The Accelerating Universe: A Gravitational Explanation

J. W. Moffat
Department of Physics, University of Toronto, Toronto, Ontario M5S 1A7, Canada

arXiv:astro-ph/0108201 11 Aug 2001

Abstract
The problem of explaining the acceleration of the expansion of the universe
and the observational and theoretical difficulties associated with dark matter
and dark energy are discussed. The possibility that GR does not correctly describe the large-scale structure of the universe is considered and an alternative
gravity theory is proposed as a possible resolution to the problems.

e-mail: [email protected]

Introduction

The recent surprising observational discovery that the expansion of the universe is
accelerating [1] has led to an increasing theoretical effort to understand this phenomenon. The attempt to interpret the data by postulating a non-zero positive
cosmological constant is not satisfactory, because it is confronted by the two serious
issues of why the estimates from the standard model and quantum field theory lead
to preposterously large values of the cosmological constant, and the coincidence of
matter and dark energy dominance today [2].
If we simply postulate a repulsive force in the universe associated with a charge
density, then we might expect that this force could be responsible for generating
the acceleration of the universe. However, for a homogeneous and isotropic universe
the net charge density would be zero, although for a finite range force with a small
mass there will exist a non-zero charge density [3]. The effect of a Maxwell-type
force would be to lower or raise the total energy, leaving the form of the Friedmann
equation unchanged. Thus, we would still have to invoke exotic forms of energy with
an equation of state, p = w, where w is negative and violates the positive energy
theorems. For a non-zero cosmological constant w = 1.
In addition to the dark energy problem, we are still confronted with the puzzle of
dark matter. Any observational detection of a dark matter candidate has eluded us
and the fits to galaxy halos using dark matter models are based on several parameters
depending on the size of the galaxy being fitted. The dark matter model predictions
disagree with observable properties of galaxies [4].
Another problem with the dark energy hypothesis is the serious challenge to
present day particle physics and string theory from the existence of cosmological
horizons, which arise in an eternally accelerating universe [5]. Several resolutions
1

of this problem have been proposed [6] but a cosmological horizon does produce a
potentially serious crisis for modern particle physics and string theory.
Challenging experimental results are often the precursors of a shifting of scientific
paradigms. We must now entertain the prospect that the discovery of the mechanism
driving the acceleration of the universe can profoundly change our description of the
universe.
Given the uneasy tension existing between observational evidence for the acceleration of the universe and the mystery of what constitutes dark matter and dark
energy, we are tempted to reconsider the question of whether Einsteins gravity theory (GR) is correct for the large scale structure of the universe. It agrees well with
local solar system experimental tests and for the data obtained for observations of
the binary pulsar PSR 1913+16. However, this does not preclude the possibility
of a breakdown of the conventional Einstein equations for the large-scale structure
of the universe. The standard GR cosmological model agrees well with the abundances of light elements from big bang nucleosynthesis (BBN), and the evolution
of the spectrum of primordial density fluctuations, yielding the observed spectrum
of temperature anisotropies in the cosmic microwave background (CMB). Also, the
age of the universe and the power spectrum of large-scale structure agree reasonably
well with the standard cosmological model. However, it could be that additional
repulsive gravitational effects from an alternative gravity theory could agree with all
of the results in the early universe and yet lead to significant effects in the present
universe accounting for its acceleration [7].
When contemplating alternative gravity theories, one is impressed with the mathematical and physical robustness of GR. It is not easy to change the structure of GR
without running into consistency problems. A fundamental change in the predictions
of the observational data will presumably only come about from a non-trivial alteration of the mathematical and geometrical formalism that constitutes GR. From the
cosmological standpoint, such theories as Jordan-Brans-Dicke [8] theories of gravity
will not radically change the Friedmann equation in the present universe. Recent
developments in brane-bulk cosmological models [9] have led to alterations of the
Friedmann equation but only for the very early universe corresponding to high energies.
This tempts us to return to a physically non-trivial extension of GR called the
nonsymmetric gravity theory (NGT). This theory was extensively studied over a
period of years, and a version of the theory was discovered that was free of several
possible inconsistencies such as ghost poles, tachyons, exotic asymptotic behaviour
and other instabilities [10, 11, 12, 13]. Further research is needed to fully understand
such problems as its Cauchy development and the deeper meaning of the basic gauge
symmetries underlying the theory.
As we shall see in the following, NGT can describe the current data on the
accelerating universe and the dark matter halos of galaxies, gravitational lensing
and cluster behaviour, as well as the standard results such as BBN, the solar system
tests and the binary pulsar PSR 1913+16, without invoking the need for dominant,
2

exotic dark matter and dark energy.

NGT Action and Field Equations

The nonsymmetric g and are defined by [10, 11, 12, 13, 14]:
1
g() = (g + g ),
2

1
g[] = (g g ),
2

(1)

and
= () + [] .

(2)

The contravariant tensor g is defined in terms of the equation

g g = g g = .

(3)

The Lagrangian density is given by

LN GT = L + LM ,

(4)

where

1
L = g R (W ) 2 g 2 g g[]
4
1
g W W + g J[ ] ,
6

(5)

and LM is the matter Lagrangian density (G = c = 1):

LM = 8g T .

(6)

W
,
) W
W
+ W
+ W,
(W,
R (W ) = W,
2

(7)

Here, g = gg , g = Det(g ), is the cosmological constant and R (W ) is

the NGT contracted curvature tensor:

defined in terms of the unconstrained nonsymmetric connection

2

W
= W ,
3
where

W = (W
W
).
2

Eq.(8) leads to the result

= [] = 0.
3

(8)

The contracted tensor R (W ) can be written as

2
R (W ) = R () + W[,] ,
3
where
R () = ,
The term in Eq.(5):


1
(), + (), + () .
2

g J[ ] ,

(9)

contains the Lagrange multiplier fields and a source vector J .

A variation of the action
Z
S = d4 xLNGT
yields the field equations in the presence of matter sources:

G (W ) + g + S = 8(T + K ),
1
g[] , = g()W ,
2

g , + g W + g W
g W

2
1

+ g W[]
+ (g()W g() W ) = 0.
3
6
1
Here, we have G = R 2 g R, and

(10)
(11)

(12)

1
1
1
1
S = 2 (g[] + g g [] g[] + g [] g g ) (W W g g W W ). (13)
4
2
6
2

Moreover, the contribution from the variation of (9) with respect to g and g
is given by
1
1
K = [J[] g (g []J[] )].
(14)
8
2
The variation of yields the constraint equations
g[] J = 0.

(15)

We have not varied the source vector J .

If we use (15), then (14) becomes
K[] =

1
J[ ] .
8

(16)

If we specify J to be J = (0, 0, 0, J0 ), then (15) corresponds to the three

constraint equations
g[i0] = 0.
(17)
4

After eliminating the Lagrange multiplier field from the field equations (10),
we get
G() (W ) + g() + S() = 8T() ,
(18)
 J (G[] (W ) + g[] + S[] ) = 8 J T[] ,

(19)

where 
is the Levi-Civita symbol.
The generalized Bianchi identities
[g G () + g G ()], + g , G = 0,

(20)

give rise to the matter response equations

g T , + g T , + (g, + g, g, )T = 0.

(21)

A study of the linear approximation has proved that the present version of NGT
described above does not possess any ghost poles or tachyons in the linear approximation [12]. This cures the inconsistencies discovered by Damour, Deser and
McCarthy in an earlier version of NGT [15]. Moreover, the instability discovered
by Clayton [16] for both massless and massive NGT in a Hamiltonian formalism,
associated with three of the six possible propagating degrees of freedom in the skew
symmetric sector, is eliminated from the theory. This is implemented in the NGT
action by the covariant constraint equations (15).

Cosmological Solutions

For the case of a spherically symmetric field, the canonical form of g in NGT is
given by

0
0
w
0

f sin
0

,
(22)
g =
0
f sin sin2 0
w
0
0

where , , and w are functions of r and t. We have

g = sin[( w2 )( 2 + f 2 )]1/2.
For a comoving coordinate system, we obtain for the velocity vector u which satisfies
the normalization condition g() u u = 1:
1
u0 = ,

ur = u = u = 0.

(23)

From Eq.(17), we get w = 0 and only the g[23] component of g[] is different from
zero. The vector W can be determined from
2
W = s g[] , ,
g
5

(24)

where s g () = . For the spherically symmetric field with w = 0, it follows from

(17) and (24) that W = 0.
The energy-momentum tensor for a fluid is
T = ( + p)u u pg + B [] ,

(25)

where B [] is a skew symmetric source tensor.

We can prove from a Killing vector analysis that for a homogeneous and isotropic
universe massless NGT requires that f (r, t) = 0 [14]. It follows that for massless
NGT all strictly homogeneous and isotropic solutions of NGT cosmology reduce to
the Friedmann-Robertson-Walker (FRW) solutions of GR. For the case of massive
NGT, it will be possible to obtain strictly homogeneous and isotropic solutions. In
the following, we shall solve the field equations for a spherically symmetric inhomogenous universe and then approximate the solution by assuming that the inhomogeneities are small. We expand the metric g() as
HI
g() = g()
+ g() ,

(26)

HI
where g()
denotes the homogeneous and isotropic solution of g() and g() are
small quantities which break the maximally symmetric solution with constant Riemannian curvature. We shall simplify our calculations by assuming that the density
and the pressure p only depend on the time t. Moreover, we assume that the mass
parameter 0 and we neglect any effects due to the antisymmetric source tensor
B [] .
It is assumed that a solution can be found by a separation of variables

(r, t) = h(r)R2 (t),

(r, t) = r2 S 2 (t).

(27)

From the field equations, we get

S
1
R
= Zr,
R S
2

(28)

where R = R/t and Z is given by

Z=

0f 2
f 0 f f0 3f 0 f
0f 2 5
0f 2 2f

+
2
3
2 4
2 3
3

2 2
+

f
f 0 2 0 f f
+
.
2 2
3

(29)

Let us assume that Z 0, then from (28) we find that R S and the metric
line-element takes the form


ds2 = dt2 R2 (t) h(r)dr2 + r2 (d2 + sin2 d2 ) .

6

(30)

From the conservation laws, we get

p =

1 3
[R ( + p)].
R3 t

(31)

If we assume that  f , then the equations of motion become [13, 14]

2b(r) + R(t)R(t)
+ 2R 2 (t) R2 (t)W (r, t) = 4R2 (t)[(t) p(t)],

(32)

1
4 2

R (t)[(t) + 3p(t)],
R(t)R(t)
+ R2 (t)Y (t) =
3
3

(33)

where
2b(r) =

h0(r)
.
rh2 (r)

The functions W and Y are given by

W =

2 5 02f 2
0 0f 2 00f 2 f
f
f

+
+

22 3
3
2 3
2 4
2 2

0 f f 0
f f 00 4f f 0 0
3f 02

+
,
22 2 2
3
2 2
f f f
2 5 2f 2 3f2 4f
f

+
2.
Y = 3

2 4
2 2
3

Within our approximation scheme, W and Y can be expressed in the form

W =

(34)
(35)

f
h0 f 2
2f 2
2R 2 f 2
10f 2
h0 f f 0
Rf

+
+

h2r5 R6 hr6 R6
r 4 R6
hr6 R6 r4 R5 2h2 r4 R6
0

8f f 0
3f 2
f f 00
,
4 6 5 6+
h4 R
hr R
2hr4 R6
2 10R 2 f 2
f
2(R 2 + RR)f
3f2
8Rf
f f
Y =

.
r 4 R6
r 4 R6
2r4 R4
r 4 R5
r 4 R4
by adding (32) and (33), we get
Eliminating R

where
From (33) we obtain

(36)
(37)

8 2
R 2 + b =
R + QR2 ,
3

(38)

1
1
Q = W Y.
2
6

(39)

= 4 R( + 3p) + 1 RY.
R
3
3

(40)

We can write Eq.(38) as

H2 +

b
= H 2 ,
2
R
7

(41)


where H = R/R,
= M + Q ,
and
M =

8
,
3H 2

Q =

(42)
Q
.
H2

(43)

If b = 0, then we get = 1 and

H2 =

8
+ Q.
3

(44)

The line element now takes the approximate form of a flat, homogeneous and
isotropic FRW universe
ds2 = dt2 R2 (t)[dr2 + r2 (d2 + sin2 d2 )].

(45)

Accelerating Expansion of the Universe

> 0 when Y > 4( + 3p). If we assume that there is a

It follows from (40) that R
solution for Q and Y , such that they are small and constant in the early universe,
then we will retain the good agreement of GR with the BBN era with rad 1/R4
and M 1/R3 . As the universe expands beyond the BBN era at the temperatures,
T 1 MeV-60 kev, then Q begins to increase and reaches a slowly varying value
with 0Q 0.7 and 0M 0.3, where 0M and 0Q denote the present values of M
and Q , respectively. These values can fit the combined supernovae, cluster and
CMB data [1].
We observe from (36) and (37) that the dependence of Q and Y as the universe
expands is a function of the behaviour of R and f and their derivatives. If f grows
sufficiently with R as t increases, then Q and Y can dominate the matter contribution
M as the universe evolves towards the current epoch. A detailed solution of the
field equations is required to determine the dynamical behaviour of R, f , Q and Y .
However, we can obtain some knowledge of the qualitative behaviour of Y and Q
by assuming that f f 0 0 and h = 1. Then, from (36) and (37) we obtain
Y
and

2f 2
4R 2 ),
(RR
r 4 R6


(46)


f2
12
.
+ 7R 2 RR
(47)
4
6
3r R r2
If, as the universe expands, the behaviour of f is f Ra with a 2, then we can
> 0 and R will increase as we approach the present epoch
satisfy Y > 0, Q > 0, R
corresponding to an accelerating expansion.
We can explain the evolution of Hubble expansion acceleration within NGT,
without violating the positive energy conditions. Both and p remain positive
Q

throughout the evolution of the universe. There is no need for a dark energy and a
cosmological constant. Thus, we avoid having to explain the unnatural and mysterious coincidence of matter and dark energy domination. The Q contribution to the
expansion of the universe increases at a slow rate up to a constant value today with
Y > 4 (p 0), and can then decrease to zero as the universe continues to expand,
avoiding an eternally accelerating universe. During this evolution, the cosmological
constant = 0. It is then possible to avoid the existence of a cosmological horizon
and the problems it produces for quantum field theory and string theory [5].

NGT and Dark Matter

Galaxy dynamics observations continue to pose a problem for gravitational theories

and cosmology. The data for spiral galaxies are in sharp contradiction with Newtonian dynamics, for virtually all spiral galaxies have rotational velocity curves which
tend towards a constant value. The standard assumption is that dark matter exists
in massive, almost spherical halos surrounding galaxies. The standard hypothesis
is that about 90% of the mass is in the form of dark matter and dark energy and
this explains the flat rotational velocity curves of galaxies. This explanation is not
economical, for it requires three or more parameters to describe different kinds of
galactic systems and no satisfactory model of galactic halos exists [4].
A possible explanation of the galactic rotational velocity curves problem has been
obtained in NGT [17]. A derivation of the motion of test particles yields the total
radial acceleration experienced by a test particle in a static spherically symmetric
gravitational field for r 0.2 kpc, due to a point source (we reinsert G and c) [19]:


r
GM
Cc2 exp(r/r0 )
1+
,
a(r) = 2 + 2
2
r
r0
r
r0

(48)

where , C and r0 =
1/ are constants which remain to be fixed.
We choose C M and fix to give
a(r) =

exp(r/r0 )
r
G M
,
+
G
1+
0 M
2
2
r
r
r0

(49)

where G is defined to be the gravitational constant at infinity



G = G 0 1 +

M0
M

(50)

and G0 is Newtons gravitational constant.

These formulas were applied to explain the flatness of rotation curves of galaxies,
as well as the Tully-Fisher law [20], L v 4, where v is the rotational velocity of a
galaxy and L is the luminosity. A derivation of v gives


3G0 L
v2 =
1+
r



L0
r
[1 exp(r/r0 ) 1 +
] ,
L
r0
9

(51)

where L0 = 3M0 . For distances less than 0.5 4 kpc, the standard Newtonian law
of gravity will apply. For r0 = 25 kpc and L0 = 250 1010 L , an excellent fit to
spiral galaxies was found [17, 18]. Moreover, a good fit to the Tully-Fisher law was
also obtained.
Consider now the giant spiral galaxy M31 and our Galaxy in the local group. The
center of M31 is approaching the center of the Galaxy at a velocity 119 km/sec.
The total mass of the local group
should be very large, namely, the mass-to-light

ratio should be 100 M/L . This big ratio is normally explained using the dark

matter hypothesis. The distance between M31 and the Galaxy is 700 kpc, so the
additional exponential force is vanishingly small, but what is left is the renormalized
gravitational constant. Thus, the gravitational acceleration becomes
G0 M
,
r2

a(r) =

(52)

where M 17M and from the observed mass-to-light ratio:



M
M
100
L
L
we predict

(53)

M
M
.
(54)
6
L
L 
This agrees with the estimated ratio for luminous matter without using the dark
matter assumption.
Gravitational lensing effects can also be accounted for in NGT. We find for the
angle of deflection , obtained in the post-Newtonian approximation


4G0 1 +
=

M0 /M

c2 R

(55)

where M is the mass of the galaxy and R is the distance between the galaxy center
and the deflected ray. This prediction is close to the one obtained from the dark
matter hypothesis.
If we calculate the acceleration expected in our solar system, we obtain
 
a aNGT
1 M0 r 2
.
(56)
a =

aNewton
2 M r0
For solar and terrestrial experiments, we find for r0 25 kpc, a < 1013 , which is
too small a deviation from Newtons law to be detected with current experiments.
We must still acount for the estimated value of M 0.33 0.053 [21]. Measurements of total baryon density give B h2 = 0.020 0.001 [22]. For h 0.7 ,we get
B 0.04. The usual hypothesis states that cold dark matter particles contribute
CDM 0.3. We do expect that there is some dark matter in the universe in the
10

form of dark baryons and neutrinos with non-vanishing mass ( 10% ). It remains
to be seen whether an alternative gravity theory such as NGT can provide an explanation for the discrepancy between visible baryon matter and dark matter. At
the era of structure formation, the contribution of Q could be of order Q 0.3,
growing to its present day value of 0Q 0.95 so that 0 1. NGT would then
be required to explain the formation of galaxy structure without CDM. These are
issues that require further investigation.
Acknowledgments
This work was supported by the Natural Sciences and Engineering Research
Council of Canada.

References
[1] S. Perlmutter et al. Ap. J. 483, 565 (1997), astro-ph/9608192; A. G. Riess,
et al. Astron. J. 116, 1009 (1998), astro-ph/9805201; P. M. Garnavich, et al.
Ap. J. 509, 74 (1998), astro-ph/9806396; S. Perlmutter et al. Ap. J. 517,
565 (1999), astro-ph/9812133; A. G. Riess, et al. to be published in Ap. J.,
astro-ph/0104455.
[2] R. R. Caldwell, R. Dave, and P. J. Steinhardt, Phys. Rev. Lett. 80, 1582
(1998), astro-ph/9708069.
[3] M. Brisudova and R. P. Woodard, gr-qc/0105072 v2.
[4] J. A. Sellwood and A. Kosowsky, astro-ph/0009074.
[5] W. Fischler, A. Kashani-Poor. R. McNees and S. Paban, hep-th/0104181;
S. Hellerman, N. Kaloper, and L. Susskind, JHEP 0106 (2001) 003, hepth/0104180.
[6] J. W. Moffat, hep-th/0105017 v2; J. M. Cline, hep-th/0105251 v2; E. Haylo,
hep-ph/0105216; C. Kolda and W. Lahneman, hep-th/0105300; C. Deffayet,
G. Dvali and G. Gabadadze, hep-th/0105068; J. Ellis, N. E. Mavromatos, and
D. V. Nanapolous, hep-th/0105206, Xiao-Gang He, hep-th/0105005 v2.
[7] S. M. Carroll, astro-ph/0107571; S. M. Carroll, and M. Kaplinghat, astroph/0108002.
[8] S. Weinberg, Gravitation and Cosmology: Principles and Applications of the
General Theory of Relativity, John Wiley and Sons, New York, 1972.

11

[9] P. Binetruy, C. Deffayet and D. Langlois, Nucl. Phys. B565, 269 (2000), hepth/9905012; D. J. H. Chung and K. Freese, Phys.Rev. D61 61 (2000) 023511,
hepph/9910235; C. Csaki, M. Graesser, C. Kolda and L. Terning, Phys. Rev.
Lett. B462, 34 (1999), hep-ph/9906513, K. Maeda, astro-ph/0012313; G.
Huey and J. E. Lidsey, astro-ph/0104006.
[10] J. W. Moffat, Phys. Letts. B 335, 447 (1995); J. W. Moffat, J. Math. Phys.
36, 3722 (1995); Erratum, J. Math. Phys. 36, 7128 (1995).
[11] J. W. Moffat, J. Math. Phys. 36, 5897 (1995).
[12] J. W. Moffat, gr-qc/9605016 v3.
[13] J. W. Moffat, astro-ph/9510024 v2.
[14] J. W. Moffat, astro-ph/9704300.
[15] T. Damour, S. Deser, and J. McCarthy, Phys. Rev. D47, 1541 (1993).
[16] M. A. Clayton, Class. Quant. Grav. 13, 2851 (1996), gr-qc/9603062; J. Math.
Phys. 37, 395 (1996), gr-qc/9505005; Int. J. Mod. Phys. A12, 2437 (1997),
gr-qc/9509028; Massive Nonsymmetric Gravitational Theory: A Hamiltonian
Approach, Ph.D. thesis, University of Toronto, 1996.
[17] J. W. Moffat and I. Yu. Sokolov, Phys. Lett. B378, 59 (1996), astroph/9509143 v3.
[18] Fits to more spiral galaxies and clusters have been obtained that agree well
with the data, private communication, I. Yu. Sokolov.
[19] J. Legare and J. W. Moffat, Gen. Rel. and Grav. 27, 761 (1995); grqc/9412009; gr-qc/9509035.
[20] R. B. Tully and J. R. Fisher, Astr. Ap. 54, 661 (1977).
[21] M. S. Turner, Physica Scripta T85, 210, 2000.
[22] S. Burles, K. M. Nollett, and M. S. Turner, ApJ, 552, L1; C. Pryke, et al.,
astro-ph/0104490; C. B. Netterfield et al., astro-ph/0104460.

12