Problem sets - General Relativity

Solutions by Sergei Winitzki
Last modified: 2008-09-23
Contents
1 Coordinates and 1-forms 2
1.1 Invertible transformations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 Examples of coordinate transformations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Basis in tangent space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.4 Differentials of functions as 1-forms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.5 Basis in cotangent space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.6 Linearly independent 1-forms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.7 Transformation law for 1-forms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.8 Examples of transformations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.9 Supplementary problem sheet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2 Tensors 7
2.1 Definition of tensor product . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.3 Example of tensor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.5 Contraction of tensor indices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.8 Examples of spaces with a metric . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.9 Supplementary problem sheet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
3 The Christoffel symbol 10
3.1 Transformations 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
3.2 Transformations 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
3.3 Covariant derivatives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
3.4 The Leibnitz rule . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
3.5 Locally inertial reference frame . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
4 Geodesics and curvature 12
4.1 Geodesics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
4.1.1 First derivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
4.1.2 Second derivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
4.2 Commutator of covariant derivatives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
4.3 Parallel transport . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
4.4 Riemann tensor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
4.5 Lorentz transformations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
5 Gravitation theory applied 15
5.1 Redshift . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
5.2 Energy-momentum tensor 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
5.3 Energy-momentum tensor 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
5.4 Weak gravity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
5.5 Equations of motion from conservation law . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
6 The gravitational field 18
6.1 Degrees of freedom . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
6.2 Spherically symmetric spacetime . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
6.2.1 Straightforward solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
6.2.2 Solution using conformal transformation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
6.3 Equations of motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
1
7 Weak gravitational fields 23
7.1 Gravitational bending of light . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
7.2 Einstein tensor for weak field . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
7.3 Gravitational perturbations I . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
7.4 Gravitational perturbations II . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
8 Gravitational radiation I 27
8.1 Gauge invariant variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
8.2 Detecting gravitational waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
8.2.1 Using distances between particles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
8.2.2 Using geodesic deviation equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
8.3 Poisson equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
8.4 Metric perturbations 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
8.5 Metric perturbations 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
9 Gravitational radiation II 30
9.1 Projection of the matter tensor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
9.2 Matter sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
9.3 Energy-momentum tensor of gravitational waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
9.4 Power of emitted radiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
10 Derivation: gravitational waves in flat spacetime 34
A GNU Free Documentation License 35
A.0 Applicability and definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
A.1 Verbatim copying . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
A.2 Copying in quantity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
A.3 Modifications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
This is a complete set of solutions to the exercises offered in the lecture course “General Relativity” (Allgemeine
Relativit¨atstheorie) taught in the winter semester of 2006-2007 by Prof. V. F. Mukhanov at the Department of Physics,
Ludwig-Maximilians-University, in Munich, Germany. A good textbook corresponding to the level of this course: General
Relativity: An Introduction for Physicists by M. P. Hobson, G. P. Efstathiou, and A. N. Lasenby (Cambridge University
Press, 2006). Many of the exercises and all of the solutions were authored by Dr. Sergei Winitzki. This file contains only
the solutions; the problem statements are available separately.
This text is copyrighted by Sergei Winitzki (2008) and distributed under the GNU Free Documentation License. In
particular, it is permitted to print, copy, and distribute the entire text (or parts thereof) at no charge. The text may be
modified at will, and modified versions may be distributed again under the terms of the same license. The license also
stipulates that the entire machine-readable and human-editable source code for this text should be made available when
the text is distributed in any form. The full text of the license is in Appendix A. The full L
A
T
E
X source for this text is
attached inside the PDF file.
1 Coordinates and 1-forms
1.1 Invertible transformations
The inverse function theorem guarantees that the equations ξ
α
= ξ
α
(x) are solvable near a point x
0
if det
_
∂ξ
α
(x)/∂x
β
_
,=
0 at x
0
. Under this condition, the coordinate transformation is invertible at x
0
. Note: we are inverting not just one
function ξ = ξ(x), but we are determining x from a system of n equations, say ξ
α
(x) = C
α
, where C
α
are n given values.
1.2 Examples of coordinate transformations
1. a) Since x = u
_
1 +v
2
_
+u
3
/3, it is clear that x has range (−∞, +∞) for any fixed value of v as u varies in the range
(−∞, +∞). Similarly, y has the range (−∞, +∞). To verify that the coordinate system (x, y) covers the entire plane,
it is sufficient to show that x has the full range at every fixed value of y. It is sufficient to consider y
0
> 0 (else change
v → −v). At fixed y = y
0
> 0, we have y
0
= v + vu
2
+ v
3
/3 and thus the admissible values of u are from −∞ to +∞,
2
while the admissible values of v are from 0 to v = v
max
such that y
0
= v
max
+
1
3
v
3
max
. Then
u = ±
¸
y
0
−v −
1
3
v
3
v
(we have
y
0
−v −
1
3
v
3
v
≥ 0 for 0 < v < v
max
),
x = ±
_
1 +v
2
+
y
0
−v −
1
3
v
3
3v
_
¸
y
0
−v −
1
3
v
3
v
= ±
_
2
3
+
8
9
v
2
+
y
0
3v
_
¸
y
0
−v −
1
3
v
3
v
.
We have now expressed x as a function of v, i.e. x = x(v). When v varies from 0 to v
max
, x(v) varies from ±∞ to 0.
Since x(v) is nonsingular for v > 0, it follows that x has the full range. Therefore, the coordinates (x, y) cover the entire
two-dimensional plane.
b) The coordinate transformation is nonsingular if
det
∂ (x, y)
∂ (u, v)
,= 0.
Compute:
det
_
∂x
∂u
∂y
∂u
∂x
∂v
∂y
∂v
_
= det
_
1 +u
2
+v
2
2uv
2uv 1 +u
2
+v
2
_
(1)
= 1 + 2
_
u
2
+v
2
_
+
_
u
2
−v
2
_
2
> 0.
Therefore there are no singular points.
2. a) To determine the range, first consider φ = 0. Then x = r sinh θ, y = 0, z = r cosh θ. It is clear that z
2
−x
2
= r
2
.
Since r ≥ 0, the coordinates (x, y, z) cover only the domain [z[ > [x[. With arbitrary φ, it is clear that the coordinates
(x, y, z) cover the domain [z[ >
_
x
2
+y
2
.
b) Compute the determinant:
det
_
_
sinh θ cos φ r coshθ cos φ −r sinh θ sin φ
sinh θ sinφ r coshθ sin φ r sinh θ cos φ
cosh θ r sinh θ 0
_
_
= r
2
sinh θ.
The coordinates are singular if r = 0 or θ = 0.
c) The singularity at r = 0 is due to the fact that the set ¦r = 0, θ, φ¦ corresponds to a single point x = y = z = 0.
This is similar to the singularity of the spherical coordinates at r = 0. Points along the cone [z[ =
_
x
2
+y
2
are not
covered because they correspond to θ → ∞, r → 0. The singularity at θ = 0, r ,= 0 is due to the fact that the set
¦r, θ = 0, φ¦ corresponds to the point x = y = 0, z = r at fixed r ,= 0. This is similar to the polar coordinate singularity.
3. a) To determine the range, note that r sin θ ≥ 0 for the given range of θ and r. However, this is immaterial since the
factors cos φ and sin φ will make x, y cover the full range (−∞, +∞). The coordinates (x, y, z) are a slight modification
of the standard spherical coordinates. These coordinates cover the whole space (x, y, z).
b) Compute the determinant:
det
_
_
−sinθ cos φ −r cos θ cos φ r sin θ sin φ
−sinθ sin φ −r cos φsin φ −r sin θ cos φ
cos θ −r sin θ 0
_
_
= r
2
sin θ.
This is nonzero unless r = 0 or θ = 0.
c) The singularities are completely analogous to those in the spherical coordinates.
1.3 Basis in tangent space
Suppose that the vectors e
α
=

∂x
α
are linearly dependent, then there exist constants c
α
, not all zero, such that the vector
c
α
e
α
equals zero. Act with this vector on the coordinate function x
1
:
c
α
e
α
x
1
= c
α

∂x
α
x
1
= c
1
.
By assumption, c
α
e
α
= 0, therefore c
1
= 0. It follows that every c
α
equals zero, contradicting the assumption.
1.4 Differentials of functions as 1-forms
d(x) = dx, d
_
x
2
_
= 2xdx,
d (xy) = xdy +ydx, d (x +y) = dx +dy,
d
_
4x
2
y +x
3
z
_
=
_
8xy + 3x
2
z
_
dx + 4x
2
dy +x
3
dz, d
_
3
_
x
2
+y
2
_
= 3
xdx +ydy
_
x
2
+y
2
.
3
Now let us compute dh by first finding
d (arctan(x ±y)) =
dx ±dy
1 + (x ±y)
2
,
d
_
arctan
2x
x
2
−y
2
−1
_
=
1
1 +
4x
2
(x
2
−y
2
−1)
2
_
2dx
x
2
−y
2
−1

4x(xdx −ydy)
(x
2
−y
2
−1)
2
_
=
−2
_
x
2
+y
2
+ 1
_
dx + 4xydy
(x
2
−y
2
−1)
2
+ 4x
2
.
Adding these together and noting that
_
1 + (x +y)
2
__
1 + (x −y)
2
_
=
_
x
2
−y
2
−1
_
2
+ 4x
2
,
we get
dh(x, y) = d
_
arctan(x + y) + arctan(x −y) + arctan
2x
x
2
−y
2
−1
_
= 0.
This means that h(x, y) is a constant. By using the tangent sum rule, we can easily show that h(x, y) = 0.
1.5 Basis in cotangent space
Note that the relation
_
dx
α
,

∂x
β
_
= δ
α
β
is the definition of how the 1-form dx
α
acts on vectors ∂/∂x
β
. Now, it is clear that any 1-form is decomposed as a linear
combination of the 1-forms dx
1
, ..., dx
n
. It remains to show that all these forms are linearly independent. If this were
not so, there would exist a linear combination c
α
dx
α
= 0 such that not all c
α
= 0. Act with this on a vector ∂/∂x
1
and
obtain
0 =
_
0,

∂x
1
_
=
_
c
α
dx
α
,

∂x
1
_
= c
1
.
Therefore c
1
= 0. Similarly, we find that every other c
α
= 0, which contradicts the assumption.
1.6 Linearly independent 1-forms
1. Two 1-forms d(e
x
cos y), d(e
x
siny) are linearly independent for every x, y because
d (e
x
cos y) = e
x
cos ydx −e
x
sinydy,
d (e
x
sin y) = e
x
sinydx +e
x
cos ydy,
and the following determinant is always nonzero,
det
_
e
x
cos y −e
x
siny
e
x
siny e
x
cos y
_
= e
2x
,= 0.
2. Two 1-forms (1 +y)dx −2xydy, 8dx are linearly independent if the following determinant is nonzero,
det
_
1 +y −2xy
8 0
_
= 16xy.
This happens for xy ,= 0.
3. Three 1-forms dx +dy, dx +dz, dy +dz are always linearly independent.
4. Three 1-forms dx −dy, dy −dz, dz −dx are always linearly dependent (their sum is zero).
1.7 Transformation law for 1-forms
The transformation law for 1-forms,
d˜ x
α
=
∂˜ x
α
∂x
β
dx
β
,
under a coordinate transformation x
α
→ ˜ x
α
= ˜ x
α
(x
β
), is merely a different interpretation of the definition of the 1-form
d˜ x
α
(see Problem 1.4), where ˜ x
α
(x) is considered a scalar function in the coordinates
_
x
β
_
.
4
1.8 Examples of transformations
a) First compute dx and dy:
dx =
_
1 +u
2
+v
2
_
du + (2uv) dv,
dy = (2uv) du +
_
1 +u
2
+v
2
_
dv.
Then it is easy to compute xdx +ydy,etc. For instance,
d
1
x +y
= −
dx +dy
(x +y)
2
= −
(du +dv)
_
1 + (u +v)
2
_
(u +v)
2
_
1 +
1
3
(u +v)
2
_
2
.
b) The component transformation matrix is given in Eq. (1).
1.9 Supplementary problem sheet
1A Tangent plane
If the tangent plane is at angle α with the horizontal, then the acceleration is g sin α (from elementary mechanics). Since
0 ≤ α <
π
2
, we need to maximize α or, equivalently, tan α, which equals
∂z
∂r
=
rh
σ
2
e

1
2
r
2
σ
−2
, r ≡
_
x
2
+y
2
.
The maximum of ∂z/∂r is at r
0
= σ. For example, a point with maximum acceleration is x
0
= σ, y
0
= 0, z
0
= −he

1
2
.
The tangent plane at a point (x
0
, y
0
, z
0
) is given by the equation
n
x
(x −x
0
) +n
y
(y −y
0
) +n
z
(z − z
0
) = 0,
where (n
x
, n
y
, n
z
) are the components of the normal vector,
(n
x
, n
y
, n
z
) =
_
∂z
∂x
,
∂z
∂y
, 1

¸
¸
¸
x0,y0,z0
=
_
e

1
2
h
σ
, 0, 1
_
.
Therefore the equation of the tangent plane is
x
σ
e

1
2
+
z
h
= 0.
1B Induced metric
The induced metric is found by taking the expression for the bulk metric, g = dx
2
+dy
2
+dz
2
, and computing dx, dy, dz
through the forms dθ and dφ:
dx = 2a sinθ cos θ cos φdθ −a sin
2
θ sin φdφ,
dy = 2a sinθ cos θ sin φdθ +a sin
2
θ cos φdφ,
dz = 2a cos 2θdθ.
Therefore
g = dx
2
+dy
2
+dz
2
= a
2

2
+a
2
(sin θ)
4

2
.
The metric is degenerate at θ = 0 and θ =
π
2
. The singularities at these points are not merely coordinate singularities
that disappear when choosing a different coordinate system; but the reason is subtle.
To figure out the nature of these singularities, let us visualize the surface in a neighborhood of θ = 0. The y = 0
section of the surface corresponds to sin φ = 0, so φ = 0 or φ = π. Then
x = ±sin
2
θ, y = sin θ cos θ, x = ±
1 −cos 2θ
2
= ±
1 −

1 −4z
2
2
.
This is a union of two circles touching at x = z = 0. Hence, the surface is a torus with zero inner radius, i.e. intersecting
itself at x = y = z = 0. The rotational symmetry around the z axis leads to a “cusp” at θ = 0: the surface has a sharp
corner and the metric cannot be made smooth and nondegenerate by any choice of local coordinates. The situation near
θ =
π
2
is similar.
5
1C Embedded waves
The surface is defined by
x =
cos v

2 −sin u
, y =
sin v

2 −sin u
, z =
cos u

2 −sin u
.
1. For y = 0, we have v = 0 and thus
x =
1

2 −sin u
, z =
cos u

2 −sinu
.
To visualize this line in the (x, z) plane, we eliminate u from these equations and find
sin u =

2 −
1
x
,
_
x −

2
_
2
+z
2
= 1.
Therefore the line is a circle of radius 1 centered at
_
x =

2, z = 0
_
. This circle does not intersect the z axis since

2 > 1. Now we see that (x, y) is obtained from
_√
2 −sin u
_
−1
by multiplying with cos v and sin v. Therefore,
the full surface is a rotation surface, where we need to use the x coordinate as the radius. Therefore, the figure in
the (x, z) plane needs to be rotated around the z axis. The resulting surface is a torus. It may be described by the
equation
_
_
x
2
+y
2


2
_
2
+z
2
−1 = 0.
Note that
_
x
2
+y
2
=
1

2 −sin u
> 0.
Also
_
x
2
+y
2


2 =
1 −

2
_√
2 −sinu
_

2 −sin u
=

2 sin u −1

2 −sin u
.
2. Since the surface is now given by an equation of the form F(x, y, z) = 0, the normal vector (up to a constant factor
C) can be found as
(n
x
, n
y
, n
z
) = C
_
∂F
∂x
,
∂F
∂y
,
∂F
∂z
_
= C
_
2
_
x
2
+y
2


2
_
x
2
+y
2
x, 2
_
x
2
+y
2


2
_
x
2
+y
2
y, 2z
_
.
Multiplying by C ≡
1
2
_
x
2
+y
2
(this factor is chosen for simplicity), we have
(n
x
, n
y
, n
z
) =
_
x
_
_
x
2
+y
2


2
_
, y
_
_
x
2
+y
2


2
_
, z
_
x
2
+y
2
_
.
Expressed through the coordinates (u, v), this becomes
n
α
= (n
x
, n
y
, n
z
) =
_
cos v

2 sin u −1
_√
2 −sinu
_
2
, sin v

2 sinu −1
_√
2 −sin u
_
2
,
cos u
_√
2 −sin u
_
2
_
.
The equation of the tangent plane at point x
0
is
n
α
_
x
α
−x
α
(0)
_
= 0,
where n
α
must be computed at x = x
0
.
3. We compute
∂x
∂u
=
cos v cos u
_√
2 −sin u
_
2
,
∂x
∂v
= −
sin v

2 −sin u
,
∂y
∂u
=
sin v cos u
_√
2 −sin u
_
2
,
∂y
∂v
=
cos v

2 −sin u
,
∂z
∂u
=
1 −

2 sin u
_√
2 −sin u
_
2
,
∂z
∂v
= 0.
Now we can expand
dx =
∂x
∂u
du +
∂x
∂v
dv, etc.
6
Therefore the induced metric is
g = dx
2
+dy
2
+dz
2
=
_
_
_
cos u
_√
2 −sinu
_
2
_
2
+
_
1 −

2 sin u
_√
2 −sinu
_
2
_
2
_
_
du
2
+
dv
2
_√
2 −sin u
_
2
=
du
2
+dv
2
_√
2 −sin u
_
2
.
The vector V
a
= (cos v, sin v) is not a unit vector because
g(V, V ) =
cos
2
v + sin
2
v
_√
2 −sin u
_
2
=
1
_√
2 −sin u
_
2
,= 1.
The Cartesian components of the vectors ∂/∂u, ∂/∂v are found from

∂u
=
∂x(u, v)
∂u

∂x
+
∂y(u, v)
∂u

∂y
+
∂z(u, v)
∂u

∂z
,

∂v
=
∂x(u, v)
∂v

∂x
+
∂y(u, v)
∂v

∂y
+
∂z(u, v)
∂v

∂z
.
Therefore, the vector V
a
has the following Cartesian components,
V = cos v
_

sin v

2 −sin u

∂x
+
cos v

2 −sin u

∂y
_
+ sin v
_
cos v cos u
_√
2 −sin u
_
2

∂x
+
sin v cos u
_√
2 −sin u
_
2

∂y
+
1 −

2 sin u
_√
2 −sin u
_
2

∂z
_
= sinv cos v
_
sin u + cos u −

2
_√
2 −sinu
_
2
_

∂x
+
_
sin
2
v cos u +

2 cos
2
v −sin u cos
2
v
_√
2 −sin u
_
2
_

∂y
+ sin v
1 −

2 sin u
_√
2 −sin u
_
2

∂z
.
This vector is within the tangent plane because n
α
V
α
= 0,
n
α
V
α
= cos v

2 sin u −1
_√
2 −sinu
_
2
sin v cos v
_
sinu + cos u −

2
_√
2 −sin u
_
2
_
+ sin v

2 sinu −1
_√
2 −sin u
_
2
_
sin
2
v cos u +

2 cos
2
v −

2 sin u cos
2
v
_√
2 −sin u
_
2
_
+
cos u
_√
2 −sin u
_
2
sinv
1 −

2 sin u
_√
2 −sin u
_
2
(after simplification) = 0.
2 Tensors
2.1 Definition of tensor product
a),b) A direct calculation using the property
¸
dx
i
, ∂/∂x
k
_
= δ
i
k
gives:
¸ω
1
, v
1
¸ =
_
dx + 2ydy, 3

∂x
_
= 3, etc.
The results:
T(v
1
, v
1
) = 0, T(v
1
, v
2
) = 6x.
c) d) First, show that the sum of two linear functions is again a linear function: If A(x) and B(x) are linear functions,
i.e. if
A(x +λy) = A(x) +λA(y)
and likewise for B, then A+B obviously has the same property. Now, since a tensor is defined as a multi-linear function,
it is clear that tensors form a vector space.
7
2.3 Example of tensor
a) An obvious example of such T is the vector product, T(u, v) = uv, defined in three-dimensional space. To determine
the rank of T, we need to represent T as a multilinear number-valued function of some number of 1-forms and vectors,
e.g. A(f
1
, ..., f
r
, v
1
, ..., v
s
). It is clear that T(v
1
, v
2
) itself is not such a function because it has vector values instead of
scalar (number) values. So we need to add a 1-form to the list of arguments. We can define
A(f
1
, v
1
, v
2
) = ¸f
1
, T(v
1
, v
2

and then it’s clear that A is multilinear. Therefore T is a tensor of rank 1+2.
b) The calculation may go as follows. We need to determine the components T
α
βγ
such that
[T(u, v)]
α
= T
α
βγ
u
β
v
γ
.
So we rewrite the given definition of T(u, v) in the index notation, e.g. like this:
[T(u, v)]
α
= 2ε
α
βγ
u
β
v
γ
−u
α
n
β
v
β
.
Now we would like to move u
β
v
γ
out of the brackets and so determine T
α
βγ
. However, the expression above contains
u
α
v
β
instead of u
β
v
γ
. Therefore we rename the index β to γ and also introduce a Kronecker symbol δ
α
β
, so as to rewrite
identically
u
α
n
β
v
β
= u
β
v
γ
n
γ
δ
α
β
.
Therefore
[T(u, v)]
α
= 2ε
α
βγ
u
β
v
γ
−u
β
v
γ
n
γ
δ
α
β
=
_

α
βγ
−n
γ
δ
α
β
_
u
β
v
γ
,
T
α
βγ
= 2ε
α
βγ
−n
γ
δ
α
β
.
2.5 Contraction of tensor indices
a) Using the definition of a tensor as a multilinear function, it is easy to show that linear combinations of tensors are also
multilinear functions. Tensor products and contractions are also multilinear. The arguments are much simpler than the
proof of tensor transformation law for components.
b) Contracting two lower indices, e.g. T
ααβ
, gives components of a quantity which is not a tensor because these
components do not transform correctly under changes of basis. If T
ααβ
were a tensor it would transform as
˜
T
ααγ
=
∂x
λ
∂˜ x
γ
T
µµλ
.
However, this does not agree with the contraction of the tensor T
αβγ
, which transforms as
˜
T
αβγ
=
∂x
λ
∂˜ x
α
∂x
µ
∂˜ x
β
∂x
ν
∂˜ x
γ
T
λµν
.
The contraction over α = β yields

α
˜
T
ααγ
=

α
∂x
λ
∂˜ x
α
∂x
µ
∂˜ x
α
∂x
ν
∂˜ x
γ
T
λµν
,=
∂x
λ
∂˜ x
γ
T
µµλ
.
c) Calculation gives
T
α
αγ
= 2ε
α
αγ
−n
γ
δ
α
α
= −3n
γ
because ε
α
αγ
= 0 and δ
α
α
= 3.
2.8 Examples of spaces with a metric
a) We perform the calculation in components,
g(u, v) = u
α
v
α
− ε
αβγ
ε
α
λµ
u
β
n
γ
v
λ
n
µ
.
We would like to write g(u, v) = g
αβ
u
α
v
β
, where g
αβ
are the components of the metric tensor. Using the known identity
for the ε-symbol,
ε
αβγ
ε
α
λµ
= δ
βλ
δ
γµ
−δ
βµ
δ
γλ
,
we find
g(u, v) = u
α
v
α
−(δ
βλ
δ
γµ
−δ
βµ
δ
γλ
) u
β
n
γ
v
λ
n
µ
= u
α
v
α
−u
λ
v
λ
n
µ
n
µ
+u
λ
n
λ
v
µ
n
µ
.
8
Denote n
2
≡ n
µ
n
µ
≡ g(n, n), and then we need to relabel indices such that v
α
u
β
can be moved outside the parentheses.
The result is
g(u, v) =
_
δ
αβ
−n
2
δ
αβ
+n
α
n
β
_
v
α
u
β
.
Therefore
g
αβ
= δ
αβ
−n
2
δ
αβ
+n
α
n
β
.
To analyze the conditions under which det g
αβ
,= 0, we can choose an orthonormal basis such that n
α
is parallel to the
first basis vector. Then the components of the vector n in this basis are ([n[ , 0, 0) and the matrix g
αβ
has the following
simple form:
g
αβ
=
_
_
1 0 0
0 1 −n
2
0
0 0 1 −n
2
_
_
.
Then it is clear that det g
αβ
=
_
1 −n
2
_
2
. Therefore, the matrix g
αβ
is nondegenerate if n
2
,= 1.
b) Similar calculations give
g
αβ
=
_
3 +n
2
0
0 3
_
.
Therefore the determinant of g
αβ
is always nonzero.
c) Considerations are analogous to b), except that the size of the matrix g
αβ
is larger.
d) The metric is (dx)
2
+ (dy)
2
+ (dz)
2
, and we need to express dx, dy, dz through du and dv. A calculation gives
g = dx
2
+dy
2
+dz
2
= R
2
_
sinh
2
u + cosh
2
u
_
du
2
+R
2
cosh
2
u dv
2
.
2.9 Supplementary problem sheet
2A Vector equations
a) The equation contains two given vectors A
α
and B
α
. The solution x
α
can be found as a linear combination of A
α
,
B
α
, and the cross product ε
αβγ
A
β
B
γ
with unknown coefficients. Using vector notation, we have
x = αA+βB+ γ (AB) .
Substituting this expression into the given equation,
kx +x A = B,
and using the known identity
(AB) C = B(A C) −A(B C) , (2)
we find
A(kα −A B) +B(kβ + A A−1) + (A B) (kγ −β) = 0.
On purpose, we write this equation in the form of linear combination of the three vectors A, B, and A B. Since we
are considering the generic case, these three vectors are independent and so each of the coefficients above must be zero:
kα −A B = 0, kβ +A A−1 = 0, kγ −β = 0.
Solving this system of equations, we find (assuming k ,= 0 in the generic case)
α =
A B
k
, β =
1 −A A
k
, γ =
1 −A A
k
2
.
b) We have in vector notation
x A = B, x C = k.
Multiply C:
(x A) C = BC.
Simplify using the identity (2),
A(x C) −x(A C) = kA−x(A C) = BC.
Therefore
x =
kA−BC
A C
.
c), d) The equations have the form x
α
M
α
β
= A
β
, where M
α
β
is a matrix and A
β
is a known vector. The solution is
x = M
−1
A, where M
−1
is the inverse matrix (it exists in the generic case).
9
2B Tensor equations
a) Since the vectors A
α
and B
α
are a basis in two-dimensional space (they are nonzero and orthogonal), then the
symmetric tensor X
αβ
can be written generally as
X
αβ
= fA
α
A
β
+g
_
A
α
B
β
+A
β
A
α
_
+hB
α
B
β
,
where the coefficients f, g, h are unknown. It remains to determine these coefficients. Using A
α
B
α
= 0 and denoting
A
α
A
α
≡ [A[
2
, we get the system of equations
X
αβ
A
α
= B
β
⇒ f [A[
2
A
β
+g [A[
2
B
β
= B
β
,
X
αβ
δ
αβ
= 0 ⇒ f [A[
2
+h[B[
2
= 0.
The result is g = [A[
−2
, f = h = 0, so
X
αβ
=
A
α
B
β
+B
α
A
β
[A[
2
.
b) An antisymmetric tensor X
αβ
in three dimensions can be always expressed as
X
αβ
= ε
αβγ
u
γ
,
where u
γ
is an unknown vector that we need to determine. We can now rewrite the conditions on X
αβ
in a vector form,
X
αβ
A
α
= B
β
⇒ u A = B
X
αβ
B
α
= 0 ⇒ u B = 0
It follows that u is parallel to B and then the condition uA = B leaves the only solution u = 0, and therefore X
αβ
= 0
is the only admissible solution.
2C Degeneracy of the metric
a) The metric can be written in the basis ¦dx, dy¦ as the matrix
g =
_
y
2
1 +x
2
1 +x
2
0
_
.
The determinant of this matrix is −
_
1 +x
2
_
2
which is always nonzero.
b) In the basis where A
α
is parallel to the first basis vector, the vector A
α
has components (A, 0, 0, 0, ...) and therefore
the metric g
αβ
has the form
_
_
_
_
_
−r
2
0 0 ...
0 1 0
0 0 1
.
.
.
.
.
.
_
_
_
_
_
.
The metric is degenerate if r = 0 (i.e. at the origin).
3 The Christoffel symbol
3.1 Transformations 1
We are considering a flat space where Euclidean coordinates exist. Suppose ¦x
α
¦ and ¦˜ x
α
¦ are two coordinate systems,
while ¦ξ
α
¦ is the standard Euclidean coordinate system. The Christoffel symbols are defined as
Γ
µ
αβ
=

2
ξ
ν
∂x
α
∂x
β
∂x
µ
∂ξ
ν
,
˜
Γ
µ
αβ
=

2
ξ
ν
∂˜ x
α
∂˜ x
β
∂˜ x
µ
∂ξ
ν
.
The relationship between
˜
Γ and Γ can be found as follows. Assuming that the functions x
α
(˜ x) and also ˜ x
α
(x) are known,
we may express the partial derivative operators using the chain rule,

∂x
µ
=
∂˜ x
α
∂x
µ

∂˜ x
α
,

∂˜ x
µ
=
∂x
α
∂˜ x
µ

∂x
α
.
10
Also we can express
∂˜ x
µ
∂ξ
ν
=
∂˜ x
µ
∂x
λ
∂x
λ
∂ξ
ν
.
Therefore we can calculate
˜
Γ
µ
αβ
(when Γ
µ
αβ
is known) as follows,
˜
Γ
µ
αβ
=
∂˜ x
µ
∂x
λ
∂x
λ
∂ξ
ν
_
∂x
γ
∂˜ x
α

∂x
γ
__
∂x
δ
∂˜ x
β

∂x
δ
_
ξ
ν
=
∂˜ x
µ
∂x
λ
∂x
λ
∂ξ
ν
_
∂x
γ
∂˜ x
α
∂x
δ
∂˜ x
β

2
ξ
ν
∂x
γ
∂x
δ
+

2
x
δ
∂˜ x
β
∂˜ x
γ
∂ξ
ν
∂x
δ
_
=
∂˜ x
µ
∂x
λ
∂x
γ
∂˜ x
α
∂x
δ
∂˜ x
β
Γ
λ
γδ
+

2
x
δ
∂˜ x
β
∂˜ x
γ
∂˜ x
µ
∂x
λ
∂x
λ
∂ξ
ν
∂ξ
ν
∂x
δ
=
∂˜ x
µ
∂x
λ
∂x
γ
∂˜ x
α
∂x
δ
∂˜ x
β
Γ
λ
γδ
+

2
x
λ
∂˜ x
β
∂˜ x
γ
∂˜ x
µ
∂x
λ
. (3)
Note that the Euclidean coordinate system ξ
ν
is not needed to determine the transformation of Γ.
3.2 Transformations 2
Consider a vector field u
µ
. Assume that ∇
ν
u
µ
obeys the correct transformation law for rank (1,1) tensors,

ν
u
µ
=
_
˜

α
˜ u
β
_
∂x
µ
∂˜ x
β
∂˜ x
α
∂x
ν
,
and substitute

ν
u
µ


∂x
ν
u
µ
+ Γ
µ
αν
u
α
,
˜

ν
˜ u
µ


∂˜ x
ν
˜ u
µ
+
˜
Γ
µ
αν
˜ u
α
.
We can now express Γ through
˜
Γ. Note that
∂˜ x
α
∂x
β
∂x
β
∂˜ x
γ
= δ
α
γ
because the matrices are ∂˜ x
α
/∂x
β
and ∂x
β
/∂˜ x
γ
are inverse to each other. The result is
Γ
λ
αβ
=
˜
Γ
µ
νρ
∂˜ x
ν
∂x
α
∂˜ x
ρ
∂x
β
∂x
λ
∂˜ x
µ
+

2
x
λ
∂˜ x
ν
∂˜ x
ρ
∂˜ x
ν
∂x
α
∂˜ x
ρ
∂x
β
.
3.3 Covariant derivatives
The rule is that every upper index gets a +Γ and every lower index gets a −Γ. Each term with Γ replaces one index in
the original tensor by one of the indices in Γ. Therefore we can write the answer as
T
αβ
γδµ;ν
= ∂
ν
T
αβ
γδµ
+ Γ
α
λν
T
λβ
γδµ
+ Γ
β
λν
T
αλ
γδµ
−Γ
λ
γν
T
αβ
λδµ
−Γ
λ
δν
T
αβ
γλµ
−Γ
λ
µν
T
αβ
γδλ
3.4 The Leibnitz rule
Perform an explicit calculation,
A
α;γ
B
β
+A
α
B
β

= B
β
_

γ
A
α
−Γ
λ
αγ
A
λ
_
+A
α
_

γ
B
β
+ Γ
β
λγ
B
λ
_
;
_
A
α
B
β
_

= ∂
γ
_
A
α
B
β
_
−Γ
λ
αγ
A
α
B
β
+ Γ
β
λγ
A
α
B
λ
.
This proves the required property.
3.5 Locally inertial reference frame
In this problem (unlike problem 3.1) the coordinate system ¦ξ
α
¦ is not a flat Euclidean coordinate system, but it is just
a coordinate system which is like Euclidean at one point. Now we want to use the formula (3), which will enable us to
compute the Christoffel symbol
˜
Γ in the coordinate system ξ, given the Christoffel symbol Γ in the original coordinate
system ¦x
α
¦. To use that formula, we need to compute some derivatives. Denoting ¦˜ x
α
¦ ≡ ¦ξ
α
¦, we find
∂˜ x
α
∂x
β
= δ
α
β
+
_
x
µ
−x
µ
(0)
_
δ
ν
β
Γ
α
(0)µν
= δ
α
β
+
_
x
µ
−x
µ
(0)
_
Γ
α
(0)µβ
.
11
The inverse derivative, ∂x
α
/∂˜ x
β
, can be found by inverting this matrix; the result can be found simply by assuming that
∂x
α
∂˜ x
β
= δ
α
β
+
_
x
µ
−x
µ
(0)
_
A
α
µβ
+O
_
(x −x
(0)
)
2
_
,
where A
α
µβ
is an unknown matrix. So up to quadratic terms we find
∂x
α
∂˜ x
β
= δ
α
β

_
x
µ
−x
µ
(0)
_
Γ
α
(0)µβ
+O
_
(x −x
(0)
)
2
_
.
Therefore

2
x
λ
∂˜ x
β
∂˜ x
γ
= −Γ
λ
(0)βγ
+O
_
x −x
(0)
_
.
Finally, we find
˜
Γ
µ
αβ
=
∂˜ x
µ
∂x
λ
∂x
γ
∂˜ x
α
∂x
δ
∂˜ x
β
Γ
λ
γδ
+

2
x
λ
∂˜ x
β
∂˜ x
γ
∂˜ x
µ
∂x
λ
= Γ
µ
αβ
−Γ
µ
(0)αβ
+O
_
x −x
(0)
_
.
At x = x
0
we have Γ
µ
αβ
= Γ
µ
(0)αβ
. It follows that the new Christoffel symbol is equal to zero at x = x
0
.
Alternatively, one can use the transformation law for the Christoffel symbol in the inverse direction, Γ =
˜
Γ... + ...,
i.e. one denotes ¦x
α
¦ ≡ ¦ξ
α
¦, ¦˜ x
α
¦ ≡ ¦x
α
¦. This has the advantage that only derivatives ∂
2
ξ/∂x∂x need to be computed,
and not the derivatives ∂
2
x/∂ξ∂ξ. Since all derivatives only need to be evaluated at x = x
0
, the first-order derivatives
∂x/∂ξ at x = x
0
can be found as the inverse matrix to ∂ξ
α
/∂x
β
= δ
α
β
, i.e. ∂x
α
/∂ξ
β
= δ
α
β
. This considerably simplifies
the calculations.
4 Geodesics and curvature
4.1 Geodesics
(a) Note that d/ds is the ordinary (not “covariant”) derivative in the direction of u
α
. The geodesic equation can be
rewritten for the 1-form u
α
as
u
γ
u
α;γ
= 0 =
du
α
ds
−Γ
β
αγ
u
β
u
γ
.
An explicit formula for Γ
β
αγ
yields
Γ
β
αγ
u
β
u
γ
= Γ
β αγ
u
β
u
γ
=
1
2
(g
βα,γ
+g
βγ,α
−g
αγ,β
) u
β
u
γ
.
Note that g
βα,γ
− g
αγ,β
is antisymmetric in (β ↔γ). Therefore these terms will cancel after a contraction with u
β
u
γ
.
The remaining term yields
Γ
β
αγ
u
β
u
γ
=
1
2
g
αβ,γ
u
β
u
γ
.
b) We give two derivations; the first one is direct and the second one uses the property (a).
4.1.1 First derivation
Note that
d
ds
_
g
αβ
u
α
u
β
_
= u
γ
_
g
αβ
u
α
u
β
_

where we must use an ordinary derivative instead of the covariant derivative (according to the definition of d/ds). So we
find
u
γ
_
g
αβ
u
α
u
β
_

= g
αβ,γ
u
α
u
β
u
γ
+ 2g
αβ
u
α
u
γ
u
β

.
Now we need to simplify an expression containing u
β

. By assumption, the derivative of the vector field u
α
satisfies
d
ds
u
α
+ Γ
α
βγ
u
β
u
γ
= u
γ
u
α

+ Γ
α
βγ
u
β
u
γ
= 0
= u
γ
u
α

+
1
2
g
αλ
(g
λµ,γ
+g
λγ,µ
−g
µγ,λ
) u
γ
u
µ
.
Therefore
u
α
u
γ
u
α

= −
1
2
u
α
g
αλ
(g
λµ,γ
+g
λγ,µ
−g
µγ,λ
) u
γ
u
µ
= −
1
2
u
λ
(g
λµ,γ
+g
λγ,µ
−g
µγ,λ
) u
γ
u
µ
= −
1
2
g
λµ,γ
u
λ
u
µ
u
γ
.
What remains is a straightforward computation:
u
γ
_
g
αβ
u
α
u
β
_

= g
αβ,γ
u
α
u
β
u
γ
+ 2g
αβ
u
α
u
γ
u
β

= g
αβ,γ
u
α
u
β
u
γ
−g
λµ,γ
u
λ
u
µ
u
γ
= 0.
12
4.1.2 Second derivation
We write
d
ds
_
g
αβ
u
α
u
β
_
=
d
ds
_
g
αβ
u
α
u
β
_
= u
γ
g
αβ

u
α
u
β
+ 2g
αβ
u
α,γ
u
β
u
γ
.
Now we need to express u
α,γ
u
γ
≡ du
α
/ds. To do that, we use the property derived in (a),
du
α
ds
=
1
2
g
βγ,α
u
β
u
γ
,
and find
d
ds
_
g
αβ
u
α
u
β
_
=
dg
αβ
ds
u
α
u
β
+g
αβ
g
λγ,α
u
λ
u
β
u
γ
=
dg
αβ
ds
u
α
u
β
+
dg
λγ
ds
u
λ
u
γ
.
It remains to express the derivative of g
αβ
through the derivative of g
αβ
. We use the identity
d
ds
_
g
αβ
g
βγ
_
=
d
ds
_
δ
α
γ
_
= 0,
thus
dg
αβ
ds
g
βγ
= −
dg
βγ
ds
g
αβ

dg
αβ
ds
= −
dg
λµ
ds
g
αλ
g
βµ
.
Therefore
dg
αβ
ds
u
α
u
β
= −
dg
λµ
ds
g
αλ
g
βµ
u
α
u
β
= −
dg
λµ
ds
u
λ
u
µ
,
and thus
d
ds
_
g
αβ
u
α
u
β
_
= 0.
4.2 Commutator of covariant derivatives
First compute
u
α

= u
α

+ Γ
α
βλ
u
λ
,
u
α
;βγ
=
_
u
α

+ Γ
α
βλ
u
λ
_

+ Γ
α
µγ
_
u
µ

+ Γ
µ
βλ
u
λ
_
−Γ
µ
βγ
_
u
α

+ Γ
α
µλ
u
λ
_
= u
α
,βγ
+ Γ
α
βλ,γ
u
λ
+ Γ
α
βλ
u
λ

+ Γ
α
µγ
u
µ

+ Γ
α
µγ
Γ
µ
βλ
u
λ
−Γ
µ
βγ
u
α

.
Since we want to compute the commutator u
α
;βγ
− u
α
;γβ
, we can omit the terms that are symmetric in (β ↔γ). These
terms are the following:
u
α
,βγ
, Γ
α
βλ
u
λ

+ Γ
α
µγ
u
µ

, Γ
µ
βγ
u
α

.
The remaining terms are
u
α
;βγ
= Γ
α
βλ,γ
u
λ
+ Γ
α
µγ
Γ
µ
βλ
u
λ
+ (symmetric in β ↔γ) ,
which yields
u
α
;βγ
−u
α
;γβ
= u
λ
_
Γ
α
λβ,γ
−Γ
α
λγ,β
+ Γ
µ
λβ
Γ
α
γµ
−Γ
µ
λγ
Γ
α
βµ
_
= u
λ
R
α
λγβ
.
4.3 Parallel transport
The parallel-transported vector can be represented by a 1-form A
α
(s) such that
dA
α
ds
−Γ
β
αg
A
β
u
γ
= 0, u
γ

dx
γ
ds
.
However, the closed curve is assumed to cover only a very small neighborhood of one point x
0
, so we can approximate
A
α
by a constant, A
α
(x
0
), along the curve. Therefore
δA
α
=
_
dA
α
ds
ds =
_
Γ
β
αγ
(x)A
β
(x)dx
γ
≈ A
β
(x
0
)
_
Γ
β
αγ
(x)dx
γ
.
Now, in a locally inertial system at x
0
we have Γ
β
αγ
(x
0
) = 0. Therefore we can Taylor expand Γ
β
αγ
(x) near x
0
as
Γ
β
αγ
(x) =
_
x
λ
−x
λ
0
_
Γ
β
αγ,λ
+O
_
(x −x
0
)
2
_
.
Therefore
δA
α
≈ A
β
(x
0
)
_
Γ
β
αγ,λ
_
x
λ
−x
λ
0
_
dx
γ
≈ A
β
(x
0

β
αγ,λ
(x
0
)
_
x
λ
dx
γ
,
where we have again approximated Γ
β
αγ,λ
(x) by its value at x = x
0
, and also used the identity
_
dx
γ
= 0. Further,
_
d
_
x
γ
x
λ
_
= 0 =
_
x
γ
dx
λ
+
_
x
λ
dx
γ
.
13
Therefore we may rewrite
δA
α
≈ A
β
Γ
β
αγ,λ
_
x
λ
dx
γ
=
1
2
A
β
_
Γ
β
αγ,λ
−Γ
β
αλ,γ
_
_
x
λ
dx
γ
=
1
2
A
β
R
β
αλγ
_
x
λ
dx
γ
.
4.4 Riemann tensor
a) It is more convenient to consider the fully covariant tensor R
αβγδ
. This tensor has the following symmetries,
R
αβγδ
= −R
βαγδ
= −R
αβδγ
, (4)
R
αβγδ
+R
βγαδ
+R
γαβδ
= 0, (5)
R
αβγδ
= R
γδαβ
. (6)
However, it is known that the property (6) follows from (4)-(5), therefore it is sufficient to consider these two properties
[note that (5) does not follow from (4), (6)]. Let us first consider the property (4). For fixed γ, δ, we have that R
αβγδ
is an antisymmetric n n matrix (indices α, β). This matrix has
1
2
n(n −1) independent components. Likewise for fixed
α, β. Therefore, the number of independent components of R
αβγδ
is reduced to
N
1
=
_
1
2
n(n −1)
_
2
.
Now we use the property (5). Let us see whether the property (5) is nontrivial at fixed δ. If α = γ, then the property (5)
becomes
R
αβαδ
+R
βααδ
+R
ααβδ
= 0 (no summation),
which is already a consequence of (4). Likewise for β = γ or for α = β. Therefore, the property (5) is a new constraint
only if all three indices α, β, γ are different (i.e. α ,= β, α ,= γ, β ,= γ). Suppose that α, β, γ are different. There are
N
2
=
1
6
n(n −1)(n −2)
choices of such α, β, γ. Therefore, for each δ = 1, ..., n we obtain N
2
additional constraints. Finally, let us check that
every such constraint is nontrivial for every δ (even if δ is equal to one of α, β, γ). Suppose δ = α, then (5) becomes
R
αβγα
+R
βγαα
+R
γαβα
= 0 (no summation).
This is a nontrivial constraint (equivalent to R
αβαγ
= R
αγαβ
). Therefore, the number of constraints is nN
2
, and thus
the total number of independent components of R
αβγδ
is
N = N1 −nN
2
=
n
2
_
n
2
−1
_
12
.
b) Weinberg, Chapter 6, ¸8
c) There is only one independent component of R
αβγδ
in two dimensions. For instance, we can choose R
1212
as the
independent parameter. Then we can express the Ricci tensor as
R
αβ
= g
λµ
R
λαµβ
.
Calculating component by component, we find
R
11
= g
22
R
1212
, R
12
= −g
12
R
1212
, R
22
= g
11
R
1212
,
R = g
αβ
R
αβ
=
_
2g
11
g
22
−2g
12
g
21
_
R
1212
= 2gR
1212
.
Note that the matrix
_
g
22
−g
12
−g
12
g
11
_
is equal to the inverse matrix to g
αβ
(which is g
αβ
), multiplied by the determinant det g
αβ
; since det g
αβ
= 1/g, we have
_
g
22
−g
12
−g
12
g
11
_
= gg
αβ
.
Therefore
G
αβ
= R
αβ

1
2
g
αβ
R = gg
αβ
R
1212

1
2
g
αβ
2gR
1212
= 0.
14
4.5 Lorentz transformations
A Lorentz transformation is represented by a matrix Λ
α
β
such that
Λ
α
λ
Λ
β
µ
g
αβ
= g
λµ
.
Consider an infinitesimal Lorentz transformation,
Λ
α
λ
= δ
α
λ
+εH
α
λ
.
The number of parameters in Lorentz transformations is the same as the number of parameters in H
α
β
. The condition
for H
α
β
is

α
λ
+εH
α
λ
)
_
δ
β
µ
+εH
β
µ
_
g
αβ
= g
λµ
.
Disregarding terms of order ε
2
, we find
0 = δ
β
µ
H
α
λ
g
αβ

α
λ
H
β
µ
g
αβ
= H
µλ
+H
λµ
.
Therefore, H
λµ
is an antisymmetric n n matrix, which has
1
2
n(n − 1) independent components. For n = 4 we get 6
components. These can be interpreted as three spatial rotations and three Lorentz rotations (boosts).
5 Gravitation theory applied
5.1 Redshift
In the weak field limit, the Newtonian gravitational potential near a mass M is
Φ =
GM
r
,
while the component g
00
of the metric is
g
00
= 1 +

c
2
.
(We write the units explicitly.) Therefore the redshift factor z(r) at distance r from the center of the Earth is
z(r) =
_
1 +
2GM
c
2
r
≈ 1 +
GM
c
2
r
.
To compare the redshift factors at the surface of the Earth, denote by R
E
the radius of the Earth. We know that the
gravitational acceleration at the surface is
g
E
=
GM
R
2
E
≈ 9.81
m
s
2
.
Therefore, it is convenient to express GM = g
E
R
2
E
. For a vertical distance L between sender and receiver, we find
z(R
E
)
z(R
E
+L)
=
1 +g
E
R
E
c
−2
1 +g
E
R
E
c
−2
RE
RE+L
≈ 1 +
g
E
R
E
c
2
_
1 −
R
E
R
E
+L
_
= 1 +
g
E
R
E
L
c
2
(R
E
+L)
.
Since in our problem L ≪R
E
, we may approximate
z(R
E
)
z(R
E
+L)
≈ 1 +
g
E
L
c
2
≈ 1 + 1.1
L
10
16
m
.
5.2 Energy-momentum tensor 1
In the nonrelativistic limit, we may disregard gravitation; g
αβ
= η
αβ
. The EMT of an ideal fluid is
T
αβ
= −pη
αβ
+ (p +ρ) u
α
u
β
,
where u
α
is the 4-velocity vector of the fluid motion. In the nonrelativistic limit, u
α
≈ (1, v), where v is the 3-vector of
velocity and [v[ ≪1 in the units where c = 1.
The conservation law is
0 = T
αβ

= −p

+ (p +ρ)

u
α
u
β
+ (p +ρ) u
α
u
β

+ (p +ρ) u
α

u
β
.
Let us simplify this expression by introducing the time derivative along the fluid flow,
d
dt
≡ u
α

α
.
15
Then we find
0 = −p

+u
α
_
˙ p + ˙ ρ + (p +ρ) u
β

_
+ (p +ρ) ˙ u
α
. (7)
Contracting with u
α
and using u
α
˙ u
α
= 0, we find
˙ ρ + (p +ρ) u
β

= 0. (8)
This is the relativistic continuity equation. Using this equation, we find from Eq. (7) that
0 = −p

+u
α
˙ p + (p +ρ) ˙ u
α
. (9)
Now let us apply the nonrelativistic limit, u
α
≈ (1, v), to Eqs. (8) and (9). In our notation, for any quantity X we
have
˙
X ≡
d
dt
X ≡

∂t
X + (v ∇) X.
The continuity equation (8) gives

dt
+ (p +ρ) divv = 0.
This is the ordinary, nonrelativistic continuity equation.
1
Finally, Eq. (9) gives

∇p +v ˙ p + (p +ρ)
˙
v = 0.
(Note that p
,j
= −∇
j
p.) This is the Euler equation,
dv
dt
=
1
p +ρ
_

∇p −v
dp
dt
_
.
5.3 Energy-momentum tensor 2
Compute the covariant derivative,
T
αβ

=
_
Φ

Φ


1
2
g
αβ
Φ

Φ

_

= Φ


Φ

+ Φ

Φ


−g
αβ
Φ

Φ
;λα
= Φ


Φ

.
Here we used Φ
;αβ
= Φ
;βα
which follows from Γ
λ
αβ
= Γ
λ
βα
(note that Φ is a scalar; covariant derivatives do not commute
when applied to vectors!) and also the property
Φ


X
α
= Φ
;αβ
X
α
,
which is due to g
αβ;µ
= 0. Therefore, we get
Φ


Φ

= 0;
this entails Φ


= 0 (since Φ

= 0 is a weaker condition than Φ


= 0, i.e. if Φ

= 0 then also Φ


= 0, so it is
sufficient to write the latter).
5.4 Weak gravity
A very short solution is to write R
00
directly through Γ and note that only Γ
α
00,α
comes in (if we disregard terms of
second order). Then compute Γ
α
00
explicitly through Φ. (Assume that g
αβ,0
= 0.) We may disregard terms of order ΓΓ
because Γ is of order Φ, and also we may raise and lower indices using η
µν
instead of g
µν
. (This is somewhat heuristic;
see below.) The calculation goes like this:
R
00
= Γ
α
00,α
−Γ
α
0α,0
+O( Γ Γ),
Γ
α
00
=
1
2
η
αβ
(g
0β,0
+g
β0,0
−g
00,β
) = −η
αβ
Φ

,
therefore (using Φ
,0
= 0)
R
00
= −η
αβ
Φ
,αβ
= −Φ = −Φ
,00
+ Φ
,11
+ Φ
,22
+ Φ
,33
= ∆Φ.
Here is another, somewhat more comprehensive solution. In the weak field limit, we write
g
µν
= η
µν
+h
µν
.
1
Note that in the usual, nonrelativistic continuity equations as they are written in most books, there is no p + ρ - just ρ. This is so
because in most cases the matter is nonrelativistic, so p ≪ρ and p +ρ ≈ ρ. This is, however, not true for relativistic matter, such as photons
(electromagnetic radiation) for which p =
1
3
ρ.
16
Then we only compute everything up to first order in h
µν
. Therefore, we may raise and lower indices using the Minkowski
metric η
αβ
instead of g
αβ
.
Note: the Newtonian limit does not determine the components of g
µν
except for g
00
= 1 +2Φ. The actual metric g
µν
is equal to η
µν
plus a small first-order deviation, h
µν
, but this deviation cannot be expressed just through the Newtonian
potential Φ ≡
1
2
h
00
. In principle, one needs to solve the full Einstein equations to find h
µν
; in other words, one needs to
determine other, “post-Newtonian potentials” and not just the Newtonian potential Φ. However, when one only wants
to compute effects of gravitation on motion of slow bodies, only g
00
is necessary. So it is sufficient to compute just the
Newtonian potential Φ. But e.g. trajectories of light rays cannot be computed accurately in the Newtonian limit (because
light does not move slowly). To compute trajectories of light rays, one needs all components of h
µν
, not just h
00
.
Let us do the computation through h
µν
in a more general way. First we compute the Christoffel symbol and the Ricci
tensor:
Γ
λ
αν
=
1
2
η
λγ
(h
αγ,ν
+h
νγ,α
−h
αν,γ
) ; ⇒ Γ ∼ O(Φ);
R
αβ
= R
λ
αλβ
= Γ
λ
αβ,λ
−Γ
λ
αλ,β
+ ΓΓ −ΓΓ ≈ Γ
λ
αβ,λ
−Γ
λ
αλ,β
.
(We may disregard the ΓΓ terms since they are second order in Φ.) Now we compute (again up to first order in Φ)
Γ
λ
αλ,β
=
1
2
η
λγ
(h
αγ,λβ
+h
λγ,αβ
−h
αλ,γβ
) =
1
2
η
λγ
h
λγ,αβ
,
R
αβ
=
1
2
η
λγ
(h
αγ,βλ
+h
βγ,αλ
−h
αβ,γλ
−h
λγ,αβ
) .
Let us now compute just the component R
00
, recalling that h
µν
is time-independent (so h
µν,0
= 0) and h
00
= 2Φ:
R
00
=
1
2
η
λγ
(h
0γ,0λ
+h
0γ,0λ
− h
00,γλ
−h
λγ,00
) = −
1
2
η
λγ
h
00,γλ
= −η
λγ
Φ
,λγ
= ∆Φ.
5.5 Equations of motion from conservation law
We would like to rewrite the covariant conservation law T
µν

= 0 through ordinary derivatives. The given relations are
useful; let’s derive them first.

∂x
ν

−g = −
1
2

−g
_

∂x
ν
g
_
= −
1
2

−g
_
gg
αβ
g
αβ,ν
_
=
1
2

−gg
αβ
g
αβ,ν
;
Γ
µ
µν
=
1
2
g
µα
(g
µα,ν
+g
να,µ
−g
νµ,α
) =
1
2
g
µα
g
µα,ν
=
1

−g

∂x
ν

−g.
Now rewrite the covariant derivative of T
µν
explicitly:
T
µν

= T
µν

+ Γ
µ
αµ
T
αν
+ Γ
ν
αµ
T
µα
=
1

−g
_√
−gT
µν
_

+ Γ
ν
αβ
T
αβ
.
Apply this to the given T
µν
:
0 = T
µν

=
m
0

−g
__
ds
dx
µ
ds
dx
ν
ds
δ
(4)
(x
σ
−x
σ
(s))
_

+
m
0

−g
_
ds
dx
α
ds
dx
β
ds
Γ
ν
αβ
(x)δ
(4)
(x
σ
−x
σ
(s)).
Since in the first term the dependence on x
σ
is only through δ
(4)
, we can rewrite
dx
µ
ds

∂x
µ
_
δ
(4)
(x
α
−x
α
(s))
_
= −
d
ds
_
δ
(4)
(x
α
−x
α
(s))
_
(this is easily understood if read from right to left) and then integrate by parts,
_
ds
dx
ν
ds
d
ds
δ
(4)
(x
σ
−x
σ
(s)) = −
_
ds
d
2
x
ν
ds
2
δ
(4)
(x
σ
−x
σ
(s)).
Finally,
0 =

−g
m
0
T
µν

=
_
ds
_
d
2
x
ν
ds
2
+ Γ
ν
αβ
dx
α
ds
dx
β
ds
_
δ
(4)
(x
σ
−x
σ
(s)).
This is a function of x
σ
which should equal zero everywhere. Therefore, the integrand should vanish for every value of s,
d
2
x
ν
ds
2
+ Γ
ν
αβ
dx
α
ds
dx
β
ds
= 0.
17
Remark: in general, equations of motion do not follow from conservation law, but they do follow if there is only one
field. (e.g. one fluid, or one scalar field, or some number of point particles). The situation in ordinary mechanics is similar:
e.g. the equation of motion for a particle follow from the conservation of energy only if the motion is in one dimension:
E =
mv
2
2
+V (x) = const,
0 =
dE
dt
= (m˙ v +V

(x)) v ⇒m˙ v = −V

(x).
However, equations of motion do not follow from conservation of energy if there is more than one degree of freedom.
Similarly, equations of motion for say two scalar fields Φ, Ψ do not follow from the conservation of their combined T
µν
.
These fields have two different equations of motion, and one cannot hope to derive them from a single conservation law.
6 The gravitational field
6.1 Degrees of freedom
The electromagnetic field is described by a 4-vector potential A
µ
(x). This would give 4 degrees of freedom. However,
there is also a gauge symmetry,
A
µ
→A
µ


,
where φ(x) is an arbitrary function of spacetime. Using this gauge symmetry, we may e.g. set the component A
0
(x) = 0.
Then only three functions of spacetime (A
1
, A
2
, A
3
) are left. Hence the electromagnetic field has 3 degrees of freedom.
There are additional gauge symmetries involving functions φ(x) that do not depend on time. Since these functions φ(x)
are functions only of three arguments, they do not change the number of degrees of freedom.
6.2 Spherically symmetric spacetime
6.2.1 Straightforward solution
A direct computation listing all the possible Christoffel symbols and components of the Ricci tensor is certainly straight-
forward but very long. Here is a way to compute the curvature tensor without writing individual components. Since the
metric has a diagonal form, let us denote
g
αβ
= η
αβ
A
α
, g
αβ
= η
αβ
1
A
α
(no summation!), (10)
where
A
α

_
e
N
, e
L
, r
2
, r
2
sin
2
θ
_
(11)
is a fixed array of four functions. For this calculation, we do not use the Einstein summation convention any more; every
summation will be written explicitly. However, we make heavy use of the fact that η
αβ
,= 0 only if α = β, and that
η
λλ
= η
λλ
. At the end of the calculation of the Ricci tensor R
αβ
, we shall substitute the known functions A
α
and use
the resulting simplifications.
We begin with the calculation of the Christoffel symbols,
Γ
λ
αβ
=

µ
1
2
η
λµ
1
A
λ

αµ
A
µ,β

βµ
A
µ,α
−η
αβ
A
α,µ
]
=
1
2
δ
λ
α
A
λ,β
A
λ
+
1
2
δ
λ
β
A
λ,α
A
λ

1
2
η
λλ
η
αβ
A
α,λ
A
λ
. (12)
Note that the summation over µ results in setting λ = µ due to η
λµ
, and that we have relations such as η
λλ
η
αλ
= δ
λ
α
and
δ
β
α
η
αα
= η
αβ
, which hold without summation. For convenience, we rewrite Eq. (12) as
Γ
λ
αβ
=
1
2
_
δ
λ
α
B
λ,β

λ
β
B
λ,α
−η
λλ
η
αβ
A
α
A
λ
B
α,λ
_
,
where we defined the auxiliary function
B
α
≡ ln A
α
.
As a check, we compute the “trace” of the Christoffel symbols and compare with the known formula,

λ
Γ
λ
αλ
=
1
2
_
B
α,α
+

λ
B
λ,α

A
α
A
α
B
α,α
_
=
1
2

λ
B
λ,α
=
_
ln

−g
_

.
Let us also denote for brevity
C ≡ ln

−g =
1
2

λ
B
λ
;

λ
Γ
λ
αλ
= C

.
18
We proceed to the computation of the Ricci tensor. We use the formula (with Landau-Lifshitz sign conventions)
R
αβ
=

λ
_
Γ
λ
αβ,λ
−Γ
λ
αλ,β
_
+

λ,µ
_
Γ
µ
λµ
Γ
λ
αβ
−Γ
λ
αµ
Γ
µ
βλ
_
.
We now compute the necessary terms:

λ
Γ
λ
αβ,λ
=
1
2
B
α,αβ
+
1
2
B
β,αβ

1
2
η
αβ

λ
η
λλ
_
A
α
A
λ
B
α,λ
_

,

λ
Γ
λ
αλ,β
= C
,αβ
,

λ,µ
Γ
µ
λµ
Γ
λ
αβ
=

λ
C

1
2
_
δ
λ
α
B
λ,β
+
1
2
δ
λ
β
B
λ,α

1
2
η
λλ
η
αβ
A
α
A
λ
B
α,λ
_
=
1
2
_
C

B
α,β
+C

B
β,α
−η
αβ

λ
C

η
λλ
A
α
A
λ
B
α,λ
_
,

λ,µ
Γ
λ
αµ
Γ
µ
λβ
=
1
4

λ,µ
_
δ
λ
α
B
λ,µ

λ
µ
B
λ,α
−η
λλ
η
αµ
A
α
A
λ
B
α,λ
_ _
δ
µ
λ
B
µ,β

µ
β
B
µ,λ
−η
µµ
η
λβ
A
λ
A
µ
B
λ,µ
_
=
1
4

µ
B
α,µ
_
δ
µ
α
B
µ,β

µ
β
B
µ,α
−η
µµ
η
αβ
A
α
A
µ
B
α,µ
_
(here set λ = α)
+
1
4

µ
B
µ,α
_
B
µ,β

µ
β
B
µ,µ
−η
µµ
η
µβ
A
µ
A
µ
B
µ,µ
_
(here set λ = µ)

1
4

λ
η
λλ
η
αα
A
α
A
λ
B
α,λ
_
δ
α
λ
B
α,β

α
β
B
α,λ
−η
αα
η
λβ
A
λ
A
α
B
λ,α
_
(here set µ = α)
=
1
4
B
α,α
B
α,β
+
1
4
B
α,β
B
β,α

1
4
η
αβ

µ
η
µµ
A
α
A
µ
B
α,µ
B
α,µ
+
1
4

µ
B
µ,α
B
µ,β
+
1
4
B
β,α
B
β,β

1
4
B
β,α
B
β,β

1
4
B
α,α
B
α,β

1
4
η
αβ

λ
η
λλ
A
α
A
λ
B
α,λ
B
α,λ
+
1
4
B
α,β
B
β,α
=
1
2
B
α,β
B
β,α

1
2
η
αβ

µ
η
µµ
A
α
A
µ
B
α,µ
B
α,µ
+
1
4

µ
B
µ,α
B
µ,β
.
Finally, we put all the terms together:
R
αβ
=
_
1
2
B
α
+
1
2
B
β
− C
_
,αβ

1
2
η
αβ
A
α

λ
η
λλ
_
_
B
α,λ
A
λ
_

+C

B
α,λ
A
λ
_
+
1
2
C

B
α,β
+
1
2
C

B
β,α

1
2
B
α,β
B
β,α

1
4

µ
B
µ,α
B
µ,β
.
We can simplify this expression by considering separately diagonal and off-diagonal components:
R
αα
= (B
α
−C)
,αα

1
2
η
αα
A
α

λ
η
λλ
_
_
B
α,λ
A
λ
_

+C

B
α,λ
A
λ
_
+
_
C


1
2
B
α,α
_
B
α,α

1
4

µ
B
µ,α
B
µ,α
;
R
αβ
=
1
2
_
B
α
+B
β

λ
B
λ
_
,αβ
+
1
2
(C

B
α,β
+C

B
β,α
−B
α,β
B
β,α
)

1
4

µ
B
µ,α
B
µ,β
. (only for α ,= β)
19
Now we need to simplify this expression further by using the specific form of the metric (10)-(11). We have
A
t
= e
N
, A
r
= e
L
, A
θ
= r
2
, A
φ
= r
2
sin
2
θ;
B
t
= N, B
r
= L, B
θ
= 2 lnr, B
φ
= 2 lnr + 2 ln sin θ;
C = ln

−g =
N +L
2
+ 2 lnr + ln sin θ;
C
,t
=
˙
N +
˙
L
2
, C
,r
=
N

+L

2
+
2
r
, C

= cot θ, C

= 0.
Note that the term
1
2
_
B
α
+B
β

λ
B
λ
_
,αβ
(13)
always vanishes when α ,= β. We find (after some omitted algebra):
R
φφ
=
1
2
r
2
sin
2
θ

λ
η
λλ
_
_
__
ln
_
r
2
sin
2
θ


A
λ
_

+C

_
ln
_
r
2
sin
2
θ


A
λ
_
_
= −r
2
sin
2
θ
__
N

−L

2r
+
1
r
2
_
e
−L

1
r
2
_
;
R
θθ
= 1 −r
2
e
−L
_
N

−L

2r
+
1
r
2
_
;
R

= R

= R
θφ
= R

= 0;
R

=
1
2
C

B
θ,r

1
4
B
φ,r
B
φ,θ
=
1
2
cot θ
2
r

1
4
2
r
2 cot θ = 0;
R
tr
=
1
2
_
˙
N +
˙
L
2
N

+
_
N

+L

2
+
2
r
_
˙
L −
˙
NL

_

1
4
˙
NN


1
4
˙
LL

=
˙
L
r
;
R
rr
=
−N
′′
2
+
2
r
2
+N

L

−N

4
+
1
2
e
L−N
_
¨
L +
˙
L −
˙
N
2
˙
L
_
+
1
r
L

;
R
tt
=

¨
L
2
+
˙
L
˙
N −
˙
L
4
+
1
2
e
N−L
_
N
′′
+
_
N

−L

2
+
2
r
_
N

_
.
Finally, we compute the Ricci scalar,
R =

λ
η
λλ
1
A
λ
R
λλ
= e
−N
_

¨
L
2
+
˙
L
˙
N −
˙
L
4
_
+
1
2
e
−L
_
N
′′
+
_
N

−L

2
+
2
r
_
N

_
−e
−L
_
−N
′′
2
+
2
r
2
+N

L

−N

4
+
1
r
L

_

1
2
e
−N
_
¨
L +
˙
L −
˙
N
2
˙
L
_

1
r
2
+e
−L
_
N

−L

2r
+
1
r
2
_
+
_
N

−L

2r
+
1
r
2
_
e
−L

1
r
2
= e
−N
_

¨
L +
˙
L
˙
N −
˙
L
2
_
−e
−L
_
−2
N

−L

r
−N
′′
+N

L

−N

2
_

2
r
2
.
Hence, the nonzero components of the Einstein tensor are
G
tt
= e
N−L
L

r
+
e
N
r
2
;
G
tr
=
˙
L
r
;
G
rr
=
2
r
2
+
N

r

e
L
r
2
;
G
θθ
= −r
2
e
−L
_

N

−L

2r
+
1
r
2
+−
N
′′
2
+N

L

−N

4
_
+r
2
e
−N
2
_

¨
L +
˙
L
˙
N −
˙
L
2
_
;
G
φφ
= −
1
2
r
2
sin
2
θ
_
e
−N
_

¨
L +
˙
L
˙
N −
˙
L
2
_
−e
−L
_

N

−L

r
+
2
r
2
−N
′′
+N

L

−N

2
_
_
.
20
6.2.2 Solution using conformal transformation
A more clever way to reduce the amount of computation is to notice that the metric g
µν
is simplified after a conformal
transformation,
g
µν
= r
2
h
µν
= r
2
_
_
_
_
r
−2
e
L
0
−r
−2
e
N
1
0 sin
2
θ
_
_
_
_
.
The metric h
µν
separates into the r −t components and the θ − φ components. We shall first compute the Ricci tensor
for the metric h
µν
and then determine how R
αβ
changes under a conformal transformation. The calculation of R
αβ
for
the metric h
µν
is much simpler because h
µν
is a direct (“block”) sum of two metrics defined on 2-dimensional spaces. It
is clear that R
αβ
will also be a direct sum of the corresponding two-dimensional Ricci tensors.
Let us first compute the Ricci tensor for a diagonal metric γ
ab
= diag
_
e
A
, e
B
_
in two dimensions; set A ≡ A
1
, B ≡ A
2
,
and indices a, b, c,... range from 1 to 2. For a two-dimensional metric γ
ab
, we know that the Ricci tensor is proportional
to γ
ab
, namely (see Problem 4.4c)
R
ab
=
1
2
γ
ab
R, R ≡ γ
ab
R
ab
= 2 (det γ
ab
) R
1212
.
So it is sufficient to compute say R
11
,
R
11
= Γ
a
11,a
−Γ
a
1a,1
+ Γ
a
ba
Γ
b
11
−Γ
b
1a
Γ
a
1b
,
and afterwards we will have
R
ab
= γ
ab
R
11
γ
11
; R = γ
ab
R
ab
= 2
R
11
γ
11
.
The necessary Christoffel symbols are found as (no implicit summation from now on!)
Γ
a
1b
=

c
1
2
γ
ac

1c,b
+ γ
bc,1
−γ
1b,c
) =
1
2
_
A
1,b
δ
a
1

a
b
A
b,1
−δ
1
b
e
A1−Aa
A
1,a
_
;
Γ
1
12
=
1
2
A
,2
; Γ
2
12
=
1
2
B
,1
;
Γ
a
11
= A
1,1
δ
a
1

1
2
e
A1−Aa
A
1,a
; Γ
1
11
=
1
2
A
,1
; Γ
2
11
= −
1
2
e
A−B
A
,2
.

a
Γ
a
ba
=
1
2

b
ln (det γ
cd
) =
1
2
(A
1,b
+A
2,b
) =
1
2
(A +B)
,b
.
Then the component R
11
of the Ricci tensor is
R
11
= Γ
a
11,a
−Γ
a
1a,1
+ Γ
a
ba
Γ
b
11
−Γ
b
1a
Γ
a
1b
= Γ
1
11,1
+ Γ
2
11,2

1
2
(A+B)
,11
+
1
2
(A +B)
,1
Γ
1
11
+
1
2
(A +B)
,2
Γ
2
11
−Γ
1
11
Γ
1
11
−2Γ
1
12
Γ
2
11
−Γ
2
12
Γ
2
12
=
1
2
A
,11

1
2
_
e
A−B
A
,2
_
,2

1
2
(A+B)
,11
+
1
2
(A+ B)
,1
1
2
A
,1

1
2
(A +B)
,2
1
2
e
A−B
A
,2

1
4
A
,1
A
,1
+ A
,2
1
2
e
A−B
A
,2

1
4
B
,1
B
,1
= −
1
2
B
,11

1
2
e
A−B
A
,22

1
4
A
,2
(A −B)
,2
e
A−B
+
1
4
B
,1
(A−B)
,1
.
We also find (note the symmetry apparent in this formula; this shows that at least this is not obviously wrong)
R = 2
R
11
γ
11
= −e
−A
B
,11
−e
−B
A
,22

1
2
A
,2
(A −B)
,2
e
−B

1
2
e
−A
B
,1
(B −A)
,1
.
Well, we are not going to finish this calculation here, anyway. But this is roughly how it goes. Let us at
least derive the useful formula below.
The relationship between the curvature tensors under a conformal transformation is found as follows. First we define
the conformally transformed metric for convenience as follows,
˜ g
αβ
= e
2Ω
g
αβ
.
Then the Christoffel symbols receive a correction which is a tensor B
λ
αβ
,
˜
Γ
λ
αβ
= Γ
λ
αβ
+B
λ
αβ
; B
λ
αβ
≡ δ
λ
α



λ
β


−g
αβ


.
The Riemann and the Ricci tensors are defined (in Landau-Lifshitz sign convention) by
R
λ
αµβ
= Γ
λ
αβ,µ
−Γ
λ
αµ,β
+ Γ
λ
µν
Γ
ν
αβ
−Γ
λ
βν
Γ
ν
αµ
,
R
αβ
= R
λ
αλβ
= Γ
λ
αβ,λ
−Γ
λ
λα,β
+ Γ
λ
λν
Γ
ν
αβ
−Γ
λ
βν
Γ
ν
αλ
.
21
The same relation holds for
˜
R
λ
αµβ
and R
αβ
through
˜
Γ
λ
αβ
(note that these relations do not involve the metric g
αβ
explicitly).
Therefore
˜
R
λ
αµβ
−R
λ
αµβ
= B
λ
αβ,µ
−B
λ
αµ,β
+B
λ
µν
Γ
ν
αβ
+ Γ
λ
µν
B
ν
αβ
+B
λ
µν
B
ν
αβ
−B
λ
βν
Γ
ν
αµ
−Γ
λ
βν
B
ν
αµ
−B
λ
βν
B
ν
αµ
;
˜
R
αβ
−R
αβ
= B
λ
αβ,λ
−B
λ
λα,β
+B
λ
λν
Γ
ν
αβ
+ Γ
λ
λν
B
ν
αβ
+B
λ
λν
B
ν
αβ
−B
λ
βν
Γ
ν
αλ
−Γ
λ
βν
B
ν
αλ
−B
λ
βν
B
ν
αλ
.
We shall only compute the expression for the Ricci tensor R
αβ
. As a preparation, we compute
B
λ
λα
= δ
λ
λ


≡ NΩ

,
where N = δ
λ
λ
= g
αβ
g
αβ
is the number of spacetime dimensions. We shall always raise and lower indices using the original
metric g
αβ
. So we compute term by term,
˜
R
αβ
−R
αβ
=
_
δ
λ
α



λ
β


−g
αβ


_

−NΩ
,αβ
+NΩ

Γ
ν
αβ
+ Γ
λ
λν
_
δ
ν
α



ν
β


−g
αβ


_
+NΩ

_
δ
ν
α



ν
β


−g
αβ


_

_
δ
λ
β



λ
ν


− g
βν


_
Γ
ν
αλ
−Γ
λ
βν

ν
α



ν
λ


−g
αλ


) −
_
δ
λ
β



λ
ν


−g
βν


_

ν
α



ν
λ


−g
αλ


)
= 2Ω
,αβ
−g
αβ



−g
αβ,λ


−NΩ
,αβ
+NΩ

Γ
ν
αβ
+ Γ
λ
λα


+ Γ
λ
λβ


−g
αβ
Γ
λ
λν


+ 2NΩ



−Ng
αβ




−Γ
ν
αβ


−Γ
λ
λα


+g
βν
Γ
ν
αλ


− Γ
λ
αβ


−Γ
λ
λβ


+g
αλ
Γ
λ
βν


−(2 +N) Ω



+ 2g
αβ




= −g
αβ



−g
αβ
Γ
λ
λν


− Ng
αβ




+ 2g
αβ




= −(N −2)
_

,αβ
−Ω

Γ
ν
αβ
¸
+ (N −2) Ω



−g
αβ
_
(N −2) Ω



+ Ω


+ Γ
λ
λν


_
+
_
g
βν
Γ
ν
αλ


+g
αν
Γ
ν
βλ


−g
αβ,λ


¸
.
Now we note that some of the Γ terms can be absorbed into covariant derivatives, and also that the terms in the last
bracket cancel,
_
g
βν
Γ
ν
αλ


+g
αν
Γ
ν
βλ


−g
αβ,λ


¸
= 0,
so the resulting formula can be written more concisely as
˜
R
αβ
−R
αβ
= (N −2) [Ω



−Ω
;αβ
] −g
αβ
_
(N −2) Ω



+ Ω


¸
.
The modified Ricci scalar is
˜
R = ˜ g
αβ
˜
R
αβ
= e
−2Ω
g
αβ
R
αβ
+e
−2Ω
g
αβ
_
(N −2) [Ω



−Ω
;αβ
] −g
αβ
_
(N −2) Ω



+ Ω


¸_
= e
−2Ω
R +e
−2Ω
_
(N −2) [Ω



−Ω


] −N
_
(N −2) Ω



+ Ω


¸_
= e
−2Ω
¦R − (N −2) (N −1) Ω



−2 (N −1) Ω


¦ .
The Einstein tensor is modified as follows,
˜
G
αβ
=
˜
R
αβ

1
2
˜ g
αβ
˜
R = R
αβ
+ (N −2) [Ω



−Ω
;αβ
] −g
αβ
_
(N −2) Ω



+ Ω


¸

1
2
g
αβ
[R −(N −2) (N −1) Ω



−2 (N −1) Ω


]
= G
αβ
+ (N −2) [Ω



−Ω
;αβ
] +g
αβ
_
(N −2) (N −3)
2




+ (N −2) Ω


_
.
Note that there is no change in G
αβ
in two dimensions (since the Einstein tensor is always equal to zero).
6.3 Equations of motion
The equation for the covariant component u
1
(s) is
du
1
ds

1
2
u
α
u
β

∂r
(g
αβ
) = 0.
Using the metric g
αβ
= diag
_
f, −1/f, −r
2
, −r
2
sin
2
θ
_
, where f ≡ 1 −r
g
/r, and u
α
=
_
˙
t, ˙ r,
˙
θ,
˙
φ
_
, where ˙ ≡ d/dλ, we find
d

_
−f
−1
˙ r
_

1
2
_
df
dr
˙
t
2

d
dr
_
1
f
_
˙ r
2
−2r
˙
θ
2
−2r
˙
φ
2
sin
2
θ
_
= 0. (14)
To derive this equation from other equations given in the lecture, we transform in a clever way the expression
0 =
d

/ =
d

_
f
˙
t
2
−f
−1
˙ r
2
−r
2
˙
θ
2
−r
2
˙
φ
2
sin
2
θ
_
.
22
Namely, we try to separate terms of the form
d

(u
α
) out of the terms of the form
d

(u
α
u
α
) in the following way,
d

_
u
1
u
1
_
=
d

_
g
11
u
1
u
1
_
= 2u
1
d

_
g
11
u
1
_
−u
1
u
1
d

g
11
(no summation!).
For example,
d

_
f
˙
t
2
_
= 2
˙
t
d

_
f
˙
t
_

˙
t
2
d

f, etc.
We find
0 =
d

/ =
d

_
f
˙
t
2
−f
−1
˙ r
2
−r
2
˙
θ
2
−r
2
sin
2
θ
˙
φ
2
_
= 2
˙
t
d

_
f
˙
t
_

˙
t
2
df

−2 ˙ r
d

_
f
−1
˙ r
_
+ ˙ r
2
d

f
−1
−2
˙
θ
d

_
r
2
˙
θ
_
+
˙
θ
2
d

_
r
2
_
−2
˙
φ
d

_
r
2
sin
2
θ
˙
φ
_
+
˙
φ
2
d

_
r
2
sin
2
θ
_
.
Now we substitute the given equations (2)-(4), and also evaluate derivatives of the metric, e.g. df/dλ = f

˙ r, so
0 = −
˙
t
2
f

˙ r −2 ˙ r
d

_
f
−1
˙ r
_
− ˙ r
2
f

f
2
˙ r −2
˙
θr
2
˙
φ
2
sin θ cos θ +
˙
θ
2
2r ˙ r +
˙
φ
2
2r sin
2
θ ˙ r + 2
˙
φ
2
r
2
˙
θ sinθ cos θ
= ˙ r
_
d

_
−2f
−1
˙ r
_
−f

˙
t
2

f

f
2
˙ r
2
+ 2r
˙
θ
2
+ 2r
˙
φ
2
sin
2
θ
_
.
This is obviously equivalent to Eq. (14).
Note: the reason one of the equations follows from other equations is that the equation u
α
u
α
= const is a consequence
of the four geodesic equations, u
β
u
α

= 0, and the fact that g
αβ;µ
= 0. Therefore, when we consider the four geodesic
equations and the equation u
α
u
α
= const, any one of these five equations is a consequence of four others.
7 Weak gravitational fields
7.1 Gravitational bending of light
In the lecture it was shown that the trajectory of a light ray in polar coordinates satisfies the equation
d
2

2
_
1
r
_
+
1
r
=
3
2
r
g
r
2
, r
g

2GM
c
2
≈ 3km,
where M is the mass of the Sun. Introduce an auxiliary variable v(φ) ≡ r
−1
and solve the equation
v
′′
+v =
3
2
r
g
v
2
perturbatively, assuming that v is small,
v(φ) = v
0
(φ) +v
1
(φ) +...
The unperturbed solution is
v
0
(φ) =
1
R
0
cos φ,
where R
0
is the distance of closest approach to the Sun. Then
v
′′
1
+v
1
=
3
2
r
g
R
2
0
cos
2
φ =
3
4
r
g
R
2
0
(1 + cos 2φ) .
The solution is found with undetermined coefficients,
v
1
(φ) = A+Bcos 2φ, A =
3
4
r
g
R
2
0
, B = −
1
4
r
g
R
2
0
.
The total deflection angle is found as δ = φ
1
−φ
2
−π, where φ
1,2
are fixed by the condition v(φ) = 0. We find a quadratic
equation
cos
2
φ −
2R
0
r
g
cos φ −2 = 0, cos φ =
R
0
r
g
±
¸
R
2
0
r
2
g
+ 2.
Only the solution with the minus sign is meaningful (cos φ < 1). Since r
g
≪R
0
, we may expand this in Taylor series and
find
cos φ ≈ −
r
g
R
0
+O(r
3
g
/R
3
0
).
23
Therefore, the angle φ is very close to π/2,
φ
1,2
= ±
_
π
2

_
, ε ≈
r
g
R
0
, ⇒ δ = 2ε =
2r
g
R
0
.
This formula can be rewritten as
δ =
2r
g
/R

R
0
/R

=
_
2r
g
R

_
R

R
0
.
For the Sun we have R

= 6, 96 10
5
km and r
g
= 2, 954 km, therefore
2r
g
/R

= 8, 489 10
−6
=
__
8, 489 10
−6
360

/2π
¸
3600
¸
[arc seconds]
= 1, 751”
(see R. Oloff “Geometrie der Raumzeit,” 2nd German edition, page 151).
7.2 Einstein tensor for weak field
For this problem Chapter 4 from the book Norbert Straumann “General Relativity and Relativistic Astrophysics” is
useful. We have g
µν
= η
µν
+h
µν
and
R
µν
= Γ
λ
µν,λ
−Γ
λ
λµ,ν
,
where (...)

denotes a derivative ∂
µ
(...). Here one can ask the students about the symmetry of this tensor.
Furthermore
Γ
α
µν
=
1
2
η
αβ
[h
µβ,ν
+h
νβ,µ
−h
µν,β
] =
1
2
_
h
α
µ,ν
+h
α
ν,µ
−h

µν
¸
, (15)
where as usual we use the convention that indices are raised or lowered with η
µν
; thus e.g. h
α
β
≡ η
αλ
h
λβ
. Using Eq. (15)
we have
R
µν
=
1
2
_
h
λ
µ,νλ
+h
λ
ν,µλ
−h
µν
−h
,µν
¸
,
where = η
µν

µ

ν
and h = h
λ
λ
= η
λα
h
αλ
. And for the Ricci scalar we obtain
R = η
µν
R
µν
= h
λν
,νλ
−h.
Thus in the linear approximation we have
G
µν
= R
µν

1
2
η
µν
R =
1
2
_
h
λ
µ,νλ
+h
λ
ν,µλ
−h
µν
−h
,µν
−η
µν
h
λβ
,βλ

µν
h
_
.
Let us introduce a new variable γ
µν
≡ h
µν

1
2
η
µν
h. The traces of two tensors h and γ are related by γ = −h, thus
h
µν
≡ γ
µν

1
2
η
µν
γ. Inserting the last expression for h
µν
in G
µν
, we have
G
µν
=
1
2
_
γ
λ
µ,νλ

λ
ν,µλ
−γ
µν
−η
µν
γ
λβ
,βλ
_
=
=
1
2
_
γ

µλ,ν


νλ,µ
−γ
µν
−η
µν
γ
,λβ
βλ
_
,
or finally
G
µ
ν
=
1
2
_
γ
µ,λ
λ,ν

λ,µ
ν,λ
−γ
µ
ν
−δ
µ
ν
γ
β,λ
λ,β
_
.
7.3 Gravitational perturbations I
The metric is written as g
µν
= η
µν
+δg
µν
, i.e.
g
00
= 1 + 2Φ, g
0i
= B
,i
+S
i
, g
ij
= −δ
ij
+ 2Ψδ
ij
+ 2E
,ij
+F
i,j
+F
j,i
+h
ij
, (16)
where S
,i
i
= F
,i
i
= h
,i
ij
= h
ij
η
ij
= 0, h
ij
= h
ji
. We shall use the formula for G
µ
ν
derived in Problem 7.2. All 3-dimensional
indices are raised and lowered using δ
ij
, so we can write these indices in any position, as convenient:
δg
0
j
= δg
0j
= −δg
j
0
, δg
j
i
= −δg
ij
.
Also note that for any quantity X we have
X
,j
=
_
˙
X, −X
,j
_
.
We need to write the components of
γ
µ
ν
= δg
µ
ν

1
2
δ
µ
ν
¯
h,
¯
h ≡ δg
µ
µ
,
24
using the 3+1 decomposition:
¯
h = δg
µ
µ
= η
µν
δg
µν
= 2 (Φ −3Ψ−∆E) ,
γ
0
0
= Φ + 3Ψ+ ∆E, γ
0
j
= B
,j
+S
j
= −γ
j
0
= γ
0j
,
γ
i
j
= −(Φ −Ψ−∆E) δ
ij
−2E
,ij
−F
i,j
−F
j,i
−h
ij
= γ
j
i
.
Now we compute
γ
0,λ
λ
= ˙ γ
0
0
−γ
0
j,j
=
˙
Φ + 3
˙
Ψ + ∆
_
˙
E −B
_
;
γ
j,λ
λ
= −γ
λ
j,λ
= −˙ γ
0j
−γ
j
i,i
= −
˙
B
,j

˙
S
j
−(−(Φ −Ψ−∆E) δ
ij
−2E
,ij
−F
i,j
−F
j,i
−h
ij
)
,i
= −
˙
B
,j

˙
S
j
+ (Φ −Ψ + ∆E)
,j
+ ∆F
j
;
γ
β,λ
λ,β
=
_
γ
0,λ
λ
_
,0
+
_
γ
j,λ
λ
_
,j
=
¨
Φ + 3
¨
Ψ+ ∆
_
¨
E −
˙
B
_
+
_

˙
B
,j

˙
S
j
+ (Φ −Ψ + ∆E)
,j
+ ∆F
j
_
,j
=
¨
Φ + 3
¨
Ψ + ∆
¨
E −2∆
˙
B + ∆(Φ −Ψ + ∆E) .
Then we compute each component of G
µ
ν
separately:
2G
0
0
= 2γ
0,λ
λ,0
−γ
0
0
−δ
0
0
γ
β,λ
λ,β
= 2
_
¨
Φ + 3
¨
Ψ+ ∆
_
¨
E −
˙
B
__
−∂
0

0
(Φ + 3Ψ + ∆E) + ∆(Φ + 3Ψ+ ∆E)

_
¨
Φ + 3
¨
Ψ+ ∆
¨
E
_
+ 2∆
˙
B −∆(Φ −Ψ + ∆E)
= 4∆Ψ;
2G
0
j
= γ
0,λ
λ,j

λ,0
j,λ
−γ
0
j
−δ
0
j
γ
β,λ
λ,β
=
_
˙
Φ + 3
˙
Ψ + ∆
_
˙
E −B
__
,j


_

¨
B
,j

¨
S
j
+
_
˙
Φ −
˙
Ψ + ∆
˙
E
_
,j
+ ∆
˙
F
j
_

_
¨
B
,j
+
¨
S
j
_
+ ∆(B
,j
+S
j
) = 4
˙
Ψ
,j
+ ∆S
j
−∆
˙
F
j
,
2G
i
j
= γ
i,λ
λ,j

j,λ
λ,i
−γ
i
j
−δ
i
j
γ
β,λ
λ,β
=
_

˙
B
,j

˙
S
j
+ (Φ −Ψ + ∆E)
,j
+ ∆F
j
_
,i
+
_

˙
B
,i

˙
S
i
+ (Φ −Ψ + ∆E)
,i
+ ∆F
i
_
,j
+((Φ −Ψ−∆E) δ
ij
+ 2E
,ij
+F
i,j
+F
j,i
+h
ij
) −δ
ij
_
¨
Φ + 3
¨
Ψ + ∆
¨
E −2∆
˙
B + ∆(Φ −Ψ + ∆E)
_
= 2
_
Φ −Ψ−
˙
B +
¨
E
_
,ij
+
¨
F
i,j
+
¨
F
j,i

˙
S
i,j

˙
S
j,i
+h
ij
−2δ
ij
_
2
¨
Ψ+ ∆
_
Φ −Ψ−
˙
B +
¨
E
__
.
* - the origin of the minus sign here is γ
λ,0
j,λ
= −γ
j,λ
λ,0
.
7.4 Gravitational perturbations II
Under an infinitesimal transformation x
µ
→x
µ

µ
, the metric changes as
g
αβ
→g
αβ
−g
αγ
ξ
γ

−g
βγ
ξ
γ

= g
αβ
−ξ
α,β
−ξ
β,α
. (17)
(This can be easily found from the standard formula for the change of coordinages, involving ∂˜ x
µ
/∂x
ν
.) Now let
us write Eq. (17) in full, using the perturbation variables (16), the covariant components ξ
µ
, and the decomposition
ξ
µ
=
_
ξ
0
, ξ
⊥i

,i
_
. We can write the transformation of g
αβ
component by component using the 3+1 decomposition, and
we use the fact that the background metric is diagonal,
g
00
→g
00
−2ξ
0
,0
; g
0i
→g
0i
−ξ
0
,i
−ξ
i,0
; g
ij
→g
ij
−ξ
i,j
−ξ
j,i
.
To simplify calculations, we adopt the convention of raising and lowering the spatial indices i, j, ... by the Euclidean
spatial metric δ
ij
rather than by η
ij
. This will get rid of some minus signs. We also denote ∂
0
≡ ∂
t
by the overdot. Thus
we have
g
00
→g
00
−2
˙
ξ
0
; g
0i
→g
0i
−ξ
0
,i

˙
ξ
i
; g
ij
→g
ij
− ξ
i,j
−ξ
j,i
.
Substituting the perturbation variables from Eq. (16), we get
Φ →Φ −
˙
ξ
0
, (18)
B
,i
+S
i
→B
,i
+S
i
−ξ
0
,i

˙
ξ
⊥i

˙
ζ
,i
, (19)
2Ψδ
ij
+ 2E
,ij
+F
i,j
+F
j,i
+h
ij
→2Ψδ
ij
+ 2E
,ij
+F
i,j
+F
j,i
+h
ij
−ξ
⊥i,j
−ξ
⊥j,i
−2ζ
,ij
. (20)
Now we need to separate these equations and derive the transformation laws for the individual perturbation variables.
This is easy to do if we perform a Fourier transform of Eqs. (18)-(20) and pass to the Fourier space (where every variable
is a function of a 3-vector k). A vector V
i
is decomposed into scalar and vector components as follows,
V
j
= ik
j
V
(S)
+V
(V )
j
; V
(S)

V
l
k
l
k
2
, V
(V )
j
≡ V
j
−ik
j
V
l
k
l
k
2
= V
j
−ik
j
V
(S)
. (21)
25
The idea is first, to project the given vector V
i
(k) onto the direction of k
i
, and second, to subtract the projection from
V
i
and to obtain the component of V
i
which is transversal to k
i
. The imaginary unit factors are added as coefficients at
k
j
for convenience: with these factors, the decomposition (21) translates to real space as
V
j
= ∂
j
V
(S)
+V
(V )
j
.
The same procedure applied to a symmetric tensor T
ij
leads to a decomposition into scalar, vector, and tensor
components. Let us go through this procedure in more detail. First, we subtract the trace and obtain the traceless part
T
(1)
of the tensor T,
T
(1)
ij
≡ T
ij

1
3
δ
ij
T
ll
; T
(1)
ii
= 0.
Note the coefficient
1
3
that depends on the number of spatial dimensions (three). Now we project T
(1)
ij
onto k
i
k
j
and
obtain the scalar component T
(S)
proportional to k
i
k
j
and the tensor T
(2)
ij
orthogonal to k
i
k
j
:
T
(1)
ij

_
−k
i
k
j
+
1
3
δ
ij
k
2
_
T
(S)
+T
(2)
ij
; T
(S)
≡ −
3
2
k
i
k
j
T
(1)
ij
k
4
; T
(2)
ij
k
i
k
j
= 0.
Note that T
(2)
ij
is again a trace-free tensor, T
(2)
ii
= 0, due to the subtraction of
1
3
k
2
δ
ij
in the first term. Finally, we project
T
(2)
ij
onto k
i
and k
j
separately, to obtain a “vector” part T
(V )
j
such that T
(V )
j
k
j
= 0, and a completely traceless (“tensor”)
part T
(T)
ij
such that T
(T)
ij
k
j
= 0 and T
(T)
ii
= 0:
T
(2)
ij
= ik
i
T
(V )
j
+ ik
j
T
(V )
i
+T
(T)
ij
; T
(V )
j
≡ i
k
l
k
2
T
(2)
jl
, T
(T)
ij
≡ T
(2)
ij
−i
_
k
i
T
(V )
j
+k
j
T
(V )
i
_
.
In real space, the full decomposition is
T
ij
=
1
3
T
ll
δ
ij
+
_

i

j

1
3
δ
ij

_
T
(S)
+∂
i
T
(V )
j
+∂
j
T
(V )
i
+T
(T)
ij
.
T
(S)

3
2
1

2

i

j
_
T
ij

1
3
T
ll
δ
ij
_
; T
(2)
ij

_
T
ij

1
3
T
ll
δ
ij
_

_

i

j

1
3
δ
ij

_
T
(S)
;
T
(V )
j
=
1


i
T
(2)
ij
, T
(T)
ij
= T
(2)
ij
−∂
i
T
(V )
j
−∂
j
T
(V )
i
.
It may be convenient to gather the “trace” terms (the terms containing δ
ij
) as one term,
T
ij
= T
(tr)
δ
ij
+∂
i

j
T
(S)
+∂
i
T
(V )
j
+∂
j
T
(V )
i
+T
(T)
ij
, T
(tr)

1
3
T
ll

1
3
∆T
(S)
.
Note that the perturbation variables Ψ, E, F
i
, h
ij
are obtained by this decomposition method, starting from the
symmetric perturbation tensor δg
ij
, with slight modifications: there are some cosmetic factors of 2 and some minus signs.
Applying the decomposition method to Eqs. (18)-(20), we get the following transformation laws for the perturbation
variables,
Φ →Φ −
˙
ξ
0
, B →B −ξ
0

˙
ζ, S
i
→S
i

˙
ξ
⊥i
,
E →E −ζ, Ψ →Ψ, F
i
→F
i
−ξ
⊥i
, h
ij
→h
ij
.
Remarks:
1. It is clear that one can set F
i
= 0, B = E = 0 with a coordinate transformation. Other components will then show
whether the geometry is really perturbed or it’s just a coordinate transformation of a flat space. In general, there will
remain 6 independent components of perturbations (Φ, Ψ, S
i
, h
ij
).
2. These considerations depend rather crucially on the silently made assumption that all the metric perturbations
vanish, δg
µν
→0, at spatial infinity. These boundary conditions are implicitly used when defining the Fourier transforms
necessary for the tensor/vector/scalar decompositions (a Fourier transform is undefined without this boundary condition).
Alternatively, one may do without Fourier transforms but then one still needs boundary conditions to solve the relevant
Poisson equations for components. Without boundary conditions, there is no unique decomposition of the form
X
i
= A
,i
+B
i
, A =
1

X
i,i
,
because the function A is defined up to solutions of ∆A = 0. So the tensor/vector/scalar decomposition is actually
undefined without a fixed assumption about the boundary conditions. The boundary conditions δg
µν
→ 0 at spatial
infinity is a natural, physically motivated set of boundary conditions. An explicit counterexample where these boundary
conditions are not satisfied: g
µν
= diag (A, −B, −B, −B), where A ,= 1, B ,= 1 are constants. This metric is flat
but one cannot see this by using the perturbation formalism! (The component Ψ ,= 0 cannot be removed by a gauge
transformation.) The reason is that this g
µν
is a “perturbation” of flat metric with Φ and Ψ that do not decay to zero at
spatial infinity. So a coordinate transformation with ξ
µ
decaying to zero cannot bring this metric to η
µν
.
26
8 Gravitational radiation I
8.1 Gauge invariant variables
Using the equations derived in Problem 7.4, it is very easy to verify that D = Φ −Ψ −
˙
B +
¨
E and S
i

˙
F
i
are invariant
under infinitesimal changes of coordinates (i.e. invariant under infinitesimal gauge transformations).
8.2 Detecting gravitational waves
8.2.1 Using distances between particles
(This solution follows Hobson-Efstathiou-Lasenby [2006], ¸18.4.)
Consider a plane wave moving in the z direction, (all other components of h
µν
are zero)
h
xx
= −h
yy
= A
+
e
−iω(t−z)
, h
xy
= h
yx
= A
×
e
−iω(t−z)
. (22)
To detect the presence of this gravitational wave, let us imagine a cloud of particles initially at rest at different positions.
The 4-vectors describing the particles are u
µ
= (1, 0, 0, 0), so one can easily see that these particles move along geodesics:
u
ν
u
µ

= u
ν
u
µ

+ Γ
µ
να
u
ν
u
α
= Γ
µ
00
,
Γ
µ
να
=
1
2
η
λµ
(h
λν,α
+h
λα,ν
−h
αν,λ
) ,
Γ
µ
00
=
1
2
η
λµ
(h
λ0,0
+h
λ0,0
−h
00,λ
) = 0.
Therefore the coordinates x
µ
of each particle remain constant with time. However, the distance between each pair of
particles is determined through the spacelike vector ∆x
µ
≡ x
µ
(1)
−x
µ
(2)
as
∆L
2
≈ (η
µν
+h
µν
) ∆x
µ
∆x
ν
and will change with time because of the dependence on h
µν
. Since the only nonzero components of h
µν
are the x, y
components, it is clear that only changing lengths are between particles that have some separation in the x, y directions.
Therefore it is sufficient to consider a ring of particles situated in the x − y plane. The physically measured distances
between the particles in the ring will change with time, i.e. the ring will experience a deformation.
To visualize the deformation, it is convenient to make a local coordinate transformation (local in the neighborhood
of the ring) such that the metric becomes flat, g
µν
x
µ
x
ν
= η
µν
˜ x
µ
˜ x
ν
(up to second-order terms). The trick that performs
this transformation is the following,
˜ x
µ
= x
µ
+
1
2
h
µ
λ
x
λ
≡ x
µ
+
1
2
h
αλ
x
λ
η
αµ
.
It is easy to check that
g
µν
x
µ
x
ν
≡ (η
µν
+h
µν
) x
µ
x
ν
= η
µν
˜ x
µ
˜ x
ν
+O(h
2
).
Therefore, ˜ x
µ
can be understood as the (approximate) Cartesian coordinates where the length is given by the usual
Pythagorean formula. Now if we compute the shape of the ring in these coordinates, it will be easy to interpret this
shape in a straightforward way.
Consider a particle with constant 3-coordinates (x, y, z). After the coordinate transformation, we have
˜ x = x +
1
2
(A
+
x +A
×
y) e
−iω(t−z)
,
˜ y = y +
1
2
(A
×
x −A
+
y) e
−iω(t−z)
,
˜ z = z.
To visualize the deformation, it is convenient to consider first the case A
+
,= 0, A
×
= 0 and then the opposite case.
The deformation of the ring is squeezing in one direction and expansion in the orthogonal direction. It follows that A
+
describes a deformation in the two vertical directions, while A
×
describes a deformation in the directions at 45

.
Note that the deformations change the shape of the ring in the same way, except for the rotated orientation. This
can be verified by performing a rotation by
π
4
,
_
˜ x
˜ y
_

1

2
_
1 −1
1 1
__
˜ x
˜ y
_
,
and then it is straightforward to see that this will exchange A
+
and A
×
.
27
8.2.2 Using geodesic deviation equation
PLEASE NOTE: The commonly found arguments that use the geodesic deviation equation are suspect because the
geodesic deviation equation uses coordinates ξ
µ
rather than gauge-invariant quantities. A cloud of particles at rest in
the gravitational field h
µν
described by Eq. (22) will stay indefinitely at rest in the coordinate system (ξ
µ
= const)
even though the distances between particles will change with time. See arxiv:gr-qc/0605033 for nice explanations. The
solution given above is simple and straightforward. The argument using the geodesic deviation (see Carroll, Chapter 6,
p. 152-154) goes like this:
The geodesic deviation equation can be simplified for a deviation vector S
σ
corresponding to nonrelativistic (almost
stationary) particles moving with 4-velocity approximately equal to (1, 0, 0, 0),
d
2
S
σ
dt
2
= R
σ
00λ
S
λ
.
The Riemann tensor to first order in h can be expressed as
R
σ
00λ
=
¨
h
σ
λ
(note that h
µ0
= 0). Therefore, the geodesic deviation equation becomes
¨
S
σ
=
¨
h
σ
λ
S
λ
,
¨
S
x
= ω
2
(h
xx
S
x
+h
xy
S
y
) = ω
2
(A
+
S
x
+A
×
S
y
) e
−iω(t−z)
,
¨
S
y
= ω
2
(h
yx
S
x
+h
yy
S
y
) = ω
2
(A
×
S
x
−A
+
S
y
) e
−iω(t−z)
,
and there is no change in the z direction.
8.3 Poisson equation
The general solution of the Poisson equation,
∆φ = 4πρ,
with boundary conditions φ →0 at infinity, is easy to find using the Fourier transform:
−k
2
φ(k) = 4πρ(k),
φ(x) = −
_
d
3
k
(2π)
3
e
ik·x
4πρ(k)
k
2
= −
_
d
3
k
(2π)
3
e
ik·x

k
2
_
d
3
ye
−ik·y
ρ(y) =
_
d
3
y ρ(y)G(x −y),
where G(x) is the Green’s function,
G(x) = −
_
d
3
k
(2π)
3
e
ik·x

k
2
= −
1
π
_

0
dk
_
π
0
dθ sin θ e
ik|x| cos θ
= −
2
π [x[
_

0
dk
k
sinkx = −
1
[x[
.
Here we used the known integral
_

0
sinz
z
dz =
1
2
_
+∞
−∞
sin z
z
dz =
1
2
π.
Therefore
φ(x) = −
_
d
3
y
[x −y[
ρ(y). (23)
One can denote this integral more concisely,
φ = 4π
1

ρ,
where the operator
1

is just a shorthand notation for the integral in Eq. (23).
Note that the function ρ must fall off sufficiently rapidly as [x[ →∞ or else the integral (23) will not converge. It is
sufficient that [ρ(x)[ ∼ [x[
−2−ε
at large [x[ (where ε > 0).
8.4 Metric perturbations 1
An arbitrary 3-vector X
i
(such as T
0
i
) is decomposed into scalar and vector parts as follows,
X
i
= a
,i
+b
i
, b
i,i
= 0.
To determine an explicit expression for a, let us compute the divergence of X
i
,
X
i,i
= a
,ii
= ∆a.
Therefore
a(x) = −
1

_
d
3
y
[x −y[
X
i,i
(y).
One can write more concisely
a =
1

X
i,i
.
28
8.5 Metric perturbations 2
The energy-momentum tensor T
µν
is decomposed as
T
0
i
= α
,i

i
, α ≡
1

T
0
k,k
, β
i
≡ T
0
i

_
1

T
0
k,k
_
,i
;
T
i
k
= µδ
ik

,ik

i,k

k,i
+T
(T)i
k
,
β
i,i
= σ
i,i
= 0, T
(T)i
i
= 0, T
(T)i
k,i
= 0.
We need to verify that the equation

1
16πG
(
˙
S
i

¨
F
i
) = σ
i
, (24)
which represents the vector part of the spatial Einstein equation (here σ
i
is the vector part of the spatial T
ij
), also follows
from the conservation of T
µν
and from the other Einstein equations.
To calculate the components λ, µ of the EMT, we compute
T
i
i
= ∆λ + 3µ,
T
i
k,i
= ∆λ
,k

,k
+ ∆σ
k
,
T
i
k,ik
= ∆∆λ + ∆µ.
Now we solve this system of equations and find
µ =
1
2
_
T
i
i

1

T
i
k,ik
_
, λ =
3
2
1

T
i
k,ik

1
2
T
i
i
,
σ
j
=
1

T
i
j,i

1

_
1

T
i
k,ik
_
,j
,
(T)
T
i
k
= T
i
k
−µδ
i
k
−λ
,ik
−σ
i,k
−σ
k,i
.
Note that the operator
1

2
applied to a function f(x) is defined only if the function f has a sufficiently fast decay at
[x[ → ∞. It is sufficient that [f(x)[ ∼ [x[
−3−ε
with ε > 0 at large [x[. This is a faster decay than that required by the
operator
1

.
The Einstein equations are
2∆Ψ = 8πGT
0
0
,
2
˙
Ψ
,i
+
1
2

˜
S
i
= 8πGT
0
i
= 8πG(α
,i

i
) ,
D
,ij
−δ
ij
_
∆D + 2
¨
Ψ
_

1
2
_
˙
˜
S
i,j
+
˙
˜
S
j,i
_
+
1
2
h
ij
= 8πGT
i
j
= 8πG
_
µδ
ik

,ik

i,k

k,i
+T
(T)i
k
_
,
where we have denoted for brevity
˜
S
i
≡ S
i

˙
F
i
, D ≡ Φ −Ψ +B −
˙
E,
which are gauge-invariant variables. In the 3+1 decomposition, the Einstein equations become
∆Ψ = 4πGT
0
0
, (25)
˙
Ψ = 4πGα, (26)

˜
S
i
= 16πGβ
i
, (27)
D = 8πGλ, (28)
∆D + 2
¨
Ψ = −8πGµ, (29)
˙
˜
S
i
= −16πGσ
i
, (30)
h
ij
= 16πGT
(T)i
j
. (31)
The conservation law of the EMT in 3+1 decomposition looks like this,
T
0
0,0
+T
j
0,j
= 0, T
0
i,0
+T
j
i,j
= 0. (32)
This gives
˙
T
0
0
= ∆α, ˙ α
,i
+
˙
β
i
+ ∆λ
,i

,i
+ ∆σ
i
= 0,
therefore
˙
T
0
0
= ∆α, ˙ α + ∆λ +µ = 0,
˙
β
i
+ ∆σ
i
= 0. (33)
Then it is easy to see that Eqs. (26), (29), and (30) are consequences of Eqs. (25), (27), (28), and the conservation
laws (33). In particular,
˙
˜
S
i
= ∂
t
1

16πGβ
i
= −
1

16πG∆σ
i
= −16πGσ
i
.
29
9 Gravitational radiation II
9.1 Projection of the matter tensor
a) First note that P
ab
is a projector,
P
ab
P
bc
= P
ac
,
and its image has dimension 2, that is, the trace of P
ab
is 2,
P
ii
= 3 −n
i
n
i
= 2.
Therefore, for any X
ab
we have
(T)
X
ii
= P
ia
X
ab
P
bi

1
2
P
ii
P
ab
X
ab
= P
ab
X
ab

1
2
2P
ab
X
ab
= 0.
b) We compute
(T)
X
ik,i
=
_
P
ia
X
ab
P
bk

1
2
P
ik
P
ab
X
ab
_
,i
=
_
P
ia
P
bk

1
2
P
ik
P
ab
_
,i
X
ab
+
_
P
ia
P
bk

1
2
P
ik
P
ab
_
X
ab,i
. (34)
Note that the projection kills any component proportional to R
i
because P
ia
R
i
= 0. At the same time, X
ab,i
is propor-
tional to R
i
because
X
ab,i
=
_
X(t −[

R[)
_
,i
= −
R
i
R
X

.
Therefore the second term in Eq. (34) vanishes:
_
P
ia
P
bk

1
2
P
ik
P
ab
_
R
i
= 0.
So only the first term remains,
(T)
X
ik,i
=
_
P
ia
P
bk

1
2
P
ik
P
ab
_
,i
X
ab
.
However, this term contains derivatives of P
ab
, which are also sometimes proportional to R
i
. We compute
P
ik,a
= −n
i,a
n
k
−n
i
n
k,a
; n
i,a
=
_
R
i
R
_
,a
=
R
i,a
R

R
i
R
2
[R[
,a
=
δ
ia
R

R
i
R
a
R
3
=
1
R
P
ia
,
P
ik,a
= −
1
R
(P
ai
n
k
+P
ak
n
i
) , P
ik,i
= −
2
R
n
k
, (note that P
ak
n
k
= 0)
_
P
ia
P
bk

1
2
P
ik
P
ab
_
,i
= P
ia,i
P
bk

1
2
P
ik,i
P
ab
+P
ia
P
bk,i

1
2
P
ik
P
ab,i
= −
2
R
n
a
P
bk
+
1
R
n
k
P
ab

1
R
P
ia
(P
bi
n
k
+P
ik
n
b
) +
1
2R
P
ik
(P
ai
n
b
+P
bi
n
a
)
=
1
R
_
−2n
a
P
bk
+n
k
P
ab
−P
ab
n
k
−P
ak
n
b
+
1
2
P
ak
n
b
+
1
2
P
bk
n
a
_
= −
1
R
_
3
2
P
bk
n
a
+
1
2
P
ak
n
b
_
.
This is higher-order in 1/[

R[ than P
ab
, as required.
9.2 Matter sources
The question is to verify the following property,
(T)
X
ik
=
(T)
Q
ik
,
where
Q
ik
= X
ik

_
1
3
δ
ik
r
2
T
0
0
d
3
r.
It is easy to see that X
ik
differs from Q
ik
only by a term of the form A(t, R)δ
ij
. The transverse-traceless part of δ
ij
is
zero,
_
P
ia
P
bk

1
2
P
ik
P
ab
_
δ
ab
= 0.
Therefore the transverse-traceless parts of X
ik
and Q
ik
are the same.
30
9.3 Energy-momentum tensor of gravitational waves
See Hobson-Efstathiou-Lasenby [2006], ¸17.11.
We need to compute the second-order terms in the Einstein tensor. The idea is to separate the second-order terms
already in the Ricci tensor. We will also try to simplify things by using the fact that h
µν
is purely transverse-traceless;
h

= 0, h
ii
= 0, h
ik,i
= 0. It follows that
η
µν
h
µν
= 0, h
µν

= 0.
Also, it is given that the EMT of matter vanishes, T
µν
= 0, which we will use below.
First we decompose the metric,
g
µν
= η
µν
+h
µν
, g
µν
= η
µν
−h
µν
;
note that now indices are always raised and lowered using η
µν
. We need to compute the Ricci tensor to second order.
The Christoffel symbol up to second order is
Γ
λ
αβ
=
1
2
_
η
λµ
−h
λµ
_
(h
µα,β
+h
µβ,α
− h
αβ,µ
) = Γ
(1)λ
αβ
+ Γ
(2)λ
αβ
,
Γ
(1)λ
αβ
=
1
2
η
λµ
(h
αµ,β
+h
βµ,α
−h
αβ,µ
) ,
Γ
(2)λ
αβ
= −
1
2
h
λµ
(h
αµ,β
+h
βµ,α
−h
αβ,µ
) .
The Ricci tensor is
R
αβ
= Γ
λ
αβ,λ
−Γ
λ
λα,β
+ Γ
λ
λν
Γ
ν
αβ
−Γ
λ
βν
Γ
ν
αλ
= R
(1)
αβ
+R
(2)
αβ
,
R
(1)
αβ
= Γ
(1)λ
αβ,λ
−Γ
(1)λ
λα,β
,
R
(2)
αβ
= Γ
(2)λ
αβ,λ
−Γ
(2)λ
λα,β
+ Γ
(1)λ
λν
Γ
(1)ν
αβ
−Γ
(1)λ
βν
Γ
(1)ν
αλ
.
Let us now evaluate these expressions and simplify as much as possible, as early as possible:
Γ
(1)λ
αλ
=
1
2
η
λµ
h
λµ,α
=
1
2
_
η
λµ
h
λµ
_

= 0,
R
(1)
αβ
= Γ
(1)λ
αβ,λ
−Γ
(1)λ
λα,β
=
1
2
η
λµ
(h
αµ,βλ
+h
βµ,αλ
−h
αβ,µλ
) = −
1
2
h
αβ
,
because of the transverse traceless property of h
µν
. Now, since R
(1)
αβ
is found from the first-order Einstein equation
R
(1)
αβ

1
2
η
αβ
R
(1)
= 8πGT
αβ
,
and it is given that T
αβ
= 0. Hence, we have h
αβ
= 0.
Let us now evaluate derivatives of the second-order terms in the Christoffel symbols:
Γ
(2)λ
αλ
= −
1
2
h
λµ
(h
αµ,λ
+h
λµ,α
−h
αλ,µ
) = −
1
2
h
λµ
h
λµ,α
,
−Γ
(2)λ
αλ,β
=
1
2
h
λµ
h
λµ,αβ
+
1
2
h
λµ

h
λµ,β
,
Γ
(2)λ
αβ,λ
= −
1
2
h
λµ
(h
αµ,βλ
+h
βµ,αλ
−h
αβ,µλ
)
(in the last line we used h
λµ

= 0). Finally, we tackle the term Γ
(1)λ
βν
Γ
(1)ν
αλ
. In this term, it helps to write
Γ
(1)λ
αβ
=
1
2
_
h
λ
α,β
+h
λ
β,α
−h

αβ
_
,
where again the indices are raised via η
µν
since we only need this term to first order. Then we can simplify this expression
by grouping together terms where α, β appear in similar positions:

(1)λ
βν
Γ
(1)ν
αλ
=
_
h
λ
β,ν
+h
λ
ν,β
−h

βν
_
_
h
ν
α,λ
+h
ν
λ,α
−h

αλ
_
(expand brackets) = h
λ
β,ν
h
ν
α,λ
+h
λ
β,ν
h
ν
λ,α
−h
λ
β,ν
h

αλ
+h
λ
ν,β
h
ν
α,λ
+h
λ
ν,β
h
ν
λ,α
−h
λ
ν,β
h

αλ
−h

βν
h
ν
α,λ
−h

βν
h
ν
λ,α
+h

βν
h

αλ
(move, rename λ, ν) = h

βλ
h

αν
+h
βλ,ν
h
λν

−h
λ
β,ν
h

αλ
+h
λν

h
αν,λ
+h
λν

h
λν,α
−h
λν

h
αλ,ν
−h
λ
β,ν
h

λα
−h
βν,λ
h
λν

+h

βλ
h

αν
(gather terms) = 2h

βλ
h

αν
+ (h
βλ,ν
−h
βν,λ
) h
λν

−2h
λ
β,ν
h

αλ
+h
λν

(h
αν,λ
−h
αλ,ν
) +h
λν

h
λν,α
(symmetry of h
µν
) = h
λν,α
h
λν

+ 2h

βλ
h

αν
−2h
λ
β,ν
h

αλ
.
Finally, we put together the expression for R
(2)
αβ
:
R
(2)
αβ
=
1
2
h
λµ
(−h
αµ,βλ
−h
βµ,αλ
+h
αβ,µλ
+h
λµ,αβ
) +
1
2
h
λµ

h
λµ,β

1
4
_
h
λν,α
h
λν

+ 2h

βλ
h

αν
−2h
λ
β,ν
h

αλ
_
=
1
2
h
λµ
(−h
αµ,βλ
−h
βµ,αλ
+h
αβ,µλ
+h
λµ,αβ
) +
1
4
h
λµ

h
λµ,β
+
1
2
_
h

αλ
h
λ
β,ν
−h

βλ
h

αν
_
. (35)
31
The Ricci scalar is
R
(2)
= η
αβ
R
(2)
αβ
=
1
2
h
λµ
h
λµ
+
1
4
h
λµ

h

λµ
+
1
2
_
h

αλ
h
αλ

−h
αλ,ν
h
αν,λ
_
=
3
4
h
λµ,α
h
λµ,α

1
2
h
αλ,ν
h
αν,λ
, (36)
where we again used the transverse traceless property of h
µν
and also h
αβ
= 0. Note that the first-order Ricci scalar
is zero, R
(1)
= 0, since T
αβ
= 0. For this reason we may use η
αβ
in Eq. (36), otherwise we would have to write
(η +h)
_
R
(1)
+R
(2)
_
and pick up a second-order term hR
(1)
.
Again, since R
(1)
= 0, we may use η
µν
rather than g
µν
to compute the Einstein tensor:
G
(2)
αβ
= R
(2)
αβ

1
2
η
αβ
R
(2)
.
We do not write the answer explicitly since it is a combination of Eqs. (35) and (36).
Now let us perform an averaging of the quantity G
(2)
αβ
over both space and time. In other words, we integrate G
(2)
αβ
over a 4-dimensional region such that h
µν
= 0 and h
µν,α
= 0 on the boundary of that region. Then ¸∂
µ
(...)¸ = 0 and so
we may integrate by parts, for example
¸A
µ
B
ν,α
¸ = −¸A
µ,α
B
ν
¸ ,
as long as A
µ
B
ν
contains first powers of h
αβ
or h
αβ,γ
, so that boundary terms vanish. Then, for example,
¸
h
λµ
h
λµ,αβ
_
= −
¸
h
λµ

h
λµ,β
_
, (37)
¸
h
λµ
h
αν,βλ
_
= −
_
h
λµ

h
αν,β
_
= 0,
¸h

αλ
h
βµ,ν
¸ = −¸h
αλ
h
βµ
¸ = 0,
by h
αβ
= 0 and by the transverse traceless property of h
µν
. Many terms cancel in this way; for instance,
¸
R
(2)
_
= 0.
Finally, we get
_
G
(2)
αβ
_
=
_
1
2
h
λµ
(−h
αµ,βλ
−h
βµ,αλ
+h
αβ,µλ
+h
λµ,αβ
) +
1
4
h
λµ

h
λµ,β
+
1
2
_
h

αλ
h
λ
β,ν
−h

βλ
h

αν
_
_
=
1
2
¸
h
λµ
h
λµ,αβ
_
+
1
4
¸
h
λµ

h
λµ,β
_
= −
1
4
¸
h
λµ

h
λµ,β
_
using Eq. (37). Finally, we obtain the required equation,
(GW)
T
µν
= −
1
8πG
_
G
(2)
αβ
_
=
1
32πG
¸
h
λµ

h
λµ,β
_
=
1
32πG
¸
h
ij

h
ij,β
_
.
9.4 Power of emitted radiation
To derive the relations
_
n
l
n
m
dΩ

=
1
3
δ
lm
,
_
n
l
n
m
n
k
n
r
dΩ

=
1
15

lm
δ
kr

lk
δ
mr

lk
δ
mr
),
let us consider the generating function
g

(q
l
) ≡
_
dΩ

exp
_
−in
l
q
l
¸
,
which is a function of a vector argument q
l
. After computing g

(q
l
) it will be easy to obtain integrals such as the above:
_
n
l
n
m
dΩ

= i

∂q
l
i

q
m
g

(q
j
)
¸
¸
¸
¸
qj=0
, etc.
The computation is easy if we introduce spherical coordinates with the z axis parallel to the vector q
l
, then n
l
q
l
= [q[ cos θ,
where [q[ ≡

q
l
q
l
, and then we have
g

(q
l
) =
1

_

0

_
π
0
dθ sin θ exp [−i [q[ cos θ] =
1


−2i sin [q[
−i [q[
=
sin [q[
[q[
= 1 −
1
3!
q
l
q
l
+
1
5!
(q
l
q
l
)
2

1
7!
(q
l
q
l
)
3
+...
32
We have used the Taylor expansion for convenience of evaluating derivatives at [q[ = 0. These derivatives can be found
as follows,
∂g

∂q
l
= −
2
3!
q
l
+
4
5!
q
l
[q[
2
−... =
_

1
3
+
1
30
[q[
2
_
q
l
,

2
g

∂q
k
∂q
l
=
_

1
3
+
1
30
[q[
2
_
δ
kl
+
1
15
q
l
q
k
,

3
g

∂q
j
∂q
k
∂q
l
=
1
15
(q
j
δ
kl
+q
k
δ
jl
+q
l
δ
jk
) ,

4
g

∂q
j
∂q
k
∂q
l
∂q
m
=
1
15

jm
δ
kl

km
δ
jl

lm
δ
jk
) .
Now we compute the intensity of radiation. The flux of radiation in the direction n
k
is
(GW)
T
0k
n
k
, and we need to
integrate this flux through a sphere of radius R:
dE
dt
= R
2
_
d
2

(GW)
T
0k
n
k
=
R
2
32πG
_
d
2

_
h
ij
,0
h
ij,k
_
n
k
.
The perturbation h
ij
is found from the Einstein equation. It was derived in the lecture that, in the leading order in 1/R,
we have
h
ij
= 2G
(TT)
¨
Q
ij
(t −[

R[)
[

R[
,
(TT)
Q
ij

_
P
ai
P
bj

1
2
P
ab
P
ij
_
Q
ab
.
The projection tensor P
ij
is defined in Problem 9.1. The tensor Q
ij
is defined by
Q
ij
(t) ≡
_
d
3
x
_
x
i
x
j

1
3
[x[
2
δ
ij
_
T
00
(x, t)
and is by definition trace-free, Q
ii
= 0. Thus we have
h
ij,k
=
16πG
R
(TT)
¨
Q
ij
(t −[R[)
R
k
R
,
dE
dt
=
G

_
d
2

_
(TT)
...
Q
ij
(TT)
...
Q
ij
_
.
It remains to compute the average over the sphere of
(TT)
...
Q
ij
(TT)
...
Q
ij
.
Consider any symmetric, trace-free tensor A
ij
instead of
...
Q
ij
; the transverse-traceless part of A
ij
is defined by
(TT)
A
ij

_
P
ai
P
bj

1
2
P
ab
P
ij
_
A
ab
.
Since A
ii
= 0, we have A
ab
P
ab
= −A
ab
n
a
n
b
and so
(TT)
A
ij
(TT)
A
ij
=
_
P
ai
P
bj

1
2
P
ab
P
ij
__
P
ci
P
dj

1
2
P
cd
P
ij
_
A
ab
A
cd
=
_
P
ac
P
bd

1
2
P
ab
P
cd
_
A
ab
A
cd
=
_
P
ac
P
bd

1
2
n
a
n
b
n
c
n
d
_
A
ab
A
cd
= A
ab
A
ab
−2A
ac
A
bc
n
a
n
b
+
1
2
A
ab
A
cd
n
a
n
b
n
c
n
d
.
After integration over the sphere, according to formulas derived above, we have (again note that δ
ab
A
ab
= 0 and A
ab
=
A
ba
)
1

_
d
2
ΩA
ac
A
bc
n
a
n
b
=
1
3
A
ac
A
bc
δ
ab
=
1
3
A
ab
A
ab
,
1

_
d
2
ΩA
ab
A
cd
n
a
n
b
n
c
n
d
=
1
15
A
ab
A
cd

ab
δ
cd

ac
δ
bd

ad
δ
bc
) =
2
15
A
ab
A
ab
,
and thus
1

_
d
2

(TT)
A
ij
(TT)
A
ij
= A
ab
A
ab
_
1 −
2
3
+
1
2
2
15
_
=
2
5
A
ab
A
ab
.
33
Finally, substituting
...
Q
ij
instead of A
ij
, we find
dE
dt
=
G
2
1

_
d
2

_
(TT)
...
Q
ij
(TT)
...
Q
ij
_
=
G
5
¸...
Q
ij
...
Q
ij
_
. (38)
The angular brackets ¸¸ indicate that we must perform an averaging over spacetime domains. This means, for us, that we
need to average over time (since Q
ij
is a function only of time). Averaging is performed over timescales larger than the
typical timescale of change in the source. For instance, if the source is a rotating body, then averaging must be performed
over several periods of rotation.
10 Derivation: gravitational waves in flat spacetime
The metric is assumed to be of the form
g
µν
= η
µν
+h
µν
,
where η
µν
= diag(1, −1, −1, −1) is the Minkowski metric for flat space and h
µν
is a small perturbation which is assumed
to fall off to zero quickly at infinity. We start with a 3+1 decomposition of the metric perturbation h
µν
and compute
the Einstein tensor (see Problems 7.2, 7.3, 7.4) in terms of the perturbation variables Φ, Ψ, etc. We also decompose the
matter energy-momentum tensor T
µν
and obtain the Einstein equations separately for each component (8.5). The result
is that (a) the variables E, B, F
i
can be set to zero by choosing a coordinate system; (b) if there is no matter (vacuum)
the scalar and vector components of the metric perturbation are equal to zero; (c) the tensor component h
ij
satisfies the
wave equation (31).
Solutions of the wave equation in four dimensions with retarded boundary condition can be written using the known
Green’s function. For instance, if
f(t, r) = A(t, r) ⇒ f(t, R) = −
1

_
d
3
r
A(t −[r −R[ , r)
[r −R[
.
We will use this formula for f ≡
(T)
h
ij
and A ≡ 16πG
(T)
T
i
j
. Now, we are interested in describing the radiation sent far
away by a matter distribution, so we take the limit [R[ ≫[r[, and then we can approximately set
(T)
h
ij
≈ −
4G
[R[
_
d
3
r
(T)
T
i
j
(t −[r −R[ , r).
Now we use a trick (See Hobson-Efstathiou-Lasenby, ¸17.9) to express the components T
j
i
through T
0
0
; it is much easier
to compute with T
0
0
because this is just the energy density of matter. Consider first the tensor T
i
j
rather than its
transverse-traceless part
(T)
T
i
j
. The trick is to write the integral (out of sheer luck)
_
d
3
r ∂
a

b
_
r
i
r
j
_
T
ab
= 2
_
d
3
r T
ij
.
Then we integrate by parts and use the conservation laws (32),
T
i
j,i
= −T
0
j,0
, T
i
j,ij
= −T
0
j,0j
= T
0
0,00

¨
T
0
0
; T
ij
,ij
= −
¨
T
0
0
.
2
_
d
3
r T
ij
=
_
d
3
r T
ab
,ab
r
i
r
j
= −
_
d
3
r r
i
r
j
¨
T
0
0
.
Now, we need to obtain the transverse-traceless part of the tensor. In principle, we have the formulas for this (see
Problem 8.5). But they are very complicated. A shortcut is to notice that the projection operator P
ab
does the job
(Problems 9.1 and 9.2), at least in the leading order in 1/[R[. (We are only interested in everything to leading order in
1/ [R[ since all smaller terms will not give any flux of radiated energy.) The result is
(T)
h
ik
(R, t) =
2G
[R[
(TT)
¨
Q
ik
(t −[R[),
Q
ik
(t) ≡
_
d
3
r T
0
0
(r, t)
_
r
i
r
k

1
3
r
2
δ
ik
_
. (39)
The tensor Q
ik
is the quadrupole moment of energy distribution; it is a traceless and symmetric tensor. In principle, we
could just use the integral
_
d
3
r T
0
0
(r, t)r
i
r
k
, (40)
because the transverse-traceless parts of (40) and of Q
ik
are the same, but it is more convenient to use Q
ik
.
Since we found the tensor perturbation
(T)
h
ij
, now we would like to compute the energy radiated in the gravitational
waves. For this we need the energy-momentum tensor of gravitational waves. This is a rather nontrivial object, since
in general the gravitational field does not have any energy-momentum tensor. In the case of gravitational waves in
flat background spacetime, one can define some quantity
(GW)
T
µν
which looks like the energy-momentum tensor of
34
gravitational waves (but actually is not even a generally covariant tensor). We will compute this quantity below. This
quantity is useful because it gives the correct value of the energy after one integrates over a large region of spacetime.
The real justification for using this procedure is complicated and is beyond the scope of this introductory course of
General Relativity. We will only show a heuristic justification, which is the following. Gravitation is sensitive to every
kind of energy, because the energy-momentum tensor acts as a “source” for gravity (it is on the right-hand side of the
Einstein equation). So gravitation should be also sensitive to the energy in gravitational waves. One expects that the
energy-momentum tensor for gravitational waves,
(GW)
T
µν
(if we know how to compute it), will act as an additional
source for gravity, like every other energy-momentum tensor for other kinds of matter. We will guess the formula for
(GW)
T
µν
as follows. We can write the Einstein equation and expand it in powers of the perturbation h
µν
:
G
α
β

µν
+h
µν
] = G
(1)α
β
[h] +G
(2)α
β
[h] +... = 8πGT
α
β
. (41)
Here G
(1)
is the first-order Einstein tensor, G
(2)
is the second-order etc. First we solve only to first-order in h (this is
what we have been doing so far) and then we will get an approximate solution h
(1)
µν
:
G
(1)α
β
[h
(1)
] = 8πGT
α
β
. (42)
This solution disregards the effect of gravitational waves and only takes into account the effect of matter T
α
β
. We can
try to get a more precise solution by using the second-order terms in Eq. (41). Then we will get a correction h
(2)
to the
solution; the solution g = η +h
(1)
+h
(2)
will be more precise. From Eq. (41) we find
G
(1)α
β
[h
(1)
+h
(2)
] +G
(2)α
β
[h
(1)
] = 8πGT
α
β
.
Now this is similar to Eq. (42), but it looks as if there is an additional term in the energy-momentum tensor, which we
may rewrite as
G
(1)α
β
[h
(1)
+h
(2)
] = 8πG
_
T
α
β
+
(GW)
T
α
β
_
,
(GW)
T
α
β
≡ −
1
8πG
G
(2)α
β
[h
(1)
].
This motivates us to say that the EMT for gravitational waves is given by this formula. But of course this is not a real
derivation because this does not show why the quantity
(GW)
T
µν
has anything to do with the energy carried by waves.
The second-order terms G
(2)α
β
are computed in Problem 9.3. The result is used to compute the power radiated in
gravitational waves (Problem 9.4). Note that the calculation of G
(2)α
β
uses averaging over spacetime in an essential way.
Thus, the result is an averaged power radiated during a long time—much longer than the typical time scale of change
in the sources—and averaged over large distances, much larger than the typical length scale of the sources. This kind of
averaging is assumed in Eq. (38). It remains unclear exactly how one performs averaging over space and time.
One can use the formula (38) to compute the gravitational radiation emitted by nonrelativistic matter far away from
those places where the matter is contained. The distribution of the energy density, T
0
0
(r, t), should be given. Then one
computes the tensor Q
ij
according to Eq. (39), by integrating over space where the matter is contained. Finally, one
computes the third derivative
...
Q
ij
, the trace, and averages over long time, as indicated in Eq. (38). If we want to insert
factors of c, we replace G by Gc
−9
.
A GNU Free Documentation License
Version 1.2, November 2002
Copyright (c) 2000,2001,2002 Free Software Foundation, Inc. 59 Temple Place, Suite 330, Boston, MA 02111-1307,
USA
Everyone is permitted to copy and distribute verbatim copies of this license document, but changing it is not allowed.
Preamble
The purpose of this License is to make a manual, textbook, or other functional and useful document free in the sense
of freedom: to assure everyone the effective freedom to copy and redistribute it, with or without modifying it, either
commercially or noncommercially. Secondarily, this License preserves for the author and publisher a way to get credit
for their work, while not being considered responsible for modifications made by others.
This License is a kind of “copyleft”, which means that derivative works of the document must themselves be free in
the same sense. It complements the GNU General Public License, which is a copyleft license designed for free software.
We have designed this License in order to use it for manuals for free software, because free software needs free
documentation: a free program should come with manuals providing the same freedoms that the software does. But this
License is not limited to software manuals; it can be used for any textual work, regardless of subject matter or whether
it is published as a printed book. We recommend this License principally for works whose purpose is instruction or
reference.
35
A.0 Applicability and definitions
This License applies to any manual or other work, in any medium, that contains a notice placed by the copyright holder
saying it can be distributed under the terms of this License. Such a notice grants a world-wide, royalty-free license,
unlimited in duration, to use that work under the conditions stated herein. The “Document”, below, refers to any such
manual or work. Any member of the public is a licensee, and is addressed as “you”. You accept the license if you copy,
modify or distribute the work in a way requiring permission under copyright law.
A “Modified Version” of the Document means any work containing the Document or a portion of it, either copied
verbatim, or with modifications and/or translated into another language.
A “Secondary Section” is a named appendix or a front-matter section of the Document that deals exclusively with
the relationship of the publishers or authors of the Document to the Document’s overall subject (or to related matters)
and contains nothing that could fall directly within that overall subject. (Thus, if the Document is in part a textbook of
mathematics, a Secondary Section may not explain any mathematics.) The relationship could be a matter of historical
connection with the subject or with related matters, or of legal, commercial, philosophical, ethical or political position
regarding them.
The “Invariant Sections” are certain Secondary Sections whose titles are designated, as being those of Invariant
Sections, in the notice that says that the Document is released under this License. If a section does not fit the above
definition of Secondary then it is not allowed to be designated as Invariant. The Document may contain zero Invariant
Sections. If the Document does not identify any Invariant Sections then there are none.
The “Cover Texts” are certain short passages of text that are listed, as Front-Cover Texts or Back-Cover Texts, in the
notice that says that the Document is released under this License. A Front-Cover Text may be at most 5 words, and a
Back-Cover Text may be at most 25 words.
A “Transparent” copy of the Document means a machine-readable copy, represented in a format whose specification
is available to the general public, that is suitable for revising the document straightforwardly with generic text editors
or (for images composed of pixels) generic paint programs or (for drawings) some widely available drawing editor, and
that is suitable for input to text formatters or for automatic translation to a variety of formats suitable for input to text
formatters. A copy made in an otherwise Transparent file format whose markup, or absence of markup, has been arranged
to thwart or discourage subsequent modification by readers is not Transparent. An image format is not Transparent if
used for any substantial amount of text. A copy that is not “Transparent” is called “Opaque”.
Examples of suitable formats for Transparent copies include plain ASCII without markup, Texinfo input format,
L
A
T
E
X input format, SGML or XML using a publicly available DTD, and standard-conforming simple HTML, PostScript
or PDF designed for human modification. Examples of transparent image formats include PNG, XCF and JPG. Opaque
formats include proprietary formats that can be read and edited only by proprietary word processors, SGML or XML
for which the DTD and/or processing tools are not generally available, and the machine-generated HTML, PostScript or
PDF produced by some word processors for output purposes only.
The “Title Page” means, for a printed book, the title page itself, plus such following pages as are needed to hold,
legibly, the material this License requires to appear in the title page. For works in formats which do not have any title
page as such, “Title Page”means the text near the most prominent appearance of the work’s title, preceding the beginning
of the body of the text.
A section “Entitled XYZ” means a named subunit of the Document whose title either is precisely XYZ or contains
XYZ in parentheses following text that translates XYZ in another language. (Here XYZ stands for a specific section
name mentioned below, such as “Acknowledgements”, “Dedications”, “Endorsements”, or “History”.) To “Preserve the
Title” of such a section when you modify the Document means that it remains a section “Entitled XYZ” according to this
definition.
The Document may include Warranty Disclaimers next to the notice which states that this License applies to the
Document. These Warranty Disclaimers are considered to be included by reference in this License, but only as regards
disclaiming warranties: any other implication that these Warranty Disclaimers may have is void and has no effect on the
meaning of this License.
A.1 Verbatim copying
You may copy and distribute the Document in any medium, either commercially or noncommercially, provided that this
License, the copyright notices, and the license notice saying this License applies to the Document are reproduced in all
copies, and that you add no other conditions whatsoever to those of this License. You may not use technical measures
to obstruct or control the reading or further copying of the copies you make or distribute. However, you may accept
compensation in exchange for copies. If you distribute a large enough number of copies you must also follow the conditions
in section A.2.
You may also lend copies, under the same conditions stated above, and you may publicly display copies.
A.2 Copying in quantity
If you publish printed copies (or copies in media that commonly have printed covers) of the Document, numbering more
than 100, and the Document’s license notice requires Cover Texts, you must enclose the copies in covers that carry, clearly
and legibly, all these Cover Texts: Front-Cover Texts on the front cover, and Back-Cover Texts on the back cover. Both
covers must also clearly and legibly identify you as the publisher of these copies. The front cover must present the full title
36
with all words of the title equally prominent and visible. You may add other material on the covers in addition. Copying
with changes limited to the covers, as long as they preserve the title of the Document and satisfy these conditions, can
be treated as verbatim copying in other respects.
If the required texts for either cover are too voluminous to fit legibly, you should put the first ones listed (as many as
fit reasonably) on the actual cover, and continue the rest onto adjacent pages.
If you publish or distribute Opaque copies of the Document numbering more than 100, you must either include a
machine-readable Transparent copy along with each Opaque copy, or state in or with each Opaque copy a computer-
network location from which the general network-using public has access to download using public-standard network
protocols a complete Transparent copy of the Document, free of added material. If you use the latter option, you must
take reasonably prudent steps, when you begin distribution of Opaque copies in quantity, to ensure that this Transparent
copy will remain thus accessible at the stated location until at least one year after the last time you distribute an Opaque
copy (directly or through your agents or retailers) of that edition to the public.
It is requested, but not required, that you contact the authors of the Document well before redistributing any large
number of copies, to give them a chance to provide you with an updated version of the Document.
A.3 Modifications
You may copy and distribute a Modified Version of the Document under the conditions of sections A.1 and A.2 above,
provided that you release the Modified Version under precisely this License, with the Modified Version filling the role of
the Document, thus licensing distribution and modification of the Modified Version to whoever possesses a copy of it. In
addition, you must do these things in the Modified Version:
A. Use in the Title Page (and on the covers, if any) a title distinct from that of the Document, and from those of
previous versions (which should, if there were any, be listed in the History section of the Document). You may use the
same title as a previous version if the original publisher of that version gives permission.
B. List on the Title Page, as authors, one or more persons or entities responsible for authorship of the modifications
in the Modified Version, together with at least five of the principal authors of the Document (all of its principal authors,
if it has fewer than five), unless they release you from this requirement.
C. State on the Title page the name of the publisher of the Modified Version, as the publisher.
D. Preserve all the copyright notices of the Document.
E. Add an appropriate copyright notice for your modifications adjacent to the other copyright notices.
F. Include, immediately after the copyright notices, a license notice giving the public permission to use the Modified
Version under the terms of this License, in the form shown in the Addendum below.
G. Preserve in that license notice the full lists of Invariant Sections and required Cover Texts given in the Document’s
license notice.
H. Include an unaltered copy of this License.
I. Preserve the section Entitled “History”, Preserve its Title, and add to it an item stating at least the title, year, new
authors, and publisher of the Modified Version as given on the Title Page. If there is no section Entitled “History” in the
Document, create one stating the title, year, authors, and publisher of the Document as given on its Title Page, then add
an item describing the Modified Version as stated in the previous sentence.
J. Preserve the network location, if any, given in the Document for public access to a Transparent copy of the
Document, and likewise the network locations given in the Document for previous versions it was based on. These may
be placed in the “History” section. You may omit a network location for a work that was published at least four years
before the Document itself, or if the original publisher of the version it refers to gives permission.
K. For any section Entitled “Acknowledgements” or “Dedications”, Preserve the Title of the section, and preserve in
the section all the substance and tone of each of the contributor acknowledgements and/or dedications given therein.
L. Preserve all the Invariant Sections of the Document, unaltered in their text and in their titles. Section numbers or
the equivalent are not considered part of the section titles.
M. Delete any section Entitled “Endorsements”. Such a section may not be included in the Modified Version.
N. Do not retitle any existing section to be Entitled “Endorsements” or to conflict in title with any Invariant Section.
O. Preserve any Warranty Disclaimers.
If the Modified Version includes new front-matter sections or appendices that qualify as Secondary Sections and contain
no material copied from the Document, you may at your option designate some or all of these sections as invariant. To
do this, add their titles to the list of Invariant Sections in the Modified Version’s license notice. These titles must be
distinct from any other section titles.
You may add a section Entitled “Endorsements”, provided it contains nothing but endorsements of your Modified
Version by various parties—for example, statements of peer review or that the text has been approved by an organization
as the authoritative definition of a standard.
You may add a passage of up to five words as a Front-Cover Text, and a passage of up to 25 words as a Back-Cover
Text, to the end of the list of Cover Texts in the Modified Version. Only one passage of Front-Cover Text and one of
Back-Cover Text may be added by (or through arrangements made by) any one entity. If the Document already includes
a cover text for the same cover, previously added by you or by arrangement made by the same entity you are acting on
behalf of, you may not add another; but you may replace the old one, on explicit permission from the previous publisher
that added the old one.
37
The author(s) and publisher(s) of the Document do not by this License give permission to use their names for publicity
for or to assert or imply endorsement of any Modified Version.
Combining documents
You may combine the Document with other documents released under this License, under the terms defined in section
4 above for modified versions, provided that you include in the combination all of the Invariant Sections of all of the
original documents, unmodified, and list them all as Invariant Sections of your combined work in its license notice, and
that you preserve all their Warranty Disclaimers.
The combined work need only contain one copy of this License, and multiple identical Invariant Sections may be
replaced with a single copy. If there are multiple Invariant Sections with the same name but different contents, make the
title of each such section unique by adding at the end of it, in parentheses, the name of the original author or publisher
of that section if known, or else a unique number. Make the same adjustment to the section titles in the list of Invariant
Sections in the license notice of the combined work.
In the combination, you must combine any sections Entitled “History” in the various original documents, forming
one section Entitled “History”; likewise combine any sections Entitled “Acknowledgements”, and any sections Entitled
“Dedications”. You must delete all sections Entitled “Endorsements.”
Collections of documents
You may make a collection consisting of the Document and other documents released under this License, and replace the
individual copies of this License in the various documents with a single copy that is included in the collection, provided
that you follow the rules of this License for verbatim copying of each of the documents in all other respects.
You may extract a single document from such a collection, and distribute it individually under this License, provided
you insert a copy of this License into the extracted document, and follow this License in all other respects regarding
verbatim copying of that document.
Aggregation with independent works
A compilation of the Document or its derivatives with other separate and independent documents or works, in or on a
volume of a storage or distribution medium, is called an “aggregate”if the copyright resulting from the compilation is not
used to limit the legal rights of the compilation’s users beyond what the individual works permit. When the Document is
included an aggregate, this License does not apply to the other works in the aggregate which are not themselves derivative
works of the Document.
If the Cover Text requirement of section A.2 is applicable to these copies of the Document, then if the Document is less
than one half of the entire aggregate, the Document’s Cover Texts may be placed on covers that bracket the Document
within the aggregate, or the electronic equivalent of covers if the Document is in electronic form. Otherwise they must
appear on printed covers that bracket the whole aggregate.
Translation
Translation is considered a kind of modification, so you may distribute translations of the Document under the terms
of section A.3. Replacing Invariant Sections with translations requires special permission from their copyright holders,
but you may include translations of some or all Invariant Sections in addition to the original versions of these Invariant
Sections. You may include a translation of this License, and all the license notices in the Document, and any Warrany
Disclaimers, provided that you also include the original English version of this License and the original versions of those
notices and disclaimers. In case of a disagreement between the translation and the original version of this License or a
notice or disclaimer, the original version will prevail.
If a section in the Document is Entitled“Acknowledgements”, “Dedications”, or“History”, the requirement (section A.3)
to Preserve its Title (section A.0) will typically require changing the actual title.
Termination
You may not copy, modify, sublicense, or distribute the Document except as expressly provided for under this License.
Any other attempt to copy, modify, sublicense or distribute the Document is void, and will automatically terminate your
rights under this License. However, parties who have received copies, or rights, from you under this License will not have
their licenses terminated so long as such parties remain in full compliance.
Future revisions of this license
The Free Software Foundation may publish new, revised versions of the GNU Free Documentation License from time to
time. Such new versions will be similar in spirit to the present version, but may differ in detail to address new problems
or concerns. See http://www.gnu.org/copyleft/.
Each version of the License is given a distinguishing version number. If the Document specifies that a particular
numbered version of this License “or any later version” applies to it, you have the option of following the terms and
38
conditions either of that specified version or of any later version that has been published (not as a draft) by the Free
Software Foundation. If the Document does not specify a version number of this License, you may choose any version
ever published (not as a draft) by the Free Software Foundation.
ADDENDUM: How to use this License for your documents
To use this License in a document you have written, include a copy of the License in the document and put the following
copyright and license notices just after the title page:
Copyright (c) <year> <your name>. Permission is granted to copy, distribute and/or modify this document under
the terms of the GNU Free Documentation License, Version 1.2 or any later version published by the Free Software
Foundation; with no Invariant Sections, no Front-Cover Texts, and no Back-Cover Texts. A copy of the license is
included in the section entitled “GNU Free Documentation License”.
If you have Invariant Sections, Front-Cover Texts and Back-Cover Texts, replace the “with...Texts.” line with this:
with the Invariant Sections being <list their titles>, with the Front-Cover Texts being <list>, and with the Back-Cover
Texts being <list>.
If you have Invariant Sections without Cover Texts, or some other combination of the three, merge those two alterna-
tives to suit the situation.
If your document contains nontrivial examples of program code, we recommend releasing these examples in parallel
under your choice of free software license, such as the GNU General Public License, to permit their use in free software.
Copyright
Copyright (c) 2000, 2001, 2002 Free Software Foundation, Inc. 59 Temple Place, Suite 330, Boston, MA 02111-1307, USA
Everyone is permitted to copy and distribute verbatim copies of this license document, but changing it is not allowed.
39

7 Weak gravitational fields 7.1 Gravitational bending of light 7.2 Einstein tensor for weak field 7.3 Gravitational perturbations I 7.4 Gravitational perturbations II

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

. . . . . . . . . . . . . . .

23 23 24 24 25 27 27 27 27 28 28 28 29 30 30 30 31 32 34

8 Gravitational radiation I 8.1 Gauge invariant variables . . . . . . . . . 8.2 Detecting gravitational waves . . . . . . . 8.2.1 Using distances between particles . 8.2.2 Using geodesic deviation equation 8.3 Poisson equation . . . . . . . . . . . . . . 8.4 Metric perturbations 1 . . . . . . . . . . . 8.5 Metric perturbations 2 . . . . . . . . . . .

9 Gravitational radiation II 9.1 Projection of the matter tensor . . . . . . . . . . 9.2 Matter sources . . . . . . . . . . . . . . . . . . . 9.3 Energy-momentum tensor of gravitational waves 9.4 Power of emitted radiation . . . . . . . . . . . . .

10 Derivation: gravitational waves in flat spacetime A GNU Free Documentation License A.0 Applicability and definitions . . . . A.1 Verbatim copying . . . . . . . . . . A.2 Copying in quantity . . . . . . . . A.3 Modifications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

35 36 36 36 37

This is a complete set of solutions to the exercises offered in the lecture course “General Relativity” (Allgemeine Relativit¨tstheorie) taught in the winter semester of 2006-2007 by Prof. V. F. Mukhanov at the Department of Physics, a Ludwig-Maximilians-University, in Munich, Germany. A good textbook corresponding to the level of this course: General Relativity: An Introduction for Physicists by M. P. Hobson, G. P. Efstathiou, and A. N. Lasenby (Cambridge University Press, 2006). Many of the exercises and all of the solutions were authored by Dr. Sergei Winitzki. This file contains only the solutions; the problem statements are available separately. This text is copyrighted by Sergei Winitzki (2008) and distributed under the GNU Free Documentation License. In particular, it is permitted to print, copy, and distribute the entire text (or parts thereof) at no charge. The text may be modified at will, and modified versions may be distributed again under the terms of the same license. The license also stipulates that the entire machine-readable and human-editable source code for this text should be made available when A the text is distributed in any form. The full text of the license is in Appendix A. The full L TEX source for this text is attached inside the PDF file.

1
1.1

Coordinates and 1-forms
Invertible transformations

The inverse function theorem guarantees that the equations ξ α = ξ α (x) are solvable near a point x0 if det ∂ξ α (x)/∂xβ = 0 at x0 . Under this condition, the coordinate transformation is invertible at x0 . Note: we are inverting not just one function ξ = ξ(x), but we are determining x from a system of n equations, say ξ α (x) = C α , where C α are n given values.

1.2

Examples of coordinate transformations

1. a) Since x = u 1 + v 2 + u3 /3, it is clear that x has range (−∞, +∞) for any fixed value of v as u varies in the range (−∞, +∞). Similarly, y has the range (−∞, +∞). To verify that the coordinate system (x, y) covers the entire plane, it is sufficient to show that x has the full range at every fixed value of y. It is sufficient to consider y0 > 0 (else change v → −v). At fixed y = y0 > 0, we have y0 = v + vu2 + v 3 /3 and thus the admissible values of u are from −∞ to +∞,

2

1 3 while the admissible values of v are from 0 to v = vmax such that y0 = vmax + 3 vmax . Then 1 y0 − v − 3 v 3 v

u=±

(we have

y0 − v − 1 v 3 3 ≥ 0 for 0 < v < vmax ), v y0 − v − 1 v 3 3 v

x = ± 1 + v2 + =±

y0 − v − 1 v 3 3 3v

2 8 2 y0 + v + 3 9 3v

1 y0 − v − 3 v 3 . v

We have now expressed x as a function of v, i.e. x = x(v). When v varies from 0 to vmax , x(v) varies from ±∞ to 0. Since x(v) is nonsingular for v > 0, it follows that x has the full range. Therefore, the coordinates (x, y) cover the entire two-dimensional plane. b) The coordinate transformation is nonsingular if det Compute: det
∂x ∂u ∂x ∂v ∂y ∂u ∂y ∂v

∂ (x, y) = 0. ∂ (u, v)

= det

1 + u2 + v 2 2uv

2uv 1 + u2 + v 2
2

(1)

= 1 + 2 u2 + v 2 + u2 − v 2

> 0.

The coordinates are singular if r = 0 or θ = 0. c) The singularity at r = 0 is due to the fact that the set {r = 0, θ, φ} corresponds to a single point x = y = z = 0. This is similar to the singularity of the spherical coordinates at r = 0. Points along the cone |z| = x2 + y 2 are not covered because they correspond to θ → ∞, r → 0. The singularity at θ = 0, r = 0 is due to the fact that the set {r, θ = 0, φ} corresponds to the point x = y = 0, z = r at fixed r = 0. This is similar to the polar coordinate singularity. 3. a) To determine the range, note that r sin θ ≥ 0 for the given range of θ and r. However, this is immaterial since the factors cos φ and sin φ will make x, y cover the full range (−∞, +∞). The coordinates (x, y, z) are a slight modification of the standard spherical coordinates. These coordinates cover the whole space (x, y, z). b) Compute the determinant:   − sin θ cos φ −r cos θ cos φ r sin θ sin φ det  − sin θ sin φ −r cos φ sin φ −r sin θ cos φ  = r2 sin θ. cos θ −r sin θ 0 This is nonzero unless r = 0 or θ = 0. c) The singularities are completely analogous to those in the spherical coordinates.

Therefore there are no singular points. 2. a) To determine the range, first consider φ = 0. Then x = r sinh θ, y = 0, z = r cosh θ. It is clear that z 2 − x2 = r2 . Since r ≥ 0, the coordinates (x, y, z) cover only the domain |z| > |x|. With arbitrary φ, it is clear that the coordinates (x, y, z) cover the domain |z| > x2 + y 2 . b) Compute the determinant:   sinh θ cos φ r cosh θ cos φ −r sinh θ sin φ det  sinh θ sin φ r cosh θ sin φ r sinh θ cos φ  = r2 sinh θ. cosh θ r sinh θ 0

1.3

Basis in tangent space

∂ Suppose that the vectors eα = ∂xα are linearly dependent, then there exist constants cα , not all zero, such that the vector α c eα equals zero. Act with this vector on the coordinate function x1 :

cα eα x1 = cα

∂ 1 x = c1 . ∂xα

By assumption, cα eα = 0, therefore c1 = 0. It follows that every cα equals zero, contradicting the assumption.

1.4

Differentials of functions as 1-forms
d(x) = dx, d (xy) = xdy + ydx, d 4x2 y + x3 z = 8xy + 3x2 z dx + 4x2 dy + x3 dz, d x2 = 2xdx, d (x + y) = dx + dy, xdx + ydy d 3 x2 + y 2 = 3 . x2 + y 2

3

where xα (x) is considered a scalar function in the coordinates xβ . 4. 2 1 + (x ± y)2 1 1+ dx ± dy . y) is a constant. Two 1-forms d(ex cos y). If this were not so.6 Linearly independent 1-forms d (ex cos y) = ex cos ydx − ex sin ydy.7 Transformation law for 1-forms d˜α = x ∂ xα β ˜ dx . Now. 3. we find that every other cα = 0.. ∂ ∂xβ = δα β Note that the relation is the definition of how the 1-form dxα acts on vectors ∂/∂xβ . det ex cos y ex sin y −ex sin y ex cos y = e2x = 0.4). is merely a different interpretation of the definition of the 1-form ˜ ˜ ˜ d˜α (see Problem 1. we can easily show that h(x.Now let us compute dh by first finding d (arctan (x ± y)) = d arctan 2x x2 − y 2 − 1 = = Adding these together and noting that 1 + (x + y) we get dh(x. y because 2. ∂xβ The transformation law for 1-forms. dy + dz are always linearly independent. dx + dz. 1. and the following determinant is always nonzero. d (ex sin y) = ex sin ydx + ex cos ydy. dxn . 8dx are linearly independent if the following determinant is nonzero. y) = d arctan (x + y) + arctan(x − y) + arctan 2x x2 − y 2 − 1 = 0. 1 = c1 . it is clear that any 1-form is decomposed as a linear combination of the 1-forms dx1 . dy − dz. 2dx 4x (xdx − ydy) − 2 x2 − y 2 − 1 (x2 − y 2 − 1) . Act with this on a vector ∂/∂x1 and obtain ∂ ∂ 0 = 0. By using the tangent sum rule. 1 = cα dxα . dz − dx are always linearly dependent (their sum is zero). which contradicts the assumption. Three 1-forms dx + dy. 1+y 8 −2xy 0 = 16xy.. Three 1-forms dx − dy. ∂x ∂x Therefore c1 = 0. 1. 4x2 (x2 −y 2 −1)2 −2 x2 + y 2 + 1 dx + 4xydy (x2 − y 2 − 1)2 + 4x2 1 + (x − y) 2 = x2 − y 2 − 1 2 + 4x2 . x 4 .5 Basis in cotangent space dxα . . This means that h(x. det This happens for xy = 0. 1.. 1. It remains to show that all these forms are linearly independent. under a coordinate transformation xα → xα = xα (xβ ). y) = 0. Similarly. d(ex sin y) are linearly independent for every x. Two 1-forms (1 + y)dx − 2xydy. there would exist a linear combination cα dxα = 0 such that not all cα = 0.

and computing dx. y0 . For example. Since 0 ≤ α < π . a point with maximum acceleration is x0 = σ. x = ± sin2 θ. The singularities at these points are not merely coordinate singularities 2 that disappear when choosing a different coordinate system. Then √ 1 − 1 − 4z 2 1 − cos 2θ =± . so φ = 0 or φ = π. intersecting itself at x = y = z = 0. dz through the forms dθ and dφ: dx = 2a sin θ cos θ cos φdθ − a sin2 θ sin φdφ. i. The tangent plane at a point (x0 . then the acceleration is g sin α (from elementary mechanics). (nx . The rotational symmetry around the z axis leads to a “cusp” at θ = 0: the surface has a sharp corner and the metric cannot be made smooth and nondegenerate by any choice of local coordinates. 0. Hence. equivalently. we need to maximize α or. y0 = 0.e. which equals 2 ∂z rh 1 2 −2 = 2 e− 2 r σ . 1. σ x −1 z e 2 + = 0. 1 . 2 5 . (du + dv) 1 + (u + v)2 dx + dy 1 =− d 2 =− x+y 2 2 1 (x + y) (u + v) 1 + 3 (u + v) b) The component transformation matrix is given in Eq. (1). x = ± 2 2 This is a union of two circles touching at x = z = 0. The metric is degenerate at θ = 0 and θ = π . σ h 1B Induced metric The induced metric is found by taking the expression for the bulk metric. dz = 2a cos 2θdθ. 2. y = sin θ cos θ. a) First compute dx and dy: Then it is easy to compute xdx + ydy. Therefore g = dx2 + dy 2 + dz 2 = a2 dθ2 + a2 (sin θ)4 dφ2 .9 Supplementary problem sheet 1A Tangent plane If the tangent plane is at angle α with the horizontal. The y = 0 section of the surface corresponds to sin φ = 0. z0 = −he− 2 . 1 The maximum of ∂z/∂r is at r0 = σ.8 Examples of transformations dx = 1 + u2 + v 2 du + (2uv) dv. The situation near θ = π is similar. ny .1 ∂x ∂y = x0 . let us visualize the surface in a neighborhood of θ = 0. where (nx . but the reason is subtle.1.z0 e− 2 1 h . dy = (2uv) du + 1 + u2 + v 2 dv. tan α. z0 ) is given by the equation nx (x − x0 ) + ny (y − y0 ) + nz (z − z0 ) = 0. nz ) are the components of the normal vector. ny . dy. . dy = 2a sin θ cos θ sin φdθ + a sin2 θ cos φdφ. nz ) = Therefore the equation of the tangent plane is ∂z ∂z . To figure out the nature of these singularities. the surface is a torus with zero inner radius.etc. For instance. g = dx2 + dy 2 + dz 2 . ∂r σ r≡ x2 + y 2 .y0 .

v). √ 2 − sin u 2 − sin u 2 − sin u . √ Therefore the line is a circle of radius 1 centered at x = 2. 2 x2 + y 2 − √ 2 x2 + y 2 y.z = √ . For y = 0. ny . Therefore. the normal vector (up to a constant factor C) can be found as (nx .1C Embedded waves The surface is defined by cos v . 2 − sin u 2 − sin u Also 2. sin v √ cos v √ 2 2. we have x2 + y 2 − √ 2 . 2z . ∂u ∂v etc. (nx . z) = 0. x2 + y 2 (this factor is chosen for simplicity). y. ∂u 2 − sin u √ ∂z 1 − 2 sin u = √ 2. = √ ∂v 2 − sin u ∂z = 0. 2 The equation of the tangent plane at point x0 is nα xα − xα = 0. ∂x ∂y ∂z =C 2 x2 + y 2 − √ 2 x2 + y 2 x. we eliminate u from these equations and find sin u = √ 1 2− . nz ) = C Multiplying by C ≡ 1 2 ∂F ∂F ∂F . = −√ ∂v 2 − sin u cos v ∂y . It may be described by the equation √ 2 x2 + y 2 − 2 + z 2 − 1 = 0. ∂v ∂x ∂x du + dv. z = 0 . y) is obtained from 2 − sin u the full surface is a rotation surface. z) plane. x= √ 2 − sin u 1. this becomes nα = (nx . 2 − sin u cos u z=√ . Note that 1 x2 + y 2 = √ > 0. This circle does not intersect the z axis since √ √ −1 by multiplying with cos v and sin v. Therefore. ny . (0) where nα must be computed at x = x0 . Since the surface is now given by an equation of the form F (x. The resulting surface is a torus. the figure in the (x. 2 − sin u x2 + y2 − √ 2= 1− √ √ √ 2 2 − sin u 2 sin u − 1 √ = √ . we have v = 0 and thus sin v y= √ . 2 − sin u To visualize this line in the (x. nz ) = x Expressed through the coordinates (u. 6 . Now we see that (x.y x2 + y 2 − √ 2 . We compute ∂x cos v cos u = √ 2. ny . nz ) = √ √ 2 sin u − 1 2 sin u − 1 cos u .z x2 + y 2 . . where we need to use the x coordinate as the radius. 2 − sin u 2 − sin u + z 2 = 1. ∂u 2 − sin u Now we can expand dx = ∂x sin v . z) plane needs to be rotated around the z axis. x x− √ 2 2 1 cos u x= √ . 2 > 1. ∂u 2 − sin u sin v cos u ∂y = √ 2. 3.

e. ∂v ∂v ∂x ∂v ∂y ∂v ∂z 2 = 1. √ 2 sin u − 1 sin u + cos u − 2 = cos v √ √ 2 sin v cos v 2 2 − sin u 2 − sin u √ √ √ sin2 v cos u + 2 cos2 v − 2 sin u cos2 v 2 sin u − 1 + sin v √ √ 2 2 2 − sin u 2 − sin u √ cos u 1 − 2 sin u + √ 2 sin v √ 2 2 − sin u 2 − sin u √ 2 2. T (v1 . 2 − sin u  √ cos u 2 − sin u 2 2 + √ 1 − 2 sin u √ 2 2 − sin u 2  2  du2 + √ dv 2 − sin u 2 The vector V a = (cos v. Therefore. sin v) is not a unit vector because cos2 v + sin2 v 1 g(V. v) ∂ ∂z(u. then A + B obviously has the same property. v) ∂ = + + . the vector V a has the following Cartesian components. v1 = The results: dx + 2ydy. v) ∂ ∂z(u. etc. nα V α (after simplification) = 0. since a tensor is defined as a multi-linear function. it is clear that tensors form a vector space. i. 2 2 − sin u ∂z √ ∂ sin v cos u ∂ 1 − 2 sin u ∂ + √ 2 ∂x + √ 2 2 2 − sin u ∂y 2 − sin u ∂z √ √ sin2 v cos u + 2 cos2 v − sin u cos2 v sin u + cos u − 2 ∂ + √ √ 2 2 ∂x 2 − sin u 2 − sin u ∂ ∂y This vector is within the tangent plane because nα V α = 0. ∂u ∂u ∂x ∂u ∂y ∂u ∂z ∂ ∂x(u.1 Tensors Definition of tensor product i a). c) d) First. v) ∂ ∂y(u. T (v1 . ∂ ∂ sin v cos v V = cos v − √ +√ 2 − sin u ∂x 2 − sin u ∂y + sin v cos v cos u √ 2 − sin u = sin v cos v √ 1 − 2 sin u ∂ + sin v √ . if A(x + λy) = A(x) + λA(y) and likewise for B. ∂/∂xk = δk gives: ω1 . v) ∂ ∂y(u. Now. show that the sum of two linear functions is again a linear function: If A(x) and B(x) are linear functions. v1 ) = 0.Therefore the induced metric is g = dx2 + dy 2 + dz 2 =  du2 + dv 2 = √ 2. v2 ) = 6x.b) A direct calculation using the property dxi . 3 ∂ ∂x = 3. 7 . V ) = √ 2 = √ 2 − sin u 2 − sin u The Cartesian components of the vectors ∂/∂u. v) ∂ = + + . ∂/∂v are found from ∂ ∂x(u.

v) = u× v.g.g. Therefore T is a tensor of rank 1+2. v) = uα v α − εαβγ εα λµ uβ nγ v λ nµ .3 Example of tensor a) An obvious example of such T is the vector product. like this: [T (u. α α Tαγ = 2εα αγ − nγ δα = −3nγ 2. e. α Therefore α α [T (u. e. fr . However. Using the known identity for the ε-symbol.. v) = uα v α − (δβλ δγµ − δβµ δγλ ) uβ nγ v λ nµ = uα v α − uλ v λ nµ nµ + uλ nλ vµ nµ . defined in three-dimensional space. gives components of a quantity which is not a tensor because these components do not transform correctly under changes of basis. v) in the index notation. b) Contracting two lower indices.. We can define A(f1 . If Tααβ were a tensor it would transform as ∂xλ ˜ Tααγ = Tµµλ . v1 . vs ). v) = gαβ uα v β .8 Examples of spaces with a metric g(u.2. a) We perform the calculation in components. ∂ xα ∂ xα ∂ xγ ˜ ˜ ˜ ∂ xγ ˜ c) Calculation gives α because εα αγ = 0 and δα = 3. The arguments are much simpler than the proof of tensor transformation law for components. v)] = Tβγ uβ v γ . . where gαβ are the components of the metric tensor. v1 . we need to represent T as a multilinear number-valued function of some number of 1-forms and vectors.g. Tααβ . α Now we would like to move uβ v γ out of the brackets and so determine Tβγ . T (v1 .. A(f1 . so as to rewrite identically α u α n β v β = u β v γ n γ δβ . Therefore we rename the index β to γ and also introduce a Kronecker symbol δβ .. the expression above contains α β β γ α u v instead of u v . T (u. it is easy to show that linear combinations of tensors are also multilinear functions. e. α 2. ∂ xγ ˜ However. v2 ) = f1 . v2 ) itself is not such a function because it has vector values instead of scalar (number) values. We would like to write g(u.. To determine the rank of T . Tensor products and contractions are also multilinear. v)] = 2εα βγ uβ v γ − uβ v γ nγ δβ = 2εα βγ − nγ δβ uβ v γ . 8 . α b) The calculation may go as follows. v2 ) and then it’s clear that A is multilinear.5 Contraction of tensor indices a) Using the definition of a tensor as a multilinear function.. We need to determine the components Tβγ such that α [T (u. εαβγ εα λµ = δβλ δγµ − δβµ δγλ . Tαβγ = ∂ xα ∂ xβ ∂ xγ ˜ ˜ ˜ The contraction over α = β yields ˜ Tααγ = α α ∂xλ ∂xλ ∂xµ ∂xν Tλµν = Tµµλ . . It is clear that T (v1 . So we need to add a 1-form to the list of arguments. v)] = 2εα βγ uβ v γ − uα nβ v β . this does not agree with the contraction of the tensor Tαβγ . α So we rewrite the given definition of T (u. which transforms as ∂xλ ∂xµ ∂xν ˜ Tλµν . we find g(u. α α Tβγ = 2εα βγ − nγ δβ .

0 gαβ =  0 1 − n2 2 0 0 1−n Then it is clear that det gαβ = 1 − n2 . Then the components of the vector n in this basis are (|n| . The result is g(u. 0 3 2 Therefore the determinant of gαβ is always nonzero. dy. Therefore. A·C α α c). n). c) Considerations are analogous to b). kA − B × C . 0) and the matrix gαβ has the following simple form:   1 0 0 . where Mβ is a matrix and Aβ is a known vector. 2. 0. (2) Solving this system of equations. The solution xα can be found as a linear combination of Aα . Substituting this expression into the given equation. where M is the inverse matrix (it exists in the generic case). Multiply ×C: Simplify using the identity (2). B. and then we need to relabel indices such that v α uβ can be moved outside the parentheses. k β= 1−A·A . except that the size of the matrix gαβ is larger. d) The metric is (dx)2 + (dy)2 + (dz)2 . d) The equations have the form xα Mβ = Aβ . kx + x × A = B. kβ + A · A − 1 = 0. these three vectors are independent and so each of the coefficients above must be zero: kα − A · B = 0. The solution is −1 −1 x = M A. dz through du and dv. k2 (x × A) × C = B × C. and we need to express dx. A·B . and the cross product εαβγ Aβ B γ with unknown coefficients. A calculation gives g = dx2 + dy 2 + dz 2 = R2 sinh2 u + cosh2 u du2 + R2 cosh2 u dv 2 . Using vector notation. and using the known identity (A × B) × C = B (A · C) − A (B · C) . Since we are considering the generic case. To analyze the conditions under which det gαβ = 0. 9 .Denote n2 ≡ nµ nµ ≡ g(n. we have x = αA + βB + γ (A × B) . b) Similar calculations give 3 + n2 0 gαβ = . we find (assuming k = 0 in the generic case) α= b) We have in vector notation x × A = B. On purpose.9 Supplementary problem sheet 2A Vector equations a) The equation contains two given vectors Aα and Bα . we find A (kα − A · B) + B (kβ + A · A − 1) + (A × B) (kγ − β) = 0. we write this equation in the form of linear combination of the three vectors A. k γ= 1−A·A . kγ − β = 0. v) = δαβ − n2 δαβ + nα nβ v α uβ . we can choose an orthonormal basis such that nα is parallel to the first basis vector. A (x · C) − x (A · C) = kA − x (A · C) = B × C. the matrix gαβ is nondegenerate if n2 = 1. x= Therefore x · C = k. Bα . Therefore gαβ = δαβ − n2 δαβ + nα nβ . and A × B.

where the coefficients f.  0 0 1   . The metric is degenerate if r = 0 (i..   0 1 0   . x ˜ we may express the partial derivative operators using the chain rule. at the origin). g. 3 3. ∂ξ ν ˜ The relationship between Γ and Γ can be found as follows. x while {ξ α } is the standard Euclidean coordinate system. The Christoffel symbols are defined as Γµ = αβ ˜ Γµ αβ ∂ 2ξν ∂xα ∂xβ ∂ 2ξν = ∂ xα ∂ xβ ˜ ˜ ∂xµ . The result is g = |A|−2 . b) In the basis where Aα is parallel to the first basis vector. X αβ Aα = B β X αβ Bα = 0 ⇒ u×A =B ⇒ u×B= 0 It follows that u is parallel to B and then the condition u × A = B leaves the only solution u = 0. 2C Degeneracy of the metric a) The metric can be written in the basis {dx. . Using Aα B α = 0 and denoting 2 Aα Aα ≡ |A| . . and therefore X αβ = 0 is the only admissible solution. . ∂ xα ∂ ˜ ∂ = . Suppose {xα } and {˜α } are two coordinate systems. µ ∂x ˜ ∂ xµ ∂xα ˜ 10 . where uγ is an unknown vector that we need to determine. . h are unknown.2B Tensor equations a) Since the vectors Aα and Bα are a basis in two-dimensional space (they are nonzero and orthogonal). f = h = 0. The determinant of this matrix is − 1 + x2 which is always nonzero. b) An antisymmetric tensor X αβ in three dimensions can be always expressed as X αβ = εαβγ uγ .1 The Christoffel symbol Transformations 1 We are considering a flat space where Euclidean coordinates exist. We can now rewrite the conditions on X αβ in a vector form. .. so X αβ = Aα B β + B α Aβ |A| 2 ⇒ f |A| Aβ + g |A| B β = B β . µ ∂x ∂xµ ∂ xα ˜ ∂ ∂xα ∂ = . 2 2 2 2 . ∂ξ ν ∂ xµ ˜ . 0. we get the system of equations X αβ Aα = B β X αβ δαβ = 0 ⇒ f |A| + h |B| = 0.. then the symmetric tensor X αβ can be written generally as X αβ = f Aα Aβ + g Aα B β + Aβ Aα + hB α B β .. 0. Assuming that the functions xα (˜) and also xα (x) are known. dy} as the matrix g= 2 y2 1 + x2 1 + x2 0 .. the vector Aα has components (A.) and therefore the metric gαβ has the form   −r2 0 0 . It remains to determine these coefficients.e. 0.

5 Locally inertial reference frame In this problem (unlike problem 3. Therefore we can write the answer as T αβ γδµ. we find x ∂ xα ˜ µ α α µ ν α = δβ + xµ − xµ δβ Γα (0)µν = δβ + x − x(0) Γ(0)µβ .1) tensors. Γλ = Γµ αβ ∂xα ∂xβ ∂ xµ ˜ ∂ xν ∂ xρ ∂xα ∂xβ ˜ ˜ 3.γ B β + Aα B β = B β ∂γ Aα − Γλ Aλ + Aα ∂γ B β + Γβ B λ . Assume that ∇ν uµ obeys the correct transformation law for rank (1.3 Covariant derivatives The rule is that every upper index gets a +Γ and every lower index gets a −Γ. and substitute ∂ µ u + Γµ u α .4 The Leibnitz rule Perform an explicit calculation.ν = ∂ν T αβ γδµ + Γα T λβ γδµ + Γβ T αλ γδµ λν λν − Γλ T αβ λδµ − Γλ T αβ γλµ − Γλ T αβ γδλ γν δν µν 3. Note that ∂ xα ∂xβ ˜ α = δγ ∂xβ ∂ xγ ˜ because the matrices are ∂ xα /∂xβ and ∂xβ /∂ xγ are inverse to each other. αγ λγ 3. αν ∂xν ∂ µ ˜µ α ˜ ˜ u + Γαν u . Denoting {˜α } ≡ {ξ α }. Each term with Γ replaces one index in the original tensor by one of the indices in Γ. but it is just a coordinate system which is like Euclidean at one point.γ = ∂γ Aα B β − Γλ Aα B β + Γβ Aα B λ . To use that formula. Aα. .1) the coordinate system {ξ α } is not a flat Euclidean coordinate system. ∂xλ ∂ 2 xδ ∂ xµ ∂xγ ∂xδ λ ˜ Γγδ + β γ ∂xλ ∂ xα ∂ xβ ˜ ˜ ∂x ∂x ˜ ˜ γ δ µ ∂ 2 xλ ∂ x ∂x ∂x λ ˜ Γγδ + β γ = ∂xλ ∂ xα ∂ xβ ˜ ˜ ∂x ∂x ˜ ˜ (3) Note that the Euclidean coordinate system ξ ν is not needed to determine the transformation of Γ. (0) ∂xβ 11 . The result is ˜ ˜ 2 λ ˜ν ˜ρ ˜ν ˜ρ λ ˜ νρ ∂ x ∂ x ∂x + ∂ x ∂ x ∂ x . ∂ξ ν ∂xλ ∂ξ ν ˜ Therefore we can calculate Γµ (when Γµ is known) as follows. given the Christoffel symbol Γ in the original coordinate system {xα }.Also we can express ∂ xµ ˜ ∂ xµ ∂xλ ˜ = . we need to compute some derivatives. ∇ν uµ = ∇α uβ β ∂xν ∂x ˜ Consider a vector field uµ .2 Transformations 2 µ ˜α ˜ ˜ ∂x ∂ x .γ λγ Aα B β This proves the required property. αβ αβ ∂ xµ ∂xλ ˜ ˜ Γµ = αβ ∂xλ ∂ξ ν = = ∂xγ ∂ ∂ xα ∂xγ ˜ ∂xδ ∂ ∂ xβ ∂xδ ˜ ξν ∂ xµ ∂xλ ∂xγ ∂xδ ∂ 2 ξ ν ˜ ∂ 2 xδ ∂ξ ν + β γ δ ∂xλ ∂ξ ν ∂ xα ∂ xβ ∂xγ ∂xδ ˜ ˜ ∂ x ∂ x ∂x ˜ ˜ ∂ xµ ∂xλ ∂ξ ν ˜ ∂xλ ∂ξ ν ∂xδ ∂ xµ ˜ . which will enable us to ˜ compute the Christoffel symbol Γ in the coordinate system ξ. Now we want to use the formula (3). 3. αγ . ˜ ˜ ∇ν uµ ≡ ∂ xν ˜ ∇ν uµ ≡ ˜ We can now express Γ through Γ.

4 4. Therefore these terms will cancel after a contraction with uβ uγ .λ ) uγ uµ .γ + gλγ. ∂ xβ ∂ xγ ˜ ˜ where Aα µβ At x = x0 we have Γµ = Γµ αβ (0)αβ . one denotes {xα } ≡ {ξ α }. ∂ 2 xλ = −Γλ (0)βγ + O x − x(0) . By assumption.µ − gµγ. Γ = Γ. the result can be found simply by assuming that ˜ ∂xα α = δβ + xµ − xµ Aα + O (x − x(0) )2 .γ + gλγ. = δβ − xµ − xµ Γα (0)µβ + O (x − x(0) ) (0) ∂ xβ ˜ Therefore Finally. ∂xα /∂ xβ . The geodesic equation can be rewritten for the 1-form uα as duα uγ uα.β ) uβ uγ . i. It follows that the new Christoffel symbol is equal to zero at x = x0 .e. the first-order derivatives α α ∂x/∂ξ at x = x0 can be found as the inverse matrix to ∂ξ α /∂xβ = δβ . Γβ uβ uγ = Γβ αγ uβ uγ = αγ 4. ∂xα /∂ξ β = δβ .γ βγ ds 1 = uγ uα + g αλ (gλµ.γ ds where we must use an ordinary derivative instead of the covariant derivative (according to the definition of d/ds). i. 12 .e. µβ (0) ∂ xβ ˜ is an unknown matrix.γ 2 Therefore 1 uα uγ uα = − uα g αλ (gλµ.1 Geodesics and curvature Geodesics (a) Note that d/ds is the ordinary (not “covariant”) derivative in the direction of uα .γ = gαβ. This has the advantage that only derivatives ∂ 2 ξ/∂x∂x need to be computed.. ˜ Alternatively. .γ + gλγ.γ = gαβ.λ ) uγ uµ . αγ ds An explicit formula for Γβ yields αγ 1 (gβα. So we find uγ gαβ uα uβ . one can use the transformation law for the Christoffel symbol in the inverse direction. we find ˜ ∂ 2 xλ ∂ xµ ∂ xµ ∂xγ ∂xδ λ ˜ ˜ Γγδ + β γ λ Γµ = αβ λ ∂ xα ∂ xβ ∂x ˜ ˜ ∂ x ∂ x ∂x ˜ ˜ = Γµ − Γµ αβ (0)αβ + O x − x(0) . the derivative of the vector field uα satisfies . αγ 2 b) We give two derivations.µ − gµγ. . the first one is direct and the second one uses the property (a).γ uα uβ uγ + 2gαβ uα uγ uβ .γ uλ uµ uγ .1 First derivation d gαβ uα uβ = uγ gαβ uα uβ . 2 2 What remains is a straightforward computation: uγ gαβ uα uβ . + .λ ) uγ uµ = − gλµ.γ uα uβ uγ + 2gαβ uα uγ uβ .γ − gαγ. can be found by inverting this matrix.. x and not the derivatives ∂ 2 x/∂ξ∂ξ.α − gαγ.γ d α u + Γα u β u γ = u γ u α + Γα u β u γ = 0 βγ .γ uβ uγ .γ Note that = gαβ.γ uλ uµ uγ = 0.γ + gβγ.µ − gµγ.β is antisymmetric in (β ↔ γ).γ Now we need to simplify an expression containing uβ .γ uα uβ uγ − gλµ.The inverse derivative. Since all derivatives only need to be evaluated at x = x0 . {˜α } ≡ {xα }. This considerably simplifies the calculations.γ 2 1 1 = − uλ (gλµ.1.. So up to quadratic terms we find ∂xα 2 α . The remaining term yields 1 Γβ uβ uγ = gαβ. 2 Note that gβα...γ = 0 = − Γβ u β u γ .

2 Commutator of covariant derivatives u α = u α + Γα u λ .α uλ uβ uγ = uα uβ + u u . ds ds γ thus dgβγ αβ dg αβ dgλµ αλ βµ dg αβ gβγ = − g ⇒ =− g g . duα 1 = gβγ. ds ds Now we need to express uα.λ (x0 ) xλ dxγ .λ d xγ xλ = 0 = 13 xγ dxλ + xλ dxγ .4. Therefore δAα = dAα ds = ds Γβ (x)Aβ (x)dxγ ≈ Aβ (x0 ) αγ Γβ (x)dxγ . .β µγ βλ βγ . We use the identity d d α g αβ gβγ = δ = 0. These .γ uγ ≡ duα /ds.γ µγ βλ α uα − uα = uλ Γα − Γα + Γµ Γα − Γµ Γα = uλ Rλγβ .β βλ = u α + Γα u λ + Γα u λ + Γα u µ + Γα Γµ u λ − Γµ u α . ds 2 and find dg αβ dg αβ dgλγ λ γ d g αβ uα uβ = uα uβ + g αβ gλγ. .βγ βλ. so we can approximate Aα by a constant. ds The parallel-transported vector can be represented by a 1-form Aα (s) such that uγ ≡ However.βγ . and also used the identity αγ. in a locally inertial system at x0 we have Γβ (x0 ) = 0.γβ terms are the following: uα .βγ βλ.2 Second derivation d d gαβ uα uβ = g αβ uα uβ = uγ g αβ . Γα u λ + Γα u µ . we can omit the terms that are symmetric in (β ↔ γ). dxγ = 0.αuβ uγ .µ The remaining terms are which yields uα = Γα uλ + Γα Γµ uλ + (symmetric in β ↔ γ) .γ λγ. Γβ (x) = xλ − xλ Γβ αγ 0 αγ. ds ds ds d g αβ uα uβ = 0.βγ . ds We write Therefore and thus 4. .γ µγ . Further. Therefore we can Taylor expand Γβ (x) near x0 as αγ αγ 2 .λ + O (x − x0 ) Therefore δAα ≈ Aβ (x0 ) β γ λ λ Γβ αγ. To do that.γ First compute + Γα u µ + Γµ u λ − Γµ u α + Γα u λ . ds ds ds ds It remains to express the derivative of g αβ through the derivative of gαβ .β . αγ Now.γ βλ .γ µγ .γ uβ uγ .γβ λβ γµ λγ βµ 4.λ x − x0 dx ≈ Aβ (x0 )Γαγ. λβ. . . ds ds ds ds dg αβ dgλµ αλ βµ dgλµ λ µ uα uβ = − g g uα uβ = − u u .β βλ u α = u α + Γα u λ . αg ds dxγ .β .µ Since we want to compute the commutator uα − uα .3 Parallel transport dAα − Γβ Aβ uγ = 0.βγ .βγ βλ . the closed curve is assumed to cover only a very small neighborhood of one point x0 . along the curve. Γµ u α .β βλ . we use the property derived in (a).1.γ uα uβ + 2g αβ uα.µ µλ µγ βγ .β βγ . Aα (x0 ). where we have again approximated Γβ (x) by its value at x = x0 .

which is already a consequence of (4). (4) (5) (6) However.. Rαβγδ = −Rβαγδ = −Rαβδγ . (6)]. For fixed γ. therefore it is sufficient to consider these two properties [note that (5) does not follow from (4). Finally.λ − Γαλ. R12 = −g 12 R1212 . Therefore. let us check that every such constraint is nontrivial for every δ (even if δ is equal to one of α. n we obtain N2 additional constraints. the number of independent components of Rαβγδ is reduced to N1 = 1 n(n − 1) 2 2 . γ. Therefore.γ 2 1 β = Aβ Rαλγ xλ dxγ . δ. This is a nontrivial constraint (equivalent to Rαβαγ = Rαγαβ ). β = γ). the number of constraints is nN2 . . β). β. we have −g 12 g 11 = ggαβ . R = g αβ Rαβ = 2g 11 g 22 − 2g 12 g 21 R1212 = 2gR1212 . For instance. Therefore. γ are different. Therefore. This tensor has the following symmetries. α = γ. since det g αβ = 1/g. Likewise for β = γ or for α = β. for each δ = 1. β. Then we can express the Ricci tensor as Rαβ = g λµ Rλαµβ .Therefore we may rewrite δAα ≈ Aβ Γβ αγ. Rαβγδ + Rβγαδ + Rγαβδ = 0. we find R11 = g 22 R1212 . Now we use the property (5). the property (5) is a new constraint only if all three indices α. §8 c) There is only one independent component of Rαβγδ in two dimensions. 12 b) Weinberg. and thus the total number of independent components of Rαβγδ is N = N1 − nN2 = n2 n2 − 1 . γ are different (i. α = β.λ xλ dxγ = 1 β Aβ Γβ αγ. There are N2 = 1 n(n − 1)(n − 2) 6 choices of such α. This matrix has 1 n(n − 1) independent components. If α = γ. then the property (5) becomes Rαβαδ + Rβααδ + Rααβδ = 0 (no summation). Suppose δ = α. Note that the matrix g 22 −g 12 g 22 −g 12 Therefore −g 12 g 11 is equal to the inverse matrix to g αβ (which is gαβ ).4 Riemann tensor a) It is more convenient to consider the fully covariant tensor Rαβγδ . Rαβγδ = Rγδαβ . we can choose R1212 as the independent parameter. we have that Rαβγδ is an antisymmetric n × n matrix (indices α. β. Calculating component by component. β. R22 = g 11 R1212 . it is known that the property (6) follows from (4)-(5). then (5) becomes Rαβγα + Rβγαα + Rγαβα = 0 (no summation). Let us first consider the property (4). multiplied by the determinant det g αβ . 2 xλ dxγ 4. Suppose that α. Likewise for fixed 2 α.. β. Let us see whether the property (5) is nontrivial at fixed δ.. γ).e. 2 2 14 . 1 1 Gαβ = Rαβ − gαβ R = ggαβ R1212 − gαβ 2gR1212 = 0. Chapter 6.

which has 1 n(n − 1) independent components.2 Energy-momentum tensor 1 T αβ = −pη αβ + (p + ρ) uα uβ .1 Gravitation theory applied Redshift GM .) Therefore the redshift factor z(r) at distance r from the center of the Earth is g00 = 1 + z(r) = 1+ 2GM GM ≈1+ 2 . denote by RE the radius of the Earth. We know that the gravitational acceleration at the surface is m GM gE = 2 ≈ 9. where v is the 3-vector of velocity and |v| ≪ 1 in the units where c = 1. The condition β for H α is β (δ α + εH α ) δ β + εH β gαβ = gλµ . RE s 2 Therefore. Let us simplify this expression by introducing the time derivative along the fluid flow. µ λ µ Therefore. it is convenient to express GM = gE RE .β = −p. v). For a vertical distance L between sender and receiver. r In the weak field limit. The EMT of an ideal fluid is where uα is the 4-velocity vector of the fluid motion. d ≡ uα ∂α . For n = 4 we get 6 2 components. These can be interpreted as three spatial rotations and three Lorentz rotations (boosts). In the nonrelativistic limit.β uα uβ + (p + ρ) uα uβ .81 2 . Hλµ is an antisymmetric n × n matrix. the Newtonian gravitational potential near a mass M is Φ= while the component g00 of the metric is 2Φ . we may approximate 1− RE RE + L = 1+ gE RE L . λ µ Consider an infinitesimal Lorentz transformation. gαβ = ηαβ . we find z(RE ) 1 + gE RE c−2 gE RE = ≈1+ −2 RE z(RE + L) c2 1 + gE RE c RE +L Since in our problem L ≪ RE . z(RE + L) c 10 m 5. uα ≈ (1. 5 5. λ λ µ µ Disregarding terms of order ε2 . c2 (RE + L) z(RE ) L gE L ≈ 1 + 2 ≈ 1 + 1.β + (p + ρ) uα . c2 (We write the units explicitly. Λα = δ α + εH α . λ λ λ The number of parameters in Lorentz transformations is the same as the number of parameters in H α .1 16 .4.5 Lorentz transformations A Lorentz transformation is represented by a matrix Λα such that β Λα Λβ gαβ = gλµ . In the nonrelativistic limit. The conservation law is 0 = T αβ . we find α 0 = δ β H α gαβ + δλ H β gαβ = Hµλ + Hλµ .β uβ . 2r c c r To compare the redshift factors at the surface of the Earth.α + (p + ρ). we may disregard gravitation. dt 15 .

In our notation. we get Φ.) The calculation goes like this: R00 = Γα − Γα + O( Γ Γ ). ˙ This is the relativistic continuity equation.β ) = −η αβ Φ. somewhat more comprehensive solution.0 − g00. dv 1 = dt p+ρ −∇p − v dp dt . not true for relativistic matter.) This is the Euler equation. covariant derivatives do not commute αβ βα when applied to vectors!) and also the property Φ.λα Here we used Φ.α .α Φ. see below. Eq. v).α + uα p + (p + ρ) uα . for any quantity X we have ∂ d ˙ X ≡ X ≡ X + (v · ∇) X.α Φ . to Eqs.β .β .αβ X α .αβ = − Φ = −Φ. we find from Eq.just ρ.β Compute the covariant derivative. dt ∂t The continuity equation (8) gives dρ + (p + ρ) divv = 0.j = −∇j p. such as photons (electromagnetic radiation) for which p = 1 ρ.00 + Φ.α . (8) and (9). so it is . which is due to gαβ.α = 0. we find ˙ ρ + (p + ρ) uβ .α 0α.α . we write gµν = ηµν + hµν .βα which follows from Γλ = Γλ (note that Φ is a scalar.0 1 αβ α Γ00 = η (g0β.0 = 0) R00 = −η αβ Φ.β = 0 then also Φ. Using this equation. 3 16 . 2 therefore (using Φ. dt This is the ordinary.β = 0. (7) that 0 = −p.1 Finally. and also we may raise and lower indices using ηµν instead of gµν . ˙ (Note that p. 5.α Φ.β + (p + ρ) uα .33 = ∆Φ.) We may disregard terms of order ΓΓ 00 because Γ is of order Φ. (This is somewhat heuristic.4 Weak gravity A very short solution is to write R00 directly through Γ and note that only Γα 00.λ Φ.α αβ Φ.β − g αβ Φ.α + uα p + ρ + (p + ρ) uβ . there is no p + ρ . 1 Note that in the usual. −g . ˙ ˙ ˙ Contracting with uα and using uα u = 0.λ 2 =Φ . if Φ.α .11 + Φ. so p ≪ ρ and p + ρ ≈ ρ.0 + gβ0.β = 0 is a weaker condition than Φ. however.β X α = Φ.αβ = Φ. this entails Φ.0 = 0. i. uα ≈ (1.β = 0. This is so because in most cases the matter is nonrelativistic.α sufficient to write the latter). In the weak field limit. Therefore. Here is another.e.α Φ. nonrelativistic continuity equation.α .α comes in (if we disregard terms of second order). ˙ ˙ (9) (8) α (7) Now let us apply the nonrelativistic limit.α = 0 (since Φ. nonrelativistic continuity equations as they are written in most books.α = 0.α = Φ. 00.α . (9) gives ˙ ∇p + v p + (p + ρ) v = 0.λ Φ. +Φ Φ .α .22 + Φ.β .µ = 0.3 Energy-momentum tensor 2 1 T αβ .α = Φ. (Assume that gαβ. 5.Then we find 0 = −p. This is. Then compute Γα explicitly through Φ.

ds Finally.αβ .βλ + hβγ. So it is sufficient to compute just the Newtonian potential Φ. not just h00 . ⇒ Γ ∼ O(Φ).α − hαν. The actual metric gµν is equal to ηµν plus a small first-order deviation.ν + hνγ.g.γ ) .0λ + h0γ.λβ + hλγ. one needs all components of hµν .ν = 0 through ordinary derivatives. Let us do the computation through hµν in a more general way.γλ − hλγ.Then we only compute everything up to first order in hµν . we can rewrite dxµ ∂ d (4) α δ (x − xα (s)) δ (4) (xα − xα (s)) = − ds ∂xµ ds (this is easily understood if read from right to left) and then integrate by parts. In principle. when one only wants to compute effects of gravitation on motion of slow bodies.ν . Therefore.λ − Γαλ.0 = 0) and h00 = 2Φ: R00 = 1 λγ 1 η (h0γ. αβ ds2 ds ds 17 . but this deviation cannot be expressed just through the Newtonian 1 potential Φ ≡ 2 h00 . 0= √ dxν d (4) σ δ (x − xσ (s)) = − ds ds ds d2 xν (4) σ δ (x − xσ (s)). 2 Let us now compute just the component R00 . ∂ 1 ∂ √ 1 −g = − √ gg αβ gαβ.γλ 2 2 = −η λγ Φ. 2 λ λ λ = Γλ αβ. The given relations are useful.5 Equations of motion from conservation law We would like to rewrite the covariant conservation law T µν . ds ds αβ Since in the first term the dependence on xσ is only through δ (4) .µ + Γµ T αν + Γν T µα = √ −gT µν αµ αµ −g Apply this to the given T µν : m0 0 = T µν .) Now we compute (again up to first order in Φ) Γλ αλ. 2 2 1 = η λγ (hαγ. To compute trajectories of light rays.ν + gνα.αλ − hαβ. dxα dxβ d2 xν + Γν = 0.γβ ) = η λγ hλγ. (We may disregard the ΓΓ terms since they are second order in Φ. But e.µ = m0 ds dxα dxβ (4) σ d2 xν δ (x − xσ (s)). only g00 is necessary. one needs to solve the full Einstein equations to find hµν . ds2 −g µν T .0λ − h00. we may raise and lower indices using the Minkowski metric ηαβ instead of gαβ . trajectories of light rays cannot be computed accurately in the Newtonian limit (because light does not move slowly). Therefore.λγ = ∆Φ.α ) = g µα gµα. + Γν αβ 2 ds ds ds This is a function of xσ which should equal zero everywhere. However.αβ ) . αβ .µ = √ −g m0 +√ −g ds dxµ dxν (4) σ δ (x − xσ (s)) ds ds α β .µ − gνµ. in other words.00 ) = − η λγ h00.µ ds dx dx ν Γ (x)δ (4) (xσ − xσ (s)). 5.αβ − hαλ. First we compute the Christoffel symbol and the Ricci tensor: Γλ = αν λ Rαβ = Rαλβ 1 λγ η (hαγ.β + ΓΓ − ΓΓ ≈ Γαβ. one needs to determine other. “post-Newtonian potentials” and not just the Newtonian potential Φ. the integrand should vanish for every value of s. recalling that hµν is time-independent (so hµν.ν = √ −g. let’s derive them first.β = Rαβ 1 1 λγ η (hαγ.γλ − hλγ.ν + Γν T αβ . µν 2 2 −g ∂xν Now rewrite the covariant derivative of T µν explicitly: √ 1 µν T µν . 2 1 ∂ √ 1 1 Γµ = g µα (gµα.β .µ = T. hµν .λ − Γαλ.ν g =− √ ∂xν 2 −g ∂xν 2 −g 1√ = −gg αβ gαβ. Note: the Newtonian limit does not determine the components of gµν except for g00 = 1 + 2Φ.

where φ(x) is an arbitrary function of spacetime. and that we have relations such as ηλλ ηαλ = δα and β δα ηαα = ηαβ . Using this gauge symmetry. (12) as Γλ = αβ where we defined the auxiliary function Aα 1 λ λ δ Bλ. there is also a gauge symmetry. Γλ = αβ µ g αβ = η αβ 1 Aα (no summation!). 2 α Aλ Bα ≡ ln Aα . or one scalar field. ˙ dE = (mv + V ′ (x)) v ˙ dt However.Remark: in general. Then only three functions of spacetime (A1 . Since the metric has a diagonal form.α = Aα 2 λ √ Bλ.β + ηβµ Aµ. or some number of point particles). (10) 1 λµ 1 [ηαµ Aµ. Since these functions φ(x) are functions only of three arguments. A2 .α . and one cannot hope to derive them from a single conservation law. Let us also denote for brevity C ≡ ln √ −g = Bλ . the equation of motion for a particle follow from the conservation of energy only if the motion is in one dimension: E= 0= mv 2 + V (x) = const. The situation in ordinary mechanics is similar: e. let us denote gαβ = ηαβ Aα .α − ηαβ Aα. Γλ = αλ λ 1 Bα. Ψ do not follow from the conservation of their combined Tµν . This would give 4 degrees of freedom.β + δβ − ηλλ ηαβ . = δα 2 Aλ 2 Aλ 2 Aλ λ Note that the summation over µ results in setting λ = µ due to η λµ .1 Spherically symmetric spacetime Straightforward solution A direct computation listing all the possible Christoffel symbols and components of the Ricci tensor is certainly straightforward but very long. where Aα ≡ eN . However.2.α . we shall substitute the known functions Aα and use the resulting simplifications. Aµ → Aµ + φ. αλ 18 . we rewrite Eq. We begin with the calculation of the Christoffel symbols. These fields have two different equations of motion. they do not change the number of degrees of freedom.α − 1 2 1 Aα Bα.α + 2 λ Bλ. Hence the electromagnetic field has 3 degrees of freedom.λ . For convenience.α = ln −g . 2 ⇒ mv = −V ′ (x). r2 . At the end of the calculation of the Ricci tensor Rαβ . For this calculation.g.µ . equations of motion do not follow from conservation of energy if there is more than one degree of freedom. Similarly. equations of motion for say two scalar fields Φ. However. (e. which hold without summation. There are additional gauge symmetries involving functions φ(x) that do not depend on time. we compute the “trace” of the Christoffel symbols and compare with the known formula. A3 ) are left. one fluid.g. 6 6. we do not use the Einstein summation convention any more. r2 sin2 θ (11) is a fixed array of four functions. and that ηλλ = η λλ . equations of motion do not follow from conservation law. set the component A0 (x) = 0. we may e. eL . but they do follow if there is only one field.λ 1 λ Aλ.2 6. we make heavy use of the fact that ηαβ = 0 only if α = β.β + δβ Bλ. Here is a way to compute the curvature tensor without writing individual components. every summation will be written explicitly. λ λ Γλ = C. 6.g.α 1 Aα.1 The gravitational field Degrees of freedom The electromagnetic field is described by a 4-vector potential Aµ (x). As a check.µ ] η 2 Aλ (12) 1 λ Aλ.α − ηλλ ηαβ Bα.

β + δβ Bλ.α Bµ.β Bβ.β µ Aα Bα.β Bβ.λ .α Bα.λ Bα.β Bβ.λ − Γαλ.β − Bβ.µ Aµ 1 + 4 1 1 + Bβ.β Bβ. we put all the terms together: Rαβ = 1 Aα Bα.µ Bα.α Bµ.β + C. We use the formula (with Landau-Lifshitz sign conventions) Rαβ = λ λ Γλ αβ.µ Aµ ηλλ ηαα λ Aλ Aα α α Bα.β Bβ.α − 2 2 2 4 Bµ. αµ βλ λµ αβ We now compute the necessary terms: Γλ αβ.α − ηλλ ηαµ µ µ δλ Bµ.α − ηµµ ηαβ µ Bµ.µ − ηµµ ηµβ Aα Bα.λ − ηµµ ηλβ 1 = 4 + − = 1 4 1 4 µ µ µ Bα.α Aλ 4 1 Aα Bα. .α − ηαβ 2 2 Finally.λ Bα.α ) 2 1 − 4 (only for α = β) 19 .β .λ + Bα.λ Aλ 1 1 + C.α − ηαβ 2 Aα Bα.µ 1 4 λ.αα − ηαα Aα 2 ηλλ λ Bα.µ Aµ (here set λ = α) (here set λ = µ) (here set µ = α) µ Aµ Bµ.β + δβ Bµ.αβ 1 − ηαβ Aα 2 ηλλ λ Bα.λ .β 4 4 ηλλ λ 1 1 − Bα.λ 1 λ Aα 1 1 λ δα Bλ.β + C.α Bα.β .β = C.µ + Aµ 4 Bµ.λ Aλ + C.β + δβ Bα.λ δλ Bα. λ Γµ Γλ = λµ αβ λ.λ Aλ 1 1 1 1 + C.α Aλ Aα ηµµ µ 1 1 1 Bα.β + Bα.α Bβ. µ Bλ λ .β .µ Aµ Γλ Γµ = αµ λβ λ.α Bβ.α Bα.α − ηαβ 4 4 4 Bµ.αβ + Bβ.β + λ.α − Bα.We proceed to the computation of the Ricci tensor.λ .λ Aλ Aα Bα.µ λ λ δα Bλ.µ δα Bµ. µ ηµµ µ 1 1 Bα + Bβ − C 2 2 . µ We can simplify this expression by considering separately diagonal and off-diagonal components: 1 Rαα = (Bα − C).β − ηαβ 4 4 1 1 = Bα.λ Aλ .λ − ηαα ηλβ Bλ.λ 2 2 2 Aλ C.β Bβ.α Bµ.µ Γµ Γλ − Γλ Γµ .α Bα.α Bµ.β + δβ Bµ. µ Bµ.β Bβ.λ Γλ αλ.λ = λ 1 1 1 Bα.µ + δµ Bλ.α Bµ.µ Bα.β Bβ.α Bα.α Bµ.αβ + 1 (C.α − Bα.λ ηλλ λ = 1 C.λ Aλ + C.α − ηλλ ηαβ Bα.α − 2 4 Rαβ = 1 2 Bα + Bβ − Bµ.α − Bα.α Bα.µ λ C.α .αβ − ηαβ 2 2 2 ηλλ λ Aα Bα.β + δβ Bµ.β + C.λ Bα.αβ . Aλ Aλ Bλ.

2r r = Rrφ = Rθφ = Rtθ = 0. √ N +L + 2 ln r + ln sin θ.θ = cot θ − 2 4 2 r 4r ˙ ˙ ˙ 1 N +L ′ 1 ˙ 1˙ L 2 ˙ N ′ + L′ ˙ = L − N L′ − N N ′ − LL′ = . 2r r r N ′ − L′ 1 = 1 − r2 e−L + 2 . the nonzero components of the Einstein tensor are Gtt = eN −L Gtr = Grr Gθθ L′ eN + 2. C. We have At = eN .φ = 0. 1 Gφφ = − r2 sin2 θ e−N 2 N ′ − L′ L′ − N ′ 2 + 2 − N ′′ + N ′ r r 2 20 . N + + 2 2 2 r 4 4 r = −r2 sin2 θ L′ − N ′ 2 1 −N ′′ + 2 + N′ + eL−N 2 r 4 2 ¨ ˙ ˙ −L ˙ N − L + 1 eN −L N ′′ + +L 2 4 2 ˙ ˙ ¨ L−NL ˙ L+ 2 N ′ − L′ 2 + 2 r 1 + L′ . Br = L. 1 1 2 12 1 2 cot θ = 0.θ = cot θ. Bφ = 2 ln r + 2 ln sin θ.λ 1 2 2  Rφφ = r sin θ ηλλ 2 Aλ λ + C. r2 Hence. r r r N ′′ 1 L′ − N ′ N ′ − L′ + 2 +− + N′ = −r2 e−L − 2r r 2 4 ˙ ˙ ¨ ˙N −L −L + L 2 − e−L − + r2 e−N 2 ˙ ˙ ¨ ˙N −L −L + L 2 .λ . we compute the Ricci scalar. Finally.r = + . C = ln −g = 2 ˙ ˙ N +L N ′ + L′ 2 C.λ ln r2 sin2 θ Aλ Rθθ Rtφ Rrθ Rtr 1 1 N ′ − L′ + 2 e−L − 2 .Now we need to simplify this expression further by using the specific form of the metric (10)-(11). C. Bθ = 2 ln r. We find (after some omitted algebra):  ln r2 sin2 θ . .r Bφ. Aθ = r2 . Ar = eL . r r ˙ L . r . Aφ = r2 sin2 θ.t = .λ   Rrr = Rtt = N′ .θ Bθ. = C. R= λ ηλλ 1 Rλλ Aλ ˙ ˙ ¨ L ˙ N −L +L 2 4 1 + e−L N ′′ + 2 N ′ − L′ 2 + 2 r N′ = e−N − e−L − − −N ′′ 1 L′ − N ′ 2 1 + 2 + N′ + L′ − e−N 2 r 4 r 2 ˙ ˙ ¨ L−NL ˙ L+ 2 1 + e−L r2 = e−N 1 N ′ − L′ N ′ − L′ 1 1 + 2 + + 2 e−L − 2 2r r 2r r r ′ ′ ′ ′ ˙ ˙ ¨ ˙ N − L − e−L −2 N − L − N ′′ + N ′ L − N −L + L 2 r 2 − 2 . r 2 N′ eL = 2 + − 2. Bt = N. 2 2 r Note that the term 1 2 Bα + Bβ − Bλ λ .r − Bφ. C.αβ (13) always vanishes when α = β.

22 − A.1 − δb eA1 −Aa A1.1 A.2 eA−B + B. The relationship between the curvature tensors under a conformal transformation is found as follows. γ11 1 γab R. It is clear that Rαβ will also be a direct sum of the corresponding two-dimensional Ricci tensors.c ) = 2 2 Γ2 = 12 1 B. Γ2 = − eA−B A.1 . set A ≡ A1 .b . 11.2 .b + γbc. we know that the Ricci tensor is proportional to γab .2 . and indices a. λ λ ν λ ν λ Rαβ = Rλ αλβ = Γαβ.2 . For a two-dimensional metric γab .2 eA−B A.b ) = (A + B). γ11 2 2 Well. 2 Γ1 12 1 = A. 2 a 1 1 1 a Γa = A1.. λ ˜ αβ Γλ = Γλ + Bαβ . range from 1 to 2.  −2 L  r e 0   −r−2 eN .1 . 2 2 2 Then the component R11 of the Ricci tensor is R11 = Γa − Γa + Γa Γb − Γb Γa 11.1 . γ11 R = γ ab Rab = 2 R11 . namely (see Problem 4..2 − B. gαβ = e2Ω gαβ . Let us at least derive the useful formula below.1 ba 11 1a 1b 1 1 1 = Γ1 + Γ2 − (A + B). this shows that at least this is not obviously wrong) R=2 R11 1 1 = −e−A B.λ − Γλα.22 − A. A more clever way to reduce the amount of computation is to notice that the metric gµν is simplified after a conformal transformation.1 ba 11 1a 1b and afterwards we will have Rab = γab R11 . 2 R ≡ γ ab Rab = 2 (det γab ) R1212 . The calculation of Rαβ for the metric hµν is much simpler because hµν is a direct (“block”) sum of two metrics defined on 2-dimensional spaces.1 A.2 Γ2 − Γ1 Γ1 − 2Γ1 Γ2 − Γ2 Γ2 11 11 11 11 12 11 12 12 11.β + δβ Ω.b δ1 + δb Ab.4c) Rab = So it is sufficient to compute say R11 . We shall first compute the Ricci tensor for the metric hµν and then determine how Rαβ changes under a conformal transformation. First we define the conformally transformed metric for convenience as follows.β + Γµν Γαβ − Γβν Γαµ . c. 2 2 4 4 We also find (note the symmetry apparent in this formula.1 11.6.11 + (A + B). γ (γ1c.2 (A − B).11 − 2 2 2 2 2 2 2 4 2 4 1 1 1 1 = − B.1 (B − A).2 (A − B). R11 = Γa − Γa + Γa Γb − Γb Γa .1 − γ1b. gµν = r2 hµν = r2    1 0 sin2 θ The necessary Christoffel symbols are found as (no implicit summation from now on!) Γa = 1b c 1 1 ac a a 1 A1. The Riemann and the Ricci tensors are defined (in Landau-Lifshitz sign convention) by ν λ ν λ λ λ Rλ αµβ = Γαβ.11 − eA−B A.1 = A.2 − A.a .a 1a. 11 11 11 2 2 2 1 1 1 a Γba = ∂b ln (det γcd ) = (A1. 21 .1 (A − B).a 1a.2 eA−B A.λ .2 Solution using conformal transformation The metric hµν separates into the r − t components and the θ − φ components.µ − Γαµ.2 e−B − e−A B. B ≡ A2 . ˜ λ Then the Christoffel symbols receive a correction which is a tensor Bαβ . b.11 − e−B A.2 − (A + B).. eB in two dimensions.1 Γ1 + (A + B).2. Γ1 = A. Let us first compute the Ricci tensor for a diagonal metric γab = diag eA . But this is roughly how it goes. αβ λ λ λ Bαβ ≡ δα Ω.1 − (A + B).1 B.2 2 2 2 1 1 1 A−B 1 1 1 1 1 1 1 e A. we are not going to finish this calculation here.β + Γλν Γαβ − Γβν Γαλ .α − gαβ Ω. anyway.1 .11 + (A + B).a .b + A2.1 δ1 − eA1 −Aa A1.1 + A.

λ .λ (δα Ω.ν + 2gαβ Ω.λ + δλ Ω.˜ ˜λ The same relation holds for Rλ αµβ and Rαβ through Γαβ (note that these relations do not involve the metric gαβ explicitly).λ − (2 + N ) Ω.λ = 0.β + 2gαβ Ω.α .ν Γν + (N − 2) Ω.α .α .λ + Ω.α Ω.α − gαλ Ω.λ Ω. where f ≡ 1 − rg /r.β + δβ Ω.α Ω.µ − Bαµ.λ + Γλ Ω.α − 2 (N − 1) Ω.λ Ω.λ .α − Ω.α − gαβ Ω.λ Ω. we compute λ λ Bλα = δλ Ω.α Ω.α Ω.β − gβν Ω. αβ λν αλ βν We shall only compute the expression for the Ricci tensor Rαβ .β + Γλ Ω.λ .λ Ω.β + Bλν Γν + Γλ Bαβ + Bλν Bαβ − Bβν Γν − Γλ Bαλ − Bβν Bαλ . αλ βλ so the resulting formula can be written more concisely as ˜ Rαβ − Rαβ = (N − 2) [Ω.λ 6.λ Γν αλ = 2Ω.α + gαλ Γλ Ω. The modified Ricci scalar is ˜ ˜ ˜ R = g αβ Rαβ = e−2Ω g αβ Rαβ + e−2Ω g αβ (N − 2) [Ω.β − Ω.λ = − (N − 2) Ω.α + (N − 2) Ω.λ + gαν Γν Ω. As a preparation.β − Ω.α Ω.α . we transform in a clever way the expression 0= d d ˙ ˙ ˙ f t2 − f −1 r2 − r2 θ2 − r2 φ2 sin2 θ .α − gαβ Ω. −1/f.α Ω.ν Ω.α . Therefore λ λ λ λ ν λ ν λ ν λ ν λ ν λ ν ˜ Rλ αµβ − R αµβ = Bαβ.λ .λ + Ω.ν + δν Ω.β + δβ Ω.ν Γν + Γλ δα Ω. λ λ ˜ Rαβ − Rαβ = δα Ω. λ λ λ ν λ ν λ ν λ ν ˜ Rαβ − Rαβ = Bαβ.λ + Ω. θ.λ ˜ ˜ 2 1 − gαβ [R − (N − 2) (N − 1) Ω. −r2.λ − Γλ Ω.λ − Bλα.α − 2 (N − 1) Ω.β + Bµν Γαβ + Γµν Bαβ + Bµν Bαβ − Bβν Γαµ − Γβν Bαµ − Bβν Bαµ .β − gαβ (N − 2) Ω.β + gβν Γν Ω. −r2 sin2 θ . = e−2Ω R + e−2Ω (N − 2) [Ω.ν + δν Ω.ν ) βν + 2N Ω.λ + Ω.αβ − gαβ Ω.λ Ω.αβ ] − gαβ (N − 2) Ω.β − N gαβ Ω.ν − δβ Ω.β − Ω. So we compute term by term.λ + Ω.λ Ω. 2 Note that there is no change in Gαβ in two dimensions (since the Einstein tensor is always equal to zero).α − gαβ Ω.λ − Γλ Ω.λ − gαβ.λ .ν βν αλ αβ λβ αβ λα = −gαβ Ω.α } .αβ ] − gαβ (N − 2) Ω.β − Ω.λ ν ν − N Ω.α Ω. ds 2 ∂r The equation for the covariant component u1 (s) is ˙ ˙ ˙ ˙ Using the metric gαβ = diag f. αλ βλ Now we note that some of the Γ terms can be absorbed into covariant derivatives.β + δβ Ω.ν − Γλ Ω.ν αβ λν .λ ν ν λ λ ν ν − Γλ (δα Ω.β − gβν Ω.α Ω.ν Γν + Γλ Ω. gβν Γν Ω.λ − gαβ Γλ Ω. We shall always raise and lower indices using the original metric gαβ .λ Ω.λ − gαβ.λ − N Ω. and uα = t. we find 1 d −f −1 r − ˙ dλ 2 df ˙2 d t − dr dr 1 f ˙ ˙ r2 − 2rθ2 − 2rφ2 sin2 θ ˙ = 0. φ . ˙ K= dλ dλ 22 .α ≡ N Ω.αβ ] − gαβ (N − 2) Ω.λ + δλ Ω.α Ω.ν − Γν Ω.λ + gβν Γν Ω.λ + gαν Γν Ω.α Ω. where ˙ ≡ d/dλ.α − gαβ Γλ Ω.ν − N gαβ Ω.αβ + N Ω. and also that the terms in the last bracket cancel.3 Equations of motion du1 1 ∂ − uα uβ (gαβ ) = 0.ν ) − δβ Ω.λ Ω. 1 ˜ ˜ Gαβ = Rαβ − gαβ R = Rαβ + (N − 2) [Ω.αβ ] + gαβ Ω.αβ + N Ω.λ = e−2Ω {R − (N − 2) (N − 1) Ω.λ .λ λν . (14) To derive this equation from other equations given in the lecture.λ − gαβ.ν δα Ω.ν αβ λν .α − gαλ Ω.λ Ω.α Ω.α .α ] − N (N − 2) Ω.λ λ λ ν ν + N Ω.α ] 2 (N − 2) (N − 3) = Gαβ + (N − 2) [Ω.αβ − Ω. The Einstein tensor is modified as follows. λ where N = δλ = gαβ g αβ is the number of spacetime dimensions.λ Ω.ν λν αβ λα λβ .ν Ω. r.

4 R0 The total deflection angle is found as δ = φ1 − φ2 − π. we try to separate terms of the form d dλ (uα ) out of the terms of the form d dλ (uα uα ) in the following way.1 Weak gravitational fields Gravitational bending of light d2 dφ2 1 r 1 3 rg . where φ1. Since rg ≪ R0 . Therefore. any one of these five equations is a consequence of four others. The unperturbed solution is v0 (φ) = 1 cos φ.g. ˙ ˙ ˙ f t2 = 2 t dλ dλ dλ etc. cos φ = ± + 2.. and also evaluate derivatives of the metric.β = 0. (no summation!).. 2 rg rg rg Only the solution with the minus sign is meaningful (cos φ < 1). = r 2 r2 2GM ≈ 3km. 7 7. Then ′′ v1 + v1 = 3 rg 3 rg 2 2 cos φ = 4 R2 (1 + cos 2φ) .µ = 0. we may expand this in Taylor series and find rg 3 3 + O(rg /R0 ). d d d d u1 u1 = g11 u1 u1 = 2u1 g11 u1 − u1 u1 g11 dλ dλ dλ dλ For example. Introduce an auxiliary variable v(φ) ≡ r−1 and solve the equation v ′′ + v = perturbatively. R0 3 rg v 2 2 where R0 is the distance of closest approach to the Sun. − 2φ dλ dλ Now we substitute the given equations (2)-(4). uβ uα .Namely. A= 3 rg 2. c2 In the lecture it was shown that the trajectory of a light ray in polar coordinates satisfies the equation + rg ≡ where M is the mass of the Sun. 4 R0 B=− 1 rg 2. assuming that v is small.2 are fixed by the condition v(φ) = 0. We find 0= d d ˙ ˙ ˙ f t2 − f −1 r2 − r2 θ2 − r2 sin2 θφ2 ˙ K= dλ dλ ˙ ˙ ˙ ˙ ˙ ˙ d f t − t2 df − 2r d f −1 r + r2 d f −1 − 2θ d r2 θ + θ2 d r2 ˙ ˙ ˙ = 2t dλ dλ dλ dλ dλ dλ ˙ ˙ ˙ d r2 sin2 θφ + φ2 d r2 sin2 θ . d ˙ d f t − t2 d f. We find a quadratic equation 2 2R0 R0 R0 cos2 φ − cos φ − 2 = 0. v(φ) = v0 (φ) + v1 (φ) + . cos φ ≈ − R0 23 . ˙ dλ f This is obviously equivalent to Eq. df /dλ = f ′ r. (14). so ˙ ˙ ˙ 0 = − t2 f ′ r − 2 r ˙ =r ˙ d f′ ˙ ˙ ˙ ˙ ˙ ˙ f −1 r − r2 2 r − 2θr2 φ2 sin θ cos θ + θ2 2rr + φ2 2r sin2 θr + 2φ2 r2 θ sin θ cos θ ˙ ˙ ˙ ˙ ˙ dλ f d f′ ˙ ˙ ˙ ˙ −2f −1 r − f ′ t2 − 2 r2 + 2rθ2 + 2rφ2 sin2 θ . v1 (φ) = A + B cos 2φ. 2 R0 0 The solution is found with undetermined coefficients. Note: the reason one of the equations follows from other equations is that the equation uα uα = const is a consequence of the four geodesic equations. e. when we consider the four geodesic equations and the equation uα uα = const. and the fact that gαβ.

i. where (. Also note that for any quantity X we have ˙ X .2 Einstein tensor for weak field For this problem Chapter 4 from the book Norbert Straumann “General Relativity and Relativistic Astrophysics” is useful.µ denotes a derivative ∂µ (.µ − hµν. h µ .. R0 2rg R⊙ 2rg /R⊙ = . thus 1 hµν ≡ γµν − 2 ηµν γ.i + Si .λβ = γ + γνλ.µν .µλ 2 2 µ. Here one can ask the students about the symmetry of this tensor.” 2nd German edition.λ λ. . 7.µν − ηµν hλβ + ηµν h . the angle φ is very close to π/2. we have Gµν = 1 λ λ γ + γν. (16) . 489 × 10−6 = = 1. φ1..β ] = ν. 489 × 10−6 × 360◦ /2π × 3600 [arc seconds] (see R.λ .βλ h. 2 ε≈ rg . Oloff “Geometrie der Raumzeit. g00 = 1 + 2Φ..νλ hµν − h.µ µν 2 2 µ.e.g. or finally Gµ = ν 7. 96 × 105 km and rg = 2.j . therefore 2rg /R⊙ = 8.).ν 1 µ. We have gµν = ηµν + hµν and Rµν = Γλµν. so we can write these indices in any position.βλ = 2 µ.µλ 2 µ.2. Using Eq.νλ Thus in the linear approximation we have 1 1 λ Gµν = Rµν − ηµν R = h + hλ − ν.i = hij = hij η ij = 0.).ν (15) where as usual we use the convention that indices are raised or lowered with η µν .λ .i = Fi . All 3-dimensional ν indices are raised and lowered using δij .3 Gravitational perturbations I The metric is written as gµν = ηµν + δgµν . Rµν = ν. 2 µλ. We shall use the formula for Gµ derived in Problem 7.i where Si. 1 Let us introduce a new variable γµν ≡ hµν − 2 ηµν h. (15) β we have 1 λ h + hλ − hµν − h.j = X.Therefore.λ γν − δν γλ.ν µ µ β. R0 /R⊙ R⊙ R0 For the Sun we have R⊙ = 6.νλ 1 . And for the Ricci scalar we obtain λ R = η µν Rµν = hλν − .λ − 2 λ. page 151). Furthermore 1 α 1 . j δgi = −δgij .ν + hνβ. −X. as convenient: j 0 δgj = δg0j = −δg0 . 954 km.ij + Fi.. 751” 8. The traces of two tensors h and γ are related by γ = −h.µ γ + γν.νλ where = η µν ∂µ ∂ν and h = hλ = η λα hαλ .i + hij . gij = −δij + 2Ψδij + 2E.ν .µλ − γµν − ηµν γ λβ . hij = hji . Γα = η αβ [hµβ.2 = ± This formula can be rewritten as δ= π +ε . hα ≡ η αλ hλβ . g0i = B. thus e.j + Fj. R0 ⇒ δ = 2ε = 2rg .α h + hα − hµν . 2 24 ¯ ≡ δg µ .λ − Γλλµ.β .µ − γµν − ηµν γβλ . Inserting the last expression for hµν in Gµν . We need to write the components of 1 µ¯ µ µ γν = δgν − δν h.

h µ 0 γ0 = Φ + 3Ψ + ∆E.the origin of the minus sign here is γj. (16).λ 0. 2Ψδij + 2E.j − ξj.j + Fj.i .0 . Under an infinitesimal transformation xµ → xµ + ξ µ .j + γj. (17) in full.j = Φ + 3Ψ + ∆ E − B .λ 2Gi = γλ.i . k2 25 Vj (V ) ≡ Vj − ikj Vl kl = Vj − ikj V (S) . (18)-(20) and pass to the Fourier space (where every variable is a function of a 3-vector k).i − j i i β.λ = −γ0j − γi. V (S) ≡ Vl kl .0 j.j + ∆Sj − ∆Fj .i − Si + (Φ − Ψ + ∆E).0 .j − ξ⊥j.j + ((Φ − Ψ − ∆E) δij + 2E.j − Sj + (Φ − Ψ + ∆E). k2 (21) .j + ∆Fj . and we use the fact that the background metric is diagonal.0 − 0 0 0 β.ij . and the decomposition ξµ = ξ 0 .ij ¨ ¨ ¨ ˙ Φ + 3Ψ + ∆E − 2∆B + ∆ (Φ − Ψ + ∆E) .λ + γλ . 0 g0i → g0i − ξ. Then we compute each component of Gµ separately: ν 0. j 0 γj = B. B.4 Gravitational perturbations II γ γ gαβ → gαβ − gαγ ξ.j ˙ + ∆Fj ˙ ˙ ¨ ¨ − B.i − ξi .λ = −γλ.i + ∆Fi .β − ξβ. by the Euclidean spatial metric δij rather than by ηij .i + hij → 2Ψδij + 2E. ξ⊥i + ζ.j + Fj.λ 0 ˙ ˙ ˙ γλ = γ0 − γj. the covariant components ξµ .i . .λ j λ ˙ ˙ γλ = −γj.i + Si → B.j − ξj.i − hij )..i ˙ ˙ ˙ = −B.i − ξi.λ λ.λ j.j + Sj ) = 4Ψ.i − hij = γi .β − gβγ ξ.β = 2 Φ + 3Ψ + ∆ E − B − ∂0 ∂0 (Φ + 3Ψ + ∆E) + ∆ (Φ + 3Ψ + ∆E) ¨ ¨ ¨ ˙ − Φ + 3Ψ + ∆E + 2∆B − ∆ (Φ − Ψ + ∆E) = 4∆Ψ. we adopt the convention of raising and lowering the spatial indices i.i − 2ζ.using the 3+1 decomposition: ¯ = δg µ = η µν δgµν = 2 (Φ − 3Ψ − ∆E) . i..i + ¨ ˙ ¨ hij − 2δij 2Ψ + ∆ Φ − Ψ − B + E λ.j − Sj + (Φ − Ψ + ∆E).j + Sj + ∆ (B. A vector Vi is decomposed into scalar and vector components as follows. (18) Substituting the perturbation variables from Eq. 0.j + ∆Fj . This will get rid of some minus signs. gij → gij − ξi.j ¨ ¨ ˙ ˙ ˙ −∗ −B.j + γλ.β = −B.λ ¨ ¨ ¨ ˙ γ0 − δ0 γλ.α = gαβ − ξα.λ ˙ ˙ ˙ γj − δj γλ.j − Sj + (Φ − Ψ + ∆E). This is easy to do if we perform a Fourier transform of Eqs.i = −B.i ˙ ˙ + −B.j β.j − Sj − (− (Φ − Ψ − ∆E) δij − 2E. j i γj = − (Φ − Ψ − ∆E) δij − 2E. Thus we have ˙ g00 → g00 − 2ξ 0 . We also denote ∂0 ≡ ∂t by the overdot.ij − Fi. we get ˙ Φ → Φ − ξ0. gij → gij − ξi. 7. 0 ˙ g0i → g0i − ξ.j + ∆Fj .i (19) (20) Now we need to separate these equations and derive the transformation laws for the individual perturbation variables. ˙ ˙ − ξ⊥i − ζ.j ¨ ¨ ¨ ˙ = Φ + 3Ψ + ∆E − 2∆B + ∆ (Φ − Ψ + ∆E) .ij + Fi.j − Fj.0 j.j + Fj.ij − Fi.ij + Fi.α . We can write the transformation of gαβ component by component using the 3+1 decomposition.i + Si − 0 ξ. Vj = ikj V (S) + Vj (V ) . j. ˙ ¨ = 2 Φ−Ψ−B+E ¨ ¨ ˙ ˙ + Fi.λ − j 0 0 β.β = Φ + 3Ψ + ∆ E − B .i + hij − ξ⊥i.i + hij ) − δij .0 .j + Sj = −γ0 = γ0j . the metric changes as (17) (This can be easily found from the standard formula for the change of coordinages.β = γλ ¨ ¨ ¨ ˙ ˙ ˙ = Φ + 3Ψ + ∆ E − B + −B. 0 g00 → g00 − 2ξ.j − Sj. Now we compute 0. To simplify calculations.λ * .) Now let ˜ us write Eq. .j − Sj + Φ − Ψ + ∆E .j + Fj.ij + Fi. using the perturbation variables (16). involving ∂ xµ /∂xν .0 2G0 = γλ.λ 2G0 = 2γλ.λ γλ.i .j − Fj.λ ˙ ˙ γj − δj γλ. ˙0 j.i − Si.

δgµν → 0. −B. to project the given vector Vi (k) onto the direction of ki . vector. These considerations depend rather crucially on the silently made assumption that all the metric perturbations vanish. and tensor components. (18)-(20). Without boundary conditions. A= 1 Xi. Tij = T (tr) δij + ∂i ∂j T (S) + ∂i Tj (V ) + ∂j Ti (V ) + Tij . So the tensor/vector/scalar decomposition is actually undefined without a fixed assumption about the boundary conditions. with slight modifications: there are some cosmetic factors of 2 and some minus signs. −B). Tj (V ) ≡i kl (2) T . Now we project Tij (2) (1) (1) onto ki kj and obtain the scalar component T (S) proportional to ki kj and the tensor Tij orthogonal to ki kj : Tij ≡ (2) (2) Tij (1) 1 (2) −ki kj + δij k 2 T (S) + Tij . there is no unique decomposition of the form Xi = A. ∆ because the function A is defined up to solutions of ∆A = 0. physically motivated set of boundary conditions. An explicit counterexample where these boundary conditions are not satisfied: gµν = diag (A. Tij ≡ Tij − Tll δij − ∂i ∂j − δij ∆ T (S) . where A = 1. Applying the decomposition method to Eqs.) The reason is that this gµν is a “perturbation” of flat metric with Φ and Ψ that do not decay to zero at spatial infinity. starting from the symmetric perturbation tensor δgij . −B. Let us go through this procedure in more detail. E → E − ζ. Tij = Tij − ∂i Tj − ∂j Ti . hij are obtained by this decomposition method. ∆ It may be convenient to gather the “trace” terms (the terms containing δij ) as one term. ˙ Si → Si − ξ⊥i . there will remain 6 independent components of perturbations (Φ. Other components will then show whether the geometry is really perturbed or it’s just a coordinate transformation of a flat space. 1 (1) (1) Tij ≡ Tij − δij Tll . we get the following transformation laws for the perturbation variables. It is clear that one can set Fi = 0. hij ).The idea is first. Alternatively. and second. 1. In real space. B = 1 are constants. Fi . 3 3 3 1 1 1 1 (2) ≡ ∂i ∂j Tij − Tll δij . we project onto ki and kj separately. to subtract the projection from Vi and to obtain the component of Vi which is transversal to ki . First. The same procedure applied to a symmetric tensor Tij leads to a decomposition into scalar. E. due to the subtraction of 3 k 2 δij in the first term. Tii = 0. 26 . The imaginary unit factors are added as coefficients at kj for convenience: with these factors. 2 k4 Tij ki kj = 0. 3 3 Note that the perturbation variables Ψ. at spatial infinity. one may do without Fourier transforms but then one still needs boundary conditions to solve the relevant Poisson equations for components. (T ) T (tr) ≡ 1 1 Tll − ∆T (S) . Fi → Fi − ξ⊥i . Ψ. k 2 jl Tij (T ) ≡ Tij − i ki Tj (2) (V ) + kj Ti (V ) .i . we subtract the trace and obtain the traceless part T (1) of the tensor T . ˙ B → B − ξ 0 − ζ. This metric is flat but one cannot see this by using the perturbation formalism! (The component Ψ = 0 cannot be removed by a gauge transformation. These boundary conditions are implicitly used when defining the Fourier transforms necessary for the tensor/vector/scalar decompositions (a Fourier transform is undefined without this boundary condition). Si . The boundary conditions δgµν → 0 at spatial infinity is a natural. hij → hij . the decomposition (21) translates to real space as Vj = ∂j V (S) + Vj (V ) . 3 (2) (V ) T (S) ≡ − 3 ki kj Tij . 3 Note the coefficient 1 3 that depends on the number of spatial dimensions (three).i + Bi . (2) 1 Note that Tij is again a trace-free tensor. Tii = 0. B = E = 0 with a coordinate transformation. 2 ∆2 3 3 3 1 (V ) (V ) (2) (T ) (2) = ∂i Tij . In general. So a coordinate transformation with ξ µ decaying to zero cannot bring this metric to ηµν . ˙ Φ → Φ − ξ0. Remarks: Ψ → Ψ. and a completely traceless (“tensor”) part such that (2) (T ) Tij kj = 0 and (T ) Tii (V ) = 0: (T ) Tij = iki Tj (V ) + ikj Ti + Tij . 2. Finally. to obtain a “vector” part Tj (T ) Tij such that Tj (V ) kj = 0. the full decomposition is Tij = T (S) Tj (V ) 1 1 (T ) (V ) (V ) Tll δij + ∂i ∂j − δij ∆ T (S) + ∂i Tj + ∂j Ti + Tij .

˜ 2 z = z. A× = 0 and then the opposite case. Note that the deformations change the shape of the ring in the same way. The 4-vectors describing the particles are uµ = (1. we have x=x+ ˜ 1 (A+ x + A× y) e−iω(t−z) . y components. ˜ ˜ Therefore. 0.ν + Γµ uν uα = Γµ . Now if we compute the shape of the ring in these coordinates. while A× describes a deformation in the directions at 45◦ .) Consider a plane wave moving in the z direction. The physically measured distances between the particles in the ring will change with time. hxy = hyx = A× e−iω(t−z) .2. The deformation of the ring is squeezing in one direction and expansion in the orthogonal direction. Consider a particle with constant 3-coordinates (x. It follows that A+ describes a deformation in the two vertical directions. ˜ 2 λ 2 It is easy to check that gµν xµ xν ≡ (ηµν + hµν ) xµ xν = ηµν xµ xν + O(h2 ). the ring will experience a deformation. 8. i. 0).1 Detecting gravitational waves Using distances between particles (This solution follows Hobson-Efstathiou-Lasenby [2006]. it is convenient to make a local coordinate transformation (local in the neighborhood of the ring) such that the metric becomes flat. This can be verified by performing a rotation by π . Therefore it is sufficient to consider a ring of particles situated in the x − y plane. and then it is straightforward to see that this will exchange A+ and A× .λ ) . z). it will be easy to interpret this shape in a straightforward way. ˜ To visualize the deformation.λ ) = 0. 2 1 y = y + (A× x − A+ y) e−iω(t−z) . it is clear that only changing lengths are between particles that have some separation in the x. However. 2 Therefore the coordinates xµ of each particle remain constant with time. 1 1 xµ = xµ + hµ xλ ≡ xµ + hαλ xλ η αµ . 4 x ˜ y ˜ 1 →√ 2 1 1 −1 1 x ˜ y ˜ .e.4.4.ν − hαν. y. it is convenient to consider first the case A+ = 0.8 8.e. let us imagine a cloud of particles initially at rest at different positions. except for the rotated orientation. Since the only nonzero components of hµν are the x. (22) To detect the presence of this gravitational wave. After the coordinate transformation. y directions. να 2 1 µ Γ00 = η λµ (hλ0. To visualize the deformation. the distance between each pair of particles is determined through the spacelike vector ∆xµ ≡ xµ − xµ as (1) (2) ∆L2 ≈ (ηµν + hµν ) ∆xµ ∆xν and will change with time because of the dependence on hµν . so one can easily see that these particles move along geodesics: uν uµ . 27 .α + hλα.0 − h00. xµ can be understood as the (approximate) Cartesian coordinates where the length is given by the usual ˜ Pythagorean formula. invariant under infinitesimal gauge transformations). (all other components of hµν are zero) hxx = −hyy = A+ e−iω(t−z) . it is very easy to verify that D = Φ − Ψ − B + E and Si − Fi are invariant under infinitesimal changes of coordinates (i. gµν xµ xν = ηµν xµ xν (up to second-order terms).0 + hλ0. να 00 1 λµ Γµ = η (hλν.1 Gravitational radiation I Gauge invariant variables ˙ ¨ ˙ Using the equations derived in Problem 7. 0.ν = uν uµ .2 8. §18. The trick that performs ˜ ˜ this transformation is the following.

(23). 0. let us compute the divergence of Xi . p. Note that the function ρ must fall off sufficiently rapidly as |x| → ∞ or else the integral (23) will not converge. where G(x) is the Green’s function. d2 S σ = Rσ S λ .2. 0. and there is no change in the z direction.i + bi . Xi. It is sufficient that |ρ(x)| ∼ |x|−2−ε at large |x| (where ε > 0). Therefore. z 2 (23) Therefore One can denote this integral more concisely. with boundary conditions φ → 0 at infinity. See arxiv:gr-qc/0605033 for nice explanations. the geodesic deviation equation becomes ¨ ¨ S σ = hσ S λ . φ(x) = − d3 k 3e ik·x 4πρ(k) k2 (2π) d3 k =− ∞ d3 k (2π) π 3e ik·x 4π k2 d3 ye−ik·y ρ(y) = d3 y ρ(y)G(x − y). ∆ 28 . Therefore a(x) = − One can write more concisely a= 1 4π d3 y Xi. An arbitrary 3-vector Xi (such as T 0 i ) is decomposed into scalar and vector parts as follows. 8. d3 y ρ(y). 00λ dt2 The Riemann tensor to first order in h can be expressed as ¨ R σ = hσ 00λ λ (note that hµ0 = 0). bi.i . A cloud of particles at rest in the gravitational field hµν described by Eq. The argument using the geodesic deviation (see Carroll.i = 0.3 Poisson equation ∆φ = 4πρ. 152-154) goes like this: The geodesic deviation equation can be simplified for a deviation vector S σ corresponding to nonrelativistic (almost stationary) particles moving with 4-velocity approximately equal to (1. To determine an explicit expression for a.4 Metric perturbations 1 Xi = a. k |x| Here we used the known integral sin z 1 dz = z 2 φ(x) = − 1 sin z dz = π. |x − y| 1 Xi. The solution given above is simple and straightforward.2 Using geodesic deviation equation PLEASE NOTE: The commonly found arguments that use the geodesic deviation equation are suspect because the geodesic deviation equation uses coordinates ξ µ rather than gauge-invariant quantities. is easy to find using the Fourier transform: −k 2 φ(k) = 4πρ(k). λ ¨x = ω 2 (hxx S x + hxy S y ) = ω 2 (A+ S x + A× S y ) e−iω(t−z) . φ = 4π 8.i (y). The general solution of the Poisson equation. Chapter 6.8.ii = ∆a. |x − y| 1 ρ. G(x) = − 1 ik·x 4π =− 3e 2 k π (2π) ∞ 0 dk 0 0 dθ sin θ eik|x| cos θ = − +∞ −∞ 2 π |x| ∞ 0 1 dk sin kx = − . S ¨ S y = ω 2 (hyx S x + hyy S y ) = ω 2 (A× S x − A+ S y ) e−iω(t−z) . (22) will stay indefinitely at rest in the coordinate system (ξ µ = const) even though the distances between particles will change with time. 0).i = a. ∆ 1 where the operator ∆ is just a shorthand notation for the integral in Eq.

8. σj = Tj. (24) 16πG which represents the vector part of the spatial Einstein equation (here σi is the vector part of the spatial Tij ).i + Tk .j µ= 1 Note that the operator ∆2 applied to a function f (x) is defined only if the function f has a sufficiently fast decay at −3−ε |x| → ∞. Now we solve this system of equations and find 3 1 i 1 i 1 1 Tii − Tk. In particular.i + βi + ∆λ.ik 2 i 1 1 i 1 i i i i T . j 0 Ti.ik . and (30) are consequences of Eqs. 1 1 ˙ ˜ Si = ∂t 16πGβi = − 16πG∆σi = −16πGσi . In the 3+1 decomposition.i . hij = ˙ ˜ Si = −16πGσi . (29).k − σk.i + hij = 8πGTji = 8πG µδik + λ.j 2 ˜ ˙ Si ≡ Si − Fi .j = 0.ik . ˙ α. It is sufficient that |f (x)| ∼ |x| with ε > 0 at large |x|. i Tk. To calculate the components λ.i = σi. ¨ ∆D + 2Ψ = −8πGµ. ˜ ∆Si = 16πGβi .ik = ∆∆λ + ∆µ. (T )i 16πGTj .i + Tk . 1 ˙ ¨ (Si − Fi ) = σi .5 Metric perturbations 2 1 0 T . (26). i Tk. ˙ D ≡ Φ − Ψ + B − E. also follows from the conservation of Tµν and from the other Einstein equations.i − ∆ ∆ ∆ k. ˙ Ψ = 4πGα.k + σk.i + βi ) . βi. µ of the EMT. α≡ βi ≡ Ti0 − (T )i . The Einstein equations are 0 2∆Ψ = 8πGT0 .k + σk.i = 0. j 0 T0.j = 0.i = 0.i + µ. ˙ α + ∆λ + µ = 0.k The energy-momentum tensor Tµν is decomposed as Ti0 = α. ˙ ˙ βi + ∆σi = 0.i = ∆λ. .k + ∆σk . (33) Then it is easy to see that Eqs.i + βi .0 + Ti. The conservation law of the EMT in 3+1 decomposition looks like this. (27). We need to verify that the equation (T )i Ti = 0.i + ∆Si = 8πGTi0 = 8πG (α.ij − δij ∆D + 2Ψ − where we have denoted for brevity 1 ˜ ˙ Si. 2 ∆ 2 ∆ k.k + µ. (25). ∆ k.k 1 0 T ∆ k.ik − σi. (25) (26) (27) (28) (29) (30) (31) which are gauge-invariant variables. (32) This gives therefore ˙0 T0 = ∆α.0 + T0.i i Tk = µδik + λ. (T ) Tk = Tk − µδk − λ. ∆ ∆ 29 . (T )i Tk. ˙0 T0 = ∆α. 2 ¨ D. the Einstein equations become 0 ∆Ψ = 4πGT0 . This is a faster decay than that required by the 1 operator ∆ . and the conservation laws (33). (28).ik + σi. D = 8πGλ. we compute − Tii = ∆λ + 3µ. λ = T − T i. 2 1 (T )i ˙ ˜ + Sj.i + ∆σi = 0.ik + σi. 1 ˜ ˙ 2Ψ.

9 9. We compute Pik.a δia 1 Ri Ri Ra = Pia . 9. 3 where Qik = Xik − It is easy to see that Xik differs from Qik only by a term of the form A(t.i − Pik Pab.i is proportional to Ri because Ri Xab. 2 So only the first term remains. The transverse-traceless part of δij is zero.i = 1 Pia Pbk − Pik Pab 2 Xab .i 2 2 1 1 1 2 Pik (Pai nb + Pbi na ) = − na Pbk + nk Pab − Pia (Pbi nk + Pik nb ) + R R R 2R 1 1 1 −2na Pbk + nk Pab − Pab nk − Pak nb + Pak nb + Pbk na = R 2 2 1 3 1 =− Pbk na + Pak nb .2 Matter sources The question is to verify the following property. 1 δik r2 T 0 0 d3 r. this term contains derivatives of Pab . for any Xab we have (T ) 1 1 Xii = Pia Xab Pbi − Pii Pab Xab = Pab Xab − 2Pab Xab = 0. Pik.i Pbk − Pik.i 1 2 = − (Pai nk + Pak ni ) . which are also sometimes proportional to Ri . R 2 2 This is higher-order in 1/|R| than Pab . 2 Therefore the transverse-traceless parts of Xik and Qik are the same. 30 .a = Ri R = .i = 1 Pia Xab Pbk − Pik Pab Xab 2 = . Xab.i 1 Pia Pbk − Pik Pab 2 1 Xab + Pia Pbk − Pik Pab Xab.i .a = −ni. .a Ri. 1 Pia Pbk − Pik Pab δab = 0.a . that is. (note that Pak nk = 0) R R 1 1 = Pia.i Pab + Pia Pbk. and its image has dimension 2.a 1 Pia Pbk − Pik Pab 2 ni.i = X(t − |R|) = − X ′ . Pii = 3 − ni ni = 2. At the same time. R)δij . 2 . Pik. − 2 |R|. (34) vanishes: 1 Pia Pbk − Pik Pab Ri = 0. a) First note that Pab is a projector.i Therefore the second term in Eq. Therefore. as required. (T ) Xik. R .i However.i = − nk . (T ) Xik = (T ) Qik .a = − 3 R R R R R . 2 2 b) We compute (T ) Xik.1 Gravitational radiation II Projection of the matter tensor Pab Pbc = Pac .i (34) Note that the projection kills any component proportional to Ri because Pia Ri = 0. the trace of Pab is 2.a nk − ni nk.

α ν.i = 0. 2 2 (1) because of the transverse traceless property of hµν .λ β.β − .β + hβµ.ν ) + hλν hλν.9.λ hν − h.ν αλ .ν − hβν. In this term.µ ) . 2 and it is given that Tαβ = 0.ν h.β λ. (2) (2)λ (2)λ (1) (1)λ (1)λ Rαβ = Γαβ.λ β.β .3 Energy-momentum tensor of gravitational waves See Hobson-Efstathiou-Lasenby [2006]. Rαβ = Γαβ.µ ) = − hλµ hλµ. η hλµ. we tackle the term Γβν Γαλ .α − hαβ. The Christoffel symbol up to second order is Γλ = αβ Γαβ (1)λ (2)λ Γαβ The Ricci tensor is 1 λµ (1)λ (2)λ η − hλµ (hµα. rename λ. The idea is to separate the second-order terms already in the Ricci tensor.α . 2 (1) (2) λ λ ν λ ν Rαβ = Γλ αβ.λ − hαλ.11. .α 4 1 λµ 1 .αλ + hαβ.λ .ν + hλν hαν.µ = 0.ν + hλ hν + hλ hν − hλ h.α = 2 2 1 1 (1)λ (1)λ = Γαβ.αβ ) + 2 31 1 λµ 1 hλν. Also.ν h.µλ ) = − hαβ .ν α. ν) = h.ν h.µλ + hλµ.αλ + hαβ.λ ) hλν − 2hλ h. Finally.λ hλν + h.µ ) .α − hαλ. since Rαβ is found from the first-order Einstein equation 1 (1) Rαβ − ηαβ R(1) = 8πGTαβ .α βλ αν βλ αν (symmetry of hµν ) = hλν.λ λ.β + βλ αν 4 .ν hλν − hλ h.β α. h0α = 0. We need to compute the second-order terms in the Einstein tensor.αβ ) + 2 1 = hλµ (−hαµ.λ + hλν hλν.αλ − hαβ.λ hν + h.ν λ.α αλ (move.ν λ h h − h.λ h.αλ − hαβ.β β.β .λ .β + Γλν Γαβ − Γβν Γαλ = Rαβ + Rαβ . g µν = η µν − hµν . note that now indices are always raised and lowered using ηµν .λ − Γλα. gµν = ηµν + hµν . hii = 0. Let us now evaluate these expressions and simplify as much as possible.β + hβµ.β .α β.λ 1 λ h + hλ − h.ν β.β αλ βν α.λ − Γλα.α β. β appear in similar positions: Γαβ = (1)λ (1)λ (1)ν 4Γβν Γαλ = hλ + hλ − h. it helps to write .β where again the indices are raised via η µν since we only need this term to first order.λ + (hβλ.ν h. it is given that the EMT of matter vanishes. as early as possible: Γαλ = Rαβ (1) (1)λ (1)λ (1)ν (1)λ (1)ν 1 λµ 1 λµ η hλµ .α hλµ.λ ν.µλ ) 2 (in the last line we used hλµ = 0). It follows that η µν hµν = 0. hµν .ν αλ βλ αν 1 λµ h (−hαµ.ν h.ν αλ βλ αν Finally.α hλν + 2h.β β. We will also try to simplify things by using the fact that hµν is purely transverse-traceless. Hence.λ + hλµ.α = 0.β + Γλν Γαβ − Γβν Γαλ . hik. which we will use below.βλ − hβµ. h hλµ.β βν (1)λ (1)ν hν + hν − h.ν α.α − hαβ. First we decompose the metric. We need to compute the Ricci tensor to second order.λ .β .α hλν + 2h.β + hµβ.β β.α β.ν (35) .α αβ 2 α.λ βν λ. §17.β . Then we can simplify this expression by grouping together terms where α.ν h.λ = − h (hαµ.α βν αλ (gather terms) = 2h. 2 1 = − hλµ (hαµ.ν − h.α .λ − Γλα. Let us now evaluate derivatives of the second-order terms in the Christoffel symbols: 1 1 (2)λ Γαλ = − hλµ (hαµ.ν λα .ν .β .α − hλν hαλ. 2 2 1 λµ 1 λµ (2)λ −Γαλ.λ − Γλα. Now.β = η λµ (hαµ. we put together the expression for Rαβ : Rαβ = (2) (2) (expand brackets) = hλ hν + hλ hν − hλ h.α 2 αλ β.β = h hλµ.βλ + hβµ.λ − 2hλ h.ν ν.αβ + h. Tµν = 0. we have hαβ = 0.ν αλ ν.ν αλ βλ αν 2 .ν − hλ h.µλ + hλµ.ν − hβν.λ − 2hλ h.ν + hλν (hαν. β.α − hαβ.βλ + hβµ.ν h hλµ.λ + hβλ. 2 2 1 λµ (2)λ Γαβ. 2 1 = η λµ (hαµ.µ ) = Γαβ + Γαβ .βλ − hβµ.

α Bν .γ . since Tαβ = 0. (36). h hλµ + hλµ h..) = 0 and so we may integrate by parts. αλ by hαβ = 0 and by the transverse traceless property of hµν .ν λ 1 λµ h h − h. then nl ql = |q| cos θ. we may use ηµν rather than gµν to compute the Einstein tensor: 1 (2) (2) Gαβ = Rαβ − ηαβ R(2) .α 32πG .ν 1 λµ 1 λµ 1 = h hλµ.α using Eq. so that boundary terms vanish. Then.αβ = − hλµ hλµ. for instance.β = 0.ν = − hαλ hβµ = 0. .4 Power of emitted radiation 1 dΩ = δ lm .β + .α + . for example Aµ Bν.β = hij hij.β = − hλµ hλµ. Finally. hλµ hλµ.α λµ 2 4 2 4 2 (36) where we again used the transverse traceless property of hµν and also hαβ = 0. otherwise we would have to write (η + h) R(1) + R(2) and pick up a second-order term hR(1) .αβ ) + hλµ hλµ. R(1) = 0.. and then we have gΩ (ql ) = 2π π −2i sin |q| 1 1 2π dφ dθ sin θ exp [−i |q| cos θ] = 4π 0 4π −i |q| 0 1 1 1 sin |q| 2 3 = 1 − ql ql + (ql ql ) − (ql ql ) + . In other words.αβ + h. qj =0 etc. we integrate Gαβ over a 4-dimensional region such that hµν = 0 and hµν. .α (37) hλµ hαν.λ = hλµ.ν h.µλ + hλµ. (2) (2) Now let us perform an averaging of the quantity Gαβ over both space and time.λ h. (35) and (36).α = − Aµ.βλ − hβµ. Finally.ν hαν. Gαβ = 8πG 32πG .α = 0 on the boundary of that region. The computation is easy if we introduce spherical coordinates with the z axis parallel to the vector ql .α − hαλ.α βλ αν 2 4 2 αλ β.β . = |q| 3! 5! 7! 32 .β 2 4 4 . For this reason we may use η αβ in Eq.βλ = − hλµ hαν. 4π 15 nl nm To derive the relations let us consider the generating function gΩ (ql ) ≡ dΩ exp −inl ql . √ where |q| ≡ ql ql .β .λ .ν hαν. Note that the first-order Ricci scalar is zero. since R(1) = 0...αλ + hαβ.The Ricci scalar is R(2) = η αβ Rαβ = (2) 1 λµ 3 1 1 . R(2) = 0. (GW) Tµν = − 1 1 1 (2) hλµ hλµ. Again. Many terms cancel in this way. 4π 3 dΩ 1 lm kr nl nm nk nr = (δ δ + δ lk δ mr + δ lk δ mr ). as long as Aµ Bν contains first powers of hαβ or hαβ. for example. After computing gΩ (ql ) it will be easy to obtain integrals such as the above: nl nm ∂ ∂ dΩ i gΩ (qj ) =i 4π ∂ql qm .ν − hαλ. (37). we obtain the required equation. we get Gαβ = (2) 1 1 . 2 We do not write the answer explicitly since it is a combination of Eqs.ν hβµ.α hλµ. 4π which is a function of a vector argument ql . Then ∂µ (.λ h (−hαµ.ν αλ 1 hαλ h.α 9.α hλµ.

3! 5! 3 30 1 1 1 2 = − + |q| δkl + ql qk . R R . .. trace-free tensor Aij (T T ) Aij ≡ 1 Pai Pbj − Pab Pij 2 Aab . in the leading order in 1/R.. Since Aii = 0. we have (again note that δab Aab = 0 and Aab = Aba ) 1 4π 1 4π and thus 1 1 Aac Abc δab = Aab Aab . 15 1 = (δjm δkl + δkm δjl + δlm δjk ) . instead of Qij .... we have (T T ) hij = 2G (T T ) Qij ≡ |R| 1 Pai Pbj − Pab Pij 2 ¨ Qij (t − |R|) .1. The flux of radiation in the direction nk is integrate this flux through a sphere of radius R: dE = R2 dt d2 Ω (GW ) T0k nk = R2 32πG d2 Ω hij hij. and we need to The perturbation hij is found from the Einstein equation.k nk . It was derived in the lecture that. t) 3 and is by definition trace-free. Thus we have hij. The projection tensor Pij is defined in Problem 9. . 3 3 2 1 Aab Acd (δab δcd + δac δbd + δad δbc ) = Aab Aab .. Qii = 0.... ∂gΩ ∂ql 2 ∂ gΩ ∂qk ∂ql ∂ 3 gΩ ∂qj ∂qk ∂ql ∂ 4 gΩ ∂qj ∂qk ∂ql ∂qm =− 2 4 1 1 2 2 ql + ql |q| − .. (T T ) .k = 16πG (T T ) ¨ Rk Qij (t − |R|) . = − + |q| ql .0 T0k nk . G dE = d2 Ω (T T ) Qij Qij . 5 .We have used the Taylor expansion for convenience of evaluating derivatives at |q| = 0. These derivatives can be found as follows.. d2 Ω Aab Acd na nb nc nd = 15 15 d2 Ω Aac Abc na nb = 1 4π d2 Ω (T T ) Aij (T T ) Aij = Aab Aab 1 − 33 2 1 2 + 3 2 15 = 2 Aab Aab .. The tensor Qij is defined by Qij (t) ≡ d3 x xi xj − 1 2 |x| δij T00 (x. the transverse-traceless part of Aij is defined by (T T ) It remains to compute the average over the sphere of Consider any symmetric. 3 30 15 1 = (qj δkl + qk δjl + ql δjk ) . 2 After integration over the sphere. 15 (GW ) Now we compute the intensity of radiation. Qab . dt 8π . Qij Qij . according to formulas derived above. we have Aab Pab = −Aab na nb and so (T T ) Aij (T T ) Aij = 1 1 Pai Pbj − Pab Pij Pci Pdj − Pcd Pij Aab Acd 2 2 1 1 = Pac Pbd − Pab Pcd Aab Acd = Pac Pbd − na nb nc nd Aab Acd 2 2 1 = Aab Aab − 2Aac Abc na nb + Aab Acd na nb nc nd . (T T ) .

¨0 T ij . (b) if there is no matter (vacuum) the scalar and vector components of the metric perturbation are equal to zero. then averaging must be performed over several periods of rotation. G . But they are very complicated. t) = Qik (t) ≡ 2G (T T ) ¨ Qik (t − |R|). −1) is the Minkowski metric for flat space and hµν is a small perturbation which is assumed to fall off to zero quickly at infinity. For instance.. i 0 0 ¨0 Tj. 0 Now we use a trick (See Hobson-Efstathiou-Lasenby. In principle. In the case of gravitational waves in flat background spacetime. we could just use the integral 0 d3 r T0 (r... A shortcut is to notice that the projection operator Pab does the job (Problems 9. The result is that (a) the variables E. Fi can be set to zero by choosing a coordinate system.5).1 and 9. d3 r T ij .) The result is (T ) hik (R. 2 d3 r T ij = d3 r T ab .i = −Tj.00 ≡ T0 . For this we need the energy-momentum tensor of gravitational waves. at least in the leading order in 1/|R|. The metric is assumed to be of the form where ηµν = diag(1. (T T ) . it is a traceless and symmetric tensor. In principle. 10 Derivation: gravitational waves in flat spacetime gµν = ηµν + hµν . and then we can approximately set (T ) hij ≈ − 4G |R| d3 r (T ) Tji (t − |r − R| . now we would like to compute the energy radiated in the gravitational waves. Now. This is a rather nontrivial object. we have the formulas for this (see Problem 8. Averaging is performed over timescales larger than the typical timescale of change in the source.. Consider first the tensor Tji rather than its transverse-traceless part (T ) Tji . if the source is a rotating body. The trick is to write the integral (out of sheer luck) d3 r ∂a ∂b ri rj T ab = 2 Then we integrate by parts and use the conservation laws (32). (c) the tensor component hij satisfies the wave equation (31). i 0 Tj.3. so we take the limit |R| ≫ |r|.ab ri rj = − ¨0 d3 r ri rj T0 . we need to obtain the transverse-traceless part of the tensor. 7. −1.2. (We are only interested in everything to leading order in 1/ |R| since all smaller terms will not give any flux of radiated energy. r) ⇒ f (t.4) in terms of the perturbation variables Φ. we find dE G 1 = dt 2 4π d2 Ω (T T ) . |r − R| We will use this formula for f ≡ (T ) hij and A ≡ 16πG(T ) Tji . Ψ. for us.. For instance. that we need to average over time (since Qij is a function only of time). but it is more convenient to use Qik .. substituting Qij instead of Aij . t) ri rk − r2 δik . Qij Qij . . since in general the gravitational field does not have any energy-momentum tensor. Now.. r) = A(t. we are interested in describing the radiation sent far away by a matter distribution. t)ri rk . We start with a 3+1 decomposition of the metric perturbation hµν and compute the Einstein tensor (see Problems 7. Solutions of the wave equation in four dimensions with retarded boundary condition can be written using the known Green’s function..ij = −Tj. §17. one can define some quantity (GW ) Tµν which looks like the energy-momentum tensor of 34 . etc..9) to express the components Tij through T0 . We also decompose the matter energy-momentum tensor Tµν and obtain the Einstein equations separately for each component (8..2).. Since we found the tensor perturbation (T ) hij .5). |R| 1 0 d3 r T0 (r.0 .0j = T0. B.ij = −T0 . 7. 3 (39) The tensor Qik is the quadrupole moment of energy distribution. Qij Qij = 5 (38) The angular brackets indicate that we must perform an averaging over spacetime domains. it is much easier 0 to compute with T0 because this is just the energy density of matter. R) = − 1 4π d3 r A(t − |r − R| . if f (t. Finally. r) . r). This means. −1. (40) because the transverse-traceless parts of (40) and of Qik are the same.

We have designed this License in order to use it for manuals for free software. We will compute this quantity below.4). Inc. But of course this is not a real derivation because this does not show why the quantity (GW) Tµν has anything to do with the energy carried by waves. regardless of subject matter or whether it is published as a printed book. USA Everyone is permitted to copy and distribute verbatim copies of this license document. Then we will get a correction h(2) to the solution. This kind of averaging is assumed in Eq. one . The result is used to compute the power radiated in gravitational waves (Problem 9. 8πG β This motivates us to say that the EMT for gravitational waves is given by this formula.. From Eq. Suite 330. but it looks as if there is an additional term in the energy-momentum tensor. If we want to insert factors of c. which means that derivative works of the document must themselves be free in the same sense.. by integrating over space where the matter is contained. much larger than the typical length scale of the sources. (38). One can use the formula (38) to compute the gravitational radiation emitted by nonrelativistic matter far away from 0 those places where the matter is contained. So gravitation should be also sensitive to the energy in gravitational waves. the solution g = η + h(1) + h(2) will be more precise. T0 (r. 59 Temple Place. 35 . because free software needs free documentation: a free program should come with manuals providing the same freedoms that the software does. should be given. Secondarily. the trace. like every other energy-momentum tensor for other kinds of matter. with or without modifying it. computes the third derivative Qij . (2)α A GNU Free Documentation License Version 1. (42). Gravitation is sensitive to every kind of energy..gravitational waves (but actually is not even a generally covariant tensor). We will only show a heuristic justification. (41) Here G(1) is the first-order Einstein tensor. Note that the calculation of Gβ uses averaging over spacetime in an essential way. it can be used for any textual work. = 8πGTβ . But this License is not limited to software manuals.2. We will guess the formula for (GW ) Tµν as follows. and averages over long time. The real justification for using this procedure is complicated and is beyond the scope of this introductory course of General Relativity. (38).2002 Free Software Foundation. We can try to get a more precise solution by using the second-order terms in Eq. G(2) is the second-order etc. will act as an additional source for gravity. as indicated in Eq. (GW) α Tβ ≡ − 1 (2)α (1) G [h ].3. because the energy-momentum tensor acts as a “source” for gravity (it is on the right-hand side of the Einstein equation). either commercially or noncommercially. Preamble The purpose of this License is to make a manual. (41) we find Gβ (1)α [h(1) + h(2) ] + Gβ (2)α α [h(1) ] = 8πGTβ .. which is the following.2001. t). we replace G by Gc−9 . the result is an averaged power radiated during a long time—much longer than the typical time scale of change in the sources—and averaged over large distances. We recommend this License principally for works whose purpose is instruction or reference. The distribution of the energy density. Thus. First we solve only to first-order in h (this is (1) what we have been doing so far) and then we will get an approximate solution hµν : Gβ (1)α α [h(1) ] = 8πGTβ . textbook. (39). This License is a kind of “copyleft”. Then one computes the tensor Qij according to Eq. Boston. but changing it is not allowed. (41). It complements the GNU General Public License. (GW ) Tµν (if we know how to compute it). One expects that the energy-momentum tensor for gravitational waves. or other functional and useful document free in the sense of freedom: to assure everyone the effective freedom to copy and redistribute it. which is a copyleft license designed for free software. This quantity is useful because it gives the correct value of the energy after one integrates over a large region of spacetime. Finally. which we may rewrite as Gβ (1)α α α [h(1) + h(2) ] = 8πG Tβ + (GW) Tβ . It remains unclear exactly how one performs averaging over space and time. MA 02111-1307. Now this is similar to Eq. while not being considered responsible for modifications made by others. We can write the Einstein equation and expand it in powers of the perturbation hµν : Gα [ηµν + hµν ] = Gβ β (1)α [h] + Gβ (2)α α [h] + . this License preserves for the author and publisher a way to get credit for their work. November 2002 Copyright (c) 2000. (42) α This solution disregards the effect of gravitational waves and only takes into account the effect of matter Tβ . (2)α The second-order terms Gβ are computed in Problem 9.

SGML or XML using a publicly available DTD. The “Cover Texts” are certain short passages of text that are listed. You accept the license if you copy. The “Document”.2. Examples of suitable formats for Transparent copies include plain ASCII without markup. you must enclose the copies in covers that carry. the copyright notices. Both covers must also clearly and legibly identify you as the publisher of these copies.0 Applicability and definitions This License applies to any manual or other work. represented in a format whose specification is available to the general public.) To “Preserve the Title” of such a section when you modify the Document means that it remains a section “Entitled XYZ” according to this definition. under the same conditions stated above. either commercially or noncommercially. in the notice that says that the Document is released under this License. and you may publicly display copies. A copy made in an otherwise Transparent file format whose markup. commercial. SGML or XML for which the DTD and/or processing tools are not generally available. “Endorsements”. ethical or political position regarding them. that contains a notice placed by the copyright holder saying it can be distributed under the terms of this License. legibly. numbering more than 100. and standard-conforming simple HTML.A. below. in any medium. If a section does not fit the above definition of Secondary then it is not allowed to be designated as Invariant. The front cover must present the full title 36 . you may accept compensation in exchange for copies. or with modifications and/or translated into another language. Opaque formats include proprietary formats that can be read and edited only by proprietary word processors. A “Transparent” copy of the Document means a machine-readable copy. (Here XYZ stands for a specific section name mentioned below. You may also lend copies. or of legal. but only as regards disclaiming warranties: any other implication that these Warranty Disclaimers may have is void and has no effect on the meaning of this License. philosophical. However. A section “Entitled XYZ” means a named subunit of the Document whose title either is precisely XYZ or contains XYZ in parentheses following text that translates XYZ in another language. For works in formats which do not have any title page as such. all these Cover Texts: Front-Cover Texts on the front cover. A Front-Cover Text may be at most 5 words. that is suitable for revising the document straightforwardly with generic text editors or (for images composed of pixels) generic paint programs or (for drawings) some widely available drawing editor. plus such following pages as are needed to hold. as Front-Cover Texts or Back-Cover Texts. or absence of markup. If you distribute a large enough number of copies you must also follow the conditions in section A. clearly and legibly. modify or distribute the work in a way requiring permission under copyright law. the material this License requires to appear in the title page. and a Back-Cover Text may be at most 25 words. Such a notice grants a world-wide. A. as being those of Invariant Sections. A. A “Modified Version” of the Document means any work containing the Document or a portion of it. and that you add no other conditions whatsoever to those of this License. provided that this License. the title page itself. The Document may contain zero Invariant Sections. and the license notice saying this License applies to the Document are reproduced in all copies. in the notice that says that the Document is released under this License.1 Verbatim copying You may copy and distribute the Document in any medium. preceding the beginning of the body of the text. and that is suitable for input to text formatters or for automatic translation to a variety of formats suitable for input to text formatters. (Thus. if the Document is in part a textbook of mathematics. such as “Acknowledgements”. Any member of the public is a licensee. unlimited in duration. “Title Page” means the text near the most prominent appearance of the work’s title. “Dedications”. royalty-free license. a Secondary Section may not explain any mathematics.) The relationship could be a matter of historical connection with the subject or with related matters. A “Secondary Section” is a named appendix or a front-matter section of the Document that deals exclusively with the relationship of the publishers or authors of the Document to the Document’s overall subject (or to related matters) and contains nothing that could fall directly within that overall subject. If the Document does not identify any Invariant Sections then there are none. An image format is not Transparent if used for any substantial amount of text. The Document may include Warranty Disclaimers next to the notice which states that this License applies to the Document. A copy that is not “Transparent” is called “Opaque”. or “History”. The “Title Page” means. You may not use technical measures to obstruct or control the reading or further copying of the copies you make or distribute. Examples of transparent image formats include PNG. Texinfo input format. PostScript or PDF designed for human modification. and is addressed as “you”. PostScript or PDF produced by some word processors for output purposes only. to use that work under the conditions stated herein. XCF and JPG. These Warranty Disclaimers are considered to be included by reference in this License.2 Copying in quantity If you publish printed copies (or copies in media that commonly have printed covers) of the Document. refers to any such manual or work. and the machine-generated HTML. for a printed book. The “Invariant Sections” are certain Secondary Sections whose titles are designated. has been arranged to thwart or discourage subsequent modification by readers is not Transparent. and the Document’s license notice requires Cover Texts. A L TEX input format. and Back-Cover Texts on the back cover. either copied verbatim.

and publisher of the Document as given on its Title Page. as long as they preserve the title of the Document and satisfy these conditions. immediately after the copyright notices. To do this. If you publish or distribute Opaque copies of the Document numbering more than 100. Preserve the Title of the section. you should put the first ones listed (as many as fit reasonably) on the actual cover. as the publisher. on explicit permission from the previous publisher that added the old one. and add to it an item stating at least the title. to the end of the list of Cover Texts in the Modified Version. Preserve all the Invariant Sections of the Document. or state in or with each Opaque copy a computernetwork location from which the general network-using public has access to download using public-standard network protocols a complete Transparent copy of the Document. You may omit a network location for a work that was published at least four years before the Document itself. These titles must be distinct from any other section titles. You may add a passage of up to five words as a Front-Cover Text. D. unless they release you from this requirement. but not required. if any) a title distinct from that of the Document. List on the Title Page. K. A. if any. Preserve the network location. If there is no section Entitled “History” in the Document. a license notice giving the public permission to use the Modified Version under the terms of this License. 37 . and publisher of the Modified Version as given on the Title Page. when you begin distribution of Opaque copies in quantity. Do not retitle any existing section to be Entitled “Endorsements” or to conflict in title with any Invariant Section. H. you must do these things in the Modified Version: A. if there were any. If the Document already includes a cover text for the same cover. You may use the same title as a previous version if the original publisher of that version gives permission. you must either include a machine-readable Transparent copy along with each Opaque copy. These may be placed in the “History” section. provided that you release the Modified Version under precisely this License. but you may replace the old one. F. if it has fewer than five). C. Copying with changes limited to the covers. statements of peer review or that the text has been approved by an organization as the authoritative definition of a standard. O. Include.3 Modifications You may copy and distribute a Modified Version of the Document under the conditions of sections A. with the Modified Version filling the role of the Document. E. If the required texts for either cover are too voluminous to fit legibly. and preserve in the section all the substance and tone of each of the contributor acknowledgements and/or dedications given therein. be listed in the History section of the Document). M. then add an item describing the Modified Version as stated in the previous sentence.2 above. Only one passage of Front-Cover Text and one of Back-Cover Text may be added by (or through arrangements made by) any one entity.1 and A. and from those of previous versions (which should. State on the Title page the name of the publisher of the Modified Version. Such a section may not be included in the Modified Version. or if the original publisher of the version it refers to gives permission. free of added material. as authors. in the form shown in the Addendum below. If you use the latter option. that you contact the authors of the Document well before redistributing any large number of copies. add their titles to the list of Invariant Sections in the Modified Version’s license notice. given in the Document for public access to a Transparent copy of the Document. year. Preserve all the copyright notices of the Document. new authors. provided it contains nothing but endorsements of your Modified Version by various parties—for example. together with at least five of the principal authors of the Document (all of its principal authors. one or more persons or entities responsible for authorship of the modifications in the Modified Version. to ensure that this Transparent copy will remain thus accessible at the stated location until at least one year after the last time you distribute an Opaque copy (directly or through your agents or retailers) of that edition to the public. and a passage of up to 25 words as a Back-Cover Text. year. Section numbers or the equivalent are not considered part of the section titles. B. and likewise the network locations given in the Document for previous versions it was based on. If the Modified Version includes new front-matter sections or appendices that qualify as Secondary Sections and contain no material copied from the Document. J. create one stating the title. Preserve in that license notice the full lists of Invariant Sections and required Cover Texts given in the Document’s license notice. For any section Entitled “Acknowledgements” or “Dedications”. Delete any section Entitled “Endorsements”. L. you may at your option designate some or all of these sections as invariant. You may add other material on the covers in addition. thus licensing distribution and modification of the Modified Version to whoever possesses a copy of it. N. you may not add another. Preserve any Warranty Disclaimers. I. previously added by you or by arrangement made by the same entity you are acting on behalf of. In addition. can be treated as verbatim copying in other respects.with all words of the title equally prominent and visible. authors. and continue the rest onto adjacent pages. to give them a chance to provide you with an updated version of the Document. Preserve its Title. You may add a section Entitled “Endorsements”. unaltered in their text and in their titles. Include an unaltered copy of this License. Preserve the section Entitled “History”. Use in the Title Page (and on the covers. Add an appropriate copyright notice for your modifications adjacent to the other copyright notices. G. It is requested. you must take reasonably prudent steps.

you must combine any sections Entitled “History” in the various original documents. the name of the original author or publisher of that section if known. Replacing Invariant Sections with translations requires special permission from their copyright holders. in parentheses. Such new versions will be similar in spirit to the present version.2 is applicable to these copies of the Document. Aggregation with independent works A compilation of the Document or its derivatives with other separate and independent documents or works. See http://www. is called an “aggregate” if the copyright resulting from the compilation is not used to limit the legal rights of the compilation’s users beyond what the individual works permit. You may include a translation of this License. or “History”.” Collections of documents You may make a collection consisting of the Document and other documents released under this License. or rights. from you under this License will not have their licenses terminated so long as such parties remain in full compliance. but you may include translations of some or all Invariant Sections in addition to the original versions of these Invariant Sections. but may differ in detail to address new problems or concerns. or else a unique number. modify. and any sections Entitled “Dedications”. If a section in the Document is Entitled “Acknowledgements”.The author(s) and publisher(s) of the Document do not by this License give permission to use their names for publicity for or to assert or imply endorsement of any Modified Version. then if the Document is less than one half of the entire aggregate. or the electronic equivalent of covers if the Document is in electronic form. likewise combine any sections Entitled “Acknowledgements”. Future revisions of this license The Free Software Foundation may publish new. the original version will prevail. If the Document specifies that a particular numbered version of this License “or any later version” applies to it. forming one section Entitled “History”. Make the same adjustment to the section titles in the list of Invariant Sections in the license notice of the combined work. and all the license notices in the Document. in or on a volume of a storage or distribution medium.org/copyleft/. You must delete all sections Entitled “Endorsements. sublicense. and follow this License in all other respects regarding verbatim copying of that document. you have the option of following the terms and 38 .0) will typically require changing the actual title. and replace the individual copies of this License in the various documents with a single copy that is included in the collection. If there are multiple Invariant Sections with the same name but different contents. and multiple identical Invariant Sections may be replaced with a single copy. under the terms defined in section 4 above for modified versions. In case of a disagreement between the translation and the original version of this License or a notice or disclaimer. When the Document is included an aggregate. However. sublicense or distribute the Document is void. parties who have received copies. The combined work need only contain one copy of this License. and distribute it individually under this License.gnu. Termination You may not copy. the Document’s Cover Texts may be placed on covers that bracket the Document within the aggregate. modify. If the Cover Text requirement of section A. so you may distribute translations of the Document under the terms of section A. and any Warrany Disclaimers. and list them all as Invariant Sections of your combined work in its license notice. provided that you include in the combination all of the Invariant Sections of all of the original documents.3) to Preserve its Title (section A. provided that you follow the rules of this License for verbatim copying of each of the documents in all other respects. the requirement (section A. Each version of the License is given a distinguishing version number. provided you insert a copy of this License into the extracted document. and that you preserve all their Warranty Disclaimers. or distribute the Document except as expressly provided for under this License. Otherwise they must appear on printed covers that bracket the whole aggregate.3. Any other attempt to copy. this License does not apply to the other works in the aggregate which are not themselves derivative works of the Document. make the title of each such section unique by adding at the end of it. provided that you also include the original English version of this License and the original versions of those notices and disclaimers.“Dedications”. In the combination. unmodified. You may extract a single document from such a collection. and will automatically terminate your rights under this License. Combining documents You may combine the Document with other documents released under this License. Translation Translation is considered a kind of modification. revised versions of the GNU Free Documentation License from time to time.

USA Everyone is permitted to copy and distribute verbatim copies of this license document. Boston. replace the “with. Version 1. If the Document does not specify a version number of this License. Copyright Copyright (c) 2000.2 or any later version published by the Free Software Foundation. 59 Temple Place.. 39 . 2001. you may choose any version ever published (not as a draft) by the Free Software Foundation. to permit their use in free software. Inc.Texts. distribute and/or modify this document under the terms of the GNU Free Documentation License. MA 02111-1307. we recommend releasing these examples in parallel under your choice of free software license.. merge those two alternatives to suit the situation. A copy of the license is included in the section entitled “GNU Free Documentation License”. but changing it is not allowed. Permission is granted to copy. Front-Cover Texts and Back-Cover Texts. no Front-Cover Texts. If you have Invariant Sections. If you have Invariant Sections without Cover Texts.conditions either of that specified version or of any later version that has been published (not as a draft) by the Free Software Foundation. 2002 Free Software Foundation. and with the Back-Cover Texts being <list>. or some other combination of the three. with the Front-Cover Texts being <list>. with no Invariant Sections. ADDENDUM: How to use this License for your documents To use this License in a document you have written. and no Back-Cover Texts. include a copy of the License in the document and put the following copyright and license notices just after the title page: Copyright (c) <year> <your name>. Suite 330.” line with this: with the Invariant Sections being <list their titles>. If your document contains nontrivial examples of program code. such as the GNU General Public License.

Sign up to vote on this title
UsefulNot useful