You are on page 1of 27

Department of Physics

Gravitational Waves from Slow Motion Binary Systems

Daniel Aho, Ask Ellingsen


danaho@kth.se
askel@kth.se
SA114X Degree Project in Engineering Physics, First Cycle
Department of Physics
KTH Royal Institute of Technology
Supervisor: Mattias Blennow

20th May 2017


Abstract

In this thesis we study gravitational waves in the domain of linearised general relativity. We
present the fundamental ideas and theory of general relativity, then, using the traditional
means of quadrupole approximation, we derive an expression for the power radiated as grav-
itational waves in a binary system. We limit ourselves to binary systems in the Newtonian
limit, and can therefore use Kepler’s laws in our calculations. We then use the derived expres-
sion to explicitly predict the power radiated by the Sun-planet systems in our solar system
specifically, regarding each system as a binary consisting of the planet and the Sun. Finally,
we calculate the resulting orbital decay and find an expression for the time it would take for
the bodies to come into contact if gravitational radiation were the only means by which the
system lost energy. As expected, the power radiated by gravitational waves is found to be
very small for systems in the Newtonian limit, and the corresponding time until impact is
found to be on the order of many times the age of the Universe.
Sammanfattning

I detta kandidatexamensarbete studerar vi gravitationsvågor utifrån linjäriserad allmän


relativitetsteori. Vi presenterar först de grundläggande idéerna bakom relativitetsteorin, sen
använder vi den traditionella kvadrupolapproximationen för att härleda ett uttryck för den
effekt som strålas ut i form av gravitationsvågor från ett binärt system. Vi begänsar oss
till system i den Newtonska gränsen och kan därmed använda oss av Keplers lagar i våra
beräkningar. Uttrycket vi härleder använder vi sedan för att explicit förutspå hur mycket
effekt som strålas ut av de olika sol-planetsystemen i vårt eget solsystem, där vi betraktar
varje system som ett binärt system bestående av planeten och solen. Slutligen beräknar vi
omloppsbanans förfall till följd av den utstrålade energin, och hittar ett uttryck för hur lång
tid det skulle ta för kropparna att komma i kontakt med varandra om de endast förlorade
energi genom gravitationsstrålning. Som förväntat finner vi att den utstrålade effekten är
mycket liten för system i den Newtonska gränsen, och den motsvarande tiden till kollision
uppskattas till många gånger Universums ålder.
Acknowledgements
We would like to thank our supervisor Dr. Mattias Blennow for insightful super-
vision and for being willing to spend his entire annual budget on red ink.
Contents

1 Introduction 2

2 Background Material 4
2.1 Conventions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.2 A new structure of space and time . . . . . . . . . . . . . . . . . . . . . . . . 4
2.3 The equivalence principle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.4 Curved spacetime as Riemannian manifolds . . . . . . . . . . . . . . . . . . 6
2.5 The Einstein field equations . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

3 Investigation 9
3.1 Gravitational waves . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
3.2 Generation of gravitational waves . . . . . . . . . . . . . . . . . . . . . . . . 12
3.3 Mathematical model of the binary system . . . . . . . . . . . . . . . . . . . 13
3.4 Analytical calculations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
3.4.1 Energy loss for circular orbits . . . . . . . . . . . . . . . . . . . . . . 14
3.4.2 Correction for elliptical orbits . . . . . . . . . . . . . . . . . . . . . . 15
3.4.3 Orbital decay due to energy loss . . . . . . . . . . . . . . . . . . . . . 15
3.5 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
3.6 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3.6.1 A new spectrum to study . . . . . . . . . . . . . . . . . . . . . . . . 19
3.6.2 Slow motion binaries . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

4 Summary and Conclusions 21

Bibliography 22

1
1. Introduction

The purpose of this thesis is to study gravitational waves from slow-motion (SM) binary
systems of heavenly bodies using Einstein’s general theory of relativity (GR), with primary
focus on the energy radiated away from these systems.
GR is a classical field theory of gravitation [1]. It extends Newton’s laws of gravitation to
strong gravitational potentials and velocities close to the speed of light. Unlike almost every
other theory in physics the development of GR was originally the result of pure theoretical
research of one single mind - Albert Einstein’s. What motivated Einstein to search for a new
theory of gravity to begin with was the following [2]:
(i) To have a completely relativistic theory of gravitation, a theory that replaces Newton’s
concept of an “action at distance” force.
(ii) To have a deeper understanding of the observed equivalence of gravitational mass and
inertial mass.
(iii) To emphasize that space itself is not a thing. The laws of physics should not depend
on some arbitrary choice of reference frame, they should be the same in any reference
frame.
GR lies at the heart of astronomy and cosmology and can be used to predict and explain
many exotic physical phenomena such as the perihelion shifts of planetary motions, grav-
itational lensing, black holes, gravitational time dilation, and of course the phenomenon of
interest for this thesis: gravitational waves.
An analogy with classical electrodynamics may be helpful. According to Maxwell’s theory
of electromagnetism, an electrically charged particle gives rise to an electric field everywhere
in space. This field is described mathematically by Maxwell’s equations [3]. In turn, the field
itself acts on other charged particles - called test particles - telling them how to move. This
is the same kind of two-step process
Source Ñ field Ñ test particle
that occurs in GR. Matter and energy curve spacetime, and this curvature is what we call a
gravitational field. This field is then described by the Einstein field equations, and particles
move according to these equations following so called geodesic curves, or “the shortest and
straightest path possible”. The basics of GR are presented in somewhat more detail in Section
2.

2
The motion of the gravitational sources can cause “ripples” in spacetime that propagate
outward from the source. This is what we call gravitational waves. This phenomenon was
first predicted by Einstein himself in a 1916 paper [4]. However, in his discussion of gravita-
tional waves he made a mistake in the calculations [5]. This, as well as judging his previous
presentation to be “not sufficiently transparent”, led him to publish a follow up paper in 1918
focusing exclusively on gravitational waves, aptly named “On Gravitational Waves”. In this
paper Einstein worked with gravitational waves using the same linearised equations we will
use in this thesis.
Whereas Einstein predicted the existence of gravitational waves as early as 1916, they
were not experimentally confirmed to exist until the 1970:s. In 1974 Russell Hulse and
Joseph Taylor discovered an unusual binary system named PSR 1913+16 [6, 7] consisting of
two neutron stars. Over the years they managed to observe indirect evidence of gravitational
radiation by measuring the orbital decay of this binary pulsar. The experimental results
obtained from these observations agree exactly with the predictions of GR. This was the first
ever indirect evidence for the existence of gravity waves.
More recently, on September 14 2015, the two detectors of the Laser Interferometer
Gravitational-wave Observatory (LIGO) both simultaneously observed the binary black hole
merger GW(150914) [8], marking the first direct detection of gravitational waves. This
groundbreaking experiment meant that virtually all of the predictions made by GR had now
been confirmed.
Just like electromagnetic waves, gravitational waves carry energy away from their source.
In this thesis, we will focus on the energy loss of slow motion (SM) binary systems∗
This thesis is structured as follows: In Section 2 we give a deeper introduction to GR. In
Section 3 we show how the gravitational wave equation is obtained by linearising Einstein’s
field equations. We then use an analogy to electromagnetic theory to study the motion of
a general SM binary system, as well as analyze some special cases; namely the Sun-planet
systems in our solar system. We finally calculate the energy loss of these systems as well as
the resulting orbital decay.


“Slow motion” in an astrophysical, relativistic context can be very fast indeed compared to everyday
speeds.

3
2. Background Material

2.1 Conventions
In this section we present some of the basics of GR required to understand gravitational radi-
ation. In order to work more efficiently with GR, we employ a couple of common conventions.
First, we use so called geometrised units. In these units, we define

G “ c “ 1, (2.1)

where G is Newton’s constant and c is the speed of light in vacuum.


In order to simplify expressions involving sums over indices we also employ the Einstein
summation convention. Whenever a term in an expression has a repeated index with one
index up and one index down, this index is implicitly summed over. We furthermore make
use of the convention that Latin indices take on the values 1, 2 and 3, whereas Greek indices
range from 0 to 3. In other words, we let
3
ÿ
xi y i “ xi y i , (2.2)
i“1
ÿ3
xµ y µ “ xµ yµ . (2.3)
µ“0

These conventions simplify many calculations and are used throughout this thesis.

2.2 A new structure of space and time


Einstein’s perhaps greatest insight in the formulation of special relativity (SR) was that
space and time were not, as was previously believed, separate entities. Rather, they together,
on equal footing, form a larger unit called spacetime. SR predicts that this equivalence of
space and time means that arbitrary choices of reference frames can and will cause counter-
intuitive phenomena like time dilation and length contraction. Besides the obvious difficulties
of reconciling intuition based on the very slow speeds of everyday events with the new, exotic
effects of relativity, this also poses a problem of notation: If space and time are not absolute,
how do we label points in spacetime? Einstein’s answer was to use events as labels [9]. By

4
specifying what happens at each point in spacetime, all of it can be catalogued. The distance
between events in different directions might change with the reference frame, but the events
and their causal correlations to each other remain the same. This is the physical intuition
behind what spacetime is.

2.3 The equivalence principle


The step from SR to GR is taken by incorporating the effects of gravity and acceleration into
the theory of relativity. It is a well known fact that the inertial mass and gravitational mass
of an object are equal. The inertial mass is the mass that determines the reluctance of that
object to be moved by an external force, showing up in Newton’s second law

F “ ma,

whereas the gravitational mass is the mass governing gravitational interaction between bodies,
showing up in Newton’s law of gravity
m1 m2
F “ ,
r2
(note the use of geometrised units). A priori, there is no reason these two masses should be
equal. Yet it is an empirical fact that they are. This statement, that the inertial mass and
gravitational mass of an object are equal, is known as the weak equivalence principle.
GR improves on this by introducing the strong equivalence principle. It has several
different but equivalent formulations. This one is from Misner, Thorne & Wheeler [9]:
[In] any and every local Lorentz frame, anywhere and anytime in the Universe,
all the (nongravitational) laws of physics must take on their familiar special-
relativistic form.
A local Lorentz frame is a frame of reference where gravity can be treated as absent. In
SR, a Lorentz frame is a non-accelerating frame, an SR-compatible generalisation of the
Galilean frames of classical mechanics. In GR, global Lorentz frames do not exist because
of the curvature of spacetime. In a large enough region, different parts of that region will
be curved in different directions by gravity, causing large-scale gravitational effects, known
as tidal forces. However, the principle of equivalence states that local Lorentz frames, also
known as locally flat frames, still exist. In such a frame, objects are in free fall, and the region
under consideration is small enough so that all objects within fall in the same direction. The
principle of equivalence states that around any point in spacetime, a local Lorentz frame can
be chosen, and the laws of physics in such a frame are precisely the same as in any other
such frame.
Physically, while the weak equivalence principle states that inertial mass and gravitational
mass are equivalent, the strong equivalence principle states that acceleration and gravity are
equivalent, given that the considered region is small enough. A non-accelerating frame with
gravity is exactly the same as a frame without gravity but with acceleration. Standing on

5
the floor of an accelerating spaceship and being pushed toward the floor by the acceleration
is indistinguishable from standing on the surface of a massive object such as the Earth and
being pulled toward the ground by gravity. The strong equivalence principle states that there
is no experiment whatsoever that may distinguish these two “different” situations. In both
cases, the only real force felt is the contact force from the spaceship or the Earth respectively
pushing you from your natural free falling state.

2.4 Curved spacetime as Riemannian manifolds


Mathematically, spacetime is described by what is known as a smooth manifold. Essentially,
this can be viewed as a warped version of Rn which looks locally like Euclidean space. The
basic mathematical techniques that go into working with such a manifold in GR are described
in this section.
In the study of differential geometry and general relativity, heavy use is made of tensors,
a generalization of scalars, vectors and linear operators. The perhaps most important tensor
in GR is the so called metric tensor. It serves as a form of “ruler” which assigns distances to
pairs of neighbouring points on the manifold. It is absolutely essential in general relativity
as it carries information about spatial distances and the rate of time’s passing. A smooth
manifold equipped with a metric tensor field is called Riemannian.
More specifically, the metric gµν tensor defines the inner product of tangent vectors to
the manifold, and thereby allows for the local measurement of lengths. That is, if Eµ are the
tangent vectors at a point on the manifold in some choice of coordinates, then
gµν “: Eµ ¨ Eν . (2.4)
Note the direction of definition; the metric tensor defines the inner product - not the other
way around. The inner product of two general vectors is then
x ¨ y “ xµ Eµ ¨ y ν Eν “ xµ y ν pEµ ¨ Eν q “ xµ y ν gµν . (2.5)
Importantly, the metric depends both on the manifold itself and the choice of coordinates
used to describe it. In three-dimensional Euclidean space R3 - which is itself a manifold -
the metric tensor is given in Cartesian coordinates as gij “ δij . In relativity, spacetime is
described as a four -dimensional manifold, where the extra coordinate represents time. The
convention used in this thesis is to denote the time coordinate by t or x0 . In SR, where
gravity (which is spacetime curvature) is not accounted for, the metric is
$
&´1, µ “ ν “ 0;

gµν “ ηµν :“ 1, µ “ ν ‰ 0; (2.6)

0, µ ‰ ν,
%

in so called Minkowski coordinates, essentially ordinary Cartesian coordinates x1 , x2 and x3


together with a time coordinate x0 . Of course, the metric may have other matrix represent-
ations if other coordinate systems are chosen. The point here is that given the proper choice
of coordinate system, a matrix representation of this form exists.

6
An important property of the metric is that it allows us to raise and lower indices of
tensors. A tensor is either covariant (component index down) or contravariant (component
index up) in each of its indices. The nature of the tensor (contravariant or covariant) then
determines how that tensor transforms under coordinate changes. In general spaces there
is no intrinsic way to relate contravariant tensors to covariant ones. Introducing a metric
however allows us to do just this via the definitions

σµ :“ gµν σ ν , λµ :“ g µν λν , (2.7)

where σ ν and λν are arbitrary contravariant and covariant vectors respectively (if the tensors
have more indices, more factors of the metric are used). In the above definition we have
introduced the inverse metric g µν , defined to fulfill the relation

g µα gνα “ δ µν . (2.8)

We can therefore move unhindered between contravariant and covariant tensors as long as
the metric is known. Note that this means that if x “ xµ Eµ and y “ y ν Eν are two arbitrary
vectors and the metric is given by gµν “ ηµν , then
3
ÿ
µ µ ν 0 0
x yµ “ x ηµν y “ ´x y ` xi y i , (2.9)
i“1

to be compared with the general expression in Eq. (2.5). The placement of indices becomes
important when gµν ‰ δµν .
While we are open to the idea that space and time are not absolute in SR in the way they
are in classical mechanics, we still insist that inner products between vectors, being scalars,
are independent of the choice of reference frame. This means that the quantity

x2 “ xµ Eµ ¨ xν Eν “ xµ xν ηµν “ ´t2 ` x2 ` y 2 ` z 2 , (2.10)

must be invariant under Lorentz transformations.


In an intrinsically curved (non-affine) space things get more complicated, as vectors at
different points cannot be compared in a canonical way. As the manifold is smooth, we do
however expect it to be “almost flat” locally, so for infinitesimal separations ds the above
discussion should still be valid. The infinitesimal quantity

ds2 “ gµν dxµ dxν , (2.11)

called the invariant length element, should then be independent of our choice of coordinates.
Finally, we need some mathematical object that characterizes the curvature of the man-
ifold. We will not derive this since it involves an introduction to covariant differentiation,
Christoffel symbols, and parallel transport; but we can write out an expression for it by using
just the metric and its derivatives [10]. The Riemann tensor can be defined in a locally flat
coordinate system as
1
Rαβµν “ pgαν,βµ ´ gαµ,βν ` gβµ,αν ´ gβν,αµ q , (2.12)
2

7
where the comma stands for differentiation with the respect to the variables to the right
of it. In a non-flat reference frame the Riemann tensor will have extra terms involving
Christoffel symbols. These however vanish in a local Lorentz frame, which can always be
chosen according to the equivalence principle. The Riemann tensor contains all information
about the curvature of the manifold at every point, and vanishes everywhere if spacetime is
flat. From the Riemann tensor we can obtain the Ricci tensor by contracting the first and
third indices
Rβν “ g αµ Rαβµν , (2.13)
and the Ricci scalar by contracting the Ricci tensor

R “ g βν Rβν . (2.14)

2.5 The Einstein field equations


Einstein’s field equations can be motivated in multiple ways. One way (see for example
Schutz [10]) is to require them to generalise Newton’s equation

∇2 φ “ 4πρ, (2.15)

where ρ is the mass density, make no reference to a “special” coordinate system, and locally
conserve energy-momentum independently of the metric. Another way is to minimise the
Einstein-Hilbert action [11], which is expressed as
ż a
1
S“ R |g| d4 x (2.16)
16π
in geometrical units, where R is the Ricci scalar and g is the determinant of the metric tensor
in the given coordinates. However, the Einstein-Hilbert action cannot be derived from more
fundamental principles, but is more of an educated guess based on the expected form of the
Einstein equations and the possible variables involved. Thus, the Einstein equations can be
seen as fundamental, just as Newton’s equation for gravity is to classical physics.
No matter which approach is used, the Einstein equations fall out in component form as

Gµν “ 8πT µν (2.17)

in geometrised units, where the Einstein tensor Gµν is defined as


1
Gµν “ Rµν ´ g µν R. (2.18)
2
The tensor T µν in Eq. (2.17) is the stress-energy tensor, carrying the information about
energy, momentum, and stress as a tensor field. This is a particularly interesting quantity
for us since it acts as the source of gravity which is the reason for curved spacetime.

8
3. Investigation

We study the power radiated from a SM binary system in the form of gravitational waves,
with focus on the various Sun-planet systems in the solar system. We do this by first deriving
the wave equation from the Einstein field equations. Then we take a look at the generation of
gravitational waves from masses in motion. With this in mind we can study some examples in
the solar system by using Kepler’s laws of planetary motion, a non-relativistic approximation
valid for SM systems. We study the influence of different parameters such as the masses of
the rotating bodies and their separation. We also consider the eccentricity of the orbits as a
parameter. Finally we look at the total energy of a binary system and calculate the orbital
decay due to energy lost by gravitational waves.

3.1 Gravitational waves


If the gravitational fields involved in the problem at hand are weak, the corresponding
curvature of spacetime will be small, and we can linearise the Einstein equations around
the Minkowski vacuum solution gµν “ ηµν . By this we mean that we can regard the metric
gµν as “almost flat”, in the sense that its components can be expressed as

gµν “ ηµν ` hµν , |hµν | ! 1, (3.1)

where ηµν is the Minkowski metric of special relativity and hµν is a small perturbation. If we
perform a Lorentz transformation on Eq. (3.1) we find that

gµ1 ν 1 “ Λµµ1 Λν ν 1 gµν “ Λµµ1 Λν ν 1 ηµν ` Λµµ1 Λν ν 1 hνµ “ ηµ1 ν 1 ` Λµµ1 Λν ν 1 hνµ , (3.2)

i.e., hµν transforms just as a tensor in special relativity, where spacetime is flat. Hence, we
can view the effect of gravity as a tensor defined on Minkowski space. We therefore raise
and lower indices with the Minkowski metric ηµν and treat the perturbation as a field in
spacetime, rather than as being a part of spacetime itself.
Working in Minkowski spacetime allows us to make gauge transformations in order to
simplify the calculations. We can write a very small shift of the coordinates as∗
1
xµ “ xµ ` χµ pxq. (3.3)

Please note that Eq. (3.3) is not a tensor equation, but is a relation of the transformed and pre-transformed
coordinates.

9
where we demand that |χµ ,ν | ! 1. The Lorentz transformation matrix components are now
obtained through differentiation as
1
1 Bxµ Bxµ
Λµ α1 “
` µ 2˘
Λµ α “ “ δ µα ` χµ,α , 1 “ δ µ
α ´ χµ
,α ` O |χ ,α | . (3.4)
Bxα Bxα
We see that the metric tensor transforms as

gµ1 ν 1 “ ηµν ` hµν ´ χµ,ν ´ χν,µ . (3.5)

Clearly, our small fictional gravity tensor transforms as

hµν Ñ hµν ´ χµ,ν ´ χν,µ , (3.6)

which looks a lot like the well known gauge transformation of Maxwell’s electromagnetic
4-potential Aµ Ñ Aµ ´ Bµ χ from special relativity. Since we demand that |χµ ,ν | is small, the
new hµν will be small, and this means that condition in Eq. (3.1) will still be satisfied in the
new coordinate system.
An immediate implication of the construction of the new linearised metric tensor is that
we can express the linearised Riemann tensor (2.12) as
1
Rαβµν “ phαν,βµ ´ hαµ,βν ` hβµ,αν ´ hβν,αµ q , (3.7)
2
the Ricci tensor Eq. (2.13) as
1` α ˘
Rµν “ η αβ Rαµβν “ h µ,αν ` hαν,µα ´ lhµν ´ h,µν , (3.8)
2
and finally the Ricci scalar in Eq. (2.14) as

R “ hµν ,µν ´ lh, (3.9)

where the d’Alambertian l, also known as the wave operator, is the differential operator
B
l :“ B µ Bµ “ ´ ` ∇2 , (3.10)
Bt
and
h :“ hαα
is the trace of hµν .
We are now set to use these linearised quantities to get an expression for the Einstein
tensor Gµν . To simplify this expression we use the Lorentz gauge condition, requiring that
the equality
h̄µν ,µ “ 0 (3.11)
holds. Here h̄µν is the trace reverse of hµν defined as
1
h̄µν :“ hµν ´ ηµν h.
2

10
Equation (3.11) holds if we chose our transformation to be

h̄1 µν Ñ h̄µν ´ χν,µ ´ χµ,ν ` ηµν B λ χλ ,

which simplifies the Ricci tensor and Ricci scalar to


1 1
Rµν “ ´ lhµν , R “ ´ lh. (3.12)
2 2
This reduces the Einstein field equations to

lh̄µν “ ´16πT µν . (3.13)

But this is nothing but the usual three-dimensional inhomogenous wave equation for each
component of h̄µν . In the vacuum of space, the stress-energy tensor T µν vanishes, so Eq. (3.13)
reduces to
lh̄µν “ 0. (3.14)
Equations (3.13) and (3.14) can be solved in exactly the same manner as the analogous
problem from electromagnetism, for example using Green’s functions. The Green’s function
for the d’Alambertian is the (possibly generalised) function f defined to fulfill

lf pxσ ´ y σ q “ δ p4q pxσ ´ y σ q, (3.15)

where δ is the Dirac delta function. If the function f is known, the general solution to
Eq. (3.13) can be written as
ż
h̄µν px q “ ´16π f pxσ ´ y σ qTµν py σ qd4 y.
σ
(3.16)

Since the waves left the source possibly long time ago, we are therefore interested in the
retarded Green’s function, known from literature [12] to be
1
f pxσ ´ y σ q “ δp|x ´ y| ´ px0 ´ y 0 qqHpx0 ´ y 0 q, (3.17)
4π|x ´ y|

where the boldface vector denotes the vector of spatial coordinates xi only and Hpx0 ´ y 0 q is
the Heaviside step function. Plugging this choice of Green’s function into our general solution
in Eq. (3.16) leaves us with
ż
1
h̄µν pt, xq “ 4 Tµν pt ´ |x ´ y|, yqd3 y, (3.18)
|x ´ y|

where, just like in Maxwell’s theory, the quantity t ´ |x ´ y| is referred to as the “retarded
time”. Thus, the solutions to Eqs. (3.13) and (3.14) will describe waves or “ripples” in
spacetime, propagating at the speed of light. This is the mathematical motivation behind
the phenomenon of the gravitational wave theory.

11
3.2 Generation of gravitational waves
In this section we discuss how gravitational waves are emitted in the first place. The dis-
cussion mirrors the one found in Misner, Thorne & Wheeler [9]. In order to estimate how
much gravitational radiation a SM binary system emits we can use the similarities between
the linearised theory of gravity and Maxwell’s theory of electromagnetism, known as grav-
itoelectromagnetism, to perform a multipole expansion of the stress-energy tensor T µν and
identify the leading term.
In electromagnetism the electric dipole term is the leading one. The power output, or
luminosity (denoted L), from an electric-dipole configuration is in geometrical units
2
Led “ p:, (3.19)
3
where p: is the second time derivative of the dipole moment p [9]. The gravitational analogue
of the electric dipole moment is the mass dipole moment d given by
ÿ
d“ mi xi . (3.20)
i

But no gravitational radiation can then be generated from a mass dipole. For
ÿ
d9 “ mi x9 i “ p
i

is just the total momentum of the system (not to be confused with the electric dipole moment
denoted with the same symbol), so the second derivative d : must vanish as momentum is a
conserved quantity.
The second strongest source of electromagnetic radiation is the magnetic dipole. Magnetic
dipole radiation is generated by the second derivative of the magnetic moment [3]. The
gravitational analogue to the magnetic dipole moment is the angular momentum
ÿ
J“ xi ˆ pmi vi q,
i

which like momentum is a constant of motion. Thus there can be no radiation from this term
either. We therefore see that there can be no gravitational radiation whatsoever from dipole
sources.
Next in the multipole expansion are the quadrupole terms. The luminosity of an electro-
magnetic quadrupole source is
1 ; ;
Leq “ Q jk Qjk ,
20
where
ÿ ˆ 1
˙
2
Qjk “ qi xij xik ´ δjk xi
i
3

12
y

r1
θ
x
r2

Figure 3.1: A binary system consisting of point masses m and M , described in polar coordin-
ates.

is the electromagnetic quadrupole moment tensor. The gravitational analogue to this is the
mass quadrupole moment tensor defined as
ÿ ˆ 1
˙ ż ˆ
1
˙
I jk “ mi xij xik ´ δjk xi “ ρ xj xk ´ δjk x d3 x,
2 2
(3.21)
i
3 3
and we can use this to obtain the the power output, or luminosity, as
1 2 1
Lmq “ x;I y “ x; I jk ;
I jk y. (3.22)
5 5
This is the first nonzero term, which will dominate in the SM approximation, and we will use
this later on in our calculations for binary systems. Notice that this expression is given in
geometrical units; if we want to compare our result with the order of magnitudes in everyday
life it would be neat to be able to convert our result to SI units. We can use the expression
c5
LpSIq “ Lpgeometrisedq (3.23)
G
to convert between these units [10].

3.3 Mathematical model of the binary system


Consider the following model of a general binary system. Let two masses M and m be in
elliptical orbit about one another in the xy-plane, see figure 3.1. The reduced mass of the
binary system is
Mm
µ“ . (3.24)
M `m

13
If we use polar coordinates with the origin placed in the center of mass, the positions of the
masses are pr1 ptq, θptqq, pr2 ptq, θ ` πq, both depending on time. The total separation is then
dptq “ r1 ptq ` r2 ptq, and by the definition of center of mass it must hold that mr1 “ M r2 .
Solving for r1 and r2 gives
M m
r1 “ d, r2 “ d. (3.25)
M `m M `m
We can now use Newton’s extension of Kepler’s laws of planetary motions, which gives us
ap1 ´ ε2 q
dptq “ , (3.26)
1 ` ε cos θptq
where ε is the eccentricity of the elliptical path, a is the semi-major axis of the ellipse, and θ
is defined as in figure 3.1. Kepler’s laws also give us the expression for angular velocity ω as
a
apM ` mqp1 ´ ε2 q2
ω “ θ9 “ . (3.27)
dptq2
Equation (3.21) allows us to calculate the quadrupole tensor for the system as
¨ 2 ˛
µd pcos2 θ ´ 13 q 12 µd2 sinp2θq 0
I “ ˝ 12 µd2 sinp2θq µd2 psin2 θ ´ 31 q 0 ‚. (3.28)
1
0 0 ´ 3 µd2
It is in principle straightforward to calculate the derivatives and use Eq. (3.22) to get the
power radiated by the system. When we are interested in the order of magnitude of the
radiation in SI units we can use Eq. (3.23).

3.4 Analytical calculations


3.4.1 Energy loss for circular orbits
Calculating the derivatives of Eq. (3.28) can be rather tedious since both d and θ are time
dependent. The calculation, as well as the result, simplify drastically if one instead considers
circular orbits, where d “ a and the angular velocity ω is constant. If we calculate the third
time derivative of Eq. (3.28) with these conditions we obtain
¨ ˛
4 sinp2θq 4 cosp2θq 0
;I “ µd2 ω 3 ˝4 cosp2θq ´4 sinp2θq 0‚ (3.29)
0 0 0
in matrix representation, where ω is given as Eq. (3.27) and with d “ a kept constant.
We can now insert this result in Eq. (3.22), together with Eq. (3.24) for the reduced mass
of the system in order to calculate the gravitational radiation luminosity L. This falls out as
B F
dE 32m2 M 2 pm ` M q
L“ “ , (3.30)
dt 5d5

14
where x¨y denotes the average taken over one period. In SI-units we get

32c5 m2 M 2 pm ` M q
L“ . (3.31)
5Gd5

3.4.2 Correction for elliptical orbits


As stated in the previous section, taking the time derivatives of the components of the
quadrupole tensor for a general elliptical orbit can be quite tedious. This was however done
once and for all in 1963 by P. C. Peters and J. Matthews [13]. Taking the time derivatives of
Eq. (3.28), averaging over one period and plugging it into Eq. (3.22) we then, for a general
SM binary system, obtain the luminosity as
ˆ ˙
32 m2 M 2 pm ` M q 73 2 37 4
Lelliptical “ Lcircular f pεq “ 1` ε ` ε , (3.32)
5 a5 p1 ´ ε2 q7{2 24 96

where a is the semi-major axis and


ˆ ˙
1 73 37
f pεq “ 1 ` ε2 ` ε4 (3.33)
p1 ´ ε2 q7{2 24 96

is a correction factor depending only on the eccentricity of the orbital motion. We see in
figure 3.2 that this factor can be quite significant when the eccentricity is large. For most
planetary systems in the solar system however, the orbits are almost circular and f pεq « 1.
Again, if we want to express the luminosity in SI-units, we can use the conversion factor
Eq. (3.23).

3.4.3 Orbital decay due to energy loss


The energy loss discussed in the previous sections will cause the bodies to slowly but surely
start spiraling into each other. We here quantify this effect for the special case of circular
orbit. The total Newtonian energy of a binary system in circular orbit is
1 Mm
E “ µω 2 d2 ´ . (3.34)
2 d
Here the angular frequency ω is given by Eq. (3.27) with a “ d and ε “ 0. Plugging this into
Eq. (3.34) leaves us with
mM
E“´ . (3.35)
2d
We now use Kepler’s third law, which relates the distance d and masses M and m to the
orbital period T as c
d3 2π
T “ 2π , T “ . (3.36)
M `m ω

15
10 3

10 2

1
10

10 0
0 0.2 0.4 0.6 0.8 1

Figure 3.2: The correction factor as a function of eccentricity. Note that the y-axis is logar-
ithmic. The correction factor is close to unity for small eccentricities, but grows rapidly and
without bound as ε Ñ 1.

This enables us to eliminate the distance d in favour of T in Eq. (3.35). Using these relations
in Eq. (3.35) yields
ˆ ˙1{3
mM 4π 2
E“´ .
2 T 2 pM ` mq
If we now differentiate this expression with respect to time we obtain
ˆ ˙1{3
dE mM 4π 2
“ T9 . (3.37)
dt 3 T 5 pM ` mq
Since dE{dt “ L by definition, we can equate the above expression with Eq. (3.30) - keeping
in mind that Eq. (3.30) gives the energy radiated away from the system in order to get the
sign right - where we can plug in the relation in Eq. (3.36) for d, and solve for T9 . After some
simplification the expression obtained is
ˆ 2 ˙1{3
9 384π 2 mM 4π
T “´ 1{3
. (3.38)
5 pm ` M q T5
Note that the above expressions are all in geometrised units.
We will now find the explicit time dependence of T . For notational brevity, let us introduce
the time-independent variable
384π 2 mM ` 2 ˘1{3
k“ 1{3
4π . (3.39)
5 pm ` M q
We can then write Eq. (3.38) as
dT
“ ´kT ´5{3 . (3.40)
dt

16
Equation (3.40) is a separable first order ODE for T . Separating variables and integrating† ,
we obtain ż ż
5{3
T dT “ ´ k dt.

Carrying out the integration and solving for T we now have


ˆ ˙3{8
8
T ptq “ ´ kt ` C , (3.41)
3

where C is a constant of integration. Setting t “ 0 to be the present day (or generally any
given time), we find that
8{3
C “ T0 , (3.42)
where T0 is the current period of the system.
Plugging this expression for T into Eq. (3.36) gives us an expression for the separation d
as a function of time:
ˆ ˙1{3 ˆ ˙1{4
M `m 8 8{3
dptq “ ´ kt ` T0 .
4π 2 3
Reinserting the definition of k and simplifying gives
˜ ˆ ˙4{3 ¸1{4
256 m`M 8{3
dptq “ ´ mM pm ` M qt ` 2
T0 . (3.43)
5 4π

In particular, the bodies will crash into each other when d “ R1 ` R2 , where R1 and R2 are
the radii of the respective bodies. However, since d0 " R1 ` R2 for most SM binary systems
(where d0 is the current distance between the bodies), to good approximation R1 ` R2 « 0.
Then, solving for t we see that the time t1 of the collision is given by

5pM ` mq1{3 8{3


t1 “ T . (3.44)
256p4π 2 q4{3 mM 0

We plot the shape of dptq in figure 3.3. The bodies will spiral into one another more
and more quickly as time progresses. The time scales involved are however quite large, see
Section 3.6.

3.5 Results
In order to put Eq. (3.32) into perspective, let us look at what it says about the planets in our
own solar system. For this discussion, we assume that we can treat all Sun-planet systems
as binary - meaning that corrections to the orbits of the planets due to other planets and

Note that this differential equation would not be this easy to solve if the orbits were not circular. The
eccentricity ε has a long term time dependence.

17
1

0.8

0.6
d/d0

0.4

0.2

0
0 0.2 0.4 0.6 0.8 1
t/t1

Figure 3.3: The time evolution of the distance d between the bodies. Note that the time t1
will be enormous for most cases.
Mean distance Semimajor
Mass Mass
from Sun axis Eccentricity
(1024 kg) (m)
(106 km) (106 km)
Mercury 0.330 0.245 ¨ 10´3 57.9 57.91 0.205
Venus 4.87 3.60 ¨ 10´3 108.2 108.21 0.007
Earth 5.97 4.40 ¨ 10´3 149.6 149.60 0.017
Mars 0.642 0.477 ¨ 10´3 227.9 227.92 0.094
Jupiter 1898 1.41 778.6 778.57 0.049
Saturn 568 0.422 1433.5 1433 0.057
Uranus 86.8 0.0645 2872.5 2872 0.046
Neptune 102 0.0757 4495.1 4495 0.011

Table 3.1: Observed data from NASA for all the planets in our solar system. Note that the
values for the masses in the second column are given in geometrised units.

moons are ignored. These effects would of course in reality be much larger than the impact
of gravitational radiation on the orbits.
The motions of the planets in our solar system are very well studied. Since they are all
matching our criteria of SM binary system, we can use our model with data from NASA [14]
to calculate how much energy the different systems radiate in gravitational waves. To get the
mass in geometrised units we multiply the SI-unit in kg by a factor of G{c2 [m/kg], where
G is the gravitational constant and c is the speed of light. The observed values for all the
planets in our solar system are given in Table 3.1.
We can now use these values to calculate the luminosity of all the Sun-planet systems in
our solar system. Since the eccentricity of all the orbits is close to zero we can start with
our approximation for pure circular orbits in Eq. (3.30) and then see how big the elliptical
correction of Eq. (3.32) is compared to this. The calculated luminosity together with the

18
Lcircular (W) Lelliptical (W) Difference (W) f pεq
Mercury 69 90 22 1.312
Venus 659 659 0 1.0003
Earth 196 196 0 1.0019
Mars 0.28 0.29 0.01 1.0593
Jupiter 5189 5271 83 1.0158
Saturn 22 22 0 1.0215
Uranus 0.016 0.016 0 1.0139
Neptune 0.002 0.002 0 1.0008

Table 3.2: Power radiated from the different Sun-planet systems in our solar system, and the
eccentricity correction factor f pεq.

elliptical correction for these systems is presented in SI-units in Table 3.5. We can see
that Jupiter is clearly the most radiating, but the inner three Sun-planet systems radiate
much more than the outer three gas giants do. Furthermore we see that the correction for
eccentricity is quite small for all planets except Mercury, which is to be expected.
To gain perspective on the orbital decay, we look at the rate at which Jupiter’s (being the
most radiating planet) orbit changes as a consequence of emitting gravitational radiation.
Plugging in numerical values in Eq. (3.38) and converting to seconds per year, we obtain

T9 “ ´5.70 ¨ 10´16 s/year. (3.45)

By Eq. (3.44) this corresponds to a time t1 until collision with the Sun of

t1 “ 3.61 ¨ 1023 years, (3.46)

Or roughly 26 trillion times the age of the Universe.

3.6 Discussion
3.6.1 A new spectrum to study
Almost every observation ever made of our universe has been possible due to the fact that
we have been able to detect electromagnetic waves. In the beginning of the 20th century
astronomers were restricted to observing the visible light region of the electromagnetic spec-
trum only. As technology evolved this region expanded and today the whole electromagnetic
spectrum has been explored to some degree. Because of this we have been able to detect
many astrophysical phenomena such as quasars, pulsars, supermassive black holes at galactic
centers, and the cosmic background radiation.
The gravitational wave spectrum, on the other hand, has been left almost completely
untouched. The first ever direct detection of gravitational waves was made by LIGO in 2015,
and there more to come for the future, such as the space based LISA detector [15]. It is

19
in fact very reasonable to expect new discoveries to be made in the future using gravita-
tional waves considering the structure of our universe. Only 4 % of the total mass-energy in
the Universe is contained in charged particles and thus interacts via electromagnetic waves,
whereas the remaining 96 % can only be detected through their gravitational interaction
and thus cannot be studied by traditional means. This remaining mass-energy could feasibly
be studied using gravitational waves [10]. In addition, electromagnetic waves are scattered
on their passage through the universe countless of times by interfering objects before even
reaching Earth, while gravitational waves are uninhibited by such obstacles. The weakness
of the gravitational interaction, captured in the very small value of Newton’s constant G,
of course makes such studies difficult, but with experimental technology constantly evolving
and precision measurements becoming better and better there is still reason to be hopeful
about the future.

3.6.2 Slow motion binaries


The power radiated by gravitational waves in binary systems is found in this thesis to be
extremely small in comparison to the kinetic energy of these systems. Jupiter, radiating by
far the most, still only outputs a power on the same order of magnitude as a regular household
oven or refrigerator. Saturn hardly even outputs enough to power a conventional light bulb.
The effect of gravitational waves on the planets of the solar system and equivalent systems
are, as expected, vanishingly unimportant.
For all planets except Mercury the correction for eccentricity was almost or completely
negligible. However, the eccentricity correction factor f pεq grows quickly and without bound
as ε Ñ 1. We can therefore expect an SM binary system in strongly elliptical orbit to lose a
much larger amount of energy through emission of gravitational waves. Eventually however,
given that Kepler’s second law says that a line segment joining a planet and the Sun sweeps
out equal areas during equal intervals of time, we can expect a binary in extremely elliptic
orbit to become too fast on closest passage for the SM approximation to hold. We do not
treat this limit in this thesis.

20
4. Summary and Conclusions

We have motivated the occurrence of gravitational waves by linearising the Einstein equations
around the Minkowski vacuum solution and showing that this results in the field equations
reducing to the three-dimensional wave equation for the metric perturbation.
We saw that this linearised theory of general relativity has close analogues to Maxwell’s
electromagnetic field theory, and we could use this fact to gain a better understanding for
gravitational waves. While Einstein’s theory is needed to describe the exact motions of
celestial bodies, we could use Kepler’s laws of motion as a good approximation for planets
moving with velocities much less than the speed of light in weak gravitational potentials.
We have derived an expression for the power (i.e., luminosity) radiated in the form of
gravitational waves by a slow motion binary system, and applied this expression to the
planets of our solar system. We found this energy loss to be exceedingly small in comparison
to the total energy of the systems for all the Sun-planet systems
Last of all we studied the orbital decay due to the energy loss. This was, not surprisingly,
found to be an extremely slow process. This of course agrees quite well with our current view
of a very stable and static solar system.
In conclusion, whereas gravitational waves are interesting from a theoretical perspective,
it is unlikely that waves coming from less exotic events than the merging of neutron stars or
black holes will ever have any major effects.

21
Bibliography

[1] A. Einstein, Fundamental Ideas of the General Theory of Relativity and the Applica-
tion of this Theory in Astronomy, Preussische Akademie der Wissenschaften, Sitzungs-
berichte, 1915 Part 1, 315

[2] T.P. Cheng, Relativity, Gravitation and Cosmology, Oxford (2010).

[3] D.J. Griffiths, Introduction to Electrodynamics, Pearson (2014)

[4] A. Einstein, Approximative Integration of the Field Equations of Gravitation, Preuss-


ische Akademie der Wissenschaften, Sitzungsberichte, (1916): 688-696

[5] A. Einstein, On Gravitational Waves, Preussische Akademie der Wissenschaften,


Sitzungsberichte, 1918 Part 1: 154-167

[6] R. A. Hulse, J. H. Taylor, Discovery of a pulsar in a binary system, Astrophysical


Journal, vol. 195, 51-53, (1975)

[7] Joel M. Weisberg, Joseph H. Taylor & Lee A. Fowler, Gravitational Waves from an
Orbiting Pulsar, Scientific American 245, 74 - 82 (1981)

[8] LIGO and Virgo Collaborations, B.P. Abbott et al., GW150914: First results from the
search for binary black hole coalescence with Advanced LIGO, Phys. Rev. D 93, 122003
(2016)

[9] C.W. Misner, K.S. Thorne, J.A. Wheeler, Gravitation, W.H. Freeman and Company
(1973)

[10] B. Schutz, A First Course In General Relativity, Cambridge (2009).

[11] S. Weinberg, Gravitation and Cosmology: Theory and Applications of the General The-
ory of Relativity, John Wiley & sons (1972)

[12] S. Carrol, Spacetime and Geometry: An introduction to General Relativity, Pearson


(2004)

[13] P.C. Peters, J. Matthews Gravitational Radiation from Point Masses in a Keplerian
Orbit, Physical Review, vol 131, Jul. 1, 1963

22
[14] NASA (2016), Planetary Fact Sheet, [online] Available at: ht-
tps://nssdc.gsfc.nasa.gov/planetary/factsheet/index.html [Accessed 30 Mars 2017]

[15] LISA (2017), A proposal in response to the ESA call for L3 mission concepts, [online]
Available at: https://www.elisascience.org/files/publications/LISA_L3_20170120.pdf
[Accessed 17 May 2017]

23

You might also like