A background-free lattice gauge theory of quantum gravity

T. D. Andersen
(Dated: Received: date / Accepted: date)
Abstract
The search for a quantum theory of gravity has become one of the most well-known problems in
theoretical physics. Problems quantizing general relativity because it is not renormalizable have
led to a search for a new theory of gravity that, while still agreeing with measured observations, is
renormalizable. In this paper, I show that, given a vortex model of elementary particles in which
rest mass derives from intrinsic spin and polarization, a Yang-Mills force with a SO(4,1) group
symmetry predicts post-Newtonian N-body motion such as solar system observations of gravita-
tional behavior as well as binary pulsar precession and orbital speed-up caused by gravitational
radiation-reaction. Yang-Mills theory of the SO(4,1) group is dened on a lattice graph such that
a background manifold is not requireda property unique to the (anti-) de Sitter and Poincare
groups. It predicts the accelerating expansion of the universe as a consequence of the de Sitter group
Lie algebraan acceleration that I show does not occur in the Poincare group approximation
suggesting that the de Sitter group symmetry explains dark energy. In addition, with quantized
mass, because it is a generic massless, semi-simple Yang-Mills theory, it is mathematically proved
to be a perturbatively renormalizable quantum theory.
PACS numbers:
1
I. INTRODUCTION
That general relativity is able to subsume Newtonian gravity and explain phenomena
that do not t into the Newtonian framework such as Mercurys perihelion precession, light
deection, and gravitational red-shift as well as being compatible with special relativity has
brought it wide acceptance. In recent years, however, as measurement tools have grown more
accurate and new observations made, the necessity of introducing either tunable parameters
or exotic forms of matter and energy to make general relativity t those measurements has
brought the theory into question. Observations of galactic rotational curves and gravita-
tional lensing [39] [27] [4] have demanded the introduction of dark matter and accelerating
expansion of the universe [21] dark energy. At present there is no consensus on what these
substances are.
Besides the problems with general relativity at the macroscopic scale, achieving a quan-
tum theory of gravity has become one of the most signicant unsolved problems in physics.
Attempts to place the Einstein-Hilbert action,
S
EH
=
_
d
4
x

gR, (1)
where R is the Ricci scalar, into a functional path integral ensemble, have all failed, giving
nonsense results. The immediate source of the diculty is that the theory is not renor-
malizable creating an innite number of counterterms in perturbation expansions. Another,
less theoretically troubling but mathematically dicult issue is that the expansion of the
Lagrangian fails to terminate because of the volume element

g and the inverse metric
g

. Therefore, unlike the actions of other forces, the gravitational action is not nite poly-
nomial. A disaster also occurs quantizing the weak force but disappears when it is unied
with the electromagnetic force, motivating the quest to unify gravity with the other forces
in the hope that it will become renormalizable [38].
Although elegant and surprisingly accurate until recent decades, because the dieren-
tiable manifold assumption about spacetime causes such deep, unresolvable problems at the
quantum level, there is some motivation to nd an alternative to dieomorphism as the
source of gravity. Despite its current acceptance, dieomorphism was a counterintuitive
choice for a theory of gravity given the success of electromagnetism in the 19th century. As
Wald points out [31],
2
Maxwells theory is a remarkably successful theory of electricity, magnetism, and
light which is beautifully incorporated into the framework of special relativity.
Therefore, one might expect that the next logical step would have been to develop
a new theory of the other classical force, gravitation, which would generalize
Newtons theory and make it compatible with special relativity in the same way
that Maxwells theory generalized Coulombs electrostatics. However, Einstein
chose an entirely dierent path[.]
Because dieomorphic symmetry has never been precisely measured, the path not taken still
lies open.
The precision to which general relativity has been conrmed is two-fold: (1) if general
relativitys eld equations were as simple as Maxwells equations or not very much more
complicated, then it could be considered a well-conrmed theory because its linearized ver-
sion has a great deal of evidence behind it, and (2) although observation has conrmed
many eects, many of the highly non-linear portions of the theory have been unobservable
in any precise way because of the weakness of gravity. All precision experiments (classic
tests) of general relativity have been done within the solar system (with the exception of
binary pulsar precession) where spherically symmetric, weak gravity prevails and only the
rst order post-Newtonian Einstein-Infeld-Homan (EIH) equations have been conrmed
[19] [31]. An example of the best recent evidence for strong eld general relativity is the
periastron precession of the double pulsar system PSR J0737-3039A/B [13], which is a rst
post-Newtonian order conrmation. Furthermore, recent measurements of the cosmic mi-
crowave background, where spacetime curvature is most likely to appear, show none at all
[12] with the inationary theory as a potential explanation [32].
The dierence between the EIH equations and the full equations of GR has invited a large
number of alternative theories of gravity such as are discussed in [35], but virtually all are
metric theories and retain the dieomorphic spacetime picture with the associated problems
mentioned above. Non-metric theories, by contrast, oer the hope of eliminating gravitys
problems with quantization at the cost of also eliminating the elegant curved spacetime
approach. It has been suggested that the Standard Model approach, where forces have
Yang-Mills actions, may be more suitable for a quantum theory of gravity because of its
success at explaining the other forces. The tetrad or veirbein formulation of the Einstein-
Hilbert action, for example, has a structure similar to the Yang-Mills [38]. Conformal gravity
3
has a Yang-Mills structure as well but tends to diverge strongly from Newtonian gravity
over large length scales [25][18]. None of these theories have achieved signicant success in
making predictions that general relativity is not capable of making nor in resolving, entirely,
the problems with the quantum theory. Thus, there has been, as yet, no motivation for
replacing general relativity with any of these nor of accepting any as quantum gravitys true
representation.
In this paper, I present a Yang-Mills theory of gravity with a SO(4,1) or de Sitter group
without a background metric, where the Yang-Mills potential is itself responsible for space-
time curvature and determining geodesics. Thus, Einsteins curved spacetime approach is
retained without sacricing renormalizability at the cost of dieomorphism. This paper rep-
resents the rst time that a Yang-Mills theory based on this group has been shown to agree
with all precise observations of gravitational phenomena. Because a non-Abelian SO(4,1)
theory has local Lorentz symmetry, the theory complies with the Einstein Equivalence Prin-
ciple (EEP) and the Strong Equivalence Principle (SEP) (which even most alternate metric
theories do not satisfy [35]). All physical theories are Poincare symmetric, but, given that
the Poincare group is not semi-simple, the de Sitter group is a natural choice for a semi-
simple extension. The anti-de Sitter group would serve as well, and there is no compelling
evidence to choose one over the other. Thus, everything in this paper applies equally well
to that group.
Yang-Mills theories have a nite, polynomial action with a renormalizable quantization
provided the coupling constant has non-negative mass dimension. Since G has a mass-
dimension of 2, a non-dimensional gravitational constant, a
g
, is needed for quantum pre-
dictions. Because the coupling constant is non-dimensional, mass, energy, momentum, and
angular momentum are quantized, which has already been suggested in the context of other
theories of gravity [34]. Any quantum eld can be decomposed into nitely many modes
with a quantum of momentum between successive modes k,
=
M/2

n=M/2
a
n
e
ik

n
x
, (2)
which constrains the number of degrees of freedom of the eld to be a nite number, M.
The quantity M is simply a regularization (energy cuto) which is taken care of by renor-
malization, and, if the k
n
are closely spaced (k is small) then, we can take the continuum
4
approximation,
(x

)
_
M/2
M/2
dk

(2)
4

(k

)e
ik

x
. (3)
This is an approximation only however, similar to the thermodynamic limit approximation
of classical uids, taking the number of molecules to innity, and, at a fundamental level,
the number of degrees of freedom of the eld remains nite.
With a lattice method, I show that the background coordinate system normally present
in Yang-Mills theory can be replaced with an articial, bookkeeping coordinate system
where all distances, time intervals, and torsions in spacetime are derived from the Yang-Mills
potential. In this way, the Yang-Mills action is the same as the standard Yang-Mills action
for SU(N) theories, but its potential alone is the measurement of spacetime.
The second component to the theory is that rest mass is derived from the energy of
spin and polarization. In any frame, these vectors form an anti-symmetric tensor S

such
that polarization is p
i
= S
0i
and spin is s
i
=
ijk
S
jk
where summation is implied. Spin
and polarization have a conserved current (similar to the symmetric electromagnetic stress
energy tensor [9]),
T

=
_
S

+
1
4

_
, (4)
where

= diag(1, 1, 1, 1) is the Minkowski metric. This tensor is both symmetric and

traceless. This implies that mass derives relativistically from an intrinsic kinetic energy of
spin and oscillation.
With these denitions, this Yang-Mills theory, called vortex gravity because of the vortex
particle interpretation that spin and oscillation implies, predicts N-body observations such
as pertain to the Solar System and binary pulsars including radiation reaction, all with a
nite polynomial, positive denite action that is renormalizable.
The theory oers a cosmology slightly dierent from the CDM model. For example,
there is a distinction between Doppler and gravitational redshift, i.e. redshift caused by
matter moving within space and redshift caused by the expansion of space itself. For the
expansion of matter, a linear coasting cosmology is derived. The linear cosmology has been
shown to t observations such as Type-Ia supernovae and the Cosmic Microwave Background
related to the later universe well [2] [11]. The gravitational potential determines the degree
of gravitational redshift which gives the expansion of space itself. The combination of
the expansion of matter with the expansion of space leads to the following novel result:
5
while in the Poincare group approximation of the de Sitter theory, the linear coasting model
dominates the later universe with the gravitational redshift vanishing, in the de Sitter theory
the combination of the Doppler and gravitational redshift predicts a linear acceleration in
total observed redshift.
In general relativity, the de Sitter solution to the Einstein eld equations is an ad hoc
explanation for the acceleration of distant galaxies [32]. For the de Sitter Yang-Mills theory
it is anything but ad hoc. It derives automatically from the quantum theory and applies to
both the cosmological and quantum realms (as well as everything in between) and provides
a dynamical formula for the acceleration. Therefore, it has a compelling theoretical basis for
the accelerating expansion on par with vacuum energy but without the problem the vacuum
energy theory has with a discrepancy between the observed and predicted cosmological
constant of 120 orders of magnitude [38].
The de Sitter model for accelerating expansion in this paper is explained from a particle
as well as a geometric point of view by observing that, in the de Sitter Lie algebra, mo-
mentum produces spin by the coupling of momentum generators [V

, V

] = iM

where
V

are momentum generators and M

are spin generators. In the Poincare Lie algebra

spin can only be produced if spin is already present because [V

, V

] = 0. Because the de
Sitter theory produces spin from momentum without an initial source of spin, it causes the
expansion of the universe to produce gravitons that in turn couple to spin which in turn
create gravitons that couple back to momentum. The back and forth production of gravitons
leads to an accelerating redshift. This explanation is similar to the explanation for Thomas
precession where the orbital motion of an electron increases its spinan anomaly explained
in the early 20th century which can be shown from the Lorentz Lie algebra [9]. Hence, gravi-
tons produce dark energy because of the relationship between momentum and spin in a
de Sitter symmetry group. No cosmological constant is required, and no vacuum energy is
modelled. The accelerating expansion eect is produced directly from the Yang-Mills equa-
tions. Observations currently attributed to dark matter such as galactic rotational curves
and anomalous lensing are not addressed in this paper.
The paper is organized as follows: Sec. II A describes the conserved quantities of mass
and linear momentum, oscillation momentum, and angular momentum, how these are rep-
resented in the theory, and how they appear intrinsically through the vortex model; Sec.
II B derives the Yang-Mills theory with only a lattice and the gravitational potential and
6
no background manifold. This section also explains how the bookkeeping coordinate system
works; Sec. II C shows a derivation of the eld equations for the SO(4,1) Yang-Mills theory
and its formulation as two covariant eld equations; Sec. II E derives the equation for parti-
cle motion and shows that it is the same as the geodesic equation; Sec. II F gives a discussion
for how the theory satises the Strong Equivalence Principle (SEP) which GR also satises;
Sec. V discusses novel predictions made by the theory; Sec. II G briey mentions the mathe-
matical justication for a renormalizable quantization; Sec. III shows that the theory agrees
with measurements within the solar system including redshift, perihelion precession, time
dilation and light bending (III B), binary pulsar precession (III C), the quadrupole formula
for gravitational radiation as well as showing that there are no dipole/monopole moments
(III C1), and the equations for an expanding Robertson-Walker universe (III D); Sec. IV
explores any possible experimental disagreement with the theory; Sec. V discusses tests that
may conrm or deny the theorys predictions; and Sec. VI discusses related work. There
are also two Appendices.
II. THE THEORY
Although ad hoc gravitational theories exist, a good theory tends to be based on a guiding
principle. In developing this theory, the guiding principle is that gravity is an ordinary force
in the standard, SU(3)SU(2)U(1) model. This does not imply, as is frequently mentioned
in the literature, that gravity needs a background manifold to be dened as a non-metric
theory. It does not provided one accept a lattice graph construction to spacetime. (A lattice
graph, unlike a lattice imposed on a manifold, does not break Lorentz covariance.) The
assumptions of the lattice graph are equivalent in a discrete sense to Einsteins dierential
manifold and point-coincidence assumptions [22]. They both assume that events take place
at points and that points have neighboring points. Indeed, the lattice assumptions are
weaker because they do not require continuity, meaning that neighboring points need not
be close. Quantum entanglement may imply that neighboring event points are not always
close in space, which is possible in the lattice approach but not in general relativity.
7
A. Conserved quantities of the vortex model in the classical limit
The vortex model of elementary particles was rst introduced by Lord Kelvin for atoms
and since expanded with modern intrinsic quantum properties of particles [15]. Even pre-
dating the development of special relativity, the mass of the electron has been presumed to
derive from its electrical energy [17], and hence the stress-energy tensor of an electron, and,
by extension to quarks and other particles, all matter, must, classically, take the general
form of the Maxwell stress-energy tensor. For neutral particles, their stress energy is based
on spin and oscillation rather than magnetic and electric elds, but the critical point is that
spin/oscillation has, not only mass, but kinetic energy in the form of static pressure. In
other words, particles cannot be represented as point masses, blobs of matter where the
only non-zero component of their stress-energy tensors is T
00
= m (in the rest frame). The
stress-energy tensors have space-space components as well,
jk
T
jk
= m. Because they are
positive, these space-space components cannot be cancelled out in scaling up to macroscopic
bodies.
The vortex model proposes that all mass derives from intrinsic quantum properties of
spin and oscillation. Spin and oscillation are not separate entities, and each particle has a
covariant, anti-symmetric spin tensor, s

= s

, forming a single 4-dimensional spin in

the six planes of a Minkowski space. In any frame, oscillation occurs in the direction of the
polarization vector P
i
= s
0i
and spin in the plane with spin vector S
i
=
ijk
s
jk
.
The 5 5 matrix current of the SO(4,1) Yang-Mills theory, J

, describes the motion of

particles (see A for an overview of the de Sitter group). This current, written in component
form, is,
J

= T

(V

) +
1
2
S

, (5)
where S

represents Lorentz boosts (oscillations) and rotations (total angular momentum),

and T

represents the stress energy momentum tensor which is conserved by translations,

where , = 0, 1, 2, 3.
Classical particles oscillate and spin at the speed of light, c = 1. Let rest frame polariza-
tion vector be p
i
and rest frame spin vector be s
i
.
8
From the spin/polarization vectors, s
i
and p
i
, the anti-symmetric spin tensor is,
S

0 p
1
p
2
p
3
p
1
0 s
3
s
2
p
2
s
3
0 s
1
p
3
s
2
s
1
0

. (6)
The total energy is given by,
T

= [S

+
1
4

]. (7)
Like the electromagnetic stress energy tensor, this has the form (in Cartesian coordinates
(x, y, z)) [9],
T

1
2
(p
2
+ s
2
) U
x
U
y
U
z
U
x
P
xx
P
xy
P
xz
U
y
P
yx
P
yy
P
yz
U
z
P
zx
P
zy
P
zz

, (8)
where

U = p s, P
ij
= p
i
p
j
+ s
i
s
j

1
2
(p
2
+ s
2
)
ij
, p
2
= (p
x
)
2
+ (p
y
)
2
+ (p
z
)
2
, and s
2
=
(s
x
)
2
+ (s
y
)
2
+ (s
z
)
2
.
Similar to the electromagnetic Poynting vector,

S
Poynting
=
1
2

E

B, the vector

U/m is a
particles ux or velocity. The energy ux or momentum is

U. Particles in the rest frame
have no energy ux. This gives a denition of a massive particles rest frame: it is the frame
where the polarization and spin vectors are parallel, p s = 0. This means that in every
particles rest frame, the plane of oscillation is orthogonal to the plane of spin. Although
spin and oscillation of particles are essentially quantum parameters rather than properties of
motion, the point particle is analogous to a body following a periodic helical path revolving
about its axis of spin while oscillating normal to the plane of spin like a tiny vortex. As the
particle increases velocity with respect to the observer, the angle between the planes of spin
and oscillation decreases until they become parallel at the speed of light (which explains
why light waves are transverse). Therefore, the two 3-vectors are plus or minus the same
direction, p
i
= s
i
.
The conserved current above for massive particles in the rest frame has a convenient
representation in spherical coordinates,
T
00
= T
rr
= (p
2
+ s
2
) (9)
9
Because p
2
+ s
2
= m (with c = 1) in the rest frame,
T
00
= T
rr
= m (10)
Finally, the conserved currents described here should not be confused with the denition
of stress-energy-momentum in general relativity. General relativity provides a denition of
energy based on the symmetries of spacetime geometry. If we have an action that depends
on the metric eld, o(g

), then energy is simply the variation of that Lagrangian with

respect to the eld,
1

g
o
g

= T

. (11)
This is not the tensor, T

, above which is given by,

o
G

= T

, (12)
where G

is the gravitational potential discussed in the next section.

Only this tensor is conserved because the SO(4,1) symmetry pre-empts the coordinate
dieomorphic symmetry. General relativity assumes dieomorphism-covariance which im-
plies that the quantity 11 is conserved by Noethers theorem. Kretschmann pointed out to
Einstein himself (and Einstein conceded), however, that general covariance, i.e. coordinate
covariance, was a ctitious symmetry group generating no physical predictions [22]. Many
theories can be made that are generally covariant and predict dierent physical outcomes.
The physical restrictions on the theory are important and x the theory. Dieomorphism-
covariance in general relativity is an active symmetry that is unrestricted. Imposing an
SO(4,1) gauge symmetry on spacetime implies that, once it is xed, the gauge prefers a
particular coordinate system. With a preferred set of reference frames, although the coor-
dinate system can be changed at will, the theory is not actively dieomorphism-covariant.
Hence, Noethers theorem does not apply to the metric, and the stress-energy-momentum
tensor of General Relativity is not a conserved quantity. This is not the same as having a
preferred coordinate system overall, only that xing the gauge destroys the dieomorphic
freedom that general relativity assumes.
B. Derivation of a Yang-Mills theory without a background metric
Yang-Mills theory is typically dened with a Minkowski background metric,

. In this
section, A Yang-Mills theory is derived without a background manifold using a dual labelling
10
approach where each vertex on a four dimensional lattice is associated with two labels: (1) 4-
vector, y
a
= (y
0
, y
1
, y
2
, y
3
), labels derived from the gravitational potential that represent true
distances and torsions between vertices and (2) a second set of labels, x
a
= (x
0
, x
1
, x
2
, x
3
),
that are equidistant between neighboring vertices. The mapping x
a
y
a
is one-to-one and
onto since there is exactly one of each label for each vertex.
In this section only, I will use the beginning letters of the latin alphabet a, b, c, . . . , h
to represent 4-vectors. (I prefer not to use the usual Greek letters , because they are
typically used with a metric and a dieomorphic manifold. Here we assume no general
covariant structure because there is no manifold.) These letters range over 0, 1, 2, 3. I will
use y
4
and x
4
to refer to the additional, 5th dimension that completes de Sitter group vectors.
I will use the middle latin letters i, j, k, l (in this section only) to label entire vectors. When
I do so, the vectors will be bolded, e.g., y
i
.
Let time be imaginary t i. Consider a four dimensional lattice and label each vertex
with four numbers y
0
, y
1
, y
2
, y
3
. These labels are undened for now. Consider three vertices
in a row y,z, and w, and let U
yw
(z) be a small rotation matrix with > 0 the small
parameter. U
yw
= exp(A
a
) can be split into two parts: G
ab
and H
abc
by the following: an
SO(5) potential can be written as a sum of generators:
A
a
= G
ab
V
b
+
1
2
H
abc
M
bc
, (13)
where V
b
are generators for rotations in the y
b
-y
4
plane and M
bc
= M
cb
are generators for
SO(4) for rotations in the y
b
-y
c
plane where a, b, c = 0, 1, 2, 3. We will refer to G
ab
and H
abc
as the gravitational potentials.
To linear order in ,
U
yw
= I + A
a
= I + (G
ab
V
b
+
1
2
H
abc
M
bc
).
Each of the ten generator matrices, V
b
and M
bc
, is a 5 by 5 matrix. Let y = (y
0
, y
1
, y
2
, y
3
, y
4
)
and w = (w
0
, w
1
, w
2
, w
3
, w
4
), then we dene
w := [I + (G
ab
V
b
+
1
2
H
abc
M
bc
)]y. (14)
Eliminating the generator matrices, the equation 14 implies the relationship between the
4-vectors,
w
a
=
1
2
H
abc
y
c
+ G
ab
+ y
b
, (15)
11
y z w
U
yw
(z)
FIG. 1: The de Sitter operator U
yw
(z) transforms vectors in parallel transport from y to w. In
this picture, a vector starting at y both rotates and translates as it moves in parallel transport to
w. The dotted lines represent intervening states of the vector, but these are never realized because
points other than y, z, and w are not dened. The translation of the vector as well as the rotation
is accomplished by U
yw
(z) because there is no underlying metric space.
again, all to linear order in . The rst term in the right hand side of 15 represents the
torsion or twisting of a vector as it travels from y through z to end at w. The second term
on the right hand side represents the translation. (Note: if G
ab
(z) = 0 then there is no
translation and the two lattice points are eectively in the same location. This means that
it is possible for spacetime to collapse in on itself as in the Big Bang.) Figure 1 illustrates
how the potentials translate and rotate vectors in parallel transport from y to w, dening
the relationship between the points in spacetime y and w.
The labels of the lattice, y, are dened in terms of the gravitational potentials by hopping
from point to point outward from the only pre-dened point y (which is arbitrary of course).
Note: a similar procedure can dene the natural numbers in terms of a successor function
S, where 0 is the only predened point, from Peanos axioms. The gravitational potentials,
G and H, are successor functions that dene the spacetime lattice.
Having dened the labels as coordinates in a discrete spacetime, the Yang-Mills theory
can be derived using lattice gauge theory. First, dene the second labelling system by the
same method but using trivial potentials G

ab
=
ab
, the Kronecker delta such that
aa
= 1
and
ab
= 0 for a ,= b, and H

abc
= 0. This labelling system is denoted by x
a
and corresponds
to a lattice where each vertex is a distance /2 from each of its eight neighbors.
The relationship between the two coordinate systems, the articial one x and the true
one y, is illustrated in gure 2.
To dene the action of gravity, consider a plaquette p between four vertices in the a b
plane between x
i
, x
j
, x
k
, x
l
, and back to x
i
, and assume that each vertex on the plaquette
12
x
1
x
2 x
3
x
4
x
5 x
6
x
7
x
8
x
9

ab
y
1
y
2
y
3
y
4
y
5 y
6
y
8
y
7
y
9
G
ab
Artificial coordinate labels True coordinate labels
FIG. 2: This picture shows a 2D example of how the coordinates are dened. On the left is
the articial coordinate system while on the right is the true coordinate system. The subscripts
indicate which point maps to which, so x
1
maps to y
1
and so on. The distances, , between the
articial points on the right are equal (implying a Cartesian system) while those on the left are
given by the translation potential G
ab
scaled by . Only the distances on the right are measurable.
The lattice points on which the potentials are dened in between the coordinate points are not
shown.
is a distance (two lattice edges) from its two neighbors on the plaquette. (In other words,
there are lattice points at the midpoints in between the plaquettes vertices at which the
transformation are dened.) The action on the plaquette is given by S
p
= Re tr U
ij
U
jk
U
kl
U
li
.
Now, U
ij
= exp(A
a
(x)) where x = (x
i
+x
i
)/2 is the midpoint between x
i
and x
j
and likewise
for U
jk
, U
kl
, and U
li
. The remaining calculations can be found in a standard text such as
[38]:
U
ij
U
jk
= e
Aa
e
A
b
= e
(Aa+A
b
)+
1
2

2
[Aa,A
b
]+O(
3
)
U
kl
U
li
= e
A

a
e
A

b
= e
(A

a
+A

b
)+
1
2

2
[Aa,A
b
]+O(
3
)
Some matrix algebra gives,
U
ij
U
jk
U
kl
U
li
= e

2
{(AaA

a
)/(A
b
A

b
)/+[Aa,A
b
]}+O(
4
)
which simplies to
U
ij
U
jk
U
kl
U
li
= I+
2
_
(A
a
A

a
)

(A
b
A

b
)

+ [A
a
, A
b
]
_
+
_

2
_
(A
a
A

a
)

(A
b
A

b
)

+ [A
a
, A
b
]
__
2
,
13
to the required order. Summing over all plaquettes, there is a constant term, the vacuum
energy (note that here the gravitational potential does not couple to vacuum energy), the
middle term cancels out, and an equation for the total action is,
S =

p
Re tr
_

2
_
(A
a
A

a
)

(A
b
A

b
)

+ [A
a
, A
b
]
__
2
Taking 0 gives us the Yang-Mills potential and continuing to real time from imaginary
time changes the
ab
operators into
ab
operators,
S =
1
4g
2
_
d
4
x
ac

bd
F
ab
F
cd
, (16)
where g is the coupling constant and
F
ab
=
a
A
b

b
A
a
+ [A
a
, A
b
]. (17)
The Minkowski metric comes from continuing from the SO(5) group to the SO(4,1) group
and represents the coordinate system in which the de Sitter group operators are represented
rather than the metric of space time.
The only requirement of the articial coordinates is that they map every coordinate
point x
a
to every true coordinate point y
a
one-to-one and onto. The coordinate system, x
a
,
is generally covariant in the continuum limit, but the dieomorphic symmetry is vacuous
because the coordinate points are not measurable. Only the rst coordinate system, y
a
, is
measurable. Distances and time are measured with rods and clocks. Torsion is measured by
detecting particle spin and oscillation. Hence, the coordinates y
a
are observables, but x
a
is
invented for convenience, a bookkeeping system.
A note on indexes: 15 implies that all the indexes of the gravitational potentials (including
the rst index) refer to true coordinate indexes. Indeed, by using the gravitational potential
to dene both the true and articial sets of coordinates, we avoid invoking another type of
index. In the rest of the paper, Greek letters are used for all indexes in order to conform
to the convention set by relativity, e.g., G

rather than G
ab
. The gravitational potentials,
however, are pseudotensors not tensors.
C. Yang-Mills theory and derivation of the eld equations
In this section, I begin with standard Yang-Mills theory and derive a locally de Sitter
theory of quantum gravity using the de Sitter Lie algebra. Yang-Mills theory describes
14
forces as exchanges of gauge bosons based on a group symmetry. The Standard Model of
quantum eld theory currently contains three forces, electromagnetism, the weak force, and
the strong force in a U(1)SU(2)SU(3) group symmetry, all using the Lie algebras of
semi-simple groups.
A conventional SU(N) Yang-Mills theory has action,
S =
1
4g
2
_
d
4
xF

, (18)
where g is the coupling constant and
F

i[A

, A

], (19)
is the force (a misnomer carried over from the electromagnetic Lorentz force) where A

is
a matrix potential from the SU(N) group. For the SO(4,1) group these are 5 5 matrices.
(Note that in the previous section we absorbed a factor of i into A

to make the matrices real.)

The theory is invariant under a gauge transformation U(x) such that A

= U(x)A

(x)
i/g

U(x)U

(x). For innitesimal transformations, U(x) = e

i(x)
I + i(x), A

= A

(x) + i[(x), A

] is the gauge transformation.

Yang-Mills equations, expressed in group component form, have the action,
S =
1
4g
2
_
d
4
xF
a

F
a
, (20)
F
a

A
a

A
a

+ f
abc
A
b

A
c

, (21)
and f
abc
is the group symbol.
The SO(4,1) potential is written as a sum of generators of Lorentz rotations and boosts
and of boost/rotations with respect to x
4
,
A

= G

V

+
1
2
H

, (22)
where the pseudotensor potential G

couples to T

and pseudotensor H

to S

. In
this sense, the theory resembles Einstein-Cartan theory but not in the same way as the
Macdowell-Mansouris action which is neither de Sitter invariant nor renormalizable [37].
Following the de Sitter algebra of A, the forces of gravity decompose into a covariant
form with two equations,
E

(G

+ G

) , (23)
15
and
F

+ G

(24)
where

=
1
2

(H

+ H

) . (25)
Thus, the action of gravity is,
S
gravity
=
1
4a
g
_
d
4
xE

+
1
2
F

. (26)
Here the coupling constant is a
g
= g
2
.
The classical equations of motion of any Yang-Mills theory can be found by the variation
of the action, S = 0, via the Euler-Lagrange equations,

_
/
(

a
)
_

/
A

a
= 0, (27)
where / = F
a

a
is the Lagrangian. Evaluating the Euler-Lagrange equations for the
Yang-Mills Lagrangian, the equations of motion are,

i[F

, A

] = 8g
2
J

, (28)
where J

is the conserved current.

Applying the denition of the group symbol again and the previous denitions 23 and
24 to the eld equations (28) yields the following equations, analogous to the Einstein eld
equations and the Cartan torsion equations respectively:

(E

+ G

) = 8T

(29)

+ E

= 8S

(30)
where

=
1
2

(F

+ F

) . (31)
16
The usual negative sign on the RHS is ipped to positive because the i is absorbed into the
potential. This is what makes like signed masses attract rather than repel. The coupling
constant, a
g
= g
2
, is non-dimensional, but, because the size of the mass quantum is not
known, the standard coupling constant, = G/c
4
, in units such that = 1, sucies for the
classical equations.
Boundary conditions are:
G

0 (32)
and
H

0,

0 (33)
as x

.
Gauge transformations can be done in component form. Consider the SO(4, 1) de Sitter
transformation (x) =

(x)V

+
1
2

. The gauge transformation is

G

= G

(x) +

(x)H

(x)H

(x)G

) (34)
and
H

= H

(x)+

(x)G

+
1
2

(x)H

(x)H

(x)H

(x)H

),
(35)
which can be found from the Lie algebra.
D. Equations for evolution of matter
The continuity equations determine the evolution of matter. By Noethers theorem, for
any matter action, o
M
, e.g., Diracs, the relations between the potential G

and the stress-

energy-tensor T

and the potential H

and the spin-density S

are,
T

=
o
M
(G

)
, (36)
S

=
o
M
(H

)
, (37)
and the continuity equations are,
D

= 0, (38)
D

= 0, (39)
17
[10], where, for a generic source J

a
and potential A

a
, D

i[A

, J

] is the
covariant derivative. Applying the de Sitter Lie algebra to 38 and 39, the full continuity
equations are,

(G

+ T

) = 0, (40)

+ G

= 0 (41)
where

=
1
2

(H

+ H

) . (42)
There are nonlinearities in the eld equations 29-30 and the continuity equations because
gravitons have spin, polarization and momentum and couple to themselves. These nonlin-
earities are too small to detect except at the cosmological scale (Sec. III D) and close to
strongly gravitating objects moving at high speeds or rates of rotation.
When H

= 0, the eld equations without an angular momentum source revert to the

Abelian equation,

) = 8T

, (43)
which, under the harmonic gauge condition, is,
G

= 8T

, (44)
which relates the tensor T

to the pseudotensor G

. This equation is most useful for

experimental predictions since H

is usually too small to detect.

E. Equation of test particle motion
There are two forms of test particle: (1) a parameterized path with no structure which
models massless particles such as photons and (2) a body with internal static pressure which
models massive particles.
18
1. The geodesic of light and other massless particles
A test particle with null geodesic and no internal structure is represented as a parame-
terized path x

(). Given the basic structure of the theory from Section II B, the potential,
G

, takes the place of the metric normally found in Lagrangians for test particle motion.
Indexes are raised and lowered with the Minkowski metric

. The Lagrangian for motion

is:
/ = G

dx

d
dx

d
+ H

dx

d
dx

d
dx

d
. (45)
This describes both the curvature and twisting of the path as it moves through spacetime
along its shortest path. (One can derive this equation by looking at the discrete lattice in
Section II B and describing a path as a sequence of connected edges in the lattice.)
No observations of gravitational phenomena, save perhaps cosmological observations, are
sensitive enough to measure torsion. Therefore, we concern ourselves in this section only with
geodesic motion. Let H

= 0. (A similar assumption is made in applying Einstein-Cartan

theory.) Let the Lagrangian for particle motion be,
/ = G

dx

d
dx

d
. (46)
The equations of motion are given by the Euler-Lagrange equation,
d
d
_
/
(dx

/d)
_

/
x

= 0. (47)
The equation of motion for the particle (the geodesic) is,
d
2
x

d
2
+
1
2
(G
1
)

]
dx

d
dx

d
= 0, (48)
which is the geodesic equation. Note: there is a distinction between raising indexes and
inverting G

. Its inverse is (G
1
)

such that (G
1
)

, not G

.
2. The geodesic of massive bodies
For a test particle with internal static pressure (which applies to bodies such as planets),
let the Lagrangian for particle motion be,
/ =
1
2m
G

, (49)
19
where T

is the particles current given its energy of spin and polarization as dened in the
section on conserved currents (Sec. II A).
Let the rest frame of the particle (specically, the frame where the energy ux, T
0
and
T
0
, vanishes) be x

= (, r,

,

) where the particle is at (, 0, 0, 0). (Both the gauge and
the coordinate system must be changed to the rest frame.) Let

G

be the potential and

, the current in frame x

. From 10 the rest frame current is

T
00
=

T
r r
= m; therefore,
49 is,
/ =
1
2
(

G

+

G
r r
). (50)
At the particles location, r = 0, the metric degenerates into a 2-D metric, g

=
(1, 1, 0, 0). Let the potential be traceless, i.e. g

= 0,

G

=

G
r r
.
Remark: The assumption that the potential is traceless is not a restrictive assumption
because one can always nd a gauge where the potential is traceless if H

= 0. Without
loss of generality, assume a space in Cartesian coordinates (t, x, y, z). Given a potential G

with trace

= f(t, x, y, z), choose a function

0
=
_
t
0
dt

f(t

, x, y, z). Now,
0

0
= f.
Change the gauge, G

0
= G
0

0
. Since o-diagonals do not aect the trace, the resulting
potential is traceless.
Replacing

G
r r
with

G

in the Lagrangian, we have,

/ =

G

, (51)
or, in the observers frame,
/ = G

dx

d
dx

d
, (52)
which is the same as 49. Thus, in both test particle cases the geodesic equation is identical
to that of GR.
F. Strong Equivalence Principle
Equivalence principles are dening features of gravitational theories. Newtons theory
included the Weak Equivalence Principle (WEP) where an objects weight is proportional
to its mass. Einstein developed the equivalence principle named for him, the Einstein Equiv-
alence Principle (EEP), which states that local experiments in free fall are independent of
position and velocity. When strongly gravitating bodies are taken into account, the validity
20
of the equivalence principles come into doubt because internal structure violates WEP by,
e.g., the (never observed) Nordtvedt eect [35].
The Strong Equivalence Principle (SEP) is a natural extension of early equivalence prin-
ciples. It has three conditions: (a) an objects weight is proportional to its mass (WEP)
for self-gravitating bodies as well as test bodies, (b) the outcome of any local experiment
is independent of the velocity of the (freely falling) apparatus, and (c) the outcome of any
local experiment is independent of where and when in the universe it is performed [35].
Although a rigorous proof has not been found, GR is the only known metric theory that
appears to satisfy SEP strictly. Because experiment has never found a violation of SEP, any
new gravitational theory must satisfy strict constraints on SEP violations. In the following,
the Yang-Mills theorys gauge symmetry allows the required transformations to satisfy SEP.
The Yang-Mills theory satises the rst condition of SEP, (a), by the arguments of
Section 20.6 of [19] also found in [36]: If the eld of a self-gravitating body asymptotically
approaches atness (G

(R) =

) at some distance, R, considered to be the boundary

of the local system, this is sucient to guarantee that a bodys self-gravitation and other
internal structure does not aect its motion. Because it is always possible to nd a de Sitter
gauge that eliminates the gravitational eld at the boundary between a local, spherically
symmetric, compact system and the external environment, a spherically symmetric self-
gravitating body, even a neutron star or black hole, can be regarded as a point particle.
In bimetric theories, which are dieomorphism-covariant, changing the gauge to eliminate
the eld at the boundary changes the Minkowski metric, so a coordinate system cannot
be found that satises this requirement. With the Yang-Mills theory, however, the gauge
can be changed without changing the coordinate system. It is simple to nd a gauge such
that G

(R) =

for a spherically symmetric solution. Therefore, the Yang-Mills theory

satises condition (a).
The Yang-Mills theory also satises the second condition of SEP, (b), by local Lorentz
covariance (a subset of local de Sitter covariance) which means that experimental outcomes
are independent (by gauge covariance) of Lorentz boosts and rotations. It also satises the
third condition of SEP, (c), by having no preferred location/time in the eld equations.
Unlike Rosens bimetric theory which has a local gravitational constant that depends on the
eld [35], the Yang-Mills theory has a non-location specic gravitational constant, a
g
. Be
these arguments, all three conditions of SEP are met.
21
G. Quantization
The quantization of the theory is given by the generating functional:
Z[T, S] =
_
DGDH exp
_
i
4
_
d
4
xE

+ F

+ ia
g
_
d
4
xG

+ H

_
(53)
Although it is derived without a background manifold, the action is identical to those of
Yang-Mills theories that do have a background. Because it is a massless gauge boson Yang-
Mills theory, the theory of gravity given in this paper has a nite polynomial, positive denite
action (20). Like all Yang-Mills theories on semi-simple groups the theory is renormalizable
[29].
The standard quantization scheme is appropriate with Fadeev-Popov ghost elds used to
derive Feynman rules including the G

graviton propagator [38]:

D

(p) =
i

p
2
+ i
_

(1 )p

p
2
+ i
_
. (54)
Other than proving renormalizability, which was achieved in the 1970s, any quantum pre-
dictions are left for future work.
III. EXPERIMENTS AND OBSERVATIONS
Numerous observations, starting with light bending in 1919, have been made to attempt to
conrm predictions of general relativity. These include gravitational time dilation, redshift,
and light bending all of which have been measured within the solar system. Perihelion
precession of the planets and binary pulsar precession have also been measured to good
rst order accuracy [31]. Other observations do not agree with the original theory and
have required modications to the Einstein eld equations. In the following, are derived
(1) a static, spherically symmetric solution to the eld equations, (2) the 1PN equations of
motion for a binary system, (3) the radiation reaction for binary pulsar inspiral, and (4) the
accelerating expansion of the universe via a homogeneous, isotropic model.
22
A. Spherically symmetric solution
The static, spherically symmetric Schwarzchild solution is one of the most important
solutions to the Einstein eld equations. The solution to the de Sitter abelian eld equations
44 is identical up to linear order. Let spacetime be covered by spherical coordinates (t, r, , )
where is colatitude and let the metric be the at spacetime metric [31],
ds
2
= dt
2
+ dr
2
+ r
2
d
2
+ r
2
sin
2
()d
2
. (55)
Choosing a harmonic gauge the equations for the potentials can be found by the general
Greens function solution [36] to the relaxed eld equations 44,
h

(t, x) =
_
C
d
3
x

(t [x x

[, x

)
[x x

[
, (56)
where h

= (G

)/2 and ( is the past lightcone. Given a form for the stress-energy
tensor, T
00
= M(r) and T
rr
= M(r), the solutions to the integral equation (in spherical
coordinates) give the potential pseudotensor:
G
00
=
_
1
2M
r
_
, G
rr
= 1 +
2M
r
, G

= r
2
, G

= r
2
sin
2
(), (57)
with the boundary conditions such that G

as r . This is also the linearized

Schwarzchild solution to the Einstein equations [31].
B. Particle trajectories
Particle trajectories in a static, spherically symmetric potential eld are identical to those
of general relativity. The only dierence is in the radial potential. The Schwarzchild radial
potential g
rr
= (1 2M/r)
1
diers from the radial potential derived in Section III A to
quadratic order,
G
rr
= 1 + 2M/r (1 2M/r)
1
+ O((M/r)
2
). (58)
No tests within the Solar System, whether with respect to the Sun, Earth, or another body
(the most precise tests have been done near the Earth) have achieved better than linear order
in Schwarzschild coordinates [19][31] because of the weak elds involved. If M = Gm/c
2
is
the Schwarzchild radius and m is the mass in other units (e.g., kilograms) at the surface of
the Sun, 2M

/r 4.25 10
6
and at the surface of the Earth 2M

/r 1.4 10
9
. Both
23
are much too small for higher order eects to be measured, and imperfections in density
(e.g., mountains) would make measurements dicult to verify.
Predictions of gravitational time dilation and redshift, meanwhile, are mathematically
identical in both theories (to any order) given a static, spherically symmetric eld with
redshift proportional to 1 + 2M/r. Recent high precision measurements of redshift using
cesium atoms in a laboratory agree with predictions of both theories [20]. In fact, gravita-
tional redshift is identical in both theories to all orders for a static, spherically symmetric
source. Measurements of the geodetic eect and frame dragging by Gravity Probe B are
likewise rst order accuracy measurements [8]. Therefore, no tests done within the Solar
system to date disprove either theory. For strong eld tests, require looking outside the
solar system to binary pulsars and, further on, cosmology where the theories have the best
chance of being tested.
C. Post-Newtonian Equations of Motion
The binary pulsar system B1913+16 discovered in 1974 provided one of the rst tests of
strong eld general relativity [36]. Because the regular radio pulses of the star, orbiting a
relatively inert body, possibly a dead pulsar, allowed precise measurements of the pulsars
orbit, the small deviations of the two body orbit from Keplers laws can be measured,
including orbital damping caused by gravitational radiation, which is a higher order eect
than orbital precession [36]. Observations of this system over the past thirty-ve years have
increased the accuracy of the measurements. Recently, the double binary pulsar system
PSR J0737-3039A/B with two active pulsars orbiting each other has provided even greater
precision [13]. The following demonstrates that the eld equations of Sec. II C agree with
the observations of these pulsar systems.
The parameterized post-Newtonian (PPN) equations of motion of gravity are the primary
vehicle by which not only dierent theories of gravity (including GR) are compared but
provide the equations for predicting the motion of two or more gravitating bodies. Tests of
these equations include the measurement of planetary motion within the Solar system for
weak elds and binary pulsar precession and gravitational radiation for stronger elds [19]
[35].
For a binary pulsar system the GR metric in the post-Newtonian coordinate system for
24
general relativity is ([35], 11.52)
g
00
= 1 + 2

a=1,2
m
a
/[x x
a
(t)[ + O(
4
), (59)
g
0j
= O(
3
), (60)
g
ij
=
ij
_
1 + 2

a=1,2
m
a
/[x x
a
(t)[
_
+ O(
4
), (61)
where is a small parameter such that v/c and GM/c
2
r
2
with m
1
and m
2
being the
masses of the bodies and x
1
and x
2
their positions in the appropriate harmonic coordinate
system. Because GR satises SEP the self-gravitation of the two bodies along with their
motion can be included in their masses and they can be modelled as test bodies with the
geodesic equation.
The same method used to derive the GR metric can be used on the relaxed eld equations
of the Yang-Mills theory (44). The calculations are somewhat involved and are included in
Appendix B. They result in the following potential:
G
00
= 1 + 2

a=1,2
m
a
[x x
a
[
+ O(
4
), (62)
G
0j
= O(
3
), (63)
and
G
ij
=
ij
_
1 + 2

a=1,2
m
a
[x x
a
[
_
+ O(
4
). (64)
The velocities of the bodies are v
1
= (v
11
, v
12
, v
13
) and v
2
= (v
21
, v
22
, v
23
) and v
a
= |v
a
|.
(See Appendix B for a derivation of these potentials.)
Since the theory satises SEP, for any set of compact nearly-spherical bodies where
tidal forces may be neglected, the geodesic equation 48 predicts orbital motion. The post-
Keplerian parameters of periastron advance, , time delay,

, and Shapiro delay param-

eters, r and s, are also, consequently, the same in both theories. Higher order terms in the
PPN equations do not match, however, (see Appendix B) and oer the best chance of ruling
out one of the theories with continued observations of binary pulsars.
The nal post-Keplerian parameter, orbital speed-up caused by gravitational radiation,

P
b
, derives from the quadrupole formula from the Yang-Mills eld equations discussed in
the following section.
25
1. Orbital speed-up of a binary pulsar system
Gravitational radiation was rst addressed by Einstein shortly after the publication of
general relativity, where he and others demonstrated that the primary radiation is quadrupo-
lar in contrast to the dipolar radiation from electromagnetic sources [7]. The energy loss
of a binary system of compact stars caused by radiation was rst demonstrated in a paper
by Peters and Matthews [24], who derived the energy loss of binary stars in Keplerian orbit
leading to the formula for the orbital speed-up,

P
b
=
192G
5/3
5c
5
_
P
b
2
_
5/3
(1 e
2
)
7/2

_
1 +
73
24
e
2
+
37
96
e
4
_
m
p
m
c
(m
p
+ m
c
)
1/3
, (65)
with m
p
and m
c
the masses of the pulsar and companion body respectively and e the orbital
eccentricity [33]. Although this equation was developed for Keplerian orbits, it applies to
post-Keplerian orbits since the only requirement is that the orbit be elliptical, and, although
much higher order derivations have been made [35], only this equation has been tested. The
equation derives from a multipole expansion of the wave equations for general relativity, and
a detailed discussion of multipole expansions for gravitational radiation in general relativity
can be found in [30]. For measured radiation-reaction only the lowest order quadrupole term
is relevant. The reaction which is what is measured is caused by the conservation of energy
which requires a system expelling radiation to slow in some fashion. Orbital speed-up is a
reaction to the loss of gravitational potential energy.
Orbital speed-up is a function of energy loss, i.e. radiation-reaction, which to lowest
multipole order in General Relativity is ([19], 36.31),
dE
dt
=
1
5

d
3
Q
jk
dt
3
d
3
Q
jk
dt
3
, (66)
where Q

is the reduced quadrupole moment (the trace free part of the second moment of
the mass distribution) such that Q

= q

1
3

q,
q

=
_
d
3
x
0
x

, (67)
and q = q

. The Peters-Matthews formula for energy loss of a binary system ([35], 10.80),
dE
dt
=
_

2
m
2
r
4
8
15
_
12v
2
11 r
2
_
_
, (68)
26
where m = m
1
+ m
2
, = m
1
m
2
/m, v = [v
1
v
2
[, and r = [x
1
x
2
[, can be found from
the linearized vacuum equations of general relativity,

= 0, [24][19] by the quadrupole

relation,

h
jk
(t, x) =
2
r
q
jk
(t r), (69)
for the spatial part of the radiation eld ([19], 36.50) where r is the distance from the source.
Peters and Matthews derives the energy loss using the linearized Einstein vacuum equa-
tions:
g

+ h

, (70)
such that

= 0, (71)
where

h

= h

1
2

.
The radiation formula can be found by a plane wave solution to the linearized equations.
The linearized Yang-Mills equations for an unpolarized, spinless source,
G

= 8T

, (72)
is the same as the linearized equation for GR.
Let G

= h

. A plane wave solution to the wave equation,

h

= 0, (73)
is
h

= ae

cos(t

k x), (74)
where a is the amplitude, e

is a symmetric, traceless, transverse, and unitary polarization

tensor (as dened in [24]), and is the frequency.
The solution can be expanded by multipoles folllowing the outline in [35], Chapter 10.
Several steps are left out here but have been worked out there.
The multipole expansion of the plane wave solution (74) is,
h

= 4r
1

m=0
(1/m!)(/t)
m
_
d
3
x

(t r, x

)(n x

)
m
. (75)
to 1PN order, where n = x/r. Because of the gauge condition,

= 0, and the retarded

potential, the following relations apply,

0
h

0k
= n
j

0
h

jk
,
0
h

00
= n
j
n
k

0
h

jk
, (76)
27
and only need to determine the h

jk
components [35]. Because, to order,

= 0 by
conservation and the source is symmetric, a useful relation is,
(
2
/t
2
)
_
d
3
xT
00
x
j
x
k
= 2
_
d
3
xT
jk
(77)
Therefore, 75 becomes,
h

ij
= 2r
1
(
2
/t
2
)
__
d
3
xT
00
(t r, x)x
i
x
j
_
+ higher order. (78)
This means that monopole and dipole moments of T
ij
can be expressed as time derivatives
of quadrupole moments of T
00
. To post-Newtonian order T
00
= , the mass density, and
the quadrupole approximation follows.
For the binary system the integral simplies to,
h

ij
=
2
r
q
jk
(t r). (79)
Since from 69 the solution is the same as for GR, h

ij
=

h
ij
and, because

h
ij
solves the
vacuum equation,

h
ij
= h
ij
obtaining the speed-up formula (65) follows as for general
relativity: take a Taylor expansion of 79 in powers of r to extract the radiation-reaction
potential and applying the geodesic equation 48, determine the reaction acceleration of the
bodies ([19], Sec. 36.11). Then arrive at 65. See [35] and [19] for detailed calculations. This
proves that the energy loss caused by radiation is the same in GR and the YM theory. The
reaction, then, can be modelled by a loss of potential energy between the two bodies.
D. Cosmology
If this paper had been written as recently as 15-20 years ago, cosmology could have
been addressed more fully within the scope of this paper. In recent years, however, the
amount of data available to match to cosmological models has exploded thanks to projects
such as WMAP [21], observations of Type 1a supernovas, and baryon acoustic oscillation
measurements. In this section, the Robertson-Walker model is addressed. This is the starting
point of the two most prominent numerical approaches, both centered around the Friedmann
equations: perturbations of the Einstein eld equations and N-body cosmology which models
the universe as an N-body Newtonian system.
The Friedmann equations can be derived from Newtons laws [32], suggesting that at
slow speeds and very weak elds they are valid for the universe as a wholecertainly the
28
present universe is essentially Newtonian at large scales where inhomogeneities are small. In
the early universe, however, relativistic speeds and strong gravitational elds predominated
and the Newtonian model breaks down for times close to the Big Bang.
Cosmological observations have not found major anomalies in the empirical CDM model,
the main cosmological model at this time, but the constraints on alternative theories of grav-
ity are much weaker than those from N-body observations. The Yang-Mills theory predicts
some deviations from the standard theorynone of which are challenged by observation
because of all the unknowns involvedbut it also explains accelerating expansion without
a cosmological constant.
Whether one assumes general relativity or an alternative, the accelerating expansion, in
the de Sitter point of view, is a natural result of translation occurring over a negatively curved
hypersurface. The locally de Sitter model of this paper treats each point of the universe as if
it were on the hypersurface of a ve dimensional hyperbola. In order to move from one point
to another, a particle translates over the hypersurface rather than translating in a straight
line. In an isotropic, homogeneous universe, the entire universe exists on the surface of a
single hyperbola. The standard cosmological de Sitter model applied to Einsteins theory
implies a cosmological constant as a consequence of a hyperbolic surface of xed curvature
[32]. In contrast, the Yang-Mills theory implies that the hyperbolic curvature is an active
potential that changes with time. Thus, the two theories make dierent predictions about
the past and future of the universe.
1. Isotropic, Homogeneous Universe
Observations of the cosmic microwave background and statistical counts of the distribu-
tion of galaxies imply that the universe is statistically isotropic [31]. By the Copernican
principle that human beings do not occupy a privileged location in the universe, the uni-
verse must also be homogeneous. An isotropic, homogeneous universe has a simple dynam-
ical description. For a perfect uid universe, the stress-energy-momentum tensor in special
relativity has the standard form,
T

= u

+ p(u

+ g

), (80)
29
where is mass density and p is pressure [31]. The fundamental Friedmann equations for
general relativity are,
a
2
+ k =
8a
2
3
, (81)
3 a
a
= 8( + 3p), (82)
where k is the curvature constant [32] and
g
00
= 1, g
rr
=
a
2
()
1 kr
2
, g

= a
2
()r
2
, g

= a
2
()r
2
sin
2
, (83)
is the metric. Recent observations that the universes density, the sum of baryon, dark
matter, and dark energy densities, is near or at the critical value,
b
+
c
+

crit
, and
the universe is spatially at, k 0 [21][12]. Therefore, for the rest of this section, k = 0.
From the conservation of energy,

+
3 a
a
( + p) = 0, (84)
is a fundamental equation for the evolution of matter and energy.
Turning to the Yang-Mills theory, the equations of motion for the universe are somewhat
dierent and require explanation within the constraints of observation. Modern cosmology
relies on perturbation methods with the Robertson-Walker assumptions of isotropy and
homogeneity the zeroth order equations. Therefore, equations based on these assumptions
are critical to any treatment of cosmology.
The isotropic, homogeneous universe under the General Relativity model does not dis-
tinguish between matter moving apart within space and space itself moving apart. Hence,
redshifts are interpreted as being caused by matter moving apart as space drags it along
in its expansion. It is possible, however, to distinguish between the two types of redshift,
gravitational and doppler, in ordinary, non-cosmological circumstances such as within the
Solar System.
In this section, the theory predicts that, in the Poincare approximation, gravitational
redshift expansion is constant in the later universe while Doppler redshift is linear. Thus, in
that approximation all observed redshift is caused by the Doppler eect of matter expanding.
In the de Sitter theory, however, the inuence of the torsion potential, H, causes gravitational
redshift to increase with time causing total redshift (gravity + doppler) to accelerate with
time. Based on the discussion of true vs. articial coordinates (Sec. II B), this implies that
30
matter is expanding as well as space, causing a dual expansion and, hence, acceleration of
the universes expansion.
As discussed in section II B, the coordinate system is a set of degrees of freedom, that,
while having no physical meaning, may be used to simplify equations. Comoving coordinates,
g
00
= 1, g
ij
= a
2
(t)
ij
, expand with matter moving within space. The expansion of space
itself is given by the potential, G
00
= 1 and G
ij
= b(t)a
2
(t)
ij
. The torsion potential is
non-zero, H

,= 0 as a consequence of the de Sitter Lie algebra, [V

, V

] ,= 0, but vanishes
in the Poincare approximation where [V

, V

] = 0 (see B). Let H

i0i
= H
ii0
= c(t)a
2
(t) and
H
iij
= H
iji
= d(t)a
2
(t) be the isotropic, homogeneous torsion potential.
In the comoving coordinates, the conserved current is T
00
= and T
ij
= pa
2

ij
. The
spin density is zero by isotropy, S

= 0. Evaluating the eld equations 29, the time-time

equation is,
3 a(a

b + b a a + ca)
a
2
= 8, (85)
and the next three (all the same) space-space equations are,
3 a

ba

ba
2
aab a
2
b + a a

ba
2
ca
2
2 aac
a
2
= 8p. (86)
From the eld equations 30, the torsion equation for oscillatory torsion is,
ca
2
+ 3 c aa 3 a
2
c + aac aab + a a

ba
2
ca
2
= 0 (87)
and the equation for twisting torsion is,

da
2
a
2
d + a a

d a
2
b
2
d = 0. (88)
By the vortex model all matter has p = /3 which simplies 40 to
= 4
a
a
(89)
This does not imply all matter is radiation nor does it support theories such as hot dark
matter. Matter can be slow moving, but its internal static pressure does aect its behavior
in a gravitational eld.
The system of three dierential equations (85,86, and 89) govern the gravitational behav-
ior of matter in this universe. The boundary condition are given in terms of present day, t
0
,
and origin, t = 0, parameters: a(t
0
) = a
0
, a(0) = 0, b(0) = 0, c(0) = 0, and c(t
0
) = c
0
. The
31
assumption of a(0) = 0 is a classical assumption. Quantum eects take over at t 1 and
could predict a dierent initial size of the universe or a quantum bounce where quantum
pressures could reverse a collapsed state.
The equations 85,86, and 89 have the non-trivial solution,
a(t) =
a
0
t
0
t, (90)
(t) =
0
a
4
0
a
4
, (91)
b(t) =
1
12t
2
_
Ct
4
3Ct
3
+ (12 + 8
0
t
4
0
)t
2
40
0
t
4
0
t 32
0
t
4
0

, (92)
c(t) =
1
12
_
3Ct + 8
0
t
4
0

, (93)
where C =
1
3t
0
(12c
0
+ 8
0
t
4
0
). The equation for d(t) is not expressible in closed form but
is the solution to a second order ODE. Note: these results were obtained with Maple
TM
s
tensor package and PDE solver.
The scale factor, a(t), conforms to a universe with a Minkowski metric sometimes called
the Milne universe or simply the linear model [16]. The Milne cosmological model has
recently seen a resurgence as an alternative to the standard CDM model and has some
compelling features: It gives the same age for the universe as the CDM model [14]. Nu-
cleosynthesis has been t to observations for the linear model in [28] with some problems
that I discuss below [11]. Predictions for Type 1a supernovae are very close to those of the
standard model. The Milne universe has angular distances in the CMB of about 1.1 degrees,
also close to the observed value [2] [3].
Although the rate of matter expansion is linear, the gravitational potential, b(t), shows an
acceleration in total expansion rate (space + matter), related to
_
G
ij

ij
/3 =
_
b(t)a
2
(t)
O(t
2
), meaning that the redshift of distant galaxies should be accelerating linearly with time.
It is clear that the theory deviates from the linear or Milne theory (against which there have
been objections based on Big Bang Nucleosynthesis [11]) in relation to how strongly the
universe deviates from the Poincare approximation. At the present time, the acceleration is
not well understood enough to determine whether it is linear, exponential (as in models with
a cosmological constant), or some other function of time. Hence, should better measurements
become available that show the rate of acceleration, they may test the theorys predictions.
32
2. The Poincare approximation
If the hypersurface over which particles translate in the de Sitter model has innite size,
translations become straight, and the Poincare group is the result. When the Poincare
approximation is made such that we change the equations to conform to the Lie algebra
with [V

, V

] 0 (see A), the expansion no longer accelerates:
a(t) = a
0
t/t
0
, (94)
(t) =
0
_
a
0
a
_
4
, (95)
b(t) = 1
8
0
3(a
4
0
/t
4
0
)t
2
+
K
(a
0
/t
0
)t
, (96)
where K = (b
0
+ 8
0
/(3a
4
0
/t
2
0
) 1)a
0
and b
0
= b(t
0
). Note: b(t) 1 as t indicating
that the Poincare universe has no accelerating expansion at its present age. The de Sitter
Lie algebra oers an explanation: when [V

, V

] = iM

(see A), momentum, related to

the V

generators, produces spin. Because [M

, P

] = i(

) spin gravitons
couple to momentum gravitons to create more momentum gravitons. Since the potential
for momentum gravitons is G

, which relates to the expansion of space, this relationship

causes accelerating redshift.
Another physical phenomenon has a similar explanation: Thomas Precession. In that
case, the Lorentz rather than de Sitter algebra works to relate orbital momentum of the
electron to changes in its spin [9]. Thus, dark energy can be explained as the production
of gravitons by an interplay between spin and momentum in de Sitter group theory.
IV. POTENTIAL DISAGREEMENTS
Observations of gravity are classed into a four basic types: (1) Newtonian gravity which
applies to very weak elds and slow velocities that pertain to most galaxies, stars, and orbits
of the outer planets, (2) weak eld gravity responsible for classic eects such as redshift,
light bending, time dilation, and perihelion precession, (3) strong eld gravity responsible
for binary pulsar precession, orbital damping, and gravitational waves, and (4) cosmology
where assumptions such as homogeneity and isotropy greatly simplify equations as well as
perturbative cosmology (not addressed in this paper). Included are tests of the Strong
33
Equivalence Principle (SEP).
Going through each of the four types of tests:
Newtonian gravity. Both theories subsume Newtonian gravity at very weak eld strengths
and slow velocities, easily demonstrated from the eld equations. There is no disagreement
here which is not true of all alternatives to GR. Conformal gravity is an example of a theory
that diverges from Newtonian gravity [18].
Weak Fields. As shown in section III B, both theories agree with classic eects up to linear
order in the Schwarzschild radius over the distance. Measurements of redshift, light bend-
ing, perihelion precession, and time dilation have not been made to quadratic accuracy in
Schwarzschild coordinates (as opposed to isotropic coordinates where the coordinate change
introduces an articial measurable quadratic term); therefore, none of these experiments
contradict either theory.
Strong Fields. As [13] mentions, measurements of binary pulsar precession, while some of
the most precise measurements of relativistic gravity ever made, are not suciently accurate
to conrm beyond the 1PN equations of motion to the order given in Sec. III C. Orbital
damping caused by radiation reaction is not conrmed beyond the quadrupole approxima-
tion, and neither GR nor the YM theory predict a dipole moment. Therefore, as shown
in section III C, none of these measurements contradicts either theory nor any other mea-
surements of multibody systems. The YM theorys geodesic equation 48 is also the same
as in relativity; therefore, since SEP is satised (Sec. II F) and compact self-gravitating
bodies behave as if they are test bodies, it agrees with the 1PN equations of motion of gen-
eral relativity. Gravitational wave measurements promise to provide higher order estimates
which may show a violation of one of the theories, but these, as yet, are not available. Few
alternatives to GR satisfy SEP and some have been ruled out by N-body observations that
have constrained the PPN parameters.
Cosmology. Several sources including Type Ia supernovae (SN), baryon acoustic oscilla-
tion (BAO), and the cosmic microwave background (CMB) provide data that can be used
to constrain various theories [12]. The Robertson-Walker model is essentially a toy or zeroth
order model because it assumes no uctuations or deviations from perfect uniformity and
isotropy in the universe. First order perturbation of the eld equations about the Robertson-
Walker solution combined with numerical solvers can indicate whether observed uctuations
in the CMB agree with the theory, and this is an important future direction of research for
34
the YM theory.
Studies of Big Bang Nucleosynthesis (BBN) where the predicted quantities of light el-
ements such as Helium, Lithium, and Deuterium are compared with predicted quantities
can also constrain some theories and several studies of Tensor-Scalar theory have been done
[6][5]. These studies tend to focus on the eects of the scalar eld on nucleosynthesis and
are not directly applicable to the YM theory which has no scalar eld.
While the linear model explains features of the later universe as well as the standard
CDM model [2], a single power law scale factor has proved insucient to explain the early
universe where the power is sharply constrained to around a t
0.55
[11] unless additional
modes of production of deuterium are invoked in the later universe. Because the YM theory
predicts a multi-term polynomial gravitational redshift evolution on top of the Doppler
redshift evolution, however, the theory has a (at least) dual epoch model. Because the
present values of b(t) and c(t) are unknown, however, whether the theory is able to predict
abundances of light elements is left for future work.
V. FUTURE TESTS
General Relativity and the Yang-Mills theory agree for the linearized Schwarzschild metric
predictions and 1PN equations of motion but, they fundamentally diverge at higher order
motion. The 2PN order should be measurable for binary pulsars within the next few decades
as additional data is collected about known systems (the more data collected the smaller
the error bound). It is doubtful that the 2nd order of the Schwarzschild geometery will
be measurable within the Solar system any time soon because the required accuracy would
have to be thousands of times better than current, already extremely sensitive, instruments
provide, at which point signicant and potentially unknown variations in Earths or the
Suns density would make it dicult to predict the measurements. Gravitational wave
detectors such as LIGO and VIRGO may also be capable of detecting much higher order
eects (well into the quadratic or even cubic realm) from inspiraling black holes and other
closely spaced, highly massive objects. Because general relativity and the Yang-Mills theory
no longer agree at this level, these observations will be able to rule out one of the theories.
Another important test is a quantitative measurement of the values of b
0
and c
0
which may
allow a quantitative prediction of the rate of acceleration of the universe, an important
35
prediction of the theory.
Quantum predictions are signicant because the theory, in order to be renormalizable,
requires mass, energy, momentum, and angular momentum to be quantized so that the
coupling constant has zero mass dimension. Without that assumption, the coupling constant
would have negative mass dimension of -2 and the theory would not be renormalizable [38].
The size of the quantum must be exceptionally small, but an experiment may be devised
to detect it similar to how Millikans oil drop experiment detected the indivisibility of the
electron charge. In order to do so the energy of a process or wave would have to be measured
to much greater accuracy than currently possible and shown to be a multiple of a small
quantum.
VI. RELATED WORK
While this paper has focused on the problems with the prevailing theory of gravity,
general relativity, several other theories have been proposed to solve quantum gravity and
explain macroscopic observations. The most important empirical theory is the Lambda-
CDM model which combines the Einstein eld equations with a cosmological constant and
cold dark matter. The eld equations 29-30 do away with the cosmological constant and
challenge some of the assumptions of the model, but the YM theory does not challenge the
prevailing theory of dark matter.
The most prominent theory of quantum gravity is string theory and its derivative M-
theories which, rather than being a simple theory of gravity, is an extensive modication of
physical law positing that all matter is composed of strings, attempting to unify all forces
[38][26]. Although unication is not its purpose, in representing gravity identically to the
other three forces, the theory be unied with them at a high enough energy without requir-
ing any signicant additional physical assumptions such as strings or additional spacetime
dimensions.
The next most signicant model of quantum gravity is loop quantum gravity [26]. Loop
quantum gravity has a dierent approach from that given in this paper. In making the
assumption that gravity has only dieomorphism covariance and is dominated by a metric
geometry, it retains the symmetry of general relativity and adds to its complexity by in-
troducing spin foams, i.e. discretizations of spacetime, in order to avoid blow-ups of the
36
quantum variables. It currently makes no predictions that are testable and, furthermore,
has not been shown to agree with classical general relativity in its entirety.
The main Yang-Mills approach to gravity is conformal gravity [25][18], and is not directly
relevant to the theory in this paper which makes signicantly dierent physical and mathe-
matical assumptions. Conformal gravity is a four derivative Lagrangian theory and predicts
signicantly dierent large scale behavior than either general relativity or Newtonian gravity.
The Macdowell-Mansouri action is also based on the de Sitter group but is not locally de
Sitter invariant, is not a Yang-Mills theory, and is not renormalizable [37].
VII. CONCLUSION
The assumption of dieomorphism, while elegant, is not a necessary component given
current experimental evidence, and, therefore, we are motivated to consider alternatives to
it in a theory of gravity. By assuming a de Sitter symmetry group, this theory not only
agrees with the observations conrming general relativitys predictions but also includes a
renormalizable quantum theory of gravity, a denition of mass as relativistic energy, and
a novel formulation of Yang-Mills theory without a background manifold. It does this
by establishing gravity on the theoretical foundation of the Standard Model. Indeed, the
Standard Model SU(3) SU(2) U(1) is conjectured to be part of a unied, covariant
theory in SU(5) at some energy since SU(5) is the smallest group that can contain all
three. With the addition of the de Sitter group, the Standard Model could be updated to
be SO(4, 1) SU(3) SU(2) U(1). The smallest group to contain this would be SU(10)
with 99 degrees of freedom. Another possibility would be SO(15) which has 105 degrees of
freedom. Either of these might represent a unied eld theory at high enough energy.
The theory, in addition, solves at least two problems with General Relativity:
1. Quantum renormalizability. General Relativity has a non-renormalizable action while
the YM theory has a renormalizable action provided mass and momentum are quan-
tized. Although such quantization has not been measured, the weakness of gravity
suggests that it is extremely small, much smaller than the quantization of charge.
2. Vacuum energy. Although vacuum energy is the predominant explanation for the
source of dark energy, the vacuum energy has a density approaching /x where
37
x = Planck length
3
Planck time [38]. This density is the density the universe
would have if Planck mass black holes were set adjacent to one another, event horizon
to event horizon at every point in space. Clearly, this is not so. It has been suggested
that something cancels the vacuum energy out, but that has not yet been found. By not
coupling to the vacuum energy in its quantum formalism and explaining accelerating
expansion via group theory rather than a constant, the YM theory does not have the
same problem as GR.
The most important prediction is the prediction of the accelerating expansion of the uni-
verse from the de Sitter group. In the context of GR, the de Sitter model for cosmology is ad
hoc, but here it is a fundamental outgrowth of gauge symmetry (not to mention necessary for
renormalizability). Although qualitative, it has the potential to be a quantitative prediction
with additional data to constrain the current conditions of the universe within the context
of this theory, particularly values of b
0
and c
0
. Further conrmations may be obtained from
detection of gravitational waves, observations of supermassive objects such as black holes,
and additional cosmological measurements.
[1] R. Aldrovandi and J. G. Pereira. An Introduction to Geometrical Physics. World Scientic,
1995.
[2] A. Benoit-Levy and G. Chardin. Observational constraints of a milne universe. In Proc. of
the 43rd Rencontres de Moriond, 2008.
[3] A. Benoit-Levy and G. Chardin. A symmetric milne universe: a second concordant universe?
In SF2A 2008, 2008.
[4] D. Clowe and et. al. A direct empirical proof of the existence of dark matter. Astrophysical
Journal Letters, 648:109114, 2006.
[5] A. Coc, K. Olive, J-P. Uzan, and E. Vangioni. Big bang nucleosynthesis constraints on scalar-
tensor theories of gravity. Phys. Rev. D, 73(083525), 2006.
[6] T. Damour and B Pichon. Big bang nucleosynthesis and tensor-scalar gravity. Phys. Rev D,
59(123502), 1999.
[7] A. Einstein. Preuss. Akad. Wiss. Berlin, Sitzber, 154, 1918.
38
[8] C. W. F. Everitt, et al. Gravity probe b: Final results of a space experiment to test general
relativity. Phys. Rev. Lett., 106:221101, May 2011.
[9] J. D. Jackson. Classical Electrodynamics. Wiley, Hoboken, NJ, 1999.
[10] D. C. Joshi, M. P. Benjwal, and S. Kumar. Classical yang mills theory in presence of electric
and magnetic charges. Acta Physica Polonica, B16(10):901908, 1985.
[11] M. Kaplinghat, G. Steigman, and T. P. Walker. Nucleosynthesis in power-law cosmologies.
Phys. Rev. D, 61(103507), 2000.
[12] E. Komatsu, et al. Seven-year wilkinson microwave anisotropy probe (wmap) observations:
Cosmological interpretation. The Astrophysical Journal Supplement Series, 192(2):18, 2011.
URL http://stacks.iop.org/0067-0049/192/i=2/a=18.
[13] M. Kramer, I. H. Stairs, and et. al. Tests of general relativity from timing double pulsar.
Science, 314(5796):97102, 2006.
[14] M. Kutschera and M. Dyrda. Coincidence of age in cdm and milne cosmologies. Acta Phys.
Polonica B, 38(1):215217, 2007.
[15] L. S. Levitt. The spin kinetic energy and intrinsic mass of elementary particles. Lett. Al Nuovo
Cimento, 36:167170, 1983.
[16] D-E Liebscher. Cosmology. Springer, Heidelberg, 2005.
[17] H. A. Lorentz. Electromagnetic phenomena in a system moving with any velocity smaller than
that of light. Proceedings of the Royal Netherlands Academy of Arts and Sciences, 6:809831,
1904.
[18] Philip D. Mannheim. Alternatives to dark matter and dark energy. Progress in Particle and
Nuclear Physics, 56(2):340 445, 2006.
[19] C. W. Misner, K. S. Thorne, and J. A. Wheeler. Gravitation. W. H. Freeman, San Francisco,
1973.
[20] H. M uller, A. Peters, and S. Chu. A precision measurement of the gravitational redshift by
the interference of matter waves. Nature, 463:926, 2010.
[21] NASA. Wmap mission results, 2008. Updated: March 7, 2008. Accessed: March 18, 2009.
http://map.gsfc.nasa.gov/news.
[22] J. Norton. General covariance, gauge theories and the Kretschmann objection. Cambridge
UP, Cambridge, UK, 2003.
[23] T. Ohlsson. Relativistic Quantum Physics: From Advanced Quantum Mechanics to Introduc-
39
tory Quantum Field Theory. Cambridge UP, 2011.
[24] P. C. Peters and J. Matthews. Gravitational rational from point masses in a keplerian orbit.
Phys. Rev., 131(1):435440, 1963.
[25] R. J. Riegert. Classical and Quantum Conformal Gravity. PhD thesis, UCSD, 1986.
[26] C. Rovelli. Quantum Gravity. Cambridge UP, Cambridge, UK, 2004.
[27] V. Rubin. Rotation of the andromeda nebula from a spectroscopic survey of emission regions.
Astrophysical J., 159:379, 1970.
[28] M. Sethi, A. Batra, and D. Lohiya. Comment on observational constraints on power-law
cosmologies. Phys. Rev. D, 60(108301), 1999.
[29] Gerardus t Hooft. 50 Years of Yang-Mills theory. World Scientic, Singapore, 2005.
[30] K. S. Thorne. Multipole expansions of gravitational radiation. Rev. Mod. Phys., 52:299324,
1980.
[31] R. Wald. General Relativity. U. Chicago UP, Chicago, 1984.
[32] S. Weinberg. Cosmology. Oxford UP, Oxford, UK, 2008.
[33] J. M. Weisberg and J. H. Taylor. Relativistic binary pulsar b1913+16: Thirty years of obser-
vations and analysis. ASP Conf. Series, 328:2531, 2005.
[34] P. S. Wesson. Is mass quantized? Mod. Phys. Lett. A, 19:19952000.
[35] C. Will. Theory and experiment in gravitational physics. Cambridge UP, Cambridge, UK,
1993.
[36] C. Will. The confrontation between general relativity and experiment. Living Review: Rela-
tivity, 9, 2006.
[37] Derek K Wise. Macdowellmansouri gravity and cartan geometry. Classical and Quantum
Gravity, 27(15):155010, 2010.
[38] A. Zee. Quantum Field Theory in a Nutshell. Princeton UP, Princeton, 2003.
[39] F. Zwicky. On the masses of nebulae and of clusters of nebulae. Astrophysical J., 86:217,
1937.
APPENDIX A: THE DE SITTER GROUP LIE ALGEBRA
In addition to rotations and boosts, all physical theories are translation invariant. This
conrms the assumption that there is no special place in the universe. By Noethers
40
theorem, translation invariance causes momentum to be conserved (including energy). A
translation by an amount v

is achieved by a 55 matrix, T = v

such that V

= (V
t
, V
i
)
are four generators. Given a 4-vector u

if w = (u

, 1), then w

= Tw = (u

+v

, 1). It is well
known, however, that prior to the introduction of Lorentz covariance, mechanical theories
such as Newtonian gravity were Galilean invariant, R
3
SO(3) rather than Lorentz invariant,
SO(3,1). A reasonable extension of the Poincare group, then, is from R
(3,1)
SO(3,1) to
SO(4,1) assuming that the radius of curvature in the fth dimension is large so that it
appears to be Poincare invariant for small rotations.
In the de Sitter group, x
4
is a fourth spatial dimension and has ordinary rotations with
respect to the other three spatial dimensions and boosts with respect to time. The anti-de
Sitter group has x
4
as a second time dimension. Let V

be rotations/boosts in the x

-x
4
plane.
The Poincare group has the covariant Lie algebra,
[V

, V

] = 0 (A1)
[M

, P

] = i(

) (A2)
[M

, M

] = i(

), (A3)
for

the Minkowski metric where U = exp[

i
2

] is a Lorentz operation and U =

exp[ia

V

] is the translation operation [23]. The only alteration that the de Sitter group
makes is that the translation operators now generate a Lorentz rotation or boost,
[V

, V

] = iM

, (A4)
in which case
U = exp[
i
2

+ ia

V

],
is the de Sitter operation [1].
APPENDIX B: DERIVATION OF THE 1PN POTENTIAL
The potential at a spacetime point (t, x) in the given inertial frame can be solved by the
integral equations [36], h

= (G

)/2:
h

(t, x) =
_
C
d
3
x

(t [x x

[, x

)
[x x

[
, (B1)
41
where ( is the past-light cone and bodies have small polarized acceleration and angular
momentum is small.
The integral equation may be solved by iteration in which a trial potential h

= 0 is
inserted into the integrand and the integral solved to arrive at a new potential h

which is
then reinserted to achieve a potential h

and so on until the required accuracy is achieved.

In order to derive the 1PN equations of motion general relativity requires two iterations but
here we only require one, meaning that the 1PN equations for this theory are linear.
Let the baryon density be
0
, and the Newtonian potential is given by,
U =
_
d
3
x

0
(t, x

)
[x x

[
. (B2)
Let U
2
and v/c where v is the average velocity and c is the speed of light in
vacuum. The rst post-Newtonian corrections require that h
00
be given up through order

4
, h
0j
through order
3
and h
ij
through
2
, correcting the Newtonian order,
h
00
= U, h
0j
= 0, h
ij
= 0. (B3)
The G-eld derives from the equations of Sec. II C above and the stress-energy-
momentum tensor for baryon dust. Each baryon particle or compact spherical mass (such
as a neutron star or black hole) p has a stress-energy-momentum tensor in its rest frame in
a spherical coordinate system with the particle at the origin,
T
00
= m
p

3
(x), T
rr
= m
p

3
(x). (B4)
The latter is related to the potential h
rr
which has a transformation,
h
ij
= h
rr
x
i
x
j
[x[
2
, (B5)
which is
h
ij
= m
p
(x
i
x

i
)(x
j
x

j
)
[x x

[
3
(B6)
for a particle at x

. To obtain the potential for all the dust, sum all the potentials for all
the baryons by superposition of the potentials,

h
ij
=
_
d
3
x

0
(t, x

)(x
i
x

i
)(x
j
x

j
)
[x x

[
3
+ O(
4
). (B7)
By a non-Abelian gauge transformation to order [19],
h
ij
=

G
ij
+
i

j
, (B8)
42
where

and
=
_
d
3
x

0
(t, x

)[x x

[, (B9)
we have
h
ij
=
ij
U + O(
4
). (B10)
The next two parts of the G-eld, h
0j
and h
00
, require particle velocities and time dilation
to be included. The current tensors, T

(p) of each baryon p, are in the rest frame, but,

because the bodies are moving with respect to the observer at innity and subject to the
gravitational elds of one another, the tensors must be (1) boosted from comoving frame
x

to the observers frame, x

, and (2) subject to a non-Abelian gauge transformation from

the zero gravitational eld (free fall) in the rest frame of the body to the gravitational eld
in the rest frame of the observer,
T

(p) =
d x

dx

d x

dx

(p) + h

m
p

3
(x x
p
), (B11)
where

= d x

/dx

is a Lorentz transformation matrix for a boost density v(t, x) (c =

1)[19] and h

is the G-eld not including the contribution of the mass at that point. These
two transformations are equivalent to the two step, boost-and-coordinate-transform method
of general relativity [19] which reects the eects of both velocity and gravitational elds
on measurements of stress-energy-momentum tensors.
For a baryon p with mass m
p
at x

with a rest frame potential

h

and velocity v (c = 1),

h
0j
=

h
00
x
0
x
0
x
0
x
j
+

h
ij
x
i
x
0
x
j
x
j
, (B12)
where

h
ij
=
m
p
(x
i
x

i
)(x
j
x

j
)
[x x

[
3
. (B13)
To order this becomes,

h
0j
=
m
p
[x x

[
_
v
j
+
[(x x

) v](x
j
x

j
)
[x x

[
2
_
+ O(
5
). (B14)
Under the same gauge transformation as above,
h
0j
=

h
0j

0

j
, (B15)
and summing over all baryons, the potential is,
h
0j
= 2V
j
+ O(
5
), (B16)
43
where
V
j
=
_
d
3
x

0
(t, x

)v
j
(t, x

)
[x x

[
. (B17)
The time-time eld is,

h
00
= U + 2 + + O(
6
), (B18)
where
(t, x) =
_
d
3
x

0
(t, x

)(v
2
+ U)
[x x

[
, (B19)
and
(t, x) =
_
d
3
x

0
(t, x

)
[x x

[
_
[(x x

) v]
2
[x x

[
2
v
2
_
. (B20)
The gauge transformation, h
00
=

G
00
+
2
0
(which we have already applied to the other
potentials and hence must carry over to this one), eliminates the last term, and gives
h
00
= U + 2 + O(
6
). (B21)
Missing is a term of order U
2
that appears in metric PPN formalism which indicates the
nonlinearity in the superposition of the potential. In GR, however, this term is an artifact
of the choice of isotropic coordinates and vanishes with a change of coordinates.
In general relativity, the change of coordinates, r = r(1 + M/2 r)
2
, switches the
Schwarzschild solution to isotropic coordinates [19],
ds
2
=
_
1 M/2 r
1 + M/2 r
_
2
dt
2
+ (1 + M/2 r)
4
[d r
2
+ r
2
(d
2
+ sin
2
d
2
)]. (B22)
Because the change in coordinates, however, depends on the potential itself, this introduces
a nonlinearity that previously did not exist in the solution. Like the Schwarzschild solution
in Schwarzschild coordinates, the Yang-Mills theory 1PN solution does not contain any
nonlinearity because it is not in this nonlinear coordinate system. The Yang-Mills spherically
symmetric solution given in spherical coordinates,
h
00
= h
rr
= M/r, (B23)
under the same change of variables, becomes to 1PN order,
h
00
= M/ r 2(M/ r)
2
+ O(
6
), (B24)
h
0i
= h
i0
= O(
5
) (B25)
44
and
h
ij
=
ij
M/ r + O(
4
). (B26)
While one could make a change of the at spacetime coordinates to bring the YM potential
in line with the GR metric, there is no purpose in doing so. Since both GR and the Yang-
Mills theory are generally covariant, changing coordinates does not change predictions. It
does, however, require that observations be input in the appropriate coordinate system since
the Schwarzschild r is not the same as the 1PN r. In General Relativity, the Schwarzschild
r is simply a coordinate not a distance. Meanwhile, because of the potential dependent
metric in isotropic coordinates, r is just a coordinate in both theories and does not refer
to distance directly.
The remaining equations (which can be derived from the eld equations) are the equation
of motion. The continuity equations (38) to 1PN order are the four equations,

= 0, (B27)
each of which is the same as the electromagnetic continuity equation. Linearized general
relativity has the same problem [31] and care needs to be taken that the nonlinear, gauge
covariant equations of motion (38) be used instead. For N-body problems, which are gener-
ally the ones that gravitation deals with, however, the geodesic equation is sucient. This
is because the Yang-Mills theory satises the Strong Equivalence Principle (SEP) and inter-
nal structure does not aect the motion of bodies. Therefore, any spherical, compact body
moves as a test particle.
45