Classical and Quantum Gravity

PAPER You may also like

- Junction conditions in scalar–tensor
Viable constraint on scalar field in scalar–tensor theories
Luis Avilés, Hideki Maeda and Cristián
theory Martínez

- Invariant slow-roll parameters in

scalar–tensor theories
To cite this article: Chao-Qiang Geng et al 2020 Class. Quantum Grav. 37 115001 Piret Kuusk, Mihkel Rünkla, Margus Saal
et al.

- Universal inflationary attractors

implications on static neutron stars
V K Oikonomou
View the article online for updates and enhancements.

Recent citations
- Inverse-chirp imprint of gravitational wave
signals in scalar tensor theory
Chao-Qiang Geng et al

- Structure of Neutron Stars in Massive

Scalar-Tensor Gravity
Roxana Rosca-Mead et al

This content was downloaded from IP address 131.216.162.40 on 08/01/2022 at 01:55

 Classical and Quantum Gravity
Class. Quantum Grav. 37 (2020) 115001 (14pp) https://doi.org/10.1088/1361-6382/ab86fb

Viable constraint on scalar field in

scalar–tensor theory
Chao-Qiang Geng1,2,3,4,6 , Hao-Jui Kuan3 and Ling-Wei
Luo5
1
School of Fundamental Physics and Mathematical Sciences, Hangzhou Institute for
Advanced Study, UCAS, Hangzhou 310024, People’s Republic of China
2
International Centre for Theoretical Physics Asia-Pacific, Beijing/Hangzhou,
People’s Republic of China
3
Department of Physics, National Tsing Hua University, Hsinchu 300, Taiwan
4
Physics Division, National Center for Theoretical Sciences, Hsinchu 300, Taiwan
5
Institute of Physics, Academia Sinica, Taipei 11529, Taiwan

E-mail: [email protected]

Received 12 December 2019, revised 1 April 2020

Accepted for publication 6 April 2020
Published 4 May 2020

Abstract
The scalar–tensor theory can be formulated in both Jordan and Einstein frames,
which are conformally related together with a redefinition of the scalar field.
As the solution to the equation of the scalar field in the Jordan frame does not
have the one-to-one correspondence with that in the Einstein frame, we give
a criterion along with some specific models to check if the scalar field in the
Einstein frame is viable or not by confirming whether this field is reversible
back to the Jordan frame. We further show that the criterion in the first parame-
terized post-Newtonian approximation can be determined by the parameters of
the osculating approximation of the coupling function in the Einstein frame and
can be treated as a viable constraint on any numerical study in the scalar-tensor
scenario. We also demonstrate that the Brans–Dicke theory with an infinite
constant parameter ωBD is a counterexample of the equivalence between two
conformal frames due to the violation of the viable constraint.

Keywords: scalar–tensor theory, Jordan frame, Einstein frame, Brans–Dicke

theory

(Some figures may appear in colour only in the online journal)

6
Author to whom any correspondence should be addressed.

1361-6382/20/115001+14$33.00 © 2020 IOP Publishing Ltd Printed in the UK 1

Class. Quantum Grav. 37 (2020) 115001 C-Q Geng et al

1. Introduction

General relativity (GR) is a great theory as its predictions agree well with experiments. How-
ever, some strong-field phenomena may deviate from those of GR Jordan proposed a scalar
field which couples to the spacetime curvature to fulfill the hypothesis about the variation of
the constant of gravitation [1, 2]. Bergmann [3] and Fierz [4] used the Kaluza–Klein theory
[5] to get an effective gravitational constant. Among various alternative gravity theories, the
scalar-tensor (ST) one is the most natural extension to GR since it can be obtained by reducing
a higher dimensional theory into a four dimensional one together with a scalar field.
The most famous formulation of the ST theory was done by Brans and Dicke (BD) [6,
7], in which GR is explicitly modified by introducing a scalar field variably to determine the
universal Newtonian coupling. Subsequently, the BD theory was generalized by Bergmann [8]
and Wagoner [9] to include a scalar potential, which has been widely used to realize inflation
and the late-time cosmic acceleration of the Universe. Damour and Esposito-Farese [10] further
extended it with couplings between gravity and multi-scalar fields, in which, besides the tensor
gravitational field, one or more scalar fields are added in the gravitational sector with non-
minimal couplings.
There are two formalisms of the ST theory. One is given in the so called Jordan frame,
where the scalar field φ couples non-minimally to the Ricci scalar R but not directly to matter
with the kinetic term for the scalar field involving an arbitrary function ω(φ). The other is in
the Einstein frame, where the canonical scalar field ϕ is introduced to minimally couple to the
Ricci scalar R⋆ . The later one makes the field equations mathematically less complicated due
to the separation of the second-order derivatives of the gravitational variables g⋆µν and ϕ, but
the matter couplings depend on the scalar field. In order to have a frame transformation, we
need to assume that each scalar field is a functions of the other. In other words, we should have
φ(ϕ) to rephrase the system from the Jordan frame into the Einstein one, and vice versa. The
connection between these two frames is through a Weyl (conformal) transformation by taking
the Jordan frame metric gµν into the Einstein frame one g⋆µν , together with a redefinition of the
original scalar field φ into ϕ to have its kinetic term being a canonical form.
The Einstein frame is convenient to perform the numerical simulations and transparent to
discuss the cosmological applications. However, the issues on the physical interpretation and
equivalence of these two frames have been debated for a long time (e.g. [11, 12]). On the other
hand, there has been much less attention to the redefinition of the scalar field. In spite of many
efforts to show that these two conformal frames are equivalent, in general there are mainly two
problems from the redefinition of the scalar field, which are shown as follows:
(a) The existence of the Einstein frame is determined by whether the scalar field φ can be
described in terms of the new field ϕ. Furthermore, we need to transform the system back
to the Jordan frame. These operations are based on the requirement of the inequalities,
given by
dφ
6= 0 (Jordam to Einstein frame), (1a)
dϕ

dϕ
6= 0 (Einstein to Jordam frame), (1b)
dφ
which are related to each other. Note that the non-zero values of the derivatives should
also be finite. We call (1a) as the derivative constraints. In practice, one often study the
ST theories in the Einstein frame even though most of them do not perceive the inverse
problem of the solution in the Jordan frame. In other words, they do not consider the
2
Class. Quantum Grav. 37 (2020) 115001 C-Q Geng et al

possibility that their results may not be able to be transformed back into the Jordan frame.
Since the Jordan frame is assumed to be the physical frame, those fields in this case do
not bear any physical meanings. It is one of the purposes of the present work to give an
essential criterion to see if a solution in the Einstein frame can uniquely correspond to a
Jordan one. It will be shown that (1a) leads to the required condition.
(b) The coefficient of the kinetic term of ϕ determines the value of dϕ/dφ through the Weyl
transformation. However, the sign of the differential relation dϕ/dφ can be either ‘+’
or ‘−’, which has been mentioned in [13] but without further discussions therein, while
many just consider the positive sign as in [7, 14, 15]. Therefore, starting from a given ST
model in the Einstein frame, it usually corresponds to two different models in the Jordan
(physical) frame. We will give an example on this issue. In addition, the condition that
the relation dϕ/dφ is regular indicates that there is an irrelevant value of ϕ by which the
solution space of the scalar fields in the Einstein frame is divided into two branches [16].
Conversely, the value of ϕ can restrict the choice of the sign of the relation dϕ/dφ in some
cases. As a result, it arises the uniqueness problem of a model in the Jordan frame.
In this study, we explore the ST theory in the Jordan and Einstein frames by examining
the relations between them. In particular, we would like to find out the possible constraints
on the theory by requiring the one-to-one correspondence for the physical quantities after the
conformal transformation between the two frames.
The paper is organized as follows. In section 2, we first write down the ST action. We
then display the transformation between the Jordan and Einstein frames and demonstrate that
there is a constraint on the scalar transformation. In section 3, we take two specific models as
examples to illustrate how the constraint manifest itself as a criterion to check if the solution
in the Einstein frame is viable or not. We present our conclusions in section 4.

2. Scalar-tensor theories

2.1. Conformal frame freedom

A general action of the ST theory with a dimensionless single scalar field can be written as7
Z √
−g
4
F(φ)R − B(φ)gµν ∂µ φ∂ν φ − U(φ) + Sm [ψm , e2γ(φ) gµν ],


S= d x
16πG
(2)
where g and R associated with gµν respectively stand for the determinant and Riemann scalar
curvature, F(φ), B(φ), U(φ) and γ(φ) represent functions of φ and ψ m denotes the non-
gravitational fields. We note that F(φ) > 0 in the Jordan frame. The corresponding equations
of motion for gµν and φ are
   
1 1 1
Rµν = 8πG Tµν − gµν T + gµν U + B ∂µ φ∂ν φ , (3a)
F 2 2
and
1 
−Fφ R + Uφ − Bφ gµν ∂µ φ∂ν φ − 16πGαT ,

φ = (3b)
2B

7 The metric signature is (− + ++), while the Riemann tensor is given in terms of the Christoffel symbol by
Rα βµν := Γα βν,µ − Γα βµ,ν + Γα σµ Γσ βν − Γα σν Γσ βµ .

3
Class. Quantum Grav. 37 (2020) 115001 C-Q Geng et al

respectively, where the subscript φ denotes the partial derivative with respect to φ. Here, we
have used the relation
δSm √
:= −g αT (4)
δφ
where

α := γφ , (5)

and T := gµν Tµν with the stress energy tensor, given by

−2 δSm
Tµν := √ . (6)
−g δgµν
This form of the action is unchanged under a group of field redefinitions through a Weyl
transformation. Specifically, with the new metric g⋆µν and scalar field ϕ, defined by

φ = φ(ϕ), (7a)

gµν = e2Γ(φ(ϕ)) g⋆µν , (7b)

along with the coupling function Γ and necessary condition (1a), the action (1b) has the same
form up to a boundary term [17, 18], i.e.,
Z √ ⋆
−g ⋆
d4 x F ⋆ (ϕ)R⋆ − B ⋆ (ϕ)g⋆µν ∂µ ϕ∂ν ϕ − U ⋆ (ϕ) + Sm [ψm , e2γ (ϕ) g⋆µν ],


S=
16πG
(8)

through the definition of

1/2 dφ √
eΓ B − 6 FΓ2φ − 6 Fφ Γφ := ± B ⋆ . (9)
dϕ
The equation (9) yields that
s
dφ e−2Γ B ⋆
=± , (10)
dϕ B − 6 FΓ2φ − 6 Fφ Γφ

which should be non-zero and finite. Here, the transformed functions of F⋆ (ϕ), B ⋆(ϕ), U⋆ (ϕ)
and γ ⋆ (ϕ) are given by

F ⋆ (ϕ) = e2Γ(φ(ϕ)) F (φ(ϕ)) , (11a)

B ⋆(ϕ) = e2Γ(φ(ϕ)) φ2ϕ (ϕ) B (φ(ϕ)) − 6 F (φ(ϕ)) Γ2φ (φ(ϕ)) − 6 Fφ (φ(ϕ)) Γφ (φ(ϕ)) ,

(11b)
U ⋆ (ϕ) = e4Γ(φ(ϕ)) U (φ(ϕ)) , (11c)
γ ⋆ (ϕ) = γ (φ(ϕ)) + Γ (φ(ϕ)) , (11d)
respectively. Two special frames of the action often used in the literature are:
4
Class. Quantum Grav. 37 (2020) 115001 C-Q Geng et al

The Jordan f rame, which is characterized by γ = 0 in (2). In this frame, matter obeys the
weak equivalent principle (WEP) [6, 7, 19], meaning that freely falling objects follow the
geodesics of the Jordan frame metric [14]. But the shortcuts of this frame violates the energy
conditions. However, it is not a negative kinetic energy to be problematic, but an energy that
is unbounded from below. For instance, although a negative energy is usually associated with
the instability and runaway solution, the Minkowski space is stable against inhomogeneous
perturbations in ST [20]. Furthermore, a positive energy theorem has been shown to hold for a
special ST theory in the Jordan frame [21]. We regard the Jordan frame as the physical frame
in this work.
The Einstein f rame, which is characterized by F⋆ = 1, B ⋆ = constant and γ ⋆ 6= 0 in (8). By
contrast to the Jordan one, the energy conditions are satisfied in this frame, whereas the WEP
is violated [22].

2.2. Transformation between the Jordan and Einstein frames

In the Jordan frame, B(φ) = ω(φ)/φ and γ(φ) = 0 in (2), the action is given by
√  
ω(φ) µν
Z
−g
S = d4 x F(φ)R − g ∂µ φ∂ν φ − U(φ) + Sm [ψm , gµν ]. (12)
16πG φ
This action can be cast into a conformal frame by a Weyl transformation

gµν = A2 (φ)g⋆µν , (13)

where A(φ) = eΓ(φ) is the coupling function. We can now rewrite action (12) as
Z √ ⋆
−g
S= d x 4
R⋆ − 6 ⋆ (ln A) − 6g⋆µν (−1)2∂µ (ln A)∂ν (ln A)
16πG

ω
− A2 g⋆µν ∂µ φ∂ν φ − A4 U(φ) + Sm [ψm , A2 g⋆µν ], (14)
φ

where we have used the relation of FA2 = 1 to obtain the action in the Einstein frame.
Consequently, by using
1
∂µ (ln A) = − Fφ ∂µ φ, (15)
2F
we can simplify (14) to

√ ⋆" ! #
3Fφ2 ω
Z
−g
S= 4
d x R⋆ − 2g⋆µν + ∂µ φ∂ν φ − A U(φ) + Sm [ψm , A2 g⋆µν ]
4
16πG 4F 2 2φF
(16)

up to a divergence term of 6⋆ (ln A−1 ). Through the redefinition of the scalar field with
s
dϕ 3Fφ2 ω
:= ± + , (17)
dφ 4F 2 2φF

we can obtain the derivative of the canonical field ϕ = ϕ(φ) in the Einstein frame. The
equation (17) is important and should satisfy the derivative constraints in (1a). It will be shown
5
Class. Quantum Grav. 37 (2020) 115001 C-Q Geng et al

that the sign can determine the range of ϕ in particular models in section 3. As a result, (16)
reads as
Z √ ⋆
−g
S = d4 x R⋆ − 2g⋆µν ∂µ ϕ∂ν ϕ − 4 V(ϕ) + Sm [ψm , A2 g⋆µν ],


(18)
16πG

where V(ϕ) := A4 U(φ)/4. Obviously, the matter fields couple non-minimally to the scalar field
through A2 (φ) in (18).
By variating (18) with respect to g⋆µν and ϕ, we derive the equations of motion of the tensor
and scalar fields to be
 
1 ⋆ ⋆
⋆ ⋆
Rµν = 8πG Tµν − T gµν + 2∂µ ϕ∂ν ϕ + 2Vg⋆µν , (19a)
2

dV
⋆ ϕ = −4πGα(ϕ)T ⋆ + , (19b)
dϕ
respectively, where α(ϕ) is a function of ϕ in the Einstein frame, defined by

d ln A Fφ dφ
α(ϕ) = =− , (20)
dϕ 2F dϕ
which is closely related to the derivative constraints. It can be checked that the equations of
motion can be read by substituting F = 1, B = 2, φ = ϕ and U = 4V as well as γ = ln A into
(3a), (3b) and (5).
Clearly, it is non-trivial that the equations of (11c) derived by variating (18) with g⋆µν and ϕ
in the Einstein frame are equivalent to those from the variations of the action (12) with respect
to gµν and φ in the Jordan frame. From the inverse function theorem, φ(ϕ) exists as long as
(17) never vanishes, i.e., the derivative constraints. Hence, the solution to the scalar equation
in the Einstein frame must satisfy (1a). Otherwise, it is not a solution to the scalar equation in
the Jordan frame, which is identified as the physical equation of motion.
There is another way to understand the derivative constraints. It is clear that U(φ) is a phys-
ical potential as it is defined in the Jordan frame, whereas V(ϕ) is an auxiliary one. In general,
if there is a potential V(ϕ) in the Einstein frame, one cannot claim that a solution ϕ of (19b)
is physical unless U(φ) = 4A−4 (φ)V(ϕ(φ)) exists and is unique. It is apparent that V can be a
function of φ if and only if ϕ is that of φ. As a result, the derivative constraints should hold.
From the derivative constraints, we can further show that
dϕ dϕ
>0 or < 0, (21)
dφ dφ
representing the choice of the sign in the RHS of (17), which cannot cross the critical value
of dϕ/dφ = 0 during the evolution of the scalar field [23]. Moreover, to bear the one-to-one
correspondence between F and φ, Fφ must be non-vanishing, so that the sign of Fφ cannot
flip either. Under the condition of Fφ 6= 0, the action (12) can be rewritten into that in the
Brans–Dicke–Bergmann–Wagoner (BDBW) theory [24].

2.3. PPN parameters in scalar–tensor theories

As discussed in the literature [18, 23, 25–27], the deviation from GR for a ST theory in the
parameterized post-Newtonian (PPN) regime can be expressed in terms of the asymptotic value
of α at spatial infinity as well as its successive derivatives.
6
Class. Quantum Grav. 37 (2020) 115001 C-Q Geng et al

At the first order of PPN, it has been shown that the coupling function is sufficient to be
determined by the osculating approximation [13]
1
ln A = α0 (ϕ − ϕ0 ) + β0 (ϕ − ϕ0 )2 (22)
2
and
α = α0 + β0 (ϕ − ϕ0 ) (23)

where ϕ0 is the asymptotic value of ϕ at spatial infinity and α0 and β 0 are constants, defined
as

α0 := α(ϕ0 ), (24a)

dα
β0 := (ϕ0 ). (24b)
dϕ
Classically, for the massive scalar field, ϕ0 should be determined by the ground state of ϕ,
i.e., the global minimum of the potential V(ϕ), which satisfies

dV 
lim =0 (25)
r→∞ dϕ 
ϕ=ϕ0

and
d2 V 

lim > 0. (26)
r→∞ dϕ2 ϕ=ϕ0

It turns out that if one expresses the potential in terms of the Taylor expansion around ϕ =
ϕ0 , the coefficient of the linear term should vanish. To obtain the ground state naturally, it is
convenient to consider a shifted field of

ϕ′ = ϕ − ϕ0 , (27)

which simplifies (22) to be

1
ln A = α0 ϕ′ + β0 ϕ′ 2 . (28)
2
Subsequently, we obtain

ln F = −2α0 ϕ′ − β0 ϕ′ 2 (29)

through the help of the relation FA2 = 1. Note that α(ϕ) = 0 with α0 = β 0 = 0 leads to the
result in GR with the coupling function A being identically unit, implying that the scalar field
couples to the Riemann curvature minimally in the Jordan frame. Consequently, the potential
becomes a Taylor series of ϕ′ at ϕ′ = 0 without the linear term, given by

1 d2 V 

V(ϕ′ ) = V0 + ϕ′ 2 + · · · . (30)
2! dϕ′2 ϕ′ =0

where V0 := V(ϕ0 ) is the minimum of the potential V, which gives no contribution to the
equation of motion of ϕ′ .
Furthermore, as a radiative coordinate [10, 29] can be constructed in the Einstein frame
with the metric to be the Minkowski one asymptotically, one finds that gµν (r → ∞) ≈ A20 ηµν
7
Class. Quantum Grav. 37 (2020) 115001 C-Q Geng et al

with A0 := A(ϕ0 ) in the Jordan frame. Within the conditions (25) and (26), the ground state of
the potential leads to ϕ → ϕ0 (ϕ′ → 0) asymptotically, resulting in

A0 = 1 (31)

As a result, (22) can be used to make sure that the Jordan frame metric gµν ≈ η µν is
asymptotically flat.
By following the discussions in references [15, 27, 28], α0 (β 0 ) is positive (negative) to
derive the phenomena of the spontaneous scalarization. The non-vanishing property of (1a)
implies that the parameter of α in (20) can never be zero, i.e., the α-constraint with

α 6= 0, (32)

which indicates that there exists an unacceptable critical value for ϕ′ , denoted as
α0
ϕ′c := − , (33)
β0
since this value leads to α = 0 from (23). Clearly, the derivative constraints in (1a) are
equivalent to the α-constraint in (32). Therefore, the forbidden value of ϕ is

ϕc = ϕ′c + ϕ0 . (34)

This illustrates that the solution space of the scalar field ϕ in the Einstein frame is divided into
two branches by the value of ϕc .

3. Constraints on specific models

In a specific model, the condition (1a) would manifest itself as a critical value ϕc . We will show
that the scalar field can be nowhere equal to ϕc as it cannot be a solution to the scalar equation in
the Jordan frame. Hence, the solution space of the scalar field in the Einstein frame is separated
by ϕc into two pieces. Moreover, in the PPN regime, we demonstrate that this critical value is
determined by the coefficients of the coupling function (22). In this section, we will take the
general Jordan frame with the action (12) and BDBW theories as two examples for discussions.
In particular, we concentrate on the potential in the Einstein frame with the form of

V(ϕ) = m2 ϕ2 , (35)

which has the minimum of V0 = 0 at ϕ0 = 0. Hereafter, we drop out the superscript of prime
on ϕ′ for simplicity due to ϕ′ = ϕ. As a result, the parabola of (29) is shown as figure 1.

3.1. General Jordan frame action

From (29) with ϕ0 = 0, we get that

q
α20 − β0 ln F
ϕ = ϕc ± . (36)
β0
With the two signs in (36), there are different relations for ϕ(φ) or φ(ϕ), resulting in different
possible potentials of U(φ). Therefore, a given model in the Einstein frame may correspond to
two models in the Jordan frame.
Moreover, the sign in (36) is associated with the behavior of ϕ. We consider two situations:
(a) ϕ is not always positive and (b) ϕ is always positive.
8
Class. Quantum Grav. 37 (2020) 115001 C-Q Geng et al

Figure 1. The parabola reveals the relation between ln F and ϕ, where the dashed line
represents the forbidden value ϕc for ϕ. Particularly, we have F = φ for the BDBW
theory.

For (a), like all the solutions in [15], oscillating across ϕ = 0, we should choose the ‘+’
sign in (36), providing that ϕ < ϕc 8 Therefore, once the sign in (36) is determined, the other
branch of the solution space represented by ϕ > ϕc should collapse, so that ϕc can be viewed
as a ceiling of ϕ. In addition, we note that the collapsed region of ϕ > ϕc is independent of
the potential. Any value of ϕ exceeding this ceiling is not viable as it is not a solution to the
scalar equation in the Jordan frame associated with the current Einstein frame. For example, the
evolution of the scalar field in the Einstein frame during the process of the supernovae explosion
has been investigated in [15]. The solution labeled by α0 = 10−4 violates this constraint as
illustrated in figure 2.
Although the solution space is constrained by ϕ < ϕc , we still cannot specify a unique
model in the Jordan frame due to the lack of the information about Fφ to determine the form
of the relation φ(ϕ). However, we can constrain α to be positive as ϕ < ϕc , which gives rise
to

dφ
Fφ <0 (37)
dϕ

due to (20). Clearly, the positivity of Fφ implies that the sign of (17) should be minus and
vice versa. Furthermore, we have two models in the Jordan frame based on the ‘±’ signs of
Fφ , within ϕ < ϕc . For (b), there are two cases. The first one corresponds to the ‘−’ sign in
(36) with ϕ > ϕc . Similar to the argument above, there exists two models in the Jordan frame
related to the sign of Fφ . In this case with the equivalence of α < 0, we have the relation

dφ
Fφ > 0. (38)
dϕ

By contrast, the case for the ‘+’ sign requires 0 < ϕ < ϕc . The solution space is bounded,
which is the most restrictive one among the cases.

8 If we choose the ‘−’ sign in (36), ϕ must be positive and greater than ϕc due to β 0 < 0.

9
Class. Quantum Grav. 37 (2020) 115001 C-Q Geng et al

Figure 2. Waveforms of the scalar field, extracted at rex = 5 × 109 cm away from the
supernovae core in the simulation of [15], where the solid line represents that with α0 =
10−4 in figure 1 of [15] and the dashed horizontal line corresponds to the critical value
of ϕc = 5 × 10−6 .

By differentiating both sides of (36) with respect to φ, one gets an additional equation

dϕ ∓Fφ
= q , (39)
dφ 2F α20 − β0 ln F

which suggests that Fφ determines the sign of dϕ/dφ since F > 0. We note that the sign in
(53) is opposite to the one chosen in (36). The requirement of ln F > α20 /β0 to prevent ϕ from
being imaginary is fulfilled because (29) has minimum α20 /β0 at the forbidden point ϕc , which
can easily be checked from figure 1. One can see that the discussions of the sign problem of
Fφ above are consistent with the sign in (53) .

3.2. Brans–Dicke–Bergmann–Wagoner theory

Another example is the BDBW theory with F = φ in the action (12), particularly given by
√  
ω(φ) µν
Z
−g
S= 4
d x φR − g ∂µ φ∂ν φ − U(φ) + Sm [ψm , gµν ]. (40)
16πG φ

It can be found that the sign of dϕ/dφ can totally be determined through α due to Fφ = 1.
In this subsection, we will transform the potential defined in the Einstein frame back into the
Jordan one and discuss the ambiguity arising from this transformation. In this specific case,
since F(φ) = A−2 is simply φ, we have

φ = A−2 > 0, (41)

10
Class. Quantum Grav. 37 (2020) 115001 C-Q Geng et al

leading to

ln φ = −2α0ϕ − β0 ϕ2 , (42)

which is also illustrated in figure 1.

From figure 1, it is easy to see that the condition (1a) is equivalent to ϕ 6= ϕc . As a result,
one finds that (23) can never be zero. Consequently, (20) reads as
1 dφ
α=− . (43)
2φ dϕ
From (57), we obtain the inequality by (55)
dϕ 1
α =− < 0. (44)
dφ 2φ
As mentioned early that dϕ/dφ is either positive or negative, the valid value of ϕ can fix the
sign in (17) as follows. For ϕ > ϕc , we have α < 0, so that one has
√
dϕ 3 + 2ω
= (45)
dφ 2φ
by (58). Similarly, the redefinition of the scalar fields leads to
√
dϕ 3 + 2ω
=− (46)
dφ 2φ
for ϕ < ϕc .
In the limit of the BD theory, ω in (54) is constant, i.e., ω(φ) = ω BD . Consequently, the
equation in (57) reveals that
1
α = α0 + β0 ϕ = ∓ √ (47)
3 + 2ωBD
where the minus (plus) case corresponds to the branch of ϕ > ϕc (ϕ < ϕc ). Since the right-
hand side of (61) is constant, we have two possibilities: (i) β 0 = 0 and (ii) ϕ is almost a
constant. Due to the negativity of β 0 in this paper, the solution is restricted to the second one.
It is interesting to note that ω BD → ∞ is a critical value, equivalent to ϕ = ϕc due to α = 0.
Even though this case is consistent with those in the literature and in turn reproduces the results
in GR in the sense that all its predictions become indistinguishable from GR in the Jordan
f rame [30], it cannot be transformed to the Einstein f rame due to the violation of the derivative
constraints and α-constraint. Therefore, the case of the infinite ω BD is improper to be discussed
in the Einstein f rame, which shows the inequivalence between the two frames.
We again note that the critical value of ϕc is so generic that the results of all the simulations
should obey the criterion ϕ 6= ϕc . For instance, it has been concentrated on the BDBW theory
to investigate the dynamical scalarization of the neutron star binaries in ST in [31]. The critical
value for the scalar field in their setting is ϕc = 0, and the results therein do all satisfy this
criterion, i.e., no crossing the line of ϕ = 0.
Starting from (56) in the BDBW theory, we get
q
α20 − β0 ln φ
ϕ = ϕc ± . (48)
β0
11
Class. Quantum Grav. 37 (2020) 115001 C-Q Geng et al

Since U = 4VA−4 = 4m2 ϕ2 φ2 as defined in (18), we recover the potential U in the Jordan
frame, given by
 q 
U(φ) = 4 m2 β0−2 2α20 − β0 ln φ ∓ 2α0 α20 − β0 ln φ φ2 (49)

and
U(ϕ) = 4 m2 ϕ2 exp −4α0 ϕ − 2β0 ϕ2 ,


(50)

in terms of φ and ϕ, respectively, indicating two different potentials in the Jordan frame, which
depend on the branches of the scalar field in the Einstein frame, i.e., the sign chosen in (62).
For the case of the action (12) in section 3.1, one cannot consider only the ‘+’ sign in (17),
which is equivalent to choose one model from two prospective ones.
In addition, the undetermined signs in (62) can be eliminated by considering the function of
(ϕ − ϕc )2 . Hence, the solution space has an Z2 symmetry with respect to ϕc . In other words,
the model in the Jordan frame, which is responsible for the given model in the Einstein frame,
remains the same under the transformation ϕ − ϕc ↔ −(ϕ − ϕc ). We can shift ϕ to ϕ̄ = ϕ −
ϕc , resulting in that the existence of the Z2 symmetry is characterized by an even potential
U(ϕ̄) of ϕ̄. The potential (64), which can be rewritten as

U(ϕ̄) = 4 m2 (ϕ̄ + ϕc )2 exp −2β0 ϕ̄2 − ϕ2c ,



(51)

is not a even function of ϕ̄ apparently. As a result, the one to one correspondence between
models in both frames holds strictly. In general, it is hard to have the same symmetry of Z2 at
ϕc for U(ϕ). Particularly, the direct effect of the Z2 symmetry at ϕc for the potential is that the
model in the Einstein frame is associated with a unique model in the Jordan frame.

4. Conclusions

In order to use the formulation of the ST theories in the Einstein frame, we have investigated
the redefinition of the scalar. We have found that the regularity of such redefinition (17), which
comes from the requirement dϕ/dφ 6= 0 and has been shown to be equivalent to the condition
α 6= 0, gives an irrelevant value ϕc of the scalar field in the Einstein frame. The value of ϕc
separates the solution space of the scalar field into two viable regions. Any scalar field in the
Einstein frame, which crosses the irrelevant value, cannot be the solution in the Jordan frame.
The two signs in (17) result in two different models in the Jordan frame. In some special
cases, such as that in the BDBW theory without a potential, these two models in the Jordan
frame coincide with each other. In general, the sign of (17) can be obtained by Fφ and ϕ. For
the case with one branch of the solution space collapsed, the sign of (17) can be fixed by the
sign of Fφ alone. In such a case, it is the ambiguity of the sign of Fφ that gives rise to two
prospective models in the Jordan frame. On the other hand, in some models with Fφ being
constant, such as the BDBW theory, the sign of (17) is fully determined by the branch of the
scalar fields. If we consider the asymptotic value ϕ0 6= 0, (36) should be modified as
q
α20 − β0 ln F
ϕ − ϕ0 = ϕ′c ± . (52)
β0

Together with (34), one can also conclude that if the potential U(ϕ) in the Jordan frame is an
even function with respect to the value of ϕ = ϕc , the correspondence between the models in
12
Class. Quantum Grav. 37 (2020) 115001 C-Q Geng et al

two frames is one to one. Hence, by satisfying the α-constraint in (32) the results in the Einstein
frame can be well-defined under this condition.
We have demonstrated that the critical value ϕc provides a viable constraint on the formu-
lation of the ST theories in the Einstein frame, which is even independent of the form of the
potential. Clearly, all the numerical results must obey the criterion of no crossing the line of
the critical value. Otherwise, they will not have any physical meaning in the Einstein frame. In
addition, we have also shown that the case of the infinite ω BD in the BD theory in the Einstein
frame is irrelevant due to the violation of the derivative constraints and α-constraint.
Furthermore, for the case without a potential, two branches of the solution space of the
scalar field in the Einstein frame correspond to the same model in the Jordan one. However,
the solution branch of ϕ should be specified because the α-constraint is strict with the behavior
of ϕ during the evolution in the Einstein frame.
Finally, we conclude that the critical value ϕc induced from the α-constraint or derivative
constraints and the sign determination in (17) are two important issues to study the ST theory
in the Einstein frame.

Acknowledgments

We would like to thank Patrick Chi-Kit Cheong for useful discussions. The work was partially
supported by National Center for Theoretical Sciences, Ministry of Science and Technology
(MoST-107-2119-M-007-013-MY3 and MoST-108-2811-M-001-598), and Academia Sinica
Career Development Award Program (AS-CDA-105-M06).

ORCID iDs

Chao-Qiang Geng https://orcid.org/0000-0002-7303-4581

Hao-Jui Kuan https://orcid.org/0000-0002-3480-3448

References

[1] Jordan P 1955 Schwerkraf t und weltall (Braunschweig: Vieweg)

[2] Jordan P 1959 Z. Phys. 157 112
[3] Bergmann P G 1948 Ann. Math. 49 255–64
[4] Fierz M 1956 Helv. Phys. Acta 29 128
[5] Kaluza T 1921 Sitzungsber. Preuss. Akad. Wiss. Berlin Math. Phys. 1921 966
Klein O 1926 Z. Phys 37 895
Klein O 1986 Surv. High Energy Phys. 5 241
[6] Brans C and Dicke R H 1961 Phys. Rev. 124 925
[7] Dicke R H 1962 Phys. Rev. 125 2163
[8] Bergmann P G 1968 Int. J. Theor. Phys. 1 25
[9] Wagoner R V 1970 Phys. Rev. D 1 3209
[10] Damour T and Esposito-Farese G 1992 Class. Quantum Grav. 9 2093
[11] Faraoni V and Nadeau S 2007 Phys. Rev. D 75 023501
[12] Capozziello S, Martin-Moruno P and Rubano C 2010 Phys. Lett. B 689 117
[13] Damour T 2007 arXiv:0704.0749
[14] Fujii Y and Maeda K 2007 The Scalar-Tensor Theory of Gravitation (Cambridge: Cambridge
University Press)
[15] Sperhake U, Moore C J, Rosca R, Agathos M, Gerosa D and Ott C D 2017 Phys. Rev. Lett. 119
201103
[16] Geng C Q, Kuan H J and Luo L W 2020 in preparation
[17] Flanagan E E 2004 Class. Quantum Grav. 21 3817

13
Class. Quantum Grav. 37 (2020) 115001 C-Q Geng et al

[18] Järv L, Kuusk P, Saal M and Vilson O 2014 J. Phys. Conf . Ser. 532 012011
[19] Salgado M 2006 Class. Quantum Grav. 23 4719
[20] Faraoni V 2004 Phys. Rev. D 70 081501
[21] Bertolami O 1987 Phys. Lett. B 186 161
[22] Brans C H 2005 arXiv:gr-qc/0506063
[23] Jarv L, Kuusk P and Saal M 2007 Phys. Rev. D 76 103506
[24] Clifton T, Ferreira P G, Padilla A and Skordis C 2012 Phys. Rept. 513 1
[25] Damour T and Esposito-Farese G 1996 Phys. Rev. D 53 5541
[26] Damour T and Esposito-Farese G 1996 Phys. Rev. D 54 1474
[27] Damour T and Esposito-Farese G 1993 Phys. Rev. Lett. 70 2220
[28] Novak J 1998 Phys. Rev. D 58 064019
[29] Blanchet L 1987 Proc. R. Soc. A 409 383
[30] Will C M 2014 Living Rev. Rel. 17 4
[31] Palenzuela C, Barausse E, Ponce M and Lehner L 2014 Phys. Rev. D 89 044024

14