Beruflich Dokumente
Kultur Dokumente
Natacha Leite
Center for Space Research, North-West University, Mafikeng 2735, South Africa
E-mail: natacha.leite@desy.de
Petar Pavlović
II. Institute for Theoretical Physics, Hamburg University, Luruper Chaussee 149,
22761 Hamburg, Germany
E-mail: petar.pavlovic@desy.de
1. Introduction
While various astrophysical models were developed to address this question [3, 4], the
potential existence of magnetic fields in the voids of the intergalactic medium seems to
suggest that magnetic fields could be produced cosmologically. Lower bounds on the
strength of extragalactic magnetic fields have been obtained from the non-observation
of γ-ray emission from distant TeV blazars that were expected to produce electromag-
netic cascades in the GeV range [5, 6, 7]. The obtained lower bounds on extragalactic
magnetic fields suggest that these fields are remnant from the early Universe since it is
unlikely that astrophysically originated fields at the galactic scale could have an impact
on the voids. This is the case since magnetic fields produced at the galactic scale needed
have an efficient transport mechanism to be able to account for the constraints derived
with the mentioned probing method for magnetic fields on the voids. The difficulty
in hypothesizing galactic outflows with a volume filling factor sufficient to redistribute
astrophysically produced magnetic fields into voids and filaments leads us to favour a
primordial/cosmological origin for magnetic fields. However, while observations have
been increasingly ruling out some models, it has not been possible to decisively ascer-
tain the magnetogenesis mechanism preferred by our Universe yet [8].
In order for magnetic fields to be produced, some turbulence inducing and out-of-
equilibrium process, such as a first order phase transition, was usually considered. In
the standard model, based on the determination of the Higgs boson mass, it appears
that quantum chromodynamics (QCD) and electroweak transitions are not first order
phase transitions but crossovers. There are numerous suggestions of extensions of the
standard model where these transitions remain first order, as well as various works on
a transition of this nature taking place at much higher energy scales, e.g. grand unified
theories models. Focusing ourselves on the standard model of particles – as will be as-
sumed in the remainder of this manuscript – the most natural choice for magnetogenesis
falls over inflation [9]. However, it is not straightforward to obtain magnetic fields of the
strength and correlation lengths of the presently observed fields. Conformal invariance
has to be broken in a conformally flat universe in order for the generated field strengths
to be sufficiently high [10]. Many approaches to break the conformal invariance of the
electromagnetic action during inflation are based on adding direct couplings to, for ex-
ample, curvature [11], torsion [12], scalars [13, 14, 15, 16, 17], pseudo-scalars [18, 19]
and charged scalar fields [20]. However, the generated fields are very sensitive to the
variation of the coupling constants such that parameters need to be fine-tuned [21].
The effects of backreaction on the background expansion from the produced scale in-
variant magnetic spectrum restrict the amplitude of the primordial seed magnetic fields,
constraining the viability of such models [22, 17]. There are models that address this
backreaction problem, for instance, by by using a minimal Lorentz-violating extension
of the standard model [23]. Additionally, the need to solve the strong coupling problem
at the beginning of inflationary expansion [24] usually implies more complexity, such as
an extradimensional scale factor coupling [25].
Magnetogenesis in Cyclical Universe 3
relativity. As the problem of dark energy seems to suggest, these quantum effects could
become significant already on energies much lower than the Planck scale. The future
theory of quantum gravity should be expected to wipe out the singularities existing in
standard general relativity, by virtue of the high-energy modifications that it will in-
troduce. While this more complete theory is still unknown, the approach which seems
the most justified to us is to consider the potential corrections to general relativity ef-
fectively, by investigating the possible modifications of the theory, which are in accord
with the current empirical data and with already confirmed physical principles. With
these considerations taken into account, it is very interesting to ask if the problems of
magnetogenesis are actually somehow related to the foundational problems of the dom-
inant ΛCDM paradigm, and if they could have their natural resolution in frameworks
that go beyond the Big Bang cosmology.
Bouncing cosmological solutions [41, 42, 43, 44, 45, 46, 47], representing the tran-
sition from a contracting to an expanding phase of the Universe, are also a necessary
ingredient of cyclic Universe models, but for that purpose, the bouncing phase needs
moreover to be combined with the approximately standard cosmological evolution fol-
lowing it (radiation dominated, matter dominated, effective cosmological term domi-
nated phases) and with a turnaround phase – the transition from the expanding to the
contracting phase, which happens at the end of each cycle. The problem of magnetoge-
nesis could be one of the problems and limitations of the Big Bang paradigm, and this
naturally leads to the program of solving magnetogenesis beyond the Big Bang cosmol-
ogy, i.e. in models that tend to overcome its limitations. Magnetogenesis produced out
of a non-minimal coupling has also been studied in the context of bouncing cosmologies
[48, 49, 50, 51]. Refs. [48, 49] show how non-singular bouncing cosmological models,
achieved by adding a bouncing term to the Lagrangian, can avoid the backreaction and
strong coupling problems, while [50, 51] show how scale invariant primordial magnetic
fields are obtained in non-minimal coupling bouncing scenarios. The main problem of
the various magnetogenesis models based on non-minimal coupling is that the main in-
gredient – the scalar field – is introduced without any proper physical motivation, and
its properties and the type of coupling are usually artificially constructed just to lead
to the desired results. The same criticism can also be applied to various models which
use a non-motivated coupling between curvature and vector potentials (which moreover
breaks the gauge invariance of the theory and should be thus approached with suspicion).
In the magnetogenesis model that we propose in this work, magnetic fields are dom-
inantly produced during the contraction phase of the previous cycle, and not during the
bouncing phase. Assuming that the Universe enters the contraction phase approxi-
mately empty, which solves various problems of a contracting Universe – including the
instabilities which otherwise generally arise – we argue that the proper picture of the
electromagnetic interaction during this epoch is given in terms of vacuum electrodynam-
ics on curved spacetime and not in terms of magnetohydrodynamics. We show that, in
typical conditions for the beginning of a contraction phase, that is, the beginning of a
new cycle, magnetic fields of sufficient strength to account for the present observational
bounds can be simply created, without any further assumptions and additions (such as
new scalar fields, non-minimal couplings, etc.).
The central assumption of our model is that the Universe undergoes endless cycles of
contraction and expansion, so that the Big Bang is replaced by a transition from the
contracting phase of the previous cycle into the expanding phase of the following cycle
– the cosmological bounce. We assume that effects coming from the generalization of
the standard Einstein equations lead to solutions of the field equations that enable both
the early bouncing phase and the late transition from an accelerated expansion to the
contraction of the Universe. Such generalizations of Einstein’s equations are understood
to come from quantum gravity considerations and to be related to the fact that general
relativity is not a complete theory of gravity and needs to break in the regimes of high
curvatures. In the same research framework, it is also possible to understand the still
unanswered problems of dark energy and dark matter as coming from the low-energy
corrections to the equations of Einstein’s general relativity [52, 53, 54, 55, 56, 57]. It is
important to note that in such a generalized setting the cosmological term is no longer
necessarily a constant, but becomes a dynamic quantity [58, 59, 60, 61].
The contraction phase of the Universe can lead to more physical difficulties, since
the contraction of the Universe with a non-vanishing matter-energy distribution in
Magnetogenesis in Cyclical Universe 8
see that this assumption can be relaxed, taking ρ 6= 0, p 6= 0, and solving the Friedmann
equation numerically, still using (4). It turns out that this does not introduce any
significant difference in the evolution of the spacetime geometry nor in the evolution of
the electromagnetic fields on it, as long as ρ < ρef f .
The components of the electromagnetic field will evolve according to (9)-(12), which
we apply to the FRWL metric (1). The evolution of the electrical field in this regime
will thus follow Ampre’s law
∂i E i = 4πρq , (15)
where i denotes spatial coordinates, and Gauss’s law
1
∂t E i + 3HE i = ∗ilm ∂l B m − 4πj i , (16)
a
where latin indexes run through the spatial coordinates and ∗ijk is the fully
antisymmetric symbol with ∗123 = 1. For the magnetic field, Maxwell’s equations take
the form
∂i B i = 0 (17)
Magnetogenesis in Cyclical Universe 12
1
∂t B i + 3HB i = − ∗ilm ∂l E m . (18)
a
where φ(t) and ψ(t) functions of time, such that the time and space evolution of fields
are independent. We moreover choose the spatial components of the fields to be
∂m E(i, m) − ∂l E(i, l) ∂l B(i, l) − ∂m E(i, m)
u= ,w = , (21)
B(l, m) B(l, m)
where u and w are constants, which need to be related through u = −v. Inserting
(19) and (20) in (16) and (18), while using this field configuration and considering the
vacuum case, we obtain the following differential equations, respectively
φ̇(t) + 3Hφ(t) = wψ(t) , (22)
This set of equations can be solved using the conditions adequate to the contraction
phase and the Hubble parameter thereof.
The Hubble parameter can be computed from Friedmann’s equation (2), where the
energy density can be modelled as ρ = ρmat + ρrad = ρ0mat /a3 + ρ0rad /a4 , where ρ0mat
and ρ0rad are the values of the matter and radiation densities at the beginning of the
contraction phase, and the pressure p = ρrad /3. As discussed previously, by the end
of contraction all structures of the Universe will have been ripped apart and matter
diluted to such extent that pressure and density can be effectively taken as vanishing.
If ρ = p = 0, (2) has an analytical solution. For completion, we considered also initial
matter and radiation densities and solved (2) numerically, finding that, for small values
of ρ0mat and ρ0rad consistent with the assumption of a nearly empty Universe at the
beginning of the contraction phase, the analytical approximation provides a very good
description.
As mentioned in § 2, at the beginning of the contraction phase, the Universe turns
from expanding into contracting, implying a maximum of the scale factor at a certain
time tmax . At the end of contraction, when the Universe bounces back into expansion,
the scale factor will display a minimum, at a time tmin . These mean that the Hubble
Magnetogenesis in Cyclical Universe 13
parameter will vanish at the points of beginning and end of the contracting phase and
should be negative in between. We use (4) in (2), encapsulating the modification of
gravity that our model allows for through the effective energy density ρef f that represents
the effect of the non-Hilbert component in the gravitational action. The Ricci scalar
can be easily obtained from it as well as the scale factor, since R = 6Ḣ + 12H 2 and
H ≡ ȧ(t)/a(t), respectively.
To use dimensionless quantities, we set H → Htmin , R → Rt2min , u → utmin and
normalize the scale factor to its minimum a/amin .
Since we can assume that at the late contraction phase the quantum description of
electromagnetism becomes relevant, but also in order to discuss the scale dependence
of the cosmological electromagnetic fields at this point, it is usual to consider the field
equations in terms of the quantized four-vector potential Aµ . Gauge freedom can be
fixed using the Coulomb gauge, A0 = 0 and ∂i Ai = 0, and the potential can be promoted
to an operator using the canonical quantization. It can then be written in terms of the
following Fourier decomposition
2 Z
X d3 k ir·k
Ai (t, r) = i,σ (k)a kσ A k (t)e + h.c. . (24)
σ=1
(2π)3/2
The sum in this expression goes over two polarization states for the electromagnetic
field, i,σ is the polarization vector, and akσ (a†kσ in the second term) is the typical
quantum annihilation (creation) operator, which obeys the standard commutation
relations [akσ , a†k0 σ0 ] = δσ,σ0 δ(k − k0 ), while annihilation and creation operators commute
among themselves. When this decomposition is used in the vacuum case described by
(5), assuming the FRWL spacetime, it yields the simple quantized Maxwell equation for
an individual Fourier mode, Ak , when conformal units are used, i.e. A00k (τ )+k 2 Ak (τ ) = 0,
where the derivatives are taken with respect to conformal time, dτ = dt/a(t), and
the conformal wave number, k → a(t)k, was also used. The general solution, Ak =
Ck eikτ + Dk e−ikτ , will then typically lead to a functional dependence of the magnetic
power spectrum, PB , on the scale, since
2
X k 5 | Ak (τ ) |2
PB (k, τ ) = , (25)
σ=1
2π 2 a4
and the spectrum will thus not be scale-invariant in general. However, the spectrum
can still be nearly scale-invariant in certain cases. For instance, this will be the case if
the initial field configuration is such that Ck2 ≈ Dk2 ≈ C 2 k −5 and the constant C is such
that C 1 or if the duration of magnetogenesis and the physical scales on which the
magnetic fields are created are such that τ k ≈ const. For the end of this discussion,
let us note that the relation between the scale dependence of magnetic fields during
magnetogenesis and magnetic fields today is not a simple issue since different processes
can change the spectrum of magnetic fields during the evolution of the Universe. For
example, the development of magnetohydrodynamics turbulence during the expanding
Magnetogenesis in Cyclical Universe 14
phase of the Universe can lead both to the transfer of magnetic energy from large scales
to smaller scales and vice versa, thus changing the original distribution of magnetic
energy.
The coupled equations (22) and (23) are a stable system with respect to initial
conditions, as we will see, and we have freedom in setting the parameter b, which is
directly related to the Hubble parameter evolution. This parameter has no influence in
the shape of the curves φ(t) and ψ(t), but it determines the strength of the generated
magnetic field. Using the simplified assumption that magnetic fields only decayed with
expansion, one can infer roughly the magnetic field strengths that were necessary to
be present or generated at the early Universe in order for the current limits placed on
extragalactic magnetic fields to be obtained. There have been no direct observations of
extragalactic magnetic fields, but several methods allow us to set limits on them. The
lower bounds alow to infer from TeV blazars emission that presently B(t0 ) & 10−12 T
[5]. Upper bounds depend on the coherence length of the field, but, at the limit, their
highest value cannot exceed the observations of Zeeman splitting from the radiation of
distant quasars that points to a magnetic field B(t0 ) . 0.1 T [72]. Since B µ Bµ ∝ a(t)−4
and estimating that the present scale factor is 1032 times larger than at the Planck
scale, if the mentioned bounds are taken, then 1055 & B(tP l ) & 1044 T, where tP l repre-
sents the Planck time. If we assume that the end of the contraction phase corresponds
roughly to the smallest scale we know of, viz. the Planck scale, we have an estimate
of the field strengths expected that this magnetogenesis mechanism supplies in order
for it to be a viable hypothesis. Let us consider the parameter b0up/low = κ−1 bup/low s−3
of (4) from which the Hubble parameter is computed. We obtain magnetic fields of
the strength expected to provide the current lower and upper bounds on extragalactic
magnetic fields taking b0low = 2.7 × 104 and b0up = 4 × 104 , respectively. In the following
plots, the different choice of upper and lower bound parameters is represented in solid
and dashed curves, respectively, to illustrate these two different cases.
Figure 1 shows on the right panel the modelled Hubble parameter in this case, ac-
cording to (2) and on the left panel the scale factor obtained by integrating H(t). Both
show the expected discussed behaviour for the contraction regime.
Magnetogenesis in Cyclical Universe 15
Figure 1. Scale factor and Hubble parameter during the contraction phase of the
Universe. Computed from (2) using (4), taking u = −w = 1. Solid lines represent the
possible parameter choice in (4) rendering the lowest possible bound on extragalactic
magnetic fields and dashed lines the highest, illustrating the range of normalized values
of the scale factor and Hubble parameter in these cases.
Figure 2. Magnetic and electrical field evolution during the contraction phase of
the Universe. Computed solving (23) and (22), respectively, taking u = −w = 1 and
using as initial conditions B(tmax ) = 0 T and E(tmax ) = 10−4 V/m, while the Hubble
parameter is determined through the solution of (2). Solid lines represent the possible
parameter choice in (4), rendering the lowest possible bound on extragalactic magnetic
fields and dashed lines the highest, illustrating the allowed range of generated fields in
these cases.
Magnetogenesis in Cyclical Universe 16
As can be concluded through Figure 2, in this setting, from a small seed electrical
field a strong enough current is generated that induces a magnetic field. This field will
not only be sustained but will also get amplified in the course of contraction due to the
decrease of the scale factor thereof.
The starting assumption used in our analysis was that the contracting Universe,
which is nearly empty and permeated with weak electric fields, can be described by
the FRWL metric (1). A Universe consisting only of electromagnetic fields is not
consistent with the symmetry of FRWL metric and can thus not be described by the
Friedmann equations. Therefore, although approximately empty, the matter-energy
content of the Universe still needs to dominate over the contribution coming from
the electromagnetic fields, while being significantly smaller than the contribution of
the effective energy density, modelling the effects of modified gravity. If this is the
case at the beginning of contraction, it is simple to see that it will remain to be so.
In fact, this will happen as long as the initial density values are ρ0mat,rad . 10−3 b.
Using the characteristic scalings for energy densities, at later stages of contraction
[ρB (t) + ρE (t)]/[ρrad (t) + ρmat (t)] ≈ [ρB (tmax ) + ρE (tmax )]/ρrad (tmax ). Therefore, under
this assumption for initial conditions, [ρB (tmax ) + ρE (tmax )]/ρrad (tmax ) 1, there is no
backreaction problem in the proposed model.
5. Conclusions
Let us state the assumptions at the base of this work in order to understand the ap-
plicability scope and limitations of the same. We have assumed that electromagnetic
fields are independent of the spacetime background, not affecting the metric. Isotropic
and homogeneous conditions are taken and thus an FRWL metric is used.
densities are negligible. These conditions, apart from inducing initial electrical fields,
are also favourable to the growth of magnetic fields in the subsequent contraction. We
did not assume any particular modified gravity framework nor model and kept the anal-
ysis general enough to suit a broad class of hypotheses.
The obtained results are quite robust with respect to the inclusion of more detailed
effects taking place during the evolution of the Universe (for example, to include matter
and radiation components will not change the main features of cosmological evolution as
long as the contribution of the effective dark energy is dominant). The approximations
that we have taken, namely the Universe being approximately empty at the beginning of
the contraction phase and the geometrical properties of the background considered, are
well justified and not merely introduced ad hoc. For the former, the late acceleration
phase in the Universe dominated by a growing effective dark energy contribution will
lead naturally to an effectively empty Universe afterwards. For the latter, the considered
ansatz is the most general description of the contracting regime by simply considering
Taylor expansion suitable to these cases. The generality of such approach has the clear
advantage, in comparison to analysing a large number of specific different modifications
of gravity, of highlighting the general features of magnetic field generation in a contract-
ing Universe. The results found here can now simply be introduced in any potential
modified gravity theory capable of supporting cyclical solutions. The central result of
our paper is that the creation of magnetic fields is a general feature of cyclical models
and it does not require any new physics beyond the modifications of the gravitational
sector that are needed to lead to a cyclic Universe.
The main new conclusion of this work can be summarized by stating that in the
cyclical Universe, magnetogenesis can occur during the phase of contraction resorting
only to a small seed electrical field, which is naturally created in the conditions
characterizing the beginning of the contraction phase.
Magnetogenesis in Cyclical Universe 18
Acknowledgments
This work was partially supported by the Centre for Space Research of the Faculty of
Natural and Agricultural Sciences of North-West University. NL wishes to thank the
warm hospitality received at the Centre for Physics of the University of Coimbra and
interesting discussions with Alex Blin.
References
[1] Grasso D and Rubinstein H R 2001 Phys. Rept. 348 163–266 (Preprint astro-ph/0009061)
[2] Giovannini M 2004 Int. J. Mod. Phys. D13 391–502 (Preprint astro-ph/0312614)
[3] Subramanian K 2007 PoS MRU 071 (Preprint 0802.2804)
[4] Kandus A, Kunze K E and Tsagas C G 2011 Phys. Rept. 505 1–58 (Preprint 1007.3891)
[5] Neronov A and Vovk I 2010 Science 328 73 (Preprint 1006.3504)
[6] Dolag K, Kachelriess M, Ostapchenko S and Tomas R 2011 Astrophys. J. 727 L4 (Preprint
1009.1782)
[7] Tavecchio F, Ghisellini G, Foschini L, Bonnoli G, Ghirlanda G and Coppi P 2010 Mon. Not. Roy.
Astron. Soc. 406 L70–L74 (Preprint 1004.1329)
[8] Durrer R and Neronov A 2013 Astron. Astrophys. Rev. 21 62 (Preprint 1303.7121)
[9] Turner M S and Widrow L M 1988 Phys. Rev. D37 2743
[10] Dolgov A 1993 Phys. Rev. D48 2499–2501 (Preprint hep-ph/9301280)
[11] Kunze K E 2010 Phys. Rev. D81 043526 (Preprint 0911.1101)
[12] Bamba K, Geng C Q and Luo L W Large-Scale Magnetic Fields from Inflation in Teleparallel
Gravity (Preprint http://journals.jps.jp/doi/pdf/10.7566/JPSCP.1.013123) URL http:
//journals.jps.jp/doi/abs/10.7566/JPSCP.1.013123
[13] Ratra B 1992 Astrophys. J. 391 L1–L4
[14] Gasperini M, Giovannini M and Veneziano G 1995 Phys. Rev. Lett. 75 3796–3799 (Preprint
hep-th/9504083)
[15] Giovannini M 2008 Phys. Lett. B659 661–668 (Preprint 0711.3273)
[16] Giovannini M 2001 Phys. Rev. D64 061301 (Preprint astro-ph/0104290)
[17] Martin J and Yokoyama J 2008 JCAP 0801 025 (Preprint 0711.4307)
[18] Fujita T, Namba R, Tada Y, Takeda N and Tashiro H 2015 JCAP 1505 054 (Preprint 1503.05802)
[19] Bamba K 2015 Phys. Rev. D91 043509 (Preprint 1411.4335)
[20] Giovannini M and Shaposhnikov M E 2000 Phys. Rev. D62 103512 (Preprint hep-ph/0004269)
[21] Subramanian K 2010 Astron. Nachr. 331 110–120 (Preprint 0911.4771)
[22] Fujita T and Mukohyama S 2012 JCAP 1210 034 (Preprint 1205.5031)
[23] Campanelli L 2015 Eur. Phys. J. C75 278 (Preprint 1503.07415)
[24] Demozzi V, Mukhanov V and Rubinstein H 2009 JCAP 0908 025 (Preprint 0907.1030)
[25] Atmjeet K, Seshadri T R and Subramanian K 2015 Phys. Rev. D91 103006 (Preprint 1409.6840)
[26] Ellis G F R, Maartens R and MacCallum M A H 2012 Relativistic Cosmology
[27] Carroll S M 2001 Living Rev. Rel. 4 1 (Preprint astro-ph/0004075)
[28] Baumann D and Peiris H V 2009 Adv. Sci. Lett. 2 105–120 (Preprint 0810.3022)
[29] Hawking S W 1966 Proceedings of the Royal Society of London Series A 294 511–521
[30] Hawking S W 1966 Proceedings of the Royal Society of London Series A 295 490–493
[31] Hawking S W 1967 Proceedings of the Royal Society of London Series A 300 187–201
[32] Tolman R C 1931 Phys. Rev. 38 1758
[33] Lemaı̂tre G 1933 Annales de la Société Scientifique de Bruxelles 53
[34] Markov M A 1984 Annals Phys. 155 333–357
[35] Okouma P M, Fantaye Y and Bassett B A 2013 Phys. Lett. B719 1–4 (Preprint 1207.3000)
[36] Yu H, Ratra B and Wang F Y 2018 Astrophys. J. 856 3 (Preprint 1711.03437)
Magnetogenesis in Cyclical Universe 19
[37] Ellis G F R, Platts E, Sloan D and Weltman A 2016 JCAP 1604 026 (Preprint 1511.03076)
[38] Khoury J, Ovrut B A, Steinhardt P J and Turok N 2001 Phys. Rev. D64 123522 (Preprint
hep-th/0103239)
[39] Saridakis E N, Banerjee S and Myrzakulov R 2018 (Preprint 1807.00346)
[40] Pavlovic P and Sossich M 2017 Phys. Rev. D95 103519 (Preprint 1701.03657)
[41] Battefeld D and Peter P 2015 Phys. Rept. 571 1–66 (Preprint 1406.2790)
[42] Cai Y F 2014 Sci. China Phys. Mech. Astron. 57 1414–1430 (Preprint 1405.1369)
[43] Brandenberger R and Peter P 2017 Found. Phys. 47 797–850 (Preprint 1603.05834)
[44] Astashenok A V 2014 Astrophys. Space Sci. 351 377–383 (Preprint 1308.0581)
[45] Nojiri S, Odintsov S D and Oikonomou V K 2017 Phys. Rept. 692 1–104 (Preprint 1705.11098)
[46] Ben-Dayan I 2016 JCAP 1609 017 (Preprint 1604.07899)
[47] Andrade L C G 2018 Eur. Phys. J. C78 530
[48] Koley R and Samtani S 2017 JCAP 1704 030 (Preprint 1612.08556)
[49] Qian P, Cai Y F, Easson D A and Guo Z K 2016 Phys. Rev. D94 083524 (Preprint 1607.06578)
[50] Sriramkumar L, Atmjeet K and Jain R K 2015 JCAP 1509 010 (Preprint 1504.06853)
[51] Chowdhury D, Sriramkumar L and Jain R K 2016 Phys. Rev. D94 083512 (Preprint 1604.02143)
[52] Tsujikawa S 2010 Lect. Notes Phys. 800 99–145 (Preprint 1101.0191)
[53] Bekenstein J D 2010 99–117 (Preprint 1001.3876)
[54] Joyce A, Lombriser L and Schmidt F 2016 Ann. Rev. Nucl. Part. Sci. 66 95–122 (Preprint
1601.06133)
[55] Katsuragawa T and Matsuzaki S 2017 Phys. Rev. D95 044040 (Preprint 1610.01016)
[56] Lue A, Scoccimarro R and Starkman G 2004 Phys. Rev. D69 044005 (Preprint astro-ph/0307034)
[57] Calmet X and Kuntz I 2017 Eur. Phys. J. C77 132 (Preprint 1702.03832)
[58] Borges H A and Carneiro S 2005 Gen. Rel. Grav. 37 1385–1394 (Preprint gr-qc/0503037)
[59] Shapiro I L and Sola J 2009 Phys. Lett. B682 105–113 (Preprint 0910.4925)
[60] Pan S 2018 Mod. Phys. Lett. A33 1850003 (Preprint 1712.01215)
[61] Pan S, Saridakis E N and Yang W 2017 (Preprint 1712.05746)
[62] Battefeld T J and Brandenberger R 2004 Phys. Rev. D70 121302 (Preprint hep-th/0406180)
[63] Sokolowski L M 1995 Universality of Einstein’s general relativity General relativity and gravitation.
Proceedings, 14th International Conference, Florence, Italy, August 6-12, 1995 pp 337–354
(Preprint gr-qc/9511073)
[64] Kijowski J 2016 Int. J. Geom. Meth. Mod. Phys. 13 1640008 (Preprint 1604.05052)
[65] Belinskii V, Khalatnikov I and Lifshitz E 1970 Advances in Physics 19 525–573
(Preprint https://doi.org/10.1080/00018737000101171) URL https://doi.org/10.1080/
00018737000101171
[66] Lifshitz E M and Khalatnikov I M 1963 Adv. Phys. 12 185–249
[67] Misner C W 1969 Phys. Rev. Lett. 22(20) 1071–1074 URL https://link.aps.org/doi/10.1103/
PhysRevLett.22.1071
[68] Davies P C W 1988 Ann. Inst. H. Poincare Phys. Theor. 49 297
[69] Subramanian K 2016 Rept. Prog. Phys. 79 076901 (Preprint 1504.02311)
[70] Kolb E W and Turner M S 1990 Front. Phys. 69 1–547
[71] Baym G and Heiselberg H 1997 Phys. Rev. D56 5254–5259 (Preprint astro-ph/9704214)
[72] Wolfe A M, Jorgenson R A, Robishaw T, Heiles C and Prochaska J X 2008 Nature 455 638
(Preprint 0811.2408)