆s = r(r − 2M) + 2M ln r + r − 2M .

r1
For a 1 M⊙ black hole 2M = 2.954 km in geometrized units and for spherical shells with
r1 = 5 km and r2 = 6 km, this formula yields ∆s = 1.474 km for the physical distance
between them. This is much larger than the coordinate difference of 1 km, indicating a
large distortion of spacetime relative to a flat metric (see Fig. 9.1). The following figure
illustrates the variation of the proper radial separation between spherical shells with radial
coordinates r1 and r2 = r1 + 1 km as a function of r1 around a 1 solar mass black hole.
∆s(r1, r1+1) (km)
r1(km)
At large distance the proper separation approaches the coordinate separation of 1 km, but
near the event horizon at r ∼ 2.954 km the curvature is very large and the proper separation
is much larger than the coordinate separation.
9.13 Write the integral of Eq. (9.49) in the form
1 r3/2 dr
Z
t = t0 − √ ,
2M r − 2M
where t0 is an integration constant. Then substitute r = x2 and apply tabulated results for
integrals of the form
xm dx
Z
ax2 + bx + c
√
iteratively to do the integration. Substitution of x = r and some algebra then gives Eq.
(9.50).
9.14 For simplicity, take the circle to lie in the θ = π2 plane, centered on r = 0. Then
dt = dr = d θ = 0 and sin2 θ = 1. The Schwarzschild line element (9.4) then reduces to
ds2 = r2 d ϕ 2 and the proper circumference C is
Z Z 2π
C= ds = rd ϕ = 2π r.
0
This is formally the same result as for flat space. However C is a physical (proper) distance
since it was computed from the metric, but r is a coordinate and not a physical distance.
Conversely, in flat space both C and r in C = 2π r can be interpreted as physical distances.
9.15 (a) Combine Eqs. (9.30) and (9.29) with the requirement that Veff = E to give Eq.
(9.32).
(b) The angular velocity is given by
dϕ
Ω= .
dt
Substitute Eqs. (9.22) and (9.23) to give Eq. (9.33).
(c) From Eqs. (9.25), (9.5), and (9.35),

2M
− 1− (ut )2 + r2 (ut )2 Ω2 = −1,
r
This may be solved using Eq. (9.34) to give,
3M −1/2

ut = 1 − ,
r
9.16 Since u = (ut , 0, 0, ut Ω) and s · u = 0, utilizing the metric (9.5),

µ ν 2M t t
s · u = gµν s u = − 1 − s u + R2 sϕ uϕ = 0,
R
from which
R2 sϕ uϕ R2 Ω
st = = sϕ ,
(1 − 2M/R)ut 1 − 2M/R
where Eqs. (9.35) and (9.36) was used in the last step. This is Eq. (9.61).
9.17 (i) Our solution follows that of Hartle [110], Ch. 14. Utilizing
u = (ut , 0, 0, ut Ω) s = (st , sr , 0, sϕ ),
r sϕ uϕ .
the only non-vanishing combinations for the sums in Eq. (9.62) are Γttr st ut and Γϕϕ
From Appendix C, for a general spherical metric with line element

the required connection coefficients are
Γttr = Γ100 = 21 eσ −λ σ ′ Γrϕϕ = Γ133 = −re−λ sin2 θ ,

π
where σ ′ ≡ d σ /dr. Specializing to the Schwarzschild metric with θ = 2 and letting r = R,
2M −1

2M 2M
eσ = 1 − eλ = 1 − σ = ln 1 −
R R R
2M −1

′ dσ 2M
σ = = 2 1− sin2 θ = 1,
dr R R
and the connection coefficients evaluate to

M 2M
Γttr = 2 1 − Γrϕϕ = −(R − 2M)
R R
Therefore Eq. (9.62) becomes
dsr M

2M t t
+ 2 1− s u − (R − 2M)sϕ uϕ = 0.
dτ R R
Utilizing ut = dt/d τ and Eq. (9.61), this may be written as
dsr
− (R − 3M)Ωsϕ = 0.
dτ
(ii) Now consider Eq. (9.63). By similar considerations as above, the only nonvanishing
ϕ
contribution to the sum in the second term is Γrϕ sr uϕ . From Appendix C the single required
ϕ
connection coefficient is Γrϕ = r−1 and Eq. (9.63) becomes
dsϕ Ω r
+ s = 0,
dt R
where ut = dt/d τ and uϕ = Ωut from Eq. (9.35) were used (with Ω = d ϕ /dt).
(iii) Thus the two equations
dsr dsϕ Ω r
− (R − 3M)Ωsϕ = 0 + s =0
dτ dt R
must be solved simultaneously. Take d/dt of the second equation and plug in dsr /dt from
the first equation to give
d 2sϕ
+ ω 2 sϕ = 0,
dt 2
where ω ≡ (1 − 3M/R)1/2Ω. This is the equation of a harmonic oscillator and the solutions
are given by Eq. (9.64).
9.18 From Eq. (9.68) evaluated assuming a satellite in orbit around the Earth, the geodetic
precession per orbit is

3π GM km
∆ϕ ≃ 2 = 8.6 arcsec orbit−1 .
c R R
For GP-B, setting R = a = 7027.4 km gives 1.22 × 10−3 arcsec orbit−1 . By Kepler’s 3rd
law, the period for a circular satellite orbit around the Earth is
s
4π 2 a3 a 3/2
P= = 9.95 × 10−3 s.
GM⊕ km
For GP-B with a = 7027.4 km this gives a period of 97.7 minutes, which translates to
5.38 × 103 orbits yr−1 . Therefore general relativity predicts that GP-B should exhibit a
geodetic precession rate of
∆ϕ
= (1.22 × 10−3 arcsec orbit−1 ) × (5.38 × 103 orbit yr−1 ) ≃ 6.6 arcsec yr−1 .
∆t
As discussed in Box 9.3, GP-B measured a geodetic precession rate for its gyroscopes that
was within 0.07% of this value.
9.19 The precession rate is given by Eq. (9.76). The angular momentum of the Earth can
be expressed as J⊕ = I⊕ Ω⊕ , where I⊕ is the Earth’s moment of inertia for rotation about
the polar axis and Ω⊕ is the angular velocity of the Earth. In planetary science the moment
of inertia is parameterized as I = kMR2 , where M is the mass, R is the radius, and k
indicates how much the interior mass distribution differs from uniform (for example, a
completely uniform sphere has k = 52 = 0.4, but Saturn with a centrally-concentrated mass
distribution has k = 0.21). For the Earth (and other terrestrial planets) k ≃ 0.33, so Earth’s
moment of inertia is
I⊕ = 0.33M⊕R2⊕ = 8.06 × 1044 g cm2 ,
the Earth’s angular velocity is
2π
Ω⊕ = = 0.262 rad hr−1 = 7.27 × 10−5 rad s−1 ,
24 hr
−1
and thus the Earth’s angular momentum is J = I⊕ Ω⊕ = 5.86 × 1040 g cm2 s . Hence the
Lense–Thirring precession rate for a gyroscope in free fall on Earth’s rotation axis is
1 km 3

2GJ
ΩLT = 2 3 = 5.65 × 1010 arcsec yr−1 .
c z z
For Gravity Probe B illustrated in Box 9.3, the semimajor axis of the nearly circular polar
orbit was 7027.4 km. As the satellite passes over the North Pole it is in free fall with z =
7027.4 km. Inserting this in the above equation gives a precession rate of 0.16 arcsec yr−1 .
The Lense–Thirring precession rate depends on the latitude of a satellite in polar orbit, so
the smaller general-relativistic prediction of 0.039 arcsec yr−1 per year shown in Box 9.3
represents an average over the satellite in polar orbit, which is less than our evaluation on
the z axis. At any rate, this Lense–Thirring precession is a much smaller effect than the
geodetic precession of more than 6 arcseconds per year.
10 Neutron Stars and Pulsars
10.1 We work in c = G = 1 units (Appendix B.1). The Schwarzschild radius is rs =

2M, where M is the mass. Assuming gravity to pack neutrons down to a hardcore ra-
dius r0 ≃ 0.5 × 10−13 cm, the radius of the neutron star will be R ≃ r0 A1/3 , where A is
the number of neutrons in the entire star (related to the total mass by M ≃ Am, where
m = 939 MeV = 1.2 × 10−52 cm is the neutron mass). Equating the neutron star radius to
the Schwarzschild radius implies that r0 A1/3 = 2M, from which M may be eliminated us-
ing M = Am to give A = (r0 /2m)3/2 ≃ 3 × 1057 neutrons in a typical neutron star. Thus, by
these simple considerations one estimates that the radius is R = r0 A1/3 ≃ 7 km, the mass is
M = R/2 = 3.5 km = 2.4M⊙ , and the average density is
ρ̄ = M/ 34 π R3 ≃ 0.0024 km−2 = 3.2 × 1015 g cm−3 ,

which is larger than nuclear matter density (about 2.5 × 1014 g cm−3 ). More realistic esti-
mates for actual neutron stars give similar numbers: a radius of ∼ 10 km, a mass of about
∼ 1.5M⊙ , and an average density of ∼ 1015 g cm−3 .
10.2 From Eq. (10.15),
Z r
M(r) = 4π ε (r)r2 dr ≃ 34 π ε̄ r3 ,
0
where an average energy density ε̄ = M(R)/ 34 π R3 has been assumed. The gravitational
energy is then
M(r)(4π r2 ε̄ )
Z R Z R
M(r)
Eg = dM(r) = dr
0 r 0 r
3 M2
Z R
16 2 2
= π ε̄ r4 dr = ,
3 0 5 R
where dM(r)/dr = 4π r2ε̄was used to change integration variables and the preceding ex-
pressions for M(r) and ε̄ have been inserted. Assuming a neutron star with M = 2M⊙ =
2.95 km and R = 10 km (in geometrized units) gives
Eg = 0.52 km = 6.3 × 1053 erg = 4 × 1059 MeV = 7 × 1032 g = 0.35 M⊙ .
The ratio of the gravitational energy to the rest mass energy is Eg /M = (0.52 km/2.95 km) ≃
0.18, so the gravitational energy is a significant fraction of the rest mass energy and GR
corrections to Newtonian gravity are expected to be important for neutron stars.
10.3 Neglecting the pressure term in the numerator and 2m relative to r in the denomina-
tor (weak gravity assumption) in Eq. (10.13) gives d ϕ /dr ∼ m/r2 in the Newtonian limit,
where 2ϕ ≡ σ . But this means that ϕ is just the Newtonian gravitational potential in G = 1
59
units (see Section 8.1). Thus, σ is proportional to the gravitational potential in the New-
tonian limit. This interpretation is strengthened by substituting ϕ = 21 σ and neglecting P
relative to ρ in Eq. (10.8) to give
dP 1 dσ dϕ dϕ Gmρ
= − (P + ρ ) = −(P + ρ ) ≃ −ρ =− 2 ,
dr 2 dr dr dr r
where the last step follows from substituting d ϕ /dr ∼ Gm/r2 . This is the pressure-balance
equation in Newtonian hydrostatics for a gravitational potential ϕ = 21 σ .
10.4 Solving the differential form of Eq. (10.15) for dr, substitution in Eq. (10.14), and a
little algebra gives Eq. (10.16).
10.5 From Eqs. (10.2) and (10.11) the radial metric component for the Oppenheimer–
Volkov solution is given by
2M(r) −1

g11 (r) = 1 − ,
r
which is unity at the center where M(r) = 0. Thus the ratio of g11 at the surface to that at
the center is
2M(R) −1

g11 (R)
= 1− ≃ 2.5,
g11 (0) R
where we’ve assumed a neutron star of radius 10 km and mass M = 2 M⊙ ≃ 3 km. Since
the average spacing between neutrons is ∼ 10−13 cm, this means that the metric changes
by only of order one part in 1019 over the internucleonic spacing. Thus, on that distance
scale the metric is very flat. See Glendenning [98], Section 4.4 for further discussion.
10.6 From Problem 10.2 the total gravitational binding energy is about 4 × 1059 MeV and
from Problem 10.1 the total number of nucleons is about 3 × 1057. Thus, from the ratio of
these two numbers it may be estimated roughly that the gravitational binding energy per
nucleon in a neutron star is about 133 MeV nucleon−1 . This is much larger than the nuclear
binding energy per nucleon, which is about 8 MeV per nucleon for heavy nuclei and about
16 MeV per nucleon for extended symmetric nuclear matter.
10.7 Outside a spherical neutron star the metric should be well approximated by the
Schwarzschild form. In the Schwarzschild metric the escape velocity is vesc = (2M/R)1/2.
The exact relationship between the mass and radius of a neutron star depends on the equa-
tion of state but if M = 1 M⊙ = 1.48 km and R = 12 km one obtains vesc /c = 0.5, while if
M = 2 M⊙ and R = 10 km then vesc /c = 0.77. Thus, any realistic choice of M and R for a
neutron star will give an escape velocity that is a significant fraction of the speed of light.
This is a signal that the gravitational field is very strong and general relativistic effects are
significant. On the other hand, assuming for a white dwarf that M = 1 M⊙ and R = 5000 km
gives an escape velocity of vesc /c ≃ 0.024 and general relativistic effects for a white dwarf
are small (but not completely negligible).
10.8 From Eq. (6.5),

GM −31 M km M km
ε= = 7.416 × 10 = 1.475 ,
Rc2 kg R M⊙ R
where M is the mass producing the gravitational field and R is the characteristic distance
over which it acts. The innermost white dwarf has a mass of ∼ 0.2M⊙ , for which a radius
estimate is ∼ 15, 000 km. For the neutron stars let’s take a mass of 1.4M⊙ and radius of
∼ 10 km. Then
εWD ∼ 2 × 10−5 εNS ∼ 0.2 εEarth ∼ 7 × 10−10 εMoon ∼ 3 × 10−11,
confirming the assertion that any deviations from the strong equivalence principle would
be greatly amplified in PSR J0337+1715 relative to the Solar System. See also the related
Problem 25.2.
11 Spherical Black Holes
11.1 Begin with Eq. (11.9). For the r > 2M case, ln |r/2M − 1| = ln(r/2M − 1) and dif-
ferentiating (11.9) gives
2M −1

dt = dv − 1 − dr.
r
For the r < 2M case, ln |r/2M − 1| = ln(1 − r/2M) and differentiating (11.9) gives the
same result as above. Therefore,
2M −1 2M −2 2

dt 2 = dv2 − 2 1 − dvdr + 1 − dr .
r r
Inserting this expression for dt 2 in the Schwarzschild line element (9.4) expressed in stan-
dard coordinates gives the Schwarzschild line element in Eddington–Finkelstein coordi-
nates (11.10).
11.2 From Eq. (11.6), the proper time to fall from r = 2M to r = 0 is τ = 4M/3, in
geometrized units. Restoring the G and c factors, τ → cτ and M → GM/c2 , gives

4G −6 M
τ = 3 M = 6.6 × 10 seconds.
3c M⊙
Some times estimated from this formula to fall from the event horizon to the singularity of
a spherical black hole are shown in the following table.
Type of black hole Mass (M⊙ ) Time (s)
Typical stellar black hole 10 6.6 × 10−5

GW150914 final black hole 62 4.1 × 10−4
Milky Way central black hole 4.3 × 106 28.4
AGN central engine 109 6555
See also the related Problem 11.6.

11.3 For a Schwarzschild black hole of mass M the area of the horizon is A = 4π (2M)2 =
16π M 2 [see Problem 11.9 and Eq. (13.11)]. Split the mass between two black holes M →
λ M + (1 − λ )M, with λ varying from 0 to 1. Then the total horizon area for the two
resulting black holes is A′ = 16π λ 2 M 2 + (1 − λ )2M 2 and
A′ 16π λ 2 M 2 + (1 − λ )2M 2

= = 2λ 2 − 2λ + 1,
A 16π M 2
62
which is less than one except for λ = 0 or λ = 1. Thus, any split into two finite black holes
decreases the total horizon area and is forbidden by the area theorem.
11.4 No, because of the Hawking area theorem. The mass is dropped radially and is as-
sumed uncharged, so no angular momentum or charge is added to the black hole and the
resulting configuration must settle down eventually to a new Schwarzschild black hole. But
for the original black hole the horizon area is A = 16π M 2 and this can never decrease in
any (non-quantum) physical process. If gravitational waves were emitted they would carry
away energy, and if this exceeded the amount of energy dropped in with the new mass the
horizon area would have to decrease, violating the area theorem.
11.5 This problem is adapted from one in Ref. [64]. The Lagrangian is
2
µ ν ds
L = g µν ẋ ẋ = ,
dτ
where ẋµ ≡ dxµ /d τ . Using the Schwarzschild metric defined in Eq. (9.5) and using the
spherical symmetry to choose θ = π2 so that d θ = 0 gives
2M −1 2

2M 2 2
L = − 1− c t˙ + 1 − ṙ + r2 ϕ̇ 2 = −c2 ,
r r
where L = −c2 [since L = (ds/d τ )2 and ds2 = −c2 d τ 2 ] was used. A flat-space limit may
be obtained by setting 2M/r = 0, giving
2 2 2
dt dr dϕ
Lflat = −c2 + + r2 = −c2 .
dτ dτ dτ
Utilizing that
2 2
1 2 dr 2 dϕ
dt = γ d τ γ=p v = +r ,
1 − v2/c2 dt dt
where v is the 3-velocity, this may be expressed as −c2 γ 2 + γ 2 v2 = −c2 . Finally, multiply
both sides by m2 c2 and use that the relativistic energy is E = γ mc2 and the relativistic
momentum is p = γ mvv (see Section 5.2) to write this in the form E 2 = p2 c2 + m2 c4 .
11.6 From Eq. (11.6), the proper time in free fall from the horizon to the singularity is (see
the solution of Problem 11.2)

4G M
τ = 3 M = 6.6 × 10−6 seconds.
3c M⊙
Substituting τ = 1 day and solving for M gives M = 1.3 × 1010 M⊙ .
11.7 The expression for Ω follows directly from Eqs. (9.22) and (9.23):

2M dt dϕ
ε = 1− ℓ = r2 sin2 θ ,
r dτ dτ
π
which may be combined assuming θ = 2 to give

dϕ d ϕ /d τ 1 2M ℓ
Ω= = = 2 1− .
dt dt/d τ r r ε
Note that this expression gives an angular velocity with respect to the Schwarzschild coor-
dinate time, not with respect to the proper time (which is more complicated). For a circular
orbit the variable E in Eq. (9.28) must be equal to the potential Veff (r) evaluated at the
radial coordinate r = r0 that gives the (single) minimum of the potential. Equations (9.29)
and (9.30):
ε2 − 1
2
1 2M ℓ
E= Veff (r) = 1− +1 −1 ,
2 2 r r2
then lead to the requirement
s
ℓ2

2M
ε= 1− 1+ 2
r r
for a circular orbit.
11.8 The Schwarzschild radius is rS = 2.953(M/M⊙ ) km = 8.86 × 106 km. From Problem
9.14, the coordinate radius r and proper circumference C are related by
C 6.283 × 108
r= = = 108 km.
2π 2π
If one now travels radially inward from r = 108 km to r = 107 km, the proper distance
covered ∆s is found in Problem 9.14 to be
Z 108 −1/2
8.86 × 106
Z r2
rS −1/2
∆s = 1− dr = 1− dr = 1.0565 × 108 km,
r1 r 107 r
where the integral was done numerically (for example, with Wolfram Alpha online,1 Maple,
or Mathematica). If one then traverses a circle at this coordinate radius, the proper distance
covered is C = 2π r = 2π × (107 km) = 6.283 × 107 km.
11.9 From the discussion of invariant integration in Section 3.13.1, the area of the Schwarzschild
horizon is
Z 2π Z πp
A= dϕ detg d θ ,
0 0
where the metric g for the horizon surface is 2-dimensional, corresponding to the Schwarzschild
line element (9.4) evaluated at constant time and constant r = rs :
ds2 = rs2 d θ 2 + rs2 sin2 θ d ϕ 2 .
Thus the metric is specified by the diagonal 2 × 2 matrix g = diag (rs2 , rs2 sin2 θ ), which has
det g = rs4 sin2 θ . Substituting in the above expression for A then gives a horizon area
Z 2π Z π
A= rs2 dϕ sin θ d θ = 4π rs2 = 16π M 2,
0 0
where rs = 2M was used.
11.10 The Ricci curvature scalar R is the complete contraction of the Riemann curvature
tensor for the corresponding space [see Eqs. (8.14)–(8.17)]. Assuming the surface of the
1
www.wolframalpha.com/examples/mathematics/calculus-and-analysis/
Earth to have the metric of a 2-sphere, the 2-dimensional Riemann curvature tensor reduces
to the Gaussian curvature R = 2/r2 , where r is the radius of the Earth (see Problem 7.1).
The Ricci scalar is identically zero for the Schwarzschild metric, which is a solution of the
vacuum Einstein equation, Rµν = 0. The more appropriate scalar measure of curvature is
the Kretschmann scalar,
K ≡ Rαβ γδ Rαβ γδ ,
which is derived for general black hole solutions in Ref. [116]. The non-vanishing com-
ponents of the Riemann tensor for the Schwarzschild metric evaluated in an orthonormal
basis are (see Appendix B of [110])
−2M M 2M −M
R0101 = R0202 = R0303 = R2323 = R1212 = R1313 =
r3 r3 r3 r3
and evaluation of the Kretschmann curvature scalar for the Schwarzschild metric gives
K = 48M 2 /r6 [116]. Thus for the poles of the Earth and at r = 2M in the Schwarzschild
geometry the components of the Riemann tensor and the corresponding curvature are finite
and smoothly varying, suggesting that these singularities are coordinate and not physical
singularities. Indeed in both cases explicit transformations to new coordinates are known
that remove them. But at r = 0 in the Schwarzschild spacetime the Kretschmann curvature
scalar goes to infinity. Since R is a scalar, it cannot be transformed to a finite value by a
new choice of coordinates, suggesting that r = 0 is a physical, not coordinate, singularity
in the Schwarzschild metric.
11.11 In cartesian coordinates the line element is ds2 = dx2 + dy2 + dz2 and the equation
of the sphere is x2 + y2 + z2 = R2 . Differentiate the equation of the sphere, solve for dz, and
substitute into the metric to give
(xdx + ydy)2
ds2 = dx2 + dy2 +
R2 − (x2 + y2)
for the line element on the 2-dimensional spherical surface. Introduce new coordinates
(r, ϕ ) using x = r cos ϕ and y = r sin ϕ so that
dx = −r sin ϕ d ϕ + cos ϕ dr dy = r cos ϕ d ϕ + sin ϕ dr.
Substitution in the above equation for ds2 and some algebra then gives
R2 dr2
ds2 = + r2 d ϕ 2
R2 − r 2
for the line element of the 2-sphere. This diverges at r = R, but the Gaussian curvature
is constant over the whole sphere (Problem 7.1) and the sphere can be parameterized in
other coordinates that eliminate this singularity, so this clearly is a coordinate singularity
caused by a choice of variables with a restricted domain of validity. (Generally it may be
shown that no single coordinate system is valid over the entire 2-sphere; a minimum of two
overlapping coordinate patches are required to parameterize the entire surface.)
11.12 This result is indicated by the lightcone diagrams inside and outside the horizon
shown in Chapter 11, but can be formalized concisely by requiring that for a stationary
observer dr = d θ = d ϕ = 0. Then from the metric
rS 2
ds2 = g00 dt 2 = − 1 − dt .
r
But g00 is positive inside the horizon rS and negative outside it. Therefore ds2 is positive
(spacelike) inside the horizon and negative (timelike) outside the horizon. Since causality
demands that the trajectories of particles be timelike, stationary observers are possible only
outside the horizon.
11.13 This problem was adapted from one in Ref. [110]. The causal (light cone) structure
may be examined by considering the motion of radial light rays, for which d θ = d ϕ =
ds = 0. Inserting these constraints in the line element equation gives the local equation for
lightcones in this metric
−1
r2

dt
= 1− 2 .
dr R
Thus at r = 0 one has dt/dr = 1 and the forward lightcone opens at a 45 degree angle up-
ward, but as r → R this opening angle closes continuously to zero since dt/dr → ∞. Hence
for r < R a photon emitted from some r can reach any other value of r in a finite time, but
one emitted at r = R remains at R for all time. From the line element the metric compo-
nents g00 and g11 interchange signs at r = R. As was seen in the case of the Schwarzschild
metric expressed in Schwarzschild coordinates, this implies that the lightcones rotate by
90 degrees for r > R. Thus no null or timelike trajectory originating at a point r > R can
ever cross the boundary r = R. Conclusion: r = R defines an event horizon for observers
inside it: once one passes from r < R to r > R, it is impossible on causal grounds to return
to or send a signal to the region r < R.
11.14 In flat spacetime or a local inertial frame, one would expect the 4-acceleration to be
given by
duµ dxα ∂ uµ
aµ = = = uα ∂α uµ ,
dτ d τ ∂ xα
where u is the 4-velocity and τ is the proper time measured along the particle worldline.
In curved spacetime the derivatives must be replaced by covariant derivatives, so the ac-
celeration components are given by (see Problem 11.15) aµ = uα ∇α uµ , where ∇α is the
covariant derivative (3.55). But from Eq. (9.18), for a stationary observer only the timelike
component of the 4-velocity is non-vanishing:

u = (u0 , 0, 0, 0) = (1 − 2M/r)−1/2, 0, 0 , 0
so from Eq. (3.55)

µ
∂u µ
aµ = uα ∇α uµ = u0 ∇0 uµ = u0 + Γ0α uα
∂t
µ
∂u

µ µ
= u0 + Γ00u0 = Γ00 (u0 )2 ,
∂t
where the result in the last line used that the components of u are independent of time
and that only the timelike component of u is non-zero. The connection coefficients for the
Schwarzschild metric may be deduced from the solution of Problem 8.2 (summarized in
Appendix C). Since the metric is independent of time, the only non-vanishing connection
µ
coefficient of the form Γ00 is
1 σ −λ d σ

1 2M M
Γ00 = e = 1− ,
2 dr r r2
where for the Schwarzschild metric eσ = e−λ = (1 − 2M/r). Thus, the acceleration of the
stationary observer is directed radially with
M
a1 = Γ100 (u0 )2 = ,
r2
and the magnitude of the acceleration vector is
√ 2M −1/2 M
p
p
a · a = aµ gµν aν = a1 g11 a1 = 1 − ,
r r2
which diverges at the Schwarzschild radius r = 2M. Thus infinite acceleration is required
to remain stationary at the event horizon.
11.15 In a local inertial frame,
duµ dxα ∂ µ ∂
aµ = = u = uα α uµ = uα ∂α uµ ,
dτ d τ ∂ xα ∂x
where uα = dxα /d τ . To convert to general curvilinear coordinates, the partial derivatives
∂α must be replaced with covariant derivatives ∇α , giving
µ
∂u µ γ
aµ = uα ∇α uµ = uα + Γαγ u ,
∂ xα
11.16 The loss of light and energy from the Sun would be catastrophic, but there would
be essentially no change in the Earth’s orbit. To the degree that the Sun is spherical and its
spin can be neglected, the spacetime outside the Sun is well described by a Schwarzschild
metric, so the Earth would be orbiting in the same metric for the Sun or for the black hole.
There would presumably be some tiny differences because the Sun spins slowly and is not
exactly spherical, but those differences would be extremely small. The popular idea that
black holes are unique gravitational predators sucking up everything in sight is a miscon-
ception, since the gravitational field produced by a black hole of mass M well outside the
event horizon is just that expected for any mass M, whether it is a black hole or not.
12 Quantum Black Holes
12.1 Using the ideal gas law and assuming the Sun to be made of N ionized hydrogen
atoms, the entropy can be estimated roughly as S ≃ kB N ≃ 1057kB . For the Schwarzschild
solution the entropy is given by Eq. (12.14) with constants G and c restored using h̄ →
Gh̄/c3 and M → (G/c2 )M from Table B.1,
c3 kB A M 2

A 16π G 2 77
S= = kB 2 = kB M ≃ 1.07 × 10 kB ,
4Gh̄ 4ℓp 4h̄c M⊙
where A = 16M 2 is the area of the event horizon, the Planck length is ℓp = (h̄G/c3 )1/2 =
1.62 × 10−33 cm, and M is the mass. Thus, the entropy for the black hole would be about
1020 times larger than the present entropy of the Sun.
12.2 A reasonable estimate is to construct a density from the Planck mass divided by the
cube of the Planck length:
MP c5
ρp ≃ = ≃ 1094 g cm−3 ,
ℓ3P h̄G2
which may be compared with the paltry 1014 g cm−3 for a neutron star.
12.3 The average time for decay of the black hole and its temperature are given by
Eqs. (12.11) and (12.9), respectively:
3
M M⊙
tH = 8.4 × 10−26 s T = 6.2 × 10−8 K.
1g M
Solving the first of these for the mass with tH = 500, 000 yr gives M = 5.7 × 1012 g =
2.9 × 10−21M⊙ . Assuming a spherical event horizon, rs = 2M = 8.6 × 10−14 cm, where
1M⊙ = 1.475 km in geometrized units has been used. From the second equation for the
temperature above and the mass computed above, T ≃ 2.1 × 1013 K.
12.4 From Eq. (12.8) a black hole with lifetime comparable to the age of the Universe has
a mass of order 1014 g. One meter is very far outside its event horizon so using Newtonian
gravity to estimate the gravitational acceleration gives |g| = MG/r2 ≃ 6.7 m s−2 , which is
comparable to the gravitational acceleration of the Earth at its surface.
12.5 From Eq. (12.10) with M = 1 M⊙ , the black hole radiates a minuscule Hawking power
of 9 × 10−29 W, which is about 55 orders of magnitude smaller than the actual solar power
of 3.828 × 1026 W. From Eq. (12.9) the corresponding black hole temperature is a frigid
6.2 × 10−8 K.
12.6 From Eqs. (12.11) and (12.10) the mass is 2.3 × 105 kg and the power is 6.2 × 1021 W.
68
12.7 The CMB has a temperature of about 2.7 K. From Eq. (12.4) a Hawking black hole
of mass 4.6 × 1022 kg would have the same temperature and thus be in equilibrium with the
CMB (this mass is a little less than that of the Moon, or about 0.008 that of Earth). Black
holes with less mass than this would have higher temperature than the CMB and thus could
radiate more energy than they absorb; black holes with more mass than this would absorb
more energy from the CMB than they could radiate by Hawking radiation.
12.8 From the Stefan–Boltzmann law for a blackbody radiator the power is
P = Aσ T 4 ,
where A is the surface area, the Stefan–Boltzmann constant is
π 2 k4
σ= ,
60h̄3 c2
and T is the temperature. For a Schwarzschild black hole the area of the event horizon is
16π G2M 2
A = 16π M 2 =
c4
(see results of Problem 11.9 with factors of c and G reinstated using Table B.1). Inserting
these and the temperature given by Eq. (12.4) in P = Aσ T 4 gives Eq. (12.5). The power
radiated by the black hole comes at the expense of its mass so
dM
P = −c2 ,
dt
which leads to Eq. (12.6) when Eq. (12.5) is substituted for P.
12.9 This problem is adapted from a discussion in Perkins [181]. Using Newtonian gravity,
the magnitude of the tidal force acting over a distance ∆r can be estimated as
dF 2mMG 2mMG
dF = dr = dr → ∆F ≃ ∆r,
dr r3 r3
where M is the mass of the black hole. The energy required to create the particle–hole pair
from the vacuum is E ≃ mc2 . By the uncertainty principle the virtual pair can live for a
time ∆t ∼ h̄/E, and thus could separate a maximum distance
h̄c
∆r ∼ c∆t ∼
E
in the time ∆t. Requiring that the work done ∆F · ∆r be comparable to the rest mass energy,
2mMG
∆F · ∆r = (∆r)2 ∼ E,
r3
substituting m ∼ E/c2 and ∆r ∼ h̄c/E, and solving for E gives
s
2h̄2 GM
E∼ .
r3
Evaluating this at the Schwarzschild radius r = 2MG/c2 gives
h̄c3
E∼ ,
GM
where a factor of 12 has been dropped since our approximations are crude. But up to nu-
merical factors this is the result of Eq. (12.4) for the average energy kB T associated with
the Hawking radiation.
13 Rotating Black Holes
13.1 This is straightforward but entails a considerable amount of algebra. For example,
collecting the terms proportional to dt 2 in the alternative form of the metric,
−∆ 2 sin2 θ 2 2 1
a dt = 2 a2 sin2 θ − ∆ dt 2

2
dt + 2
ρ ρ ρ
1
= 2 a2 sin2 θ − r2 + 2Mr − a2 dt 2

ρ
1
= 2 a2 (sin2 θ − 1) − r2 + 2Mr dt 2

ρ
1
−r2 − a2 cos2 θ + 2Mr dt 2

= 2 2 2
r + a cos θ

2Mr 2Mr
= − 1+ 2 dt = − 1 + 2 dt 2 ,
2
r + a2 cos2 θ ρ
which is the first term of Eq. (13.1).
13.2 From Eq. (13.10) the horizon area for a Kerr black hole may be written as
p
2
AK = 4π (rH + a2 ) rH ≡ M + M 2 − a2.
Because of entropy conservation (Hawking area theorem), the horizon area cannot de-
crease as angular momentum is extracted, and removing all the angular momentum leaves
a Schwarzschild black hole. The maximum energy extraction will occur if the horizon re-
mains constant in size, since if it grows the Schwarzschild black hole left behind will be
more massive than if the horizon stays constant. Taking the a = J/M = 0 limit of the above
expression,
2
AK = 4π rH = 4π (2M0 )2 = 16π M02 ,
where rH = 2M for the Schwarzschild black hole has been used, and where M0 represents
the mass left after all angular momentum has been extracted (sometimes termed the irre-
ducible mass). Equating the above two expressions for AK ,
p 2
16π M02 = 4π M + M 2 − a2 + a2 ,
and solving for M0 gives

1
q 2
M0 = M + (M 2 − a2)1/2 + a2 .
2
Thus, the maximum total energy that can be extracted from the Kerr black hole by remov-
71
ing its angular momentum is

1
q 2
2
(M − M0 )c = M − 2 2
M + (M − a )1/2 + a c2 .
2
2
For an extremal Kerr black hole, a = M, which yields

1
Emax = (M − M0 )c2 = 1 − √ Mc2 = 0.29 Mc2 ,
2
for the maximum energy that could be extracted from the Kerr black hole by reducing it to
a Schwarzschild black hole. In principle, this could be accomplished by a Penrose process,
but they are difficult to arrange in practical astrophysics environments. A more plausible
realization is the Blandford–Znajek mechanism [50], where compression of magnetic field
lines near the horizon can cause the rotational energy to be emitted as a stream of e+ e−
pairs.
13.3 The steps are analogous to those for the equations of motion in the Schwarzschild
metric derived in Section 9.3.
13.4 The distances are gotten by integrating the line element (13.8) around the correspond-
ing curves. For the equator, θ = π2 so d θ = 0 and only the second term contributes. The
distance around the equator is
I √ Z 2π
L= ds2 = 2M d ϕ = 4π M,
0
where sin θ = 1 has been used. For a meridian through the poles take ϕ = 0 and d ϕ = 0,
so only the first term contributes. The corresponding distance is
Z πq
√
I
L′ = ρ+ d θ = 2 2 + a2 cos2 θ d θ .
r+
0
For the special case of an extremal black hole a = M and r+ = M, so that ρ+ = M(1 +
cos2 θ )1/2 and
Z πp
L′ = 2M 1 + cos2 θ d θ ≃ 7.6M,
0
where the definite integral has been evaluated numerically using Maple. Thus, the ratio
of the equatorial to polar circumferences for the horizon of an extremal Kerr black hole
is L/L′ = 4π M/7.6M = 1.65. Although the horizon corresponds to a constant Boyer–
Lindquist coordinate r+ , it does not have a spherical geometry.
13.5 Substitution of a = J/M into Eq. (13.1), discarding terms higher than linear in J, and
a little algebra gives
4GJ
ds2 ≃ (ds2 )0 − 3 2 sin2 θ (rd ϕ )(cdt),
c r
where (ds2 )0 is the Schwarzschild line element and factors of G and c have been restored
using the conversion J → GJ/c3 from Table B.1. Utilizing the classical result that the angu-
lar momentum of a rotating body is given by J ∼ Mrv, the dimensionless ratio (GJ/c3 r2 )
governing the strength of the rotational correction may be expressed as

4GJ v GM
≃ .
c3 r 2 c rc2
The second factor GM/rc2 is the measure of spacetime curvature produced by a non-
rotating spherical mass given in Eq. (6.5). Thus the total spacetime curvature (which re-
ceives contributions from this correction plus that coming from the leading Schwarzschild
term) depends both on the spherical mass and the rotational velocity. From Eq. (6.12) the
gravitational redshift caused by the spherical mass is of order 1/c2 , but the above effect is
of order 1/c3 .
13.6 From the Kerr line element with θ = π2 and dt = d ϕ = d θ = 0
s
ds r2
= .
dr r − 2Mr + a2
2
If a → 0 the Schwarzschild result (9.7) in geometrized units is recovered. If instead one

sets a = M, then
−1/2
2M M 2

ds
= 1− + 2
dr r r
for an extremal Kerr black hole.
13.7 Let us calculate d ϕ /dr for a particle dropped from rest (ε = 1) with zero angular
momentum (ℓ = 0) into a Kerr black hole. Since ε and ℓ are conserved they retain their
initial values and from Eqs. (13.15) and (13.16) for an infalling trajectory,
s
a2

dϕ 1 2Ma dr 2M
= =± 1+ 2 ,
dτ ∆ r dτ r r
where the negative sign should be chosen on the square root because the particle is in-
falling. These may be combined to give
−1/2
a2

dϕ d ϕ /d τ 2Ma 2M
= =− 1+ 2 .
dr dr/d τ r∆ r r
Since in the general case d ϕ /dr 6= 0, the particle is dragged in ϕ as it falls radially inward,
even though no forces act on it.
13.8 Assume r >> M and r >> a, and drop terms quadratic in a. Then
ρ 2 = r2 + a2 cos θ ≃ r2 ∆ = r2 − 2Mr + a2 ≃ r2 − 2Mr.
Substituting in the Kerr metric (13.1) gives

2M 2M 4Ma 2
ds2 = − 1 − dt 2 + 1 + dr2 + r2 d θ 2 + r2 sin2 θ d ϕ 2 − sin θ d ϕ dt,
r r r
where (1 − 2M/r)−1 ∼ (1 + 2M/r) was used. Taking the limit r → ∞ at constant M and a
gives
ds2 = −dt 2 + dr2 + r2 (d θ 2 + sin2 θ d ϕ 2 ),
which is the flat-space Minkowski metric in spherical coordinates. Hence the Kerr space-
time is asymptotically flat.
Observational Evidence for Black
14
Holes
14.1 From Kepler’s third law,

1/3 " 2 #1/3
GMP2

−6 M P
a= = 7.495 × 10 AU,
4π 2 M⊙ 1 day
where a is the semimajor axis, M is the total mass, and P is the period. Taking
M = 40M⊙ P = 5.6 days
gives a ≃ 0.2 AU for the average separation. In one ms light can travel a distance
d ∼ 2 × 10−6 AU ∼ 300 km.
Thus, X-ray fluctuations on a millisecond scale set an upper limit of about 300 km for the
diameter of the X-ray emitting region.
14.2 Insert G = 6.6726 × 10−8 cm3 g−1 s−2 and convert units to give
3
PK 3

−7 P K
f (M) = = 1.036 × 10 M⊙ .
2π G 1 day km s−1
Eyeballing from Fig. 14.2 K ≃ 450 km s−1 and using the period of 0.32 days from Ta-
ble 14.1 gives f ≃ 3.02 M⊙ . The discrepancy with the value of 3.18 M⊙ quoted in Ta-
ble 14.1 is because in the literature a value K ≃ 457 km s−1 is commonly used; see for
example Refs. [150, 111]. These references may also be consulted for methods used to try
to infer the actual mass of the black hole from this lower limit.
14.3 From standard conversion factors, 1 light-minute = 0.12 AU, 1 light-hour = 7.22 AU,
and 1 light-day = 173.3 AU. Thus, from the quantities displayed in Fig. 14.5 the distance
of closest approach is 17 light-hours = 122.7 AU (about 6 times larger than the orbit of
Neptune). The event-horizon radius for a spherical black hole of mass 4.3 × 106 M⊙ is

M
R = 2.95 km ≃ 12.7 × 106 km = 0.085 AU.
M⊙
For a black hole with mass M and star with mass M∗ and radius R∗ , the tidal distortion
radius for the star’s orbit may be estimated as
1/3
M
rt ≃ R∗ .
M∗
S0-2 is a M∗ = 15 M⊙ main sequence star, which from stellar systematics implies that R∗ ∼
6.5 R⊙ ∼ 4.55 × 106 km. This gives rt ∼ 3 × 108 km ∼ 16.6 light-minutes ∼ 2 AU. Thus S0-
2 at closest approach is well outside the event horizon and the tidal distortion radius of the
75
76 Observational Evidence for Black Holes
black hole. The average velocity of Earth on its orbit can be estimated by assuming the
orbit to the circular. Then the distance around the orbit is 2π r, where r = 1 AU. Dividing
this by a period of 1 year gives an average velocity of 29.7 km s−1 , which is 168 times
smaller than the 5000 km s−1 orbital velocity of S0-2 at closest approach to the black hole.
The radius of the Sgr A∗ 4.3 × 106 M⊙ black hole is 0.085 AU, the radius of the Sun is
4.65 × 10−3 AU, and the semimajor axis of Mercury’s orbit is 0.39 AU. Thus the black
hole is 18.3 times larger than the Sun and 0.22 times the size of Mercury’s orbit.
14.4 Evaluating the constants in Eq. (14.8) gives
2
R σ
M = 3.77 × 10−11 M⊙
km km/s
(a) For M31 with R = 0.8 pc = 2.47 × 1013 km and σ = 240 km s−1 , the virial mass is
M = 3.77 × 10−11(2.47 × 1013)(240)2 M⊙ = 5.36 × 107 M⊙ .
(b) For a typical Seyfert galaxy,
M = 3.77 × 10−11(3.09 × 1015)(1000)2 M⊙ = 1.16 × 1011 M⊙ ,
where R = 100 pc = 3.09 × 1015 km and σ = 1000 km s−1 were assumed. (This suggests
that 10% of the mass of the Seyfert galaxy is contained within a central region only 100 pc
across.)
14.5 Let m1 = M and m2 = Mc , and define the mass ratio q ≡ m2 /m1 = Mc /M. Then
Kepler’s 3rd law is
4π 2 (a1 + a2)
m1 (1 + q) = m1 + m2 =
G P2
Utilizing m1 a1 = m2 a2 gives (a1 + a2)3 = a32 (1 + q)3 and thus
4π 2 a32 (1 + q)3
m1 (1 + q) = .
G P2
For a spectroscopic binary the easily observable orbital quantities are the period P and the
semiamplitude of the radial velocity K ≡ v2 sin i for the visible star 2 (see Fig. 14.2), with
2π 2π a2 2π a2 sin i 2π a2 sin i
P= = = = .
ω v2 v2 sin i K
Solving this expression for a2 and inserting in the Kepler’s law equation gives
K 3 P (1 + q)2
m1 = .
2π G sin3 i
Rearranging and substituting q = m2 /m1 gives
PK 3 m1 sin3 i m31 sin3 i
= = ,
2π G (1 + q)2 (m1 + m2 )2
which is Eq. (14.2) with M = m1 and Mc = m2 .
14.6 From Fig. 2 of Ref. [171], P = 5.6 days and K ∼ 75 km s−1 , which gives from Eq.
77 Observational Evidence for Black Holes
(14.3) F = PK 3 /2π G = 0.245. Inserting this in Eq. (14.6) and plotting M versus i and M
versus Mc gives the following diagrams.
30 30
(a) F = 0.245 M (b)
25 25
20 M c( M ) 20
20o
M (M )
15 40 15
30
30o
10 20 10
40oo
10 50
o
5 5 o 70o 60
F = 0.245 M 90o 80
0
0 0
10 15 20 25 30 35 40 45 50 0 10 20 30 40
i (deg) Mc (M )
The gray areas indicate the ranges i = 25–35◦ and Mc = 20–30 M⊙, which constrain the
mass M of the unseen object to lie in the range ∼ 10–20 M⊙. A more diligent use of
observational constraints described in Ref. [171] and Box 14.2 gives a value of M =
14.8 ± 1.0 M⊙. Since these mass estimates are far above the maximum mass expected for
a neutron star, the unseen massive and compact object in Cygnus X-1 is almost certainly a
black hole.
14.7 From Eq. (6.5), the intrinsic strength of gravity is governed by

GM M km M km
ε = 2 = 7.416 × 10−31 = 1.475 ,
Rc kg R M⊙ R
over which it acts. For the Sun, taking the mass of the Sun as M and its radius as R gives
ε⊙ ∼ 2.1 × 10−6. For the Binary Pulsar, taking the separation at closest approach (∼ 1.1R⊙)
as R and the mass of about 1.4M⊙ for the other neutron star as M gives εBP ∼ 2.7 × 10−6.
For the star S0-2, taking the distance at closest approach of 17 light–hours as R and the
mass 4.3 × 106 M⊙ of the black hole as M gives εS0-2 = 3.45 × 10−4 (which is comparable
to the strength of gravity at the surface of a white dwarf). The star S0-102 comes even
closer to the black hole so ε for it is larger but the same order of magnitude. Thus ε is
about two orders of magnitude larger for stars in orbit around the black hole at Sgr A∗ than
for gravity at the surface of the Sun or in the Binary Pulsar, and the orbits of stars like S0-2
and S0-102 can provide a test of general relativity in stronger gravity than for either Solar
System measurements or binary pulsars.
15 Black holes as Central Engines
15.1 (a) From the masses before and after in nuclear reactions burning hydrogen to helium
one finds a typical efficiency for mass to energy conversion of η ∼ 0.007. Requiring that
η ṁc2 = 1047 erg s−1 = 3.2 × 1054 erg yr−1
yields that ṁ ≃ 255M⊙ yr−1 if this luminosity is supplied by hydrogen fusion. It is difficult
to conjecture a mechanism consistent with observations that could account for this.
(b) On the other hand, for black hole accretion the mass to energy conversion efficiency
could lie in the range η ≃ 0.1 − 0.4. Then the luminosity could be sustained by accretion of
5 M⊙ − 20 M⊙ yr−1 , for which there are plausible mechanisms. The Eddington luminosity
(maximum luminosity for which radiation pressure would not reverse accretion infall) is
given by

M
LEdd = 1.3 × 10 38
erg s−1 .
M⊙
Equating this to the observed ∼ 1047 erg s−1 and solving for the mass of the central object
gives M ≃ 7.7 × 108 M⊙ .
(c) Observed light variation on timescales of days argues by causality that the source has
a maximum diameter of 1 light-day or about 170 AU (twice the diameter of the Solar
System). Thus, it may be inferred on rather general grounds that the AGN central engine
has of order 109 M⊙ concentrated in a region not much larger than the Solar System. The
most plausible explanation is a supermassive black hole. The Schwarzschild radius of a
109 M⊙ black hole would be

M
R ≃ 2.95 km ≃ 3 × 109 km ≃ 20 AU,
M⊙
which is approximately the radius of the orbit of Uranus.
15.2 From Eq. (15.1), in Newtonian approximation the energy released through accretion
is
GMm 21 M m km
∆Eacc = = 1.327 × 10 erg,
R M⊙ g R
where R is the radius and M the mass of the object onto which accretion is taking place,
and m is the mass of the accreting material. The entries in Table 15.1 result from using in
this formula 1 gram for m and the representative parameters
1. M = 1M⊙ and R = 10 km for a neutron star,
2. M = 1M⊙ and R = 104 km for a white dwarf,
3. solar parameters for a normal star, and
78
4. M = 10 M⊙ and R = 3rs for a black hole, where the Schwarzschild (event-horizon)

radius is given by

M
rs = 2.95 km
M⊙
and 3rs is the radius of the innermost stable circular orbit in a Schwarzschild spacetime;
see Eq. (9.37).
The last column of the table is normalized to the thermonuclear burning of hydrogen to
helium (“fusion”), which releases ∼ 6 × 1018 erg for each gram of hydrogen burned. These
numbers should be viewed as approximate for neutron stars and black holes since they have
been estimated using Newtonian gravity. However, they show clearly that the accretion
energy can be very large for highly compact objects.
15.3 From the geometry of the figure, the leading edge of the jet appears to move about
14 lightyears in two elapsed observing years. Thus the apparent transverse velocity is v ∼
14/2 ∼ 7c.
15.4 From the inverse Compton boost factor given in Box 15.4, a visible photon of fre-
quency 5 × 1014 Hz is boosted to a frequency
ν0 5 × 1014 Hz
ν = γ 2 ν0 = 2 2
= = 5 × 1020 Hz,
1 − v /c 1 − (0.9999995)2
which lies in the γ -ray region of the spectrum.
15.5 Neglecting the spectral energy distribution of the emitted flux,
Sobserved 1
= .
Semitted [γ (1 − β cos θ )]3
Thus, if one assumes the same γ and β for both jets and that the approaching jet makes an
angle θ with the line of sight, the ratio of observed flux densities for light emitted from the
approaching and receding jets will be
[γ (1 − β cos θ )]−3 1 + β cos θ 3

Sapproach
= = .
Srecede [γ (1 − β cos(θ + π ))]−3 1 − β cos θ
If, for example, β = 0.98 and θ = 10◦ , this ratio is ∼ 105 , implying that the counterjet
will appear to be much fainter than the jet. If one takes into account the spectral energy
distribution of the emitted flux, the above formula remains valid except that the exponent
of 3 is replaced by a somewhat smaller exponent ∼ 2.7 (see Section 8.3.3 of Ref. [204]).
15.6 By the Hubble law (Ch. 16), the distance is
v cz
d= ∼ ,
H0 H0
since z ∼ v/c for redshifts that are not too large. Using H0 = 72 km s−1 Mpc−1 ,
cz (3 × 105 km s−1 )(0.158)
d= = ≃ 660 Mpc.
H0 72 km s−1 Mpc−1
The apparent magnitude m and the absolute magnitude M for a distant object are related
by the distance modulus formula
d (pc)
M = m − 5 log ,
10
where d (pc) is the distance to the object in parsecs. Using the apparent visual magnitude
m = +12.9 and d = 660 × 106 pc gives an absolute visual magnitude of M = −26.2 for 3C
273. Absolute magnitudes M and luminosities L for objects 1 and 2 are related by
L1
= 100.4(M2 −M1 ) .
L2
For the Sun the absolute visual magnitude is +4.8 and the above formula indicates that
3C 273 is 2.5 × 1012 times more luminous than the Sun at visual wavelengths. For M31,
with absolute visual magnitude of about −21.5, the corresponding ratio is about 76, and
for M87, with absolute visual magnitude of about −22, one obtains that 3C 273 is 48 times
more luminous at visual wavelengths. Thus, 3C 273 is roughly 100 times more luminous
at visual wavelengths than large galaxies. However, the quasar emits most of its light at
nonvisual wavelengths. When the luminosities are integrated over all wavelengths 3C 273
is found to be about 1000 times more luminous than large normal galaxies.
15.7 For the thin disk radiating as a blackbody the radiation rate per unit area is σ T 4 ,
where σ is the Stephan–Boltzmann constant. Thus, if the disk has a radius R the luminosity
L is
L = 2 π R2 σ T 4 ,
where the factor of two comes from the disk having two sides. Assume the observed lu-
minosity of the disk to be a fraction η of the Eddington luminosity (15.3). Then solving
the preceding equation for T and using the expression (15.4) to approximate the Eddington
luminosity gives for the temperature of the disk
1/4
η Ledd 1/4 M/M⊙ 1/4

L 7
T= = = 7.72 × 10 η K,
2 π R2 σ 2 π R2 σ (R/ km)2
where M is the gravitational mass responsible for the accretion (which may be approxi-
mated by the mass of the compact object, since this is much larger than the mass in the
accretion disk). For a neutron star, assuming
R ∼ 10 km M ∼ 1M⊙ η ∼ 1,
this formula yields T ≃ 2.4 × 107 K. By the Wien law, the corresponding blackbody spec-
trum peaks at a wavelength
2.9 × 10−3 m K
λpeak = ≃ 0.12 nm,
T
which is in the X-ray portion of the spectrum. For a Schwarzschild (spherical) black hole,
approximate R by the radius of the innermost stable circular orbit, which from Eq. (9.34)
is given by
6GM
R= 2 ,
c
with factors of G and c restored. In convenient units, G/c2 = 1.475 km M⊙ −1 and the
preceding equation for T may be written
M⊙ 1/4

T = 2.6 × 107η 1/4 K.
M
Assuming radiation near the Eddington limit so that η ∼ 1, for a 10 M⊙ black hole this
formula and the Wien law give
T = 1.5 × 107 K λpeak = 0.20 nm,
which is dominantly in the X-ray region of the spectrum. For a 108M⊙ black hole, we find
likewise that
T = 2.6 × 105 K λpeak = 11 nm,
which is dominantly in the UV portion of the spectrum. This has been a rather crude ap-
proximation to the physics of accretion disks (for more realistic descriptions, see Refs.
[89, 183]), but it indicates correctly that accretion disks around neutron stars or stellar-
size black holes are expected to radiate in the X-ray region, but the corresponding accre-
tion disks around supermassive black holes should have lower temperatures and radiate at
longer wavelengths, largely in the UV portion of the spectrum.
15.8 Evaluation of constants allows Eq. (15.15) to be written as
f σT FD2 D 2

−40 F s 2
τ≃ 2 4
= 9.01 × 10 f −2
δ t me c cm erg cm δt
2
D F ms 2
= 8.58 × 1015 f −2
Mpc erg cm δt
2
10 ms 2

13 D F
= 7.7 × 10 f .
3000 Mpc 10−7 erg cm−2 δt
Since f and the product of quantities in parentheses are of order one in the last expression
for a typical gamma-ray burst, the resulting optical depth is huge (τ ∼ 1014 ). This is incon-
sistent with the observed nonthermal spectrum for gamma-ray bursts, since a nonthermal
spectrum typically requires a medium that is optically thin. The fallacy is that Eq. (15.15)
is invalid for a gamma-ray burst because it must be modified to account for the ultrarela-
tivistic kinematics of the burst. When that is done, as in Eq. (15.16), the above expression
is multiplied by a factor approximately equal to 1/γ 4+2α , as discussed in Section 15.7.4.
For a typical value α ∼ 2 this will yield optical depths smaller than one for γ of order 100
or larger.
16 The Hubble Expansion
16.1 This problem was adapted from a discussion in Peebles [178]. Assume a single lumi-
nosity bin i, with the number of galaxies in the bin given by ni ≡ n(Li ). Out to a distance
r, the volume of space in 4π steradians is
Z r
V (4π sr) = 4π r2 dr = 43 π r3 ,
0
so the volume in 1 sr is
V (4π sr) 1 3
V (1 sr) = = 3r .
4π
Thus the mean number of galaxies per steradian brighter than f = L/4π r2 (closer than r)
is
Li 3/2

1 3 ni
niV = 3 ni r = ,
3 4π f
where f = L/4π r2 → r3 = (L/4π f )3/2 has been used. If the galaxies are distributed uni-
formly the preceding applies to each luminosity bin and the total is gotten by summing
over all bins
Li 3/2

1
N( f> ) = ∑ niV = 3 ∑ ni ,
i i 4π f
which is proportional to f −3/2 . To convert to magnitudes m, use m = −2.5 log f + constant

to replace f with m, giving N( f> ) ∝ 100.6m . More realistically space is not euclidean and
not static, so the above discussion must be modified to reflect redshifts in the expanding
Universe and the effect of galaxy evolution over time (since distant galaxies are observed
at a younger age than more nearby galaxies). Nevertheless, the realization by Hubble that
the galaxy observations available to him (out to about 800 Mpc, corresponding to a redshift
z ∼ 0.2 and to light that was emitted more than 2 billion years ago) were approximately
consistent with the f −3/2 power law was the first strong evidence that galaxies are dis-
tributed homogeneously on large scales.
16.2 From the solution of Problem 4.7, the relativistic Doppler frequency shift is given by
s
ν0 1 + v/c 1 + v/c
=p = .
ν 2
1 − v /c 2 1 − v/c
Using ν = c/λ and z ≡ (λ − λ0)/λ0 gives

λ ν0 1 + v/c
1+z = = =p .
λ0 ν 1 − v2/c2
82
However, the cosmological redshift should not be interpreted as being due to a Doppler
shift (though it commonly is, especially in popular-level discussions). It is actually caused
by the expansion of space, as shown in Section 16.2.3.
16.3 Integration of the Hubble law dr/dt = H0 r, assuming that H0 is constant with time,
gives r(t) ∼ eH0t . The volume of a spherical region is
V (t) = 43 π r3 ≃ e3H0 t .
Since the volume expands but the density is assumed constant, matter must be created
continuously to maintain the constant density. The total mass within a volume is M = ρ V ,
where ρ is the constant density. Then Ṁ = ρ V̇ = ρ × 3H0V and the creation rate per unit
volume is
Ṁ
= 3H0 ρ ≃ 7 × 10−48 g s−1 cm−3 ,
V
where a matter density of ρ ∼ 10−30 g cm−3 and a Hubble parameter H0 ≃ 72 km s−1 Mpc−1
were assumed. This is equivalent to the creation of about one hydrogen atom per cubic me-
ter every 10 billion years.
16.4 (a) The most important point is that light from stars beyond a certain distance has not
had time to reach us if the Universe is of finite age. A secondary point is that this light is
redshifted to lower energies in an expanding universe. Thus big bang cosmology resolves
Olber’s paradox. Note that the sky actually is uniformly bright, not at visible wavelengths,
but rather at millimeter wavelengths because of the 2.725 K cosmic microwave background
radiation discussed in Section 20.4.
(b) The first alternative explanation is based on incorrect physical reasoning because inter-
vening dust would absorb energy, equilibrate, and re-emit the energy that was originally
absorbed. The second is inconsistent with the cosmological principle because it postulates
a distribution of stars on large scales that is not uniform.
16.5 From Eq. (6.5), a general relativistic description is required if GM/Rc2 ≃ 1. As a
crude estimate take the Universe to be static and euclidean, with a radius given by the
Hubble distance and a density comparable to the critical density. Then, treating the Uni-
verse as a spherical gravitating mass,
c 3H02
R∼ M ∼ 43 π R3 ρcrit ρcrit ≃ ,
H0 8π G
which implies that
GM 4 R2 ρcrit 1
2
= π G 2
≃ .
Rc 3 c 2
Therefore, a correct cosmological description is expected to involve general relativity.
16.6 Taking a Hubble constant of 72 km s−1 Mpc−1 , corresponding to h = 0.72, the reces-
sional velocity by the Hubble law would be
km
v = H0 d ≃ 100h × 16 Mpc = 1600h km s−1 = 1152 km s−1 .
s Mpc
But the observed recessional velocity is only 985 km s−1 , so the peculiar velocity is 167
km s−1 , toward us.
16.7 The erroneous Hubble constant may be expressed as
km 1 Mpc
H0 = 550 × = 1.78 × 10−17 s−1 .
s Mpc 3.086 × 1019 km
Therefore, the Hubble time is 1/H0 = 5.61 × 1016 s = 1.78 × 109 yr, which is about an
order of magnitude smaller than the currently accepted age of the Universe.
16.8 The Lyman alpha spectral line normally at λ0 = 121.6 nm, is observed to be shifted
to λ = 968.2 nm. Thus,
λ 968.2
z= −1 = − 1 = 6.96.
λ0 121.6
The scale factor corresponding to this redshift is a = (1 + z)−1 ≃ 0.125. The evolution of
the scale factor with time according to the best current cosmological parameters is given
by Fig. 19.10. From that plot, a redshift of z ∼ 7 (or scale factor relative to today of a(t) =
0.125), corresponds to a time about 0.75 billion years after the big bang (defined by time
when a = 0). A similar result may be obtained by using the cosmological calculator at Ref.
[5]. Thus, in the standard cosmology z ∼ 7 corresponds to light that was emitted about 750
million years after the big bang.
16.9 Take vp = 200 km s−1 as a representative peculiar velocity. From the Hubble law with
−1
H0 = 72 km s−1 Mpc , a distance of d = vp /H0 = 2.8 Mpc corresponds to the distance
where the cosmological recessional velocity is comparable to the average peculiar velocity
in magnitude. Thus, a distance of 10 times that or about 30 Mpc is necessary before the
peculiar velocities are only ∼ 10% of the cosmological recessional velocities.
16.10 Supernova SN 2015A was a Type Ia supernova in the Sb spiral galaxy NGC 2995
with a redshift z = 0.023329 (corresponding to cz = 6998.7 km s−1 ). The Hubble law is
approximately valid and for small distances v ∼ cz, so the distance to the supernova is
v cz
d∼ ∼ = 97.2 Mpc,
H0 H0
−1
where H0 = 72 km s−1 Mpc has been assumed.
17 Energy and Matter in the Universe
17.1 The energy density in the CMB is given by the blackbody formula with T = 2.725 K,
εCMB = aT 4 = 4.17 × 10−13 erg cm−3 = 2.6 × 10−7 MeV cm−3 .
The corresponding density parameter is
εCMB 2.6 × 10−7 MeV cm−3
ΩCMB = = = 4.8 × 10−5,
εcrit 1.05 × 10−2h2 MeV cm−3
assuming h = 0.72. For starlight we may estimate roughly,
εstar ≃ (2.6 × 10−32 erg s−1 cm−3 ) × (14 × 109 yr) × (3.16 × 107 s yr−1 )
≃ 1.15 × 10−14 erg cm−3 ,
if the Universe is 14 billion years old and therefore
εstar 1.15 × 10−14 erg cm−3
≃ = 0.028.
εCMB 4.17 × 10−13 erg cm−3
Therefore, the radiation density parameter for the present Universe,
Ωr ≃ ΩCMB ≃ O(10−5 ),
is dominated by the cosmic microwave background and is negligible compared with the
matter density (Ωm ≃ 0.3) and the vacuum energy density (ΩΛ ≃ 0.7).
17.2 From Eqs. (17.15)–(17.16),
a0
ȧ2 = a20 H02 1 + Ω0 − Ω0 .
a
Choose Ω0 = 1 (flat Universe), take the square root of both sides, and then integrate both
sides to give
Z a Z t
3/2
a1/2 da = a0 H0 dt.
0 0
Performing the integrals and solving for a/a0 gives

3/2 2/3
a(t) 2 t
= ,
a0 3 tH
where tH = H0−1 has been used. The redshift z is given by a0 /a(t) = 1 + z. Choosing by
convention a0 = 1 gives
t(z) 2
= 3 (1 + z)−3/2.
tH
Since z = 0 today, the age of a flat, dust-filled Universe is t0 ≡ t(z = 0) = 23 tH .
85
17.3 From Eq. (17.18), for a closed universe

a Ω
= (1 − cos ψ ),
a0 2(Ω − 1)
and since 1 + z = a0 /a,
2(Ω − 1)
cos ψ = 1 − .
Ω(1 + z)
Also, from Eq. (17.19),
2(Ω − 1)3/2

t
ψ − sin ψ = ,
Ω tH
where tH = 1/H0 was used. Combining these relations and a substantial amount of algebra
then gives
" #
2(Ω2 z + Ω − Ωz − 1)1/2

t(z) Ω −1 Ωz − Ω + 2
= cos − ,
tH 2(Ω − 1)3/2 Ω + Ωz Ω + Ωz
which reduces to Eq. (17.25) upon setting z = 0. For an open universe, start from Eqs.
(17.22) and (17.23) and proceed in a way analogous to above. After substantial algebra,
one obtains
" #
2(−Ω2 z − Ω + Ωz + 1)1/2

t(z) Ω Ωz − Ω + 2
= − cosh−1 ,
tH 2(1 − Ω)3/2 Ω + Ωz Ω + Ωz
which reduces to Eq. (17.24) upon setting z = 0.

17.4 Calculate as for Example 17.2 using the formulas derived in Problems 17.2 and 17.3
(which are quoted in Example 17.1).
17.5 Balance of centripetal and gravitational forces for a circular orbit requires that v2 (r) =
GM(r)/r. Thus, if the velocity is constant with increasing r the mass must be increasing
as M(r) ∝ r.
17.6 From the preceding problem, v(r)2 = GM(r)/r. Thus,
2
rv(r)2

r v
M(r) = = 2.32 × 105 M⊙ .
G kpc km s−1
At the Sun’s distance of about 8 kpc from the center the velocity v ∼ 220 km s−1 gives
M(8 kpc) ≃ 9 × 1010 M⊙ . At 60,000 light years (18.4 kpc) from the center the measured
velocity v ∼ 230 km s−1 gives M(18.4 kpc) ≃ 2.3 × 1011M⊙ .
17.7 One has that
Z +∞
1 2 2
hv2i i = √ e−vi /2σi v2i dvi
−∞ 2π σi
√
Letting x ≡ vi / 2σi gives
2σ 2
Z +∞
2
hv2i i = √ i x2 e−x dx = σi2 .
π −∞
Therefore,
hv2 i = hv2x i + hv2y i + hv2z i = σx2 + σy2 + σz2 ≡ σ 2
and hv2 i = σ 2 .
17.8 From the Hubble law v = H0 r and for small v/c the redshift is z ≃ v/c. Thus, z ≃
H0 r/c. Expand the scale factor to first order in time,

da
a(t) ≃ a0 − (t0 − t) ≡ a0 − ȧ0∆t.
dt t0
The redshift can also be written as
λ0 a0 a0 − a
z= −1 = −1 =
λ a a
a0 − a0 + ȧ0 ∆t ȧ0 ȧ0 r
≃ ≃ ∆t = ,
a0 − ȧ0 ∆t a0 a0 c
where r = c∆t was used in the last step. Comparing with the earlier expression z ≃ H0 r/c
gives ȧ0 /a0 = H0 .
17.9 The coordinate distance is
dt ′
Z t0 Z t0
c
1 − H0(t ′ − t0 ) dt ′ ,

d=c ≃
t a(t ′ ) a0 t
where in the second step the expansion (17.36) was inserted, terms quadratic and higher in
t − t0 were dropped, and the integrand was expanded in a binomial series. Performing the
integration gives
c
(t0 − t) + 21 H0 (t0 − t)2 ,

d=
a0
which is Eq. (17.39) with a0 = 1.
17.10 Equation (17.36) follows directly from inserting Eqs. (17.34)–(17.35) into Eq. (17.33).
From z ≡ a0 a−1 − 1, and (17.36) for a,
−1
z = 1 + H0(t − t0 ) − 12 H02 q0 (t − t0 )2 − 1.
Then Eq. (17.37) follows from a binomial expansion (1 + x)−1 ≃ 1 − x + x2 , with terms
higher than second order in (t − t0 ) discarded. Equation (17.37) is a quadratic equation
in (t0 − t), which gives Eq. (17.38) when solved by the usual quadratic formula with the
positive solution, and with the square root expanded according to (1 + x)1/2 ≃ 1 + 21 x− 81 x2 .
17.11 Inserting Eq. (17.38) in Eq. (17.39) and keeping terms of quadratic order or lower
gives
c
z − (1 + 12 q0 )z2 + 21 z2

d(t0 ) =
H0

cz 1 + q0
= 1− z ,
H0 2
17.12 From the identity ä = 21 d(ȧ2 )/da and Eq. (17.14),
1 d 2 − 21 H02 a30 Ω
ä = ȧ = .
2 da a2
Solving this for d ȧ2 and integrating from the present time t0 to a time t,
Z a Z a
da
d ȧ2 = −H02 a30 Ω .
a0 a0 a2
Evaluating the integrals gives

2 1 1
ȧ = ȧ20 + H02a30 Ω − ,
a a0
and since ȧ0 = a0 H0 (see Problem 17.8),

2 2 2 a0
ȧ = a0 H0 1 + Ω − Ω, ,
a(t)
18 Friedmann Cosmologies
18.1 (a) The Friedmann equations for P = 0 and k = 0 are

ȧ
3ȧ2 = 8π Gρ a2 ρ̇ + 3ρ = 0.
a
Rewrite the first equation as

r
8 π Gρ
da = a dt
3
and the second is clearly satisfied by the condition ρ (t)a(t)3 = ρ0 , where ρ0 is the present
density. Thus, upon substituting ρ = ρ0 /a3 into the expression for da and solving for dt
we obtain
s
3
dt = a1/2 da.
8 π Gρ 0
Integrating this expression gives

s
2 3
t= a3/2 + constant.
3 8 π Gρ 0
Requiring that a = 0 when t = 0 implies that the constant is zero, so
a(t) = (6π Gρ0)1/3t 2/3
for the evolution of the scale factor.

(b) From this expression for a(t), the Hubble parameter is H = ȧ/a = 2/3t, the deceleration
parameter is
1 d 2a 1
q0 = − = ,
H 2 a dt 2 2
and taking t = 0 as the beginning, the age of the universe is just t = 2/3H0, where H0 is
the current value of H.
µν
18.2 The condition T ;ν = 0 with T µν = (ε + P)u µ uν + Pgµν implies that
∂ T µν µ
+ Γαν T αν + Γαν
ν
T µα = 0.
∂ xν
The non-zero components of Tµν are obtained from Eq. (18.31) and the corresponding T µν
89
can be obtained from these by contraction with the R–W metric tensor (18.16):
T00 = ε T 00 = ε
T11 = Pa2 /(1 − kr2) T 11 = P(1 − kr2)/a2
T22 = Pr2 a2 T 22 = P/(ra)2
T33 = Pr2 a2 sin2 θ T 33 = P/(ra sin θ )2 ,
where a is the scale parameter. Consider the µ = 0 component:
∂ T 0ν
+ Γ0αν T αν + Γν0ν T 00 = 0,
∂ xν
which is explicitly
Γ000 T 00 + Γ011 T 11 + Γ022T 22 + Γ033 T 33 + Γ000T 00

+ Γ101 T 00 + Γ202T 00 + Γ303 T 00 = 0.
The required Christoffel symbols are given in Table 18.1,

aȧ
Γ000 = 0 Γ011 = Γ022 = r2 aȧ Γ033 = r2 sin2 θ aȧ
1 − kr2
ȧ ȧ ȧ
Γ101 = Γ202 = Γ303 =
a a a
and from the preceding expressions for the stress–energy tensor components
∂ T 0ν ∂ T 00
ν
= = ε̇ .
∂x ∂ x0
Inserting these results into the previous expression for the µ = 0 component and collecting
terms gives
ȧ
ε̇ + 3(ε + P) = 0,
a
which expresses conservation of mass–energy.
18.3 Let’s use the Robertson–Walker metric in the form (18.19). For positive curvature
the spatial line element is
dℓ2 = a2 (d χ 2 + sin2 χ (d θ 2 + sin2 θ d ϕ 2 )),
so the metric is diagonal with
g11 = a2 g22 = a2 sin2 χ g33 = a2 sin2 χ sin2 θ
and
p
det g = a3 sin2 χ sin θ .
Then the volume is (see Section 3.13.1; here a positive sign is used under the square root
because det g is positive for the spatial part of the metric)

Z 2π Z π Z πp
V= dϕ dθ det g d χ
0 0 0
Z 2π Z π Z π
= a3 dϕ sin θ d θ sin2 χ d χ
0 0 0
= 2π 2 a3 .
For negative curvature the line element is
dℓ2 = a2 (d χ 2 + sinh2 χ (d θ 2 + sin2 θ d ϕ 2 )),
so the metric is diagonal with
g11 = a2 g22 = a2 sinh2 χ g33 = a2 sinh2 χ sin2 θ
and
p
detg = a3 sinh2 χ sin θ .
Therefore, for negative curvature the volume is

Z ∞
V = 4π a3 sinh2 χ d χ = ∞.
0
For a flat metric the line element is
dℓ2 = a2 (d χ 2 + χ 2(d θ 2 + sin2 θ d ϕ 2 )),
and proceeding as above the volume for flat space is

Z ∞
V = 4π a3 χ d χ = ∞.
0
Thus, the volume of a spatial slice described by a Robertson–Walker metric with positive
curvature is finite but the volume of a spatial slice described by a R–W metric with negative
curvature or no curvature is infinite.
18.4 This problem is suggested by a discussion in Padmanabhan [173]. Taking a0 = a(0) =
1, the current horizon distance is given by
Z t0 Z t0
cdt cdt t0
ℓh = = = .
0 a(t) 0 tn 1−n
The corresponding Hubble radius is
−1
d(t )/dt −1
n
1 ȧ t0
dH = = = = .
H0 a t=t0 tn t=t0 n
Thus ℓh /dH = n/(1 − n) ∼ 1 for radiation-dominated (n = 12 ) or matter-dominated (n = 32 )

cosmologies. However, ℓh and dH are conceptually rather different since dH is determined
only by the behavior near t = t0 but ℓh is sensitive to the entire past history of the Universe.
Hence, for cosmologies in which the scale factor is not of the form a(t) = t n with 12 ≤ n ≤ 23
the Hubble distance could be a poor indicator of the horizon distance.
18.5 This is mostly a matter of substitution and a lot of algebra. For example, in the
positive-curvature (closed universe) case Eqs. (18.17) may be used to write
dw = − sin χ d χ
dz = − sin χ sin θ d θ + cos χ cos θ d χ
dx = − sin χ sin θ sin ϕ d ϕ + sin χ cos θ cos ϕ d θ + cos χ sin θ cos ϕ d χ
dy = sin χ sin θ cos ϕ d ϕ + sin χ cos θ sin ϕ d θ + cos χ sin θ sin ϕ d χ .
Substitution of these derivatives in
dℓ2 = dx2 + dy2 + dz2 + dw2 ,
and a substantial amount of algebra abetted by use of trigonometric identities leads to
dℓ2 = d χ 2 + sin2 χ (d θ 2 + sin2 θ d ϕ 2 ),
which gives the first of Eqs. (18.19) when substituted in Eq. (18.2). The negative curvature
and flat cases may be proved in a similar way. The equivalence of Eqs. (18.14) and (18.19)
may be proved by using the change of variables in Eq. (18.20). For example, in the negative
curvature case substitute
dr2
r = sinh χ −→ sinh2 χ = r2 dr = cosh χ d χ −→ d χ 2 =
cosh2 χ
from Eq. (18.20) in the spatial part of the third equation in Eq. (18.19) to give
dr2

dℓ2 = + r 2
(d θ 2
+ sin2
θ d ϕ 2
) .
cosh2 χ
Upon substituting cosh2 χ = 1 + sinh2 χ = 1 + r2, this becomes
dr2

dℓ2 = + r 2
(d θ 2
+ sin2
θ d ϕ 2
) ,
1 + r2
which gives upon substitution in Eq. (18.2) the line element (18.14) with the curvature
choice k = −1.
18.6 Writing the implied sum over ν in Eq. (7.30) out explicitly,
3
∂ gµν ∂ gλ ν ∂ gµλ

Γλσ µ = ∑ 21 gνσ + − .
ν =0 ∂ xλ ∂ xµ ∂ xν
But the metric is diagonal so only the term with ν = σ survives in the sum. Then, for
example,

∂ g22 ∂ g02 ∂ g20 ∂ g22
Γ202 = 21 g22 0
+ 2
− 2
= 12 g22 0
∂x ∂x ∂x ∂x
1 1 2 ∂ (a2 ) ȧ
= r = ,
2 a2 r 2 ∂t a
where the metric given in Eqs. (18.15) and (18.16) has been used and (x0 , x1 , x2 , x3 ) =
(t, r, θ , ϕ ). The other Γλσ µ may be derived in similar fashion, with the results summarized
in Table 18.1. The Ricci tensor may be constructed from the connection coefficients and
Eq. (8.16),
Rµν = Γλµν ,λ − Γλµλ ,ν + Γλµν Γλσ σ − Γσµλ Γνσ
λ
.
For example, utilizing Table 18.1 and remembering that the only time dependence is in the
scale factor a, the non-vanishing terms for µ = ν = 1 are
∂ Γ011 ∂ Γ212 ∂ Γ313
R11 = − −
∂t ∂r ∂r
+ Γ011 Γ202 + Γ011 Γ303 + Γ111 Γ212 + Γ111 Γ313 − Γ011 Γ110 − Γ212Γ212 − Γ313Γ313 .
Inserting explicit forms for the connection coefficients from Table 18.1 and a little algebra
then gives
aä + 2ȧ2 + 2k
R11 = .
1 − kr2
The Ricci scalar (8.17) is obtained by contraction with the metric tensor, R = gµν Rµν . The
metric coefficients are given in Eq. (18.16) and the Ricci tensor components are given in
Eq. (18.32). Inserting them and carrying out some algebra yields
6(aä + ȧ2 + k)
R = gµν Rµν = g00 R00 + g11R11 + g22R22 + g33R33 =
a2
for the Ricci scalar.
18.7 This proof is completely analogous to that for Problem 8.11, except that one obtains
from the Einstein equation an additional term for R00 depending on Λ:
4π G 4π G
R00 = ρ + Λg00 = 2 ρ − Λ.
c2 c
Then carrying through exactly as in Problem 8.11 yields
∇2 ϕ = 4π Gρ − Λc2 ,
which is the Poisson equation modified by an additional term proportional to the cosmo-
logical constant Λ.
µν
18.8 As shown in Problem 18.2, the constraint T; ν = 0 evaluated for µ = 0 gives
ȧ
ε̇ + 3(ε + P) = 0.
a
This is equivalent to
d d
(ε a3 ) = −P a3 .
dt dt
A comoving volume of space has a physical volume V = α a3 , where α is a constant. The
total energy contained in this volume is U = αε a3 . Substituting this in the above equation
gives

d U d V
= −P → dU = −PdV.
dt α dt α
The first law of thermodynamics is dU = −PdV + δ Q, but by the cosmological principle
there can be no heat differences δ Q on large scales, so the preceding result is just the
first law of thermodynamics in an expanding universe described by the Robertson–Walker
metric.
18.9 Consider vectors u and v. The angle between them is given by
v·u g λ σ vλ u σ
cos θ = =q .
|v||u|
q
gµν vµ vν gαβ uα uβ
Under a conformal transformation gγδ → Φgγδ , where Φ is a scalar field (see Section
18.8.1). Clearly the above expression for θ is not modified by this transformation of the
metric tensor components since the scalar field factors Φ cancel.
18.10 Introduce the conformal time η through dt = a(t)d η . The flat (k = 0) Robertson–
Walker metric (18.14) is then given by
ds2 = −dt 2 + a2(dr2 + r2 d θ 2 + r2 sin2 θ d ϕ 2 )
= −a2 d η 2 + a2(dr2 + r2 d θ 2 + r2 sin2 θ d ϕ 2 )
= a2 (−d η 2 + dr2 + r2 d θ 2 + r2 sin2 θ d ϕ 2 ),
which is the same form as the metric for a uniformly-expanding Minkowski space. For
radial light rays d θ = d ϕ = ds = 0, so the line element implies that a2 (−d η 2 + dr2 ) = 0
and hence d η = ±dr. Thus, in the η –r plane light rays move at 45-degree angles at all
times.
19 Evolution of the Universe
19.1 From the Friedmann equations,

ε̇ P ȧ
+3 1+ = 0.
ε ε a
If an equation of state P = wε is assumed, this may be written as
dε da
= −(3 + 3w) .
ε a
Assume w to be constant and integrate both sides from now (a = a0 and ε = ε0 ) until some
scale factor a to give
−3(1+w)
ε (a) a
ln = ln ,
ε0 a0
which implies that
−3(1+w)
a
ε (a) = ε0 .
a0
Thus matter (w ≃ 0), radiation (w = 13 ), and vacuum energy (w < − 13 ) have very different
histories: for matter, εm ∝ a−3 ; for radiation, εr ∝ a−4 ; for vacuum energy, εΛ is constant,
if it is assumed that w = −1 (as implied by a cosmological constant).
19.2 Take the Solar System to be a sphere with radius 40 AU, assume the energy density
of radiation to be negligible, and assume the total mass to be approximately that of the Sun.
The vacuum energy density is from observations about 70% of critical density, so
εvac ≃ 0.7 × 1.05 × 10−2h2 MeV cm−3 ≃ 3.8 × 10−3 MeV cm−3 ,
where h = 0.72 was assumed. Multiplying this number by the volume of the Solar System
gives Evac ≃ 3.4 × 1042 MeV for the total vacuum energy contained in the Solar System.
The total energy of the matter in the Solar System is then estimated as Ematter ≃ (1M⊙ )c2 ≃
1.1 × 1060 MeV. Thus, the ratio of vacuum energy to total energy in the Solar System is of
order 10−18 . This suggests that a local experiment to measure the vacuum energy would
be very difficult.
19.3 (a) Assume air to be an ideal gas of nitrogen molecules N2 with mass µ ∼ 2 × 14 =
28 amu at T = 300 K. From Eq. (19.10), w = kT / µ c2 . Thus w is basically the ratio of
the thermal energy to the rest mass energy, which will be small for a nonrelativistic gas.
Inserting the numbers gives w = 9.9 × 10−13.
(b) For an ideal gas hv2 i = 3kT /µ , implying that w = kT /µ c2 = hv2 i/3c2 . Thus w may
95
also be interpreted as the ratio of the average of the velocity squared to the speed of light
squared, which is a very small number for nonrelativistic gases.
19.4 For a flat, matter-only universe Table 19.1 indicates that

2/3
t 2
a(t) = t0 = ,
t0 3H0
and at time t the proper distance to a photon emitted from the present particle horizon at
time t = 0 is given by
Z t0 ′ −2/3
dt ′
Z t0
t
ℓ(t) = a(t) = a(t) dt ′
ta(t ′ ) t t0

2/3 1/3 1/3
= 3t0 a(t) t0 − t
" #
t 2/3 t
= 3t0 − ,
t0 t0
where t0 is the time today (age of Universe). Even though the coordinate distance of the
photon from Earth decreases continuously, the proper distance first increases after emis-
sion, eventually reaches a maximum, and then decreases to zero as the photon reaches
Earth. The maximum proper distance is obtained by requiring that dℓ/dt = 0, which gives
that the time of maximum proper distance corresponds to t(ℓmax ) = 0.296t0, and using that
the current horizon distance is 2/H0 (Table 19.1) the maximum proper separation of the
photon from Earth at t = 0.296t0 is found to be 0.148 times the current horizon distance.
19.5 This is called the Milne universe. From Eqs. (19.57)–(19.59), the equations of motion
are given by

p 1 Ωr Ωm 2 1
dq = 2(E − U)d τ U =− + + Ω Λ q E = (1 − Ω).
2 q2 q 2
The Milne universe contains only curvature, so U = 0 and Ω ≡ Ωr + Ωm + ΩΛ = 0 and
E = 12 . Inserting these in the first equation above and integrating gives q(t) − q(0) = τ .
But q ≡ a(t)/a0 so q(0) = 0 and τ ≡ H0t, and from Eq. (19.61) the age of the Universe is
t0 = H0−1 . Combining these results gives
a(t) t
= .
a0 t0
Note that this solution implies negative curvature since from Eqs. (19.23)–(19.26)
−k
Ωk = 1 − Ω = 1 =
a2 H02
and thus k = −a2 H02 . The proper distance at the time of detection is
Z t0 Z t0
dt dt t0
ℓ(t0 ) = a0 = t0 = t0 ln ,
te a(t) te t te
where a0 /a(t) = t0 /t was used and te is the time of emission. The redshift is
a0 t0
z= − 1 = − 1.
a(te ) te
The proper distance at the current time in terms of redshift is then

t0 1
ℓ(t0 ) = t0 ln = t0 ln(1 + z) = ln(1 + z),
te H0
and the proper distance at the time of emission is smaller by a factor a(te )/a0 = te /t =
(1 + z)−1 , giving
1 1 ln(1 + z)
ℓ(te ) = ℓ(t0 ) = ,
1+z H0 1 + z
for the Milne universe.
19.6 From Eq. (19.12) generalized to multiple components (with factors of c restored)
4π Ga 4π Ga
ä = −
3c 2 ∑ (εi + 3Pi) = −
3c2 ∑
εi (1 + 3wi ),
i i
where each component has an equation of state Pi = wi εi . Multiply both sides by −(aH 2 )−1
and invoke the definition (17.34) of the deceleration parameter q0 to write this as

ä 4π G
aH t=t0 3H02 c2 ∑
q0 ≡ 2
= εi (1 + 3wi),
i
where H = ȧ/a was used. But from Eq. (17.6) the critical density is εc = ρc c2 = 3H02c2 /8π G,
so
1
q0 = ∑ Ωi (1 + 3wi ),
2 i
where Ωi ≡ εi /εc has been used. Taking a 3-component Universe with matter (w = 0),
radiation (w = 31 ), and vacuum energy (w = −1), this reduces to
q0 = 12 Ωm + Ωr − ΩΛ .
This Universe can exhibit acceleration (q0 < 0) only if it contains sufficient vacuum energy
such that ΩΛ > 12 Ωm + Ωr . With the further assumption of a flat Universe (which requires
that Ωm + Ωr + ΩΛ = 1),
q0 = 32 Ωm + 2Ωr − 1.
Finally, taking for the standard cosmology Ωm = 0.3, negligible Ωr , and ΩΛ = 0.7, one
obtains q0 = −0.55. Since the deceleration parameter is negative, the standard cosmology
represents a Universe that is currently accelerating, with its density parameters satisfying
Ωm + Ωr + ΩΛ = 1.
19.7 (a) Using Eq. (18.27) with factors of c restored and assuming a(t) ∝ t n with n < 1,
the proper horizon distance is
cdt ′
Z t Z t
ct
ℓh = a(t) = ct n (t ′ )−n dt ′ = .
0 a(t ′ ) 0 1−n
The ratio of this horizon distance to the scale factor is then

ℓh (t) ct 1 c
= · = t 1−n ,
a(t) 1 − n tn 1−n
which always grows with time since it has been assumed that n < 1.
(b) For a flat, matter-filled universe, Table 19.1 gives that a(t) = (t/t0 )2/3 , where t0 =
2/(3H0 ) is the age. Thus, n = 32 and from the preceding results, taking t = t0 (time today),

ct0 2 2c
ℓh = = 3ct0 = 3c = .
1−n 3H0 H0
Assuming that H0 = 72 km s−1 Mpc−1 , this yields a horizon distance of 8333 Mpc.
19.8 The result can be obtained immediately as a special case of the general solution in
Problem 19.6 but we can obtain it directly from Eqs. (17.34), (17.6), and the Friedmann
equation (19.12) for dust (assuming P = 0),
3H02

ä ä0 ä0 4π G
q0 ≡ −a 2 =− 2
ρc = =− ρ.
ȧ t=t0 a0 H0 8π G a0 3
Combining these gives
1ρ 1
q0 = = Ω,
2 ρc 2
where (17.8) was used.
19.9 Equation (19.40) may be written as
2
t0 = τH ,
3(1 + w)
where τH = 1/H0 is the Hubble time. Hence in a single-component, flat universe the age
of the universe is greater than τH if w < − 13 and less than τH if w > − 31 .
√
19.10 For the angular part of the metric (18.19) displayed in Box 19.3, we have det g =
Sk2 sin θ (where a0 ≡ 1) and the spherical surface has a proper area Ap (t0 ) that differs from
4π r2 in curved space:
Z 2π Z πp
Ap (t0 ) = dϕ det g d θ = 4π Sk (r)2 .
0 0
This is a consequence of curvature, independent of whether the Universe is expanding. If
the Universe is flat, Sk = r and an area of 4π r2 is recovered.
20 The Big Bang
20.1 Assume the Universe to have been matter dominated for most of it history, so a(t) ∼
t 2/3 . The coordinate distance between the source and the observer is
Z t0
dt
1/3 1/3

r≃ = 3 t0 − tr ,
tr t 2/3
where t0 is the present time and tr is the time of hydrogen recombination. The coordinate
horizon distance at tr is
Z tr
dt 1/3
rH = 2/3
= 3tr
0 t
and the number of horizons separating the sources in opposite directions is
" #
2r t0 1/3
nH ≃ ≃2 − 1 ≃ 90,
rH tr
assuming t0 = 1010 yr and tr = 105 yr.

20.2 At a temperature of 109 K it is assumed that the neutron/proton ratio is 0.164 as
nucleosynthesis begins. This means that out of 1000 nucleons 141 are neutrons and 859
are protons. How many 4 He nuclei can be made from these 1000 nucleons? That is just
a matter of counting. Each 4 He consists of 2 neutrons and 2 protons. Since there are 141
neutrons available for each 1000 nucleons, 141/2 = 71 4 He nuclei (rounded to an integer)
can be made. To make these helium nuclei requires one proton for every neutron used,
so the number of protons left over after all the neutrons are bound into helium nuclei is
859 − 141 = 718 protons. Hence for every 1000 initial nucleons, the Universe is left after
the initial period of big bang nucleosynthesis with 71 helium nuclei, 718 protons, no free
neutrons, and a trace of other light nuclei. Since 4 He is about 4 times as massive as a proton,
the mass fractions are (4 × 71)/1000 = 0.28 for 4 He and 718/1000 = 0.72 for hydrogen.
Therefore, these very simple considerations suggest that after big bang nucleosynthesis the
baryonic matter of the Universe should be about 28 percent 4 He by mass, with most of the
rest hydrogen. A much more sophisticated treatment of the problem that is summarized in
Fig. 20.5 obtains results that are not too different from these estimates.
20.3 The current baryon and radiation density parameters are Ωb (t0 ) ≃ 0.04 and Ωr (t0 ) ≃
8 × 10−5 (Table 17.1), and these need to be extrapolated back to a redshift (that is, scale
factor) where they become equal. Because the radiation density scales as a−4 and the bary-
onic density as a−3 ,
Ωb (t) a(t)−3 Ωb (t0 )

Ωb (t0 ) 1 Ωb (t0 )
= = a(t) =
Ωr (t) a(t)−4 Ωr (t0 ) Ωr (t0 ) 1 + z Ωr (t0 )
99
100 The Big Bang
Requiring that the ratio on the left side be unity then gives a redshift of
Ωb (t0 )
z= − 1 ≃ 500.
Ωr (t0 )
To estimate the corresponding time, assume that the vacuum energy can be neglected in
subsequent evolution and that the Universe evolved as matter dominated from the unknown
time t until today (t0 ). For a matter-dominated Universe, a ∼ t 2/3 , so
a0 t 2/3
0
= 1+z= ,
a t
and the corresponding time is
3/2
1
t= t0 .
1+z
Taking the present time to be t0 ∼ 13.8 × 109 yr and using z = 500 from above gives t =
1.2 × 106 yr after the big bang for equality of baryonic and radiation energy densities. This
result is approximately correct but a better treatment would account for things like the
influence of the neglected vacuum energy on evolution of the more recent Universe. A
more sophisticated relationship of redshift to time gives t ∼ 1.35 × 106 yr for equality of
baryonic and radiation densities [5]. Notice that what we have calculated is the redshift
for equality of the radiation density with the baryonic part of the matter density. From
Table 17.1 the total matter density parameter Ωm is about 7.5 times larger than the baryonic
density parameter, so the above estimate applied to the total matter density would give a
redshift z = 500 × 7.5 = 3750 for matter–radiation equality. This may be compared with
the measured value of z = 3365 from Table 20.2.
20.4 (a) For baryon density parameter Ωb ≃ 0.04 the baryon energy density is
εb = Ωb εc = 2.18 × 10−4 MeV cm−3 ,
where Eq. (17.7) with h = 0.72 was used.
(b) The average baryon is a proton with a rest mass energy of Ēb ∼ 939 MeV, so the number
density of baryons is
εb
nb = ≃ 2.32 × 10−7 cm−3 .
Ēb
(c) Most photons are in the CMB, which is a near perfect blackbody, so the energy density
is
εγ = aT 4 = 2.61 × 10−7 MeV cm−3 ,
where T = 2.725 K was used.
(d) For a blackbody the mean energy is about 2.7kT so
Ēγ ≃ 2.7kT = 6.34 × 10−10 MeV
and therefore we may estimate
εγ
nγ = = 411 cm−3 ,
Ēγ
101 The Big Bang
for the number density of CMB photons.

(e) Combining the above results,
nb 2.32 × 10−7 cm−3
η≡ = ≃ 5.6 × 10−10.
nγ 411 cm−3
for the ratio of baryon number density to photon number density.
20.5 The energy density for matter scales as ε ∼ a−3 ∼ (1 + z)3 , where 1 + z = a0 /a. Thus,
for the energy density of (cold) dark matter
εdm (zℓs ) = Ωdm (t0 )εc (t0 )(1 + zℓs )3
where t0 denotes current values. Since Ωm ∼ 0.3 and Ωb ∼ 0.04, take Ωdm = 0.26. Using
zℓs = 1100 and taking εc from Eq. (17.7) with h = 0.72 gives for the dark matter at last
scattering
εdm (zℓs )
εdm (zℓs ) = 1.9 × 106 MeV cm−3 ρdm (zℓs ) = = 3.38 × 10−21 g cm−3 .
c2
The baryon energy density scales with a in the same manner as cold dark matter, so for
baryonic matter at last scattering εb (zℓs ) = (Ωb (t0 )/Ωdm (t0 ))εdm (zℓs ), which gives
εb (zℓs )
εb (zℓs ) = 2.9 × 105 MeV cm−3 ρb (zℓs ) = = 5.1 × 10−22 g cm−3 .
c2
Photon energy densities scale as (1 + z)−4, so the photon energy density at last scattering
is
Ωγ (t0 )
εγ (zℓs ) = (1 + zℓs)εdm (zℓs ) = 3.9 × 105 MeV cm−3 ,
Ωdm (t0 )
where Ωγ (t0 ) ∼ 4.8 × 10−5 has been used. Thus we estimate that εdm : εγ : εb ∼ 6.6 : 1.4 : 1
at decoupling. A more sophisticated determination of these energy densities based on CMB
observations of the Planck satellite is given in Problem 20.11.
20.6 Estimate the age of the Universe as τH ≃ H0−1 ≃ 13.6×109 yr. For a matter dominated
Universe the scale factor is a = bt 2/3 , where b is a constant and t is the time since the big
bang. Evaluate b assuming a(0) = 0 and a(13.6 × 109 yr) = 1 to give t = 1.36 × 1010a3/2 yr.
Taking the redshift at decoupling to be z ≃ 1100, the scale factor at decoupling is adec =
(1 + z)−1 ≃ 9.08 × 10−4 and the time of decoupling is
tdec = (1.36 × 1010)(9.08 × 10−4)3/2 = 3.7 × 105 yr.
Using the speed of light c = 3.067 × 10−7 Mpc yr−1 , the maximum distance light could
have traveled from the big bang to decoupling is ctdec ≃ 0.113 Mpc. This distance would
have been expanded by a factor that is the ratio of the scale factors today and at decoupling:
0.113 Mpc × (a0 /adec ) = 0.113 Mpc × (1/9.08 × 10−4) ≃ 124 Mpc.
In the time τH light could have traveled cτH ≃ 4171 Mpc. Using this to approximate the
distance to the last scattering surface, one may estimate that the maximum angular size on
the sky of a causally-connected region in the CMB is θ ≃ 124 Mpc/4171 Mpc = 0.03 rad ≃
102 The Big Bang
1.7 degrees. This illustrates the horizon problem: the observed CMB appears to be causally
correlated over much larger angular regions than this. (See also Problems 20.1 and 21.3)
20.7 The entropy density is given by (see Section 20.2.1),
2π 2
s= g∗ T 3 ,
45
where g∗ is defined in Eq. (20.14). The difference in photon and neutrino background
temperatures is because when electron–positron annihilation falls out of equilibrium at a
temperature of about 0.2 MeV the annihilation energy raises the temperature of the photons
but not of the neutrinos, because the neutrinos have fallen out of equilibrium at a somewhat
higher temperature (weak-interaction decoupling). The amount of photon reheating can be
estimated using the following considerations:
1. Entropy is expected to be conserved across the transition, so g∗ T 3 before and after must
be equivalent.
2. From Eq. (20.14), the contribution to g∗ is (a) 1 × states of polarization for bosons and
(b) 7/8 × (states of polarization) for fermions.
3. Before the transition (at higher temperature) the relevant species and their contributions
to g∗ are
• For e− there are two spin states and the contribution is 87 × 2 = 47 .
• For e+ there are two spin states and the contribution is 78 × 2 = 47 .
• For photons there are two polarization states and the relative contribution is 2.
Thus the total g∗ before the transition is g+ 7 7 11
∗ = 4 + 4 +2 = 2 .
4. After the transition there are (essentially) only photons so g−
∗ = 2.
The ratio of entropy densities before (+) and after (−) the transition must satisfy
s+ g+ 3
∗ T+
= − 3
= 1,
s− g∗ T−
because of the entropy conservation assumption. Hence
+ 1/3
11/2 1/3
1/3
g∗ 11
T− = T+ = T+ = T+
g−∗ 2 4
and the photons get reheated by a factor T− = (11/4)1/3T+ ∼ 1.4T+ , but the neutrinos
are not reheated because they have previously decoupled from equilibrium (see Exam-
ple 19.1). Since the CMB temperature today is 2.725 K, the neutrino background radiation
is expected to have a lower temperature Tν = (4/11)1/3 × (2.725 K) ≃ 1.95 K.
20.8 For a blackbody spectrum one has (in c = 1 units) a number density distribution of
the general form
8π 2 ν 2 d ν
n(ν )d ν = .
e /kB T − 1
hν
Assume at a time t ′ that the photons exhibit a blackbody spectrum. The number of photons
103 The Big Bang
per comoving volume na(t)3 is approximately conserved. Thus, as the Universe evolves
from the earlier time t ′ to a later time t,
′ 3
a(t )
n(ν ,t)d ν = n(ν ′ ,t ′ )d ν ′ ,
a(t)
where ν is a frequency at time t and ν ′ is the corresponding frequency at time t ′ . But
because of the expansion redshift,
a(t) a(t)
ν′ = ν dν ′ = dν .
a(t ′ ) a(t ′ )
Substituting the blackbody form for the number density and the preceding expressions
gives
′ 3
a(t ) 8π 2 ν ′2
n(ν ,t) d ν = ′
hν /kB T (t ′ )
dν ′
a(t) e −1
a(t) 2 2

′ 3 8π 2 ν
a(t ) a(t ′ ) a(t)
= × × ′ dν
a(t) h[a(t)/a(t ′ )]ν/k T (t ′ ) a(t )
e B −1
8π 2 ν 2 d ν
= hν .
e /kB [a(t )/a(t)]T (t ) − 1
′ ′
But as the radiation expands ε ∝ T 4 and ε ∝ a−4 , implying that T ∝ a−1 . Thus,
a(t ′ )
T (t ′ ) = T (t),
a(t)
which may be substituted in the above to give
8π 2 ν 2 d ν
n(ν )d ν = .
e /kB T (t) − 1
hν
The original blackbody spectrum evolves into a new blackbody spectrum at the new (lower)
temperature T (t) = [a(t ′ )/a(t)] T (t ′ ) as the Universe expands from time t ′ to time t.
20.9 The bulk of the contribution for photons is from the CMB and for neutrinos from the
cosmic neutrino background. For fermions the effective degeneracy parameter g∗ is given
by
g∗ = 87 nf · ns · na ,
where nf is the number of flavors, ns is the number of spin polarizations, and na is the
number of particle–antiparticle species. There are a particle and antiparticle for each of
three known neutrino flavors. Neutrinos are spin- 12 particles with two spin (helicity) states.
However, the phenomenology of the weak interactions indicates that only the left-handed
neutrino and right-handed antineutrino participate in the weak interactions (maximal parity
violation). Thus, below the electroweak symmetry-breaking scale there is only one effec-
tive spin degeneracy state and for the neutrinos and antineutrinos
gν∗ = 78 × nf · ns · na = 78 × 3 × 1 × 2 = 21
4 .
104 The Big Bang
(It is assumed here that the neutrinos are Dirac particles having distinct particle and an-
tiparticle; if they are instead Majorana particles, for which a particle is its own antiparticle,
the last factor of 2 above would be reduced to 1.) For the photons, there are two polar-
ization states and gγ = 2. Thus, the ratio of energy density parameters for neutrinos and
photons is
gν Tν 4

Ων
= ∗γ ,
Ωγ g Tγ
where the last factor is because energy densities scale as T 4 for relativistic particles. From
Problem 20.7, the cosmic neutrino background temperature Tν and CMB temperature Tγ
are related by Tν = (4/11)1/3Tγ . Therefore,
4 4/3
gν

Ων Tν 21/4 4
= ∗γ = ≃ 0.68
Ωγ g Tγ 2 11
for the ratio of neutrino and photon energy densities.
20.10 Estimating the average energy as kT gives
E 126 × 109 eV
T= = ∼ 1.5 × 1015 K.
k 8.617 × 10−5 eV K−1
From Eq. (20.17), this corresponds to a time of ∼ 1.5 × 10−11 seconds after the birth of the
Universe, assuming that g∗ ∼ 100 (see Fig. 20.2). Check: read off from Fig. 20.4 that for a
temperature ∼ 102 GeV the time is t ∼ 10−11 s.
20.11 The scaling of the energy densities with expansion is
εdm ∼ a−3 = (1 + z)3 εγ ∼ a−4 = (1 + z)4 εb ∼ a−3 = (1 + z)3
From Table 20.2 and Table 17.1, take the density parameters today to be
Ωm = 0.308 Ωr = 8 × 10−5 Ωb = 0.048 Ωdm = Ωm − Ωb = 0.26
and from Eq. (17.7) the critical energy density is
εc = 1.05 × 10−2h2 MeV cm−3 = 4.83 × 10−3 MeV cm−3 ,
where h = 0.678 inferred from Table 20.2 was used. Thus for dark matter at a redshift for
last scattering zls ,
εdm (zls ) = Ωdm εc (1 + zls )3 = 1.6 × 106 MeV cm−3 ,
where zls = 1080 was used. Likewise, for photons
εγ (zls ) = Ωγ εc (1 + zls )4 = 5.3 × 105 MeV cm−3 ,
and for baryons
εb (zls ) = Ωb εc (1 + zls)3 = 2.9 × 105 MeV cm−3 .
The ratio of energy densities at decoupling was then εdm : εγ : εb ∼ 5.5 : 1.8 : 1, and the
Universe was dominated by dark matter at decoupling. See also related Problem 20.5.
105 The Big Bang
20.12 Putting the parameters from Table 20.2 into the cosmological calculator at Ref. [5]
gives an angular size distance dA = 12.9 Mpc for z = 1080. From D = dA ∆θ in Box 20.2,
on the last scattering surface assuming this cosmology

∆θ ∆θ
Dlss = dA ∆θ = 12.9 Mpc = 3.75 kpc.
rad arcmin
The smallest angular size resolved in the Planck data is about 5 arcmin, which gives ∼ 19
kpc when inserted in the above formula for the size on the LSS. This would have scaled up
to
D(t0 ) = Dlss (1 + zlss) ∼ 20.5 Mpc,
as observed today because of the Hubble expansion, assuming zlss = 1080. The current
baryon mass density of the Universe is ρb = Ωb ρc , which gives ρb ∼ 4.15 × 10−31 g cm−3
upon using Ωb = 0.048 and h = 0.678 from Table 20.2 in Eq. (17.6). Using this density
a sphere of diameter D = 20.5 Mpc contains a total baryonic mass of 2.8 × 1013M⊙ . This
is comparable (given our crude estimates) to the baryonic mass of a cluster of galaxies,
which typically have total masses of 1014 − 1015 M⊙ . For example, the total mass of the
(fairly rich) Virgo Cluster is estimated to be about ∼ 1015 M⊙ , of which probably 5-10% is
baryonic.
20.13 Before decoupling the barons and photons were strongly coupled, with the photons
greatly outnumbering the baryons√in the fluid. Thus the speed of sound was essentially that
of a photon gas, which is vs = c/ 3. The proper distance that sound could travel from the
big bang to the time of last scattering tls was then
Z t tls cdt
ls vs (t)dt
1
Z
ℓs (tls ) = a(tls ) ≃ √ a(tls )
0 a(t) 3 0 a(t)
1 1
= √ ℓh = √ (0.25 Mpc) = 0.144 Mpc
3 3
where we’ve used that at the time of last scattering for the CMB the distance to the horizon
ℓh was approximately 0.25 Mpc. From Box 20.2 the corresponding angular size on the
CMB as viewed today is
ℓs (tls ) 0.144 Mpc
θs = = = 0.011 rad = 0.6◦ ,
dA 12.9 Mpc
where an angular size distance dA = 12.9 Mpc was computed for the parameters in Ta-
ble 20.2 at a redshift z = 1080 [5]. The corresponding length scale as viewed today is
stretched by a factor 1 + zls:
ℓs (t0 ) = (1 + zls )ℓs (tls ) = (1 + 1080)(0.144 Mpc) = 156 Mpc.
The present total mass density is given by ρc from Eq. (17.6) since the Universe is flat. The
total matter density is ρm = Ωm ρc = 2.66×10−30 g cm−3 , where h = 0.678 and Ωm = 0.308
were used. Then the mass contained within a volume of radius 156 Mpc is on average
Ms ∼ 34 π (156 Mpc)3 ρm ≃ 6.3 × 1017M⊙ .
This is larger than the total mass for superclusters of galaxies and sets the minimal scale
106 The Big Bang
that must be analyzed to observe the effect of the baryon acoustic oscillations for clustering
of visible matter.
20.14 From the last equation in Box 19.3 we have dL = (1 + z)ℓ(t0 ), where ℓ(t0 ) is the
proper distance at the present time t0 . Thus, from the last equation in Box 20.2
dL
(1 + z)dA = = ℓ(t0 ),
1+z
and using Eq. (16.12) to relate scale factors to redshift,
ℓ(t0 )
dA = = ℓ(te ),
1+z
where ℓ(te ) is proper distance at the time of emission te .
21 Extending Classical Big Bang Theory
21.1 From Eq. (19.2),

4π G
ä = − a(ε + 3P).
3
Therefore, ȧ decreases with time unless ε + 3P ≤ 0. For standard cosmologies the horizon
may be approximated as the inverse Hubble parameter, H −1 = a/ȧ, so
a
≃ ȧ
H −1
and if ȧ decreases with time the ratio of the distance scale a(t) to the horizon ℓH (t) also
decreases with time:
(
ℓH (t) t 1/2 (radiation dominated)
≃ .
a(t) t 1/3 (matter dominated)
Therefore, in a standard cosmology the horizon grows more rapidly than the Universe
expands and objects not presently in causal contact have never been in causal contact. In
the theory of inflation ε + 3P becomes negative, which permits objects to expand outside
the horizon (outside of causal contact) during inflation, but then to come back inside the
horizon during the later standard big bang evolution after inflation has stopped.
21.2 From Eqs. (19.5) and (17.6)
∆ρ ρ − ρc 3k
= = .
ρ ρ 8π Ga2ρ
But ρ ∝ a−4 for radiation dominated evolution, in which case ∆ρ /ρ ≃ a2 ≃ t, so the devi-
ation from flatness grows smaller as time is extrapolated backwards. Take t0 ∼ 4 × 1017 s
for today. Then (assuming radiation dominance to make the estimate simple), at the Planck
time of t ∼ 10−44 s
,
∆ρ ∆ρ 10−44 s
≃ ≃ 10−62.
ρ t ρ t0 4 × 1017 s
So unless the flatness is tuned to this precision at the Planck scale, the Universe does not
evolve into one that is flat today.
21.3 First we make an unsophisticated argument. At the time of decoupling td ∼ 3 × 105 yr,
corresponding to a redshift zd ∼ 1100. The size of the horizon at this time is ℓd ∼ 2ctd
for radiation dominated and ℓd ∼ 3ctd for matter dominated cosmologies. (Let’s assume
radiation dominated for simplicity in estimates.) The size ℓ corresponding to this horizon
in the current Universe would be stretched by the expansion according to ℓ/ℓd = a0 /ad .
However, zd = a0 /ad − 1, so ℓ ≃ 2ctd (1 + zd ) is the size of a causally connected region at
107
decoupling in the present Universe. Assuming a flat Universe, one may take the distance
to the last scattering surface to be close to the present horizon and given approximately
by c(t0 − td ). Thus, the approximate angular size of causally connected regions on the last
scattering surface is
2ctd (1 + zd )
θ≃ ≃ 0.047 rad.
c(t0 − td )
Hence regions in the sky separated by more than a degree or so should not have been
causally connected at any time in the past (in standard big bang cosmology).
A more sophisticated argument can be made by using the angular diameter distance
discussed in Box 20.2 in the form dA = ℓd /∆θ . In the standard cosmology the angular size
distance corresponding to z = 1100 can be computed to be about 12.9 Mpc [5], and the
horizon size at decoupling is about ℓd ∼ 2ctd ∼ 0.184 Mpc, assuming radiation dominance.
Then
ℓd 0.184 Mpc
∆θ = = = 0.014 rad ∼ 0.8◦ .
dA 12.9 Mpc
Again we conclude that regions on the last scattering surface separated by more than a
degree or so cannot have been in past causal contact in the standard cosmology, yet widely
separated regions on the sky are observed to have the same cosmic microwave background
temperature to one part in 105 . This is the horizon problem.
21.4 By analogy with the solution of Problem 21.3, the physical horizon size at the GUT
transition may be estimated as rh ≃ 2ctGUT ≃ 6 × 10−26 cm, if tGUT = 10−36 s. Assuming
one monopole per horizon volume, the number density of monopoles at the GUT transition
is then
nM ∼ (rh )−3 = 4.6 × 1075 cm−3 ,
and the energy density of monopoles will be
εM ≃ 1015 GeV × nM ≃ 4.6 × 1090 GeV cm−3 ,
where we’ve assumed the average mass of a monopole to be the GUT scale. The tempera-
ture at the GUT transition is about 1028 K, so the energy density of radiation is
4
εr = aTGUT ∼ 4.7 × 10100 GeV cm−3 .
This is 10 orders of magnitude larger than the energy density of monopoles, so at the GUT
scale the Universe is highly radiation dominated. However, the radiation energy density
scales as a−4 and the massive monopole energy density as a−3 . Thus, after the Universe
expands to a scale factor approximately 1010 times that at the GUT scale, the Universe will
begin to be dominated by the monopole energy density. From Eqs. (20.17) and (20.5), in
the early Universe T ∼ a−1 and t ∼ a−2 , so this transition will occur when the temperature
has fallen to 1028 K × 10−10 ≃ 1018 K, at a time of 1036 s × (10−10 )2 ≃ 10−16 seconds
after the Big Bang. Thus, the early Universe would have been strongly matter dominated,
contradicting the observational evidence.
21.5 This problem is suggested by a discussion in Ryden [209]. For n = 100 e-foldings,
the expansion during inflation is

a(tf )
=≃ en ≃ e100 ≃ 1043 .
a(ti )
The pure vacuum-energy solution is a de Sitter space. From Eqs. (19.5) and (19.14)–
(19.15) with Tµν = 0, it is necessary to solve ȧ/a = H 2 , with H ≡ (Λ/3)1/2 , which has
a solution a(t) ∼ eHt . The corresponding vacuum energy density is (with factors of c re-
stored)
c2 3c2 2
εΛ = Λ= H .
8π G 8π G
For the present model of inflation H ∼ 1036 s−1 , which gives upon inserting into the above
equation εΛ ∼ 10102 GeV cm−3 . On the other hand, the vacuum energy density correspond-
ing to the present accelerated expansion of the Universe is about 70% of closure density.
From Eq. (17.7) with h = 0.72, this is εΛ ∼ 3.8 × 10−6 GeV cm−3 . It may be concluded
that the present accelerated expansion is being driven by a vacuum energy density that is
some 107 orders of magnitude smaller than that for the model assumed above for inflation
in the early Universe.
22 Gravitational Waves
22.1 Contract both sides of the Einstein equation (8.21) with the metric tensor,
8π G µν
gµν Rµν − 12 gµν gµν R = g Tµν .
c4
The first term on the left reduces to R and the second term to 2R, and the term on the right
reduces to (8π G/c4)T νν . Solving for R then gives
8π G ν
R=− T ν.
c4
Inserting this back in the original Einstein equation gives
8π G
Rµν = (Tµν − 12 gµν T αα ).
c4
The vacuum Einstein equation Rµν = 0 then results from setting Tµν and T αα to zero.
22.2 (a) In linearized gravity gµν = ηµν + hµν . Under a Lorentz transformation of the
metric,
β β
g′µν = Λαµ ′ Λν ′ gαβ = Λαµ ′ Λν ′ (ηαβ + hαβ ) = ηµν
′
+ h′µν ,
β
where h′µν ≡ Λαµ ′ Λν ′ hαβ . Thus the field defined by hµν behaves as a rank-2 tensor in
Minkowski space, for which indices may be raised or lowered by contraction with the
Minkowski metric tensor ηµν .
(b) Since gµν and g µν must be matrix inverses of each other, this requires that gµν =
η µν − hµν so that
g µα gαν = (ηµα + hµα )(η αν − hαν ) = δµν + O h2 ,

to first order in h.
22.3 Substituting Eq. (22.6) in Eq. (22.7) and noting that h µν is symmetric under exchange
of indices,
∂ 1 γδ ∂ hδ µ ∂ hδ ν ∂ hµν ∂ 1 γδ ∂ hδ µ ∂ hδ γ ∂ hµγ

δ Rµν = γ η + − − ν η + −
∂x 2 ∂ xν ∂ xµ ∂ xδ ∂x 2 ∂ xγ ∂ xµ ∂ xδ
1 h i 1 h i
= ∂γ η γδ ∂ν hδ µ + ∂µ hδ ν − ∂δ hµν − ∂ν η γδ ∂γ hδ µ + ∂µ hδ γ − ∂δ hµγ

2 2
1
= −∂γ ∂δ η hµν + ∂µ ∂γ η hδ ν − ∂µ ∂ν η γδ hδ γ + ∂ν ∂δ η γδ hµγ .
γδ γδ
2
Introducing the definitions
γ γ
≡ η γδ ∂γ ∂δ Vν ≡ ∂γ hν − 12 ∂ν hγ = ∂γ η γδ hδ ν − 12 ∂ν η γδ hδ γ
110
and noting that, for example,
∂µ Vν = ∂µ ∂γ η γδ hδ ν − 12 ∂µ ∂ν η γδ hδ γ ,
the Ricci tensor to first order in h becomes

1

δ Rµν = 2 − hµν + ∂µ Vν + ∂ν Vµ .
Then to first order in the metric perturbation h the vacuum Einstein equation δ Rµν = 0 is
hµν − ∂µ Vν − ∂ν Vµ = 0,
22.4 Consider linearized gravity with the metric given by Eq. (22.2). Under the coordinate
transformation (22.13), xµ → x′ µ = xµ + ε µ (x), where it is assumed that ε µ and ∂ ε µ /∂ xν
have magnitudes comparable to or smaller than hµν , the metric tensor transforms as
∂ xα ∂ xβ
g′µν = g
∂ x′ µ ∂ x′ ν αβ !

∂ ε α ∂ ε β
β
≃ δµα − µ δν − ν gαβ
∂x ∂x
∂ εβ ∂ εα
≃ gµν − gµβ − g αν ,
∂ xν ∂ xµ
where Eq. (22.13) was used, we have assumed that to first order ∂ ε µ /∂ x′ ν = ∂ ε µ /∂ xν ,
and terms higher-order in ∂ ε /∂ x have been neglected. Hence, from Eq. (22.2)
h′µν = g′µν − ηµν

∂ εβ ∂ εα
= hµν − gµβ ν
− gαν µ
∂x ∂x
= hµν − ∂µ εν − ∂ν εµ ,
22.5 Transform to transverse traceless (TT) gauge by choosing

µ
h̄0i = 0 (i = 1, 2, 3) Tr h̄ ≡ h̄ µ = 0,
implying from Eq. (22.19) that the polarization tensors in TT gauge satisfy
µ
α0i = 0 Tr α = αµ = 0.
(Note that we are now in TT gauge where h̄ = h. Hence the bars could be dropped on h if
desired, but we will keep them for consistency.) From the gauge condition (22.17) and the
above constraints, ∂ h̄00 /∂ x0 = 0. Writing the gauge condition (22.17) for µ = 0 out term
by term subject to the preceding consraints also gives
∂ h̄01 ∂ h̄02 ∂ h̄03
+ + = 0,
∂ x1 ∂ x2 ∂ x3
and evaluating the partial derivatives using Eq. (22.19),
h̄ µν = αµν eik·x = αµν e−i(ω t−kk ·xx ) ,
leads to the requirements that
∂ h̄00 ∂ h̄01 ∂ h̄02 ∂ h̄03

= 0 → iωα00 eik·x = 0 + + = 0 → iki αi j e−ik·x .
∂ x0 ∂ x1 ∂ x2 ∂ x3
The first can be satisfied generally only if α00 = 0, so in TT gauge four of the polarization
tensor components are identically zero: α0µ = αµ 0 = 0, which is Eq. (22.23). The second
can be satisfied generally only if ki αi j = 0, which is the transversality condition (22.24).
22.6 The transversality condition k j αi j = 0 from Eq. (22.24) may be written out explicitly
as the set of equations
k1 α11 + k2 α12 + k3 α13 = 0

k1 α21 + k2 α22 + k3 α23 = 0
k1 α31 + k2 α32 + k3 α33 = 0.
But from (22.26), k1 = k2 = 0, so α13 = α23 = α33 = 0, and from (22.23), α0µ = 0. There-
fore, for the symmetric matrix αµν the only nonvanishing components are α11 , α12 = α21 ,
and α22 , and these are further constrained by the trace requirement from Eq. (22.22), so
α22 = −α11 .
22.7 For LISA take L = 2.5 × 109 m, which gives

c
f∗ = = 1.9 × 10−2 Hz.
2π L
For LIGO the laser effective path length is increased over the physical arm length by of
order 100 because of multiple reflections. Taking an effective 200 reflections as an estimate,
L ∼ 200 × 4000 m ∼ 8 × 105 m and
c
f∗ = = 59.7 Hz.
2π L
Comparing with Fig. 22.8, these are indeed the approximate optimal response frequencies
for LISA and advanced LIGO.
22.8 From Eqs. (22.19) and (22.16)

α

h̄µν = η λ σ ∂λ ∂σ αµν eikα x
α

= η λ σ ikσ ∂λ αµν eikα x
= η λ σ ikλ ikσ hµν

= −kλ kλ h̄µν = 0.
But h̄µν is not generally zero so a solution of the wave equation requires that k be a null
vector, kλ kλ = 0.
22.9 The test particle is initially at rest with a 4-velocity uµ = (c, 0, 0, 0). The geodesic
equation (7.23) thus reduces to
duµ µ µ
= −Γ00 (u0 )2 = −c2 Γ00 .
dτ
From Eq. (22.6), to first order in h,
µ
Γ00 = 12 η µν (∂0 hν 0 + ∂0hν 0 − ∂ν h00 ),
µ
but from Eqs. (22.23) and (22.19), hν 0 = h00 = 0, so Γ00 = 0 and the initial 4-acceleration
vanishes, duµ /d τ = 0. (Note that we are in TT gauge where h = h̄.) Thus, in TT gauge the
particle is stationary with respect to the coordinate system as the gravitational wave passes.
23 Weak Sources of Gravitational Waves
23.1 Use Eq. (23.18) to estimate L ∼ 8 × 10−23 erg s−1 . Obviously detection of gravita-
tional waves produced in the laboratory is not a practical experiment.
23.2 From Eq. (23.28) for circular orbits with c and G factors evaluated (see Problem 5.8),
4/3 2 10/3
M µ 1s
L = 2.3 × 1045 erg s−1 .
M⊙ M⊙ P
Inserting M ∼ 2.8 M⊙ , µ = 0.7M⊙ , and P = 7.75 h gives 6.8 × 1030 erg s−1 . However,
the orbit has eccentricity e = 0.617, which gives a correction factor f (e) = 11.84 from Eq.
(23.30). Thus, from Eq. (23.29) the gravitational wave luminosity of the Binary Pulsar may
be estimated as L ∼ (11.84) × (6.8 × 1030 erg s−1 ) ∼ 8 × 1031 erg s−1 . This is about 2% of
the photon luminosity of the Sun but it is much harder to detect gravitational waves than
photons so it isn’t feasible to observe the gravitational-wave energy emitted by the Binary
Pulsar directly. It can be inferred only indirectly from the observed decay of the orbit. See
Problem 23.9 for an estimate of whether the gravitational wave strain for the Binary Pulsar
is detectable from Earth.
23.3 From Eq. (23.7) and being careless about numerical factors,
... MR2 MR3 Mv3

I-i j ≃ 3 = 3
∼ ,
P RP R
where Eq. (23.9) has been used. Inserting this equation into Eq. (23.16) and eliminating
M in favor of the Schwarzschild radius rS = 2M gives Eqs. (23.18)–(23.19) [a numerical
factor has been omitted in (23.18)]. Utilizing Eq. (23.12), this also may be expressed as
L ≃ L0 (rS /R)5 . The total energy emitted in one period P is
r 2 v6

∆E ≃ LP ≃ L0 S2 6 P.
R c
Utilizing P = 2π R/v, (23.12), (23.13), rs = 2M, and (23.19), this gives Eq. (23.20),
r 7/2
S
∆E ≃ Mc2 = ε Mc2 ,
R
where a factor of 2π has been dropped. Thus ε is a measure of the efficiency of converting
mass to gravitational waves.
23.4 Defining the reduced mass µ ≡ m1 m2 /(m1 + m2 ), separation a ≡ a1 + a2 , and total

114
mass M ≡ m1 + m2 , and noting that by the definition of the center of mass m1 a1 = m2 a2 ,

m1 m2 m1 m2 m2 m1 m2
µa = a1 + a2 = 2 a2 + a2
M M M M
2
m2 m1 m2
= + a 2 = m2 a 2 .
M M
By an analogous proof µ a = m1 a1 and thus µ a = m1 a1 = m2 a2 . Therefore,
m1 a21 + m2 a22 = µ aa1 + µ aa2 = µ a(a1 + a2) = µ a2 .
Hence, from Eq. (23.21) for the second mass moments expressed in the coordinates (23.22),
I xx = I 11 = m1 x21 + m2 x22
= m1 a21 cos2 ω t + m2 a22 cos2 ω t
= (m1 a21 + m2 a22 ) cos2 ω t
= µ a2 cos2 ω t
= 21 µ a2 (1 + cos2ω t),
and by analogous proofs
I xy = I yx = I 12 = m1 x1 y1 + m2 x2 y2
= m1 a21 cos ω t sin ω t + m2 a22 cos ω t sin ω t
= µ a2 cos ω t sin ω t = 12 µ a2 sin 2ω t
I yy = I 22 = m1 y21 + m2 y22
= m1 a21 sin2 ω t + m2 a22 sin2 ω t
= µ a2 sin2 ω t = 21 µ a2 (1 − cos2ω t),
which are the results quoted in Eq. (23.23).
23.5 From Eq. (23.23) the non-zero components are
I 11 = I xx = µ a2 cos2 ω t = 21 µ a2 (1 + cos2ω t),
I 12 = I xy = µ a2 cos ω t sin ω t = 21 µ a2 sin 2ω t,
I 22 = I yy = µ a2 sin2 ω t = 21 µ a2 (1 − cos2ω t).
The trace-reversed amplitude is given by Eq. (23.4), which requires the second time deriva-
tives. These are easily computed from the above equations. For example,
d 1 2
I˙xx (t) = µ a (1 + cos2ω t) = −µ a2 ω sin 2ω t,

dt 2
I¨xx (t) = −2ω 2 µ a2 cos 2ω t,

2 ¨xx −4ω 2 µ a2
h̄xx = I (t − r) = cos 2ω (t − r).
r r
Computing the second time derivatives for the other components in like manner gives
− cos 2ω (t − r) − sin 2ω (t − r) 0
 
2 2
4ω µ a 
h̄i j = − sin 2ω (t − r) cos 2ω (t − r) 0 ,
r
0 0 0
which is Eq. (23.24). The triple time derivatives required to compute the gravitational wave
power,
...xx ...yy
I (t) = 4ω 3 µ a2 sin 2ω t I (t) = −4ω 3 µ a2 sin 2ω t
...xy ...yx ,
I (t) = I (t) = −4ω 3 µ a2 cos 2ω t
may be found from the second derivatives computed above.
23.6 Differentiating both sides of Eq. (23.33) leads to
1 dP 3 da
= .
P dt 2a dt
Assume by energy conservation that the decay of the orbit causing the decrease in period
results from emission of gravitational waves. The total energy of the binary orbital motion
is given in Newtonian approximation by Eq. (23.32),
Gm1 m2
E =− ,
2a
from which
1 da 1 dE
=− ,
a dt E dt
and combining the first and third equations from above gives
1 dP 3 1 dE
=− .
P dt 2 E dt
Equating the change in orbital energy with the energy carried off by gravitational waves,
dE/dt = −L, and using Eq. (23.27) to specify L gives
1 dE 64 G3 M 2 µ
= .
E dt 5 c5 a 4
Therefore,
dP 3 1 dE 96 G3 M 2 µ
=− P=− P,
dt 2 E dt 5 c5 a 4
which is Eq. (23.34). The period P and the separation a are related by Kepler’s 3rd law
a3 = (GM/4π 2 )P2 , which can be used to eliminate a, giving an expression depending only
on the period and masses
192π G5/3 m1 m2 2π 5/3

dP
=− ,
dt 5 c5 M 1/3 P
23.7 The mass of the system contributing to gravitational wave radiation is assumed to be
∼ 0.5M⊙ and the effective radius is taken to be R ≃ 2R⊙ ≃ 14 × 105 km. From Eq. (23.13)
2(G/c2 )M

2/7 rs M km
ε = = = 2.95 ,
R R M⊙ R
which gives for 44 Boo an efficiency ε 2/7 ∼ 10−6. Therefore, from Eq. (23.15) and the
period of 6.4 hours the amplitude and frequency of the expected gravitational wave metric
perturbation is
h̄ ≃ 7 × 10−21 f ≃ 8.7 × 10−5 Hz,
since [see Eq. (23.23)], the gravitational wave frequency is twice the binary frequency of
revolution. Consulting Fig. 22.8, the expected gravitational wave frequency is outside the
favorable response range of LIGO or Virgo, but within the frequency window for LISA.
Thus, it is possible that space-based arrays may be able to detect gravitational radiation
from some galactic binaries.
23.8 Restore c and G factors in Eq. (23.8) to give
4G MR2
h̄i j =
c4 rP2
and use
2/3
GM
R2 = P4/3
4π 2
from Kepler’s 3rd law to eliminate R, giving after evaluation of constants
M 5/3 s 2/3 km

ij −4
h̄ = 1.47 × 10 ,
M⊙ P r
for masses given in solar masses, periods in seconds, and distances in kilometers.
23.9 From the information in Section 10.4.1, assume that the average separation of the
neutron stars is R ∼ 2R⊙ , the effective mass entering into generation of gravitational waves
is M ∼ 1 M⊙ , and the distance is r = 6.4 kpc. Then from Eq. (23.13)
rs 2.95(M/M⊙ )
ε 2/7 = = = 2.1 × 10−6,
R 2R⊙
and from Eq. (23.15)

−17 2/7 M kpc
h̄ = 9.6 × 10 ε ≃ 3.2 × 10−23.
M⊙ r
The period of the binary is 7.75 hours, implying an orbital frequency 3.6 × 10−5 s−1 .
The gravitational wave frequency is twice that, f = 7.2 × 10−5 s−1 . From Fig. 22.8 this
is roughly in the LISA frequency window but the strain is several orders of magnitude too
small to be measurable by LISA.
24 Strong Sources of Gravitational Waves
24.1 The event corresponded to the merger of two black holes having a total mass of ∼
70 M⊙ . The observed frequency near peak was ∼ 150 Hz, implying a period for revolution
of the binary of half that or 75 Hz. Assuming Kepler trajectories for a rough estimate, the
separation between the objects was
1/3
G 2
r1 + r2 = (m1 + m2)P
4π 2
" 2 #1/3
3 m1 + m2 P
= 1.5 × 10 km.
M⊙ s
For a period and total mass
P = (75 Hz)−1 = 1.33 × 10−2 s m1 + m2 = 70 M⊙ ,
this gives 347 km for the separation. The Schwarzschild radius for a black hole of mass
M is rs = 2.95 M km, which gives 207 km for the sum of Schwarzschild radii assuming
m1 + m2 = 70M⊙ .
24.2 The chirp waveform in the bottom panel of Fig. 24.4 indicates a binary merger. The
theoretical chirp mass
µ 3/5
M=
M 2/5
from Eq. (24.2) is plotted as a function of m1 for different values of m2 in Fig. 24.1 [this
document]. The chirp mass M ∼ 28 ± 2 M⊙ (Table 24.1; see also Problem 24.5) deter-
mined observationally from the frequency and its time derivative of the gravitational wave
is indicated by the dashed horizontal line and gray uncertainty box. By summing m1 and
m2 at the intersections of the curves with the M = 28 line, one sees that the minimum total
mass of the binary consistent with the chirp mass is around 65 M⊙ − 70 M⊙ . From Prob-
lem 24.1 the separation of centers at the time of maximum frequency was about 350 km.
Only black holes or neutron stars are compact enough to be consistent with that. Assuming
neutron stars to have an upper mass limit of ∼ 2 M⊙ , two neutron stars would have far
too little chirp mass to account for the data. From the m2 = 2 M⊙ curve, for a neutron star
and black hole to give the observed value of M the mass of the black hole would have
to be huge, giving a very large total mass for the system that would lead to a much lower
gravitational wave frequency than observed. Thus a black hole and neutron star binary is
ruled out, leaving merging black holes with a total summed mass near 70 M⊙ as the only
plausible explanation.
24.3 Although the binary pulsar results of Section 23.2.3 are rather convincing, they are
118
70
m2
60 80
70
Chirp mass (solar masses)

60
50
50
40
40 35
30
25
30 Observed 20
chirp mass
20 10
10
2
0
0 10 20 30 40 50 60 70 80
m1 (solar masses)
t
Fig. 24.1 Chirp mass for Problem 24.2 as a function of m1 for different values of m2 .
not the same as a direct observation. Perhaps more important is that the GW150914 event
is the first observation that tests general relativity in the strong-field limit; all previous
tests have been for conditions in weak gravitational fields and thus have not been a full
test. Finally, the astrophysical implications are potentially enormous, since GW150914
represents the first observation of radiation emitted from near the event horizon of a black
hole, and the event suggests that there are extremely energetic things happening in the
Universe that are not readily visible in electromagnetic waves.
24.4 The gravitational wave travels at v = c so the frequency f and wavelength λ are re-
lated by λ = c/ f . At peak strain the frequency of GW150914 was about 150 Hz, implying
that λ ∼ 2000 km. Estimate the characteristic source size as 4rs for the black holes near
merger. Assuming 30M⊙ for each black hole, this gives a source size of about 350 km.
Thus λ /d ∼ 6 and the quadrupole formula should be at least approximately correct. From
Table 24.1 the redshift for the gravitational wave source was estimated as z ∼ 0.09, so the
corresponding frequency f0 in the rest frame of the source at peak strain was
f0 = (1 + z) f = (1 + 0.09)150 Hz = 163.5 Hz,
where f is the measured frequency. The rotational frequency ω of the binary is half the
gravitational wave frequency, giving ω ∼ 81.75 s−1 . From Eq. (23.31) the luminosity is
4/3 2 10/3
M µ 1s
L = 2.3 × 10 45
erg s−1 ,
M⊙ M⊙ P
where M is the total mass, µ is the reduced mass, and P is the period. Inserting P = ω −1 =
1.22 × 10−2 s and assuming m1 = 36 M⊙ and m2 = 29 M⊙ yields a peak luminosity of
L = 3.7 × 1056 erg s−1 , which is consistent with the value given in Table 24.1.
24.5 From Eq. (24.2) with constants evaluated

3/5
c3 5 3/5 −8/5 −11/5 ˙3/5 f −11/5 f˙

M= π f f = 5.53 × 103 −1 M⊙ .
G 96 s s−2
Estimating from the bottom panel of Fig. 24.4 that at peak strain f ∼ 150 s−1 and f˙ ∼
∆ f /∆t ∼ 1.6 × 104 s−2 and inserting in the above equation gives M ∼ 30 M⊙ , which is
consistent with the value quoted in Table 24.1.
24.6 Assuming the validity of Newtonian mechanics and Newtonian gravity, the total (ki-
netic plus potential) energy is
Gm1 m2 Gm1 m2
E = 21 m1 v22 + 12 m2 v22 − = 21 m1 r12 ω 2 + 21 m2 r22 ω 2 − ,
a a
where v = ω r has been used with ω ≡ 2π /P. But
m2 m1
r1 = a r2 = a M ≡ m1 + m2 ,
M M
allowing the total energy to be written
m 2 m 2 Gm1 m2
2 1
E = 21 m1 a2 ω 2 + 12 m2 a2 ω 2 −
M M a
Gm 1 m 2
= 21 µ a2 ω 2 − ,
a
where µ ≡ m1 m2 /M is the reduced mass. Eliminating the frequency ω using Kepler’s 3rd
law in the form a3 = GM/ω 2 then gives for the total orbital energy E = −Gm1 m2 /2a.
25 Tests of General Relativity
25.1 Any viable relativistic gravitational theory should agree with the results of Newtonian
gravity in the weak-field limit, as described in Section 8.1. There it was shown that the
lowest-order relativistic correction to flat space modifies only the g00 component of the
metric to g00 = −(1 − 2GM/rc2) [see Eq. (8.12)], with the other components unaltered to
lowest order. Comparing with Eq. (25.1), agreement of general relativity with Newtonian
gravity in the weak-field limit requires that to lowest order
2GM
A(r) = 1 − + ... B(r) = 1 + . . ..
rc2
which is Eq. (25.3) to this order.
25.2 From Eq. (6.5), the strength of the gravitational field is measured by

GM M km M km
ε = 2 = 7.416 × 10−31 = 1.475 ,
Rc kg R M⊙ R
over which it acts.
1. For terrestrial experiments R = R⊕ and M = M⊕ , which gives ε ∼ 7 × 10−10.
2. For Mercury, take its average distance from the Sun of 5.7 × 107 km for R and M =
1 M⊙ , which gives ε ∼ 2.6 × 10−8.
3. For light deflection at the surface of the Sun take R ∼ R⊙ and M = 1 M⊙ , leading to
ε ∼ 2.1 × 10−6.
4. For the Binary Pulsar, take M ∼ 1.4 M⊙ and a smallest separation R ∼ 1.1R⊙, to give
ε ∼ 2.7 × 10−6.
5. For S0-2 take a closest approach to the black hole of R = 17 lighthours = 1.8 × 1010 km
and M = 4.3 × 106 M⊙ , which yields ε = 3.5 × 10−4.
6. For GW150914 estimate that at closest approach in the black hole merger R ∼ 100 km
and M ∼ 70 M⊙ , which gives ε ∼ 0.5.
See also the related Problem 10.8.
121
26 Beyond Standard Models
26.1 This solution follows an example in Zwiebach [257]. For a square well in two vari-
ables (x, y) ∼ (x, y + 2π R), the Schrödinger equation is
h̄2 ∂ 2ψ ∂ 2ψ

− + 2 = Eψ.
2m ∂ x2 ∂x
Substituting ψ (x, y) = ψ (x)ϕ (y), the Schrödinger equation becomes
h̄2 1 d 2ψ (x) h̄2 1 d 2ϕ (y)

− 2
− = E.
2m ψ (x) dx 2m ϕ (y) dy 2
The solutions of this equation are

kπ x ℓy ℓy
ψk (x) = ck sin ϕℓ (y) = aℓ sin + bℓ cos
a R R
" #
2 2
h̄2 kπ ℓ
Ek,ℓ = + (k = 1, 2, 3, . . .∞; ℓ = 0, 1, 2, . . . ∞),
2m a R
where ℓ = 0 is allowed because of the boundary conditions in the y direction. As shown in

Problem 26.2, if R is small the new states introduced by the compactified y dimension will
be very high in energy.
26.2 From the spectrum obtained in the solution of Problem 26.1, if ℓ = 0

" #
h̄2 kπ 2 ℓ 2 h̄2 kπ 2

Ek,ℓ = + −→ ,
2m a R 2m a
which is the spectrum of the 1-dimensional square well. Thus the states with ℓ = 0 are the
old states of the 1-dimensional square well. The lowest-energy new state corresponds to
k = 1 and ℓ = 1, giving
" 2 #
2 2
h̄2 1 2

new h̄ π 1
Emin = E1,1 = + ≃ ,
2m a R 2m R
where in the last step R << a was assumed. This state has the energy of a state with
k = a/π R >> 1 in the original 1-dimensional spectrum, so it is very high in energy. The
following figure illustrates schematically how the compactified dimension changes the
spectrum.
122
123 Beyond Standard Models
(a) (b)
V(x) y
1st new state
Square
wells
2πR
x x
0 a 0 a
Solution
0 a domain 2R
0 a
In (a) the particle is restricted to a line between 0 and a. In (b) with the added compactified
dimension the particle is restricted to the surface of a cylinder of length a and circumfer-
ence 2π R. If R << a, particle vibrations in the y direction are strongly restricted, which by
uncertainty principle arguments means that they represent states having very high energy
compared with those associated with motion in the x direction. Thus, if the radius of com-
pactification is small compared with the characteristic length scales of a physical problem,
the compactified dimensions have negligible influence on the low-energy spectrum.

Solution Manual Modern General Relativity by Guidry PDF

Uploaded by

Copyright:

Available Formats

Solution Manual Modern General Relativity by Guidry PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Solution Manual Modern General Relativity by Guidry PDF

Uploaded by

Copyright:

Available Formats

What are the main topics covered in the document?

What are the main topics covered in the document?

What is the difference between intrinsic and extrinsic curvature?

What is the difference between intrinsic and extrinsic curvature?

Instructor Solutions Manual: Modern General Relativity

where r = R has been used, implying that dr/d ϕ = 0.

while all non-diagonal components vanish. Thus the metric tensor is

gϕϕ = b2 gϕθ = gθ ϕ = 0 gθ θ = (a + b cos ϕ )2 .

x′ = x cos θ + y sin θ y′ = −x sin θ + y cos θ z′ = z,

which gives the transformed line element

which is Eq. (2.22).

3.1 For the three cases

where we have used

= gµν dxµ dxν .

But B′ µ is an arbitrary vector that does not generally vanish.

and the inverse transformations are

Computing all the derivatives and assembling them gives

and by explicit matrix multiplication, ÛU = I.

3.9 From Eqs. (3.45) and (3.46),

Aµν Bµν = −Aν µ Bµν (Aµν is antisymmetic)

Multiply both sides of this expression by ∂ xλ /∂ x′ µ to give

which may be substituted into the first equation to give

where only terms first-order in δ u have been retained.

= X α ∂α Aµν + Aαν ∂µ X α + Aµα ∂ν X α ,

The terms involving the connection coefficients cancel exactly:

where Eq. (3.19) was used.

(d) The tensor product gives

ds′ 2 = −c2 (dt ′ )2 + (dx′ )2 + (dy′ )2 + (dz′ )2

so it is invariant under the transformation.

4.2 Consider the following diagram:

Rest frame In motion with velocity v to the right

4.3 For infinitesimal displacements the Lorentz transformation (4.26) is

dt ′ = γ dt − (v/c2)dx dx′ = γ (dx − vdt) dy′ = dy dz′ = dz.

The velocity transformations are then obtained by evaluating derivatives of displacements

with respect to time:

where we define the matrices

right in unprimed x' x

−c2 (∆t)2 = (∆x′ )2 − (c∆t ′ )2 ,

c2 dt 2 = −ds2 + dx2 + dy2 + dz2 ,

dxµ /d τ ) implies the Euler–Lagrange equation (5.18),

128 2/3 G7/3 4/3 2  π 10/3

5.10 From the parameterized equations

Now expand gαβ (x′ ) in a Taylor series around gαβ (x),

where ∆xγ = ε K γ . Combining the preceding two equations gives

Object R(km) M(kg) ρ (g cm−3 ) g(m s−2 ) rc (km) R/rc

Earth 6378 6 × 1024 5.6 9.8 9.2 × 1012 6.9 × 10−10

6.8 (a) From Eq. (6.7),

7.1 This problem is patterned after an example in Cheng [64].

corresponding to metric components gi j = diag (1, r2 , 1) and gi j = g−1 −2

and from Eq. (5.18) the equations of motion are

case the geodesic equation (7.21) reduces to

Comparison with the geodesic equation (7.21) for x0 = t,

For any of the spatial coordinates the Euler–Lagrange equation is

α A' A' A' A'

B' B' B' B'

(a) (b) (c) (d)

7.8 By the chain rule d/d σ = (dxν /d σ )∂ν , where ∂ν ≡ ∂ /∂ xν . Therefore,

Γσλ σ = 21 gνσ (gσ ν ,λ + gλ ν ,σ − gσ λ ,ν )

gνσ gλ ν ,σ = gσ ν gλ ν ,σ = gνσ gλ σ ,ν = gνσ gσ λ ,ν .

128 2/3 G7/3 4/3 2 π 10/3