The Conformal Transformation's Controversy What Are We Missing

The conformal transformation’s controversy: what are we missing?
Israel Quiros,1, a Ricardo Garcı́a-Salcedo,2, b Jose Edgar Madriz Aguilar,1, c and Tonatiuh Matos3, d
1
Departamento de Matemáticas, Centro Universitario de Ciencias Exáctas e Ingenierı́as (CUCEI),
Corregidora 500 S.R., Universidad de Guadalajara, 44420 Guadalajara, Jalisco, México.
2
Centro de Investigacion en Ciencia Aplicada y Tecnologia Avanzada (CICATA), Legaria del IPN, México D.F., México.
3
Departamento de Fı́sica, Centro de Investigación y de Estudios Avanzados del IPN, A.P. 14-740, 07000 México D.F., México.
(Dated: March 22, 2018)
An alternative interpretation of the conformal transformations of the metric is discussed according
to which the latter can be viewed as a mapping among Riemannian and Weyl-integrable spaces. A
novel aspect of the conformal transformation’s issue is then revealed: these transformations relate
complementary geometrical pictures of a same physical reality, so that, the question about which
is the physical conformal frame, does not arise. In addition, arguments are given which point out
that, unless a clear statement of what is understood by ”equivalence of frames” is made, the issue
arXiv:1108.5857v4 [gr-qc] 5 Nov 2012
is a semantic one. For definiteness, an intuitively ”natural” statement of conformal equivalence is

given, which is associated with conformal invariance of the field equations. Under this particular
reading, equivalence can take place only if the metric is defined up to a conformal equivalence class.
A concrete example of a conformal-invariant theory of gravity is then explored. Since Brans-Dicke
theory is not conformally invariant, then the Jordan’s and Einstein’s frames of the theory are not
equivalent. Otherwise, in view of the alternative approach proposed here, these frames represent
complementary geometrical descriptions of a same phenomenon. The different points of view existing
in the literature are critically scrutinized on the light of the new arguments.
PACS numbers: 02.40.-k, 02.40.Ky, 02.40.Hw, 04.20.-q, 04.20.Cv, 04.50.Kd, 04.50.+h, 11.25.Wx
I. INTRODUCTION The theory can be formulated in different sets of field

variables [2], among which we may cite BD gravity in
Brans-Dicke (BD) theory of gravity [1] represents the Jordan’s frame (JF) variables – the standard formulation
most simple generalization of general relativity (GR). of the theory [1] given by equations (1), (2), (3) – and BD
The theory is parametrized by a constant parameter ω – theory in the so called Einstein’s frame (EF) [2], which
the BD coupling. Up to a boundary term in the action can be obtained from the Jordan frame formulation by a
the vacuum BD field equations can be derived from: conformal transformation of the metric ḡµν = eϕ gµν . For
vacuum, up to a boundary term in the action, the latter
formulation of the theory can be derived from,
1 √
Z
d4 x −g eϕ R − ω(∇ϕ)2 ,

SBD = (1)
16π √

1 3
Z
S̄BD = d4 x −ḡ R̄ − ω + ¯ 2 ,
(∇ϕ) (4)
where R is the curvature scalar, (∇ϕ)2 ≡ g µν ∇ν ϕ∇µ ϕ, 16π 2
and we have introduced the BD scalar field ϕ which is re- where the over-bar means the quantities are given in
lated to the original one [1] through φ = eϕ . The derived terms of the conformal metric ḡµν . The derived field
field equations – which model the BD laws of gravity – equations, plus the KG equation are,
are

3 ¯ ¯ 1 ¯ 2

1
Ḡµν = ω+ ∂µ ϕ∂ν ϕ − ḡµν (∇ϕ) , (5)
Gµν = ω ∂µ ϕ∂ν ϕ − gµν (∇ϕ)2 + ∂µ ϕ∂ν ϕ 2 2
2
and
−gµν (∇ϕ)2 + ∇µ ∇ν ϕ − gµν ϕ, (2)
where ≡ g µν ∇µ ∇ν , and the Klein-Gordon (KG) equa- ¯ = 0,

ϕ (6)
tion for the scalar field:
respectively. Equivalence of JF and EF formulations of
BD theory of gravity under conformal transformations of
(2ω + 3)[ϕ + (∇ϕ)2 ] = 0. (3) the metric, also known as transformations of units [2],
ḡµν = Ω2 gµν , (7)

a Electronic address: [email protected]
b Electronic address: [email protected]
where Ω2 = eϕ , has been discussed in the literature since
c Electronic address: [email protected] long ago [2, 3], and more recently has been put to dis-
d Electronic address: [email protected] cussion again in connection with scalar-tensor theories of
2
gravity [4–11]. In spite of the amount of work published which can also be written in more standard notation [6,
on this subject to date (see the review [6]), the contro- 10] if make the replacement ϕ → 2 ln ξ ⇒ ξ − R ξ/6 =
versy is still open [12–15]. The discussion in the following 0. The difficulty with conformal invariance is originated
paragraphs will support this statement.1 from the transformation properties of the D’lambertian
Even if there can be different points of view on the sub- under (7), (8);
ject, it seems intuitively ”natural” to associate conformal
equivalence with invariance under the conformal transfor-
¯ ϕ̄
mations (7). If one follows this intuitive understanding of Ω−2 ϕ = +
conformal equivalence, since BD theory is not conformal- 1 − 2∂ϕ ln Ω
invariant, then there can not be dynamical equivalence
" #
∂ϕ2 ln Ω − ∂ϕ ln Ω(1 − 2∂ϕ ln Ω)
among the Jordan’s and Einstein’s frames. This is eas- 2 ¯ ϕ̄)2 ,
(∇
ily demonstrated in the simplest case of vacuum the- (1 − 2∂ϕ ln Ω)3
ory. In fact, under a conformal transformation (7), with
Ω2 = eϕ , the vacuum JFBD action (1) is mapped into the so that the KG equation (3) transforms into:
EFBD action (4), while the JF vacuum field equations ( " # )
(2), (3), are transformed into the vacuum field equations 2∂ϕ2 ln Ω
¯
(2ω̄ + 3) ϕ̄ + 1 + ¯
(∇ϕ̄)2
= 0.
in the EF: Eq.s (5), and (6). I. e., the laws of gravity – (1 − 2∂ϕ ln Ω)2
expressed through the field equations – are not invariant
under (7). This latter equation shares no resemblance with Eq. (3),
The situation is less clear when additional field and unless, Ω2 = e2kϕ , where the constant k is a real number
coupling constant redefinitions are invoked. In this lat- (k 6= 1/2). For the remaining gravitational field equa-
ter case it has been stated in the literature [18] that the tions (Eq.(2)), it is a matter of uncomplicated algebra to
action (1) is invariant under (7), plus the following redef- show that, unless ω = −3/2, in which case (2) transform
initions (see also [6, 19, 20]): into,
1 1
ω + 6∂ϕ ln Ω(1 − ∂ϕ ln Ω) Gµν = − [∂µ ϕ∂ν ϕ + gµν (∇ϕ)2 ] + ∇µ ∇ν ϕ − gµν ϕ,
ϕ̄ = ϕ − 2 ln Ω, ω̄ = . (8) 2 2
(1 − 2∂ϕ ln Ω)2
these are not invariant under (7), (8), neither. Since the
Contrary to existing claims it can be shown that, as a vacuum BD action (1) is invariant under (7), (8), while,
matter of fact, vacuum BD theory is not invariant under in general (arbitrary ω 6= −3/2), the field equations de-
the transformations (7), (8), unless ω 6= −3/2. Actually, rived from that action – Eq.s (2), (3) – are not invariant
even if it is true that the BD vacuum action (1) is in- under these transformations, then the conformal invari-
variant under (7), (8), this does not imply that the field ance of vacuum BD theory is, at most, a mirage or spu-
equations of the theory are also invariant under these rious symmetry. As a result, the dynamics is different in
transformations. To demonstrate this it suffices to write the different conformal frames, i. e. the JF and EF of
the vacuum KG equation (3) in the equivalent form: vacuum BD theory are not conformally equivalent rep-
resentations just as it happens with Brans-Dicke theory
1 R with matter sources.
ϕ + (∇ϕ)2 + = 0.
2 2ω A question then arises: what do authors who advocate
This equation is clearly not conformally invariant, since physical (and/or mathematical) equivalence among JF
the only conformal-invariant vacuum KG equation is nec- and EF representations, actually understand by equiv-
essarily of the form (note that this corresponds to the alence? Do they relate equivalence with an actual dy-
choice ω = −3/2 in the last equation), namical symmetry of the theory? We think this is a
non enough explored aspect of the conformal transfor-
1 R mation’s issue that deserves being discussed. Most part
ϕ + (∇ϕ)2 − = 0,
2 3 of the misunderstanding arising within this context is
due, precisely, to a lack of a clear definition of what to
understand by ”equivalence”. To worsen things, we will
show that there is an aspect of the issue that has not
1 Needless to say that the conformal transformation’s issue is criti- been discussed so far. It is connected with the possibility
cal for the interpretation of the predictions of given scalar-tensor
theories of gravity since these are deeply affected by the choice
to assign a different geometrical meaning to conformal
of the conformal frame [4, 6, 8, 9, 15]. It is of central importance transformations of the metric. Aim of this paper is to
also for the understanding of the physics behind the graviton- provide new arguments that might help winning a dip-
dilaton string effective theory [16] since, independent of the di- per understanding of this thorny subject.
mensionality of the spacetime and the number of compactified The novel aspect of the conformal transformation’s is-
dimensions, the string frame (SF) dilaton-gravity action is noth-
ing but JFBD action with, ω = −1 (see, however, Ref.[17]). The sue we will uncover in the first part of this paper (mainly
string effective theory may be formulated in a number of confor- in sections III, IV and Sec.V) is originated from an alter-
mal frames as well, including the SF and the EF among others. native interpretation of the conformal transformations of
3
the metric, according to which, under (7) not only the gauge is considered, the metric is not uniquely deter-
dynamical equations of BD theory are transformed but, mined by the field equations. It is clear that BD theory
simultaneously, the affine properties of space: the con- does not belong in this group, since no specific gauge is
nection, the geodesics, etc, are also modified (a fact that necessary to get a closed system of equations determining
is usually dismissed [2, 9–15]). According to this ap- the metric and the BD field: the Einstein-Brans-Dicke
proach, modification of the above mentioned affine prop- plus the Klein-Gordon equations form a closed system of
erties of space is reflected in that, in terms of the original equations in the gµν , ϕ - variables. We will discuss this
metric, for instance, the units of length may be point- topic in detail in Sec.VI B.
independent – as it is for Riemannian spaces – while, In view of new arguments explored here: i) transforma-
in terms of the conformal metric, the length units may tions (7) can be understood as a mapping among Rieman-
be point-dependent instead. Although the above fact nian and Weyl-integrable spaces, and, ii) the notion of
has been partially considered in the seminal paper by ”conformal equivalence” is to be endowed with a concrete
Dicke [2], and more recently in [10] (see also [11]), it has mathematical and physical meaning, a critical review of
passed unnoticed the fact that geometry with running the different viewpoints existing in the bibliography on
(also changing or point-dependent) units of length can the issue is performed in Sec.VII. Other relevant aspects
not be Riemannian but Weyl-integrable geometry2 in- of the conformal transformations conundrum, such as the
stead [21]. In this understanding conformal transforma- positivity of the energy problem, as well as its implica-
tions (7) can be viewed as a mapping from Riemannian tions for the singularity issue, will be discussed also (see
into Weyl-integrable spaces and vice versa, or, in other sections V and VIII respectively). Although in this pa-
words, these can be properly understood as units trans- per, for simplicity, the discussion mostly relies on (vac-
formations in the sense of Ref.[2]. The resulting trans- uum) BD theory, the results of our study can be straight-
formations of units relate complementary equally suited forwardly applied to scalar-tensor theories in general (see
geometrical pictures of a same physical reality. appendix B).3 In the next section, in order for the paper
to be self-contained, the fundamentals of Weyl geometry
Another important aspect of the conformal transfor- (WG) are exposed. Weyl-integrable geometry (WIG) is
mation’s issue that has to be carefully stated, no mat- a particular member in this latter class and is important
ter how obvious it seems, is to agree on what is to be for the novel understanding of the conformal transforma-
meant by physical/mathematical equivalence among the tions we shall reveal here.
different conformal frames in which a given gravity the-
ory can be formulated. As discussed in the former page,
unless a clear and mathematically meaningful statement
of the latter notion is given, the conformal equivalence II. FUNDAMENTALS OF WEYL GEOMETRY
issue is no more than a semantic matter. In this pa-
per, for definiteness, conformal equivalence is linked with Before we pursue the present discussion any further we
invariance of the field equations under Weyl rescalings do a step aside to expose the fundamentals of the simplest
(see Eq.(15) below), which include conformal transfor- generalization of Riemannian geometry that is able to ac-
mations (7) plus a scalar field redefinition. Even if there commodate running units: Weyl geometry. For readable
can be other semantic uses of the word ”equivalence”, and pedagogical introduction to WG we recommend the
both from the mathematical and physical stand points it classical books [22], [23], and [24], however research pa-
seems to us quite natural to associate conformal equiv- pers can be found where the subject is exposed in a more
alence with conformal invariance. The latter property or less pedagogical way [25–28]. It has to be said that,
is comprised in what we call here as ”conformal equiv- although there was a moderate revival of Weyl’s ideas
alence principle” (CEP), a clear statement of which is after Dirac’s ”large numbers hypothesis” [29] (see also
given in section VI. Whether or not this is actually a [30]), in the last decades there has been a renewed in-
fundamental principle of nature is not of importance to terest in WG [27, 28, 31–37], in connection to the search
the results of this paper and will not be discussed here. In for alternative explanations to outstanding questions of
that section we shall explore a conformal-invariant theory fundamental physics, such as the dark matter/dark en-
as an example where the different conformal frames, in ergy issues. A Weyl space (M, gµν , wµ ), is a manifold M
which the theory can be formulated, are actually (phys- endowed with a metric gµν and a (gauge) vector field wµ ,
ically/mathematically) equivalent. It will be demon- so that the following “metricity” condition is satisfied:
strated that, in general, the mentioned equivalence can
be achieved only within theories where the metric tensor
is defined up to an (conformal) equivalence class of met- ∇(w)
µ gαβ = −wµ gαβ , (9)
rics. In the framework of such theories, unless a specific
3 Here we deal with classical theories of the gravitational field.

2 Here by ”geometry” we understand the affine properties of space In consequence, arguments related with quantum properties and
such as, the affine connection, the geodesics, etc. processes will not be considered.
4
(w)
where ∇µ is the Weyl covariant derivative operator, correspond to a particular gauge [27] where ϕ = ϕ0 =
which is defined through the torsion-free affine connec- (w)
const ⇒ ∇µ → ∇µ , and the metric is convariantly
tion of the Weyl space: constant:
α α 1 α ∇µ gαβ = 0, (13)
δβ wγ + δγα wβ − gβγ wα ,

Γβγ = {βγ }+ (10)
2
where ∇µ refers to Riemannian covariant derivative oper-
where ator defined through the Christoffel symbols. As a conse-
quence, under parallel transport in a Riemannian space,
α 1 αν vectors get rotated but their length is unchanged, i. e.,
{βγ }= g (∂β gνγ + ∂γ gνβ − ∂ν gβγ ) , (11) length units are truly constant.
2
An interesting feature of WI spaces is that the metric-
are the Christoffel symbols of the metric. For every (non- ity condition (Eq.(9) with the replacement wµ → ∂µ ϕ),
vanishing) differentiable function Ω, the affine connection
(10), and the metricity condition (9), are invariant un-
der the following transformations, also known as Weyl ∇(w)
µ gαβ = −∂µ ϕ gαβ , (14)
rescalings:
the affine connection (Eq.(18) below), and several other
geometric objects, are invariant under the following Weyl
ḡµν = Ω2 gµν , w̄µ = wµ − 2∂µ ln Ω. (12) rescalings5
Thus the metric gµν , and the gauge vector wµ are far
ḡµν = Ω2 gµν , ϕ̄ = ϕ − 2 ln Ω. (15)
from unique: rather gµν belongs in an equivalence class
of metrics g, so that, for each gµν ∈ g, there exists a Hence, the metric gµν and the gauge scalar ϕ are far from
unique gauge vector wµ , such that the law (9) is satis- unique. Instead of a fixed pair (gµν , ϕ) – properly a gauge
fied. A given pair (gµν , wµ ) is called a gauge, and the – one has a whole (perhaps infinite) class of pairs
transformations (12) are gauge transformations [38].
Due to (9), under parallel transport, not only the ori- n o
entation
p of a given vector changes, but, also its length C = (gµν , ϕ)|∇(w)
µ g αβ = −∂µ ϕ g αβ , (16)
ℓ = gµν ℓµ ℓν , varies from point to point in the Weyl
manifold: dℓ/ℓ = dxµ wµ /2. Hence, for instance, upon such that, any other pair (ḡµν , ϕ̄) related with (gµν , ϕ)
returning back to the starting point, after parallel trans- by a Weyl rescaling (15), also belongs in C.
port in a closed path,H the length of a vector will not be It is sometimes useful to write several geometric ob-
the same, ℓ = ℓ0 exp dxν wν /2. This feature of WG led jects like, for instance, the WI curvature scalar R(w) ,
Einstein to argue that electrons moving in a background (w) (w)
Ricci tensor Rµν and Einstein’s tensor Gµν , respec-
of the wµ -field would produce unobserved broadening of tively, in terms of their Riemannian counterparts:
the atomic spectral lines (see, however, arguments that
overcome Einstein’s objection [31]). This broadening of
the spectral lines is known as the “second clock effect”. 3
R(w) = R − 3ϕ − (∂ϕ)2 ,
2
(w) 1
Rµν = Rµν − ∇µ ∇ν ϕ − gµν ϕ
A. Weyl-integrable geometry 2
1
+ [∂µ ϕ∂ν ϕ − gµν (∂ϕ)2 ],
There is a particular subclass of WG which is free from 2
the second clock effect; the so called Weyl-integrable G(w)
µν = Gµν − ∇µ ∇ν ϕ + gµν ϕ
(WI) geometry.4 The latter can be obtained from the 1 1
general class of WG-s by replacing the gauge vector by + [∂µ ϕ∂ν ϕ + gµν (∂ϕ)2 ], (17)
2 2
a gradient of a scalar field: wµ → ∂µ ϕ. In general
WI spaces can be represented by the triad (M, gµν , ϕ). where, in the right-hand-side (RHS) of the above equa-
In this case the length unit ℓ changes according to, tions, stand usual Riemannian magnitudes, including
dℓ/ℓ = dxµ ∂µ ϕ/2 = dϕ/2, soH that, after parallel trans- the curvature scalar R, the Ricci tensor Rµν , the Ein-
port in a closed path, since
H dϕ = 0, there is no neat stein’s tensor Gµν = Rµν − gµν R/2, the covariant deriva-
change in the length unit dℓ/ℓ = 0. Riemann spaces tive operator ∇µ , and the D’lambertian operator ≡
4 For applications of WIG in cosmology see, for instance, the re- 5 In this paper sometimes we shall call as ”scale invariance” in-
view [39], and also Ref.[40]. variance under the Weyl rescalings (15).
5
g µν ∇µ ∇ν , which are defined in terms of the Christof- us assume we apply a transformation (7) on the metric
fel symbols (11). Weyl-integrable curvature quantities, gµν of a Riemann’s space. This means, in particular, that
instead, are defined in terms of the WI affine connection the connections of the starting manifold coincide with the
Christoffel symbols of the metric (11), and, consequently,
that the Riemann “metricity” condition (13) is satisfied,
α α 1 α
δ ∂γ ϕ + δγα ∂β ϕ − gβγ ∂ α ϕ .

Γβγ = {βγ }+ (18) resulting in that the length units in the starting space
2 β are point-independent. Under (7) the Christoffel sym-
(w) (w) bols transform as:
The WI Ricci and Einstein’s tensors Rµν and Gµν are
unchanged by the Weyl rescalings (15), while R̄(w) =
¯
Ω−2 R(w) , so that the scale-invariant measure of scalar
α
µν = α
µν − Ω−1 δµα ∂ν Ω
curvature is the quantity, e−ϕ R(w) . Note, in between,
+δνα ∂µ Ω − ḡµν ḡ ασ ∂σ Ω) . (22)
that the quantity, eϕ/2 ds, is a scale-invariant measure of
spacetime separations. Other scale-invariant quantities If one compares this equation with Eq.(18), where the
of WIG are: affine connection of a WI space is defined, one is left
with two possibilities to build an affine structure into the
(w) µν (w)
αβµν conformal space.
e−2ϕ Rµν R(w) , and, e−4ϕ Rαβµν R(w) . (19)
Time-like geodesics in a WI space are described by the A. First point of view: Riemann7→Riemann
following scale-invariant equation [25]:
One possibility is just to regard the conformally re-
dxα dxµ dxν dxµ dxα lated manifolds as endowed with different Riemannian

d 1
+ Γα
µν − ∂µ ϕ = 0, (20) structure of the same conformal class, so that Eq.(22)
ds ds ds ds 2 ds ds ¯
is just the transformation law relating α

µν with αµν
where, as before, Γαµν is the affine connection of the WI under (7). Assuming this interpretation – the point of
space (18), and the third term in the LHS of the equation view adopted by most researchers in the field – then the
is originated from variations of the units of length from Riemannian metricity condition (13) is unchanged, i. e.
point to point in the manifold. The latter term can be in the conformal space,
removed by an appropriate affine parametrization σ =
σ(s) ⇒ dσ = eϕ/2 ds, so that the above geodesic equation
can be rewritten in the standard way: ¯ µ ḡαβ = 0,
∇ (23)
so that the affine properties of space are not modified. In

dxα dxµ dxν

d this case the transformation (7) is just a mapping from
+ Γα
µν = 0. (21)
dσ dσ dσ dσ Riemann’s space into Riemann’s space:
The null geodesic equation is similar to Eq.(20) but with

replacement of ds → dλ, where λ is an affine parameter γRR : Riemann 7→ Riemann ⇔
along the null geodesic path. Unlike the standard case (M, gµν ) 7→ (M, ḡµν ). (24)
discussed in the bibliography (see, for instance [10, 41]),
here the affine parameter λ shares the same transforma- Adopting this point of view amounts to consider that
tion properties with the interval ds under (7), namely the units of length in the conformal space are point-
dλ̄ = Ω dλ. independent as well. This is the most widespread view-
point and it is not consistent with considering (7) as a
transformation of units in the sense of [2]. Under this
III. TWO FACES OF CONFORMAL understanding, it happens, for instance, that time-like
TRANSFORMATIONS geodesics in the starting Riemannian space,
In this section we will explore an aspect of the con- d

dxα

α dxµ dxν
formal transformation’s issue which has not been dis- + µν = 0, (25)
ds ds ds ds
cussed so far. Transformations (7) are usually under-
stood as a mapping from Riemannian spaces into Rie- are mapped into time-like curves which are not geodesics
mannian spaces (first viewpoint below). Former studies in the conformal (also Riemannian) space,6
of the issue have been performed under the implicit as-
sumption that this interpretation of (7) is the only pos-
sible. Here we will show that an alternative geometric
interpretation of the conformal transformations is indeed 6 A different point of view on this property is exposed in [10] (see,
possible (point of view exposed in subsection III B). Let however, the discussion in Sec.VII B).
6
notice that while the metricity condition (27) is invariant

α
µ ν under the following Weyl rescalings (compare with (15)):
d dx ¯ dx dx
α
+ µν =
ds̄ ds̄ ds̄ ds̄ ḡαβ → ḡ¯αβ = λ2 ḡαβ , Ω → Ω̄ = Ω − λ,
µ α
∂µ Ω dx dx µα
− ḡ . (26) the Riemannian metricity condition (23) does not obey
Ω ds̄ ds̄
this symmetry. The consequence is that, according to the
Actually, it can be shown that, by means of an appropri- first point of view displayed in Eq.(24), a unique metric
ate parametrization, one could remove the first term in tensor ḡαβ is single out, meanwhile, according to Eq.(28)
the right-hand-side (RHS) of Eq. (26) (see, for instance, – since, as mentioned, the WI metricity property (27) is
Ref.[42]), however, the second term can not be elimi- invariant under the above Weyl rescalings – one is faced
nated [10]. In other words, the above equation does not with a whole equivalence class of conformal metrics in-
admit an affine parametrization whatsoever, signaling a stead. In the second place, it is not difficult to prove
truly non-geodesic character of (26). From the physical that, according to (28), under (7), Riemannian time-like
point of view, i. e., if identify the latter equation with geodesics (25) are mapped into time-like geodesics of the
the equation of motion of a test point-particle, the sec- conformal Weyl-integrable space:8
ond term in the RHS can be identified with an additional
force of non-gravitational origin acting on the test parti-
dxα dxµ dxν ∂µ Ω dxµ dxα

cle, commonly called ”five-force”. The latter happens to d
+ Γ̄α
µν + = 0, (29)
be an actual property of the laws of motion of a time-like ds̄ ds̄ ds̄ ds̄ Ω ds̄ ds̄
particle in the conformal frame.
or, if choose an appropriate affine parametrization, σ̄ =
σ̄(s̄) ⇒ dσ̄ = Ω−1 ds̄, the above equation can be rewritten
B. Second point of view: Riemann7→Weyl in the more standard way (see Sec.II A):
The second possibility – not explored so far in connec- dxα dxµ dxν

d
tion with the conformal transformation’s issue – can be + Γ̄α
µν = 0. (30)
dσ̄ dσ̄ dσ̄ dσ̄
consistently matched with the interpretation of (7) as a
transformation of units in the sense of Ref.[2]. It is based This is to be contrasted with the usual understanding of
on the following subtlety: take a second look at Eq.(22), the conformal transformations – displayed by Eq.(24) –
and then, by comparing with (18), notice that one can according to which, Riemannian time-like geodesics (25)
safely identify the RHS of (22) with the definition of the are mapped into curves (26), which do not admit an
affine connection, affine parametrization whatsoever and, hence, can not
α ¯ } − Ω−1 δ α ∂ν Ω + δ α ∂µ Ω − ḡµν ∂¯α Ω , be geodesics (this is clearly demonstrated, for instance,
Γ̄µν α
≡ {µν µ ν in Ref.[10]). This subtlety and the resulting alternative
of a conformal WI space (M, ḡµν , Ω). Then, under (7), interpretation of the conformal transformation (7) dis-
played in Eq.(28), has not been explored before in con-
¯ (w)
α α
µν → Γ̄µν , ∇µ → ∇ µ , so that the Riemannian nection with the conformal transformation’s issue. The
metricity condition (13) transforms into the WI metricity consequences of this novel aspect of the issue for gravity
condition of the conformal space (compare with Eq.(23)): theories (BD theory in particular) is one of the subjects
that will be investigated in the following sections.
¯ (w) ḡαβ = 2Ω−1 ∂µ Ω ḡαβ ,
∇ (27)
µ
where the conformal factor Ω plays the role of the gauge IV. CONFORMAL TRANSFORMATIONS:
scalar of the WI space.7 According to this viewpoint, RIEMANN7→RIEMANN VS RIEMANN7→WEYL
under the conformal transformation (7), the original Rie-
mannian space is mapped into a conformal WI space: It is obvious that both interpretations of the conformal
transformations (7): γRR (Eq.(24)), and γRW (Eq.(28)),
are mathematically correct, however, both have different
γRW : Riemann 7→ Weyl ⇔ geometrical (and physical) implications. In this section
(M, gµν ) 7→ (M, ḡµν , Ω). (28) we will be assuming we deal with theories which are not
conformally invariant, i. e. either the field equations are
That Eq.(27) is not just a convenient rewriting of Eq.(23) transformed by the conformal transformation (7), which
can be straightforwardly demonstrated. In the first place,
8 Compare with WI geodesic equations given in Eq.(20). Complete

7 Notice that if one suppresses the over bar and identifies ϕ ≡ resemblance with (20) is obtained if set, Ω2 = e−ϕ ⇒ ∂µ Ω/Ω =
ln Ω−2 , then, there is full resemblance with Eq.(14) of Sec.II A. −∂µ ϕ/2.
7
means that the laws of gravitation are different in the

different conformal frames (first point of view), or, fol-
lowing the second viewpoint, the laws of gravity look the
same but the geometrical interpretation is different in the 3 1
Ḡµν = ω+ ∂¯µ ϕ∂¯ν ϕ − ḡµν (∇ϕ)
¯ 2 ,
different conformal spaces. Examples are BD theory and 2 2
scalar-tensor theories of gravity (general relativity also ¯
ϕ = 0. (32)
belongs in this group).
Both sets of dynamical equations (31) and (32), are to

be tied to Riemann’s spaces (M, gµν ), and (M, ḡµν ) re-
A. First viewpoint: physical implications spectively. Hence, JFBD and EFBD are just different
theories related by a particular mathematical transfor-
mation Eq.(7). In the first case – JFBD theory given
According to the interpretation discussed in Sec.III A by equations (1), (31) – both fields, gµν , and ϕ, deter-
(see Eq.(24)), the transformations (7) can be viewed as a mine the strength of gravity, while, in the second case –
mapping among Riemannian spaces, γRR : (M, gµν ) 7→ EFBD theory depicted by Eqs.(4), (32) – only the metric
(M, ḡµν ). Consequently, from the point of view of its field propagates the gravitational interaction. The re-
geometrical implications, Eq.(7) relates constant units maining scalar field ϕ, in this case, represents a matter
of length in the starting space with constant units in source. Therefore, the (vacuum) EFBD theory is just
the conformal space. The above interpretation of (7) is general relativity with a (perhaps exotic) scalar matter
not compatible with understanding conformal transfor- source. If there were additional matter fields (other than
mations of the metric as a transformation of units in the the scalar), due to non-minimal coupling to the scalar
sense of [2]. If we follow this viewpoint, since geodesics of field, these would interact non-gravitationally with ϕ.
the starting Riemann’s space are transformed into non- This additional interaction is what should be identified
geodesic curves of the conformal, also Riemann’s space, with the so called ”five-force”.9
we have to regard (7) as merely a mathematical trans-
formation of the fields, relating spaces endowed with dif- That the two conformally related theories are, in fact,
ferent Riemannian structure of the same conformal class, different – according to this point of view which is the
and with different dynamics on them. Physically this is most widespread – is confirmed by the study of other
illustrated by the fact that free falling test particles in issues such as positivity of energy which may be a prin-
the starting Riemannian space appear as particles which ciple of physics in one of the conformal representations
are not in free falling in the conformal (also Riemannian) while being violated in the other one (see the next sec-
space, a fact that can be explained as due to the pres- tion). Which one of the different conformally related the-
ence of an additional non-gravitational interaction – the ories is the ”physical” one, is not a well-posed question in
so called ”five-force” – in the conformal variables. this understanding of the conformal transformations (7),
Under this reading of (7), it is obvious that one is deal- since each one of the conformal theories has its own set of
ing with completely different theories with very dissimilar measurable quantities – mostly related to the invariants
physical implications, which can be, in principle, tested of the geometry – so that these admit independent physi-
by contrasting with the same observational data. Recall cal interpretation.10 In this understanding the conformal
that the conformal transformations we are referring to in transformations (7) can serve, for instance, as a mathe-
this paper are not coordinate transformations, i. e., the matical tool to deal with a set of differential equations,
spacetime coincidences are not affected by the transfor- which are difficult to handle in the original variables, in
mation (7). Hence, if suitably refine the accurateness of terms of new field variables of the conformal space [43].
the observational data-sets, in principle, a given confor- However, when extracting physical consequences one has
mal representation can be picked out. Take, as an exam- to go back to the original field variables.
ple, vacuum Brans-Dicke theory of gravity (1), where the
field equations which determine the dynamics are Eq.s
(2), (3):
9 As we have shown, it is not necessary to have additional matter
sources to reveal the existence of a five-force. Recall that, under
the present viewpoint on (7), the motion of test point particles
in the conformal Riemannian space is non-geodesic.

1 2
Gµν = ω ∂µ ϕ∂ν ϕ − gµν (∇ϕ) + ∂µ ϕ∂ν ϕ 10 The corresponding measurable quantities can be contrasted with
2 the experimental/observational evidence [15]. As a result, one
−gµν (∇ϕ)2 + ∇µ ∇ν ϕ − gµν ϕ, can, at least in principle, experimentally/observationally differ-
entiate among the different theories. While in the JF of BD
ϕ + (∇ϕ)2 = 0. (31) theory one has to care about positivity of energy – besides obser-
vational constraints from Solar system experiments which yield
to unnaturally large values of the BD coupling ω > 40000 – in
Under (7), according to γRR in Eq.(24), the above equa- the EFBD theory one has to meet the tight constraints coming
tions transform into (5), (6): from five-force experiments.
8
B. Second viewpoint: physical implications where R̄(w) is the WI curvature scalar given in terms of
the conformal metric ḡµν , and so on. We can see that
The second viewpoint on the conformal transforma- this action does not look the same as the original one
tions (see Eq.(28)) is novel.11 In this case the transforma- in Eq.(1). In this case the question about which one
tion (7) relates Riemannian spaces with Weyl-integrable of the conformal representations (frames) is the physical
ones, γRW : (M, gµν ) 7→ (M, ḡµν , Ω), i. e. constant units one does not arise. Both conformal frames are equally
of length in the starting space are mapped into running ”physical” (or they are not). In particular, if positivity
units in the conformal space and vice versa. Under this of energy holds in one representation, it will hold true also
alternative interpretation Eq.(7) can be consistently un- in the conformal frame (see the demonstration of this in
derstood as a units transformation in the sense of [2]. the next section). The converse statement is also true.
In particular, geodesics of the starting Riemann’s space Additionally, since the field equations are unchanged un-
are transformed into geodesics of the conformal WI space der (7), and, besides, these transformations leave un-
(see the discussion in Sec.III B). It can be shown that, if changed the spacetime coincidences (properly the phys-
one follows this point of view, then the gravitational laws ical events), then observational testing can not differen-
– expressed through the JFBD field equations (31) – will tiate between the conformally related representations of
not be transformed. In this case what changes is the ge- the theory.12
ometric interpretation of these laws. To show this, take
as an example, vacuum BD theory in the Jordan’s frame
as the starting representation. Since, according to the
V. TRANSFORMATIONS OF UNITS AND
viewpoint displayed by Eq.(28), under (7), POSITIVITY OF ENERGY
Gµν = Ḡ(w) ¯ (w) ⇒ = Ω2

µν , ∇µ = ∇µ
¯ (w) ,
Authors who refer to non-equivalence of the different
¯ (w) ≡ ḡ µν ¯ (w) ¯ (w) conformal representations and favor, for instance, the
where ∇ µ ∇ν ,
then the vacuum JFBD field
Einstein’s frame over the Jordan’s one (see [6] for ref-
equations (31) transform into,
erences), usually invoke arguments based on positivity
of energy, and/or fulfillment of the energy conditions

1
[42, 45]. As long as one deals with the most widespread
Ḡ(w)
µν = ω ∂µ ϕ∂ν ϕ − ḡµν (∇ϕ) + ∂¯µ ϕ∂¯ν ϕ
¯ ¯ ¯ 2
understanding of the conformal transformations (first
2
point of view, subsection III A), these arguments may
¯
−ḡµν (∇ϕ) 2 ¯ (w)
+∇ ¯ (w) ¯
µ ∇ν ϕ − ḡµν (w) ϕ, be correct. However, if rely on the alternative viewpoint
¯ (w) ϕ + (∇ϕ)
¯ 2 = 0, (33) revealed in subsection III B – according to which under
(7) Riemann’s spaces are mapped into WI ones and vice
i. e. the laws of gravity look the same in the conformal versa – then these arguments may be wrong.
variables. However, since we start with a Riemannian According to the most widespread understanding of
space, while the transformations (7) bring us into a con- (7), a first apparent argument against equivalence of the
formal WI space, the geometrical interpretation of the conformal frames comes from the straightforward com-
measurements differs from one frame to the other one. parison of JF and EF Brans-Dicke actions (1), and (4),
In consequence, we are led with complementary geomet- respectively, and is referred to positivity of the scalar
rical representations of the same physics [44]. These are field’s kinetic energy. It is seen from EFBD action (4)
equally suited and are not to be contrasted. that, for −3/2 ≤ ω, the kinetic energy of the scalar field
It has to be pointed out that, in spite of the apparent has the correct sign while, for its conformal JFBD coun-
invariance of the JFBD dynamics under units transfor- terpart, provided that ω < 0, it shows the wrong sign. As
mations (7), one can not to talk about equivalence of the an illustration, let us consider the graviton-dilaton sector
conformally related frames as associated with a dynam- of the string effective action in the so called string frame
ics preserving symmetry. Actually, if consider that, un- (SF). In this frame the graviton-dilaton action coincides
der (7), γRW : (M, gµν ) 7→ (M, ḡµν , Ω), then the JFBD with the JFBD action (1) with, ω = −1, which implies
action (1) is transformed into (recall that we are con-
sidering conformal transformations among the Einstein’s
and Jordan’s frames so that Ω2 = eϕ ):
12 Even if the JFBD and EFBD formulations are mathematically
linked through a conformal transformation, ḡµν = eϕ gµν , this
1 √ h
Z
does not mean that these representations are equivalent at all.
i
(w) ¯ 2 ,
S̄BD = d4 x −ḡ R̄(w) − ω(∇ϕ) (34) Perhaps a closer notion could be “duality” or “complementarity”
16π
rather than “equivalence”. In Ref.[44], for instance, the author
relies on the notion of “geometrical duality” instead of “confor-
mal equivalence”. Duality of the conformal descriptions implies
that these are different but mathematically related. Given that
11 See, however, the reference [27], where a similar interpretation is ”duality” has been used in string theory in a quite different con-
managed through the concept of ”conformal Weyl frames”. text [16], complementarity can be a better suited synonym.
9
that, −3/2 < ω < 0. Hence, while in the SF graviton- mantic debate, no more. We want to stress, however,
dilaton effective string action the dilaton’s kinetic energy that other precise statements of this notion can be pos-
has the wrong sign, in the conformal EF formulation the sible (see, for instance, Ref.[13]). Here we will follow the
kinetic energy of the dilaton is positive definite instead. common sense and the notion of equivalence will be as-
If, alternatively, invoke the second point of view displayed sociated with a symmetry which preserves the dynamical
in Eq.(28) (subsection III B), as it can be seen by com- content of the theory.
paring equations (1) and (34), the terms under squared A concrete example where the meaning of the notion
brackets in the action are not transformed, so that the of “equivalence” is crystal clear is the famous Einstein’s
problem with non-positivity of the scalar field’s kinetic equivalence principle within special relativity (SR-EEP).
energy in the original JFBD theory, is inherited by the The physical content of the SR-EEP can be stated in
alternative Weyl-integrable EF formulation. Needless to the following simple way: the laws of physics are the
say that we are not considering here redefinition of the same no matter which one of the different inertial refer-
coupling constant ω, which plays a similar role in both ence frames, in which these can be formulated, is chosen.
conformal formulations, providing the sign for the scalar Mathematically this means that there exists a set of lin-
field’s kinetic energy. ear (homogeneous) coordinate transformations – Lorentz
Similar arguments can be used to rule out statements transformations – which leave invariant, in particular,
found in the literature which point to apparent fulfill- the differential equations that describe the given laws of
ment of the energy conditions in one frame but not in physics.
the conformal one (see Ref.[6] and references therein). It Following the above rule it is straightforward to formu-
is a well-known fact that the weak, strong and dominant late a principle of “conformal equivalence”, or “conformal
energy conditions (WEC, SEC and DEC, respectively) equivalence principle” (CEP for short), which might be
[42, 45] can all be violated by the scalar field ϕ regarded a fundamental principle of nature whenever the laws of
as a form of matter in JF formulation of Brans-Dicke the- gravity are involved. From the point of view of its phys-
ory [10]. This is due to a term arising in the RHS of the ical content, the CEP can be formulated in the follow-
JFBD field equations (2), which is linear in the second ing way: the laws of gravity look the same no matter
derivatives of ϕ, instead of being quadratic in the first which one of the different conformally related frames is
derivatives. On the contrary, the Einstein’s field equa- chosen to describe them. From the mathematical point
tions (5) – which are derivable from the EFBD action (4) of view the CEP is to be associated with invariance of
– are free of terms linear in the second derivatives of ϕ, the field equations that describe the gravitational phe-
so that there is no problem with fulfillment of the energy nomena under the Weyl rescalings (15), which contain
conditions in the EF formulation of BD gravity according conformal transformations (7). It is clear from the for-
to the first point of view in Sec.III A. If consider, alterna- mulation of the CEP given above that physical con-
tively, the point of view displayed by Eq.(28), since the formal (non)equivalence implies mathematical conformal
JFBD field equations are not transformed by (7), then (non)equivalence and vice versa. In the framework of a
the terms linear in the second derivatives of the scalar conformal-invariant theory of gravity, for instance, all of
field are preserved by the transformations of units. In the possible conformal frames in which the theory can be
this understanding of (7), (non)fulfillment of the energy formulated are equally “physical”. Even if the laws of
conditions in the original JF formulation of BD theory gravity look simpler in one given conformal frame, none
will entail (non)fulfillment of the energy conditions in is preferred over the others.
the conformal EF representation of the theory. According to the above ”natural” prescription, the
statement about conformal equivalence of the different
conformal frames will entail that the CEP is valid. The
VI. CONFORMAL EQUIVALENCE PRINCIPLE contrary statement is also true: if the CEP is not valid,
then the different conformally related frames in which
What is called as ”conformal transformation’s issue” a given theory of gravity can be formulated are neither
in the bibliography, is the apparent conundrum we face physically nor mathematically equivalent. As a particu-
when trying to seek for an answer to the question: which lar example we may cite the Brans-Dicke theory of grav-
one of the conformally related frames in which a given ity. Since BD gravity theory is not conformally invariant
theory of gravity can be formulated is the physical one? – the conformal transformation (7) maps the JFBD field
This question makes sense only for those who understand equations into the EFBD ones – then the CEP is not
that the different frames are not ”physically” equivalent. valid. This means, in turn, that Einstein’s and Jordan’s
Those who think the conformal frames are equivalent, ob- conformal frames of Brans-Dicke theory are not equiva-
viously do not face this question. Due to the importance lent. The argument can be safely applied to scalar-tensor
of a clear and meaningful statement of what is to be un- theories in general. This example shows the importance
derstood by ”equivalence” in order to resolve the contro- of clearly prescribing what is meant by conformal equiv-
versy, in this section we will endow the notion of ”confor- alence for a meaningful discussion of the issue.
mal equivalence” with a concrete physical/mathematical Although, as long as we know, it has not been as
meaning. Recall that, otherwise, we will be facing a se- strictly formulated as we have done above, the princi-
10
ple of conformal equivalence has been assumed to play geometry is recovered out of Weyl-integrable one, the ac-
an important role in the understanding of the laws of tion (35) is mapped into the standard Einstein-Hilbert
physics in many influential papers before [2, 46–48]. In action (see appendix A),
this regard we want to point out that whether the CEP
1 √
Z
(w)
is a fundamental principle of nature is not a subject of S → SEH = d4 x −g R,
interest in the present paper. In correspondence we will 16πGef f
not make judgments about its validity here, so that the where, as before, R is the Riemannian curvature scalar,
results of our discussion will not depend on the CEP be- and, Gef f = e−ϕ0 , is the effective gravitational coupling
ing valid. Even in case it were a fundamental principle of constant. I. e., in that gauge general relativity is recov-
physics, given the nature of the quantum measurement ered rather than Brans-Dicke theory. For that reason we
process, we do not expect the CEP to be valid at scales may call the resulting scale-invariant theory of gravity as
where quantum gravity effects become unavoidable. ”scale-invariant general relativity” (see [27]).
Nevertheless, it is of interest to investigate the physical In case the theory based on action (35) – whose dynam-
and mathematical implications of a conformal-invariant ics is dictated by the field equations (36) – were a correct
theory of gravity. The latter represents a counterexample theory of gravity, the CEP were a fundamental principle
where, unlike BD theory, the CEP is satisfied. This is, of nature, so that, conformal symmetry were a true sym-
precisely, the aim of the remaining part of this section. metry of the gravitational laws. This is to be contrasted
It is a matter of simple algebra to demonstrate that with vacuum BD theory where this is a mirage or spu-
the particular value of the BD coupling, ω = −3/2, is not rious symmetry instead. Put in different words: in the
transformed by (8). Hence, the corresponding action,13 gravitational theory depicted by (35), (36), the laws of
gravity look the same in the different conformal frames,
√

1 3
Z
BD while the spacetime coincidences – properly the observa-
S3/2 = d4 x −g eϕ R + (∇ϕ)2 ,
16π 2 tions – are unchanged. This is an outstanding example
of a theory where, unlike BD theory, the different con-
together with the field equations derived from it, will formal descriptions of a given phenomenon are actually
be invariant under the Weyl rescalings (15). It makes (physically and mathematically) equivalent.
sense, then, to rewrite the above action in terms of WI
quantities by using the Riemannian decomposition of the
WI curvature scalar R(w) in (17). The result is:14 A. Positivity of energy
After the analysis in section V, it is almost trivial to

1 √
Z
S (w) = d4 x −g eϕ R(w) . (35) show that the scale-invariant theory of gravity given by
16π (35), (36), is free of the positivity of energy problem, in-
cluding fulfillment of the energy conditions. In fact, the
In the present case not only the action (35) but also the first thing to notice is that, in the action (35), the kinetic
field equations that can be derived from it, energy term of the gauge field ϕ is absent. Hence there
is no problem with positivity of the scalar field’s kinetic
energy. Besides, since the gravitational field equations
G(w)
µν = 0, ⇒ Gµν − ∇µ ∇ν ϕ + gµν ϕ
(w)
are (see Eq.(36)): Gµν = 0, there are no terms linear in
1 1 the second derivatives of the ϕ-field (in fact there are no
+ [∂µ ϕ∂ν ϕ + gµν (∇ϕ)2 ] = 0,
2 2 terms containing derivatives of any order), so that there
1 R will be no problems with fulfillment of the energy con-
ϕ + (∇ϕ)2 − = 0, (36) ditions neither. This represents an additional difference
2 3
between standard vacuum Brans-Dicke theory [1] and the
are invariant under the Weyl rescalings (15). It is evident conformal-invariant theory explored in this section.
that the CEP is satisfied in this formulation of the grav-
itational laws. The resulting conformal-invariant theory
will be a fully geometrical description of the laws of grav- B. Other mathematical aspects
ity, not sharing any properties with the standard BD the-
ory. In the particular gauge, ϕ = ϕ0 , when Riemannian There are several mathematical consequences arising
from assuming WI spaces as the natural geometrical ba-
sis for a scale-invariant theory of gravity (35), (36). In
particular, due to invariance under (15), the metric gµν
13 That a scale-invariant scalar-tensor action corresponds to the and the gauge scalar ϕ are far from unique. Instead of
(singular) choice ω = −3/2 in BD theory has been shown, for a fixed pair (gµν , ϕ), one has a whole (perhaps infinite)
instance, in the very well-known paper [46] (see also [49]).
14 In references [27] and [50] a similar action was investigated in class of pairs C – defined in Eq.(16) – such that, any other
different contexts. In Ref.[27], besides the pure geometric part, pair (ḡµν , ϕ̄) related with (gµν , ϕ) by a scale transforma-
also a ”cosmological constant” and matter terms were considered. tion (15), also belongs in C. The situation is reminiscent
11
of what happens when one invokes invariance under gen-

eral coordinate transformations x̄α = f α (x0 , x1 , x2 , x3 ): 2
there is not a unique set of coordinates to describe a given 1
3 H + ϕ̇ = 0,
physical situation, but a whole (in principle infinite) class 2
of them.15 While, in the latter case, the spacetime coor- 1 1

1

dinates are meaningless – the physical meaning is trans- Ḣ + ϕ̈ − H + ϕ̇ ϕ̇ = 0, (37)
2 2 2
ferred to the invariants of the geometry under spacetime
diffeomorphisms – in the case when conformal invariance while the vacuum KG-equation (second equation in the
is invoked, the fields themselves loss independent physi- RHS of Eq.(36)):
cal meaning. In this latter case the physically meaningful
quantities are the conformal invariants of the WI space
such as, for instance, the conformal-invariant measure of 1 3 1
Ḣ + 2H 2 + ϕ̈ + ϕ̇H + ϕ̇2 = 0, (38)
scalar curvature, e−ϕ R(w) , the conformal-invariant mea- 2 2 4
sure of spacetime separations, eϕ/2 ds, as well as other WI
conformal-invariant quantities given in Eq.(19). is not an independent equation. The Friedmann equation
above can be integrated to obtain the following depen-
The above discussed scale-invariance property is re- dence of the scale factor upon the gauge field ϕ (a0 is an
flected in the mathematical structure of the field equa- arbitrary integration constant):
tions (36), which are derived from the action (35). In
fact, when written in terms of Riemannian quantities,
the first and second equations in the RHS of Eq.(36) are a = a0 e−ϕ/2 . (39)
not independent from each other. The second equation
is just the trace of the first one so that one equation If we substitute (39) back into the remaining equations
is redundant. Hence, there will be 6 independent equa- – second equation in (37) and equation (38) – these be-
tions to determine 11 unknown degrees of freedom (10 come just identities (0 ≡ 0), so that no new information
components of the metric tensor plus the gauge field ϕ). can be extracted from them. This is a consequence of
Nonetheless, in addition to the 4 degrees of freedom to scale invariance since, due to this symmetry, we have
make diffeomorphisms – four components of the metric the freedom to choose either any ϕ(t), or any a(t) we
can be transformed away – one more component can be want. Recall that one of these degrees of freedom can be
gauged away due to an additional degree of freedom to transformed in any desired way by an appropriate scale
make scale transformations (15). In other words, up to transformation of the kind (15). In particular, under
general coordinate plus scale transformations, the field the conformal transformation ḡµν = eϕ gµν , the confor-
equations (36) ”uniquely” determine the metric coeffi- mal scale factor, ā = eϕ/2 a(t) = a0 , is a constant, as it
cients. Notice, however, that specifying a given ϕ, or, should be in vacuum general relativity without cosmo-
alternatively, one of the components of the metric ten- logical constant. Consistently with this result, under the
sor, amounts to going into a particular gauge. Hence, as above transformation, the scale-invariant theory explored
a matter of fact, the field equations themselves are not in this section transforms down into standard general rel-
enough to uniquely determine the field content of the the- ativity (see appendix A).
ory. This ambiguity is a clear consequence of invariance This is, precisely, what one expects from a truly scale-
under Weyl rescalings (15). invariant theory: either the gauge (scalar) field or one
To illustrate the point, let us, briefly consider vacuum of the components of the metric tensor can be chosen
cosmology within the context of the theory (35), (36). at will, respecting, of course, mathematical consistency
Assume a Friedmann-Robertson-Walker (FRW) space- requirements such as the existence of the inverse confor-
time with flat spatial sections, given by the line element mal transformations, continuity, etc. (see [14] for a simi-
lar discussion but in a non scale-invariant context). This
ds2 = −dt2 + a2 (t)δij dxi dxj , freedom is what differentiates a truly conformally invari-
ant theory from one which does not respect the CEP.
where, as usual, t is the cosmic time and a(t) is the

scale factor. The vacuum field equations (36) can then VII. CONFORMAL EQUIVALENCE: THE
be written as follows (H ≡ ȧ/a, the dot accounts for DEBATE
t-derivative):
In this section we will examine the different approaches
to the conformal transformation’s issue encountered in
the bibliography under the light of the new arguments
15 We have to recall that invariance under diffeomorphisms and given here: i) the conformal transformations (7) admit,
scale-invariance are independent symmetry requirements: con-
formal transformations of the kind considered here are not dif- at least, two different geometrical interpretations (see
feomorphisms. It is evident that the scale-invariant theory given sections III, IV), and, ii) in order to avoid the confor-
by the action (35) is also invariant under diffeomorphisms. mal transformations controversy being a semantic issue
12
it is critical to agree on which concrete meaning is to as a units transformation. Here we will explore this ap-
be assigned under the notion of (physical/mathematical) proach in detail since, we think, there is a lot of confusion
”equivalence”. In the present discussion, for definiteness, associated with the lack of a concise and mathematically
we will assume conformal equivalence to be associated definite statement of what the authors understand by
with invariance under (15) as stated in section VI. In ”equivalence”. In what follows, for definiteness, we will
consequence physical equivalence will entail mathemati- refer only to Jordan’s and Einstein’s conformal frames.
cal equivalence and vice versa.
In the famous paper [2], for instance, Dicke stated that
Although the different approaches existing in the bib-
”...the laws of physics must be invariant under a trans-
liography have been classified into several groups [4, 6],
formation of units.” No matter whether this statement is
in the present discussion we will not follow the men-
correct or not, the obvious fact is that Brans-Dicke the-
tioned classification, and will consider only authors who
ory itself does not belong in this class of theories. Hence,
adhere to one of the following two different types of ap-
it is not understood which class of equivalence Dicke re-
proaches: i) those who consider that the JF and the EF
ferred to in [2]. In the well-known paper [10] – which
are not equivalent [4, 6, 12, 15, 51–59], and ii) those
is fully consistent with Dicke’s arguments – the following
who state that the JF and EF formulations of BD the-
statement is made: ”...the two frames are equivalent, pro-
ory, and scalar-tensor theories in general, are equivalent
vided that the units of mass, length, time, and quantities
[2, 10, 13, 14, 60, 61].
derived there from scale with appropriate powers of the
conformal factor Ω in the Einstein frame.” In the above
quotation no clear statement is made neither of what to
A. First Approach: JF and EF are not equivalent
understand by ”equivalence”. Besides, in concordance
with Dicke’s arguments, in Sec.III A of Ref.[10], the au-
Authors in this group admit that the different con- thors say: ”Since physics is invariant under a change of
formally related frames, in particular Jordan’s and Ein- units, it is invariant under a conformal transformation
stein’s ones, are not physically equivalent (we refer the provided that the units of length, time, and mass...are
reader to the reviews [4, 6] for a complete list of authors). scaled.” What do the cited authors mean by ”invariance
In this sense, if follow the point of view exposed in sub- of physics” under a conformal transformation?
section III A, according to which the transformation (7)
is just a mapping γRR : Riemann 7→ Riemann, we have to Let us briefly revise the arguments given in Ref.[10].
partially agree with these authors. In fact, if undertake In section III B, for instance, the authors study the mo-
this approach, since under (7) the JFBD field equations tion of massive particles in the so called ”Einstein frame
(31) are mapped into the EFBD ones (32), both sets of with running units”. They rely on the investigation of
equations represent different laws of gravity on Riemann time-like geodesics. It is shown that the correction to
space. Consequently, the different frames depict differ- the equation of motion – see the RHS term of the EF
ent theories of gravity with their own sets of measurable non-geodesic equation (26) – is entirely due to variation
quantities (see the discussion in sections IV, V). of the particle’s mass in the 3-space of an observer mov-
Our disagreement arises when, according to these au- ing with the particle, so that, if consider that the mass
thors, it has to be cleared which one of the conformally of the particle in the EF varies as, m̄ = Ω−1 m, due to
related frames is the ”physical” one, a problem which running units, then there is not any effective modifica-
is properly known as the conformal transformation’s is- tion of the geodesic motion. However, their argument is
sue. As it has been discussed in section IV, the state- flawed since it is based on a wrong assumption, and, on
ment of this problem is not correct. In fact, since each a miss-interpretation of the role affine parametrization
frame represents a different theory with its proper dy- plays in geodesic motion. In the first place, the authors
namics, its own set of measurables, etc., the question assume that, under (7), the spacetime coordinates are
about which one of the conformally related frames – the also modified, dx̄µ = Ω dxµ .16 Hence, according to their
JF and EF, in particular – is the physical one, has to be analysis, under (7), the affine parameter along a time-like
replaced by a more pragmatic question: which one of the geodesic transforms like, dλ̄ = Ω2 dλ, which led them to
different conformal theories fits better the existing ob- come to the obviously wrong result that the line-element
servational/experimental evidence? [15]. Besides, do not should transform like: ds̄2 = Ω4 ds2 (Eq.(3.16) of the
forget about first principles, such as positivity of energy, cited paper). This later equation is clearly inconsistent
etc. which may also be checked. with the original understanding that, under ḡµν = Ω2 gµν ,
the spacetime separations dxµ are unchanged [2] (dx̄µ =
dxµ ), which leads to: ds̄2 = ḡµν dxµ dxν = Ω2 ds2 , in-
B. Second Approach: JF and EF are equivalent stead (see, for instance, Eq.(10) of reference [2]). In the
A second group of authors believe the different confor-

mal frames are physically equivalent [2, 10, 13, 14, 60, 61] 16 See Eq.(3.1) of Ref.[10], and compare with correct equations in
so that, in correspondence, the issue does not arise. In Eq.(4) of the first entry in [11], where |ds| is involved rather than
this case, the conformal transformation (7) is understood the dxµ .
13
second place, the authors of [10] do not realize that the A very interesting approach that deserves independent
demonstration in Sec.III B of their paper, that the mo- comment is the one of Ref.[11]. In that reference the au-
tion equations for EF worldlines do not admit an affine thors introduce frame-independent quantities and apply
parametrization whatsoever, means, in fact, that, even them to situations of cosmological interest. Besides, in
assuming the EF mass changes as m̄ = Ω−1 m, time-like the paper corresponding to the second entry in [11], the
motion in the Einstein’s frame can not be geodesic at all authors study a frame invariant action (equation (7) of
(see our discussion of this matter in subsection III A). their paper). JFBD and EFBD theories correspond to
Contrary to the conclusion extracted from this demon- particular gauges of their more general theory. While
stration in [10], as a matter of fact, this result would their analysis is correct, if regard the theory under con-
entail that the motion of time-like particles in the JF sideration as a conformal-invariant theory, it is clear that
is not (dynamically) equivalent to the corresponding mo- these criteria can not be applied to Brans-Dicke theory,
tion in the conformal ”EF with running units”. Actually, and scalar-tensor theories in general (see the discussion
if understand the conformal transformations as they are on this matter in the introduction). We think discussion
usually considered (Sec.III A), we see that, even allowing of the affine properties of the underlying space (affine
for mass units variation – such as to make m̄ = Ω−1 m – connection, geodesics, etc) is lacking in [11]. Notice that
there is a neat modification of the time-like geodesics in test particles in their theory do not follow geodesics of
the Einstein’s frame. The demonstration is simple: just the metric hµν (here we use author’s symbology). This
notice that the EF motion equation (26) – which is an is seen from the matter part of the action in Eq.(6) of
alternative writing of equation (3.18) of Ref.[10] – may [11], where it is apparent that matter particles couple to
be rewritten in the following form: the conformal geometry. This hints to possible five-force
constraints on this theory.
dxα ¯ dxµ dxν

d ∂µ Ω µα
m̄ + m̄ α
µν = −m̄ ḡ ,
ds̄ ds̄ ds̄ ds̄ Ω
C. A third approach
where it has been considered that m̄ = Ω−1 m (the JF
mass of the particle m is a constant). If we compare the
If adopt the viewpoint on the conformal transforma-
latter equation of motion with its JF counterpart:
tions according to which, under (7), γRW : Riemann 7→
d

dxα

dxµ dxν Weyl ⇔ (M, gµν ) 7→ (M, ḡµν , Ω), i. e., Riemann’s space
m +m α µν = 0, (in the Jordan’s frame variables) is mapped into a Weyl-
ds ds ds ds
integrable space (in Einstein’s frame variables), then the
it is seen, that even if consider that the EF masses scale point-dependent property of the units of length is already
as m̄ = Ω−1 m, an additional term remains in the RHS encoded in the affine structure of the conformal space.
of the EF motion equation, which can not be removed Besides, while the JF time-like geodesics are mapped into
by an affine parametrization [42, 62], a fact that was time-like geodesics of the conformal WI space (EF), the
demonstrated, precisely, in Sec.III B of [10]. Hence, the JFBD field equations (31) are not transformed by (7)
five-force effect can not be removed by allowing the units (Ω2 = eϕ ). Hence, under this understanding of the con-
of length in the Einstein’s frame to vary in the way it formal transformations (7), the dynamics is unchanged,
was considered in Ref.[10]. This shows that even if allow pointing to a kind of dynamical equivalence. Recall, how-
the units of length of the conformal frame to vary from ever, that the geometric picture in the EF differs from the
point to point, the equations of motion are not the same one in the JF – in particular the action is modified by (7)
in the conformally related frames, which means, in turn, – so that, in fact, what one actually has is two comple-
that there is not any dynamical equivalence among the mentary geometrical descriptions of a same phenomenon.
JF and the EF representations. Additional arguments supporting this interpretation will
In contrast to what is done in references [2, 10], a con- be given in the next section, where it will be shown that,
sistent consideration of the conformal transformations while in one frame one inevitably encounters spacetime
(7) as transformations of units – subsection III B (see singularities, in the conformal frame these singularities
also Sec.IV and V) – would yield to dynamical equiva- might be an infinite proper time into the future/past so
lence in the sense that, under (7), JF time-like Rieman- that, in fact, these are removed from the alternative rep-
nian geodesics are mapped into Weyl-integrable time- resentation (see, also, a similar discussion in references
like geodesics (see the demonstration in subsection III B). [44, 63]).
Hence, the missing argument in the analysis of references
[2], [10] is the lack of consideration of the impact units
transformations have on the modification of the affine VIII. THE SINGULARITY ISSUE
properties of space (affine connection, geodesics, etc).
Recall that in [2, 10] it is implicitly assumed that both In view of the possible impact of the developments pre-
the starting and the conformal spaces are Riemannian sented in this paper, here we want to comment about
in nature, so that, in particular, the time-like geodesics a related very important subject: the singularity issue.
have to be those of a standard Riemannian metric. Arguments in favor of the fulfillment of the energy con-
14
ditions in one conformal frame but not in others, has is independent of the affine properties of the spacetime
led several authors to conclude that spacetime singular- manifold. In this regard we expect to show that in a
ities in one frame might be avoided in a conformally re- given (conformal) formulation of the theory one can find
lated one [44, 63, 64]. These results have been criticized “sufficiently long” time-like geodesics [42], so that (time-
in Ref.[10] based on a re-analysis on the light of the so like) geodesic incompleteness is not met and spacetime
called ”Einstein’s frame with running units”, which is singularities that are present in one theory’s formulation
in agreement with the spirit of Dicke’s paper [2]. Ac- can be avoided in its conformal representation.
cording to [10], since (following Dicke), the Jordan and For definiteness, let us suppose in EFBD theory there
Einstein frames are equivalent, singularities occur in the exists an isotropic spacetime singularity so that,17 fol-
Einstein’s frame if and only if they occur in the Jordan’s lowing a time-like geodesic of the WI geometry, this sin-
frame. We have shown, however, that several arguments gularity is necessarily met in a finite proper time into the
given in that reference are flawed, so that a new analysis future/past, dτ̄ → finite (τ̄ is the proper time in EFBD
of the subject is mandatory. variables). The above singularity is characterized by, say,
According to the alternative geometric interpretation (w)
I¯4 → ∞. According to (40) – assuming the singularity
of the conformal transformation (7) given in Sec.III B can be removed in the JFBD theory’s representation –
– which is the one being geometrically consistent with one may find a function Ω such that, as the singularity
Dicke’s understanding of (7) as a units transformation is approached, Ω8 → 0, quick enough as to make the
[2] – the study of the validity of the energy conditions, (w)
quantity, I4 = Ω8 I¯4 → finite. This entails that, since
and, consequently, of the singularity issue, is not as triv- under (7), the elements of proper time in JF and EF of
ial and straightforward as stated in [10, 44, 63, 64]. In
BD theory are related through, dτ 2 = Ω−2 dτ̄ 2 , hence –
fact, if we review the results of Sec.V in accordance with
supposing the above assumptions are correct – as the sin-
the viewpoint displayed by Eq.(28), one immediately sees
gularity is being approached, dτ → ∞. This means that
that, if the energy conditions [42, 45] are not satisfied in
in the Jordan’s frame of the Brans-Dicke theory a singu-
one frame, these will not be met in its conformal frame
larity that is met in a finite proper time (dτ̄ → finite) in
either. The contrary statement is also true. This argu-
its conformally related EF, may be avoided in principle
ment alone would support the results of the analysis in as long as dτ is large enough in the sense specified above
Ref.[10], however, as already said, the situation is not so
[42]. Besides, divergent curvature invariants are mapped
simple. Before reaching to any conclusive result one has into finite ones and vice versa.
to check, first, whether the singularity theorems in their
We want to underline that, under the novel approach
standard formulation [45] (which is given on the basis
to (7) explored in this paper, the point-dependent prop-
of Riemannian spacetimes), are valid when dealing with
erty of the units of length is already encoded in the
Weyl-integrable spaces. This issue deserves a separate
affine structure of space (affine connection, geodesics, rel-
investigation. The interesting thing is that, even with-
evant curvature invariants, etc), so that, for instance,
out a strict formulation of the singularity theorems in (w) (w) (w)
WI spacetimes, one can reach to interesting qualitative while in the EF, M̄ = (dτ̄ , I¯0 , I¯2 , I¯4 , ...), is a
results. set of measurable quantities in WI space, in the JF,
To state the qualitative discussion on solid grounds, M = (dτ, I0 , I2 , I4 , ...), is the corresponding set of mea-
lets write the relationships between several curvature in- surable quantities in Riemannian space. Therefore, as it
variants in the JF and EF of BD gravity, under the as- is usually done, a straightforward analysis of the invari-
sumption that the second point of view on (7) – Eq.(28) ants will suffice to judge about the (non)occurrence of
in subsection III B – is valid. Since, according to this al- a curvature singularity in a given representation of BD
ternative geometrical interpretation, γRW : (M, gµν ) 7→ theory. In other words, in contrast to the analysis of the
(M, ḡµν , Ω), then the following relationships are valid: singularity issue in Sec.IV of Ref.[10], no additional ”ma-
chinery” and/or technical assumptions have to be applied
to the study of this subject.
(w) (w) (w)
I0 = Ω2 I¯0 , I2 = Ω4 I¯2 , I4 = Ω8 I¯4 , (40) Although there are subtle issues related with integra-
bility (as a matter
R of fact we have to deal with integrals
where we have defined the following invariant quantities:
of the kind τ = dτ̄ Ω−1 , rather than with infinitesimal
quantities like dτ ) the above qualitative analysis is gen-
I0 ≡ R, I2 ≡ Rµν Rµν , I4 ≡ Rµνσρ Rµνσρ . (41) eral enough. It shows that, for given functions Ω, the sin-
gularities that are inherent in one frame of Brans-Dicke
In Eq.(40) the unbarred quantities refer to Riemannian theory, may be avoided in a properly chosen conformal
invariants in JFBD variables, while the quantities with representation. For concrete examples where this hap-
an over-bar are the ones given in the EF of BD theory pens we refer the reader to references [44, 63–65].
coupled to WI spaces. Recall that the upper/lower index
(w) denotes the given quantity is given in terms of the
affine connection of WIG, etc.
Our analysis of the singularity issue will be based on 17 The study of the impact conformal transformations have on
the notion of geodesic (in)completeness, a concept which anisotropic singularities requires a more careful analysis.
15
The lesson to be learned is that the singularity theo- v) The main points of view existing in the bibliogra-
rems [42, 45], which have been strictly formulated (and phy on the conformal transformation’s issue have been
applied) in contexts that involve Riemannian spaces, critically scrutinized. It has been shown, for instance,
are to be revised within the context of non-Riemannian that several arguments in favor of equivalence [10] are
spaces, WI ones being a particularly interesting case. flawed and fail to be meaningful due to lack of consid-
eration of the effect units transformations carry on the
affine properties of space. The singularity issue has been
IX. CONCLUSION re-examined.
There remain several interesting questions to be in-
In this paper we have re-examined the conformal trans- vestigated: Which is the actual geometrical/affine struc-
formation’s issue on the light of new arguments: i) con- ture of spacetimes that serve as models of our universe?
formal transformations may be understood as a mapping Are these Riemannian or non-Riemannian spaces? Is the
among Riemannian and Weyl-integrable spaces (sections CEP a fundamental principle of nature? If so were, is
III-V), and, ii) the notion of ”conformal equivalence” is scale-invariant general relativity a better suited theoret-
to be endowed with a concrete mathematical and physical framework for the description of the classical laws of
ical meaning (Sec.VI). The main results of the present gravity than standard general relativity itself?
research can be summarized in the following way. The authors thank Yun-Song Piao for pointing to
us reference [65], which might serve as an additional
i) A novel aspect of the conformal transformation’s is-
illustration of our discussion on the singularity issue.
sue has been revealed (subsection III B). It is based on
This work was partly supported by CONACyT México
consideration of the effect transformations of units [2]
under grants 49865-F and I0101/131/07 C-234/07 of
carry on the affine properties of space, such as the con-
the Instituto Avanzado de Cosmologia (IAC) collabo-
nection, the geodesics, etc.
ration (http://www.iac.edu.mx/), by the Department of
ii) In the understanding that equivalence entails a dy-
Physics, DCI, Guanajuato University, Campus León, and
namics preserving symmetry, we have shown that the
by the Department of Mathematics, CUCEI, Guadala-
JFBD and EFBD representations are not equivalent for-
jara University.
mulations of BD theory.
• If follow the most widespread understanding of APPENDIX

the conformal transformations (7) as a mapping,
γRR : Riemann 7→ Riemann, among Riemannian A. General relativity gauge
spaces, then, the JF and EF formulations of BD
theory are actually different theories (with different
dynamics) operating on Riemann spaces, each one General relativity can be obtained from (35) in a par-
with its own set of measurable quantities. Which ticular gauge when ϕ̄ = ϕ0 . In fact, after this choice, the
theory is the most adequate one is a question that solution of the equation in the RHS of Eq.(15) will be
may be answered by the experiment. The ques- ϕ − ϕ0 = 2 ln Ω ⇒ Ω2 = eϕ−ϕ0 .
tion about which one of the conformal frames is
the physical one is devoid of meaning. It can be shown that, under (7) with the latter choice
of the conformal factor, the affine connection of a Weyl-
• If follow the viewpoint explored in subsection III B, integrable space maps to the Christoffel symbols of the
conformal metric: Γα ¯
α
according to which (7) may be understood as a βγ → {βγ }, etc. In other words, WI
mapping, γRW : Riemann 7→ Weyl, then the dy- spaces transform under (7) – with Ω2 = eϕ−ϕ0 – into
namics is the same in the JF and in the EF, how- Riemannian spaces. Besides, the action (35) transforms
ever, the geometrical interpretation differs. Both into the Einstein-Hilbert action,
conformal formulations of BD theory represent 1
Z
√
complementary geometrical descriptions of a same SEH = d4 x −ḡ R̄,
16πGef f
physics. The question about which one of the con-
formal frames is the physical one does not arise. where Gef f = e−ϕ0 is the (rescaled) effective gravita-
tional coupling. Hence GR is a particular gauge in the
iii) It has been shown that, contrary to claims existing conformal equivalence class C defined in Eq.(16). Going
in the bibliography [6, 18–20], vacuum BD theory is not into this particular gauge means that conformal invari-
invariant under the conformal transformation (7) plus the ance becomes, automatically, into a broken symmetry.
redefinitions (8).
iv) We have explored an example of a theory whose
conformal frames are actually equivalent representations B. Brans-Dicke theory with matter
of the theory. It has to be necessarily a scale-invariant
theory [27, 50] so that the metric is defined up to a class Although for sake of simplicity we have considered in
of conformal equivalence. this paper only the vacuum sector of Brans-Dicke theory,
16
several qualitative aspects of introducing matter into the

theory can be discussed here. We will see that the re-
sults of the present work are not modified by the addi- ¯ ν T̄ (m) = − 1 ∂¯µ ϕ T̄(m) ,
∇ (43)
νµ
tion of ordinary matter into BD theory. We have seen, 2
in particular, that considering the motion of test parti- (m)
cles in vacuum BD theory already reveals several inter- where we have considered that, under (7), T̄µν =
(m) (m)
esting features. For instance, assuming one adheres to Ω−2 Tµν , and T̄(m) = ḡ µν T̄µν is the trace of the EF
the first viewpoint in subsection III A, the test particles stress-energy tensor. Only for traceless (massless) mat-
which follow geodesics of the JF metric do not fallow ter fields the continuity equation (42) is not transformed
geodesics of the conformal EF metric. Hence, if one de- by (7). Physically Eq.(43) means that in terms of the
cides to interpret the results of a given analysis in the EF EFBD metric there is an additional (non-gravitational)
of BD theory, since deviations from geodesic motion are interaction between matter and the scalar field (the five-
interpreted by the EF observers as due to the existence force) so that these fields exchange energy-momentum.
of additional interactions of non-gravitational origin be- If one invokes instead the point of view of section III B,
tween the test particle and the scalar field ϕ, one will γRW : (M, gµν ) 7→ (M, ḡµν , ϕ), then the JF continuity
be faced with confronting the predictions of the theory equation (42) is mapped into
with “five-force” experiments. The above analysis may
be corroborated by adding matter into Brans-Dicke the-
ory. The matter part of the
¯ ν T̄νµ
∇ (m)
= −∂¯ν ϕ T̄νµ
(m)
, (44)
R JFBD√ gravity is assumed (w)
to be given by [1], Sm = d4 x −gLm (ψi , gµν ), where
Lm is the Lagrangian density of matter, and ψi repre- which is just the continuity equation in WI spaces. Al-
sent the matter fields. The following continuity equation though the term in the RHS of (44) might seem alien, it
is obeyed in the JFBD theory: expresses the fact that the units of measure of the stresses
and energy are running units. In no case the RHS of (44)
can be interpreted as a source term. To show this notice
2 ∂ √ that Eq.(44) is the trace of the most general equation,
∇ν Tνµ
(m) (m)
= 0, Tµν = √ ( −g Lm ), (42)
−g ∂g µν ∇¯ (w) (m) ¯ (m)
α T̄νµ = −∂α ϕ T̄νµ , which is, in turn, the equiva-
lent of the metricity condition (27) – with Ω2 = eϕ – in
(m)
where Tµν - the matter stress-energy tensor. Under the matter sector of the EFBD theory. The opposite sign
(7) with Ω2 = eϕ – in the sense implied by the first in the RHS of the above equation in respect to the sign
viewpoint in Sec.III A (Eq.(24)) – the JF matter action of the RHS of Eq.(27) is a consequence of the fact that,
above
R 4 √is transformed into the EF one [2, 64], S̄m = increasing extent of the units of length and time is cor-
d x −ḡ e−2ϕ L̄m (ψ̄i , e−ϕ ḡµν ), while (42) is mapped related with the contrary effect on the units of stresses
into and energy.
[1] C. Brans, R. H. Dicke, Phys. Rev. 124 (1961) 925-935. [13] S. Capozziello, F. Darabi, D. Vernieri, Mod. Phys. Lett.
[2] R. H. Dicke, Phys. Rev. 125 (1962) 2163-2167. A 25 (2010) 3279-3289 [e-Print: arXiv:1009.2580].
[3] M. Fierz, Helv. Phys. Acta 29 (1956) 128. [14] N. Deruelle, M. Sasaki, e-Print: arXiv:1007.3563; e-
[4] G. Magnano, L. M. Sokolowski, Phys. Rev. D 50 (1994) Print: arXiv:1012.5386.
5039-5059 [e-Print: gr-qc/9312008]. [15] C. Corda, Astropart. Phys. 34 (2011) 412-419 [e-Print:
[5] S. Capozziello, R. de Ritis, A. A. Marino, Class. Quant. arXiv:1010.2086].
Grav. 14 (1997) 3243-3258 [e-Print: gr-qc/9612053]. [16] J. E. Lidsey, D. Wands, E. J. Copeland, Phys. Rept. 337
[6] V. Faraoni, E. Gunzig, P. Nardone, Fund. Cosmic Phys. (2000) 343-492 [e-Print: hep-th/9909061].
20 (1999) 121 [e-Print: gr-qc/9811047]. [17] D. Blaschke, M. P. Dabrowski, e-Print: hep-th/0407078.
[7] V. Faraoni, E. Gunzig, Int. J. Theor. Phys. 38 (1999) [18] Y. M. Cho, Phys. Rev. Lett. 68 (1992) 3133.
217-225 [e-Print: astro-ph/9910176]. [19] V. Faraoni, Phys. Lett. A 245 (1998) 26-30 [e-Print:
[8] D. N. Vollick, Class. Quant. Grav. 21 (2004) 3813-3816 gr-qc/9805057].
[e-Print: gr-qc/0312041]. [20] See Appendix A, p. 478 of Ref.[16].
[9] E. E. Flanagan, Class. Quant. Grav. 21 (2004) 3817 [e- [21] I. Quiros, R. Garcia-Salcedo, J. E. Madriz Aguilar, e-
Print: gr-qc/0403063]. Print: arXiv:1108.2911.
[10] V. Faraoni, S. Nadeau, Phys. Rev. D 75 (2007) 023501 [22] A. S. Eddington, ”The Mathematical Theory of Relativ-
[e-Print: gr-qc/0612075]. ity” (Cambridge University Press, Cambridge, England,
[11] R. Catena, M. Pietroni, L. Scarabello, Phys.Rev. D 76 1923).
(2007) 084039 [e-Print: astro-ph/0604492]; J. Phys. A 40 [23] W. Pauli, ”Theory of Relativity” (Dover, New York,
(2007) 6883-6888 [e-Print: hep-th/0610292]. 1981).
[12] S. Capozziello, P. Martin-Moruno, C. Rubano, Phys. [24] R. Adler, M. Bazin, M. Schiffer, ”Introduction to General
Lett. B 689 (2010) 117-121 [e-Print: arXiv:1003.5394]. Relativity” (McGraw-Hill, Second Edition, 1975).
17
[25] P. Bouvier, A. Maeder, Astrophys. Space Sci. 54 (1977) 11 (1994) 1919.

497-508. [44] I. Quiros, Phys. Rev. D 61 (2000) 124026 [e-Print:
[26] L. Fatibene, M. Francaviglia, e-Print: arXiv:1106.1961; gr-qc/9905071].
P. Gilkey, S. Nikcevic, U. Simon, J. Geom. Phys. 61 [45] S. W. Hawking, G. F. R. Ellis, “The Large Scale Struc-
(2011) 270-275 [e-Print: arXiv:1002.5027]; R. Carroll, ture of Spacetime” (Cambridge University Press, Cam-
e-Print: arXiv:0705.3921; L. ORaiefeartaigh, N. Strau- bridge, England, 1973).
mann, Rev. Mod. Phys. 72 (2000) 1; G. B. Folland, J. [46] S. Deser, Ann. Phys. 59 (1970) 248-253.
Diff. Geom. 4 (1970) 145; T. V. Thomas, Proc. N. A. J. [47] A. Shaukat, A. Waldron, Nucl. Phys. B 829 (2010)
12 (1926) 353 (see also references [28] and [38]). 28-47 [e-Print: arXiv:0911.2477]; A. R. Gover, A.
[27] C. Romero, J. B. Fonseca-Neto, M. L. Pucheu, Shaukat, A. Waldron, Phys. Lett. B 675 (2009) 93-97
Int. J. Mod. Phys. A 26 (2011) 3721-3729 [e-Print: [e-Print: arXiv:0812.3364]; A. R. Gover, A. Shaukat, A.
arXiv:1106.5543]; e-Print: arXiv:1201.1469. Waldron, Nucl. Phys. B 812 (2009) 424-455 [e-Print:
[28] C. Romero, J. B. Fonseca-Neto, M. L. Pucheu, e-Print: arXiv:0810.2867].
arXiv:1101.5333. [48] M. Grigoriev, A. Waldron, e-Print: arXiv:1104.4994;
[29] P. A. M. Dirac, Naturwiss. 60 (1973) 529-531. A. R. Gover, A. Waldron, e-Print: arXiv:1104.2991;
[30] N. Rosen, Found. Phys. 12 (1982) 213-248. R. Bonezzi, O. Corradini, A. Waldron, e-Print:
[31] H. Cheng, Phys. Rev. Lett. 61 (1988) 2182; e-Print: arXiv:1003.3855.
math-ph/0407010. [49] M. P. Dabrowski, T. Denkiewicz, D. Blaschke, Ann.
[32] H. Wei, R.-G. Cai, JCAP 0709 (2007) 015 [e-Print: Phys. (Leipzig) 17 (2007) 237-257.
astro-ph/0607064]. [50] F. P. Poulis, J. M. Salim, e-Print: arXiv:1106.3031.
[33] P. K. Aluri, P. Jain, S. Mitra, S. Panda, N. K. [51] K. K. Nandi, B. Bhattacharjee, S. M. K. Alam, J. Evans.
Singh, Mod. Phys. Lett. A 25 (2010) 1349-1364 [e-Print: Phys. Rev. D 57 (1998) 823-828. [arXiv:0906.0181v1];
arXiv:0909.1070]; P. Jain, S. Mitra, Mod. Phys. Lett. Philip E. Bloomfield, Phys. Rev. D 59 (1999) 088501;
A 25 (2010) 167-177 [e-Print: arXiv:0903.1683]; P. K. K. K. Nandi, Phys. Rev. D 59 (1999) 088502.
Aluri, P. Jain, N. K. Singh, Mod. Phys. Lett. A 24 [52] R. Dick, Gen. Rel. Grav. 30 (1998) 435.
(2009) 1583-1595 [e-Print: arXiv:0810.4421]; P. Jain, S. [53] V. Faraoni, E. Gunzig, Int. J. Theor. Phys. 38 (1999)
Mitra, Mod. Phys. Lett. A 24 (2009) 2069-2079 [e-Print: 217-225 [e-Print: arXiv:astro-ph/9910176].
arXiv:0902.2525]. [54] P. Teyssandier, P. Tourrenc, J. Math. Phys. 24 (1983)
[34] S. Rouhani, M. V. Takook, M. R. Tanhayi, JHEP 1012 2793; T. Damour, G. Esposito-Farése, (1992), Class.
(2010) 044 [e-Print: arXiv:0903.2670]; S. Fatemi, S. Quant. Grav. 9 (1992) 2093; T. Damour, K. Norvedt,
Rouhani, M. V. Takook, M. R. Tanhayi, J. Math. Phys. Phys. Rev. Lett. 70 (1993) 2217; Phys. Rev. D 48 (1993)
51 (2010) 032503 [e-Print: arXiv:0903.5249]. 3436.
[35] T. Moon, P. Oh, J. Sohn, JCAP 1011 (2010) 005 [55] S. Bellucci, V. Faraoni, D. Babusci, Phys. Lett. A 282
[e-Print: arXiv:1002.2549]; T. Y. Moon, J. Lee, P. (2001) 357-361 [e-Print: arXiv:hep-th/0103180].
Oh, Mod. Phys. Lett. A 25 (2010) 3129-3143 [e-Print: [56] L. Jarv, P. Kuusk, M. Saal, Phys. Rev. D 76 (2007)
arXiv:0912.0432]. 103506 [e-Print: arXiv:0705.4644].
[36] M. Israelit, Gen. Rel. Grav. 43 (2011) 751-775 [e-Print: [57] A. Bhadra, K. Sarkar, D. P. Datta, K. K. Nandi,
arXiv:1008.0767]; Int. J. Mod. Phys. A 17 (2002) 4229- Mod. Phys. Lett. A 22 (2007) 367-376 [e-Print:
4237; Found. Phys. 32 (2002) 945-961. arXiv:gr-qc/0605109].
[37] R. Punzi, F. P. Schuller, M. N. R. Wohlfarth, Phys. Lett. [58] K. Nozari, S. Davood Sadatian, Mod. Phys. Lett. A 24
B 670 (2008) 161-164 [e-Print: arXiv:0804.4067]. (2009) 3143-3155 [e-Print: arXiv:0905.0241].
[38] J. Miritzis, Class. Quant. Grav. 21 (2004) 3043-3056 [e- [59] M. P. Dabrowski, J. Garecki, D. Blaschke, Ann. der Phys.
Print: gr-qc/0402039]. (Berlin) 18 (2009) 13-32.
[39] M. Novello, S. E. Perez Bergliaffa, Phys. Rept. 463 [60] E. Alvarez, J. Conde. Mod. Phys. Lett. A 22 (2002) 413-
(2008) 127-213. 420 [e-Print: arXiv:gr-qc/0111031].
[40] M. Novello, L. A. R. Oliveira, J. M. Salim, E. Elbaz, Int. [61] Y. Fujii, Prog. Theor. Phys. 118 (2007) 983-1018 [e-
J. Mod. Phys. D 1 (1992) 641-677. Print: arXiv:0712.1881].
[41] See Appendix D, p. 445 of Ref.[42]. [62] S. M. Carroll, e-Print: gr-qc/9712019.
[42] R. M. Wald, “General Relativity” (The University of [63] I. Quiros, R. Bonal, R. Cardenas, Phys. Rev. D 62 (2000)
Chicago Press, 1984). 044042 [e-Print: gr-qc/9908075].
[43] E. R. Harrison, Phys. Rev. D 6 (1972) 2077; V. A. Be- [64] N. Kaloper, K. A. Olive, Phys. Rev. D 57 (1998) 811-822
linskii, I. M. Khalatnikov, Sov. Phys. JETP 36 (1973) [e-Print: hep-th/9708008].
531; J. D. Bekenstein, Ann. Phys. 82 (1974) 535; J. D. [65] Y.-S. Piao, e-Print: arXiv:1112.3737.
Barrow, K. Maeda, Nucl. Phys. B 341 (1990) 294; J. P.
Abreu, P. Crawford, J. P. Mimoso, Class. Quant. Grav.

The Conformal Transformation's Controversy What Are We Missing

Uploaded by

Copyright:

Available Formats

The Conformal Transformation's Controversy What Are We Missing

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

The Conformal Transformation's Controversy What Are We Missing

Uploaded by

Copyright:

Available Formats

The conformal transformation’s controversy: what are we missing?

is a semantic one. For definiteness, an intuitively ”natural” statement of conformal equivalence is

I. INTRODUCTION The theory can be formulated in different sets of field

where ≡ g µν ∇µ ∇ν , and the Klein-Gordon (KG) equa- ¯ = 0,

ḡµν = Ω2 gµν , (7)

3 Here we deal with classical theories of the gravitational field.

so that the affine properties of space are not modified. In

The null geodesic equation is similar to Eq.(20) but with

In this section we will explore an aspect of the con- d

notice that while the metricity condition (27) is invariant

8 Compare with WI geodesic equations given in Eq.(20). Complete

means that the laws of gravitation are different in the

Both sets of dynamical equations (31) and (32), are to

Gµν = Ḡ(w) ¯ (w) ⇒ = Ω2

After the analysis in section V, it is almost trivial to

of what happens when one invokes invariance under gen-

where, as usual, t is the cosmic time and a(t) is the

A second group of authors believe the different confor-

• If follow the most widespread understanding of APPENDIX

several qualitative aspects of introducing matter into the

[25] P. Bouvier, A. Maeder, Astrophys. Space Sci. 54 (1977) 11 (1994) 1919.

You might also like