JETP Lett. 88, 289 (2008) arXiv:0807.

3896v4 [gr-qc]

f (R) Cosmology from q–Theory

#1) ∗+2)
F.R. Klinkhamer and G.E. Volovik
#Institute for Theoretical Physics, University of Karlsruhe (TH), 76128 Karlsruhe, Germany
∗ Low Temperature Laboratory, Helsinki University of Technology, P.O.Box 5100, FIN-02015, HUT, Finland
+ Landau Institute for Theoretical Physics RAS, Kosygina 2, 119334 Moscow, Russia

Submitted 25 July 2008

From a macroscopic theory of the quantum vacuum in terms of conserved relativistic charges (generically
arXiv:0807.3896v4 [gr-qc] 8 Feb 2009

denoted by q (a) with label a), we have obtained, in the low-energy limit, a particular type of f (R) model
relevant to cosmology. The macroscopic quantum-vacuum theory allows us to distinguish between different
phenomenological f (R) models on physical grounds.

PACS:

1. INTRODUCTION The thermodynamic approach [5] assumes that the

quantum vacuum is a stable self-sustained equilibrium
The development of high-energy physics and cosmol- state, which is described by compressibility and other
ogy over the last years has led to the realization that, characteristics of the response to external perturbations.
most likely, Einstein’s theory of gravity needs to be In this approach, we have found that the vacuum energy
modified. At the moment, we are only able to calculate density appears in two forms.
higher-order curvature corrections to the Einstein action
First, there is the microscopic vacuum energy den-
coming from scalar, spinor, and vector quantum fields
sity which is characterized by an ultraviolet energy
propagating over a fixed classical background (see, e.g., 4
scale EUV , so that ǫ(q) ∼ EUV . Most likely, EUV
Refs. [1, 2]). The full theory of the modified-Einstein ac-
corresponds to the standard Planck energy scale de-
tion requires the underlying microscopic theory – what
fined in terms
p of Newton’s gravitational constant GN ,
is usually called quantum gravity theory. Since the latter 5 28
EPlanck ≡ ~ c /GN ≈ 1.2×10 eV. But, here, we pre-
theory is not yet established, phenomenological models
fer to stay as general as possible and keep EUV distinct
are needed. Among these are the so-called f (R) models
from EPlanck .
(see, e.g., Refs. [2, 3] for two recent reviews) which have
Second, there is the macroscopic vacuum energy
ad hoc powers of the curvature invariants added to the
density which is determined by a particular thermo-
linear term of the Einstein–Hilbert action [4].
dynamic quantity, e ǫvac (q) ≡ ǫ − q dǫ/dq, and it is this
We have proposed another approach to modified
type of energy density which contributes to the effec-
Einstein gravity, which is based on the treatment of
tive gravitational field equations at low energies. For
the Lorentz-invariant quantum vacuum as an extended
a self-sustained vacuum in full thermodynamic equilib-
self-sustained system characterized by a conserved rela-
rium and in the absence of matter, the effective (coarse-
tivistic charge q [5, 6]. Here, q is a microscopic variable
grained) vacuum energy density e ǫvac (q) is automatically
describing the physics of the deep (ultraviolet) vacuum,
nullified (without fine tuning) by the spontaneous ad-
but its thermodynamics and dynamics are described by
justment of the vacuum variable q to its equilibrium
macroscopic equations, because q is a conserved quan-
value q0 , so that eǫvac (q0 ) = 0. This implies that the
tity. This quantity q is similar to the particle density
effective cosmological constant Λ of a perfect quantum
in liquids, which describes a microscopic quantity – the
vacuum is strictly zero, which is consistent with the re-
density of atoms – but obeys the macroscopic equations
quirement of Lorentz invariance for zero external pres-
of hydrodynamics, because of particle-number conser-
sure.
vation. Different from known liquids, the quantum vac-
uum is Lorentz invariant. The quantity q must, there- The dynamic approach [6] demonstrates how, in
fore, be Lorentz invariant. This treatment has allowed a flat Friedmann–Robertson–Walker universe, the vac-
us to discuss both the thermodynamics and the dynam- uum energy density e ǫvac (effective cosmological “con-
ics of a Lorentz-invariant quantum vacuum. stant”) relaxes from its natural Planck-scale value at
early Planckian times to a naturally small value at late
2) e-mail: frans.klinkhamer@physik.uni-karlsruhe.de times.
2) e-mail: volovik@boojum.hut.fi In this Letter, we show that the macroscopic theory

1
2 Klinkhamer, Volovik

of the quantum vacuum, when applied to cosmology, with effective potential

gives rise to a specific class of f (R) models of modified n
X ∂ǫ
gravity. ǫ(F (a) , φ) ≡ ǫ(F (a) , φ) −
e F (b) (5)
∂F (b)
b=1

and scalar-field energy-momentum tensor

2. GRAVITY WITH F –FIELDS AND
M 1
LOW-ENERGY MATTER Tµν = ∂µ φ ∂ν φ − gµν ∂ρ φ ∂ ρ φ . (6)
2
We consider the general case of several conserved Using F (a)µνρσ as given by (1), the Maxwell equa-
microscopic variables q (a) , for a = 1, . . . , n, and tions (3a) can be rewritten in the following form:
corresponding chemical potentials µ(a) [7, 8]. As in  
∂ǫ ∂K
Refs. [5, 6], each variable q (a) can be represented by ∂µ +R = 0. (7)
a four-form field F (a) : ∂F (a) ∂F (a)
The solution of these 4n equations is
1 (a) (a)
(F (a) )2 ≡ − F F (a)µνρσ , (a)
Fµνρσ ≡ ∇[µ Aνρσ] . ∂ǫ ∂K
24 µνρσ +R = µ(a) , (8)
(1) ∂F (a) ∂F (a)
The action of the four-form fields F (a) (x), the matter
where the µ(a) are n integration constants. Eliminating
field φ(x), and the gravitational field gµν (x) is given by
∂K/∂F (a) from (4) and (8), one finds for the generalized
Z Einstein equations
√
S[A(a) , g, φ] = − d4 x −g    
 R4
 −2K Rµν − 21 Rgµν − 2 ∇µ ∇ν − gµν  K
1
× K(F (a) ) R + ǫ(F (a) , φ) + ∂µ φ ∂ µ φ . (2)  n
X 
2 + ǫ− M
µ(a) F (a) gµν = Tµν . (9)
a=1
For simplicity, low-energy matter is represented by a
single real scalar field φ. The generalized potential Equations (8) and (9) can also be obtained if we use,
ǫ(F (a) , φ) includes self-interactions and interactions be- instead of the original action, the following effective ac-
tween all fields F (a) and φ, but contains no derivatives tion:
Z
of these fields and explicit factors of the metric field (a) (a) √
Seff [A , µ , g, φ] = − d4 x −g
gµν or its inverse. The gravitational coupling parame- R4
n
!
ter K is determined by ultraviolet physics and, there- X 1
(a) (a) µ
fore, depends on the microscopic vacuum variables F (a) . × KR+ǫ− µ F + ∂µ φ ∂ φ . (10)
a=1
2
Here, and in the following, we adopt the conventions
of Ref. [4], in particular, those for the Riemann tensor The µ(a) F (a) terms in this action do not contribute to
and the metric signature (− + ++). Natural units with the equations of motion (3a), because they are total
~ = c = 1 are used throughout. derivatives,
The generalized Maxwell and Klein–Gordon equa- Z p Z
µ(a) κλµν (a)
tions from action (2) read d4 x |g| µ(a) F (a) = − e d4 x Fκλµν .
R4 24 R4
   (11)
√ F (a)µνρσ ∂ǫ ∂K (a)
∇µ −g +R = 0 , (3a) The constant µ is seen to play the role of a Lagrange
F (a) ∂F (a) ∂F (a)
multiplier related to the conservation of the vacuum
∂ǫ “charge” F (a) .
φ − = 0 , (3b)
∂φ Instead of the large microscopic energy density
where  denotes the invariant d’Alembertian operator ǫ(F (a) , φ) in the original action (2), a potentially smaller
and the partial derivatives ∂/∂F (a) and ∂/∂φ stand for macroscopic vacuum energy density enters the effective
pointwise differentiation. The variation of (2) over the action (10), namely,
metric gµν gives the generalized Einstein equations: n
X
ρV ≡ ǫ(F (a) , φ) − µ(a) F (a) . (12)
  Xn a=1
1 ∂K
2K Rµν − R gµν + R gµν F (a) (a) Precisely this macroscopic vacuum energy density grav-
2 a=1
∂F
  itates and determines the cosmological term in the grav-
ǫ(F (a) , φ) gµν + Tµν
+2 ∇µ ∇ν − gµν  K − e M
= 0, (4) itational field equations (9).
 f (R) cosmology from q–theory 3

3. EQUILIBRIUM VACUA AND STABILITY which can be local or global. In particular, the vacuum
CONDITIONS compressibility introduced in Ref. [5] must be positive:
 −1
The main goal of our approach is to describe the n
X 2
∂ ǫ
thermodynamics of the equilibrium vacuum [5] and to χ0 ≡  F (a) F (b)  > 0.
consider cosmology as the dynamics of relaxation to- a,b=1
∂F (a) ∂F (b)
(a)
F (a) =F0 , φ=φ0
wards an equilibrium state [6, 9, 10]. That is why, in (14)
what follows, we assume that the universe is close to Furthermore, the effective Newton’s constant must be
equilibrium and that all its parameters, including the positive in an equilibrium vacuum:
fields F (a) and the chemical potentials µ(a) , are close to
their equilibrium values. A static homogeneous equilib- 1
GN ≡ (a)
>0, (15)
rium vacuum, in the absence of thermal
matter, corre- 16πK(F0 )
(a)
sponds to a stationary point F0 , φ0 of Eqs. (3b), (8),
and (9) for Rµν = Tµν M
= 0: in order to have a physically consistent description of
an attractive gravitational force. More specifically, neg-
n
X ative K gives the wrong sign of the kinetic term for the
∂ǫ ∂ǫ
= 0, = µ(a) , ǫ− µ(a) F (a) = 0 , (13) graviton (which becomes ghostlike) and the quantum
∂φ ∂F (a) a=1 vacuum is unstable. As q–theory is based on a stable
self-sustaining vacuum, having K > 0 is a necessary
where the last equation demonstrates that the vacuum
condition for stability of the vacuum, together with the
energy (12) is zero in an equilibrium vacuum, ρV |eq = 0.
compressibility condition (14).
One can see the difference between the conventional
matter field φ and the conserved vacuum fields F (a) ,
as only the fields F (a) provide the integration constants 4. f (R) MODEL FROM q–THEORY
µ(a) which arise dynamically from the solution (8) of the
generalized Maxwell equations (7). (The field equations Modified-gravity f (R) models appeared already in
of generic matter fields do not give rise to such integra- the 1960s (see, e.g., Refs. [12, 13]) and were used to con-
tion constants.) These integration constants µ(a) play struct an inflationary model of the early universe in the
the role of chemical potentials in thermodynamics and 1980s [14]. More recently, these f (R) models have re-
are thermodynamically conjugate to the density of the ceived renewed attention as a way to explain the inferred
conserved quantities F (a) . With appropriate nonzero cosmic “dark energy” by attributing it to a modification
chemical potentials, the large vacuum energy ǫ(F (a) , φ) of Einstein gravity (see, e.g., Refs. [2, 3, 15, 16] and ref-
is reduced to ρV = 0 in a homogeneous equilibrium erences therein). These models are, in fact, purely phe-
vacuum state according to (13). Specifically, two large nomenological models, which, in their simplest form, re-
Pn
quantities, ǫ(F (a) , φ) and a=1 µ(a) F (a) , each of order place the linear function of the Ricci scalar R from the
4
EUV , cancel each other due to the self-tuning mecha- Einstein–Hilbert action term by a more general function
nism [5]. This is the main property of a self-sustained f (R). This function f (R) can, in principle, be adjusted
vacuum. to fit the astronomical observations and to produce a
For the case of a single vacuum variable F , the chem- viable cosmological model.
ical potential µ in equilibrium is completely fixed by the To obtain f (R) from q–theory [6], one can express
constraint ρV = 0. But, for the case of several vari- F (a) in terms of R, φ, and µ(a) by use of the equa-
ables F (a) , there are n − 1 degrees of freedom, since the tion system (8) and substitute the resulting functions
equation ρV = 0 gives only a single constraint on the n F (a) (R) into (9). This can be done in a general way
chemical potentials µ(a) . This allows for the existence of (see Ref. [3]); but, since we consider the relaxation to an
many different equilibrium vacua and also for the coex- equilibrium vacuum, we are only interested in the sim-
istence of several vacua [7, 8]. This last observation may pler situation of a system already close to equilibrium.
give microscopic support to the multiple point principle In addition, we will restrict ourselves to a single F -field
which postulates the existence of a number of phases (the generalization to n fields is straightforward) and we
with the same energy density (see, e.g., Ref. [11] and also omit the explicit matter φ-field, keeping only the
M
references therein). general matter energy-momentum tensor Tµν .
(a)
The stationary point (F0 , φ0 ) of the thermody- For a single F -field, (8) gives
P
namic potential W ≡ ǫ(F (a) , φ) − a µ(a) F (a) [i.e., ∂ǫ(F ) ∂K(F )
the solution of (13)] should correspond to a minimum, − µ = −R . (16)
∂F ∂F
4 Klinkhamer, Volovik

Close to the equilibrium state determined by (13), one modified-gravity models has also been noted in App. A
can expand the microscopic variables as follows: of Ref. [17].
In q–theory, the cosmological constant λ in (21b)
F = F0 + δF , µ = µ0 + δµ. (17) is induced by the deviation of the chemical potential
µ from its equilibrium value µ0 : λ ≡ 8πGN ρV =
Expressing δF in terms of R and δµ and excluding δF −8πGN F0 δµ. For the case of n charges F (a) , the cos-
from the Einstein equations (9), one obtains:
mological constant λ is given by
   n
−2K Rµν − 12 Rgµν + 2 χ e ∇µ ∇ν R − gµν  R X (a)
λ ≡ 8πGN ρV = −8πGN F0 δµ(a) . (23)
1 
a=1
+ R2 gµν − F0 δµ gµν = Tµν M
, (18)
4 For the general case of a quantum vacuum described by
in terms of the new dimensionless parameter conserved vacuum variables q (a) , one has:
Xn
!2 λ
∂K Λ≡ ≡ ρV = − q (a) δµ(a) . (24)
χ
e ≡ χ0 F . (19) 8πGN a=1
∂F
F =F0
This last equation also follows from the zero-
2
In (18), we have omitted the (δµ) term and kept only temperature Gibbs–Duhem relation [5] applied to
the leading term containing δµ. Expanding K in the the thermodynamic system characterized by several
first term of (18), K = (16πGN )−1 + δF ∂K/∂F , one conserved variables: µ(a) is the variable thermodynam-
obtains the following modified Einstein equations: ically conjugate to the conserved variable q (a) .
Note that f (R) phenomenology adds a pair of ther-
1   
modynamically conjugate variables, R and K. This fol-
− Rµν − 21 Rgµν + 2 χ
e ∇µ ∇ν R − gµν  R
8πGN lows from, e.g., the thermodynamic potential ǫ + K R −
1  P (a) (a)
M
− R2 gµν + R Rµν − F0 δµ gµν = Tµν , (20) µ F in (10). The corresponding thermodynamic
4 identities can be used to obtain, for example, the dimen-
where we have kept only the leading term F0 δµ and sionless parameter χ e in (21b), which can be interpreted
omitted terms R δµ and δµ2 . as −∂K/∂R evaluated at R = 0. One then has, at
The field equations (20) correspond to the following R = 0,
phenomenological model: ∂K X ∂K ∂F (a) X ∂K ∂K
Z   χ
e=− =− (a)
=
√ 1 ∂R a
∂F ∂R a
∂F (a) ∂µ(a)
Sphenom = d4 x −g fe(R) + LM , (21a)
R4 16πGN X ∂K ∂K ∂F (b)
=
fe(R) = −R + 16πGN χ e R2 − 2 λ , (21b) a,b
∂F (a) ∂F (b) ∂µ(a)
X ∂K ∂K  ∂2ǫ
−1
with LM the standard matter Lagrange density. The = , (25)
function fe(R) found belongs to the class of models ∂F (a) ∂F (b) ∂F (a) ∂F (b)
a,b
f (R) ∼ −R + R2 /(6M 2 ), where a bare cosmologi-
cal constant may or may not be added and M ≡ which gives, using definition (14), precisely (19) for the
(3 d2 f /dR2 )−1/2 is the scalaron mass (see Refs. [14, 15] case of a single charge F . However, this thermodynamic
and references therein; note that our sign convention for description is only applicable to f (R) phenomenology,
R is opposite to that in Ref. [15]). From q–theory, the since it does not take into account non-f (R) terms in the
scalaron mass square is given by action such as Rµν Rµν and Rµνρσ Rµνρσ which appear
due to quantum corrections [1, 2].


M 2 = 1 96πGN χ e , (22) For q–theory of the type considered in (2), there are
two stability conditions, namely, condition (15) for the
with χe defined by (19) for a single “charge” F and effective Newton’s constant GN and condition (14) for
by (25) below for multiple “charges” F (a) . At this the zero-temperature vacuum compressibility χ0 . These
moment, we have two parenthetical remarks. First, two conditions of q–theory correspond, respectively, to
the effective action for (20) does not contain Rµν Rµν the following stability conditions of f (R) models (see,
or Rµνρσ Rµνρσ terms in addition to the R2 term of e.g., Eq. (7) of Ref. [15]):
(21b). Second, a purely phenomenological connection  
 
between generalized-equation-of-state models and f (R) −f ′ (R)  > 0 , f ′′ (R)  > 0, (26)
R=0 R=0
 f (R) cosmology from q–theory 5

where the prime denotes differentiation with respect to As a result, the dimensionless parameter χ e in (19) can
R and the extra minus sign traces back to our conven- be expected to be of order unity. If the assumption of
tions, as mentioned already a few lines under (21b). a single fundamental energy scale holds true, one then
These conditions ensure the stability of the empty flat obtains that the mass parameter (22) of the induced
universe, which is a basic assumption for q–theory as f (R) = −R + R2 /(6M 2 ) model is of the order of the
it concerns the thermodynamics of the equilibrium vac- ultraviolet scale, M ∼ EUV . In this case, the R2 cor-
uum. In f (R) models, the second condition of (26), rection is of the order of the quantum R2 corrections to
i.e., the positive mass squared of the scalaron, M 2 > 0, the Einstein action [1, 2]. Moreover, as q–theory starts
implies the stability of the Minkowski vacuum, whereas from the assumption that the vacuum is a stable self-
for the negative mass squared, M 2 < 0, the Minkowski sustained medium, one has M 2 > 0, as follows explicitly
vacuum experiences the scalaron instability [15]. from (14), (19), and (21b).
For the case of M 2 > 0, the relaxation of the uni- All this agrees with the simplest f (R) ∼ −R +
verse to equilibrium is accompanied by oscillations of R R2 /M 2 model having a Planck-scale mass M [14], but
with frequency M (see Ref. [15] for the f (R) model and disagrees with certain other models which use more
Ref. [6] for the q–theory which also has δF oscillations). complicated phenomenological functions f (R). For the
In particular, the vacuum energy density in q–theory particular models suggested in Ref. [15], the functions
has been found to relax as follows [6]: f (R) expanded around R = 0 have negative M 2 (mak-
ing Minkowski spacetime unstable) and −M 2 of the or-
M2 der of λ now , where λ now = 8πGN ρV, now ≈ (10−33 eV)2
ρV ∼ sin2 M t , (27)
t2 is the present (positive) cosmological constant as de-
for cosmic time t ≫ 1/M . While the q field itself has an termined by observational cosmology [18]. These cos-
equation of state parameter w = −1, corresponding to mologically desirable f (R) models can, in principle,
vacuum energy density and a cosmological constant, the be obtained by using special choices of ǫ(F (a) , φ) and
decay of the vacuum energy density in (27) simulates the K(F (a) ). But such choices require a careful tuning of
evolution of cold dark matter with w = 0. The effective the parameters, which is unnatural from our point-of-
equation of state w = 0 is induced by the interaction of view on the properties of the self-sustained quantum
the q field with gravity [6]. vacuum.

5. COMPARISON WITH 6. CONCLUSION

PHENOMENOLOGICAL f (R) MODELS
In q–theory [5, 6] with chemical potentials µ(a) at
As explained above and, more briefly, in Ref. [6], F (a)
theory (or, more generally, q–theory) may give a mi- their equilibrium values µ0 , the vacuum energy den-
croscopic justification for the phenomenological f (R) sity ρV has been found to relax, according to (27), from
models used in theoretical cosmology and may allow its natural Planck-scale value at early times when the
for a choice between different classes of model functions system is far from equilibrium to a naturally small value
f (R) based on fundamental physics. Close to equi- at late times when the system is close to equilibrium.
librium, the effective f (R) model emerging from the This solves the main cosmological problem: the present
q–theory of quantum vacuum belongs to the class of cosmological constant is small compared to Planck-scale
f (R) ∼ −R + R2 /M 2 models. This rules out so-called values simply because the universe happens to be old
1/R models, i.e., models with f (R) ∼ −R + M 4/R (see, compared to Planck-scale times. The remaining prob-
e.g., Refs. [2, 3, 16] and references therein). lem is to understand why the cosmological constant does
Here, we have assumed that the physics of the vac- not completely relax to zero as t → ∞ or, in other words,
uum variable in q–theory [5, 6] is determined by a unique to determine the origin of the small residual part of the
microscopic energy scale EUV . This implies the follow- vacuum energy which remains (almost) constant during
ing orders of magnitude: the present epoch.
The suggested f (R) model in which the cosmolog-
(a) (a) 2
|F0 | ∼ |µ0 | ∼ EUV , (28a) ical constant appears just at the latest stage [15] does
(a) (a)
|F0 ∂K/∂F0 | ∼ G−1 2
∼ EUV , (28b) not follow from the macroscopic quantum-vacuum ap-
N
(a)
proach. That is why one needs to find another explana-
4
|ǫ(F0 , φ0 )| ∼ EUV , (28c) tion for the observed value of Λ. In order to produce a
−4
χ0 ∼ EUV . (28d) nonzero Λ, one must find, according to (23)–(24), pro-
6 Klinkhamer, Volovik

cesses which shift the chemical potentials µ(a) away from 6. F.R. Klinkhamer and G.E. Volovik, “Dynamic vacuum
(a)
the equilibrium mean-field values µ0 . variable and equilibrium approach in cosmology,” Phys.
There has been the suggestion to relate the present Rev. D 78, 063528 (2008), arXiv:0806.2805 [gr-qc].
value of Λ to quantum fluctuations of the vacuum en- 7. G.E. Volovik, “Coexistence of different vacua in the
ergy, for the case that the vacuum energy itself is nulli- effective quantum field theory and multiple point
fied (see, e.g., Ref. [19]). This approach could be helpful principle”, JETP Lett. 79, 101 (2004), arXiv:hep-
ph/0309144.
in q–theory, where the vacuum energy is necessarily zero
8. F.R. Klinkhamer and G.E. Volovik, “Coexisting vacua
in full equilibrium. The observed nonzero value of Λ
and effective gravity,” Phys. Lett. A 347, 8 (2005),
would then correspond to quantum or thermodynamic
arXiv:gr-qc/0503090.
fluctuations of the chemical potentials µ(a) compared to
(a) 9. C. Barcelo, “Cosmology as a search for overall equilib-
the equilibrium mean-field values µ0 . Still, there could
rium,” JETP Lett. 84, 635 (2007), arXiv:gr-qc/0611090.
be other processes which are able to shift µ(a) somewhat
(a) 10. F.R. Klinkhamer, “Equilibrium boundary conditions,
away from the mean-field values µ0 , one example be- dynamic vacuum energy, and the Big Bang,” Phys. Rev.
ing the backreaction of matter and another example be- D 78, 083533 (2008), arXiv:0803.0281 [gr-qc].
ing matter production by the rapid oscillations which 11. C.D. Froggatt, R. Nevzorov, H.B. Nielsen, and D.
accompany the relaxation of the vacuum energy (27). Thompson, “On the origin of Z(2) symmetry in the two–
It remains to be seen which of these processes, if any, Higgs doublet extension of the SM,” arXiv:0806.3190
is the dominant one for the inferred small but nonzero [hep-ph].
value of the vacuum energy density from observational 12. T.V. Ruzmaikina and A.A. Ruzmaikin, “Quadratic cor-
cosmology. rections to the Lagrangian gravitational field density
and singularity,” Zh. Eksp. Teor. Fiz. 57, 680 (1969);
Sov. Phys. JETP 30, 372 (1970).
ACKNOWLEDGMENTS 13. B.N. Breizman, V.Ts. Gurovich, and V.P. Sokolov,
“Possibility of setting up regular cosmological solu-
It is a pleasure to thank A.A. Starobinsky for helpful tions,” Zh. Eksp. Teor. Fiz. 59, 288 (1970); Sov. Phys.
discussions. GEV is supported in part by the Russian JETP 32, 155 (1971).
Foundation for Basic Research (grant 06–02–16002– 14. A.A. Starobinsky, “A new type of isotropic cosmologi-
a) and the Khalatnikov–Starobinsky leading scientific cal models without singularity,” Phys. Lett. B 91, 99
school (grant 4899.2008.2) (1980).
15. A.A. Starobinsky, “Disappearing cosmological con-
stant in f (R) gravity,” JETP Lett. 86, 157 (2007),
1. N.D. Birrell and P.C.W. Davies, Quantum Fields in arXiv:0706.2041 [astro-ph].
Curved Space (Cambridge University Press, Cambridge, 16. T. Kobayashi and K. Maeda, “Relativistic stars in f (R)
England, 1982). gravity, and absence thereof”, arXiv:0807.2503 [astro-
2. E.J. Copeland, M. Sami, and S. Tsujikawa “Dynamics ph].
of dark energy”. Int. J. Mod. Phys. D 15, 1753 (2006), 17. S. Nojiri and S.D. Odintsov, “Inhomogeneous equation
arXiv:hep-th/0603057. of state of the universe: Phantom era, future singular-
3. T.P. Sotiriou and V. Faraoni, “f (R) theories of gravity,” ity and crossing the phantom barrier,” Phys. Rev. D 72,
arXiv:0805.1726 [gr-qc]. 023003 (2005), arXiv:hep-th/0505215.
4. S. Weinberg, Gravitation and Cosmology (Wiley, New 18. E. Komatsu et al., “Five-year Wilkinson Microwave
York, 1972). Anisotropy Probe (WMAP) observations: Cosmologi-
5. F.R. Klinkhamer and G.E. Volovik, “Self-tuning vac- cal interpretation,” arXiv:0803.0547 [astro-ph].
uum variable and cosmological constant,” Phys. Rev. D 19. T. Padmanabhan, “Dark energy and its implications for
77, 085015 (2008), arXiv:0711.3170 [gr-qc]. gravity”, arXiv:0807.2356 [gr-qc]