Dirac-Born-Infeld Field Trapped in the Braneworld

Ricardo Garc a-Salcedo,1, a Dania Gonzalez,2, b Tame Gonzalez,3, c Claudia Moreno,4, d and Israel Quiros5, e
Centro de Investigacion en Ciencia Aplicada y Tecnologia Avanzada - Legaria del IPN, Mxico D.F., Mxico. e e 2 Departamento de Matemtica, Universidad Central de Las Villas, 54830 Santa Clara, Cuba. a 3 Departamento de F sica, Centro de Investigacin y de Estudios Avanzados del IPN, A.P. 14-740, 07000 Mxico D.F., Mxico. o e e 4 Departamento de F sica y Matemticas, Centro Universitario de Ciencias Exctas e Ingenier a a as, Corregidora 500 S.R., Universidad de Guadalajara, 44420 Guadalajara, Jalisco, Mxico. e 5 Divisin de Ciencias e Ingenier de la Universidad de Guanajuato, A.P. 150, 37150, Len, Guanajuato, Mxico. o a o e (Dated: July 24, 2011) We apply the dynamical systems tools to study the (linear) cosmic dynamics of a Dirac-BornInfeld-type eld trapped in the braneworld. We focus, exclusively, in Randall-Sundrum and in DvaliGabadadze-Porrati brane models. We analyze the existence and stability of asymptotic solutions for the AdS throat and a particular relationship between the warp factor and the potential for the DBI eld (f () = 1/V ()). It is demonstrated, in particular, that in the ultra-relativistic approximation matter-scaling and scalar eld-dominated solutions always arise. While the eect of the Randall-Sundrum brane is to re-introduce the initial curvature singularity that was removed by the non-linear eect of the Dirac-Born-Infeld eld, the extra-dimensional DGP brane eects modify the stability of the critical points associated with the late-time cosmic dynamics. It is foud that the self-accelerating solution arising in the self-accelerating branch of the DGP scenario is a saddle equilibrium point, which means that this is not always the end point of the cosmic evolution in that branch of the model.
PACS numbers: 04.20.-q, 04.20.Cv, 04.20.Jb, 04.50.Kd, 11.25.-w, 11.25.Wx, 95.36.+x, 98.80.-k, 98.80.Bp, 98.80.Cq, 98.80.Jk
1

I.

INTRODUCTION

Recent observations from the Wilkinson Microwave Anisotropy Probe (WMAP) [1] oer strong supporting evidence in favor of the inationary paradigm. In the most simple models of this kind, the energy density of the universe is dominated by the potential energy of a single (inaton) scalar eld that slowly rolls down in its selfinteraction potential [2]. Restrictions imposed upon the class of potentials which can lead to realistic inationary scenarios, are dictated by the slow-roll approximation, and hence, the result is that only suciently at potentials can drive ination. In order for the potential to be suciently at, these conventional inationary models should be ne-tuned. This simple picture of the earlytime cosmic evolution can be drastically changed if one considers models of ination inspired in Unied Theories like the Super String or M-theory. The most appealing models of this kind are the Randall-Sundrum braneworld model of type 2 (RS2) [3] and Dvali-GabadadzePorrati (DGP) brane worlds [4].

In the RS2 model a single co-dimension 1 brane with positive tension is embedded in a ve-dimensional anti-de Sitter (AdS) space-time, which is innite in the direction perpendicular to the brane. In general, the standard model particles are conned to the brane, meanwhile gravitation can propagate in the bulk. In the low-energy limit, due to the curvature of the bulk, the graviton is conned to the brane, and standard (four-dimensional) general relativity laws are recovered. RS2 braneworld models have an appreciable impact on early universe cosmology, in particular, for the inationary paradigm. In eect, a distinctive feature of cosmology with a scalar eld conned to a RS2 brane is that the expansion rate of the universe diers at high energy from that predicted by standard general relativity. This is due to a term quadratic in the energy density that produces enhancing of the friction acting on the scalar eld. This means that, in RS2 braneworld cosmology, ination is possible for a wider class of potentials than in standard cosmology [5]. Even potentials that are not suciently at from the point of view of the conventional inationary paradigm can produce successful ination. At suciently low energies (much less than the brane tension), the standard cosmic behavior is recovered prior to primordial nucleosynthesis scale (T 1 M eV ) and a natural exit from ination ensues as

a Electronic

b Electronic c Electronic

address: address: address: d Electronic address: e Electronic address:

rigarcias@ipn.mx dgm@uclv.edu.cu tamegc72@gmail.com claudia.moreno@cucei.udg.mx iquiros6403@gmail.com

2 the eld accelerates down its potential [6].1 The DGP model describes a brane with 4D worldvolume, that is embedded into a at 5D bulk, and allows for infrared (IR)/large scale modications of gravitational laws. A distinctive ingredient of the model is the induced Einstein-Hilbert action on the brane, that is responsible for the recovery of 4D Einstein gravity at moderate scales, even if the mechanism of this recovery is rather non-trivial [11]. Nevertheless, studying the dynamics of DGP models continues being a very attractive subject of research [23]. It is due, in part, to the very simple geometrical explanation to the dark energy problem and the fact that it is one of a very few possible consistent IR modications of gravity that might be ever found. The acceleration of the expansion at late times is explained here as a consequence of the leakage of gravity into the bulk at large (cosmological) scales, which is just a 5D geometrical eect. Nonlinear scalar eld theories of the Dirac-Born-Infeld (DBI) type have attracted much attention in recent years due to their role in models of ination based on string theory. DBI ination [1317] is motivated by brane inationary models [18] in warped compactications [19]. These scenarios identify the inaton with the position of a mobile D-brane moving on a warped (compact) 6dimensional submanifold of spacetime (for reviews and references see [20]), which means that the inaton is interpreted as an open string mode. In these models, the warped space slows down the rolling of the inaton on even a steep potential, making easier ination. This slowing down can also be understood as arising due to interactions between the inaton and the strongly coupled large-N dual eld theory [17]. This scenario can naturally arise in warped string compactications [14]. Usually only eective four-dimensional DBI cosmological models are studied. It is our opinion that studying the impact higherdimensional brane eects have on the cosmic dynamics of DBI-type models, is a task worthy of attention. A DBI eld trapped in a RS brane could be a nice scenario to make early-time ination easier, both, because of the interaction of the inaton with the strongly coupled (large-N) dual eld theory, and because of the UV brane eects. Meanwhile a DBI scalar eld conned to a (selfaccelerating) DGP braneworld could be a useful arena where to address unied description of early-time ination, fueled by the slowing down eect of the warped space, and late-time speed-up, originated from UV leakage of gravity into the extra-space. Aim of this paper is, precisely, to investigate the dynamics of a DBI-type eld trapped in a Randall-Sundrum brane and in a Dvali-Gabadadze-Porrati braneworld, respectively, by invoking the dynamical systems tools. The study of the asymptotic properties of these models allows to correlate such important dynamical systems concepts like past and future attractors as well as saddle equilibrium points with generic cosmological evolution. In a sense the present work might be considered as a natural completion of the one reported in Ref. [15]. For this reason, as in the above reference, here we concentrate in the case of an anti-de Sitter (AdS) throat and quadratic self-interaction potential for the inaton and a particular choice of the warp factor and of the potential for the DBI eld (f () = 1/V ()). In addition to the scalar eld we also consider a background uid trapped on the braneworld. Through the paper we use natural units (8G = 8/m2 l = h = c = 1). P
II. DBI ACTION

In the region where the back-reaction and stringy physics can be ignored, the eective action for the DBI eld has the following form [17]: SDBI = d4 x |g|{f 1 () 1 + f ()()2 f 1 () + V ()}, (1)

where is the inaton, V () - its self-interaction potential, and f () is the warping factor. For a spatially at FRW metric ()2 = 2 , where the dot accounts for derivative with respect to the cosmic time. The equation of motion for the DBI inaton can be written in the following way: 3 f 2 f + 3 2 H + L 2f f2 ) ( f 3 +L V + 2 = 0, f where the Lorentz boost L is dened as

(2)

In this scenario, reheating arises naturally even for potentials without a global minimum and radiation is created through gravitational particle production [7] and/or through curvaton reheating [8]. This last ingredient improves the brane steep inationary picture [9]. Other mechanisms such as preheating, for instance, have also been explored [10].

L =

1 1 f ()2

(3)

Alternatively the equation of motion of the DBI-type eld can be written in the form of a continuity equation:

3 associated with a critical point in the phase space of the model, this means that, independent of the initial data, the universes dynamics will evolve for a suciently long time in the neighborhood of this solution, otherwise, it represents a quite generic phase of the cosmic dynamics. In the following subsections we keep the expressions as general as possible, and then, in section IV we substitute the above mentioned expressions for f () and V ().
A. DBI-RS Model

+ 3H( + p ) = 0,

(4)

where we have dened the following energy density and pressure of the DBI scalar eld: L 1 L 1 + V (), p = V (), f L f

(5)

respectively. In this paper we concentrate just in two particularly symple cases: i) AdS throat which amounts to consider f () = /4 , where in specic string constructions is a parameter which depends on the ux numbers [17],2 and a quadratic self-interaction potential V () = m2 2 /2. ii) Assumption of a particularly interesting relationship between the warp factor and the potential for the DBI eld: f () = 1/V ().
III. AUTONOMOUS SYSTEM

Here we will be concerned with the dynamics of a DBI inaton that is trapped in a Randall-Sundrum brane of type 2. The eld equations in terms of the FriedmannRobertson-Walker (FRW) metric are the following: 3H 2 = T (1 + T ), 2[ (6) (7)

T 2H = (1 + ) L 2 + (1 + m )m ,

The dynamical systems tools oer a very useful approach to the study of the asymptotic properties of the cosmological models [22]. In order to be able to apply these tools one has to follow these steps: i) to identify the phase space variables that allow writing the system of cosmological equations in the form of an autonomous system of ordinary dierential equations (ODE).3 , ii) with the help of the chosen phase space variables, to build an autonomous system of ODE out of the original system of cosmological equations, and iii) (a some times forgotten or under-appreciated step) to identify the phase space spanned by the chosen variables, that is relevant to the cosmological model under study. After this one is ready to apply the standard tools of the (linear) dynamical systems analysis. The goal of the dynamical systems study is to correlate such important concepts like past and future attractors (also, saddle critical points) with asymptotic cosmological solutions. If a given cosmological solution can be

where m is the equation of state (EOS) parameter of the matter uid, while T = + m the total energy density on the brane. Additionally one has to consider the continuity equations for the DBI-type eld (equation (2) or, alternatively, (4)) and for the matter uid: m + 3(1 + m )Hm = 0. (8)

The model described by the above equations will be called as DBI-RS model. Our aim will be to write the latter system of second-order (partial) dierential equations, as an autonomous system of (rst order) ordinary dierential equations. For this purpose we introduce the following phase space variables [15]: L L V T , y , z , r , 3f H 3H 2 3H f V (9) 1 3/2 1/2 , 2 3/2 5/2 . V f V f 1 x H It can be realized that, in terms of the variable r, 2(1 r) T = , 0 < r 1. r

(10)

In general, inationary observables may depend on the details of the warp factor [21], however, if we assume that the last 60 e-foldings of ination occur far from the tip of the throat, the above is a good approximation. There can be several dierent possible choices, however, not all of them allow for the minimum possible dimensionality of the phase space.

This means that the four-dimensional (low-energy/infrared) limit of the Randall-Sundrum cosmological equations corresponding to the formal limit is associated with the value r = 1, while the high-energy/ultraviolet limit 0, corresponds to r 0. We write the Lorentz boost in terms of the variables of phase space as:

4
B. DBI-DGP model

1 = L

y2 . 3x2

(11)

Standard non-relativistic behavior corresponds to = 1, while the ultra-relativistic (UR) regime is associated with = 0. In terms of the variables that span the phase space (x, y, z, r, 1 , 2 ), the cosmological equations (4), (6), (7), and (8), can be written as an autonomous system of ordinary dierential equations (ODE): yz 3 y2 H 1 x = (1 + 2 ) 2 x , 2 x 2x H z3 3 y = [1 ( 2 + 1) + 2 ( 1)2 ] 2 x 3 2 H ( + 1)y y , 2 H 1 yz 2 H 2r(r 1) H z = 1 z , r = , 2 x H 2r H [ ] yz 3 ln f = 2 V , 1 1 x 2 ln V 2 [ ] 2 yz 3 5 ln V 3 2 2 = f , x3 2 ln f 2

In this section we focus our attention in a braneworld model where the DBI inaton is conned to a DGP brane. In the (at) FRW metric, the cosmological (eld) equations are the following:

Q2 = m

1 (m + ), 3 = 3(1 + m )Hm ,

(20)

(12)

(13) (14) (15) (16)

where, as before m is the EOS parameter of the matter uid, m is the energy density of the background barotropic uid and is the energy density of DBI eld. Also one has to consider the continuity equations for the DBI-type eld (equation (2) or, alternatively, equation (4)). We have used the following denition:

Q2 H 2

1 H, rc

(21)

where the prime denotes derivative with respect to the number of e-foldings ln a, while V and: [ ]} H 2r { 2 y + 3(m + 1) r (1 )x2 z 2 . = H 2r (17) It will be helpful to have the parameters of observational importance = /3H 2 the scalar eld dimensionless energy density parameter, and the equation of state (EOS) parameter = p / , written in terms of the variables of phase space: (1 )x2 z 2 = (1 )x + z , = . (1 )x2 + z 2
2 2 2 2 (V V ) (f f ) , f , 2 ( V ) ( f )2

with rc being the so called crossover scale. In what follows we will refer to this model as the DBI-DGP model. There are two possible branches of the DGP model corresponding to the two possible choices of the signs in (21): + is for the normal DGP models that are free of ghost, while - is for the self-accelerating solution. As before, our goal will be to write the latter system of second-order (partial) dierential equations, as an autonomous system of (rst order) ordinary dierential equations. For this purpose we introduce the following phase variables: L L V Q , r , , y , z 3f Q H 3Q

1 x Q

(22)

The expression determining the Lorentz boost coincides with (11): 1 = L 1 y2 . 3x2

(18)

Recall, also, that the deceleration parameter q = (1 + H /H): q = 1 + 2r 2 {y + 3(m + 1)[r (1 )x2 z 2 ]}. (19) 2r

Hence, as before, = 1 is for the non-relativistic case, while = 0 is for the UR regime. After the above choice of phase space variables the cosmological equations can be written as an autonomous system of ODE:

5 the standard behavior typical of Einstein-Hilbert theory coupled to a self-interacting scalar eld. To nalize this section we write useful magnitudes of observational interest in terms of the dynamical variables (22). The eective dimensionless density parameters m = m /3Q2 and = /3Q2 :

] y [ Q xy + z 3 r(1 + 2 ) x . (23) 2 2x Q ] 3z 3 r [ y = 1 ( 2 + 1) + 2 ( 1)2 2x Q 3y (24) ( 2 + 1) y , 2 Q ( ) Q 1 r2 Q yz 2 r 1 z , r = r . (25) z = 2x Q 1 + r2 Q x = Recall that the prime denotes derivative with respect to the number of e-foldings ln a, while 1 V f , 2 3/2 5/2 . V 3/2 f 1/2 V f

m m = 2 = 1 x2 (1 ) z 2 , r = 2 = x2 (1 ) + z 2 . r For the equation of state (EOS) = p / :

(28)

We have considered also the following relationship: H 2r Q = , H 1 + r2 Q where Q 1 {3(1 + m )[1 x2 (1 ) z 2 ] + y 2 }. Q 2

C.
2

x2 (1 ) z 2 . x2 (1 ) + z 2

(29)

Brane Eects in the Phase Space

(26)

Equations (23)-(25) have to be complemented with the addition of equations (15) and (16) above, which are the autonomous ordinary dierential equations for 1 and 2 , respectively. The equation (21) can be rewritten as: r2 = 1 1 . rc H

(27)

For the Minkowski phase, since 0 H (we consider just non-contracting universes), then 1 r . The case r 1 corresponds to the time reversal of the latter situation. For the self-accelerating phase, r2 1, but since we want real valued r only, then 0 r2 1.4 As before, the case 1 r 0 represents time reversal of the case 0 r 1 that will be investigated here.5 Both branches share the common subset (x, y, z, r = 1), which corresponds to the formal limit rc (see equation (21)), i. e., this represents just

Higher-dimensional (brane) eects are encrypted in the variable r for the DBI-RS model as well as for the DBIDGP model. In both cases the hyper-surfaces in the phase space, that are foliated by the value r = 1: = (x, y, z, r = 1, 1 , 2 ), represent the loci of the equilibrium congurations that can be associated with standard general relativity behavior. For the DBI-DGP model the phase space hyper-plane represents, additionally, the boundary in the phase space separating the Minkowski cosmological phase (r 1) from the the selfaccelerating one (0 r 1). Trajectories in the phase space that scape from are associated with higherdimensional modications of general relativity produced by the brane eects.

IV.

EQUILIBRIUM POINTS IN THE PHASE SPACE

1 In fact, tting SN observations requires H rc in order to achieve late-time acceleration (see, for instance, reference [25] and references therein). This means that r has to be real-valued. Points with r = 0 and their neighborhood have to be carefully analyzed due to the fact that at r = 0, other phase space variables (see Eq.(22)) and equations in (23-25) are undened in general.

In this section we will analyze in detail the existence and stability of critical points of the autonomous systems corresponding to both Randall-Sundrum and DvaliGabadadze-Porrari braneworld models. In both cases we study i) an AdS throat often explored in the literature f () = /4 , and a quadratic potential: V () = m2 2 /2 [15], and ii) a particular relationship between the warp factor and the potential for the DBI eld: f () = 1/V ().

6
TABLE I: Properties of the critical points for the autonomous system (30). Equilibrium point S M U x 0 0 1 y 0 0 z 0 0 0 r 0 1 1 Existence Always Undened Undened 0 m 0 1 0 0 0 1 Undened Undened 0 q 3m + 2 (3m + 1)/2 1/2

TABLE II: Eigenvalues of the linearization matrices corresponding to the critical points in table I. Equilibrium point S M U 1 3m 3(1 + m ) 3m 2 3(m + 1) 3(m 1)/2 3 3 3(m + 1) 3(m + 1)/2 3/2 4 3(m + 1) (3m + 1)/2 3

A.

AdS throat and a quadratic potential

autonomous dynamical system (12-16) can be reduced down to a four-dimensional one: yz 2 y2 H (z 2x2 ) x , 2x2 2x H 3(1 + 2 )z 3 y = 3(1 )2 xz 2x 3(1 + 2 )y H y , 2 H yz 2 H 2r(r 1) H z = z , r = , 2x H 2r H x =

Hence, the following relationship between f and V : m4 , f= 4V 2 is obtained. The above choice leads to 1 = being a constant = 2/(2/m), while 2 = 2(x2 /z 2 ). Consequently, only two of the phase space variables x, z, and 2 , are independent (say x and z). This fact leads to considerable simplication of the problem since the dimension of the autonomous system reduces from six down to four. This is one of the reasons why the present particular case is generously considered in the bibliography.6
1. DBI-RS Model

(30)

where the ratio H /H = (q + 1) is given by Eq. (17). It arises the following constraint: m = m = r (1 )x2 z 2 . 3H 2 (31)

Since m 0, then (1 )x2 + z 2 r. Besides, since m 1 then (1 )x2 + z 2 r 1. We will focus on expanding FRW universes, so that x 0 and z 0. The resulting four-dimensional phase space for the DBI-RS model is the following: = {(x, y, z, r) : r 1 (1 )x2 + z 2 r, x 0, y 2 3x2 , z 0, 0 r 1}. (32) As properly noticed in the former section equilibrium points lying on the hyper-plane = (x, y, z, r = 1) are associated with standard general relativity dynamics. The remaining points in are related with higherdimensional brane eects.

As just noticed, after considering the specic form of the functions f () and V () above, the six-dimensional

The importance of the quadratic potential in the cosmological context is remarkable [26, 27].

7
TABLE III: Properties of the equilibrium points of the autonomous system (33). Here m = m +1, while ( Equilibrium point S M K SF MS y 0 0 z 0 0 0
(2 3) 3m

2 + 12). q

r 0 1 1 1 1

Existence Always 0 1 < m < 0

m 0 1 0 0 1+
2 3m 2 m

0 0 1 1
m 2 m

Undened Undened 0 1 +
6

3m + 2
3m +1 2 1 2

3 2 3m m m

3 2

m 2m +1

3m +1 2

TABLE IV: Eigenvalues of the linearization matrices corresponding to the equilibrium points in table III. The eigenvalues corresponding to the fourth point (SF ) in Tab.III, have not been included due to their overwhelming complexity. Here 2 3 2 2 (9m + 6m + 1) + 24(m + 3m + 3m + 1). Equilibrium point S M K MS 1 3(m + 1) 3(m + 1) 3m 3m
3 (1 4

2 3(m + 1) 3(m + 1)/2 3 m + /)

3 3(2m + 1)/2 3m /2 3/2 3 (1 m /) 4

The relevant equilibrium points of the autonomous system of equations (30) are summarized in table I, while the eigenvalues of the corresponding linearization (Jacobian) matrices are shown in table II. The source critical point S in table I is the past attractor of the RS cosmological model. It is correlated with a decelerated solution whenever m > 2/3, otherwise it is an accelerated solution corresponding to a saddle critical point. Existence of this solution is a distinctive feature of the higher-dimensional (brane) contributions. Actually, according to equation (10), for nite non-vanishing T the case with r = 0 corresponds to the formal limit 0, i. e., the high-energy (ultra-violet) limit of the RS brane model. Since in this case x = y = z = r = 0, assuming nite non-vanishing f , L , V and total energy density on the brane T , amounts to assuming that the Hubble parameter has to be neccessarily innitely large H , a situation associated with the initial curvature singularity characterized by nite (non-null) energy density on the braney. Notice that although in this limit both and m vanish (in consequence T also vanishes), the usual denition for is irrelevant since the total T is not constrained to be 1 as in usual cosmology, so that there is no contradiction with the assumption that

T might be nite and non-vanishing in the 0-limit.7

10,0 7,5 5,0 r 0,55 t 2,5 0,0 2,5 0,42 5,0 0,05 0,3 0,55 y 0,8 0,22 0,02 1,05

0,8

0,3 0,62 0,05 0,05 0,3 0,42 0,22 0,55 y 0,8 0,02 1,05 z

0,62 z

This argument was suggested to us by one of the referees.

8
FIG. 1: Phase portrait generated by a given set of initial data (m = 1/3, = 1) for the autonomous system of ODE (33), corresponding to the UR approximation of the DBI-RS model left-hand panel, and the corresponding ux in time right-hand panel. The values of the free parameter has been chosen so that the scalar eld-dominated solution (point SF (1.14, 0.75, 1)) is the late-time attractor.

The matter-dominated solution (equilibrium point M ) and the ultra-relativistic (scalar eld-dominated) phase U , are always saddle critical points and are associated with standard general relativity dynamics. The latter is associated with decelerating dynamics while the former M represents decelerating expansion whenever m > 1/3 and inationary solution otherwise. Points M and U in Tab. I correspond to the equilibrium points A and B of Ref. [15], respectively. More detailed information can be retrieved only after further simplication of the case to study. One way to achieve further simplication is to study the ultrarelativistic regime where = 0 L = . In the UR regime, thanks to the relationship = 0 x = y/ 3, the autonomous system of ODE (30) simplies down to a set of three ordinary dierential equations: 3 3z 3 3y H y = y , 2y 2 H 2 3z H 2r(r 1) H z = z , r = , 2 H 2r H H 2r = {(m + 1)[3r y 2 3z 2 ] + y 2 }. (33) H 2r

The critical point S, which represents a singular cosmological solution, is associated with 5D brane eects since, the only way the Hubble parameter H can be arbitrarily large while keeping T nite, is that the brane tension 0, so that H 2 2 / . Provided that m > 1/2 this equilibrium point is always the past attractor in the phase space, i. e., it represents the source critical point from which any phase path in originates. It corresponds to decelerated expansion. In section V we will comment in more detail the dierence between this singular solution and the standard big-bang arising in general relativity. The matter-scaling solution (equilibrium point M S) is the late-time attractor provided that (the denition of the parameter can be found in the caption of the table IV) (1 m ) < < (1 m ). Otherwise, the scalar eld-dominated solution SF is the late-time attractor. We have to point out that the matter-scaling solution exists only if the equation of state (EOS) parameter of the background matter is a negative quantity: 1 < m < 0. This means that we can not have matter-scaling with background matter being dust (m = 0). Therefore, the usefulness of this equilibrium point to describe the current phase of the cosmic evolution is unclear. Unlike this, the scalar eld-dominated solution SF is always inationary ( > 0 always) and could be associated with accelerated late-time cosmic dynamics. The matterdominated solution M and the kinetic energy-dominated phase (equilibrium point K in Tab. III), are always saddle points in the phase space. Worth noticing that the scalar eld-dominated solution (point SF in Tab.III) and the matter-scaling phase M S correspond to the points C and D in reference [15]. In the standard 4D limit of the theory (r = 1) we obtain the same behavior that was obtained in [15, 27]. In the Figure 1 the trajectories in the reduced phase space dened in (34) originated by a given set of appropriated initial data, are drawn for the model of (30) in the ultra-relativistic approximation. The phase space pictures in the gures 1 reveal the actual behavior of the RS dynamics: trajectories in phase space depart from the initial curvature singularity, and, at late times, approach to the plane (y, z, 1), which is associated with standard four-dimensional behavior, in particular the trajectories approach to the SF point.
2. DBI-DGP Model

The reduced (three-dimensional) phase space in this simpler case is given by: = {(y, z, r) : 3(r 1) y 2 + 3z 2 3r, y , z 0, 0 < r 1}.

(34)

There are found ve equilibrium points of the autonomous system of ODE (33) in . These points, together with their properties, are listed in Tab.III, while the eigenvalues of the corresponding linearization matrices are shown in the table IV. By their overwhelming complexity, the eigenvalues of the linearization matrix corresponding to the fourth equilibrium point in table III (point SF ) have not been included in Tab.IV. It is worth recalling that the standard general relativity behavior is associated with points in the phase space lying on the phase plane = (y, z, r = 1).

For an AdS throat and the quadratic self-interaction potential, the autonomous system of ODE (23)-

9
TABLE V: Properties of the equilibrium points of the autonomous system (35). Equilibrium point SA M U x 0 0 1 y 0 0 z 0 0 0 r 0 1 1 Existence Always Undened 0 m 0 1 0 0 0 1 Undened 0 q 1 (1 + 3m )/2 1/2

TABLE VI: Eigenvalues of the linearization matrices corresponding to the critical points in table V. Equilibrium point SA M U 1 3(1 + m )/2 3(m 1)/2 3m 2 3(1 + m )/2 3(1 + m )/2 3 3 3(1 + m )/2 3(1 + m )/2 3/2 4 3(1 m )/2 3(1 + m )/2 3/2

(25), (15), (16), reduces down to the following fourdimensional autonomous system: x = y2 yzr 2 Q (z 2x2 ) x , 2x 2x2 Q 3( 2 + 1) y = 6 2 xzr [(z 2 2x2 )zr + 2x Q +xy] y , Q ( ) Q 1 r2 Q yz 2 r z , r =r . (35) z = 2x Q 1 + r2 Q

that can be associated with extra-dimensional eects. The remaining points lie on the hyper-plane related with standard general relativity behavior. The matterdominated solution (critical point M in Tab. V) is the past attractor if m > 1, otherwise it is also a saddle critical point.9 The ultra-relativistic regime, which is dominated by the scalar eld, mimics the cosmic evolution of a universe lled with dust. It can be the past attractor only for negative m < 0. For > 0 it is a saddle equilibrium point in the phase space. The above results are to be contracted with the results in the former subsection.

The ratio Q /Q is given by Eq. (26). Recall that standard general relativity behavior is associated with points liying on the hyper-plane = (x, y, z, r = 1). The remaining points in the bulk of the phase space are associated with higher-dimensional (brane) eects. The critical points of the autonomous system of ODE (35), together with their most important properties, are summarized in the table V. The eigenvalues of the linearization matrices corresponding to the critical points in Tab. V are shown in table VI. Three critical points are found: the self-accelerating solution (point SA), the matter-dominated solution (M ), and the UR phase (U ) which is dominated by the DBI scalar eld. The critical point SA the self-accelerating solution is always a saddle point and represents an inationary solution. Actually, for r = 0 one gets a de Sitter solution H = 1/rc .8 This is the only equilibrium point

As before, a more detailed study of the asymptotic properties of the model (35) requires additional simplication. The ultra-relativistic approximation comes to our rescue. As long as one considers just large Lorentz boosts (amounting to vanishing ) the relationship x = y/ 3 is veried. This relationship allows for further simplication of the autonomus system of ODE (35). Actually, in the UR regime the above system of equations can be simplied to the following three-dimensional autonomous system of ODE:

9 8

In fact, tting SN observations requires H 1/rc in order to

achieve late time acceleration (see, for instance, reference [28] and references therein). We do not consider unphysical situations where the EOS parameter for the background uid < 1.

10
TABLE VII: Properties of the equilibrium points of the autonomous system (36). Here ( Equilibrium point M K SF MS y 0 3 2 3m m m z 0 0
(2 3) 3m

2 + 12), while m m +1. Undened 0 1 + m

6

r 1 1 1 1

Existence 0 1 < m < 0

m 1 0 0 1+
2 3m 2 m

0 1 1
m 2 m

q
3m +1 2 1 2

3 2

3m +1 2

TABLE VIII: Eigenvalues of the linearization matrices corresponding to the critical points in table VII. 12m (6m + ) + 2 (6 ), while 24(1 + m )3 + 2 (1 + 3m )2 . Equilibrium point M K SF MS 1 3(m + 1)/2 3m /4
3 (1 2

Here

2 3(m + 1)/2 3/2

12(2+m )+3 8 3 4

3 3m /2 3/2 +
2 2 8

+ m )

( m 1 +

2 8

12(2+m )+3 8 3 4

( m 1

3 3z 3 r Q 3 y , y = y 2 2y Q 3 2 Q z = z r z , 2 Q 2 r(1 r ) Q r = . (1 + r2 ) Q

(36)

The phase space for the autonomous system (36) can be dened in the following way. For the + - branch (the Minkowski cosmological phase): + = {(y, z, r) : 0 y 2 + 3z 2 3, y , z 0, r [1, )},

(37)

a expected result since in the UR limit the DBI eects dominate over the brane eects at late times. These critical points together with their most salient features are summarized in table VII. If, m > 0, the matterdominated solution (point M ) is the past attractor, while for negative m < 0, the kinetic energy-dominated solution (point K) is the past attractor in the phase space (see in [15, 27]). The latter result is independent on which branch of the DGP is being considered since, for DGP models, at early times, the brane eects can be safely ignored which means that the standard GR cosmological dynamics is not modied. The equilibrium points SF (scalar eld-dominated solution) and M S (scaling dominated-solution) are always a saddle critical points. There is no future (late-time) attractor in the phase space of the model.

while, for the self-accelerating - - branch, it is given by: = {(y, z, r) : 0 y 2 + 3z 2 3, y , z 0, r [0, 1]}.

B.

Special relationship f () = 1/V ()

(38)

There are found four equilibrium points of the autonomous system of ODE (36). None of them is associated with extra-dimensional eects since all of the critical points lie on the phase hyper-plane with r = 1. This is

This case was studied in [23] where the authors used an approach based on an existing (formal) mathematical equivalence between the DBI model and standard tachyon cosmology, under an appropriate transformation of the DBI eld. The above relationship (f () = 1/V ()) leads to signicant simplication of the eld equations. The equation of motion of the DBI-type eld and the

11 conservation equation are the following: + 3 2 H = V (1 32 /2V ), where the modied Lorentz factor is dened as: 1 = , 1 2 /V (39) In terms of the above phase space variables the constraint (Friedmann) equation can be written as: m = r z2 . 1 a2

(40)

Other cosmological parameters such as the DBI-eld energy density parameter , the equation of state parameter , and the deceleration parameter q look like: z2 , = a2 1 1 a2 ] (2 r) [ q = 1 + 3m r + 3z 2 (m a2 ) . (43) 2r The phase space spanned by the above variables is dened as follows: = = {(a, z, r) : |a| 1, 0 z 4 1 a2 , 0 r 1}. For an exponential self-interaction potential of the form: V () = V0 exp(), since ln V = = const, then the system (42) is a closed autonomous system: a = (a2 1)[3a 3z] [ ] 3 z(2 r) 3z 2 (m a2 ) 2 z = az 3m r 4 2r 1 a2 [ 2 ] 3z (m a2 ) r = (r 1) 3m r . (44) 1 a2 The critical points corresponding to the exponential potential are summarized in Table IX, while their respective eigenvalues are displayed in Table X. These points coincide with the ones reported in [23] but for the additional high-energy/ultra-violet critical point S related with the brane eects. While the points M , T and M S show similar stability properties than reported in [23], the stability of the critical point U diers in that it is a saddle critical point unlike in standard cosmology where it was a inationary past attractor. In Figure 2 we show the trajectories in phase space for dierent sets of initial conditions. As clearly seen from this gure, the trajectories in phase space emerge from the source critical point S = (a, z, r) = (1, 0, 0) the kinetic/potential energy-scaling solution meaning that this is the past attractor of the Randall-Sundrum cosmological model. This equilibrium point is associated with a initial curvature singularity of higher-dimensional origin since, assuming nite (non-null) V , and T , a = 1, z = r = 0 H .

and where we have dened the following energy density and pressure of the DBI scalar eld: = V () , p = 1 V (). From now on we shall call the model given by equations (39), (40), as modied tachyon cosmology (MTC). by assuming f () = 1/V () the autonomous system of ordinary dirential equations is a three-dimensional one since we can introduce a new dimensionless variable a y/x (1 = V /V = 2 ).
1. DBI-RS Model

With the aim to write an autonomous system out of the cosmological equations we introduce the following dimensionless phase space variables: V T a= , z= , r= . 3H 2 3H V As before 2(1 r) T = , 0 < r 1. r Then we can write the above system as: a = (a2 1)[3a + 3 (ln V )z] 3 2 az (ln V ) z = 4 [ ] z(2 r) 3z 2 (m a2 ) 3m r 2r 1 a2 ] [ 2 3z (m a2 ) 3m r r = (r 1) 1 a2

(41)

(42)

where m = m +1 is the barotropic index for the matter (0 m 2), and the tilde accounts for derivative in respect to the cosmological time .

12
TABLE IX: Properties of the critical points for the autonomous system (42). Where z = Equilibrium Point S M U MS T a 1 0 1 m
z 3

36+4 2 . 6

z 0 0 0
3m

r 0 1 1 1 1

Existence Always

undened 0 0
3m 12 m

0 1 0 m 1
2 2 z 3

q undened 1 + 1 + 1 +
3m 2 3m 2 3m 2

1 + 3m /2

TABLE X: Eigenvalues of the linearization matrices corresponding to last four equilibrium points in Tab.IX. 2 48m 1m Here + 4 + m (17m 20). 2 Equilibrium Point M U MS T
2

1 3 3m
(

2 3m 3m /2
)

3 3m /2 6 3 [(2 m ) ] 4 ) 3m +
2 12

3m
4 +362 12

3 [(2 m ) + ] 4 3 +
2 12

4 + 36 2

4 + 36 2

2.
10,0 0,95 0,85 0,75 0,65 0,55 r 0,45 0,35 0,25 0,84 0,15 0,64 0,05 0,24 0,3 0,55 a 0,8 0,04 0,44 z 0,05 2,5 0,64 0,44 z 0,08 0,24 0,28 0,48 a 0,68 0,04 0,88 0,84 t 2,5 5,0

DBI-DGP Model

7,5

Here we use the same variables a and z, and redene r = Q/H. In terms of these variables we can write the following system of ODE: a = (a2 1)[3a + 3 (ln V )z] [ ] 3 2r2 Q z =z az (ln V ) 2 4 r 1 Q r = r(1 r2 ) Q r2 + 1 Q (45)

0,0

5,0

FIG. 2: Phase portrait generated by a given set of initial data (m = 1, = 1) for the autonomous system of ODE (44), corresponding to the UR approximation of the DBI-RS model left-hand panel, and the corresponding ux in time right-hand panel. The trajectories in phase space emerge from the point S = (a, z, r) = (1, 0, 0) the kinetic/potential energy-scaling solution.

where [ ] 3 z 2 (a2 m ) Q 2 = 2 m r + . Q 2r 1 a2 The DBI-eld energy density parameter , and the equation of state parameter are the same than in the former subsection. The deceleration parameter q is: q = 1 2r2 Q . r2 + 1 Q

13
TABLE XI: Eigenvalues of the linearization matrice corresponding to the equilibrium points U , T and M S to the system 45. 2 48m 1m Here + 4 + m (17m 20). 2 Equilibrium Point U MS T
2

1 3m /2
( 3m 2 4 +362 12 )

2 3m /2 3 4 3 + [(2 m ) + ] ( 4 + 36 2 ) 3 4 3m +

3 6 [(2 m ) ]
2 12

2 12

4 + 36 2

The phase space for the autonomous system (45) can be dened in the following way. For the + - branch (the Minkowski cosmological phase): + = {(a, z, r) : |a| 1, 0 z 4 + a2 1, z 0, r [1, )}, (46) while, for the self-accelerating - - branch, it is given by: = {(a, z, r) : |a| 1, 0 z 4 + a2 1, y , z 0, r [0, 1]}. (47) The system (45) is not a closed autonomous system and, for this reason, we focus in the exponential potential, V = V0 exp(), for which V /V = , and the system of ODE is closed. We obtain the same critical points M , U , M S and T , that were obtained in the former subsection within the Randall-Sundrum Model. However in this case the stability of the critical points U , T and M S dier from the ones in the former subsection as shown in the table XI. The equilibrium point U = (1, 0, 1) represents an inationary solution: a past attractor in the phase space. This point represents a scaling of the potential and of the kinetic energy of the scalar eld. These results coincide with the ones reported in [23]. On the contrary, the matter-scaling solution is always a saddle point unlike in the case explored in [23] were this point (whenever it exists) was a stable equilibrium point the future attractor. The equilibrium point T (the tachyon-dominated solution) is the late-time attractor in the present case.

the Randall-Sundrum brane model produces UV modications of general relativity, in the Dvali-GabadadzePorrati braneworld IR modications of the laws of gravity arise instead. In a similar way, the introduction of a non-linear Dirac-Born-Infeld type of eld might modify the cosmic dynamics at early as well as at late times. This makes even more interesting the study of the combined eect of a DBI-type eld trapped in the braneworld. Aim of the present paper has been, precisely, the study of the asymptotic properties of the latter kind of cosmological models.

A.

RS Model

V.

RESULTS AND DISCUSSION

The main result of Sec.IV A 1 and Sec.IV B 1 can be summarized as follows: The initial curvature singularity (critical point S in Tab.I) is the past attractor in the phase space, it represents the source critical point from which any phase path in originates. For a wide range of parameters it is a decelerated solution. We have to dierentiate this cosmological singularity from standard general relativity big-bang singularity. Actually, in a standard 4D model (no brane eects), the big-bang singularity which is characterized by innitely large H and total energy density T , is removed by the eect of the DBI eld. This can be corroborated by noting that on the hyper-surface where GR solutions lie, there are no critical points that can be associated with a curvature singularity.10 What the results of our study show is that the eect of the RS brane is to restore the occurrence of a curvature singularity at very high energy/curvature. The fact that the latter (initial, cosmological) singularity is correlated with a past attractor means that the bigbang is a generic property of DBI-RS models. For the

The importance of the brane eects for the cosmic dynamics is well known. These eects can modify the general relativity laws at early times (UV modications), as well as at late times (IR modications). Actually, while

10

Absence of critical points that can be correlated with a curvature singularity does not mean that particular solutions with curvature singularities might arise.

14 particularly simple relationship between the warp factor and the potential of the DBI eld f () = 1/V () (the exponential potential was considered), the past attractor is the kinetic/potential energy-scaling solution with the occurrence of a curvature singularity due to the UV brane eects. In the Sec.IV A 1 the matter dominated solution (critical point M in Tab.I) and the ultra-relativistic (scalar eld-dominated) phase (critical point U ) are always saddle critical points and are associated with standard general relativity dynamics. The late-time behavior in Sec.IV A 1 is correlated with the critical points M S (matter-scaling solution) and SF (scalar eld solution). The point M S is the late-time attractor provided that (1 m ) < < (1 m ).11 Otherwise, the scalar eld-dominated solution SF is the late-time attractor. This result coincides with the one obtained within the standard 4D limit of the theory and reported in [15, 27]. The critical points M , T and M S in Sec.IV B 1 have similar stability properties than in standard cosmology [23]. On the contrary, the stability of the point U diers from the one in standard cosmology where it was a inationary past attractor. In the case where the RS brane eect is considered this is a saddle critical point instead. In general, the dynamical behavior of the RandallSundrum model diers from the standard behavior within four-dimensional Einstein-Hilbert gravity only at early times (high-energy regime). The existence of the initial curvature singularity (the past attractor S) is to be contrasted with the standard four-dimensional result within general relativity fuelled by a DBI scalar eld (and a background uid), where either the uid-dominated or the kinetic-dominated, singularity-free solutions can be the past attractor [15, 23, 27]. The late-time cosmological dynamics, on the contrary, is not aected by the RS brane eects in any essential way. the DGP model can be correlated either with the matterdominated solution or with the kinetic energy-dominated solution as in [15, 27]. This result is independent of which branch of the DGP is being considered since at early times the DGP brane eects can be safely ignored, i. e., the standard cosmological dynamics is not modied. There is no future (late-time) attractor in the phase space of the model. The scalar eld-dominated solution (equilibrium point SF ) and the matter-scaling solution (equilibrium point M S) are always saddle equilibrium points. In Sec.IV B 2 we obtained the same four critical points correlated with GR solutions, that were obtained in the Randall-Sundrum model (Sec.IV B 1). However, in this case the stability properties of the critical points U , T and M S have been modied by the DGP brane eects. The latter result is independent of which branch of the DGP is being considered. The critical point U in Sec.IV B 2 a past attractor in the phase space represents an inationary solution. This point represents a scaling of the potential and of the kinetic energy of the DBI scalar eld. The stability properties are the same that were found and reported in [23]. Summing up: the eect of the DGP brane is to modify the late-time cosmological dynamics through changing the stability of the corresponding (late-time) critical points. Actually, in the present case the scalar elddominated solution, as well as the scaling dominatedsolution, are always saddle critical points. This result has to be confronted with the classical general relativity result where the above-mentioned solutions can be latetime attractors.12

VI.

CONCLUSION

B.

DGP Model

The following important results can be summarized in this case: The new ingredient added by the DGP brane eect is the inationary self-accelerating solution (critical point SA). It is always a saddle point in the phase space. This means that the self-accelerating solution is not always the end point of the cosmic evolution even for the self-accelerating branch of the DGP model. The matterdominated, decelerated solution (critical point M ) is always a saddle critical point. The early time behavior in

In the present paper we aimed at studying the asymptotic properties of a DBI-type eld trapped in the braneworld for two particularly interesting and somewhat simple cases: i) an AdS throat (f () = /4 ) and a quadratic potential (V () = m2 2 /2) [15], and ii) assumption of a special relationship between the warp factor and the potential for the DBI eld: f () = 1/V (). The combined eect of the non-linear nature of the DBI eld and of the higher-dimensional brane eects seem to produce a rich dynamics. Both brane contributions and nonlinear DBI eects can modify the general relativity laws of gravity at late times as well as at early times. Here we focused in Randall-Sundrum and in Dvali-Gabadadze-

12 11

The denition of the parameter can be found in the caption of the table IV.

In the DBI-DGP case, at early times, the dynamics is general relativistic so that the stability properties of the matter-dominated phase in tables III and IV just coincide with the results of [15, 27].

15 Porrati braneworlds exclusively. In a sense this work can be considered as a natural completion of the Ref. [15, 27] to consider the combined eect of the DBI-type eld and of the braneworld. We performed a thorough study of the phase space corresponding to the above two scenarios. It is revealed that RS brane eects conspire to re-introduce the initial curvature singularity this time of higher-dimesnional origin that was removed by the non-linear eects of the DBI eld at very high energy. At the oposite end of the energy/curvature spectrum, the DGP brane eects modify the late-time cosmic dynamics: while in the selfaccelerating branch the self-accelerating solution is correlated with a saddle critical point (point SA), in general the stability of the standard general relativity solutions is modied by the extra-dimensional eects. The present study can be considered as a natural complection of the work of Ref. [15]. This work was partly supported by CONACyT Mxico, under grants 49924-J, 52327, 105079, Instituto e Avanzado de Cosmologia (IAC) collaboration. R G-S acknowledges partial support from COFAA-IPN, EDIIPN and IPN grant SIP-20100610. D G, and T G aknowledge also the MES of Cuba for partial support of the research.

[1] C. L. Bennett et al., Astrophys. J. Suppl. Ser. 148 (2003) 1; G. Hinshaw et al., Astrophys. J. Suppl. Ser. 148 (2003) 135; D. N. Spergel et al., Astrophys. J. Suppl. Ser. 148 (2003) 175; H. V. Peiris et al., Astrophys. J. Suppl. Ser. 148 (2003) 213; A. Kogut et al., Astrophys. J. Suppl. Ser. 148 (2003) 161; E. Komatsu et al., Astrophys. J. Suppl. Ser. 148 (2003) 119 [2] A. A. Starobinsky, Phys. Lett. B 91 (1980) 99; A. H. Guth, Phys. Rev. D 23 (1981) 347; A. Albrecht, P. J. Steinhardt, Phys. Rev. Lett. 48 (1982) 1220; A. D. Linde, Phys. Lett. B 108 (1982) 389; Phys. Lett. B 129 (1983) 177. [3] L. Randall, R. Sundrum, Phys. Rev. Lett. 83 (1999) 3370. [4] G. R. Dvali, G. Gabadadze, M. Porrati, Phys. Lett. B485 208-214 (2000), [hep-th/0005016]. [5] R. M. Hawkins, J. E. Lidsey, Phys. Rev. D 63 (2001) 041301. [6] G. Huey, J. E. Lidsey, Phys. Lett. B514 (2001) 217. [7] L. H. Ford, Phys. Rev. D35 (1987) 2955. [8] B. Feng, M. Li, Phys. Lett. B564 (2003) 169. [9] A. R. Liddle, L. A. Urena-Lopez, Phys. Rev. D68 (2003) 043517. [10] M. Sami, V. Sahni, Phys. Rev. D70 (2004) 083513. [11] C. Deayet, G. R. Dvali, G. Gabadadze, A. I. Vainshtein, Phys. Rev. D65 (2002) 044026 [hep-th/0106001]; A. Nicolis, R. Rattazzi, JHEP 0406 (2004) 059, [hepth/0404159]. [12] I. Quiros, R. Garc a-Salcedo, T. Matos, C. Moreno, Phys. Lett. B 670 (2009) 259265. [13] E. Silverstein, D. Tong, Phys. Rev. D 70 (2004) 103505; M. Alishahiha, E. Silverstein, D. Tong, Phys. Rev. D 70 (2004) 123505. [14] X. Chen, Phys. Rev. D 71 (2005) 063506 [hepth/0408084]; JHEP 08 (2005) 045 [hep-th/0501184]. [15] Z-K. Guo, N. Ohta,JCAP 04 (2008) 035. [16] M. C. Bento, O. Bertolami, A. A. Sen, Phys. Rev. D 67 (2003) 063511; X. Chen, Phys. Rev. D 71 (2005) 063506; S. E. Shandera, S-H. H. Tye, JCAP 05 (2006) 007.

[17] X. Chen, M. Huang, S. Kachru, G. Shiu, JCAP 01 (2007) 002. [18] G. R. Dvali, S. H. H. Tye, Phys. Lett. B 450 (1999) 72 [hep-ph/9812483]. S. Kachru, R. Kallosh, A. Linde, J. M. Maldacena, L. McAllister, S. P. Trivedi, JCAP 0310 (2003) 013 [hep-th/0308055]. [19] H. Verlinde, Nucl. Phys. B 580 (2000) 264 [hepth/9906182]; S. Gukov, C. Vafa, E. Witten, Nucl. Phys. B 584 (2000) 69 [hep-th/9906070]; S. Gukov S, C. Vafa, E. Witten E, Nucl. Phys. B 608 (2001) 477 (erratum); K. Dasgupta, G. Rajesh, S. Sethi, JHEP 08(1999) 023 [hepth/9908088]; B. R. Greene, K. Schalm, G. Shiu, Nucl. Phys. B 584 (2000) 480 [hep-th/0004103]; S. B. Giddings, S. Kachru, J. Polchinski, Phys. Rev. D 66 (2002) 106006 [hep-th/0105097]. [20] J. M. Cline, hep-th/0612129; S. H. Henry Tye, Lect. Notes Phys. 737 (2008) 949-974 [hep-th/0610221]. L. McAllister, E. Silverstein, Gen. Rel. Grav. 40 (2008) 565-605 (2008) [arXiv:0710.2951]. [21] S. Kecskemeti, J. Maiden, G. Shiu, B. Underwood, JHEP 0609, (2006), 076, [arXiv:hep-th/0605189]. [22] A. A. Coley, Dynamical systems and cosmology, Dordrecht-Kluwer, Netherlands (2003). [23] I. Quiros, T. Gonzalez, D. Gonzalez, Y. Napoles, R. Garcia-Salcedo, C. Moreno, Class. Quantum Grav. 27 (2010) 215021, [arXiv:0906.2617]. [24] W. Fang, Y. Li, K. Zhang, H.-Q. Lu, Class. Quant. Grav. 26 (2009), 155005, [arXiv:0810.4193]. [25] K. Koyama, Class. Quantum Grav. 24 (2007) R231 [arXiv:0709.2399]. [26] C. Ahn, C. Kim, E. V. Linder, Phys. Rev. D80, (2009), 123016, [arXiv:0909.2637]. [27] E. J. Copeland, S. Mizuno, M. Shaeri, [arXiv:1003.2881] [28] K. Koyama, Class. Quantum Grav. 24 (2007) R231, [arXiv:0709.2399v2] [29] T. Gonzalez, T. Matos, I. Quiros, A. Vazquez-Gonzalez, Phys. Lett. B 676 (2009) 161167.