Dark Energy Oscillations in The Matter Dominated Era

Cosmic history of viable exponential gravity: Equation of state oscillations and
growth index from inflation to dark energy era

Kazuharu Bamba1,∗ , Antonio Lopez-Revelles2,3,† ,
R. Myrzakulov4,‡ , S. D. Odintsov2,4,5,§ and L. Sebastiani3,4,¶
1
Kobayashi-Maskawa Institute for the Origin of Particles and the Universe, Nagoya University, Nagoya 464-8602, Japan
2
Consejo Superior de Investigaciones Cientı́ficas,
ICE/CSIC-IEEC, Campus UAB, Facultat de Ciències,
Torre C5-Parell-2a pl, E-08193 Bellaterra (Barcelona), Spain
3
Dipartimento di Fisica, Università di Trento and Istituto Nazionale di Fisica Nucleare, Gruppo Collegato di Trento, Italia
4
Eurasian International Center for Theoretical Physics and Department of General & Theoretical Physics,
Eurasian National University, Astana 010008, Kazakhstan
5
Institució Catalana de Recerca i Estudis Avançats (ICREA), Spain
arXiv:1207.1009v2 [gr-qc] 13 Nov 2012
A generic feature of viable F (R) gravity is investigated: It is demonstrated that during the
matter dominated era the large frequency oscillations of the effective dark energy may influence
the behavior of higher derivatives of the Hubble parameter with the risk to produce some singular
unphysical solutions at high redshift. This behavior is explicitly analyzed for realistic F (R) models,
in particular, exponential gravity and a power form model. To stabilize such oscillations, we consider
the additional modification of the models via a correction term which does not destroy the viability
properties. A detailed analysis on the future evolution of the universe and the evolution history
of the growth index of the matter density perturbations are performed. Furthermore, we explore
two applications of exponential gravity to the inflationary scenario. We show how it is possible to
obtain different numbers of e-folds during the early-time acceleration by making different choices
of the model parameters in the presence of ultrarelativistic matter, which destabilizes inflation
and eventually leads to the exit from the inflationary stage. We execute the numerical analysis of
inflation in two viable exponential gravity models. It is proved that at the end of the inflation,
the effective energy density and curvature of the universe decrease and thus a unified description
between inflation and the ΛCDM-like dark energy dominated era can be realized.
PACS numbers: 04.50.Kd, 95.36.+x, 98.80.-k
I. INTRODUCTION
The current cosmic acceleration is supported by various observations such as Supernovae Ia (SNe Ia) [1], large scale
structure (LSS) [2] with baryon acoustic oscillations (BAO) [3], cosmic microwave background (CMB) radiation [4–6]
and weak lensing [7]. There exist two representative procedures to solve this problem, namely, introducing “dark
energy” in general relativity (for recent reviews in terms of dark energy, see [8–10]) and modifying the gravitational
theory like F (R) gravity (for recent reviews on modified gravity, see [11–15]). In this paper, we adopt modified gravity
approach to describe the inflation and dark energy eras.
There proposed several viable F (R) gravity models have been constructed (for concrete viable models, see, e.g., the
above reviews or [16] and references therein). The conditions for the viability are summarized as follows: (i) Positive
definiteness of the effective gravitational coupling. (ii) Matter stability condition [17–20]. (iii) In the large curvature
regime, the model is close to the Λ-Cold-Dark-Matter (ΛCDM) model asymptotically. (iv) Stability of the late-time
de Sitter point [21, 22]. (v) The equivalence principle. (vi) Solar-system tests [17, 23, 24]. It is considered to be one
of the most significant issues on cosmology in the framework of F (R) gravity to realize the unification of inflation
with the late time cosmic acceleration [11, 17] (for the first proposal of 1/R gravity as gravitational alternative for
dark energy, see [25, 26]).
In this paper, we study a generic feature of viable F (R) gravity models, in particular, exponential gravity and a
power form model. We find that the behavior of higher derivatives of the Hubble parameter may be influenced by
large frequency oscillations of effective dark energy, which makes solutions singular and unphysical at a high redshift.
Therefore, in order to stabilize such oscillations, we examine an additional correction term to the model and remove
∗ E-mail address: [email protected]

† E-mail address: [email protected]
‡ E-mail address: [email protected], [email protected]
§ E-mail address: [email protected], also at TSPU, Tomsk
¶ E-mail address: [email protected]
2
such an instability with keeping the viability properties. We also demonstrate the cosmological evolutions of the
universe and growth index of the matter density perturbations in detail. Furthermore, by applying two viable models
of exponential gravity to inflationary cosmology and executing the numerical analysis of the inflation process, we
illustrate that the exit from inflation can be realized. Concretely, we demonstrate that different numbers of e-folds
during inflation can be obtained by taking different model parameters in the presence of ultrarelativistic matter, the
existence of which makes inflation end and leads to the exit from inflation. Indeed, we observe that at the end of
the inflation, the effective energy density as well as the curvature of the universe decrease. Accordingly, a unified
description between inflation and the late time cosmic acceleration is presented. We use units of kB = c = ~ = 1 and
denote the gravitational constant 8πG by κ2 ≡ 8π/MPl 2 with the Planck mass of MPl = G−1/2 = 1.2 × 1019 GeV.
The paper is organized as follows. In Sec. II, we briefly review the formulations of F (R) gravity. We use the fluid
representation of F (R) gravity [27]. Here, in the Friedmann-Lemaı̂tre-Robertson-Walker (FLRW) background, the
equations of motion with the addition of an effective gravitational fluid are presented. In Sec. III, we explain two
well-known viable F (R) gravity models and show those generic feature occurring in the matter dominated era, when
large frequency oscillation of dark energy appears and influences on the behavior of higher derivatives of the Hubble
parameter in terms of time with the risk to produce some divergence and to render the solution unphysical. Thus, we
suggest a way to stabilize such oscillations by introducing an additive modification to the models. We also perform
a numerical analysis of the matter dominated era. In Sec. IV, we demonstrate that the term added to stabilize the
dark energy oscillations in the matter dominated epoch does not cause any problem on the viability of the models,
which satisfy the cosmological and local gravity constraints. We investigate their future evolution and show that the
effective crossing of the phantom divide, which characterizes the de Sitter epoch, takes place in the very far future.
We also analyze the growth index using three different ansatz choices. The second part of the paper is devoted to the
study of F (R) models for the unification of the early-time cosmic acceleration, i.e., inflation, and the late-time one.
In Sec. V, we explore two applications of exponential gravity for inflation. In particular, we show how it is possible to
obtain different numbers of e-folds during inflation by making different choices of model parameters in the presence of
ultrarelativistic matter in the early universe. In Sec. VI, we execute the numerical analysis of inflation and illustrate
that at the end of it the effective energy density and the curvature decrease and eventually the cosmology in the
ΛCDM model can follow. Finally, the summary and outlook for this work are given in Sec. VII. For reference, we
also explain the procedure of conformal transformation in Appendix A and asymptotically phantom or quintessence
modified gravity in Appendix B.
II. F (R) GRAVITY AND ITS DYNAMICS IN THE FLRW UNIVERSE: GENERAL OVERVIEW
In this section, we briefly review formulations in F (R) gravity and derive the gravitational field equations in the
FLRW space-time. The action describing F (R) gravity is given by
4 √

F (R)
Z
(matter)
I= d x −g +L , (2.1)
M 2κ2
where F (R) is a generic function of the Ricci scalar R only, g is the determinant of the metric tensor gµν , L(matter) is the
matter Lagrangian and M denotes the space-time manifold. In a large class of modified gravity models reproducing
the standard cosmology in General Relativity (GR), i.e., F (R) = R, with a suitable correction to realize current
acceleration and/or inflation, one represents
F (R) = R + f (R) . (2.2)
Thus, the modification of gravity is encoded in the function f (R), which is added to the classical term R of the
Einstein-Hilbert action in GR. In what follows, we discuss modified gravity in this form by explicitly separating the
contribution of its modification from GR. The field equation simply reads

1 1
F ′ (R) Rµν − Rgµν = κ2 Tµν (matter)
+ gµν (F (R) − RF ′ (R)) + (∇µ ∇ν − gµν ) F ′ (R) . (2.3)
2 2
Here, ∇µ is the covariant derivative operator associated with gµν , φ ≡ g µν ∇µ ∇ν φ is the covariant d’Alembertian for
µ(matter)
a scalar field φ, and Tν = diag (−ρm , Pm , Pm , Pm ) is the contribution to the stress energy-momentum tensor
from all ordinary matters, with ρm and Pm being the energy density and pressure of matter, respectively. Moreover,
the prime denotes the derivative with respect to the curvature R.
The flat FLRW space-time is described by the metric ds2 = −dt2 + a(t)2 dx2 , where a(t) is the scale factor of the
universe. The Ricci scalar reads
R = 12H 2 + 6Ḣ , (2.4)
3
where H = ȧ(t)/a(t) is the Hubble parameter and the dot denotes the time derivative of ∂t (≡ ∂/∂t). In the flat
FLRW background, from the (µ, ν) = (0, 0) component and the trace part of (µ, ν) = (i, j) (with i, j = 1, · · · , 3)
components in Eq. (2.3), we obtain the gravitational field equations [11]
3 2
ρeff = H , (2.5)
κ2
1
Peff = − 2 2Ḣ + 3H 2 . (2.6)
κ
Here, ρeff and Peff are the effective energy density and pressure of the universe, respectively, defined as
1 h i
ρeff ≡ ρm + 2 (F ′ R − F ) − 6H 2 (F ′ − 1) − 6H Ḟ ′ , (2.7)
2κ
1 h i
Peff ≡ Pm + 2 − (F ′ R − F ) + (4Ḣ + 6H 2 )(F ′ − 1) + 4H Ḟ ′ + 2F̈ ′ . (2.8)
2κ
In this way, we have a fluid representation of the so-called geometrical dark energy in F (R) gravity with the energy
density ρDE = ρeff − ρ and pressure PDE = Peff − P . However, it is important for us to remember that gravitational
terms enter in both left and right sides of Eqs. (2.5) and (2.6). For general relativity in which F (R) = R, ρeff = ρm
and Peff = Pm and therefore Eqs. (2.5) and (2.6) lead to Friedman equations.
We also explain basic equations that we use to carry out our analysis. In order to study the dynamics of F (R)
gravity models in the flat FLRW universe, we may introduce the variable [16, 28]
ρDE H2
yH (z) ≡ = 2 − (z + 1)3 − χ(z + 1)4 . (2.9)
ρm(0) m̃
Here, ρm(0) is the energy density of matter at the present time, m̃2 is the mass scale, given by
κ2 ρm(0)
m̃2 ≡ ≃ 1.5 × 10−67 eV2 ,
3
and χ is defined as [5]
ρr(0)
χ≡ ≃ 3.1 × 10−4 ,
ρm(0)
where ρr(0) is the current energy density of radiation and z = 1/a(t) − 1 is the redshift. Here, we have taken the
current value of the scale factor as unity. By using Eqs. (2.5) and (2.9), we find
d2 yH (z) dyH (z)

+ J1 + J2 (yH (z)) + J3 = 0 , (2.10)
dz 2 dz
where
1 − F ′ (R)

1 1
J1 = −3 − , (2.11)
(z + 1) yH + (z + 1)3 + χ(z + 1)4 6m̃2 F ′′ (R)
2 − F ′ (R)

1 1
J2 = , (2.12)
(z + 1)2 yH + (z + 1)3 + χ(z + 1)4 3m̃2 F ′′ (R)
J3 = −3(z + 1)
(1 − F ′ (R))((z + 1)3 + 2χ(z + 1)4 ) + (R − F (R))/(3m̃2 ) 1
− . (2.13)
(z + 1)2 (yH + (z + 1)3 + χ(z + 1)4 ) 6m̃2 F ′′ (R)
Furthermore, the Ricci scalar is expressed as

dyH (z)
R = 3m̃2 4yH (z) − (z + 1) + (z + 1)3 . (2.14)
dz
In deriving this equation, we have used the fact that −(z + 1)H(z)d/dz = H(t)d/d(ln a(t)) = d/dt, where H could
be an explicit function of the red shift as H = H(z), or an explicit function of the time as H = H(t). In general,
Eq. (2.10) can be solved in a numerical way, once we write the explicit form of an F (R) gravity model.
4
III. GENERIC FEATURE OF REALISTIC F (R) GRAVITY MODELS IN THE MATTER DOMINATED
ERA
In this section, we consider viable F (R) gravity models representing a realistic scenario to account for dark energy,
in particular, two well-known ones proposed in Refs. [28–33] (for more examples and detailed explanations on viable
models, see, e.g., [16, 34] and references therein). Here, we mention that in Ref. [35], the gravitational waves in
viable F (R) models have been studied, and that the observational constraints on exponential gravity have also been
examined in Ref. [36]. We show that for these models, large frequency oscillation of dark energy in the matter
dominated era appears, and that it may influence on the behavior of higher derivatives of the Hubble parameter with
respect to time. Such a oscillation has the risk to produce some divergence, and therefore we suggest a way to stabilize
the frequency oscillation by performing the subsequent numerical analysis. In these models, a correction term to the
Hilbert-Einstein action is added as F (R) = R + f (R) in (2.2), so that the current acceleration of the universe can be
reproduced in a simple way. Namely, a vanishing (or fast decreasing) cosmological constant in the flat limit of R → 0
is incorporated, and a suitable, constant asymptotic behavior for large values of R is exhibited.
A. Realistic F (R) gravity models
First, we explore the Hu-Sawicki model [28] (for the related study of such a model, see Ref. [37]),
m̃2 c1 (R/m̃2 )n m̃2 c1 m̃2 c1 /c2

F (R) = R − 2 n
=R− + , (3.1)
c2 (R/m̃ ) + 1 c2 c2 (R/m̃2 )n + 1
where m̃2 is the mass scale, c1 and c2 are positive parameters, and n is a natural positive number. The model is
very carefully constructed such that in the high curvature regime, m̃2 c1 /c2 = 2Λ can play a role of the cosmological
constant Λ and thus the ΛCDM model can be reproduced.
Moreover, in Refs. [29, 30] another simple model which may easily be generalized to reproduce also inflation has
been constructed
h i
F (R) = R − 2Λ 1 − e−R/(b Λ) , (3.2)
where b > 0 is a free parameter. Also in this model, in the flat space the solution of the Minkowski space-time is
recovered, while at large curvatures the ΛCDM model is realized. This kind of models can satisfy the cosmological
and local gravity constraints. Both of these models asymptotically approach the ΛCDM model in the high curvature
regime. Indeed, however, the mechanisms work in two different manners, i.e., via a power function of R (the first
one) and via an exponential function of it (the second one). For our treatment, we reparameterize the model (3.1) by
describing c1 m̃2 /c2 = 2Λ and (c2 )1/n m̃2 = b Λ with b > 0, so that we can obtain

1
F (R) = R − 2Λ 1 − , n = 4. (3.3)
[R/ (b Λ)]n + 1
Through this procedure, in both of these models the term b Λ corresponds to the curvature for which the cosmological
constant is “switched on”. This means b ≪ 4, so that b Λ ≪ 4Λ and hence R = 4Λ can be the curvature of de Sitter
universe describing the current cosmic acceleration. In the mode in Eq. (3.3), since n has to be sufficiently large in
order to reproduce the ΛCDM model, we have assumed n = 4 and we keep only the parameter b free.
B. Dark energy oscillations in the matter dominated era
Despite the fact that the models in Eqs. (3.2) and (3.3) precisely resemble the ΛCDM model, there is a problem
that in the matter dominated era the higher derivatives of the Hubble parameter diverge and thus this can make the
solutions unphysical. This problem originates from the stability conditions to be satisfied by these models [38] and
from dark energy oscillations during the matter phase [32] in Ref. [22]. Since in matter dominated era R = 3m̃2 (z +1)3
and yH (z) ≪ (1 + z)3 and χ(1 + z)4 ≪ (z + 1)3 in order for dark energy and radiation to vanish during this phase,
one may locally solve Eq. (2.10) around z = z0 + (z − z0 ), where |z − z0 | ≪ z. The solution reads to the first order
in terms of (z − z0 ),
′′ α β
yH (z) + y ′ (z) + yH (z) = ζ0 + ζ1 (z − z0 ) , (3.4)
(z − z0 ) H (z − z0 )2
5
where
7 (1 − F ′ (R0 ))F ′′′ (R0 )
α = − − ,
2 2F ′′ (R0 )2
1 2(1 − F ′ (R0 ))F ′′′ (R0 )
β = 2+ + , (3.5)
R0 F ′′ (R0 ) F ′′ (R0 )2
with ζ0 and ζ1 being constants and R0 = 3m̃2 (z0 + 1)3 . Thus, the solution of Eq. (3.4) is derived as
1 p
yH (z) = a + b · (z − z0 ) + C0 · exp −α ± α2 − 4β (z − z0 ) , (3.6)
2(z0 + 1)
where a, b and C0 are constants. Now, for the two models in Eqs. (3.2) and (3.3), when R ≫ b Λ, we find
F ′ (R) ≃ 1 ,
F ′′ (R) ≃ 0+ . (3.7)
These behaviors guarantee the occurrence of the realistic matter dominated era. Furthermore, since in the expanding
universe (z − z0 ) < 0, it turns out that the dark energy perturbations in Eq. (3.6) remain small around R0 , and that
we acquire
(1 − F ′ (R0 )) F ′′′ (R0 ) 7 1
>− , > 12 , (3.8)
2F ′′ (R0 )2 2 R0 F ′′ (R0 )
for both these models. Owing to the fact that F ′′ (R) is very close to 0+ , the discriminant in the square root of
Eq. (3.6) is negative and dark energy oscillates as
" p ! p !#
Λ α
− 1,2
(z−z0 ) β 1,2 β 1,2
yH (z) = + e 2(z0 +1) A sin (z − z0 ) + B cos (z − z0 ) . (3.9)
3m̃2 (z0 + 1) (z0 + 1)
Here, A and B are constants and α1,2 and β1,2 are given by Eq. (3.5), so they correspond to two models under
investigation. In particular, α1 = −3 for the model in Eq. (3.2) and α2 ≃ −29/10 for the model in Eq. (3.3), while
β1,2 ≃ 1/(R0 F ′′ (R0 )), i.e.,
R0 !
b2 Λe R̃
β1 ≃ , (3.10)
2R0
in case of exponential model in Eq. (3.2) and

h i3
R0 n

bΛ
n
R0 1 + bΛ R0 R0

R0
n
β2 ≃ n o ≃ , (3.11)
R0 n
n h i
2Λ n 1 + n

−1 + R0 2Λ n(n + 1) bΛ
bΛ bΛ
in case of model in Eq. (3.3). This means that the frequency of dark energy oscillations increases as the curvature
(and redshift) becomes large. Moreover, the effects of such oscillations are amplified in the derivatives of the dark
energy density, namely,
n
d n
y (t )
dtn H 0 ∝ (F (z0 )) , (3.12)

−1/2
where F (z) ≃ (R ∗ F ′′ (R)) /(z + 1) is the oscillation frequency and t0 is the cosmic time corresponding to the
redshift z0 . This is for example the case of the EoS parameter for dark energy defined as1
PDE 1 1 dyH (z)
ωDE (z) ≡ = −1 + (z + 1) . (3.13)
ρDE 3 yH (z) d(z)
1 Throughout this paper, we describe the EoS parameter by “ω” and not “w”.
6
For large values of the redshift, the dark energy density oscillates with a high frequency and also its derivatives become
large, showing a different feature of the dark energy EoS parameter in the models in Eqs. (3.2) and (3.3) compared
with the case of the cosmological constant in GR. During the matter dominated era, the Hubble parameter behaves
as
√ yH (z)

H(z) ≃ m̃2 (z + 1)3/2 + . (3.14)
2(z + 1)3/2
If the frequency F (z0 ) in Eq. (3.12) is extremely large, the derivatives of dark energy density could become dominant
in some higher derivatives of the Hubble parameter which may approach an effective singularity and therefore make
the solution unphysical. We see it for specific cases. In Refs. [16, 39, 40], the cosmological evolutions in exponential
gravity and the Hu-Sawicki model have carefully been explored. It has explicitly been demonstrated that the late-
time cosmic acceleration which follows the matter dominated era can occur, according with astrophysical data. A
reasonable choice is to take b = 1 for both these models. We also put Λ = 7.93m̃2 [5]. We can solve Eq. (2.10)
numerically2 by taking the initial conditions at z = zi , where zi ≫ 0 is the redshift at the initial time to execute the
numerical calculation, as follows:
dyH (z)
= 0,
d(z) zi
Λ
yH (z) = .

zi 3m̃2
Here, we have used the fact that at a high redshift the universe should be very close to the ΛCDM model. We have
set zi = 2.80 for the model in Eq. (3.2) and zi = 4.5 for the model in Eq. (3.3), such that R F ′′ (R) ∼ 10−8 at
R = 3m̃2 (zi + 1)3 . We note that it is hard to extrapolate the numerical results to the higher redshifts because of the
large frequency of dark energy oscillations.
Using Eq. (3.13) with yH , we derive ωDE . In addition, by using Eq. (2.14) we obtain R as a function of the redshift.
We can also execute the extrapolation in terms of the behavior of ΩDE , given by
ρDE yH
ΩDE (z) ≡ = 3 4 . (3.15)
ρeff yH + (z + 1) + χ (z + 1)
The numerical extrapolation to the present universe leads to the following results: For the model (3.2), yH (0) = 2.736,
ωDE (0) = −0.950, ΩDE (0) = 0.732 and R(z = 0) = 4.365, whereas for the model (3.3), yH (0) = 2.652, ωDE (0) =
−0.989, ΩDE (0) = 0.726 and R(z = 0) = 4.358. These resultant data are in accordance with the last and very accurate
observations of our current universe [5], which are
ωDE = −0.972+0.061
−0.060 ,
ΩDE = 0.721 ± 0.015 . (3.16)
Next, we introduce the deceleration q, jerk j and snap s parameters [41, 42]
1 d2 a(t) 1 Ḣ
q(t) ≡ − 2 2
= − 2 − H2
a(t) dt H(t) H
1 d3 a(t) 1 Ḧ
j(t) ≡ 3 3
= 3 − 3q − 2
a(t) dt H(t) H
4
...
1 d a(t) 1 H
s(t) ≡ = + 4j + 3q(q + 4) + 6 . (3.17)
a(t) dt4 H(t)4 H4
In what follows, we show the values of these cosmological parameters at the present time (z = 0) as the result of
numerical extrapolation in our two models, which we called Model I in Eq. (3.2) and Model II in Eq. (3.3), and the
calculation in the ΛCDM model:
q(z = 0) = −0.650 (ΛCDM) , −0.544 (Model I) , −0.577 (Model II)
j(z = 0) = 1.000 (ΛCDM) , 0.792(Model I) , 0.972 (Model II)
s(z = 0) = −0.050 (ΛCDM) , −0.171(Model I) , −0.152 (Model II) .
2 c
We have used Mathematica 7 .
7
The deviations of the parameters in Models I and II from those in the ΛCDM model are small at the present.
However, since these parameters depend on the time derivatives of the Hubble parameter, it is interesting to analyze
those behaviors at high curvature. Therefore, in Fig. 1 we plot the cosmological evolutions of q, j and s as functions
of the redshift z. From this figure, we see that there exist overlapped regions for Models I and II with those in the
ΛCDM model.
j
q 2.0
s
1000
0.4
1.5
0.2 500
z 1.0
0.5 1.0 1.5 2.0 2.5
z
0.5 1.0 1.5 2.0 2.5
-0.2
0.5
-500
-0.4
z
-0.6 0.0 0.5 1.0 1.5 2.0 2.5 -1000
(a) (b) (c)

q j
2.5 s
0.4
4000
2.0
0.2
1.5 2000
z
0.5 1.0 1.5 2.0 2.5
1.0
z
1 2 3 4
-0.2
0.5
-2000
-0.4 z
1 2 3 4
-4000
-0.6 -0.5
(d) (e) (f)
FIG. 1: Cosmological evolutions of q(z) [(a) and (d)], j(z) [(b) and (e)] and s(z) [(c) and (f)] parameters as functions of the
redshift z for Model I [(a)–(c)] and Model II [(d)–(f)] in the region of z > 0.
The deceleration parameter in Models I and II remains very close to the value in the ΛCDM model, because in the
first time derivative of the Hubble parameter the contribution of dark energy is still negligible. Hence, it guarantees
the correct cosmological evolution of these models. However, it is clearly seen that in the jerk and snap parameters the
derivatives of the dark energy density become relevant and the parameters grow up with an oscillatory behavior. Since
the frequency of such oscillations strongly increases in the redshift, it is reasonable to expect that some divergence
occurs in the past. We also remark that if from one side at high redshifts the exponential Model I is more similar to
the ΛCDM model because of the faster decreasing of exponential function in comparison with the power function of
Model II, from the other side it involves stronger oscillations in the matter dominated era.
It may be stated that the closer the model is to the ΛCDM model (i.e., as much F ′′ (R) is close to zero), the bigger
the oscillation frequency of dark energy becomes. As a consequence, despite the fact that the dynamics of the universe
depends on the matter and the dark energy density remains very small, some divergences in the derivatives of the
Hubble parameter can occur. In the models in Eqs. (3.2) and (3.3), although the approaching manners to a model
with the cosmological constant are different from each other, it may be interpreted that these models show a generic
feature of realistic F (R) gravity models, in which the cosmological evolutions are similar to those in a model with the
cosmological constant. The corrections to the Einstein’s equations in the small curvature regime lead to undesired
effects in the high curvature regime. Thus, we need to investigate additional modifications.
C. Proposal of a correction term
In order to remove the divergences in the derivatives of the Hubble parameter, we introduce a function
√ g(R) for
which the oscillation frequency of the dark energy density in Eq. (3.9) acquires a constant value 1/ δ, where δ > 0,
for a generic curvature R ≫ bΛ, and we stabilize the oscillations of dark energy during the matter dominated era with
8
1/3
the use of a correction term. Since in the matter dominated era, i.e., z + 1 = R/(3m̃2 )

, we have to require
(3m̃2 )2/3 1
=
R5/3 g ′′ (R) δ
1/3
R
g(R) = −γ̃ Λ , γ̃ > 0 , (3.18)
3m̃2
where γ̃ ≡ (9/2)δ(3m̃2 /Λ) = 1.702 δ. We explore the models in Eqs. (3.2) and (3.3) with adding these correction as
1/3
− bRΛ R
F1 (R) = R − 2Λ(1 − e ) − γ̃ Λ , (3.19)
3m̃2
1/3
1 R
F2 (R) = R − 2Λ 1 − − γ̃ Λ . (3.20)
(R/b Λ)4 + 1 3m̃2
We note that in both cases F1,2 (0) = 0 and therefore we still have the solution of the flat space in the Minkowski
space-time. The effects of the last term vanish in the de Sitter epoch, when R = 4L and these models resemble to a
model with an effective cosmological constant, provided that γ̃ ≪ (m̃2 /Λ)1/3 . We may also evaluate the dark energy
density at high redshifts by deriving ρDE = ρeff − ρm from Eq. (2.7) and by putting R = 3m̃2 (z + 1)3 such that
Λ
yH (z) ≃ [1 + γ̃(1 + z)] . (3.21)
3m̃2
According to the observational data of our universe, the current value of dark energy amount is estimated as yH ≡
Λ/(3m̃) = 2.643. With the reasonable choice γ̃ ∼ 1/1000, the effects of modification of gravity on the dark energy
density begin to appear at a very high redshift (for example, at z = 9, yH (9) = 1.01 × yH (0)), and hence the universe
seems to be very close to the ΛCDM model. However, while the pure models in Eqs. (3.2) and (3.3) mimic an effective
cosmological constant, the models in Eqs. (3.19) and (3.20) mimic (for the matter solution) a quintessence fluid.
Equation (3.13) leads to
(1 + z)γ̃
ωDE (z) ≃ −1 + , (3.22)
3(1 + (1 + z)γ̃)
so that when z ≫ γ̃ −1 , ωDE (z) ≃ −2/3.
′
Thus, it is simple to verify that all the cosmological constraints [12] are still satisfied. Since |F1,2 (R ≫ bΛ) − 1| ≪ 1,
′
the effective gravitational coupling Geff = G/ F1,2 (R) is positive, and hence the models are protected against the anti-
gravity during the cosmological evolution until the de Sitter solution (RdS = 4Λ) of the current universe is realized.
′′
Thus, thanks to the fact that |F1,2 (R ≫ bΛ) > 0|, we do not have any problem in terms of the existence of a stable
matter. In Sec. IV, we also analyze the local constraints in detail, and we see that our modifications do not destroy
the feasibility of the models in the solar system. It should be stressed that the energy density preserves its oscillation
behavior in the matter dominated p era, but that owing to the correction term reconstructed here, such oscillations
keep a constant frequency F = 1.702/γ̃ and do not diverge. Despite the small value of γ̃, in this way the high
redshift divergences and possible effective singularities are removed.
From the point of view of the end of inflation, there is another resolution of this problem. It is well known that
the scalar begins to oscillate once the mass m becomes larger than the Hubble parameter, H < m. Indeed, for a
canonical scalar, the energy density sloshes between the potential energy (w = −1, where w is the equation of state
of the canonical scalar) and the kinetic energy (w = +1). What is done usually is that the oscillations enough rapidly
(i.e., those with m ≫ H) can be averaged over giving an effective energy-momentum tensor with w = 0, i.e., dust.
The same procedure should be performed here, once the oscillations are rapid enough. In this interpretation, there
would be no problem with any strange rapidly oscillating contributions to the energy momentum tensor. A solution
is to choose the potential effectively so that the mass can not increase as the matter energy density increases.
Furthermore, it is significant to remark that in a number of models of F (R) gravity for dark energy, there exists
a well-known problem that positions in the field space are a finite distance away from the minimum of the effective
potential, so that a curvature singularity in the Jordan frame could appear. This means that large excursions of the
scalar could result in a singularity forming in a solution. It is known that the solution for this problem is also adding
the higher powers of R so that the behavior at large curvatures can be soften. The oscillations are extremely large
at small curvatures too, and the higher power of R or R itself do not change in this range of detection. We also note
that this argument is applicable to the so-called type I, II and III finite-time future singularities (where R diverges),
which has been classified in Ref. [45], while for a kind of singularities in our work, R does not become singular, and
hence the argument would become different from the above.
9
D. Analysis of exponential and power-form models with correction terms in the matter dominated era
In this subsection, we carry out the numerical analysis of the models in Eqs. (3.19) and (3.20). In both cases, we
assume b = 1 and γ̃ = 1/1000 and solve Eq. (2.10) in a numerical way, by taking accurate initial conditions at z = zi
so that zi ≫ 2. By using Eq. (3.21), we acquire
dyH (z) Λ
= γ,
d(z) zi 3m̃2
Λ
yH (z) = (1 + γ (zi + 1)) ,

zi 3 m̃ 2
where we have set zi = 9. The feature of the models in Eqs. (3.19) and (3.20) at the present time is very similar to
those of the models in Eqs. (3.2) and (3.3). With the numerical extrapolation to the current universe, for the model
in Eq. (3.19) we have yH (0) = 2.739, ωDE (0) = −0.950, ΩDE (0) = 0.732 and R(z = 0) = 4.369, while for the model
in Eq. (3.20), we find yH (0) = 2.654, ωDE (0) = −0.989, ΩDE (0) = 0.726 and R(z = 0) = 4.361. We analyze those
behaviors in the matter dominated era. It follows from the initial conditions yH (9) = 2.670 and ωDE (9) = −0.997
that the universe is extremely close to the ΛCDM model also at high redshifts. We see how the dynamical correction
of the Einstein’s equation, which corresponds to, roughly speaking, the fact of having “a dynamical cosmological
constant”, introduces an oscillatory behavior of dark energy density. Thanks to the contribution of the correction
term, we obtain a constant frequency of such oscillations without changing the cosmological evolution described by
the theory. In Fig. 2, we show the cosmological evolutions of the deceleration, jerk and snap parameters as functions of
the redshift z in these models. There is overlapped region of the evolutions with those in the ΛCDM model. We may
compare the graphics in Fig. 2 with the corresponding ones in Fig. 1 of the models in Eqs. (3.2) and (3.3) without the
correction term analyzed in Sec. III B. At high redshifts, the deceleration parameter is not influenced by dark energy
and hence the behavior in both these models in Eqs. (3.19) and (3.20) are the same as that in the ΛCDM model. On
the other hand, in terms of the jerk and snap parameters, the derivatives of dark energy density become relevant and
accordingly these parameters oscillate with the same frequency as that of dark energy, showing a different behavior
in comparison with the case of GR with the cosmological constant. However, here such oscillations
p have a constant
frequency and do not diverge. The predicted value of the oscillation frequency is F ≡ 1.702/γ̃ = 41.255. The
oscillation period is T = 2π/F ≃ 0.152. Thus, the numerical data are in good accordance with the predicted ones.
(We can also appreciate the result by taking into account the fact that the number of crests per units of the redshift
has to be 1/T ≃ 7).
Consequently, we have shown in both analytical and numerical ways that increasing oscillations of dark energy in
the past approach to effective singularities. It is not “a rapid oscillating system” but a system which becomes singular.
The effects of such oscillations are evident especially in the higher derivative of the Hubble parameter. It is not a
case that if all the numerical simulations presented in the literature start from small redshifts, at higher redshifts
this singular problem appears. Eventually, the oscillations may influence also on the behavior of the Ricci scalar
n
(which depends on the first derivative of the Hubble parameter, see Eq. (3.14) and |dn H(t)|t0 /dtn | ∝ (F (z0 )) with
n ≫ 1, following from Eq. (3.12)). Of course, the average value of the dark energy density remains negligible, but the
oscillations around this value become huge. Thus, the Ricci scalar may have an oscillatory behavior. We have also
evaluated the frequency of the oscillations, so that the result can match with the numerical simulations, and therefore
all the analyses in this work are consistent. This behavior of realistic F (R) gravity models has recently been studied
also in Ref. [43].
We remark that if the mass of the additional scalar degree of freedom, the so-called scalaron, is too large, the
predictability could be lost [44]. Clearly, the mass of the scalaron in the two models in Eqs. (3.19) and (3.20) is not
bounded, and thus it would diverge in very dense environment. We have confirmed that in the large curvature regime
compared with the current curvature the correction term g(R) in (3.18) in these two models do not strongly affect the
scalaron potential in the Einstein frame, namely, the correction term would not be the leading term in the form of
the scalaron potential, and thus the scalaron mass is not changed very much. The model parameter of the correction
term g(R) mainly related to the scalaron mass as well as its potential is γ̃. In the limit that the energy density of
the environment becomes infinity, since the contribution of the correction term to the scalaron mass, it would be
impossible to constrain the values of γ̃, for which the divergence of the scalaron mass can be avoided.
IV. COSMOLOGICAL CONSTRAINTS AND FUTURE EVOLUTION
In this section, first we show that the models in Eqs. (3.19) and (3.20) satisfy the cosmological and local gravity
constraints [24], and that the term added to stabilize the dark energy oscillations in the matter dominated epoch
10
q j s
100
0.4
1.4
0.2
50
1.2
z
2 4 6 8
1.0 z
-0.2 1 2 3 4 5 6
-0.4
0.8
-50
-0.6
z
1 2 3 4 5 6
-0.8 -100
(a) (b) (c)

q j s
0.6 100
1.4
0.4
50
0.2
1.2
z
2 4 6 8
1.0 z
1 2 3 4 5 6
-0.2
-0.4 0.8
-50
-0.6
z
1 2 3 4 5 6
-0.8 -100
(d) (e) (f)
FIG. 2: Cosmological evolutions of q(z) [(a) and (d)], j(z) [(b) and (e)] and s(z) [(c) and (f)] parameters as functions of the
redshift z for the model F1 (R) [(a)–(c)] and the model F2 (R) [(d)–(f)] in the region of z > 0.
does not cause any problem to these proprieties. The confrontation of F (R) models with SNIa, BAO, CMB radiation
and gravitational lensing has been executed in the past several works [46]. We have just seen that the models with
the choice of b = 1 can be consistent with the observational data of the universe. Here, we examine the range of
b in which the models are compatible with the observations and analyze the behavior of the models near to local
(matter) sources in order to check possible Newton law corrections or matter instabilities. Then, we concentrate on
the future evolution of the universe in the models and demonstrate that the effective crossing of the phantom divide
which characterizes the de Sitter epoch takes place in the very far future.
In the way of trying to explain the several aspects that characterize our universe, there exists the problem of
distinguishing different theories. It has been revealed that sometimes the study of the expansion history of the
universe is not enough because different theories can achieve the same expansion history. Fortunately, theories with
the same expansion history can have a different cosmic growth history. This fact makes the growth of the large scale
structure in the universe an important tool in order to discriminate among the different theories proposed. Thus, the
characterization of growth of the matter density perturbations become very significant. In order to execute it, the
so-called growth index γ [47] is useful. Therefore, in the second part of this section we study the evolution of the
matter density perturbation for our F (R) gravity model.
Again, we clearly state the main purpose of this section. Since the original models, i.e., the Hu-Sawicki model [28]
in Eq. (3.3) and exponential gravity [29, 30] in Eq. (3.2), have been studied well, we concentrate on the question
whether the corrected models in Eqs. (3.20) and (3.19) lead to any difference in the observables. These modified
models have been constructed in order not to alter the background evolution significantly except the oscillatory effect.
In Refs. [28, 30] and many follow-up studies of these pioneering works, the cosmological background evolutions and
the growth of structures in the two unmodified models in Eqs. (3.3) and (3.2) have been investigated. In order make
this work self consistent study of modified gravity, we explicitly demonstrate the cosmological background evolutions
and the growth of the matter density perturbations in the modified models in Eqs. (3.20) and (3.19). It is meaningful
to investigate these behaviors in the modified models even though the modifications on the observable quantities are
small.
A. Cosmological and local constraints
We take γ̃ = 1/1000 in the models in Eqs. (3.19) and (3.20), keeping the parameter b free. Now, the dark energy
density is a function of z and b, i.e., yH (z, b). We can again solve Eq. (2.10) numerically, taking the initial conditions
11
at zi = 9 as
dyH (z, b) Λ
= γ̃ ,
d(z) zi 3m̃2
Λ
yH (z, b) = (1 + γ̃ (zi + 1)) ,

zi 3m̃2
as we did in the previous section. We take 0.1 < b < 2. In Figs. 3 and 4, we display the resultant values of dark
energy EoS parameter ωDE (z = 0, b) and ΩDE (z = 0, b) at the present time as functions of b for the two models. We
also show the bounds of cosmological data in Eq. (3.16), namely, the lines in rose denote the upper bounds, while the
lines in yellow do the lower ones. By matching the comparison between the two graphics of every model, we find that
in order to correctly reproduce the universe where we live with exponential gravity in Eq. (3.19), 0.1 < b < 1.174,
with power-law model in Eq. (3.20), 0.1 < b < 1.699. The results are consistent with the choices in Sec. III D.
W_DEH0L Ω_DEH0L
-0.85
0.74
-0.90
0.73
-0.95
0.72
b
0.5 1.0 1.5 2.0
b
0.5 1.0 1.5 2.0
0.71
(a) (b)
FIG. 3: Behaviors of ωDE (z = 0, b) and of ΩDE (z = 0, b) as functions of b for exponential model. The observational data bounds
(horizontal lines) are also shown.
Ω_DEH0L W_DEH0L
0.745
-0.92
0.740
-0.94 0.735
0.730
-0.96
0.725
-0.98
0.720
-1.00
0.715
b b
1.0 1.5 2.0 1.0 1.5 2.0
(a) (b)
FIG. 4: Behaviors of ωDE (z = 0, b) and of ΩDE (z = 0, b) as functions of b for power-law model. Legend is the same as Fig. 3.
Newton law corrections and stability on a planet surface
In Ref. [48], it has been shown that some realistic models of F (R) gravity may lead to significant Newton law
corrections at large cosmological scales. We briefly review this result. From the trace of the field equation (2.3), if we
consider the constant background of R = R0 , such that 2F (R0 )−R0 F ′ (R0 ) = 0, by performing a variation with respect
to R = R(0) + δR and supposing the presence of a matter point source (like a planet), that is, T (matter) = T0 δ(x),
where δ(x) is the Dirac’s distribution, we find, to first order in δR,
κ2
− m2 δR =

T0 δ(x) , (4.1)
3F ′′ (R0 )
12
with
F ′ (R0 )

1
m2 = − R0 . (4.2)
3 F ′′ (R0 )
The solution is given by
κ2
δR = T0 G(m2 , |x|) , (4.3)
3F ′′ (R0 )
where G(m2 , |x|) is the correlation function which satisfies
− m2 G(m2 , |x|) = δ(x) .

(4.4)
Hence, if m2 < 0, there appears a tachyon and thus there could be some instability. Even if m2 > 0, when m2 is
small compared with R0 , δR 6= 0 at long ranges, which generates the large correction to the Newton law. For the
pure exponential model in Eq. (3.2) without correction terms, when R0 ≫ b Λ, m2 reads
(b2 Λ) R0
m2 ≃ e bΛ . (4.5)
6
Therefore, in general m2 /R0 is very large effectively. The same thing happens in the model in Eq. (3.3). Next, for
the models in Eqs. (3.19) and (3.20) with correction terms, we have
2/3
34/3 m̃2 R

R
m2 ≃ . (4.6)
2Λγ̃ m̃2
Despite the fact that in this case m2 is smaller than in Eq. (4.5), it still remains sufficiently large and the correction
to the Newton law is very small. For example, the typical value of the curvature in the solar system is R0 ≃ 10−61 eV2
(it corresponds to one hydrogen atom per cubic centimeter). In this case, from Eq. (4.6) we obtain m2 /R0 ≃ 2 × 106 .
Concerning the matter instability [18, 19], this might also occur when the curvature is rather large, as on a planet
(R ≃ 10−38 eV2 ), as compared with the average curvature of the universe today (R ≃ 10−66 eV2 ). In order to arrive at
a stability condition, we can perturb again Eq. (??) around R = Rb , where Rb is the curvature of the planet surface
and the perturbation δR is given by the curvature difference between the internal and the external solution. The
curvature Rb = −κ2 T (matter) depends on the radial coordinate r. By assuming δR depending on time only, we acquire
− ∂t2 (δR) ∼ U (Rb )δR , (4.7)
where
" 2 #
F ′′′ (Rb ) F ′′′ (Rb ) rr Rb F ′ (Rb )
U (Rb ) = − ′′ g ∇r Rb ∇r Rb − +
F ′′ (Rb ) F (Rb ) 3 3F ′′ (Rb )
F ′′′ (Rb )
(2F (Rb ) − Rb F ′ (Rb ) − Rb ) . (4.8)
3(F ′′ (Rb ))2
Here, gµν is the diagonal metric describing the planet. If U (Rb ) is negative, then the perturbation δR becomes
exponentially large and the whole system becomes unstable. Thus, the planet stability condition is
U (Rb ) > 0 . (4.9)
For our models in Eqs. (3.19) and (3.20), U (Rb ) ≃ m2 , where m2 is given by Eq. (4.6) again. Also in this case, we do
not have any particular problem. For example, by putting Rb ≃ 10−38 eV2 , we find U (Rb )/Rb ≃ 4 × 1021 . Thus, the
models under consideration easily pass these local tests.
We mention that in the past, the non-linear effects on the scalar are much more important, owing to the mechanism
of the chameleon effect [49, 50], and that only at late times the linear evolution is a good approximation. For
example, if a high-curvature solution is achieved, the Solar-System test is the examination whether the solution is
stable against the Dolgov-Kawasaki instability [18]. This is not the same as whether the high-curvature solution can
at all be achieved, which is a much more subtle issue and discussed at length by Hu and Sawicki in Ref. [28].
13
B. Future universe evolution
In de Sitter universe, we have R = RdS , where RdS is the constant curvature given by the constant dark energy
density yH = y0 , such that y0 = RdS /12m̃2 . Starting from Eq. (2.10), we are able to study perturbations around the
de Sitter solution in the models (3.19) and (3.20) which provide this solution for RdS = 4Λ and well satisfied the de
Sitter condition 2F (RdS ) = RdS F ′ (RdS ) as a consequence of the trace of the field equation in vacuum. Performing
the variation with respect to yH (z) = y0 + y1 (z) with |y1 (z)| ≪ 1 and assuming the contributions of radiation and
matter to be much smaller than y0 , at the first order in y1 (z) Eq. (2.10) reads
d2 y1 (z) α dy1 (z) β
+ + y1 (z) = 4ζ(z + 1) , (4.10)
dz 2 (z + 1) dz (z + 1)2
where
4F ′ (RdS ) 1 − F ′ (RdS )
α = −2 , β = −4 + , ζ =1+ . (4.11)
RF ′′ (RdS ) RdS F ′′ (RdS )
The solution of Eq. (4.10) is given by
yH (z) = y0 + y1 (z) , (4.12)
√
4ζ

1
1−α± (1−α)2 −4β
y1 (z) = C0 (z + 1) 2
+ (z + 1)3 , (4.13)
β
where C0 is a constant. The well-known stability condition for the de Sitter space-time, F ′ (RdS )/((RdS )F ′′ (RdS )) > 1,
is also valid. It has also been demonstrated that since in realistic F (R) gravity models for the de Sitter universe
F ′′ (R) → 0+ , F ′ (RdS )/(RdS F ′′ (RdS )) > 25/16 [51] giving negative the discriminant of Eq. (4.13) and an oscillatory
behavior to the dark energy density during this phase. Thus, in this case the dark energy EoS parameter ωDE (3.13)
becomes
3
2 (z + 1) 2
ωDE (R = RdS ) ≃ −1 + 4m̃ × (4.14)
RdS
" s ! s !#
4 4
A0 cos log(z + 1) + B0 sin log(z + 1) ,
RdS F ′′ (RdS ) RdS F ′′ (RdS )
and oscillates infinitely often around the line of the phantom divide ωDE = −1 [51]. According to various recent
observational data, the crossing of the phantom divide occurred in the near past [52]. These models possess one
crossing in the recent past [16], after the end of the matter dominated era, and infinite crossings in the future (for
detailed investigations on the future crossing of the phantom divide, see [34]), but the amplitude of such crossings
decreases as (z + 1)3/2 and it does not cause any serious problem to the accuracy of the cosmological evolution during
the de Sitter epoch which is in general the final attractor of the system [16, 39]. However, the existence of a phantom
phase can give some undesirable effects such as the possibility to have the Big Rip [53] as an alternative scenario of
the universe (in such a case, the model may suddenly exit from ΛCDM description) or the disintegration of bound
structures which does not necessarily require to having the final (Big Rip) singularity [54, 55]. In this subsection,
we show that in the models in Eqs. (3.19) and (3.20) the effective EoS parameter of the universe (for an alternative
study, see [56]) defined as
ρeff 2(z + 1) dH(z)
ωeff ≡ = −1 + (4.15)
Peff 3H(z) dz
never crosses the phantom divide line in the past, and that only when z is very close to −1 (this means in the very far
future), it coincides with ωDE and the crossings occur. We remark that ρeff and Peff correspond to the total energy
density and pressure of the universe, and hence that if dark energy strongly dominates over ordinary matter, we can
consider ωeff ≈ ωDE . In both of the models under investigation, we take again γ̃ = 1/1000 and keep the parameter
b free, such that 0.1 < b < 1.174 (model in Eq. (3.19)) and 0.1 < b < 1.699 (model in Eq. (3.20)), according to the
realistic representation of current universe. The numerical evaluation of Eq. (2.10) leads to H(z), given by
p
H(z) = m̃2 [yH (z) + (z + 1)3 + χ(z + 1)4 ] , (4.16)
and therefore ωeff (z). We depict the cosmological evolution of ωeff as a function of the red shift z and the b parameter
in Fig. 5 for the model in Eq. (3.19) and in Fig. 6 for the model in Eq. (3.20). On the left panels, we plot the effective
14
EoS parameter for −1 < z < 2. We can see that for both of the models, independently on the choice of b, ωDE starts
from zero in the matter dominated era and asymptotically approaches -1 without any appreciable deviation. Only
when z is very close to −1 and the matter contribution to ωeff is effectively zero, we have the crossing of the phantom
divide due to the oscillation behavior of dark energy. On the right panels, we display the behavior of the effective EoS
parameter around z = −1. Here, we focused on the phantom divide line and we excluded the graphic area out of the
range −1.0001 < ωeff < −0.9999. The blue region indicates that ωeff is still in the quintessence phase. We note that
especially in the model in Eq. (3.20), the first crossing of phantom divide is very far in the future. For example, with
the scale factor a(t) = exp (H0 t), where H0 ≃ 6.3 × 10−34 eV−1 is the Hubble parameter of the de Sitter universe,
z = −0.90 (when the crossing of the phantom divide may begin to appear in the exponential models) corresponds to
1026 years.
(a) (b)
FIG. 5: Cosmological evolution of ωeff as a function of the red shift z and the b parameter for the model in Eq. (3.19). The
left panel plots it for −1 < z < 2 and the right one displays around z = −1.
(a) (b)
FIG. 6: Cosmological evolution of ωeff as a function of the red shift z and the b parameter for the model in Eq. (3.20). Legend
is the same as Fig. 5.
Avoidance of the phantom crossing with (inhomogeneous) fluid
It may be of some interest to check if it is possible to avoid the crossing of the phantom divide by adding a suitable
(compensating) fluid in the future cosmological scenario described by the models (3.19) and (3.20). Here, we indicate
a possible realization of it. We examine an inhomogeneous fluid with its energy density ρ, pressure P and the Eos
15
parameter ω as a function of ρ, i.e., ω = ω(ρ). The EoS is expressed as

d 3
log ρ = (ω(ρ) + 1) . (4.17)
dz (z + 1)
We explore the simple case
ω(ρ) = A0 σ(z)ρα−1 − 1 , (4.18)
where α is a constant and A0 is a positive parameter. Moreover, σ(z) = −1 when z ≥ 0 and σ(z) = 1 when z < 0,
such that the fluid is in the phantom region for z ≥ 0 and in the quintessence region for z < 0. The fluid energy
density reads
1
ρ = ρ0 (B0 − σ(z) log(z + 1)) (1−α) , (4.19)
where ρ0 = [3(α − 1)A0 ]1/(1−α) and B0 are positive parameters depending on the initial conditions.
We note that one can choose B0 = 1 without the loss of generality and in this way the energy density is defined as
a positive quantity. If we take α > 1, when z → +∞ or z → −1+ , the energy density asymptotically tends to zero.
For z = 0, we have a maximum, ρ(z = 0) = ρ0 , so that we should require ρ0 ≪ Λ/κ2 , namely, the fluid energy density
is always small with respect to the dark energy density given by our models for the cosmological constant. A fluid in
the form of Eq. (4.19) may asymptotically produce a (Big Rip) singularity H(t) ∼ (t0 − t)β , where t < t0 and β > 1
(for general study of singularities in modified gravity, see [57]), only for β = 1/(2α − 1) [58], but in our case α > 1, so
that this kind of divergence can never appear. If we add this fluid in the scenario described by F (R) gravity models
in Eqs. (3.19) and (3.20), when z → −1 we find
PDE + P A0 ρα
ωeff = ≃ −1 + . (4.20)
ρDE + ρ Λ/κ2
This means that owing to the presence of fluid, the oscillations of the effective EoS parameter realize not around the
phantom divide but around ωeff given by the last equation, namely in the quintessence region. With an accurate
fitting of the parameters, in this way we may avoid the crossing of the phantom divide.
We can also add a fluid to the cosmological scenario in order to have an asymptotical phantom phase without the
Big Rip singularity. To this purpose, we investigate the EoS parameter of the fluid as in Eq. (4.18) with A0 > 0
and σ(z) = −1, which describes a phantom fluid. The fluid energy density is given by Eq. (4.19). We put B0 = 0
and α < 1 such that 1/(1 − α) can be an even number and one can have the energy density defined as a positive
quantity. In this way, the fluid energy density decreases until z = 0 and then it starts to grow up. We can take ρ0
sufficiently small so that the fluid contribution can become dominant only in the asymptotical limit, when z is close
to −1, avoiding the quintessence region in the final cosmological evolution of our F (R) gravity models. From the
equation of motion 3H 2 /κ2 = ρ, we obtain
Z z(t) s
3 dz ′
t=− . (4.21)
0 κ2 ρ(z ′ ) (z ′ + 1)
In our case, it is easy to verify that t ∼ | log(z + 1)(2α−1)/(2α−2) | and if α ≤ 1/2, when z(t) → −1 the integral
diverges and t → +∞, avoiding the Big Rip at a finite time. In this kind of models, the fluid energy density increases
with time, but ω → −1 asymptotically, so that there can be no future singularity. However, in Ref. [54] a careful
investigation on the conditions necessary to produce this evolution has been done, and it has been demonstrated that
this fluid can rapidly expand in the future, leading to the disintegration of all bound structures (this is the so-called
“Little Rip”). For example, a planet in an orbit of radius R̄ around a star of mass M will become unbound when
−(4π/3)(ρ + 3P )R̄3 ≃ M . In our case, −(ρ + 3P ) = A0 ρα and in the future every gravitationally bound system will
be disintegrated [53].
C. Growth of the matter density perturbations: growth index
In this subsection, we study the matter density perturbations. The equation that governs the evolution of the
matter density perturbations for F (R) gravity has been derived in the literature (see, for example, [59] and references
therein). Under the subhorizon approximation (for the case without such an approximation, see [60]), the matter
density perturbation δ = δρ
ρm satisfies the following equation:
m
δ̈ + 2H δ̇ − 4πGeff (a, k)ρm δ = 0 (4.22)

16
with k being the comoving wavenumber and Geff (a, k) being the effective gravitational “constant” given by
" #
k 2 /a2 (F ′′ (R)/F ′ (R))

G
Geff (a, k) = ′ 1+ . (4.23)
F (R) 1 + 3 (k 2 /a2 ) (F ′′ (R)/F ′ (R))
It is worth noting that the appearance of the comoving wavenumber k in the effective gravitational constant makes
the evolution of the matter density perturbations dependent on the comoving wavenumber k. It can be checked easily,
by taking F (R) = R in Eq. (4.23), that the evolution of the matter density perturbation does not have this kind of
dependence in the case of GR. In Fig. 7, we show the cosmological evolution as a function of the redshift z and the
scale dependence on the comoving wavenumber k of this effective gravitational constant for the case of model F1 (R)
in Eq. (3.19), while in Fig. 8 we depict those for the case of model F2 (R) in Eq. (3.20). In both these cases, we have
fixed b = 1 and used γ = 1/1000.
Geff
k HMpc-1L 1.0 G
0.5 1.35
0.0 1.30
1.4
1.3 1.25
Geff 1.2 1.20
G 1.1
1.15
1.0
0 1.10
2
4 1.05
z 6
8 z
2 4 6 8
(a) (b)
FIG. 7: (a) Cosmological evolution as a function of z and the scale dependence on k of the effective gravitational constant Geff
for the model F1 (R) with b = 1 and γ̃ = 1/1000. (b) Cosmological evolution of Geff as a function of z in the model F1 (R) with
b = 1 and γ̃ = 1/1000 for k = 1Mpc−1 (blue), k = 0.1Mpc−1 (green), k = 0.01Mpc−1 (red) and k = 0.001Mpc−1 (fuchsia).
Geff
k HMpc-1L 1.0 G
0.5
1.30
0.0
1.4 1.25
1.3
Geff1.2 1.20
G 1.1 1.15
1.0
0 1.10
2
4 1.05
z 6
8 z
2 4 6 8
(a) (b)
FIG. 8: (a) Cosmological evolution as a function of z and the scale dependence on k of the effective gravitational constant Geff
for the model F2 (R) with b = 1 and γ̃ = 1/1000. (b) Cosmological evolution of Geff as a function of z for the model F2 (R)
with b = 1 and γ̃ = 1/1000. Legend is the same as Fig. 7.
Another important remark is to state that in deriving Eq. (4.22), we have assumed the subhorizon approximation
(see [61]). Namely, comoving wavelengths λ ≡ a/k are considered to be much shorter than the Hubble radius H −1 as
k2
≫ H2 . (4.24)
a2
17
This means that we examine the scales of log k ≥ −3. On the other hand, as it was pointed out in Ref. [62], for large
k we have to take into account deviations from the linear regime. Hence, we do not consider the scales of log k > −1
and take the results obtained for log k close to −1.
From Figs. 7 and 8, we see that Geff measured today can significantly be different from the Newton’s constant
in the past. The Newton’s constant should be normalized to the current one as (Geff /G). This implies that the
Newton’s constant at the decoupling epoch must be much lower than what is implicitly assumed in CMB codes such
as CAMB [63, 64]. This could significantly change the CMB power spectrum because it changes, for example, the
relation between the gravitational interaction and the Thomson scattering rate. Since we use the CMB data when we
examine whether the theoretical results are consistent with the observational ones analyzed in the framework of GR,
it should be important for us to take into account this point. Therefore, strictly speaking, if we compare our results
with the observations, we has to use the observational results obtained by analyzing the CMB data with using the
present value of Geff in our F (R) gravity models instead of the Newton’s constant G in GR.
Instead of solving Eq. (4.22) for the matter density perturbation δ, we now introduce the growth rate fg ≡
d ln δ/d ln a and solve the equivalent equation to Eq. (4.22) for the growth rate in terms of the redshift z, given
by
fg (z) 3 m̃2 (1 + z)2 Geff (a(z), k)

dfg (z) 1 + z dH(z)
+ − 2 − fg (z) + = 0. (4.25)
dz H(z) dz 1+z 2 H 2 (z) G
Unfortunately, Eq. (4.25) cannot be solved analytically for the models F1 (R) and F2 (R), but it can be solved
numerically by imposing the initial conditions. Therefore, we execute the numerical calculations for both the model
F1 (R) and the model F2 (R) with the condition that at a very high redshift the growth rate becomes that in the ΛCDM
model. In Fig. 9, we illustrate the cosmological evolution as a function of the redshift z and the scale dependence on
the comoving wavenumber k of the growth rate for the model F1 (R), while we depict those of the growth rate for the
model F2 (R) in Fig. 10.
fg
1.2
1.0 1.0
fgHz, k
0.8
0.8
0.6
0.10 8
0.08 6
0.06 0.6
4
0.04
k HMpc-1L 0.02 2 z
0 z
2 4 6 8
(a) (b)
FIG. 9: (a) Cosmological evolution as a function of the redshift z and the scale dependence on the comoving wavenumber k
of the growth rate fg for the model F1 (R). (b) Cosmological evolution of the growth rate fg as a function of z in the model
F1 (R) for k = 0.1Mpc−1 (green), k = 0.01Mpc−1 (red) and k = 0.001Mpc−1 (blue).
One way of characterizing the growth of the matter density perturbations could be to use the so-called growth
index γ, which is defined as the quantity satisfying the following equation:
fg (z) = Ωm (z)γ(z) , (4.26)
with Ωm (z) = 8πGρ

3H 2
m
being the matter density parameter.
It is known that the growth index γ in Eq. (4.26) cannot be observed directly, but it can be determined from the
observational data of both the growth factor fg (z) and the matter density parameter Ωm (z) at the same redshift z.
Even if the growth index is not directly observable quantity, it could have a fundamental importance in discriminating
among the different cosmological models. One of the reasons is that in general, the growth factor fg (z), which can be
estimated from redshift space distortions in the galaxy power spectra at different z [65, 66], may not be expressed in
terms of elementary functions and this fact makes the comparison among the different models difficult. If Eq. (4.26)
is satisfied with any ansatz for the growth index γ, then its determination could provide an easy and fast way to
distinguish between cosmological models.
18
fg
1.2
1.2
1.0
1.0
fgHz, k
0.8
0.8
0.6
0.10 8
0.08 6
0.06 0.6
0.04 4
k HMpc-1L 0.02 2 z
0
z
2 4 6 8
(a) (b)
FIG. 10: (a) Cosmological evolution as a function of the redshift z and the scale dependence on the comoving wavenumber k
of the growth rate fg for the model F2 (R). (b) Cosmological evolution of the growth rate fg as a function of z for the model
F2 (R). Legend is the same as Fig. 9.
Various parameterizations for the growth index γ have been proposed in the literature. In the first stage works
on this topic, γ was taken constant (see [67]). In the case of dark fluids with the constant EoS ω0 in GR, it is
γ = 3 (ω0 − 1) / (6ω0 − 5) (for the ΛCDM model, the growth index is γ ≈ 0.545). Although taking γ constant
is very appropriated for a wide class of dark energy models in the framework of GR (for which |γ ′ (0)| < 0.02),
for modified gravity theories γ is not constant in general (the cases of some viable F (R) gravity models have been
investigated in Refs. [62, 68]) and the measurement of |γ ′ (0)| could be very important in order to discriminate between
different theories. For this reason, another parameterizations has been proposed. The case of a linear dependence
γ(z) = γ0 + γ0′ z was treated in Ref. [69]. Recently, an ansatz of the type γ(z) = γ0 + γ1 z/(1 + z) with γ0 and γ1 being
constants was explored in Ref. [70] and a generalization given by γ(z) = γ0 + γ1 z/(1 + z)α with α being a constant in
Ref. [62]. In the following, we study some of these parameterizations of the growth index for the case of the models
F1 (R) and F2 (R).
1. γ = γ0
We consider the ansatz for the growth index given by
γ = γ0 ,
where γ0 is a constant.
In Fig. 11, we display the results obtained by fitting Eq. (4.26) to the solution of Eq. (4.25) for different values of
the comoving wavenumber k for the two models F1 (R) and F2 (R). We note that in these and following plots, the
bars express the 68% confidence level (CL) and the point denotes the median value. The first important result for
both models is that the value of the growth index has a strong dependence with log k. This scale dependence seems
to be quite similar in both models.
In order to check the goodness of our fits, in Fig. 12 we show cosmological evolutions of the growth rate fg (z) and
Ωm (z)γ0 as functions of the redshift z together for several values of the comoving wavenumber k for the models F1 (R)
and F2 (R). To clarify these results, in Fig. 13 we also illustrate the cosmological evolution of the relative difference
between fg (z) and Ωm (z)γ0 as a function of z for the same values of k in these models. The first remarkable thing is
that for both models the function Ωm (z)γ0 fits the growth rate for large scales (i.e., lower k) very well, but this is not
anymore the case for larger values of k. In fact, if we do not consider lower values for z (i.e., z < 0.2), for log k = −2
the relative difference is smaller than 3% for both models, while for log k = −1 can arrive up to almost 13%. For
log k = −3, we see that the relative difference is always smaller than 1.5% for the model F1 (R) and smaller than 1%
for the model F2 (R).
19
Γ0 Γ0
0.45 0.50
0.45
0.40
0.40
0.35
0.35
0.30 0.30
0.25 0.25
0.20 0.20
Log kHMpc-1L Log kHMpc-1L
-2.5 -2.0 -1.5 -1.0 -2.5 -2.0 -1.5 -1.0
(a) (b)
FIG. 11: Constant growth index as a function of log k for the model F1 (R) (a) and for the model F2 (R) (b). The bars express
the 68% CL.
1.0 1.0
1.1
1.0 0.9 0.9
0.9
0.8 0.8
0.8
0.7 0.7
0.7
0.6 0.6
0.6
z z z
2 4 6 8 2 4 6 8 2 4 6 8
(a) (b) (c)

1.0 1.0
1.1
1.0 0.9 0.9
0.9 0.8
0.8
0.8
0.7 0.7
0.7
0.6 0.6
0.6
z z z
2 4 6 8 2 4 6 8 2 4 6 8
(d) (e) (f)
FIG. 12: Cosmological evolutions of the growth rate fg (red) and Ωγm (blue) with γ = γ0 as functions of the redshift z in
the model F1 (R) for k = 0.1Mpc−1 (a), k = 0.01Mpc−1 (b) and k = 0.001Mpc−1 (c), and those in the model F2 (R) for
k = 0.1Mpc−1 (d), k = 0.01Mpc−1 (e) and k = 0.001Mpc −1 (f).
2. γ = γ0 + γ1 z
With the same procedure used in the previous subsection, we explore a linear dependence for the growth index
γ = γ0 + γ1 z , (4.27)
where γ1 is a constant.
In Fig. 14, we depict the parameters γ0 and γ1 for several values of log k in both the models. As is the same as the
case γ = γ0 , it can easily be seen that the scale dependence of the parameters γ0 and γ1 is similar in these models.
We can also find that γ0 ∼ 0.46 for the model F1 (R) when log k ≤ −2, whereas γ0 ∼ 0.51 for the model F2 (R) when
log k ≤ −2.5. For both these models, the value of γ1 has a strong dependence on k in the range of log k > −2.25, but
in the range of log k < −2.25 this dependence becomes weaker.
In Fig. 15, we illustrate cosmological evolutions of the growth rate fg (z) and Ωm (z)γ(z) as functions of the redshift
z together for the models F1 (R) and F2 (R). We can see that the fits for log k = 0.1 have been improved in comparison
with the same fits as the case with a constant growth index. Also, for log k < 0.1 the fits continue to be quite good. In
order to demonstrate these facts quantitatively, in Fig. 16 we plot the cosmological evolution of the relative difference
between fg (z) and Ωm (z)γ(z) as a function of z for several values of k in the models F1 (R) and F2 (R). In this case,
for log k = −1 the relative difference is smaller than 7.5% in both the models if we do not consider lower values for
20
fg - WmΓ fg - WmΓ
fg fg
0.30
0.25
0.25
0.20
0.20
0.15
0.15
0.10
0.10
0.05 0.05
z z
2 4 6 8 2 4 6 8
(a) (b)
|fg −Ωγm |
FIG. 13: Cosmological evolution of the relative difference fg
with γ = γ0 for k = 0.1Mpc−1 (red), k = 0.01Mpc−1 (blue)
and k = 0.001Mpc−1 (green) in the model F1 (R) (a) and the model F2 (R) (b).
Γ0 Γ1
0.65 0.0
-0.2
0.60
-0.4
0.55
-0.6
0.50
-0.8
-1
Log kHMpc L Log kHMpc-1L
-2.5 -2.0 -1.5 -1.0 -2.5 -2.0 -1.5 -1.0
(a) (b)
Γ0 Γ1
0.65 0.0
-0.2
0.60
-0.4
0.55
-0.6
0.50 -0.8
Log kHMpc L-1

Log kHMpc-1L
-2.5 -2.0 -1.5 -1.0 -2.5 -2.0 -1.5 -1.0
(c) (d)
FIG. 14: Growth index fitting parameters in the case γ = γ0 + γ1 z as a function of log k for the model F1 (R) [(a) and (b)]
and the model F2 (R) [(c) and (d)]. Legend is the same as Fig. 11.
z (i.e., z < 0.2). We also see that the linear growth index improves the fits in both the models for log k = −2 in
comparison with those for a constant growth index. In this case, the relative difference for the model F1 (R) is always
smaller than 1%, whereas that for model F2 (R) is smaller than 2%. Finally, for log k = −3 the results obtained for a
constant growth index are quite similar to those for a linear dependence on z.
21
1.0 1.0
1.1
1.0 0.9 0.9
0.9
0.8 0.8
0.8
0.7 0.7
0.7
0.6 0.6
0.6
z z z
2 4 6 8 2 4 6 8 2 4 6 8
(a) (b) (c)

1.0 1.0
1.1
1.0 0.9 0.9
0.9
0.8 0.8
0.8
0.7 0.7
0.7
0.6 0.6
0.6
z z z
2 4 6 8 2 4 6 8 2 4 6 8
(d) (e) (f)
FIG. 15: Cosmological evolutions of the growth rate fg (red) and Ωγm (blue) with γ = γ0 + γ1 z as functions of the redshift z
in the model F1 (R) for k = 0.1Mpc−1 (a), k = 0.01Mpc−1 (b) and k = 0.001Mpc −1 (c), and those in the model F2 (R) for
fg - WmΓ fg - WmΓ
fg fg
0.25 0.25
0.20 0.20
0.15 0.15
0.10 0.10
0.05 0.05
z z
2 4 6 8 2 4 6 8
(a) (b)
|fg −Ωγm |
with γ = γ0 + γ1 z for k = 0.1Mpc−1 (red), k = 0.01Mpc−1
(blue) and k = 0.001Mpc−1 (green) in the model F1 (R) (a) and the model F2 (R) (b).
z
3. γ = γ0 + γ1 1+z
Next, we examine the following ansatz for the growth index:

z
γ = γ0 + γ1 . (4.28)
1+z
In Fig. 17, we depict the parameters γ0 and γ1 for several values of log k for both the models. The scale dependence
of these parameters on k is shown. The behavior of the parameter γ1 seems to be quite similar to that for the previous
case γ = γ0 + γ1 z, but it is worth cautioning that the scale of the figures are different from each other, and that for
the present ansatz the scale dependence of γ1 is stronger than that for the previous case. It can also be seen that
γ0 ∼ 0.465 for the model F1 (R) and γ0 ∼ 0.513 for the model F2 (R) in the scale log k < −2.5.
In Fig. 18, we plot cosmological evolutions of the growth rate fg (z) and Ωm (z)γ(z) in the models F1 (R) and F2 (R)
for several values of k, as demonstrated in the previous subsections. We can see the fits for log k ≤ −2 are quite
22
Γ0 Γ1
0.80 0.0
0.75
-0.5
0.70
0.65 -1.0
0.60
0.55 -1.5
0.50
-2.0
-2.5 -2.0 -1.5 -1.0 -2.5 -2.0 -1.5 -1.0
(a) (b)
Γ0 Γ1
0.80 0.0
0.75 -0.5
0.70
-1.0
0.65
0.60 -1.5
0.55 -2.0
-2.5 -2.0 -1.5 -1.0 -2.5 -2.0 -1.5 -1.0
(c) (d)
z
FIG. 17: Growth index fitting parameters in the case γ = γ0 + γ1 1+z as a function of log k for the model F1 (R) [(a) and (b)]
and the model F2 (R) [(c) and (d)]. Legend is the same as Fig. 11.
good, as those in the previous ansatz for the growth index. In the case of higher values of log k, it seems that the
fits are similar to those for a constant growth rate and these fits do not reach the goodness of those for the case of
γ = γ0 + γ1 z.
1.0 1.0
1.1
1.0 0.9
0.9
0.9
0.8 0.8
0.8
0.7 0.7 0.7
0.6
0.6 0.6
0.5
z z z
2 4 6 8 2 4 6 8 2 4 6 8
(a) (b) (c)

1.0 1.0
1.1
1.0 0.9 0.9
0.9
0.8 0.8
0.8
0.7 0.7 0.7
0.6
0.6 0.6
0.5
z z z
2 4 6 8 2 4 6 8 2 4 6 8
(d) (e) (f)
FIG. 18: Cosmological evolutions of the growth rate fg (red) and Ωγm (blue) with γ = γ0 + γ1 1+z
z
as functions of the redshift
−1 −1 −1
z in the model F1 (R) for k = 0.1Mpc (a), k = 0.01Mpc (b) and k = 0.001Mpc (c), and those in the model F2 (R) for
In order to analyze the fits quantitatively, in Fig. 19 we display the cosmological evolution of the relative difference
23
between fg (z) and Ωm (z)γ(z) for several values of k in the models F1 (R) and F2 (R). We see that the relative difference
for log k = −1 is smaller than 12% (if we do not consider z < 0.2) for both the models. Thus, it is confirmed that
these fits are better than those for the constant growth rate, but these are worse than those for γ = γ0 + γ1 z. For
lower values of log k, the relative difference is smaller than 2% in z > 0.2.
fg - WmΓ fg - WmΓ
fg fg
0.35 0.35
0.30 0.30
0.25 0.25
0.20 0.20
0.15 0.15
0.10 0.10
0.05 0.05
z z
2 4 6 8 2 4 6 8
(a) (b)
|fg −Ωγm |
z
with γ = γ0 + γ1 1+z for k = 0.1Mpc−1 (red), k = 0.01Mpc−1
−1
(blue) and k = 0.001Mpc (green) in the model F1 (R) (a) and the model F2 (R) (b).
As a consequence, through the investigations of these different ansatz for the growth index, it is concluded that
γ = γ0 + γ1 z is the parameterization that can fit Eq. (4.26) to the solution of Eq. (4.25) better in a wide range of
values for k. Even though the behavior of the parameters γ0 and γ1 in the models F1 (R) and F2 (R) is quite similar to
each other, in order to distinguish between these models in Fig. 14 we can see that the more differences between these
models come from the values of γ0 for log k ≤ −2. In fact, as remarked before, for log k ≤ −2.5 we have γ0 ∼ 0.46 for
the model F1 (R) and γ0 ∼ 0.51 for the model F2 (R).
V. UNIFIED MODELS FOR EARLY AND LATE-TIME COSMIC ACCELERATION
The reason why we also study inflation in F (R) gravity is that one of the most important goals on the study of
modified gravity theories is to describe the consistent evolution history of the universe from inflation in the early
universe to the dark energy dominated stage at the present time. Namely, the universe starts with an inflationary
epoch, followed by the radiation dominated era and the matter dominated universe, and finally the late cosmic
acceleration epoch is actually achieved without invoking the presence of dark components in the universe [11] (for a
first F (R) theory unifying inflation with dark energy, see Ref. [17]). In Secs. III and IV, it has been demonstrated that
the exponential gravity with the correction terms can be a realistic F (R) gravity model. Therefore, in this section we
investigate the possibility that in such exponential gravity with additional correction terms, inflation as well as the
late-time cosmic acceleration can be realized. Since we examine exponential gravity among various models of F (R)
gravity, we consider the unification model between inflation and the late-time cosmic acceleration in this work.
Models of the type (3.2) may be combined in a natural way to obtain the phenomenological description of the
inflationary epoch. For example, a ‘two-steps’ model may be the smooth version, given by
h i
F (R) = R − 2Λ 1 − e−R/(b R) − Λi θ(R − Ri ) . (5.1)
Here, θ(R − R0 ) is the Heaviside’s step distribution, Ri is the transition scalar curvature at inflationary scale and Λi
is a suitable cosmological constant producing inflation, when R ≫ Ri . The main problem associated with this sharp
model is the appearance of a possible antigravity regime in a region around the transition point between inflation and
the universe described by the ΛCDM model. The antigravity in a past epoch is not phenomenologically acceptable.
Furthermore, adding some terms would be necessary in order for inflation to end.
In this section, we study two applications of exponential gravity to achieve an unified description of the early-
time inflation and the late-time cosmic acceleration. In particular, we show how it is possible to obtain inflationary
universes with different numbers of e-folds by choosing different models parameters in the presence of ultrarelativistic
matter in the early universe.
24
Following the first proposal of Ref. [39], we start with the form of F (R) with a natural possibility of a unified
description of our universe
n
!
R
R
− R 1
F (R) = R − 2Λ 1 − e − bR
− Λi 1 − e i + γ̄ Rα , (5.2)
R̃iα−1
where Ri and Λi are the typical values of transition curvature and expected cosmological constant during inflation,
respectively, and n is a natural number larger than unity (here, we do not write the correction term for the stability of
oscillations in the matter dominated era). In Eq. (5.2), the last term γ̄(1/R̃iα−1 )Rα , where γ̄ is a positive dimensional
constant and α is a real number, works at the inflation scale R̃i and is actually necessary in order to realize an exit
from inflation.
We also propose another nice inflation model based on the good behavior of exponential function described as
n
R
− R
sin π e i !
R
1
F (R) = R − 2Λ 1 − e − bR
− Λi n + γ̄ α−1
Rα . (5.3)
πe
− RR
i
R̃i
Here, the parameters have the same roles of the corresponding ones in the model in Eq. (5.2). We note that the
second term of the model vanishes when R ≪ Ri and tends to Λi when R ≫ Ri . We analyze these models, i.e., Model
I in Eq. (5.2) and Model II in Eq. (5.3), and explore the possibilities to reproduce the phenomenologically acceptable
inflation.
A. Inflation in exponential model (Model I)
First, we investigate the model in Eq. (5.2). For simplicity, we describe a part of it as
n
!
− RR 1
fi (R) ≡ −Λi 1 − e i + γ̄ Rα . (5.4)
R̃iα−1
We note that if n > 1 and α > 1, when R ≪ Ri (∼ R̃i ), we obtain

Rn Rα
R ≫ |fi (R)| ≃ − n−1 + γ̄ α−1 , (5.5)

Ri R̃i
and the absence of the effects of inflation during the matter dominated era. We also find
α−1
Λi nRn−1 − RR n R

′
fi (R) = − n e i + γ̄α , (5.6)
Ri R̃i
2 n
Λi n(n − 1)Rn−2 − RR n nRn−1 Rα−2

− R
fi′′ (R) = − n e i + Λ i n e Ri + γ̄α(α − 1) α−1 . (5.7)
Ri Ri R̃i
1/n
Since when R = Ri [(n − 1)/n] the negative term of fi′ (R) has a minimum, in order to avoid the anti-gravity effects
′
(this means, |fi (R)| < 1), it is sufficient to require
n−1
n−1 n n−1
Ri > Λi n e− n . (5.8)
n
It is necessary for the modification of gravity describing inflation not to have any influence on the stability of the
matter dominated era in the small curvature limit. When R ≪ Ri , the second derivative of fi′′ (R), given by
" n−1 α−1 #
1 R R
fi′′ (R) ≃ −n(n − 1) + γ̄α(α − 1) , (5.9)
R Ri R̃i
must be positive, that is,
n > α. (5.10)
25
We require the existence of the de Sitter critical point RdS which describes inflation in the high-curvature regime of
fi (R), so that fi (RdS ≫ Ri ) ≃ −Λi + γ̄(1/Riα−1 ) Rα . In this case, if we put R̃i = RdS , we may solve the trace of the
field equation (2.3) in vacuum for a constant curvature, namely 2F (R) − RF ′ (0) = 0, and therefore we obtain (in
vacuum, namely, if the effective modified gravity energy density is dominant over matter),
n
2Λi RdS
RdS = , ≫ 1. (5.11)
γ̄(2 − α) + 1 Ri
The last two conditions have to be satisfied simultaneously. By using Eq. (5.8), we also acquire
n−1
2 n−1 n n−1
>n e− n . (5.12)
γ̄(2 − α) + 1 n
Instability and number of e-folds during inflation
The well-known condition to have an instable de Sitter solution (see Sec. IV B) is given by
F ′ (RdS )
< 1, (5.13)
RdS F ′′ (RdS )
which leads to
α γ̄(α − 2) > 1 , (5.14)
for our model. Here, we have considered fi (RdS ) ≃ −Λi + γ̄(1/Riα−1 ) Rα . From Eqs. (5.12)–(5.14), we have to require
2 + 1/γ̄ > α > 2 . (5.15)
Thus, we may evaluate the characteristic number of e-folds during inflation

zi + 1
N = log , (5.16)
ze + 1
where zi and ze are the redshifts at the beginning and at the end of early time cosmic acceleration. Given a small
cosmological perturbation y1 (zi ) at the redshift zi , we have from Eq. (4.13) avoiding the matter contribution
y1 (zi ) = C0 (zi + 1)x , (5.17)
with
s !
1 16F ′ (RdS )
x= 3− 25 − , (5.18)
2 RdS F ′′ (RdS )
where x < 0 if the de Sitter point is unstable. Thus, the perturbation y1 (z) in Eq. (4.13) grows up in expanding
universe as
x
(z + 1)
y1 (z) = y1 (zi ) . (5.19)
(zi + 1)
Here, we have considered C0 = y1 (zi )/(zi + 1)x . When y1 (z) is on the same order of the effective modified gravity
energy density y0 of the de Sitter solution describing inflation (we remind, y0 = RdS /(12m̃2 )), the model exits from
inflation. We can estimate the number of e-folds during inflation as

1 y1 (zi )
N ≃ log . (5.20)
x y0
A value demanded in most inflationary scenarios is at least N = 50–60.
A classical perturbation on the (vacuum) de Sitter solution may be given by the presence of ultrarelativistic matter
in the early universe. The system gives rise to the de Sitter solution where the universe expands in an accelerating
way but, suddenly, it exits from inflation and tends towards the minimal attractor at R = 0 (the trivial de Sitter
point). In this way, the small curvature regime arises and the physics of the ΛCDM model is reproduced.
26
B. Inflation in Model II
Next, we study the inflation model in Eq. (5.3). By performing a similar analysis to that in the previous subsection,
we find that also in this case, if α > 1 and n > 1, we avoid the effects of inflation at small curvatures and it does
not influence the stability of the matter dominated era. The de Sitter point exists if R̃i = RdS and it reads as in
Eq. (5.11) under the condition (R/Ri )n ≫ 1. Thus, the inflation is unstable if the condition in Eq. (5.14) is satisfied.
The bigger difference between the two models exists in those behaviors in the transition phase between the small
curvature region (where the physics of the ΛCDM model emerges) and the high curvature region. This means that
the no antigravity condition is different in the two models and such a condition becomes more critical in the transition
region. Therefore, in the following we are able to make the different choices of parameters in the two models. We
note that since dark energy sector of the above models only originates from exponential gravity, all qualitative results
in terms of the behavior of the dark energy component in exponential gravity found in the previous sections remain
to be valid.
VI. ANALYSIS OF INFLATION
In this section, we perform the numerical analysis of the early time acceleration for the unified models in Eqs. (5.2)
and (5.3), by choosing appropriate parameters according with the analysis in Sec. V. For this aim, it is worth rewriting
Eq. (2.10) by introducing a suitable scale factor M 2 at the inflation. We can choose M 2 = Λi . The effective modified
gravity energy density yH (z) is now defined as
ρDE 3H 2
yH (z) ≡ 2 2
= − χ̃(z + 1)4 . (6.1)
M /κ M2
Here, we have neglected the contribution of standard matter and supposed the presence of ultrarelativistic mat-
ter/radiation in the hot universe scenario, whose energy density ρrad at the redshift equal to zero is related with the
scale as
κ2 ρrad
χ̃ = . (6.2)
M2
Since the results are independent of the redshift scale, we set z = 0 at some times around the end of inflation. Equation
(2.10) reads
y ′ (z) 1 − F ′ (R)

′′
yH (z) − 3+
z+1 2M 2 F ′′ (R) [yH (z) + χ̃(z + 1)4 ]
yH (z) 2 − F ′ (R)
+
(z + 1) M F (R) [yH (z) + χ̃(z + 1)4 ]
2 2 ′′
(F ′ (R) − 1)2χ̃(z + 1)4 + (F (R) − R)/M 2

+ = 0. (6.3)
(z + 1)2 2M 2 F ′′ (R) [yH (z) + χ̃(z + 1)4 ]
Moreover, the Ricci scalar is expressed as

dyH (z)
R = M 2 4yH (z) − (z + 1) . (6.4)
dz
Thus, it is easy to verify that in the de Sitter universe with R = RdS the perturbation y1 (z) on the solution y0 =
RdS /(4 M 2 ) is effectively given by Eq. (5.17), i.e., y1 = C0 (z + 1)x , according with Eq. (4.13) if we neglect the
contribution of standard matter. In this derivation, we have assumed the contribute of ultrarelativistic matter to be
much smaller than y0 . However, as stated above, this small energy contribution may originate from the perturbation
y1 (zi ) at the beginning of inflation, which, if x < 0, grows up in the expanding universe making inflation unstable.
27
Model I
First, we explore the model in Eq. (5.2). We have to choose the parameters as Λi ≃ 10100−120 Λ. The dynamics of
the system is independent of this choice. Here, we summarize the conditions for inflation already stated in Sec. V A:
n−1
n−1 n n−1
Ri > Λi n e− n , (no antigravity effects)
n
n
RdS
R̃i = RdS , αγ̄(α − 2) > 1 , ≫ 1 , (existence of unstable dS solution)
Ri
with n > 1, 2 + 1/γ̄ > α > 2 and RdS = 2Λi / [γ̄(2 − α) + 1]. Since γ̄ and α are combined in γ(α − 2), we can fix
γ̄ = 1, so that RdS = 2Λi /(3 − α) and 3 > α > 2. The instability factor x in Eq. (5.18) only depends on RdS . Hence,
by studying the phenomenology of inflation, we examine the variation of α parameter (and, as a consequence, that
of R̃i ). We take n = 4 and Ri = 2Λi , which satisfy the condition for no antigravity well. We analyze three different
cases of α = 5/2, 8/3, and 11/4. In these cases, we have RdS = 4Λi , 6Λi , and 8Λi , respectively.
2.0 2 2
1.5
1 1
1.0
0.5 2 4 6 8 10
2 4 6 8
1 2 3 4 5 6 -1
-1
-0.5
-2
-1.0 -2
-1.5 -3
(a) (b) (c)
FIG. 20: Cosmological evolution of the quantity 2F (R/Λi ) − (R/Λi )F ′ (R/Λi ) as a function of the redshift z for exponential
model with α = 5/2 (a), α = 8/3 (b) and α = 11/4 (c). The “zeros” of these graphics indicate the de Sitter solutions of the
model.
In Fig. 20, we plot the cosmological evolution of the quantity 2F (R/Λi) − (R/Λi )F ′ (R/Λi ) as a function of the
redshift z in the three cases. The value of “zero” of this quantity corresponds to the de Sitter points of the model.
We can recognize the unstable de Sitter solutions of inflation and the attractor in zero (the de Sitter point of current
acceleration is out of scale).
Despite the fact that the three considered values of α are very close each other, the values of RdS and x significantly
change and the reactions of the system to small perturbations are completely different. By starting from Eq. (5.20), we
may reconstruct the rate y1 (zi )/y0 between the abundances of ultrarelativistic matter/radiation and modified gravity
energy at the beginning of inflation in order to obtain a determined number of e-folds during inflation in the three
different cases, by taking into account that x = −0.086, −0.218, and −0.270 for α = 5/2, 8/3, and 11/4, respectively.
For example, in order to have N = 70, for α = 5/2, a perturbation of y1 (zi )/y0 ∼ 10−3 is necessary; for α = 8/3,
a perturbation of y1 (zi )/y0 ∼ 10−7 is sufficient; whereas for α = 11/4, y1 (zi )/y0 ∼ 10−9 . The system becomes more
unstable, as (3 − α) is closer to zero.
In studying the behavior of the cosmic evolution in Model I for the three different cases, we set χ̃ = 10−4 y0 /(zi + 1)4
in Eq. (6.3) for the case α = 5/2 and χ̃ = 10−6 y0 /(zi + 1)4 for the cases α = 8/3, 11/4. In these choices, the effective
energy density originating from the modification of gravity is 104 and 106 times larger than that of ultrarelativistic
matter/radiation during inflation. By using Eq. (5.20), we can predict the following numbers of e-folds:
N ≃ 107 (for α = 5/2) ,

N ≃ 64 (for α = 8/3) ,
N ≃ 51 (for α = 11/4) . (6.5)
In order to solve Eq. (6.3) numerically, we use the initial conditions

dyH (z)
= 0,
d(z) zi
RdS
yH (z) = , (6.6)

zi 4Λi
28
at the redshift zi ≫ 0 when inflation starts. We put zi = 1046 , 1027 , and 1022 for α = 5/2, 8/3, and 11/4, respectively
(just for a more comfortable reading of the graphics). We also remark that the initial conditions are subject to an
n
artificial error that we can estimate to be in the order of exp [− (RdS /Ri ) ] ∼ 10−7 . This is the reason for which we
only consider χ̃ > 10−7 .
y_H y_H
y_H 1.5 2
1.00000
1.5 2
0.99999
1.5 2
0.99998
1.5 2
0.99997
0.99996 1.5 2
z z z
2 ´ 1045 4 ´ 1045 6 ´ 1045 8 ´ 1045 1 ´ 1046 2 ´ 1026 4 ´ 1026 6 ´ 1026 8 ´ 1026 1 ´ 1027 2 ´ 1021 4 ´ 1021 6 ´ 1021 8 ´ 1021 1 ´ 1022
(a) (b) (c)

Ω_MG Ω_MG
Ω_MG
-1 -1
-1
-0.999992
-1
-1
-0.999994
-1 -1
-0.999996 -1
-1
-1
-0.999998 -1
-1
z z z
2 ´ 1045 4 ´ 1045 6 ´ 1045 8 ´ 1045 1 ´ 1046 2 ´ 1026 4 ´ 1026 6 ´ 1026 8 ´ 1026 1 ´ 1027 2 ´ 1021 4 ´ 1021 6 ´ 1021 8 ´ 1021 1 ´ 1022
(d) (e) (f)
FIG. 21: Plots of yH [a-c] and ωMG [d-f] as functions of the redshift z for Model I with α = 5/2 [a-d], α = 8/3 [b-e] and
α = 11/4 [c-f].
In Fig. 21, we illustrate the cosmological evolutions of yH and the corresponding modified gravity EoS parameter
ωMG (defined as in Eq. (3.13)) as functions of the redshift z in the three cases. We can see, during inflation ωMG
is indistinguishable from the value of -1 and yH tends to decrease very slowly with respect to yH = 1, 3/2, 2 for
α = 5/2, 8/3, 11/4, so that the curvature can be the expected de Sitter one, RdS (= 4yH ) = 4Λi , 6Λi , 8Λi . The
expected values of ze at the end of inflation may be derived from the number of e-folds in (6.5) during inflation and
read ze ≃ −0.47 for α = 5/2; ze ≃ −0.74 for α = 8/3; ze ≃ −0.39 for α = 11/4. The numerical extrapolation yields
yH (ze ) = 0.83yH (zi ) , R(ze ) = 0.825RdS , (for α = 5/2)

yH (ze ) = 0.92yH (zi ) , R(ze ) = 0.911RdS . (for α = 11/4)
To confirm the exit from inflation, in Fig. 22 we plot the cosmological evolutions of yH and R/Λi as functions of the
redshift z in the region −1 < z < 1, where ze is included. The effective modified gravity energy density and the
curvature decrease at the end of inflation and the physical processes described by the ΛCDM model can appear.
Model II
Next, we investigate Model II in Eq. (5.3). Here, in order to satisfy the condition for no antigravity we choose
n = 3 and Ri = 2Λi , so that F ′ (R > 0) > 0. We take γ̄ = 1 again and we execute the same numerical evaluation
for α = 5/2, 13/5, 21/8 in this model as that in the previous case for Model I. The corresponding de Sitter curvatures
of inflation are RdS = 4Λi , 5Λi , 16Λi /3. Now, we obtain the factor in Eq. (5.18) for instability as x = −0.086,
−0.170, and −0.188 for α = 5/2, 13/5, and 21/8, respectively. Hence, we set χ̃ = 10−3 y0 /(zi + 1)4 for α = 5/2,
χ̃ = 10−4 y0 /(zi + 1)4 for α = 13/5, and χ̃ = 10−5 y0 /(zi + 1)4 for α = 21/8. As a consequence, the numbers of e-folds
during inflation result in N = 80, 54, and 61. The initial conditions are the same as those in the previous case in
(6.6). Furthermore, we put zi = 1034 , 1022 , and 1026 for α = 5/2, 13/5, and 21/8.
Through the numerical extrapolation, we acquire the expected values of ze at the end of inflation as ze = −0.80,
29
y_H y_H y_H
0.840
1.38
0.835
1.85
1.36
0.830
0.825 1.34
1.80
0.820
1.32
0.815
1.30
z
0.810 -1.0 -0.5 0.5 1.0
z
z -1.0 -0.5 0.5 1.0
-1.0 -0.5 0.5 1.0
(a) (b) (c)

RLi RLi
RLi
5.5
z
-1.0 -0.5 0.5 1.0
3.34 7
3.32 5.0
6
3.30
4.5
3.28
5
3.26
4.0
3.24
4
3.22 3.5
z z
-1.0 -0.5 0.5 1.0 -1.0 -0.5 0.5 1.0
(d) (e) (f)
FIG. 22: Cosmological evolution of yH [a-c] and R/Λi [d-f] as functions of the redshift z in the region −1 < z < 1 for Model I
with α = 5/2 [a-d], α = 8/3 [b-e] and α = 11/4 [c-f].
−0.97, and −0.71, and the following values for the effective modified gravity energy density and the Ricci scalar:

yH (ze ) = 0.79yH (zi ) , R(ze ) = 0.780RdS . (for α = 21/8)
For this model, in Fig. 23 we depict the cosmological evolutions of yH and R/Λi as functions of the redshift z in
the region −1 < z < 1 at the end of inflation. Again in this case, the effective modified gravity energy density and
curvature decrease, and therefore inflation ends and then the physical processes described by the ΛCDM model can
be realized.
Here, we note that at the inflationary stage, radiation is negligible, as in the ordinary inflationary scenario. It
causes the perturbations at the origin of instability. This point has been shown in a numerical way by using radiation,
whose energy density is six order of magnitude smaller than that of dark energy.
It should be emphasized that in this work, as a first step, we have concentrated on only the possibility of the
realization of inflation, and hence that important issues in inflationary cosmology such as the graceful exit problem of
inflation, the following reheating process, and the generation of the curvature perturbations, whose power spectrum
has to be consistent with the anisotropies of the CMB radiation obtained from the Wilkinson Microwave Anisotropy
Probe (WMAP) Observations [4–6], are the crucial future works of our unified scenario between inflation and the
late-time cosmic acceleration.
In the future works, if we analyze the power spectrum of the curvature perturbations in our models, the next
question becomes not what the total number of e-folds is, but how many e-folds one could obtain from the point when
the power-law index of the primordial power spectrum ns is close to its observed value. It is presumed that since the
equation of state w at the inflationary stage is so close to the model, e.g., with α = 11/4, the number of e-folds from
the point when ns ≃ 0.96 [5, 6] until the end of inflation is much smaller. Accordingly, we should examine whether it
is enough for the galaxy power spectrum to be reasonably close to the scale invariance of the power spectrum of the
curvature perturbations. Moreover, as a more relevant question which remains is the mechanism for reheating. The
problem is how the universe becomes the radiation dominated stage again after the inflationary period. In order to
construct complete models of inflation, we need a discussion of the reheating mechanism and that of exactly how the
power spectrum of anisotropies is transferred to the matter. These are significant future subjects in our studies.
30
y_H y_H y_H
0.830 1.14
1.10
0.825 1.13
1.12
0.820
1.11 1.05
0.815
1.10
0.810 1.09
1.00
z z
-1.0 -0.5 0.5 1.0 -1.0 -0.5 0.5 1.0 z
-1.0 -0.5 0.5 1.0
(a) (b) (c)

RLi RLi RLi
4.6
3.32
4.4
4.4
z
-1.0 -0.5 0.5 1.0 4.2
4.2
3.28
4.0
4.0
3.26
3.8
3.8
3.24
3.6 3.6
3.22
3.4 3.4
z z
3.20 -1.0 -0.5 0.5 1.0 -1.0 -0.5 0.5 1.0
(d) (e) (f)
FIG. 23: Cosmological evolution of yH [a-c] and R/Λi [d-f] as functions of the redshift z in the region −1 < z < 1 for Model II
with α = 5/2 [a-d], α = 13/5 [b-e] and α = 21/8 [c-f].
VII. CONCLUSIONS AND GENERAL REMARKS
In the present paper, we have examined a generic feature of viable F (R) gravity models, in particular, exponential
gravity as well as a power form model. We have shown that the behavior of higher derivatives of the Hubble
parameter may be affected by large frequency oscillations of effective dark energy, and consequently solutions may
become singular and unphysical at a high redshift. The analyzed models approach to a model with the cosmological
constant in a manner different from each other, and hence it is reasonable to expect that the found results can be
generalized to realistic F (R) gravity models, in which the cosmological evolutions are similar to those in a model
with the cosmological constant. To support our claim, in the first part of this paper we have explicitly demonstrated
how the origin of the problem influences the stability conditions satisfied by these models in order to reproduce the
realistic matter dominated era. Since the corrections to the Einstein equations at the small curvature regime may lead
to undesired effects at the high curvature regime, we have reconstructed a correcting (compensating) term added to
the models in order to stabilize the oscillations of the effective dark energy in the matter dominated era with retaining
the viability properties. It is emphasized that all the results we have found in an analytical way via studying the
perturbation theory are confirmed by the numerical analysis performed on the models under consideration. Moreover,
a detailed investigation on the cosmological evolutions of the universe described by those models has been executed. In
particular, we have demonstrated that our correction term does not cause any problem to the viability of the models,
and that the obtained results are consistent with recent very accurate observational data of our current universe and
easily pass the local tests of the solar system. Furthermore, we have shown that the effective crossing of the phantom
divide, which characterizes the de Sitter epoch, occurs in the very far future. A way to avoid the crossing of phantom
divide by using inhomogeneous fluids has also been explored.
After the discovery of the accelerated expansion of our universe, a lot of theories are proposed in order to explain
it. The issue of discriminating among all of these theories has become very important. The first step in order to
distinguish between theories can be the study of their expansion history, but it has been revealed that sometimes
different models exhibit the same (or very similar) expansion history. For this reason, the investigation of growth
of the matter density perturbations by using the so-called growth index can provide a significant tool in order to
distinguish among the different gravitational theories. In this context, the growth of the matter density perturbations
has been examined for our models. Several ansatz for the growth index have been considered, and consequently it
has been concluded that the choice of the growth index as γ = γ0 + γ1 z is the most appropriate parameterization for
these theories.
In addition, in the second part of this paper we have discussed the inflationary cosmology in two exponential
gravity models. It has explicitly been shown that different numbers of e-folds during inflation can be obtained by
taking different model parameters in the presence of ultrarelativistic matter, the existence of which makes inflation
31
to end and realize the exit from it. We have performed the numerical analysis of the inflationary stage in two viable
exponential gravity models. It has been found that at the end of the inflation, the effective energy density and
therefore the curvature of the universe become small. As a result, we have proved that it is possible to acquire a
gravitational alternative scenario for a unified description of inflation in the early universe with the late-time cosmic
acceleration due to the ΛCDM-like dark energy domination.
It should be cautioned that in this work, we have constructed a unified description of inflation with the late-time
cosmic acceleration in F (R) gravity by examining the cosmological evolutions of inflation in Secs. V and VI and
the late-time cosmic acceleration in Sec. III one by one, and therefore that the evolution equation expressing all
the processes from inflation to the current accelerated cosmic acceleration has not been obtained yet. In order to
obtain such a gravitational field equation, the detailed considerations on the reheating process after inflation is also
necessary (for a very recent analysis, see, e.g, [71]). Qualitatively, from our results it can presumably be considered
that at the inflationary stage the EoS parameter weff is approximately equal to −1 and after that it becomes close
to 1/3 during the reheating stage because of the appearance of radiation, and after the radiation-dominated stage
with weff ≈ 1/3 following the matter-dominated stage with weff ≈ 0, the dark energy dominated stage with weff ≈ −1
can be realized. If we successfully acquire the equation and solve it analytically or numerically, it would be possible
to plot the evolution of the Hubble expansion rate H or weff from the inflationary stage in the early universe to the
present time. This is very interesting and significant task in our aim, hence it would be one of the important future
works of our study.
We also mention that as another important future work in terms of our present investigations, at the next step we
plan to study cosmological perturbations [72, 73] in such resultant F (R) gravity theories. We calculate the power
spectrum of the cosmological perturbations as well as the tensor-to-scalar ratio in these models and compare those
with the observational data from such as WMAP satellite [6], future PLANCK satellite [74, 75], QUIET [76, 77],
B-Pol [78] and LiteBIRD [79] in terms of the polarization of the CMB radiation. Furthermore, it is meaningful to
remark that the growth of the matter density perturbations in modified gravity affects the spectrum of weak lensing
(for a concrete way of comparing the theoretical predictions with the observations, see [81]), and therefore more
precise future observations of weak lensing effects have a potential to present the chance to find out the signal of the
modification of gravity.
It is considered that the consequences obtained in this work can be a clue of explore the features of dark energy
as well as inflation. By developing this work further, it is strongly expected that we are able to construct a more
sophisticated and realistic inflation model, in which the power spectrum of the curvature perturbations is consistent
with the observations, the reheating mechanism is well understood, and the structure formation can be explained
more naturally.
Acknowledgments
We would like to thank Mark Trodden for useful comments. K.B., S.D.O. and L.S. would like to appreciate the
support and very kind hospitality at Eurasian National University, where the work was developed. K.B. also expresses
his sincere gratitude to National Center for Theoretical Sciences and National Tsing Hua University very much for
the very kind and warm hospitality, where the revision of this work was executed. A. L-R. is grateful to MICINN
(Spain) for three months visiting grant at Trento university. The work is supported in part by MICINN (Spain)
project FIS2010-15640 and AGAUR (Catalonia) 2009SGR-994 (S.D.O. and A. L-R.).
Appendix A: Conformal transformation of exponential model for inflation
In several cases, a suitable conformal frame to study inflation may be the so-called “Einstein frame”. An F (R)
gravity theory can be rewritten in the scalar field theory form via the conformal transformation. We can rewrite the
action in Eq. (2.1) by introducing a scalar field which couples to the curvature. Of course, this is not exactly physically-
equivalent formulation, but the formulation in the Einstein frame may be used to obtain some of intermediate results
in simpler form (especially, the case that the matter is not taken into account).
We introduce a scalar field A into the action
1 √
Z
IJF = 2 −g [F ′ (A) (R − A) + F (A)] d4 x . (7.1)
2κ M
Here, the subscript “JF” means “the Jordan frame” and we neglect the contribute of matter. By making the variation
32
of the action with respect to A, we have A = R. We define the scalar field σ as

√
3
σ = −√ ln[F ′ (A)] . (7.2)
2κ2
We make the conformal transformation of the metric
g̃µν = e−σ gµν , (7.3)
for which we acquire the “Einstein frame” (EF) action of the scalar field σ [80]
( 2 )
1 F ′′ (A)

R̃ 1 A F (A)
Z
4 µν
p
IEF = d x −g̃ − g̃ ∂µ A∂ν A − 2 +
M 2κ2 2 F ′ (A) 2κ F ′ (A) F ′ (A)2
!
R̃ 1 µν
Z
4
p
= d x −g̃ − g̃ ∂µ σ∂ν σ + V (σ) , (7.4)
M 2κ2 2
where

1 A F (A) 1 σ
e R(e−σ ) − e2σ F [R(e−σ )] .

V (σ) ≡ − − ′ =− (7.5)
2κ2 F (A) F (A)2
′ 2κ 2
Here, R(e−σ ) is the solution of Eq. (7.2) with A = R, becoming R a function of e−σ , and R̃ denotes the Ricci scalar
evaluated with respect to the conformal metric g̃µν . Furthermore, g̃ = e−4σ g is the determinant of conformal metric.
As an example, we explore our unified model (5.2) with γ = 1. Since we are interested in the de Sitter solution, we
take exp[−(R/Ri )n ] → 0 and neglect the cosmological constant Λ. In this case the potential V (σ) reads
" 1 #
1 e−σ̃ − 1 α−1 σ̃ 2σ̃
2σ̃
V (σ) = − 2 R̃ e − 2e + Λi e . (7.6)
2κ α
p
According with Sec. V A, we put R̃i = RdS . It is clearly seen that for R = RdS , σdS = − 3/(2κ2 ) log(1 + α) and
V ′ (σdS ) = 0, where the prime denotes the derivative with respect to the inflation field σ. Since V ′′ (σdS ) > 0, the
scalar potential has a minimum, that is a necessary condition for a slow-roll inflation. For slow-roll parameters, we
have to require
′ 2
1 V (σ)
ǫ(σ) = ≪ 1,
2κ2 V (σ)
′′
1 V (σ)
|η(σ)| = 2 ≪ 1. (7.7)
κ V (σ)
By defining the energy density and pressure of σ as ρσ = σ̇ 2 /2 − V (σ) and Pσ = σ̇ 2 /2 + V (σ), these conditions imply
that the gravitational field equations in the flat FLRW space-time are given by 3H 2 /κ2 = −V (σ), 3H σ̇ ≃ −V ′ (σ), and
that ä(t) > 0, and hence guarantee a sufficiently long time inflation. In our case, since V (σdS ) 6= 0, these two conditions
are well satisfied around the de Sitter solution. Thus, since σ̇ ≃ 0, we find HdS = RdS / [12(1 + α)] = R̃dS /12.
Appendix B: Asymptotically phantom or quintessence modified gravity
In general, realistic models of modified gravity are similar to GR with the cosmological constant, i.e., the dark
energy fluid with the EoS parameter ωDE = −1 and the de Sitter universe as the final scenario for the cosmological
evolution. Since in principle quintessence/phantom-dark energy phases are not excluded by observations, it may be
of some interest to try to reconstruct an F (R) gravity theory where the quintessence or phantom dark energy (with a
constant ωDE ) emerges. The big difficulty is due to the fact that in the dark energy density ρDE and pressure PDE of
modified gravity, the effective gravitational terms appear. In this appendix, we reconstruct the form of F (R) gravity
which resembles to a fluid with ωDE being very close but not equal to −1.
If the energy density of a quintessence/phantom fluid is given by
ρ = ρ0 (z + 1)3(1+ω) , (7.8)
33
where ω is the EoS parameter, the Hubble parameter reads

r r
κ2 κ 2 ρ0 1p 2
H(z) = ρ≃ + 3κ ρ0 (1 + ω) log[z + 1] . (7.9)
3 3 2
Here, we have taken into account that ω is very close to −1. We can write R as a function of the redshift as
1 2
R(z) = κ ρ0 [2 + 3(1 + ω) log(z + 1)] [1 − 3ω + 6(1 + ω) log(z + 1)] . (7.10)
2
In addition, from Eq. (2.7), in vacuum we find
(" −1 # " −1 #
1 dR(z) dF (z) dR(z) dF (z)
ρeff ≡ ρDE = 2 R(z) − F (z) − 6H 2 (z) −1
2κ dz dz dz dz
" −1 −2 2 #)
d2 R(z)

2 dR(z) dF (z) dR(z) d F (z)
+ 6H (z)(z + 1) + . (7.11)
dz dz 2 dz dz dz 2
Here, F (R) model is expressed as a function of the redshift F (z). By equating ρeff to ρ of Eq. (7.8), we can find the
F (R) model realizing this cosmology. For |ω − 1| ≪ 0, the solution of Eq. (7.11) is given by
6κ2 ρ0 [11 + (34 − 9ω)ω]

F (z) ≃ − 6κ2 ρ0 (1 + ω) log(z + 1) , (7.12)
(5 − 3ω)2
From Eq. (7.10), we have

( p )
ρ0 κ2 (5 − 3ω) ± (1 + ω) ρ0 κ2 [16R + 9ρ0 κ2 (1 + ω)2 ]
z = −1 + exp , (7.13)
12ρ0 κ2 (1 + ω)
where the plus sign corresponds to the quintessence solution, whereas the minus sign does to the phantom one. We
can now write the modified gravity model as a function of the Ricci scalar as
p
ρ0 κ2 (257 + 183ω + 27ω 2 − 27ω 3 ) ρ0 κ2 [16R + 9ρ0 κ2 (1 + ω)2 ]
F (R) = ± , (7.14)
2(5 − 3ω)2 2
where ρ0 is a free parameter of the theory, and ω is the EoS parameter of dark energy coming from the modification of
gravity and equivalent to ωDE . In this way, we have reconstructed the form of F (R) gravity that gives the quintessence
or phantom fluid solution in the empty universe. Remind that this reconstruction is valid for ωDE close to −1.
[1] S. Perlmutter et al. [SNCP Collaboration], Astrophys. J. 517, 565 (1999) [arXiv:astro-ph/9812133]; A. G. Riess et al.
[Supernova Search Team Collaboration], Astron. J. 116, 1009 (1998) [arXiv:astro-ph/9805201].
[2] M. Tegmark et al. [SDSS Collaboration], Phys. Rev. D 69, 103501 (2004) [arXiv:astro-ph/0310723]; U. Seljak et al. [SDSS
Collaboration], Phys. Rev. D 71, 103515 (2005) [arXiv:astro-ph/0407372].
[3] D. J. Eisenstein et al. [SDSS Collaboration], Astrophys. J. 633, 560 (2005) [arXiv:astro-ph/0501171].
[4] D. N. Spergel et al. [WMAP Collaboration], Astrophys. J. Suppl. 148, 175 (2003) [arXiv:astro-ph/0302209]; Astrophys. J.
Suppl. 170, 377 (2007) [arXiv:astro-ph/0603449].
[5] E. Komatsu et al. [WMAP Collaboration], Astrophys. J. Suppl. 180, 330 (2009) [arXiv:0803.0547 [astro-ph]].
[6] E. Komatsu et al. [WMAP Collaboration], Astrophys. J. Suppl. 192, 18 (2011) [arXiv:1001.4538 [astro-ph.CO]].
[7] B. Jain and A. Taylor, Phys. Rev. Lett. 91, 141302 (2003) [arXiv:astro-ph/0306046].
[8] M. Li, X. D. Li, S. Wang and Y. Wang, Commun. Theor. Phys. 56, 525 (2011) [arXiv:1103.5870 [astro-ph.CO]].
[9] M. Kunz, arXiv:1204.5482 [astro-ph.CO].
[10] K. Bamba, S. Capozziello, S. Nojiri and S. D. Odintsov, Astrophys. Space Sci. 342, 155 (2012) [arXiv:1205.3421 [gr-qc]].
[11] S. Nojiri and S. D. Odintsov, Phys. Rept. 505, 59 (2011) [arXiv:1011.0544 [gr-qc]]; eConf C0602061, 06 (2006) [Int. J.
Geom. Meth. Mod. Phys. 4, 115 (2007)] [arXiv:hep-th/0601213].
[12] S. Capozziello and V. Faraoni, Beyond Einstein Gravity: A Survey of Gravitational Theories for Cosmology and Astro-
physics (Springer, Berlin, 2010).
[13] T. Clifton, P. G. Ferreira, A. Padilla and C. Skordis, Phys. Rept. 513, 1 (2012) [arXiv:1106.2476 [astro-ph.CO]].
[14] S. Capozziello and M. De Laurentis, Phys. Rept. 509, 167 (2011) [arXiv:1108.6266 [gr-qc]].
34
[15] S. Capozziello, M. De Laurentis and S. D. Odintsov, arXiv:1206.4842 [gr-qc].

[16] K. Bamba, C. Q. Geng and C. C. Lee, JCAP 1008, 021 (2010) [arXiv:1005.4574 [astro-ph.CO]].
[17] S. Nojiri and S. D. Odintsov, Phys. Rev. D 68, 123512 (2003) [hep-th/0307288].
[18] A. D. Dolgov and M. Kawasaki, Phys. Lett. B 573, 1 (2003) [astro-ph/0307285].
[19] V. Faraoni, Phys. Rev. D 74, 104017 (2006) [arXiv:astro-ph/0610734].
[20] Y. S. Song, W. Hu and I. Sawicki, Phys. Rev. D 75, 044004 (2007) [arXiv:astro-ph/0610532].
[21] V. Muller, H. J. Schmidt and A. A. Starobinsky, Phys. Lett. B 202, 198 (1988); V. Faraoni and S. Nadeau, Phys. Rev. D
72, 124005 (2005) [arXiv:gr-qc/0511094].
[22] L. Amendola, R. Gannouji, D. Polarski and S. Tsujikawa, Phys. Rev. D 75, 083504 (2007) [arXiv:gr-qc/0612180].
[23] T. Chiba, Phys. Lett. B 575, 1 (2003) [arXiv:astro-ph/0307338].
[24] T. Chiba, T. L. Smith and A. L. Erickcek, Phys. Rev. D 75, 124014 (2007) [arXiv:astro-ph/0611867]; G. J. Olmo, Phys.
Rev. D 75, 023511 (2007) [gr-qc/0612047].
[25] S. Capozziello, S. Carloni and A. Troisi, Recent Res. Dev. Astron. Astrophys. 1, 625 (2003) [astro-ph/0303041].
[26] S. M. Carroll, V. Duvvuri, M. Trodden and M. S. Turner, Phys. Rev. D 70, 043528 (2004) [astro-ph/0306438].
[27] S. Capozziello, S. Nojiri and S. D. Odintsov, Phys. Lett. B634, 93 (2006), hep-th/0512118; S. Capozziello, S. Nojiri,
S. D. Odintsov and A. Troisi, Phys. Lett. B639, 135 (2006), astro-ph/0604431.
[28] W. Hu and I. Sawicki, Phys. Rev. D 76, 064004 (2007) [arXiv:0705.1158 [astro-ph]].
[29] G. Cognola, E. Elizalde, S. Nojiri, S. D. Odintsov, L. Sebastiani and S. Zerbini, Phys. Rev. D 77, 046009 (2008)
[arXiv:0712.4017 [hep-th]].
[30] E. V. Linder, Phys. Rev. D 80, 123528 (2009) [arXiv:0905.2962 [astro-ph.CO]].
[31] S. A. Appleby and R. A. Battye, Phys. Lett. B 654, 7 (2007) [arXiv:0705.3199 [astro-ph]].
[32] A. A. Starobinsky, JETP Lett. 86, 157 (2007) [arXiv:0706.2041 [astro-ph]].
[33] S. Tsujikawa, Phys. Rev. D 77, 023507 (2008) [arXiv:0709.1391 [astro-ph]].
[34] K. Bamba, C. -Q. Geng and C. -C. Lee, JCAP 1011, 001 (2010) [arXiv:1007.0482 [astro-ph.CO]].
[35] L. Yang, C. C. Lee and C. Q. Geng, JCAP 1108, 029 (2011) [arXiv:1106.5582 [astro-ph.CO]].
[36] L. Yang, C. C. Lee, L. W. Luo and C. Q. Geng, Phys. Rev. D 82, 103515 (2010) [arXiv:1010.2058 [astro-ph.CO]].
[37] J. Vainio and I. Vilja, arXiv:1205.5393 [astro-ph.CO]; R. Gannouji, M. Sami and I. Thongkool, arXiv:1206.3395 [hep-th];
C. -C. Lee, C. -Q. Geng and L. Yang, arXiv:1201.4546 [astro-ph.CO]; K. Henttunen and I. Vilja, arXiv:1110.6711 [astro-
ph.CO]; K. Nozari and F. Kiani, Phys. Lett. B 703, 395 (2011) [arXiv:1108.4519 [gr-qc]]; S. A.Appleby and J. Weller, JCAP
1012 (2010) 006 [arXiv:1008.2693 [astro-ph.CO]]; B. Jain and J. Khoury, Annals Phys. 325, 1479 (2010) [arXiv:1004.3294
[astro-ph.CO]]; A. Ali, R. Gannouji, M. Sami and A. A. Sen, Phys. Rev. D 81, 104029 (2010) [arXiv:1001.5384 [astro-
ph.CO]]; E. Elizalde and D. Saez-Gomez, Phys. Rev. D 80, 044030 (2009) [arXiv:0903.2732 [hep-th]]; N. Deruelle, M. Sasaki
and Y. Sendouda, Phys. Rev. D 77, 124024 (2008) [arXiv:0803.2742 [gr-qc]]; S. Nojiri and S. D. Odintsov, Phys. Rev. D
77, 026007 (2008) [arXiv:0710.1738 [hep-th]]; J. D. Evans, L. M. H. Hall and P. Caillol, Phys. Rev. D 77, 083514 (2008)
[arXiv:0711.3695 [astro-ph]].
[38] E. Elizalde, S. D. Odintsov, L. Sebastiani and S. Zerbini, Eur. Phys. J. C 72, 1843 (2012) [arXiv:1108.6184 [gr-qc]].
[39] E. Elizalde, S. Nojiri, S. D. Odintsov, L. Sebastiani and S. Zerbini, Phys. Rev. D 83, 086006 (2011) [arXiv:1012.2280
[hep-th]].
[40] H. W. Lee, K. Y. Kim and Y. S. Myung, Eur. Phys. J. C 71, 1748 (2011) [arXiv:1106.2865 [hep-th]].
[41] T. Chiba and T. Nakamura, Prog. Theor. Phys. 100, 1077 (1998) [astro-ph/9808022].
[42] V. Sahni, T. D. Saini, A. A. Starobinsky and U. Alam, JETP Lett. 77, 201 (2003) [Pisma Zh. Eksp. Teor. Fiz. 77, 249
(2003)] [arXiv:astro-ph/0201498].
[43] C. -C. Lee, C. -Q. Geng and L. Yang, Prog. Theor. Phys. 128, , 415 (2012) [arXiv:1201.4546 [astro-ph.CO]].
[44] A. Upadhye, W. Hu and J. Khoury, Phys. Rev. Lett. 109, 041301 (2012) [arXiv:1204.3906 [hep-ph]].
[45] S. Nojiri, S. D. Odintsov and S. Tsujikawa, Phys. Rev. D 71 (2005) 063004 [arXiv:hep-th/0501025].
[46] A. de la Cruz-Dombriz, A. Dobado, and A. L. Maroto, Phys. Rev. D 77, 123515 (2008); L. Pogosian, and A. Silvestri,
Phys. Rev. D 77, 023503 (2008); idem, Phys. Rev. D 81, 049901(E) (2010); H. Oyaizu, Phys. Rev. D 78, 123523 (2008);
H. Oyaizu, M. Lima, and W. Hu, Phys. Rev. D 78, 123524 (2008); F. Schmidt, M. Lima, H. Oyaizu, and W. Hu, Phys.
Rev. D 79, 083518 (2009); G. B. Zhao, B. Li, and K. Koyama, Phys. Rev. D 83, 044007 (2011); K. Koyama, A. Taruya,
and T. Hiramatsu, Phys. Rev. D 79, 123512 (2009); A. Taruya, T. Nishimichi, S. Saito, and T. Hiramatsu, Phys. Rev. D
80, 123503 (2009); F. Schmidt, Phys. Rev. D 78, 043002 (2008); G. B. Zhao, L. Pogosian, A. Silvestri, and J. Zylberberg,
Phys. Rev. D 79, 083513 (2009); F. Schmidt, A. Vikhlinin, and W. Hu, Phys. Rev. D 80, 083505 (2009); A. Borisov, and
B. Jain, Phys. Rev. D 79, 103506 (2009); S. Ferraro, F. Schmidt, W. Hu, Phys. Rev. D 83, 063503 (2011); K. W. Masui, F.
Schmidt,, U. L. Pen, and P. McDonald, Phys. Rev. D 81, 062001 (2011); T. Giannantonio, M. Martinelli, A. Silvestri, and
A. Melchiorri, JCAP 04, 030 (2010); K. Yamamoto, G. Nakamura, G. Hutsi, T. Narikawa, and T. Sato, Phys. Rev. D 81,
103517 (2010); E. Beynon, D. J. Bacon, and K. Koyama, Mon. Not. Roy. Astron. Soc. 403, 353 (2010); A. M. Nzioki, P.
K. S. Dunsby, R. Goswami, and S. Carloni, Phys. Rev. D 83, 024030 (2011); X. Fu, P. Wu, and H. Yu, Eur. Phys. J. C 68,
271 (2010); T. Narikawa, and K. Yamamoto, Phys. Rev. D 81, 043528 (2010); ibidem, Phys. Rev. D 81, 129903(E) (2010);
S. A. Thomas, S. A. Appleby, and J. Weller, JCAP 03, 036 (2011); L. Lombriser, F. Schmidt, T. Baldauf, R. Mandelbaum,
U. Seljak, and R. E. Smith, arXiv:1111.2020; S. Camera, A. Diaferio, and V. F. Cardone, JCAP 07, 016 (2011)
[47] E. V. Linder, Phys. Rev. D 72, 043529 (2005) [astro-ph/0507263].
[48] S. Nojiri and S. D. Odintsov, Phys. Lett. B 652, 343 (2007) [arXiv:0706.1378 [hep-th]].
[49] D. F. Mota and J. D. Barrow, Phys. Lett. B 581, 141 (2004) [arXiv:astro-ph/0306047].
[50] J. Khoury and A. Weltman, Phys. Rev. Lett. 93, 171104 (2004) [arXiv:astro-ph/0309300]; Phys. Rev. D 69, 044026 (2004)
35
[arXiv:astro-ph/0309411].
[51] H. Motohashi, A. A. Starobinsky and J. ’i. Yokoyama, JCAP 1106, 006 (2011) [arXiv:1101.0744 [astro-ph.CO]].
[52] U. Alam, V. Sahni and A. A. Starobinsky, JCAP 0406, 008 (2004) [arXiv:astro-ph/0403687]; S. Nesseris and
L. Perivolaropoulos, JCAP 0701, 018 (2007) [arXiv:astro-ph/0610092]; P. U. Wu and H. W. Yu, Phys. Lett. B 643,
315 (2006) [arXiv:astro-ph/0611507]; H. K. Jassal, J. S. Bagla and T. Padmanabhan, Mon. Not. Roy. Astron. Soc. 405,
2639 (2010) [arXiv:astro-ph/0601389].
[53] R. R. Caldwell, M. Kamionkowski and N. N. Weinberg, Phys. Rev. Lett. 91 (2003) 071301 [astro-ph/0302506].
[54] P. H. Frampton, K. J. Ludwick and R. J. Scherrer, Phys. Rev. D 84, 063003 (2011) [arXiv:1106.4996 [astro-ph.CO]].
[55] P. H. Frampton, K. J. Ludwick, S. ’i. Nojiri, S. D. Odintsov and R. J. Scherrer, Phys. Lett. B 708, 204 (2012)
[arXiv:1108.0067 [hep-th]].
[56] L. G. Jaime, L. Patino and M. Salgado, arXiv:1206.1642 [gr-qc].
[57] K. Bamba, S. Nojiri and S. D. Odintsov, JCAP 0810, 045 (2008) [arXiv:0807.2575 [hep-th]];
S. Nojiri and S. D. Odintsov, Phys. Rev. D 78, 046006 (2008) [arXiv:0804.3519 [hep-th]]; AIP Conf. Proc. 1241, 1094
(2010) [arXiv:0910.1464 [hep-th]]; K. Bamba, S. D. Odintsov, L. Sebastiani and S. Zerbini, Eur. Phys. J. C 67, 295 (2010)
[arXiv:0911.4390 [hep-th]].
[58] O. Gorbunova and L. Sebastiani, Gen. Rel. Grav. 42, 2873 (2010) [arXiv:1004.1505 [gr-qc]].
[59] S. Tsujikawa, Phys. Rev. D 76, 023514 (2007) [arXiv:0705.1032 [astro-ph]]; S. Tsujikawa, K. Uddin and R. Tavakol, Phys.
Rev. D 77, 043007 (2008) [arXiv:0712.0082 [astro-ph]]; S. Tsujikawa, R. Gannouji, B. Moraes and D. Polarski, Phys. Rev.
D 80, 084044 (2009) [arXiv:0908.2669 [astro-ph.CO]].
[60] K. Bamba, J. Matsumoto and S. Nojiri, Phys. Rev. D 85, 084026 (2012) [arXiv:1109.1308 [hep-th]]; J. Matsumoto, Phys.
Rev. D 83, 124040 (2011) [arXiv:1105.1419 [astro-ph.CO]].
[61] G. Esposito-Farese and D. Polarski, Phys. Rev. D 63 (2001) 063504 [arXiv:gr-qc/0009034].
[62] V. F. Cardone, S. Camera and A. Diaferio, JCAP 1202 (2012) 030 [arXiv:1201.3272 [astro-ph]].
[63] A. Lewis, A. Challinor and A. Lasenby, Astrophys. J. 538, 473 (2000) [astro-ph/9911177].
[64] http://camb.info/.
[65] N. Kaiser, Mon. Not. Roy. Astron. Soc. 227 (1987) 1.
[66] A. J. S. Hamilton, astro-ph/9708102.
[67] P. J. E. Peebles, Astrophys. J. 284, 439 (1984); O. Lahav, P. B. Lilje, J. R. Primack, M. J. Rees, MNRAS 251, 128 (1991).
[68] R. Gannouji, B. Moraes and D. Polarski, JCAP 0902 (2009) 034 [arXiv:0809.3374 [astro-ph]].
[69] D. Polarski and R. Gannouji, Phys. Lett. B 660 (2008) 439 [arXiv:0710.1510 [astro-ph]].
[70] A. B. Belloso, J. Garcia-Bellido and D. Sapone, JCAP 1110 (2011) 010 [arXiv:1105.4825 [astro-ph.CO]].
[71] H. Motohashi and A. Nishizawa, Phys. Rev. D 86, 083514 (2012) [arXiv:1204.1472 [astro-ph.CO]].
[72] H. Kodama and M. Sasaki, Prog. Theor. Phys. Suppl. 78, 1 (1984).
[73] V. F. Mukhanov, H. A. Feldman and R. H. Brandenberger, Phys. Rept. 215, 203 (1992).
[74] http://www.sciops.esa.int/index.php?project=PLANCK.
[75] http://www.rssd.esa.int/SA/PLANCK/docs/Bluebook-ESA-SCI(2005)1 V2.pdf.
[76] http://quiet.uchicago.edu/index.php.
[77] D. Samtleben and f. t. Q. Collaboration, Nuovo Cim. 122B, 1353 (2007) [arXiv:0802.2657 [astro-ph]].
[78] http://www.b-pol.org/index.php.
[79] http://cmbpol.kek.jp/litebird/.
[80] K. I. Maeda, Phys. Rev. D 39, 3159 (1989); Y. Fujii and K. Maeda, The Scalar-Tensor Theory of Gravitation (Cambridge
University Press, Cambridge, United Kingdom, 2003).
[81] L. Amendola, M. Kunz and D. Sapone, JCAP 0804, 013 (2008) [arXiv:0704.2421 [astro-ph]].

Dark Energy Oscillations in The Matter Dominated Era

Uploaded by

Copyright:

Available Formats

Dark Energy Oscillations in The Matter Dominated Era

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

Dark Energy Oscillations in The Matter Dominated Era

Uploaded by

Copyright:

Available Formats

Cosmic history of viable exponential gravity: Equation of state oscillations and

growth index from inflation to dark energy era

PACS numbers: 04.50.Kd, 95.36.+x, 98.80.-k

∗ E-mail address: [email protected]

d2 yH (z) dyH (z)

Furthermore, the Ricci scalar is expressed as

A. Realistic F (R) gravity models

m̃2 c1 (R/m̃2 )n m̃2 c1 m̃2 c1 /c2

B. Dark energy oscillations in the matter dominated era

in case of exponential model in Eq. (3.2) and

(a) (b) (c)

(d) (e) (f)

C. Proposal of a correction term

IV. COSMOLOGICAL CONSTRAINTS AND FUTURE EVOLUTION

(a) (b) (c)

(d) (e) (f)

A. Cosmological and local constraints

Newton law corrections and stability on a planet surface

The solution is given by

where G(m2 , |x|) is the correlation function which satisfies

− m2 G(m2 , |x|) = δ(x) .

− ∂t2 (δR) ∼ U (Rb )δR , (4.7)

U (Rb ) > 0 . (4.9)

B. Future universe evolution

Avoidance of the phantom crossing with (inhomogeneous) fluid

parameter ω as a function of ρ, i.e., ω = ω(ρ). The EoS is expressed as

C. Growth of the matter density perturbations: growth index

δ̈ + 2H δ̇ − 4πGeff (a, k)ρm δ = 0 (4.22)

with Ωm (z) = 8πGρ

We consider the ansatz for the growth index given by

1.0 0.9 0.9

(a) (b) (c)

1.0 0.9 0.9

(d) (e) (f)

Log kHMpc L-1

1.0 0.9 0.9

(a) (b) (c)

1.0 0.9 0.9

(d) (e) (f)

Next, we examine the following ansatz for the growth index:

0.7 0.7 0.7

(a) (b) (c)

(d) (e) (f)

V. UNIFIED MODELS FOR EARLY AND LATE-TIME COSMIC ACCELERATION

A. Inflation in exponential model (Model I)

We note that if n > 1 and α > 1, when R ≪ Ri (∼ R̃i ), we obtain

must be positive, that is,

Instability and number of e-folds during inflation

α γ̄(α − 2) > 1 , (5.14)

2 + 1/γ̄ > α > 2 . (5.15)

Thus, we may evaluate the characteristic number of e-folds during inflation

y1 (zi ) = C0 (zi + 1)x , (5.17)

VI. ANALYSIS OF INFLATION

(F ′ (R) − 1)2χ̃(z + 1)4 + (F (R) − R)/M 2

Moreover, the Ricci scalar is expressed as

(a) (b) (c)

N ≃ 107 (for α = 5/2) ,

In order to solve Eq. (6.3) numerically, we use the initial conditions