Some questions about Conformal Cyclic Cosmology

Paul Tod

February 23, 2022

arXiv:2202.10864v1 [gr-qc] 22 Feb 2022

Abstract
This article is an extended version of a talk given in Oxford in June 2021 as part of an
online meeting ‘Ninety minutes of CCC’ to mark the 90th birthday of Sir Roger Penrose. I
assemble some questions that I have been asked or have asked myself about CCC.

Introduction and sketch of Conformal Cyclic Cosmology

In this article I collect together some questions about Penrose’s Conformal Cyclic Cosmology [8]
(hereafter CCC) that I’ve been asked or have asked myself, particularly when giving the lectures
[13]. My point of view is that CCC needs to answer these questions but I see no reason to think
it can’t.
I’ll begin with a sketch of CCC. Assuming the presence in the universe of a positive cosmo-
logical constant Λ there is a firm consensus that the universe is expanding away from a hot big
bang with infinite density, through an era of galaxies into a period of exponential expansion.
We model the universe as a solution of a suitable Einstein equation and then, with positive Λ,
Penrose’s conformal methods indicate that there will be a conformal rescaling of the space-time
metric in the remote future which can add a future infinity I + as a space-like boundary. It is
natural to suppose that the rescaled metric is smooth there – natural in the sense that one can
construct many examples with that property, and in fact, following [3], one can give well-posed
Cauchy problems with data at a space-like I for many matter models.
The cosmological model will have a curvature singularity at the beginning, and Penrose has
long argued that, as a singularity of a Lorentzian manifold, this big bang must be rather special,
much less general than such singularities can be. Penrose’s Weyl Curvature Hypothesis [9] is
the hypothesis that the Weyl curvature is zero, or at least finite, at any initial singularity and
therefore at the big bang. It is tricky, though possible (see e.g. [7]), to give a statement of what
it means to have ‘finite Weyl tensor’ at a singularity of the Riemann tensor (for example, the
singularity is not yet a ‘place’ where things can happen) and a simpler condition to impose is that
there should be a conformal rescaling of the space-time metric which removes the singularity and
adds an initial space-like boundary hypersurface. This is a strong condition, apparently stronger
than the bare Weyl Curvature Hypothesis, but it will follow from a strong enough version of the
Weyl Curvature Hypothesis and conditions of uniformity, [7].
Now comes Penrose’s ‘outrageous suggestion’: the Weyl Curvature Hypothesis, which is
motivated by observation as Penrose explains [9], takes it simplest form if the space-time metric
can be scaled up to allow a past space-like boundary to be added; the presence of positive Λ
implies that the space-time metric can be scaled down to allow a future space-like boundary to
be added; we have an aeon of universe between initial and final space-like surfaces through both
of which the conformally-related metric can be extended; so let us hypothesise a continuation
of the conformal metric in both directions to a preceding aeon and a subsequent aeon; so
there would be a single conformal metric but different physical metrics in each aeon, all of them
conformal to the single conformal metric. At the future boundary of an aeon, the Weyl curvature

1
necessarily vanishes (though there are free degrees of freedom in its normal derivative) so with
this identification the Weyl tensor will also necessarily vanish at the big bang of any aeon – by
this means the Weyl Curvature Hypothesis becomes a deduction.
To introduce equations we consider two successive aeons, say the previous one, characterised
by hatted quantities, and the present one characterised by quantities with a check. Then the
previous aeon is a space-time M̂ with metric ĝ and the present is M̌ with ǧ. There is a conformal
metric g defined right through, including the surface I which is the future boundary of M̂ and
the past boundary of M̌ . Strictly speaking I is in neither M̂ nor M̌ but we assume as much
regularity on the manifold M := M̂ ∪ M̌ ∪ I as we need. Now we have conformal factors Ω̂, Ω̌
such that
ĝ = Ω̂2 g, ǧ = Ω̌2 g, (1)
with
Ω̂ → ∞, Ω̌ → 0 towards I.
We make another regularity assumption, that the product Ω̂Ω̌ is finite and nonzero in a neigh-
bourhood of I. The product is negative if we suppose that each of Ω̂, Ω̌ is positive in its own
M̂ , M̌ . At this stage there is conformal freedom in g: we can make the changes
(Ω̂, Ω̌, g) → (θ Ω̂, θ Ω̌, θ −2 g),
for positive smooth θ. This is a gauge freedom – the two physical metrics ĝ, ǧ are unchanged by
it – which evidently changes the product of the conformal factors
Ω̂Ω̌ → θ 2 Ω̂Ω̌,
so that we can choose θ to set
Ω̂Ω̌ = −1. (2)
This is Penrose’s Reciprocal Hypothesis [8] which I’m here regarding as a gauge condition. Given
(2), we have Ω̌ in terms of Ω̂ and therefore
ǧ = Ω̌2 g = Ω̌2 Ω̂−2 ĝ = Ω̂−4 ĝ, (3)
so that the physical metric in the present aeon is determined by the physical metric in the
previous aeon and Ω̂. This sets the scene for question 1 below.
It may be helpful at this point to recall a concrete example from [13]. Consider the FLRW
metric
ds2 = dt2 − R(t)2 dσk2 (4)
where dσk2 is one of the standard constant-curvature 3-metrics. Take the matter content to be
radiation fluid plus the cosmological constant Λ. The Einstein field equations reduce to the
conservation equation, which can be integrated to give the energy density of the radiation as
ρ = µ/R4 ,
with a constant of integration µ, together with the Friedmann equation, which we’ll write in
terms of conformal time. This is defined by
dτ = dt/R,
(remember: the great virtue of conformal time is that future light-cones of the origin in FLRW
have equations τ − r = constant). Then the Friedmann equation is1
 
dR 2 1
= µ − kR2 + ΛR4 . (5)
dτ 3
1
To avoid the possiblity of a recollapsing universe we’ll assume that µΛ > 3/4 if k = 1.

2
This equation has a striking symmetry: set S = c1 R−1 where c1 = (3µ/Λ)1/2 , then calculate
 2
dS 1
= µ − kS 2 + ΛS 4 ,
dτ 3

which is precisely the Friedmann equation back again. It’s worth noting that R−1 is integrable
at both ends (i.e. large and small positive t) so that there is only a finite amount of conformal
time to cover the infinite range in proper time from big bang to I. If we set τ = 0 at the initial
singularity, where conventionally t = 0, then at I
Z ∞
τ = τF = R−1 dt,
0

writing τF for the value of τ at I. Since for large t we have R ∼ eHt where Λ = 3H 2 we’ll also
have, for large t Z ∞
τF − τ = R−1 dt ∼ H −1 e−Ht (6)
t
which gives a measure of how the τ -coordinate compresses the range of the t-coordinate towards
I. The same point is made by noting that
1
R∼ towards I,
H(τF − τ )

so that R has a simple pole at τ = τF .

For use in Section 5, we’ll put some numbers into these expressions. Following a calculation
in [13], one could use an expression for R(t) for the actual current universe, including terms
for the known dust and radiation densities and the measured Λ in the Friedmann equation, to
calculate that at the present epoch, at a time 1.4 × 1010 years after the bang, the ratio τ /τF
is about .74 – in conformal time we are that close to the end! If we make t ten times bigger,
so 1011 years, then the fractional amount of conformal time remaining drops from .26 to 10−4 .
The whole infinite future of the universe in t-time is compressed to very little in τ -time.
To connect the FLRW metric to CCC, suppose the metric ĝ in the previous aeon takes the
form of (4), so
ĝ = dt̂2 − R̂(t̂)2 dσk2 = R̂2 (dτ 2 − dσk2 )
−1/2
where we denote proper time in M̂ as t̂, and with density ρ̂ = m/R̂4 . Choose Ω̂ = c1 R̂ with
c1 as above, then (3) gives

ǧ = Ω̂−4 ĝ = Ř2 (dτ 2 − dσk2 ) = dť2 − Ř2 (ť)dσk2 ,

with Ř = c1 R̂−1 , which by the symmetry noted therefore solves (5). Thus the CCC prescription
yields an FLRW metric in the present aeon as well, and the Einstein equations are satisfied
with the ‘same’ density, i.e. ρ̌ = m/Ř4 , and same cosmological constant. In this rather special
example, in which the Weyl tensor is zero throughout, all aeons are diffeomorphic.

1 How do you define a unique Ω̂?

For CCC to be well-defined, we need an algorithm for fixing the geometry in the present aeon
from that in the previous aeon. By consideration of (3) that means we need an algorithm for a
unique Ω̂ given ĝ, since these then fix ǧ. What is this algorithm to be? Penrose [8] considers the
rule for transforming scalar curvature under conformal rescaling: if metrics g, ĝ are related by
ĝab = ϕ2 gab , so that ϕ = Ω̂−1 in our previous convention, and is zero and regular at I, then the

3
scalar curvatures – call these s, ŝ respectively, to avoid confusion with the FLRW scale factor –
are related by
b + ŝϕ = 1 sϕ3 ,
ϕ (7)
6
where  b is the d’Alembertian w.r.t. ĝ. This equation is discussed in [13] (and elsewhere of
course, starting with [8]): if one has a Starobinsky expansion of the metric ĝ, which is to say an
expansion in powers of eHt and which is not a strong assumption, then solutions of (7) can be
expanded in negative powers as
ϕ = Σn≥1 ϕn (xi )e−nHt ,
where xi are comoving spatial coordinates. Furthermore, ϕ1 , ϕ2 are freely specifiable data de-
termining all the other coefficients. Penrose imposes ϕ2 = 0 and for reasons explained in [8]
and [13], he calls this the delayed rest mass hypothesis. Then he gives a range of possibilities
for fixing ϕ1 without settling on one of them. In [13] I suggested a different possibility which is
to choose ϕ1 so that the scalar curvature of the metric induced on I by g is constant, which is
what one did in the FLRW example. This is equivalent to solving the Yamabe problem for I
and in favourable circumstances this has a unique solution (but not always!).
Another approach to defining a unique Ω̂ (or ϕ = Ω̂−1 which is a better behaved function in
M̂ ) would be to suppose that there is a conformal scalar field, say φ, in M vanishing at I. Then
one would choose the conformal scale to give φ the value one, which could be done throughout
M̂ but not at I. Then one needs to decide whether φ is a physical field contributing to the
Einstein equation via an enegy-momentum tensor. Something quite close to this was suggested
in [6] but it’s not yet entirely satisfactory.
However it is accomplished, this is a question that needs answering for CCC.

2 What is the field content of CCC classically?

In Penrose’s original version, [8], the matter content of the previous aeon close to I is assumed
to be a radiation fluid, possibly with massless fields like Maxwell, and certainly with a positive
cosmological constant. Earlier on, in the era of galaxies, there would be other contributions to
the matter content, like a pressure-free perfect fluid, but it was important for Penrose that these
dropped out near I. Without some explicit mechanism to make the dust contribution fade away,
its density is proportional (in the FLRW case) to R−3 so it comes to dominate the radiation
which has density proportional to R−4 .
It follows from (3) how to express the Einstein tensor, say Ǧab in M̌ , in terms of the Einstein
tensor Ĝab of M̂ and derivatives of Ω̂ up to second order. Thus one at once knows the Einstein
tensor and therefore the energy-momentum tensor and matter content of M̌ from M̂ . This
expression can be found in [8] and [13] but it’s complicated and hard to decode. In particular, it
won’t be anything as simple as it was in M̂ . However it must express the matter content of M̌
near the big bang and therefore needs to be understood. In [8], Penrose proposes to understand
the field Ω̂, or a function of it, as a conformal scalar field in the present aeon, and specifically
he wants to connect this field to dark matter. As an alternative, I’ve suggested that for reasons
of naturalness, if there is a conformal scalar in the present aeon then perhaps there should be
one in the previous aeon too, in fact in all aeons. This would connect with Lübbe’s suggestion
in [6] mentioned at the end of the previous section.
It’s worth emphasising that until one has a settled view on the field content of CCC classically,
one doesn’t have any certainty about what the Einstein equations should be in M̌ .

4
3 What is the field content of CCC as a QFT?
This is an area in which I have little expertise but Penrose has made a number of suggestions
that are worth drawing together. The most radical is that he speculates that the rest-mass of
all particles goes to zero on the approach to I, as something like the Higgs mechanism, which
gives mass to the particles in the early universe, turning the mass off again. After this mass
fade-out the matter content is massless, and one has a radiation stress-tensor.
Pulling the questions together:

• Is the QFT content represented by conformally-invariant theory near I?

• Given the presence of Λ, does the right QFT near I have the de Sitter group replacing the
Poincaré group? (Recall that the de Sitter group does not have a mass Casimir operator.)
• Can one implement mass fade-out? (Massless charged particles are usually regarded as
problematical.)
• Inflation requires a period of exponential expansion just after the big bang, while CCC
offers a period of exponential expansion just before the big bang. Can one make sense of
the statement inflation happens before the Bang, in the previous aeon? And if so, can one
then derive the familiar spectrum of density perturbations?

This section is all questions and no answers!

4 Magnetic fields through I?

Everywhere astronomers look there are magnetic fields: in the solar system, in stars, in galaxies,
in clusters of galaxies and between clusters of galaxies. Given the right initial magnetic fields (i.e.
post big bang), there are calculations in the literature, typically based on dynamo mechanisms,
to amplify the initial magnetic fields and get agreement with the presently observed values but
there is no consensus about the origin of the initial values. Since electromagnetism has good
conformal properties, can these initial values be understood as coming through from the previous
aeon?
As a simple model, consider a flat FLRW metric for ĝ:

ĝ = dt̂2 − (â(t̂))2 (dx2 + dy 2 + dz 2 )

then as we saw in the Introduction, ǧ is also spatially flat FLRW. Now consider the 2-form

B̂ = dx ∧ dy.

This is closed and co-closed, so it’s a source-free Maxwell field, and it’s orthogonal to the
Hubble flow ∂t̂ so it’s a (very simple) magnetic field. Clearly it extends through I with the
conventions B̌ = B̂ (which of course works more generally). Note we can confuse ourselves by
using orthonormal bases (e.g. ê1 = âdx etc in M̂ and corresponding quantities in M̌ ) when, on
the two sides,
B̂ = (â)−2 ê1 ∧ ê2 , B̌ = (ǎ)−2 ě1 ∧ ě2
so |B̂| goes to zero at I and |B̌| diverges there...
Does this seamless passage through I work with more realistic fields (e.g. ones with currents
as sources)? Also, when the magnetic field emerges on the hot side (the present aeon) is it washed
out by astrophysical processes in its physical surroundings? (There is a conformally-invariant
form of MHD , see [10] or an account in [13], which might help answer this.)

5
5 Circles in the sky and Hawking points
There are two sets of observations, [4, 5] and [1, 2], claiming observational support for CCC by
finding traces in the CMB maps of events in the previous aeon. The idea for circles in the sky
is that a merger of supermassive black holes late in the previous aeon would send out a burst of
gravitational radiation approximately confined between expanding concentric spheres; this would
pass through I and register on the last scattering surface; our past light cone would intersect
this spherical annulus in an annulus on our sky; thus one should seek annular temperature
anomalies. Furthermore, if the events are due to a series of smaller black holes merging with a
single very large black hole, one would expect to see the annuli grouped into sets of concentric
rings. The two collaborations have different methodologies and different statistical analyses, but
they claim rather similar (and positive) findings.
The idea of Hawking points arises from consideration of the Hawking evaporation of large
black holes in the previous aeon. As we noted in the Introduction, from a time about ten times
the current age of the universe, the whole infinite future is very close to I in conformal time.
Thus the process of evaporating a black hole of say 1010 solar masses, which would take 10100
years in proper time2 , happens almost at I and all the energy from this radiation is dumped
into a small sphere on I. It will spread out in M̌ but register on the last scattering surface only
in a ball of radius about 1.6 degrees on our sky3 . These are what Penrose and collaborators [2]
call Hawking points and what they claim to have detected in the same places on the sky in the
Planck data and the WMAP data.
All these observations continue to be controversial and the debate is almost always about
statistical significance, since the occurence of the circles and points is usually agreed on by all
parties. In the context of this article the questions that arise are about the astrophysics of
it all: how energetic would a black hole merger in the previous aeon need to be in order to
produce the observed effect in the CMB? They need to be bright enough to be as detected but
not so bright as to be visible in the day-time! Also, can one use the observations to construct
a detailed picture of the black-hole population in the previous aeon – how many, how big and
how distributed?
The distribution of Hawking points on the sky claimed in [2] is strikingly inhomogeneous, as is
the distribution of centres of four or more concentric rings in [4]. Nonetheless this inhomogeneity
is not associated with a disturbance in the CMB homogeneity so one can calculate a consistency
check: assume the FLRW model and assume that a supermassive black hole of mass say M
dumps all its mass effectively into a point p (or at least, a small region around p) on I, then
this mass M will be confined to the ball B of coordinate radius r = τLS on the last scattering
surface which is the intersection of J + (p), the future of p, with the last scattering surface, which
is the 3-surface τ = τLS . Knowing the density ρ of the present aeon we can calculate the mass
in B due to the (averaged out FLRW) matter. In [13] I obtained the answer 1018 solar masses
for this but it depends on the (observed) parameters entering ρ; now provided M is less than
1013 solar masses then the perturbation in ρ is less than one in 10−5 which is what one needs.
It is the case that the largest known black holes in the present aeon have mass about 1010 solar
masses but there don’t seem to be reasons in principle why an earlier aeon shouldn’t have much
bigger ones. This leads us into the next section.
2
2 × 1097 years according to Wikipedia. √
3
the ball is small and so subtends at us an angle in radians of approximately 2τLS (τ0 − τLS )−1 where τLS
and τ0 are conformal time at last scattering and now, and these are given in [11].

6
6 Can one construct aeon-to-aeon maps?
The last two sections have been concerned with how massless fields come through from one aeon
to the next. Both these notions fall under the rubric of aeon-to-aeon or A2A maps: can one give
an algorithm for the matter density distribution, magnetic field strength and other quantities of
interest in the present aeon from knowledge of corresponding quantities in the previous aeon?
This would be an A2A map, and it would rely on more-or-less conventional relativistic cosmology
and relativistic astrophysics applied to the (correct) equations of CCC.
At first sight, CCC does not explain the approximate homogeneity and isotropy of the
universe. As noted at the end of the last section, if there was a super-duper-massive black hole
in the previous aeon it would certainly produce an inhomogeneity in the matter density in the
present aeon (this was pointed out to me in email correspondence with Sabine Hossenfelder and
Tim Palmer, which I gratefully acknowledge). The calculation mentioned above suggests that
a black hole of 1018 solar masses in the previous aeon could give δρ/ρ ∼ 1 at last scattering in
the present aeon. However the period between I and the last scattering surface gives enough
time to smooth out some lesser inhomogeneities on I, which are already reduced by the long
(infinite in proper time) period of exponential expansion in the previous aeon. For this reason
the rather striking inhomogeneity in the distribution of Hawking points on the sky found in [2]
does not lead to a corresponding inhomogeneity in the matter density measured in the CMB.
Once one has the A2A map one can think about iterating it and ask whether it gives con-
vergence to an almost-homogeneous and almost-isotropic universe? (or indeed to something
else!) Since the example of the radiation-filled FLRW universe seems, by the calculation in the
Introduction to be a fixed-point of the simplest A2A map one could speculate that convergence
to something near this was a good bet. However this certainly is speculation, at least for now.

References
[1] Daniel An, Krzysztof A. Meissner and Pawel Nurowski, Ring Type Structures in the Planck
map of the CMB, 2018, MNRAS, 473, 3251 10.1093/mnras/stx2299 , arXiv:1510.06537
and
Meissner K. A., Nurowski P., Ruszczycki B, Structures in the microwave background radiation
2013, Proc. R. Soc. A, 469, 2155, doi 10.1098/rspa.2013.0116, arXiv:1207.2498

[2] Daniel An, Krzysztof A. Meissner, Pawel Nurowski and Roger Penrose, Appar-
ent evidence for Hawking points in the CMB Sky, MNRAS, 2018 495 3403–3408,
doi.org/10.1093/mnras/staa1343, arXiv:1808.01740

[3] Helmut Friedrich, Existence and structure of past asymptotically simple solutions of Ein-
stein’s field equations with positive cosmological constant. J. Geom. Phys. 3 (1986) 101–117
and
On the global existence and the asymptotic behavior of solutions to the Einstein-Maxwell-
Yang-Mills equations. J. Differential Geom. 34 (1991) 275–345.

[4] V. G. Gurzadyan and R. Penrose, CCC-predicted low-variance circles in CMB sky and
LCDM, arXiv:1104.5675
V. G. Gurzadyan and R. Penrose, Concentric circles in WMAP data may provide evidence of
violent pre-Big-Bang activity, arXiv:1011.3706

[5] V. G. Gurzadyan and R. Penrose, On CCC-predicted concentric low-variance circles in the

CMB sky, 2013, Eur. Phys. J. Plus, 128, 22, arXiv:1302.5162

7
[6] Christian Lübbe, Conformal scalar fields, isotropic singularities and conformal cyclic cos-
mologies, arXiv:1312.2059

[7] Christian Lübbe and Paul Tod, An extension theorem for conformal gauge singularities,
J.Math.Phys.50:112501,2009, arXiv:0710.5552

[8] R Penrose, Cycles of Time: An Extraordinary New View of the Universe, Bodley Head 2010

[9] R Penrose, Singularities and time-asymmetry, in General Relativity: An Einstein Centenary

Survey, eds. Hawking S W and Israel W, (Cambridge: Cambridge University Press) 1979

[10] K Subramanian and J Barrow, Magnetohydrodynamics in the Early Universe

and the Damping of Non-linear Alfven Waves , doi 10.1103/PhysRevD.58.083502 ,
arXiv:astro-ph/9712083

[11] Paul Tod, Penrose’s circles in the CMB and a test of inflation, Gen. Relativity Gravitation
44 (2012), 2933–2938, arXiv:1107.1421

[12] Paul Tod, The equations of CCC, Gen. Relativity Gravitation 47 (2015), no. 3, Art. 17, 13
pp, arXiv:1309.7248

[13] Paul Tod, Conformal Methods in General Relativity with application to Conformal Cyclic
Cosmology: A minicourse given at the IXth IMLG Warsaw 2018, arXiv:2102.02701