PHYSICAL REVIEW D 94, 064023 (2016)

Cosmological self-gravitating fluid solutions of shape dynamics

Daniel C. Guariento1,2,* and Flavio Mercati2,†
1
Department of Physics and Astronomy, University of Waterloo, Waterloo, Ontario N2L 3G1, Canada
2
Perimeter Institute for Theoretical Physics, 31 Caroline Street North, Waterloo,
Ontario N2L 2Y5, Canada
(Received 21 June 2016; published 9 September 2016)
Shape dynamics is a 3D conformally invariant theory of gravity that possesses a large set of solutions in
common with general relativity. When looked at closely, these solutions are found to behave in surprising
ways; in order to probe the fitness of shape dynamics as a viable alternative to General Relativity one must
find and understand increasingly more-complex, less-symmetrical exact solutions on which to base
perturbative studies and numerical analyses to compare them with data. Spherically symmetric exact
solutions have been studied, but only in a static vacuum setup. In this work we construct a class of
time-dependent exact solutions of Shape Dynamics from first principles, representing a central
inhomogeneity in an evolving cosmological environment. By assuming only a perfect fluid source in a
spherically symmetric geometry, we show that this fully dynamic nonvacuum solution satisfies in all
generality the Hamiltonian structure of shape dynamics. The simplest choice of solutions is shown to be a
member of the McVittie family.

DOI: 10.1103/PhysRevD.94.064023

I. INTRODUCTION for any specific application of the theory. The simplest class
of solution are the homogeneous ones, which have been
Shape dynamics (SD) is a theory of gravity locally
studied in Refs. [4,5]. Giving up homogeneity, we can
identical to the Arnowitt-Deser-Misner (ADM) formulation
assume spherical symmetry in order to make the
[1] of general relativity (GR) in a constant-mean-extrinsic-
Hamiltonian calculable, but in that case we also need to
curvature (CMC) foliation. In such a foliation the
add some matter source, to have nontrivial shape degrees of
physical degrees of freedom of the gravitational field are
freedom (any spherically symmetric manifold is confor-
3-dimensionally conformally invariant [2–4]. SD consists
mally flat, so its shape configuration space is just a point).
of taking this symmetry as fundamental, and requiring that
The simplest thing to do is to add a pair of thin shells of dust
the 3-dimensional conformal geometry of each CMC slice
to a spherical universe [4,6], a procedure which leads to a
be regular. This gives rise to solutions that may differ from
very nontrivial system. In this paper, we are interested in
those of GR, in which the fundamental requirement is
coupling SD to a less singular kind of matter source: a
that of regularity of the pseudo-Riemannian geometry of
spherically symmetric perfect fluid. By not assuming any
4-dimensional spacetime.
details of the relation between density and pressure of this
The relational principles on which SD is based [4] only
fluid, we will be able to solve all the constraints of SD
require that its solutions be generated by a Hamiltonian that
and get a class of exact solutions, which will be a very
is invariant under 3D diffeomorphisms and conformal
valuable starting point for further perturbative analyses.
transformations of the spatial metric. This allows for a
Furthermore, by borrowing a familiar assumption in GR,
large class of Hamiltonians, generating inequivalent
namely, that the Weyl part of the Hawking-Hayward
dynamics; among these we have to select one based on
quasilocal mass takes the form that one expects in the
consistency with observations [3,4]. The value of this
presence of a (cosmological) black hole, or central mass,
Hamiltonian at each point of the phase space of ADM
we are able to derive one particular exact solution which is
variables can be determined only by solving a quasilinear
known in GR as the McVittie metric. Such a metric
elliptic differential equation, whose result depends non- describes a cosmological black hole in the presence of a
locally on the ADM variables. This equation obviously fluid, and it is interesting to observe that it is also a solution
cannot be solved exactly everywhere in phase space, and, of SD. We believe that the simplicity of the McVittie metric
therefore, for practical applications, we need to compute it will make it a very interesting playground for SD, in
as a perturbative expansion. But any perturbative approach particular for the understanding of its physical predictions
needs an exact solution to expand around, and for this regarding black holes. The McVittie metric, in fact, is a
reason exact solutions of SD are an invaluable starting point nonstatic cosmological solution of SD that has a central
concentration of mass which we can identify with a black
*
dguariento@perimeterinstitute.ca hole. The other candidate black hole solutions of SD that
†
fmercati@perimeterinstitute.ca have been found so far [7–9] are static, eternal, and

2470-0010=2016=94(6)=064023(10) 064023-1 © 2016 American Physical Society

DANIEL C. GUARIENTO and FLAVIO MERCATI PHYSICAL REVIEW D 94, 064023 (2016)
asymptotically flat, which makes them significantly less S ¼ SEH þ SM ; ð1Þ
interesting from the physical point of view.
The fact that SD allows for families of exact time- where SEH is the Einstein-Hilbert action given by
dependent solutions is of particular interest to cosmology. Z Z 
The family of solutions derived in this work may be expressed 3
SEH ¼ dt d x π ab γ_ ab
as a perturbed Friedman-Lemaître-Robertson-Walker
(FLRW) metric in the Newtonian gauge, a result that brings     
1 1 pffiffiffi
SD into the realm of gravitational theories with potential − N pffiffiffi π ab π ab − π 2 − γ R −2N a ∇b π ab
implications for cosmological evolution. γ 2
Z Z
For practical purposes, instead of working directly with
¼ dt d3 xðπ ab γ_ ab − NH − N a Ha Þ; ð2Þ
the SD dynamical system, it is convenient to simply work
in the conformal gauge in which SD is equivalent to GR in
a CMC foliation. We will therefore be studying solutions of where R is the 3-Ricci scalar and we have the usual
Arnowitt-Deser-Misner gravity in a CMC gauge. When definitions for the “super-Hamiltonian” and “supermomen-
such solutions exist, they are both solutions of GR and SD. tum,” namely,
However, there are situations in which such solutions do  
1 1 2 pffiffiffi
not correspond to a well-defined solution of Einstein’s H ≡ pffiffiffi π π ab − π − γ R;
ab
ð3Þ
equations, in particular, at the big bang singularity [5]. γ 2
However, by looking at the conformally invariant degrees
of freedom, one can check whether, as solutions of SD, they Ha ≡ 2∇b π ab ; ð4Þ
still make sense and can be continued past this breakdown
point. The strategy is to work with ADM gravity in a CMC and the matter action SM is left unspecified for now. It may
gauge as long as possible, and then focus on the shape be possible to define the Hamiltonian of an arbitrary perfect
degrees of freedom when the solutions evolve into some- fluid by defining it as a generic k-essence action. This
thing that cannot be described in GR. procedure has been carried out for some particular cases
In this work we derive a class of exact solutions of SD by [13], but the general action problem will be addressed in a
solving the ADM equations under the following assump- future work. We assume that it depends solely on the metric
components and not on their associated momenta, so that
tions: (i) spherical symmetry, (ii) comoving perfect fluid
we may define the components of the variation of SM with
source, (iii) asymptotically FLRW behavior, and (iv) a
respect to the lapse, shift, and metric as
singularity at the center. Conditions (i) and (ii) define the
Kustaanheimo-Qvist class of solutions, of which there are 1 δS
many physically interesting subcases [10], such as Wyman ρ≡ − pffiffiffi M ; ð5Þ
[11], FLRW, McVittie, and Schwarzschild–de Sitter [12]. N γ δN
Conditions (iii) and (iv) represent fixing properties of the
1 δS
poles of the spherically symmetric manifold. In particular, ja ≡ − pffiffiffi Ma ; ð6Þ
condition (iii) requires that the metric be regular at one of N γ δN
the poles, and condition (iv) requires it to be singular at the
1 δSM
opposite pole. Sab ≡ − pffiffiffi ab : ð7Þ
In Sec. II we establish the constraints and equations of N γ δγ
motion of shape dynamics to write the spherically symmetric
ansatz in its specific form. Using conditions (i) and (ii) we The Hamiltonian constraint and momentum (or diffeo-
arrive at the generic expression for the Kustaanheimo-Qvist morphism) constraint are, respectively,
class of spherically symmetric geometries in Sec. III, and pffiffiffi
then we use the 3-dimensional expression of the Hawking- H¼ γ ρ; ð8Þ
Hayward quasilocal mass to apply conditions (iii) and (iv) to pffiffiffi a
the evolution equations in Sec. IV. We find a particular Ha ¼ γj : ð9Þ
solution in Sec. V, which is shown to satisfy all the require-
ments of a solution of shape dynamics. Finally, we present The CMC condition reads
our conclusions and discuss further developments in Sec. VI. pffiffiffi
Latin indices run from 1 to 3 and in our units 16πG ¼ 1. γ ab π ab − γ hπi ¼ 0; ð10Þ

where hπi is the mean Rcanonical momentum across the 3-

II. STATEMENT OF THE PROBLEM
The gravita tional action of a system filled with a
continuous fluid with an arbitrary energy momentum tensor
is given by
d3 xγ ab π ab
surface, that is, hπi ≡ R 3 pffiffi . The evolution equations
d x γ
divide into a vacuum part, which is given by Hamilton’s
equations generated by the total Hamiltonian

064023-2
COSMOLOGICAL SELF-GRAVITATING FLUID SOLUTIONS … PHYSICAL REVIEW D 94, 064023 (2016)
R R  
dt d3 xðNH þ N a Ha Þ, and a matter contribution to π_ ab f 1 1 s
π ab
¼ diag sin θ; s sin θ; ; ð18Þ
given by the term (7). The equations read [1] μ 2 2 sin θ
 
N 1 with s ¼ sðr; tÞ and f ¼ fðt; rÞ. Spherical symmetry also
γ_ ab ¼ 2 pffiffiffi π ab − πγ ab þ 2∇ða N bÞ ; ð11Þ
γ 2 means that the source matter satisfies the properties
  ρ ¼ ρðr; tÞ; ð19Þ
pffiffiffi 1 ab
π_ ab
¼ −N γ R − γ Rab
2 ja ¼ jðr; tÞna ; ð20Þ
 ab    
N γ 1 1
þ pffiffiffi π cd π cd − π 2 −2 π ac π c b − ππ ab Sab ¼ pðr; tÞγ ab þ σðr; tÞPab ; ð21Þ
γ 2 2 2
pffiffiffi a b
þ γ ð∇ ∇ N − γ ab ∇c ∇c NÞ where Pab ≡ γ ab − γ c c na nb is the traceless projector with
pffiffiffi respect to the radial direction.
þ ∇c ðπ ab N c Þ − 2π cða ∇c N bÞ þ N γ Sab : ð12Þ
Condition (ii), that the source corresponds to a perfect
In order for a solution of SD to admit a description as a fluid, implies that there is no anisotropic stress; that is, in
spacetime solving Einstein’s equation (i.e. a solution of the fluid’s rest frame we have
GR), a lapse function defining a local notion of proper time
must be defined. Such a function can be determined by σ ¼ 0: ð22Þ
using the ADM equations of motion [Eqs. (11), (12)] to
calculate the time derivative of the CMC condition (10), In addition, condition (ii) limits our choice of fluid, in the
sense that it constrains us to place the fluid at rest with
pffiffiffi
γ respect to the observer; that is,
γ_ ab π ab þ γ ab π_ ab þ hπiγ ab γ_ ab
2
pffiffiffi j ¼ 0: ð23Þ
pffiffiffi γ
− γ h_γ ab π ab þ γ ab π_ ab i − hπihγ ab γ_ ab i ¼ 0: ð13Þ
2 It is worth noting that, while in GR a perfect fluid can
Replacing Eqs. (11) and (12) in Eq. (13) above, we get always be made comoving with a suitable choice of 4-
dimensional coordinates, in SD we have only 3-dimen-
ð8Δ − 2R − hπi2 þ SÞN sional diffeomorphism covariance; therefore, requiring that
   the fluid be comoving corresponds to a physical constraint
pffiffiffi
− 6γ −1 π ab − 13πγ ab π ab − 13πγ ab N ¼ h γ l:h:s:i: on the matter source.

ð14Þ III. SOLVING THE CONSTRAINTS

After applying the conditions from Sec. II A to the
The term on the right-hand side is a spatial constant and can
constraint equations (8), (9), and (10) we find
be written as an undetermined function of time ϖðtÞ ¼
pffiffiffi     0 2   
h γ l:h:s:i.
1 2 μ0 0 Y 1 f f s
ρ¼ 2 Y − Y 00 − − 2þ 2 − ;
μ Y μ Y Y 2Y Y 2 μ
A. Spherical symmetry and perfect fluid conditions
ð24Þ
We start by applying condition (i) so that our ansatz for
the solution is a spherically symmetric metric on 3-space.
f0
Following the notation from Ref. [6] we define μ ¼ sY 0 ; ð25Þ
Y
ds2 ¼ γ ab dxa dxb f s
hπi ¼ þ ; ð26Þ
¼ μ2 ðr; tÞdr2 þ Y 2 ðr; tÞdΩ2 ; ð15Þ Y2 μ

N ¼ Nðr; tÞ; ð16Þ and we can immediately solve the CMC condition (26) and
momentum constraint (25) to find
N a ¼ ξðr; tÞna ; ð17Þ  
f
s ¼ μ hπi − 2 ; ð27Þ
where na ≡ δar is a unit vector in the radial direction and Y
dΩ2 ≡ dr2 þ rsin2 θdθ2 is the usual 2-sphere element. The
canonical momentum conjugate to the metric is defined in hπiY 2 A
f¼ þ ; ð28Þ
spherical symmetry as [6] 3 Y

064023-3
DANIEL C. GUARIENTO and FLAVIO MERCATI PHYSICAL REVIEW D 94, 064023 (2016)
where AðtÞ is a spatially homogeneous integration constant.   
1 0 Y3 0 Y4 N0 0
The homogeneity of A is a consequence of the assumption MW ¼ − ξf Y − ðsξÞ þ
j ¼ 0 in Eq. (23) due to the fact that j acts as a source for 6N 2μ μ μY
  0 0 
the radial derivative of A. Inserting these results back into Y 1 2 _ Y
þ ðY s_ þ NsfÞ − fμ þ NY
the Hamiltonian constraint (24), we find μ 2 μ
    0 2   02
 2
1 2 μ0 0 Y 1 hπi2 3 A2 1 Y 1f
ρ¼ 2 Y −Y − 00
− 2− þ : þ Y 1− 2 þ ; ð32Þ
μ Y μ Y Y 12 4 Y6 6 μ 4Y
ð29Þ and using Eq. (30) we find that the two parts sum
up to
A. Hawking-Hayward mass  
The Hawking-Hayward quasilocal mass [14] is a Y Y 02 f2
M HH ¼ 1− 2 þ ; ð33Þ
measure of the energy content inside a codimension-2 2 μ 8Y
compact hypersurface in general relativity, defined as the
as expected [20]. The full derivation of this result may be
Hamiltonian in a 2 þ 2 foliation of spacetime. It coincides
found in the Appendix.
with the Misner-Sharp mass [15] in spherical symmetry and
with the Bondi and ADM masses [16,17] if the metric is
asymptotically flat. Despite the fact that in shape dynamics B. Solving the Hamiltonian constraint
there is no spacetime, it is still useful to use the Hawking- An immediate first application of the expression of the
Hayward quasilocal mass as a guide to the determination of Hawking-Hayward mass in spherical symmetry is to
the energy content inside a spatial volume. In the following algebraically solve the Hamiltonian constraint for μ
section we use the ADM formalism to cast the Hawking- without making use of the equations of motion. By
Hayward mass in terms of quantities contained in the substituting the definition from Eq. (33) into Eq. (24),
hypersurface, which will render it a meaningful quantity we may cast it as
to use in shape dynamics. We also interpret the different
contributions to the mass in terms of hypersurface quantities.  
f s f0 2M 0
It has been pointed out in the literature that in general ρ¼ 2 − 0 þ 2 HH0 : ð34Þ
2Y μ YY Y Y
relativity, because the Hawking-Hayward mass may ulti-
mately be written as a projection of the Riemann tensor, it
Applying the momentum constraint (28) and the CMC
may therefore be split in two distinct contributions: one
condition (27), we find
from the Ricci tensor and one from the Weyl tensor
[17–19]. In light of the Einstein equations, these contri-
2M0HH
butions can then be traced as coming from the energy- ρ¼ ; ð35Þ
momentum tensor distributed in the medium (the Ricci Y2Y0
part) and from a pointlike source or otherwise compact
which allows us to write M HH in integral form in terms
object (the Weyl part). In shape dynamics there is no well-
of ρ as
defined spacetime, but once we are in a foliation we may
use the ADM splitting to define an analogue to the
M HH ¼ η; ð36Þ
Hawking-Hayward mass using only quantities defined in
the hypersurface. After performing the splitting we may
define the Hawking-Hayward mass MHH as where

1
MHH ¼ M R þ MW ; ð30Þ η0 ¼ ρY 2 Y 0 : ð37Þ
2
where the Ricci component M R and Weyl component MW ,
Inserting this result back into the definition of the
in our spherically symmetric metric, read
Hawking-Hayward mass (33) and noting that it can be
   solved algebraically for μ, we find
1 0 Y3 0 Y4 N0 0
MR ¼ ξf Y − ðsξÞ þ
6N 2μ μ μY 6YY 0
  0 0  μ ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi : ð38Þ
Y 1 2 _ þ NY Y
þ ðY s_ þ NsfÞ − fμ Y 4 ðhπi þ 3A Y3
Þ2 þ 9Yð4Y − 8ηÞ
μ 2 μ
 02
 2
1 Y 1f
þ Y 1− 2 − ; ð31Þ The usefulness of this result will become clear when
3 μ 8Y solving the lapse-fixing equation in Sec. VA.

064023-4
COSMOLOGICAL SELF-GRAVITATING FLUID SOLUTIONS … PHYSICAL REVIEW D 94, 064023 (2016)
IV. SOLVING THE EQUATIONS OF MOTION 6 μ_
N¼− : ð47Þ
The evolution equations for the metric (11) read hπi μ

We now move on to the momentum equations of motion.

Nf
Y_ ¼ − þ Y 0 ξ; ð39Þ The independent components of Eq. (12) are
2Y
 2
  0

  _f ¼ N 1 þ Y 2 p − 3 f þ ξ f 0 − f μ
N μf
μ_ ¼ − s þ ðξμÞ0 : ð40Þ 4 Y2 μ
2 Y2  
μ_ sN 2YY N þ NY 02
0 0
þf þ − ξ0 − ð48Þ
Inserting the solutions of the constraint equations, Eqs. (27) μ 2μ μ2
and (28), we then find     
f2 2 Y0 0
  s_ ¼ N μ 2p þ 4 − þ ðsξÞ0
_Y ¼ − N hπi Y þ 1 A þ Y 0 ξ; ð41Þ 
2Y

Y μ
6 2 Y2 2N 0 μ0 Y 0 2N 00
þ − − : ð49Þ
  μ μ Y μ
hπi A
μ_ ¼ − N − 3 þ ðξμÞ0 : ð42Þ
6 Y Inserting the results from the constraints, Eqs. (28) and
(27), and the equations of motion of the metric, Eqs. (46)
The presence of a nonzero A implies that there is a and (47), the radial and angular components of the
singularity in the metric at areal radius Y ¼ 0, that is, at momentum evolution equations read, respectively,
both antipodes of the compact spacetime [6]. Although  0 
condition (iv) implies that we have a central singularity, 1 _ 6 μ 0 μ0 μ_ R0
hπi ¼ 3 2_μ − þ 2 μ_ 0
condition (iii) means that far from the singularity the 3 μ hπi μ μ R
spacetime is expected to satisfy, among other properties,   
μ_ 02 μ_ 6p 1
regularity. Therefore, because A is constant at every leaf þ 2 ð1 − R Þ þ − hπi ; ð50Þ
R μ hπi 2
due to our choice of j in Eq. (23), in order to guarantee a
regular spacetime on the antipodal point we choose A ¼ 0.  0 
1 _ 6 μ 0 μ0 μ_
With this choice, the equations of motion simplify signifi- hπi ¼ 3 − 2_μ − þ μ_ 00
cantly, and we now have 3 μ hπi μ μ
  
1 0 0 00 μ_ 6p 1
þ ðR μ_ þ R μ_ Þ þ − hπi : ð51Þ
Y_ hπi Y0 R μ hπi 2
¼− Nþ ; ð43Þ
Y 6 Y
The difference between these two components, also known
hπi as pressure isotropy condition [12], or spatial Ricci isotropy
μ_ ¼ − Nμ þ ðξμÞ0 ; ð44Þ [18], yields
6
  
which also implies that 6 00 μ0 0 μ0 μ_ R0 0
μ
_ þ 2 2_
μ − − μ_
μ3 hπi μ μ R
 0  
Y_ μ_ Y μ0 1
− ¼ξ − − ξ0 : ð45Þ þ 2 ðRR00 − R02 þ 1Þ ¼ 0; ð52Þ
Y μ Y μ R

If we choose a gauge in which there is no shift (by setting so we are motivated to choose a gauge for which [21,22]
ξ ¼ 0), the metric evolution Eq. (45) results in
RR00 − R02 þ 1 ¼ 0 ⇒ R02 ¼ 1 − kR2 ð53Þ
Y ¼ μRðrÞ; ð46Þ
so that
where RðrÞ appears as an integration constant, and is a 8
function of the radial coordinate only. Equation (46) splits < sinh r k ¼ −1;
the areal radius Y such that the metric becomes conformally R¼ r k ¼ 0; ð54Þ
:
equivalent to a static metric. The time evolution of the sin r k ¼ 1:
metric γ ij (15) is now encoded in the conformal factor μ. By
setting ξ ¼ 0 in Eq. (44), we find an expression for the Our gauge choice may be further specified by choosing a
lapse, namely, value for k, which will remain undetermined in our

064023-5
DANIEL C. GUARIENTO and FLAVIO MERCATI PHYSICAL REVIEW D 94, 064023 (2016)
analysis. For μ_ ≠ 0, Eq. (52) can then be rewritten as a total By inserting the results of the pressure isotropy con-
derivative, namely, dition, Eqs. (53) and (56), into one of the momentum
  equations of motion, we find
μ02 R0  
∂ t μ00 − 2 − μ0 ¼ 0; ð55Þ 1 1 μ _ μ_ 0 R0 μ0
μ R 2
p ¼ − hπi − hπihπi þ 2 2 þ
12 18 μ_ μ μ_ R μ
which reduces to the ordinary differential equation  
1 μ02
− 2 kþ 2 ; ð63Þ
μ μ
μ02 R0
μ00 − 2 − μ0 ¼ χðrÞ: ð56Þ
μ R which may interpret as the definition of the fluid pres-
sure p.
Inserting these results back into the Hamiltonian con-
straint (24), we find
V. FINDING A PARTICULAR SOLUTION
2
 02
  0 
hπi 3 1−R 2 χ 0 R μ0 The Ricci part MR of the Hawking-Hayward quasilocal
ρ¼− þ 2 − þ μ þ ; ð57Þ
12 μ R2 μ 3 R μ mass is associated with the presence of a source field in the
equations of motion, since it vanishes in a vacuum solution.
and taking the radial derivative of Eq. (57) and using again The Weyl part MW is therefore associated with the presence
Eqs. (56) and (53), we find of a central compact object, as well as the presence of
spacetime singularities. In particular, we might choose a
R0 1 finite interval for the radial coordinate by setting k ¼ 1 in
χ 0 þ 3χ ¼ − ρ0 μ3 ; ð58Þ
R 2 Eq. (54), which for particular forms of μ would be akin to
closed FLRW models. In this case, for the foliation to be
whose solution is regular at the antipode r ¼ π, in addition to setting A ¼ 0 in
 Z  Sec. IV, the Weyl tensor must also be finite at the antipode.
1 1 0 3 On the other hand, at the coordinate center r ¼ 0 a
χ¼ 3 − ρ ðμRÞ dr þ m0 : ð59Þ
R 2 singularity is guaranteed if the Weyl tensor diverges, there-
fore satisfying condition (iv). The simplest choice of a
We may use the results from this section to rewrite the function χ that satisfies both requirements while relying only
Hawking-Hayward mass in terms of the new metric on the already defined gauge-fixing function R is given by
functions. After applying the results from the constraint
equations, Eqs. (27) and (28), and the equations of motion, w02
χ ¼ 3m ; ð64Þ
Eqs. (46), (47), and (56), we find that the two parts of w3
Eq. (30) reduce to
where m is a constant, and
 2 3   
μR hπi μ χ μ02 r
MR ¼ ð1 − R02 Þ − R2 μ0 R0 þ R3 − − ; wðrÞ ≡ 2R : ð65Þ
2 72 3 2μ 2
ð60Þ By making this choice, we have fixed the mass of the central
object MW to be equal to m,1 and also ensured that condition
R3 (iv) holds.
MW ¼ χ; ð61Þ
3 Applying Eq. (64) to Eq. (56), and noting that Eq. (53)
implies that w02 ¼ 1 − k4 w2 , we have
with χ given by Eq. (56). We may interpret this result as the
fact that a nonzero χ corresponds to a contribution to the μ02 R0 w02
energy of the system from a source other than the fluid, μ00 − 2 − μ0 ¼ 3m 3 : ð66Þ
μ R w
such as a compact object or a black hole, as we will see in
the following section. The pressure isotropy condition (66) now admits a solution
Now, inspecting Eq. (58), and using Eq. (46) we notice of the form
that χ and η defined in Eq. (37) are related by  
m 2
μ¼a 1þ ; ð67Þ
R3 χ ¼ 6η − ρR3 μ3 þ BðtÞ; ð62Þ 2aw

where BðtÞ is an integration constant. Notice that the left- 1

This interpretation becomes clear if one chooses a spatially
hand side is independent of time. flat asymptotic spacetime by setting k ¼ 0 in Eq. (54).

064023-6
COSMOLOGICAL SELF-GRAVITATING FLUID SOLUTIONS … PHYSICAL REVIEW D 94, 064023 (2016)
 
with a ¼ aðtÞ an arbitrary function of time only. Now that ϖhπi μ 1 6χR0 þ 2Rχ 0 0
we have both χ and μ, we can calculate the Hawking- ¼ þ Rρ μ
24 μ_ R0 μ þ Rμ0 μ2
Hayward mass of this solution; we get   
6 R0 μ0 χ 2μ0 − 1 R0 μ0
    þ 2 þ −kþ þ
μ R μ μ μ R μ
w3 2 32 ak H2 a3 m 6 m
MHH ¼ ð4 − kw Þ þ 1þ þ 3 ; þ 3ðρ þ pÞ: ð73Þ
8 2 2 2aw w
ð68Þ Inserting the solution (70) into the lapse-fixing equation,
we find that it is identically satisfied. This proves that the
and Eq. (62) results in B ¼ 0. McVittie lapse (69) is a solution of the lapse fixing
Equation (47) now yields equation (14). Our solution is consistent and is (locally)
both a solution of shape dynamics and general relativity.
6 1 − 2aw
m
a_
N¼− : ð69Þ
hπi 1 þ 2aw a
m
VI. CONCLUSIONS
In this work we have found a new solution of shape
For condition (iii) to hold, the solution requires that N → 1 dynamics by building on the fact that in spherical symmetry
as r ≫ m, and because this must be satisfied for all slices a CMC foliation is equivalent to a shear-free comoving
we are required to choose hπi ¼ −6 aa_ . The 4D metric flow; this is part of potentially an entire new class of
finally reads solutions to the theory. There are many implications of this
 4 result, and the next step is now to fully characterize this
2 2 m solution. Although we know the causal structure of
ds ¼ aðtÞ 1 þ ½dr2 þ RðrÞ2 dΩ2 ; ð70aÞ
2aðtÞwðrÞ spatially flat McVittie spacetimes in general relativity
[24,25], the spatially compact counterpart, despite having
1 − 2aðtÞwðrÞ
m been previously studied in the context of general relativity
N¼ ; ð70bÞ [26], requires a completely new interpretation in the context
1 þ 2aðtÞwðrÞ
m
of shape dynamics.
Whether one or more members of this class of solutions
N i ¼ 0: ð70cÞ in fact contain a black hole remains an open question. In
GR, the 4-dimensional compact McVittie metric does not
This is the well-known McVittie metric [23] written in a possess an event horizon, but it may contain apparent
compact foliation. horizons that are not covered by the coordinate patch used
in previous analyses [26]. As in previously studied shape
dynamics solutions, the event horizon or otherwise margin-
A. Lapse-fixing equation ally trapping surfaces may well give way to a throat into
Rewriting Eq. (14) in our spherically symmetric ansatz another region of space which possesses no general-
and after applying the solutions of the CMC [Eq. (27)], relativistic analogue [6], which may extend the explorable
Hamiltonian [Eq. (24)], and momentum [Eq. (28)] con- region up to a possible universal horizon or other locus
straints, where the CMC foliation can no longer be extended [27].
This is a fundamental feature displayed by shape dynamics
  solutions that is not present in their general-relativistic
ϖ hπi 2A2
− ¼ N ρ þ 3p þ þ 6 counterparts. Moreover, it has been shown that McVittie
4 6 Y
0
 0 0
 spacetimes admit spatially bound timelike and null orbits
2N μ 2Y 2N 00 [28], an “astrophysical” property which would be interest-
þ 2 − − 2 : ð71Þ
μ μ Y μ ing to study in the context of shape dynamics. Finally,
regarding singularities, the known McVittie spacetime
Using Eq. (38), we may write it as singularities all stem from divergences of 4-dimensional
quantities, in regions where 3-dimensional quantities often
 2 02  0 0  
1 4Y Y Nμ Yρ0 ϖ remain well behaved. Therefore, they may very well
2 0 3 0 − N 3ðρ þ pÞ þ 0 ¼ : represent perfectly regular and traversable regions in a
2Y Y μ Y Y 4
shape dynamics interpretation, as it has been shown to be
ð72Þ the case in other solutions, such as Bianchi IX [5],
“Schwarzschild-like” [7], and “Kerr-like” [8] solutions.
After applying the results from gauge-fixing the metric Another important question that has been left open here
evolution equations [Eqs. (46) and (47)] and the pressure is the action of the matter source. It may be possible to
isotropy condition [Eqs. (54) and (56)], we find follow a procedure similar to previous Hamiltonian

064023-7
DANIEL C. GUARIENTO and FLAVIO MERCATI PHYSICAL REVIEW D 94, 064023 (2016)
analyses of k-essence models [13] in order to find a generic an example of which in spherical symmetry is a 2-sphere of
perfect-fluid source, so we may characterize the source radius r.
from a field theory perspective and provide an analogy with
the cuscuton source of the McVittie spacetime [29]. 1. Hypersurface decomposition
Finally, it must be noted that a much broader class of
solutions of Eq. (56) has been studied in Ref. [30]. Their The Hawking-Hayward quasilocal mass is defined in
applicability as solutions of shape dynamics, as well as terms of radial null geodesics on a compact spacelike 2-
whether they can be related to general solutions of the surface S as the integral of the Hamiltonian two-form over
lapse-fixing equation (72), will be the object of future work. the surface S, multiplied by the length scale given by the
area of S [14],
ACKNOWLEDGMENTS pffiffiffiffi Z 
1 A
M HH ≡ dS ð2Þ R þ θðþÞ θð−Þ
ð4πÞ2 4 S
3
We thank N. Afshordi, H. A. Gomes, S. Gryb, A. Maciel,
L. Smolin, and R. Sorkin for insights and valuable 
1 ðþÞ μν μ
discussions. D. C. G. is supported by CNPq Grant − σμν σð−Þ − 2ωμ ω ; ðA4Þ
No. 206101/2014-7. This research was supported in part 2
by Perimeter Institute for Theoretical Physics. Research at R
the Perimeter Institute is supported by the Government of where A ≡ S dS is the area of S, θðÞ are the expansion
Canada through Innovation, Science and Economic scalars of the ingoing (−) and outgoing (þ) null geodesics,
ðÞ
Development Canada and by the Province of Ontario σμν is their respective shear tensor of the geodesic
through the Ministry of Research, Innovation and congruence, and ωμ the twist of the surface S. In spherical
Science. This research was also partly supported by a symmetry ωμ vanishes, and we may use the contracted
grant from the John Templeton Foundation. Gauss equation to write

1 ðþÞ
ð2Þ
R þ θþ θ− − σμν σμν αγ βδ
ð−Þ ¼ h h Rαβγδ ; ðA5Þ
APPENDIX: INTERPRETATION OF THE 2
HAWKING MASS AS A TWO-COMPONENT
QUANTITY with hμν as defined in Eq. (A3). In this context, we can use
Eq. (A5) to rewrite the Hawking-Hayward mass from
In this a ppendix we perform the derivation and split of
Eq. (A4) as
the Hawking-Hayward mass from 4-dimensional spacetime
in terms of 3-dimensional quantities defined in our space- pffiffiffiffi Z
1 A
like foliation. To do so, we use some of the techniques M HH ¼ dShac hbd Rabcd : ðA6Þ
ð4πÞ2 4 S
3
developed in Refs. [17–19]. To avoid cluttering the
notation, all quantities represent 4-dimensional objects
unless stated otherwise. Greek indices run from 0 to 3 Moreover, using the decomposition of the Riemann
and we use the signature ð−; þ; þ; þÞ. tensor into its Ricci and Weyl parts, that is,
We start by defining a few projectors into the hypersur-
face. The comoving flow of a 4-dimensional metric g is R
Rμναβ ¼ Cμναβ þ gμ½α Rβν − gν½α Rβμ − g g ; ðA7Þ
defined as 3 μ½α βν

1 we finally recover Eq. (30) with the contributions from the

uμ ≡ pffiffiffiffiffiffiffiffiffiffi δμt ; ðA1Þ Ricci part and Weyl part defined as
−g00
pffiffiffiffi Z 
and the orthogonal projection with respect to uμ gives the A1 μα νβ
MR ≡ dSh h gμ½α Rβν
ð4πÞ2 4 S
3
spatial slices

R
γ μν ≡ gμν þ uμ uν : ðA2Þ − gν½α Rβμ − gμ½α gβν ; ðA8Þ
3
We may also define a unit vector nμ orthogonal to the flow. pffiffiffiffi Z
To do so, we use the acceleration u_ μ ≡ uα uμ;α to write 1A
MW ≡ dShμα hνβ Cμναβ : ðA9Þ
_μ
nμ ≡ puffiffiffiffiffiffiffi, which we use to construct the induced metric ð4πÞ
3
2 4 S
α
u_ u_ α
on a codimension-2 hypersurface, In fact, the integrand in Eq. (A9) is the electric part of the
Weyl tensor, which prompts the interpretation of a
hμν ≡ γ μν − nμ nν ; ðA3Þ “Newtonian” character of the Hawking-Hayward mass [19].

064023-8
COSMOLOGICAL SELF-GRAVITATING FLUID SOLUTIONS … PHYSICAL REVIEW D 94, 064023 (2016)
2. Back to the hypersurface Thus, we may cast the individual contributions from the
Using the contracted Gauss equation, the Hawking- first part of the Appendix by writing the contributions from
Hayward quasilocal mass can be cast entirely in terms Eqs. (A8) and (A9) as
of quantities within a 3-dimensional surface [19]. Since the pffiffiffiffi Z  
1 A R
mass depends on the projection of the Riemann tensor on MR ¼ dS hab
Rab − ; ðA14Þ
ð4πÞ2 4 S 3
3
the 2-surface, it is useful to compute the projection of
Eq. (A7) on the 2-surface, that is, pffiffiffiffi Z  
1 A ab R þ R :
MW ¼ dS h ab hcd R
abcd − h ab
ð4πÞ2 4 S 3
3
R ab cd
hab hcd Rabcd ¼ hab hcd Cabcd − h h ga½c gdb
3 ðA15Þ
þ hab hcd ðga½c Rdb − gb½c Rda Þ: ðA10Þ
Using the contracted Gauss-Codazzi equations, we write
We start by noticing the following identity: the Ricci scalar and projected Ricci tensor in ADM form as

hab hcd ga½c gdb ¼ 1: ðA11Þ ab ð3Þ
h Rab ¼ h
ab
Rab þ KK ab − 2K c b K ca

Additionally, for any symmetric rank-2 tensor T, we have 1
− ð∂ t K ab − Lξ K ab þ ∇a ∇b NÞ ; ðA16Þ
N
hab hcd ðga½c T db − gb½c T da Þ ¼ hab T ab : ðA12Þ
R ¼ ð3Þ R þ K 2 − 3K ab K ab
Making use of these results, the 2-surface projected 2 ab
Riemann tensor reads − γ ð∂ t K ab − Lξ K ab þ ∇a ∇b NÞ; ðA17Þ
N
R and we use Eqs. (A16) and (A17) to cast the Ricci and Weyl
hab hcd Rabcd ¼ hab hcd Cabcd þ hab Rab − : ðA13Þ
3 contributions from Eqs. (A14) and (A15) as

pffiffiffiffi Z   
1A ab ð3Þ 1 _
MR ¼ dS h Rab þ KK ab − 2K b K ca − ðK ab − Lξ K ab þ ∇a ∇b NÞ
c
ð4πÞ2 4 S
3
N
 
1 ð3Þ 2 2 ab _
− R þ K − 3K K ab − γ ðK ab − Lξ K ab þ ∇a ∇b NÞ ;
ab ðA18Þ
3 N
pffiffiffiffi Z 
1 A
MW ¼ dS hac hbd ðð3Þ Rabcd þ K ac K bd − K ad K bc Þ
ð4πÞ2 4 S
3

 
1
− hab ð3Þ Rab þ KK ab − 2K c b K ca − ðK_ ab − Lξ K ab þ ∇a ∇b NÞ
N
 
1 ð3Þ 2 2 ab _
þ R þ K − 3K K ab − γ ðK ab − Lξ K ab þ ∇a ∇b NÞ ;
ab
ðA19Þ
3 N

which depend solely on quantities defined in the hypersurface. By applying this result to our spherically symmetric ansatz
we arrive at the expressions from Eqs. (31) and (32).

[1] R. Arnowitt, S. Deser, and C. W. Misner, in Gravitation: An
Introduction to Current Research, edited by L. Witten (John
Wiley & Sons Inc., New York, 1962), Chap. 7, pp. 227–265;
L. WittenGen. Relativ. Gravit. 40, 1997 (2008).
[2] J. W. York, Phys. Rev. Lett. 26, 1656 (1971); Y.
Choquet-Bruhat, Symposia Math. 12, 317 (1973); N.
O’Murchadha and J. W. York, J. Math. Phys. (N.Y.) 14,
1551 (1973).
[3] H. Gomes, S. Gryb, and T. Koslowski, Classical Quantum
Gravity 28, 045005 (2011); H. Gomes and T. Koslowski,
Classical Quantum Gravity 29, 075009 (2012).
[4] F. Mercati, arXiv:1409.0105.

064023-9
DANIEL C. GUARIENTO and FLAVIO MERCATI
[5] T. A. Koslowski, F. Mercati, and D. Sloan, arXiv:
1607.02460.
[6] H. Gomes, T. Koslowski, F. Mercati, and A. Napoletano,
arXiv:1509.00833.
[7] H. Gomes, Classical Quantum Gravity 31, 085008 (2014).
[8] H. Gomes and G. Herczeg, Classical Quantum Gravity 31,
175014 (2014).
[9] G. Herczeg, arXiv:1508.06704.
[10] P. Kustaanheimo and B. Qvist, Comment. Phys. Math. 13, 1
(1948); Gen. Relativ. Gravit. 30, 663 (1998).
[11] M. Wyman, Phys. Rev. 70, 396 (1946).
[12] H. Stephani, D. Kramer, M. MacCallum, C. Hoenselaers,
and E. Herlt, Exact Solutions of Einstein’s Field Equations,
Cambridge Monographs on Mathematical Physics, 2nd ed.
(Cambridge University Press, Cambridge, 2009).
[13] L. R. Abramo and N. Pinto-Neto, Phys. Rev. D 73, 063522
(2006).
[14] S. W. Hawking, J. Math. Phys. (N.Y.) 9, 598 (1968); S. A.
Hayward, Phys. Rev. D 49, 831 (1994).
[15] C. W. Misner and D. H. Sharp, Phys. Rev. 136, B571
(1964).
[16] H. Bondi, M. G. J. van der Burg, and A. W. K. Metzner,
Proc. R. Soc. A 269, 21 (1962); R. K. Sachs, Proc. R. Soc. A
270, 103 (1962).
064023-10
PHYSICAL REVIEW D 94, 064023 (2016)
[17] R. L. Arnowitt, S. Deser, and C. W. Misner, Phys. Rev. 116,
1322 (1959); M. Carrera and D. Giulini, Phys. Rev. D 81,
043521 (2010).
[18] M. Carrera and D. Giulini, Rev. Mod. Phys. 82, 169 (2010).
[19] V. Faraoni, Symmetry 7, 2038 (2015).
[20] R. M. Misra and D. C. Srivastava, Phys. Rev. D 8, 1653
(1973).
[21] R. A. Sussman, J. Math. Phys. (N.Y.) 28, 1118 (1987).
[22] G. C. McVittie, Ann. Inst. Henri Poincaré, A 40, 235 (1984);
Mon. Not. R. Astron. Soc. 93, 325 (1933).
[23] G. C. McVittie, Mon. Not. R. Astron. Soc. 92, 500 (1932).
[24] N. Kaloper, M. Kleban, and D. Martin, Phys. Rev. D 81,
104044 (2010); K. Lake and M. Abdelqader, Phys. Rev. D
84, 044045 (2011).
[25] A. M. da Silva, M. Fontanini, and D. C. Guariento, Phys.
Rev. D 87, 064030 (2013).
[26] R. A. Sussman, J. Math. Phys. (N.Y.) 29, 1177 (1988); B. C.
Nolan, Phys. Rev. D 58, 064006 (1998).
[27] A. Maciel, Phys. Rev. D 93, 104013 (2016); M. Meiers, M.
Saravani, and N. Afshordi, Phys. Rev. D 93, 104008 (2016).
[28] B. C. Nolan, Classical Quantum Gravity 31, 235008 (2014).
[29] E. Abdalla, N. Afshordi, M. Fontanini, D. C. Guariento, and
E. Papantonopoulos, Phys. Rev. D 89, 104018 (2014).
[30] M. Wyman, Canad. Math. Bull. 19, 343 (1976).