See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/307636814

Fully stable cosmological solutions with a non-

singular classical bounce

Article · September 2016

CITATIONS READS

2 7

2 authors, including:

Anna Ijjas
Princeton University
12 PUBLICATIONS 191 CITATIONS

SEE PROFILE

Available from: Anna Ijjas

Retrieved on: 21 November 2016
 Fully stable cosmological solutions with a non-singular classical bounce

Anna Ijjasa,∗, Paul J. Steinhardta,b

a Princeton Center for Theoretical Science, Princeton University, Princeton, NJ 08544 USA
b Department of Physics, Princeton University, Princeton, NJ 08544, USA

Abstract
We recently showed how it is possible to use a cubic Galileon action to construct classical cosmological solutions that enter a
contracting null energy condition (NEC) violating phase, bounce at finite values of the scale factor and exit into an expanding NEC-
satisfying phase without encountering any singularities or pathologies. A drawback of these examples is that singular behavior
arXiv:1609.01253v1 [gr-qc] 5 Sep 2016

is encountered at some time either just before or just after the NEC-violating phase. In this Letter, we show that it is possible
to circumvent this problem by extending our method to actions that include the next order L4 Galileon interaction. Using this
approach, we construct non-singular classical bouncing cosmological solutions that are non-pathological for all times.
Keywords: bouncing cosmology, non-singular bounce, Galileon, ghost instability, gradient instability

1. Introduction In [6], we showed that it is possible to construct classi-

Cosmological scenarios that involve a phase of contraction
followed by a bounce to a phase of expansion are of great in-
terest since they can smooth and flatten the cosmological back-
ground [1] and generate a nearly scale-invariant spectrum of
super-horizon curvature modes [2, 3] while avoiding the multi-
verse and initial conditions problems of inflationary cosmology.
In these theories, the smoothing contraction phase is fully de-
scribed by Einstein gravity and the quantum generation of cur-
vature modes is described by standard semi-classical perturba-
tion theory. The challenge has been to find a non-pathological
theoretical framework for describing the bounce, i.e., the transi-
tion from contraction to expansion. One possibility is to realize
the transition through a singular (‘quantum’) bounce in which
the scale factor passes through or tunnels through zero, as was
proposed in [4]; this idea is intriguing, though the approach re-
lies on some as-yet unproven assumptions about the analyticity
of quantum gravity [5].
Another approach is a ‘classical bounce,’ in which the uni-
verse bounces after contracting to a small but finite size with
energy density well below the Planck scale such that quan-
tum gravity effects can be neglected. The transition occurs
through violation of the null energy condition (NEC) over a
finite period of time surrounding the bounce. On a smooth and
flat Friedmann-Robertson-Walker (FRW) cosmological back-
ground with ds2 = −dt2 + a2 (t)dxi dxi (where a(t) is the scale
factor), NEC violation means that the Hubble parameter H =
ȧ/a increases with time, Ḣ > 0, where dot denotes differentia-
tion with respect to time t. The classical bounce has the advan-
tage of not requiring any knowledge of quantum gravity. How-
ever, it is a well-known problem that NEC violation is prone to
ghost or gradient instabilities or leads to singular behavior.
∗ Correspondingauthor.
Email address: aijjas@princeton.edu (Anna Ijjas)
Preprint submitted to Elsevier
cal solutions that enter a contracting NEC-violating phase,
bounce, and exit in an expanding NEC-satisfying phase with-
out introducing pathologies, by realizing the NEC-violating
stage through a scalar field described by the generalized cubic
Galileon action. We presented an ‘inverse method’ for con-
structing the solutions and used it to construct explicit exam-
ples. The examples show that the universe can undergo a cos-
mological bounce at finite values of the scale factor and low
energies well below the Planck scale without encountering sin-
gular behavior or requiring superluminal sound speed of co-
moving curvature modes during the NEC-violating phase.
A feature of the examples based on the cubic Galileon ac-
tion, though, is that singular behavior is always encountered
at some time either shortly before or shortly after the NEC-
violating phase. The remaining open issue is whether these
pathologies are inevitable or if a stable NEC-violating bounce
stage can be embedded into a cosmology that is stable and non-
singular throughout cosmic evolution. In this Letter, we show
explicitly that it is possible to construct a fully stable bouncing
cosmology by naturally extending our inverse method to actions
that include the next order L4 Galileon interaction.
This Letter is organized as follows: First, we give a brief
review of Galileon cosmology and derive the stability condi-
tions for linear-order scalar and tensor perturbations. Next, we
explain why the cubic Galileon action inevitably leads to diver-
gences and/or other singular behavior either just before or just
after the stable NEC-violating phase in Sec. 3. Then, we show
that, in principle, by extending the action to include the L4
Galileon interaction, the pathological behavior can be avoided
for all times. In Sec. 4, we use our inverse method to construct
explicit solutions. (Readers only interested in the existence of
non-pathological bouncing solutions may wish to jump to the
figures.)
September 6, 2016
2. Galileon cosmology 2.1. Background
On a spatially-flat, FRW cosmological background, the cor-
responding homogeneous equations of motion take the form
We consider bouncing Galileon cosmologies (also known as 1 2 1
3H 2 kφ̇ + 3q − 2b′ φ̇4 + V − 3 f1′ H φ̇


Horndeski theories) described by the action of a single scalar = (11)
2 4
field,  
3 b − 3 f2′ H φ̇3 − 3 f1 − 3 f2 φ̇2 H 2 ,


+
Z !
4 √ 1 2 k + f1′′ φ̇2 + q − b′ + f2′′ φ̇4 + f1′ φ̈



S = d x −g MPl R + Lφ , (1) −2Ḣ = (12)
2 ′
2 ′ ′
3
+ 3 f2 − b φ̈φ̇ − f1 H φ̇ + 3b − 11 f2 H φ̇
 
where MPl is the reduced Planck mass; R is the Ricci scalar; and − 4 f2 H φ̇φ̈ + 6 f2 φ̇2 H 2 + 2Ḣ f1 − f2 φ̇2 ,
g is the metric determinant. Throughout, we work in reduced where prime denotes differentiation with respect to the scalar φ.
2
Planck units (MPl = 1). The Lagrangian density of the Galileon The first Friedmann equation characterizes the different contri-
field φ is given by butions to the total energy density ρtot while the second Fried-
5
X mann equation gives the sum of total energy density and pres-
Lφ = Li , (2) sure ptot . The ratio −Ḣ/H 2 is equal to the equation of state
i=2 ǫ ≡ (3/2)(ρtot + ptot )/ρtot . Notably, it is a distinctive feature of
the Galileon that there is a non-trivial mixing between spatial
with curvature terms and the scalar field, leading to possible ambi-
guities in defining the stress-energy tensor [9]. Here, we have
followed the convention and derived T µν by varying the covari-
L2 = G2 (X, φ), (3)
antized theory with respect to the metric.
L3 = −G3 (X, φ)φ, (4)
 2 
2.2. Perturbations

2
L4 = G4 (X, φ)R + G4 ,X (X, φ) (φ) − ∇µ ∇ν φ , (5)
On a homogeneous background, the source of leading-order
L5 = G5 (X, φ)Gµν∇µ ∇ν φ + (6) inhomogeneities is linear fluctuations of the metric and the
G 5 ,X   2  3  scalar field. To ensure stability, the linear theory must not have
− (φ)3 − 3φ ∇µ ∇ν φ + 2 ∇µ ∇ν φ .
6 any pathologies.
To study the stability behavior of the linear theory, we expand
Here, X = −(1/2)∇µφ∇µ φ is the canonical quadratic kinetic the ADM decomposition of the Galileon action up to second or-
term and the functions Gi (X, φ) characterize the ith Galileon in- der in perturbation theory. The ADM metric is defined through
teraction. The action was shown by Horndeski [7] to lead to   
classical equations of motion that are only second order in time ds2 = −Ndt2 + gi j N i dt + dxi N j dt + dx j , (13)
derivatives, thus avoiding an Ostragadski ghost. Horndeski and
where N is the lapse, Ni is the shift, and gi j is the spatial metric.
Galileon theories have had numerous applications in cosmology
We parametrize linear-order perturbations to the shift and lapse
in the past, including massive gravity and attempts to explain
as
late-time acceleration [8]. Here we apply them to construct sta-
δN = N − N̄ = α, δNi = Ni − N̄i = ∂i β , (14)
ble cosmological solution that violate the NEC and bounce.
For the purposes of illustration, we shall work with the con- with N̄ = 1, N̄i = 0 and ḡi j = a2 δi j being the FRW background
crete example defined through the coupling functions metric. Since throughout we consider a single field, to study the
perturbed scalar sector, we are free to choose the unitary gauge
in which all spatial inhomogeneities are promoted to the metric,
G2 (X, φ) = k(φ)X + q(φ)X 2 − V(φ) , (7) !
G3 (X, φ) = b(φ)X , (8) 1
gi j = a2 e2ζ δi j + hi j + hik hk j , (15)
1 2
G4 (X, φ) = f1 (φ) + f2 (φ)X , (9)
2 while the scalar does not carry any perturbations, δφ ≡ 0. Here,
G5 (X, φ) = 0, (10) ζ is the co-moving curvature mode and hi j is the linear-order
tensor perturbation with ∂ j hi j = 0, hii = 0. Note that both ζ and
where k(φ) is the dimensionless quadratic coupling; q(φ) is the hi j are gauge-invariant.
dimensionless quartic coupling; b(φ) is the dimensionless cou- The second-order action for ζ, α and β is given by the La-
pling of the scalar field φ to the cubic Galileon term; and V(φ) is grangian
the scalar potential. The positive definite coupling to the Ricci
(∂i ζ)2
scalar (1/2) f1 (φ) + f2 (φ)X is distinctive of the L4 Galileon in- L(2)
ζ,α,β = a3 − 3Ah ζ̇ 2 + Bh + mα α2 (16)
teraction. Note that positivity of the non-minimal coupling en- a2
sures that anti-gravity regions and associated instabilities are ∆ζ ∆β  !
+ 6γαζ̇ − 2Ah α 2 + 2 2 Ah ζ̇ − γα ,
avoided. a a
2
where restore the NEC without encountering any instability or singu-
larity during this bounce phase. However, there remained bad
Ah (t) = 1 + f1 − f2 φ̇2 , (17) behavior either shortly before or shortly after the NEC-violating
Bh (t) = 1 + f1 + f2 φ̇2 , (18) phase, depending on whether γ was positive or negative when
1 2 1 entering the NEC-violating phase. In this section, we will show
kφ̇ + 3q − 2b′ φ̇4 − 3 f1′ φ̇H


mα (t) =
2 2 how expanding the action to include the next order Galileon
interaction L4 can naturally remove this remaining singular be-
 
6 b − 3 f2′ H φ̇3 − 3 1 + f1 − 6 f2 φ̇2 H 2 ,


+ (19)
havior.
  1
γ(t) = 1 + f1 − 3 f2 φ̇2 H + f1′ φ̇ (20) The key equation for analyzing the gradient instability,
2 Eq. (25), can be rearranged and integrated from t to t0 , where
1
b − 3 f2′ φ̇3 . t0 is the time at the beginning of the NEC-violating phase and


−
2 t < t0 , to yield the relation
Varying the perturbed action with respect to the shift yields the
aA2h  aA2h  aA2h 
Z t0
momentum constraint which is, at the same time, a closed-form
  
 =  − a(B + Bh )dt ≡  − I(t) . (27)
expression for the lapse α, γ t γ t0 t γ t0
Ah (t)ζ̇ = γ(t)α. (21) Without loss of generality, we take γ(t0 ) = γ0 positive.
Let us first consider the case of the cubic Galileon action for
The equation for the shift follows from substituting into the which Ah = Bh = 1. The first term on the right hand side is pos-
Hamiltonian constraint, itive. Note that, for a non-singular bounce where a > 0 for all
∆β   ∆ζ t, the integrand in the second term is positive definite, so I(t) is
γ2 (t) = m α (t)A h (t) + 3γ 2
(t) ζ̇ − γ(t)Ah (t) 2 . (22) divergent as t → −∞. Consequently, aA2h /γ must become nega-
a2 a
tive at some finite t < t0 , which is where trouble is encountered.
Note that, if γ = 0, ζ̇ = 0 or Ah (t) = 0. Either γ diverges (which requires couplings and/or kinetic en-
Substituting the expression for α and β back into Eq. (16), we ergies to diverge) or γ passes through zero, which requires that
obtain the second-order action for co-moving curvature modes a or B in the integrand become negative and diverge. As em-
ζ, phasized in [6], this trouble occurs outside the NEC-violating
(∂i ζ)2
Z !
(2) 4 3 2 regime, more precisely at t < t0 , and is not directly related to
Sζ = d xa A(t)ζ̇ − B(t) 2 , (23)
a NEC or the bounce; it is a feature of the cubic Galileon action.
The introduction of the next order Galileon interaction, L4 ,
where the coefficients of the kinetic and gradient terms are de-
provides the freedom for Ah and Bh to become functions of
fined as
!2 time, which makes it possible to satisfy Eq. (27) for all time
Ah (t) without encountering any bad behavior. First, as we will il-
A(t) = mα (t) + 3Ah (t) , (24)
γ(t) lustrate in the next section, it is possible for B and Bh to ap-

A2 (t) 
 proach zero rapidly enough for I(t) to converge to a finite value
d 
B(t) = a−1 (t) a(t) h  − Bh (t) . (25) as t → −∞. (We note that T. Kobayashi, in presenting his no-go
dt γ(t) theorem in Ref. [11], included a footnote suggesting that there
Similarly, the second-order action for tensor modes takes the might be a loophole if Bh → −∞; however, he never developed
simple form the idea to determine all the requisite conditions or checked if
they could be satisfied.) For example, if a(t) ≤ |t| p for t < t0
  2 
Z  ∂ l h i j
 and some finite p, then it suffices that B and Bh converge to
S h(2)i j = d4 xa3 Ah (t)ḣ2i j − Bh (t) , (26) zero faster than 1/|t| p+1 . Provided γ0 is chosen such that
 
 a2 
aA2h 

where the coefficients of the kinetic and gradient terms are as  − I(−∞) > 0 , (28)
defined above in Eqs. (17-18). (The perturbed Horndeski action
was previously obtained in [10]; here we re-derived the action
for our specific example to emphasize some pedagogical points,
in particular, the role of the quantity we call γ.)
For the theory to be linearly stable, both the scalar and the
tensor sector have to be stable, i.e., the coefficients A, B, Ah, or
Bh must not become negative.
3. Singular behavior and its resolution
In Ref. [6], we showed that it is possible with the cubic
Gailileon action to enter a NEC-violating phase, bounce, and
3
γ t0
which is possible since I(−∞) is finite, the left hand side of
Eq. (27) can remain positive for all t < t0 ; and since there was
no trouble for t > t0 , γ can now remain positive for all finite t.
To keep B finite, we need the additional condition that aA2h /γ
remain finite; since a diverges as t → −∞, this requires Ah
approach zero faster than 1/|t| p/2 . If, as in the examples in the
following section, γ also approaches zero as |t|q , then Ah must
approach zero faster, as |t|q−(p/2) . As we will show next, we can
extend the inverse method used in [6] to show there is no barrier
to achieving these conditions on B, Ah , Bh or γ and to construct
complete non-singular bouncing solutions for all t.
 q(t)

.
H(t)
f2(t)
b(t)
t
1000 2000 3000 -3000 - 2000 - 1000 1000 2000 3000
-3000 - 2000 - 1000 t
k(t)
.
V(t) φ(t)
f1(t)

bounce stage bounce stage

Figure 1: A plot of the input background solutions for the fig- Figure 2: A plot of the dimensionless kinetic coefficients in
ures in this paper: Ḣ (dashed yellow curve), φ̇(t) (dotted blue Eqs. (29-33) obtained using the inverse method. The time co-
curve) and V(φ(t)) ≡ V(t) (solid green curve). The time coor- ordinate is given in Planck units; the units for the y-axis are
dinate is given in reduced Planck units; the units for the y-axis arbitrary and the functions are rescaled for the purposes of il-
are arbitrary and the functions are rescaled for the purposes of lustrating their shapes. The bounce stage indicated on the figure
illustrating their shapes. The range around t = 0 where Ḣ > 0 is is the period where Ḣ > 0. See Fig. 1.
the bounce stage that encompasses the period of NEC-violation.
2 2
The specific functions are H(t) = (c1 + c2 e−c3 (t−t¯) )/(1+c4 t ) where Ah + Bh
−6 −6 −4 −5 f1 (t) = −1, (29)
c1 = 5.6 × 10p , c2 = 8.6 × 10 , c3 = 10 and c4 = 1.6 × 10 ;
−6 2
φ̇(t) = −1/ 1 + c5 t2 with c5 = 5 × 10 ; and the potential has a Bh − Ah
typical ekpyrotic form, V(t) = c6 [tanh{c7 (t + 200)} + tanh{c7 (t + f2 (t) = , (30)
2 φ̇2
1000)}], with c6 = 5 × 10−6 and c7 = 5 × 10−3 ; where reduced !
Planck units are used throughout. We note that these are only 2 1 3
b(t) = 3
−γ + (2Ah − Bh ) H + f˙1 + f˙2 φ̇2 , (31)
particular examples for the purposes of illustration; there is sig- φ̇ 2 2
nificant flexibility in designing non-pathological solutions. 2 d


k(t) = − 2 γ̇ + 3Hγ + Ah + Bh H (32)
φ̇ dt
3 
+ (Ah + Bh ) H 2 + f¨1 − 2V ,
To keep A(t) positive, we must ensure that the right hand side 2
of Eq. (24) remains positive. This is particularly easy to achieve
without affecting B(t) by including a potential, V(φ(t)) ≡ V(t), 4  d
q(t) = γ̇ + 9Hγ + ((Ah + Bh ) H) (33)
which contributes to mα in Eq. (24) but has no direct impact on 3φ̇4 dt
B(t) in Eq. (25). 9  2
+ (Bh − Ah ) H 2 + f¨1 − 3V + b′ ;
2 3
the couplings as a function of φ follow immediately after sub-
stituting the inverse function t(φ) for the variable t. The first
three relations follow from the definition of Ah , Bh and γ while
4. Examples the last two relations can be easily derived from the back-
ground equations (11-12). We note that the inverse method is
not constrained to the specific Lagrangian given by Eqs. (7-10)
but straightforwardly applies to arbitrary Galileon couplings
To construct examples of fully stable cosmological solutions Gi (X, φ).
with a non-singular classical bounce using the ‘inverse method,’ Finally, we can compute mα as a function of the input dy-
we make use of the fact that, once L4 is included, we have namical variables H, φ̇, Ah , Bh , γ and V(t),
enough degrees of freedom that we can independently choose d
= 3 (Ah + Bh ) H 2 − 4V +


the background behavior H(t), the behavior of γ(t) derived from mα (t) (Ah + Bh ) H (34)
dt
the shift constraint, the dynamics of the tensor sector Ah (t) and
+ γ̇ + 3Hγ + 3H f˙1 + f¨1 ,
Bh (t), and the potential V(t) and that this freedom is enough to
find solutions in which A(t) and B(t) are non-negative for all t. substitute into Eq. (24) and check that A(t) is non-negative. As
See Fig. 1. In addition, we can specify φ(t), which, together noted above, the freedom to introduce a potential makes this
with the other input functions, suffices to determine the cou- condition easy to attain. In our example, for the purposes of
plings f1 , f2 , b, k, and q as a function of time t via the following illustration, we selected V(φ) corresponding to a typical ekpy-
relations rotic potential [12]; see Fig. 1. For Ah = Bh = 1, Eqs. (29-33)
4
 1.0
cs2(t) =B(t)/A(t)
0.5
bounce stage
-3000 - 2000 - 1000 NEC NEC 1000 2000
t 3000
begins ends
Figure 3: A plot of the sound speed c2S ≡ B(t)/A(t) (solid blue
curve) for co-moving curvature modes as a function of time t,
corresponding to the example defined in Figs. 1 and 2. The
time coordinate is given in Planck units and c2S is given in units
where the speed of light is unity. Throughout, the sound speed
is real and subluminal. The bounce stage indicated on the figure
is the period where Ḣ > 0. See Fig. 1. The tickmark indicted
with the black triangle shows where B/A would have become
negative if the L4 interaction had not been included. The ratio
of tensor mode coefficients Bh (t)/Ah (t) is also positive for all
finite t.
reduce to the corresponding expressions for the generalized cu-
bic Galileon that we derived in Ref. [6].
Employing the inverse method, the analysis of the linear the-
ory dramatically simplifies. It is now possible to systematically
and rapidly search for fully stable, non-singular bouncing back-
ground solutions that show the asymptotic behavior outlined in
the previous section. Using the example in Figure 1, the result-
ing kinetic coefficients are shown in Fig. 2. In Figure 3, we
have plotted the corresponding sound speed for co-moving cur-
vature modes c2S . For all finite times, the sound speed is real
(A(t), B(t) > 0) and subluminal, with 0 < c2S < 1. (An analysis
of the asymptotic limit t → −∞ will be discussed elsewhere;
see [13].)
We emphasize that, in applying the inverse method, one must
specify more than the desired background solutions H(t) and
φ(t) as input since they do not form a complete set for charac-
terizing the linear theory. Rather, it is important to choose the
right combination of dynamical background variables that fully
describe the linear theory, as exemplified above.
5. Discussion
In this Letter, we presented fully stable non-singular bounc-
ing cosmological solutions, confirming and extending our pre-
vious result in Ref. [6] where we demonstrated that it is possible
to construct a stable NEC-violating bounce stage. The key was
to identify the source of the singular behavior in cubic Galileon
cosmologies. What we have shown is that the bad behavior is
not directly related to the NEC-violating bounce stage, but to
the fact that the Hubble parameter H(t) switches sign at some
5
point during cosmic evolution. Before or after H(t) changes
sign, the dynamical quantity derived from the shift constraint,
γ(t), has to change sign as well, which is what causes the patho-
logical behavior. In fact, pathological behavior arises in cubic
Galileon cosmologies that smoothly transit from expansion to
contraction without any bounce or NEC violation. This ob-
servation explains why earlier authors were able to find non-
pathological NEC-violating solutions in cubic Galileon genesis
models where H(t) does not change sign (e.g., see [14]), but
failed to find fully stable solutions with a bounce.
Notably, the pathology of the cubic Galileon action is re-
solved in a natural way, simply by extending the action to in-
clude the next-order L4 interaction. Another natural extension
is to include additional degrees of freedom corresponding to
NEC-satisfying matter and radiation [13]. The resulting con-
struction of a fully stable bouncing solution should give one
pause. Together with Ref. [6], it removes the last major road-
block that has been holding back interest in cosmologies that
explain the origin of the large-scale structure of the universe in
terms of a contracting phase connecting to the current expand-
ing phase through a cosmological bounce.
Acknowledgements. We thank Frans Pretorius and Vasileios
Paschalidis for helpful discussions. This research was partially
supported by the U.S. Department of Energy under grant num-
ber DEFG02-91ER40671.
References
[1] J. Khoury, B. A. Ovrut, P. J. Steinhardt, and N. Turok, Phys. Rev. D64,
123522 (2001), hep-th/0103239.
[2] J.-L. Lehners, P. McFadden, N. Turok, and P. J. Steinhardt, Phys.Rev.
D76, 103501 (2007), hep-th/0702153.
[3] E. I. Buchbinder, J. Khoury, and B. A. Ovrut, Phys.Rev. D76, 123503
(2007), hep-th/0702154.
[4] S. Gielen and N. Turok, Phys. Rev. Lett. 117, 021301 (2016),
1510.00699.
[5] I. Bars, P. J. Steinhardt, and N. Turok, Phys.Lett. B726, 50 (2013),
1307.8106.
[6] A. Ijjas and P. J. Steinhardt (2016), 1606.08880.
[7] G. W. Horndeski, Int. J. Theor. Phys. 10, 363 (1974).
[8] C. de Rham, Comptes Rendus Physique 13, 666 (2012), 1204.5492.
[9] J. Khoury, J.-L. Lehners, and B. A. Ovrut, Phys.Rev. D84, 043521 (2011),
1103.0003.
[10] T. Kobayashi, M. Yamaguchi, and J. Yokoyama, Phys. Rev. Lett. 105,
231302 (2010).
[11] T. Kobayashi, Phys. Rev. D94, 043511 (2016), 1606.05831.
[12] P. J. Steinhardt and N. Turok, Phys.Rev. D65, 126003 (2002),
hep-th/0111098.
[13] A. Ijjas (to appear).
[14] D. Pirtskhalava, L. Santoni, E. Trincherini, and P. Uttayarat, JHEP 12,
151 (2014).