Skip to main content

Full text of "Nonsingular 4d-flat branes in six-dimensional supergravities"

See other formats


CCNY-HEP-04/7 
August 2004 



Nonsingular 4d-flat branes in six-dimensional 

supergravities 



V. P. NAIR 

Physics Department 
City College of the CUNY 
New York, NY 10031 
E-mail: vpn@sci.ccny.cuny.edu 



S. RANDJBAR-DAEMI 

Abdus Salam International Centre for Theoretical Physics 

Trieste, Italy 
E-mail: seif@ictp.trieste.it 



Abstract 

We show that six-dimensional supergravity models admit nonsingular so- 
lutions in the presence of flat three-brane sources with positive tensions. 
The models studied in this paper are nonlinear sigma models with the target 
spaces of the scalar fields being noncompact manifolds. For the particular 
solutions of the scalar field equations which we consider, only two brane 
sources are possible which are positioned at those points where the scalar 
field densities diverge, without creating a divergence in the Ricci scalar or 
the total energy. These solutions are locally invariant under | of D=6 super- 
symmetries, which, however, do not integrate to global Killing spinors. Other 
branes can be introduced by hand by allowing for local deficit angles in the 
transverse space without generating any kind of curvature singularities. 



1 Introducion 



The idea of solving the cosmological constant problem by regarding our uni- 
verse as a brane in more than four dimensions opens up new interesting pos- 
sibilities for attacking this fundamental question PJ. In the context of the 
brane-world models the uniform part of the cosmic energy is indistinguishable 
from the brane tension. The nonzero tension of a physical three-brane, on 
which the particles of the standard model are localized, may be compensated 
by a higher dimensional curvature or a bulk cosmological constant, without 
generating curvature in the brane. 

In a recent work it was shown that in a nonsupersymmetric six-dimensional 
model of gravity coupled to a sigma model targeted on any Kahler target 
space one can obtain nonsingular solutions in the presence of flat three-branes 
with positive tensions [2]. The branes in that paper could be viewed as vor- 
tices distributed on a two-dimensional compact manifold with the topology of 
a two-sphere S 2 . The vorticity of a brane can be thought of as an Aharonov- 
Bohm phase acquired by a scalar field when a complete rotation around the 
brane is made. There is a single relation between the tensions of the three- 
branes and their vorticities and it was speculated that the time-variation 
of the vorticity of each brane may account for the adjustment of the four- 
dimensional cosmological constant. To obtain such a solution, however, it 
was necessary to assume that the six-dimensional cosmological constant was 
zero. It was indeed shown subsequently in [3] that a nonzero six-dimensional 
cosmological term disturbs the flatness of the three-branes. 

In this paper we shall find solutions in six-dimensional supergravity mod- 
els where a bulk cosmological constant is forbidden by supersymmetry. We 
shall show that the tensions of the branes can be arbitrary and positive sub- 
ject to a single linear relation of the type stated in [2| . The solution for 
the scalar field will be implicit and more involved than the simple mappings 
given in [U 121 El- Flat three-brane solutions with a singularity at the bound- 
ary of the transverse space are known to exist in D = 6 supergravities with 
scalars described by a nonlinear sigma model j3J |5] . This singularity is, un- 
fortunately, located at a finite proper distance from the brane or from any 
other point on the transverse space. Further, although the Killing spinor 
equations have local solutions far away from the brane, they do not integrate 
to a single- valued global solution 0j. 

The aim of this paper is to present nonsingular solutions in a supergravity 
model, where like in jlj, apart from gravity, the only other active fields 



2 



will be a set of scalar fields with dynamics governed by a nonlinear sigma 
model Lagrangian targeted on some noncompact hyperbolic or quaternionic 
manifold. Such scalar fields exist in all supergravity models in six dimensions. 
Our construction will lead to a smooth transverse space with no boundaries 
and with an Euler number of +2. Like the solution in jHE] our solution in this 
paper will also be only locally invariant under 1/2 of D=6 supersymmetries. 
These local supersymmetries, however, do not integrate to globally single 
valued Killing spinors. 

Similar attempts to solve the cosmological constant problem have been 
made in the past by invoking the magnetic monopole compactifiction in a six 
dimensional theory of gravity coupled to Maxwell field and in the presence of 
a six-dimensional cosmological constant Ej . The Kaluza-Klein solution of 
this system needs a very special tuning of six-dimensional cosmological term 
versus the radius of the internal S* 2 in order to have a flat four- dimensional 
part jHJ. A similar tuning is also necessary in the (1,0) six-dimensional 
gauged supergravities of the type constructed in [TUl HTj . although in this 
case the Minkowski^ x S 2 turns out to be the unique maximally symmetric 
solution which also preserves | of the supersymmetries [12]. These theories 
also have three-brane solutions but with negative tensions and/or singular 
transverse spaces IT2"1 ITi]. 1 

The solution of the scalar field equation in this paper is akin to the one 
studied in the context of the string comic string in [T^j. There, the complex 
scalar field is the modulus of a T 2 , therefore, modular invariance plays a very 
important role. In our case the restriction imposed by modular invariance 
is needed for the single valuedness of the space time metric. For this reason 
our solution for the metric will be somewhat different and, unlike the metric 
in ^Hj, the singularities in our metric will be smoothed out. This smoothing 
out necessarily introduces delta function type terms in the Einstein equations 
whose natural interpretation is the presence of three-branes. Demanding 
regularity at infinity and positivity of the tensions of the branes will impose 
more restrictions on our solution to the scalar field equation compared to 
the solution in [T5|. However, unlike the monopole solutions of [HI El, which 
require tuning of the parameters to achieve flatness,we shall need a single 
linear relation on the tensions. The single relationship between the tensions 
is dictated by the requirement that the geometry becomes spherical away 

1 Some of these issues will be further discussed in a forthcoming paper by C. Burgess et 

al. 



3 



from the branes, as in j2]. This reference also introduced the notion of 
vorticity for the brane solutions, but, because of the complexity of the scalar 
field ansatz, it is not clear to us how we can extend this idea to the present 
case in a straightforward way. 

The plan of this paper is as follows: In section 2 we present the solution 
and show its regularity. In section 3 we examine the conditions for unbroken 
supersymmetries. Section 4 is dedicated to a brief summary. 



2 The Solutions 

Turning to the details of the solutions, consider a D = 6 supergravity with 
scalars in the hyper or tensor multiplets. Such scalars cover a noncompact 
hyperbolic or quaternioic manifold of the type SO(n, 1) / SO{n) for the scalars 
in the tensor multiplet or Sp(n, 1)/Sp(n) x Sp(l) for the scalars in the hy- 
permultiplet. (In this latter case some other quaternionic Kahler manifolds 
are also possible [15] . Our discussion will be applicable to all of them.) We 
shall set all of the scalar fields to zero apart from a single complex one which 
we shall denote by ip. The metric in the two-dimensional subspace of the 
scalar manifold is given by 

2 dtp d<p 

(i-M 2 ) 2 [) 

A simple holomorphic change of coordinates ip = (r — + i) will then 
bring it to the form 

, 9 drdf 

We shall look for solutions of the Einstein's equations of the form 

ds 2 = Vliu dx^dx u + e^ z ~ z) dzdz (3) 

where fi, v — 0, 1, 2, 3 and (z, z) is a local complex coordinate in the extra 
two dimensions. In this coordinate system the relevant field equations reduce 
to 

ddr + 2 ^— - = (4) 

T — T 

for the scalar field and 



4 



for the metric function of the transverse two-space. In the last equation, M is 
the six-dimensional Planck mass and Tj are the tensions of the brane sources. 
Note that, unlike in the case of the nonsupersymmettric model, in the present 
case, the D = 6 supersymmetry does not allow for an independent coupling 
constant for the sigma model. Apart from the brane source terms the form 
of the field equations are completely fixed by supersymmetry. We also note 
that the field equations are invariant under an SL(2, R) group acting on r 
by fractional linear transformations, viz., 



where a, b, c, d are real numbers satisfying ad — be = 1. 

We first consider the scalar field equation. Clearly any function of z which 
is independent of z solves that equation. In particular, we take the matter 
field configuration to be given by 



where J(r) is the modular function known as the absolute invariant [TJj. By 
solving this for r in terms of z, we get the matter field configuration. Since J 
is invariant under modular transformations SL(2, Z) of r, the inverse will be 
multivalued unless we restrict to the fundamental domain F of r. Writing 
t = T\ + ir 2 , this domain is defined by 



F = {t\t 2 > 0, |r| > 1, - I < Tl < i, n < for|r| = 1} (8) 



In this domain, there is a unique inverse. We will define our solution as the 
configuration r(z) where r takes values in the fundamental region. Other 
regions correspond to other solutions which are degenerate with this; clas- 
sically they can be treated as separate and different solutions. Also, since 
the equations of motion are invariant under a global SL(2, R) action, we can 
obtain new solutions by acting with this group on any given solution. 

Note that, as far as the r field equations are concerned, there is an infinite 
class of solutions of the type suggested here. In principle, we could consider 
an implicit equation of the form J(r(z)) = f(z), where / is any holomorphic 
function of z. Here we have taken / to be a simple rational function which 
approaches unity as z goes to infinity. We shall see later that the requirement 
of the positivity of the tensions of the branes and regularity of the metric 




(6) 



J(r(z)) 



z — a 



(7) 



z-b 



5 



as \z\ goes to infinity exclude every other possibility for f(z) apart from the 
simple one given above. 

The ansatz for the metric is given by 



7 



2 —2 
T 2 T] 7] ■ 



z-b\ 



\z-a\- 2a \z-b\- 2f3 



x exp 



J (t-t)(z') 



(9) 



where a, /3, 7 are constants and the Green function G(z,z') satisfies 

8dG(z,z') = 5 {2) (z-z') (10) 

The exponential involving the Green function tends to a constant whenever z 
approaches a or b, which, as we shall see, will be identified with the positions 
of the three-branes, and it simply vanishes at every other point. i](r) is the 
Dedekind 77-function. The first factor T2f] 2 f] 2 \z — 6| — 12 is needed for modular 
invariance. Since for z — > b the 77 function behaves like (z — 6) 1 / 24 , the factor 

2 

\z — b\~™ ensures that the metric does not vanish as z approaches b. 

Notice that, apart from r 2 and the exponential factor involving G(z,z'), 
is of the form g(z)g(z), where g(z) is holomorphic in z. As a result, in the 
calculation of dd<p, the function g only contributes (^-functions, which can be 
identified as the brane contribution to the energy-momentum tensor. Since 
the factor 77/ (z — b) 1 / 12 has no singularity at z = b, the contribution due to the 
brane at z = b comes entirely from the factor \z — b\~~ 2lS . The contribution 
due to r 2 exp[J cPz'G(z, z')ddr(z') / (f — t){z')] will exactly match the field 
part of the energy- momentum tensor. Also the constant prefactor 7 in e^, 
which measures the size of the transverse space, is undetermined by the field 
equations and is thus a modulus. 

We will now turn to an analysis of the scalar field, metric and curvature 
at various special points, which will yield constraints on the exponents a 
and (3. But before doing so, let us note that the tensions at these points are 
simply given by 

T a = 2nM 4 a T b = 2nM A f3 (11) 

This result follows from the substitution of our ansatz for in the Einstein 
equation and then equating the coefficients of the delta functions at z = a 
and z = b. 



6 



We will now require that the transverse geometry in the vicinity of the 
location of the branes should have no curvature singularities; this will put 
restrictions on a and (3. There are three special points to analyze ^7j. These 
points are essentially characterized by the condition of the vanishing of J(r) 



or tj-. By analyzing our equations above, it is easy to see that these are the 



dr 



only points where there is the danger of curvature singularities. 



1. At z 



a, J and 4r vanish and r = uj = exp(27rz/3). Near this point, 



dr 



we can expand r as 



r ~ + c (z — a) 3 



(12) 



where c is a constant, and the exponential factor in are finite at 
z = a, so that the metric has the behavior 



\z — a 



2a 



The geometry has a conical defect at z 
matter fields behaves as 



(13) 



The energy density for 



£ 



1 



27 



■ 4 

a\ s 



(14) 



The energy density has singularity; this will be matched exactly by a 
similar singularity in dd<p so that the Einstein equation will be satisfied. 
Further, this is an integrable singularity, so that there is no pathology 
in the total energy. In fact, the curvature is given by 



R 



c 



1 2a 



27 \ z 
\c\ 2 



■ 4 

a 3 



£ 



27 \z — a\ 3 
naS^ 2 \z — a 



- 7ra5 ( - 2 \z — o 
+ ira5 {2) (z - a) 



(15) 



We notice that there is a singularity in the curvature if we do not have 



the term 
taking 



1 2a 



With this term, we can avoid the singularity by 



a > 



(16) 



7 



In view of the aforementioned relationship between the tension of the 
brane at z = a and the parameter a we thus obtain a lower bound on 
T a , namely 

Air 

T a > —M 4 (17) 

The behavior of r near z = a requires another comment. Evidently, 
(z — a) 3 is not single- valued as we go around the point z = a; also the 
value of t after such a rotation, namely, r(Ae 27n ), A = z — a, is not 
within the fundamental modular region. However up to first order in 
A, there is a modular transformation given by 

r(Ae-) = T^±l (18) 
-r(A) 

which maps r back into the fundamental region. Since the fields are 
defined up to such a modular transformation, this does not change the 
physical fields. The factor T2if'ff'\z — in (JHJ) is modular invariant, 
making the metric invariant, as in |15j . 

2. As \z\ — > oo, J — > 1, and r — > i. Thus asymptotically, we can expand 
t(z) as 

l 

r K,i + c 2 z 2 (19) 
It is then easily seen that the metric behaves as 

In order for this not to lead to deficit angles at infinity, we need to 
behave as \z\~ A . This gives a constraint 

a + (3+^ = 2 (21) 

which translates into a constraint among the tensions, viz, 

T a + T b = 1 -^ 2vrM 4 , (22) 
o 

an identity similar to the constraint derived in 0. Note that this 
behavior as \z\ goes to infinity also insures that the Euler number of 
our transverse space is +2. 



8 



3. As z — > b, J — > oo. In this case, the metric has the form 

\z-b\- 2 ? (23) 



log 1 2 — 61 

2tt 1 1 



Without the extra term \z — b\ 2/3 , there is a curvature singularity, since 
the Ricci scalar is given by 



R m \z — b\ 



2/3 



log 1 2 - b\) 



? + np5 {2) (z - b) 

A\z-b\>{\og\z-b\f V ; 

(24) 



Even though the Einstein equation can be satisfied, balancing the singular 
term of £ with the similar term in e^R, the curvature itself will be singular 
unless we choose 

P > 1 (25) 
In terms of the tension T& this implies 

T b > 2ttM 4 (26) 

Thus, we see that, to obtain regular geometry not only the tensions should be 
bounded from below by positive quantities, their sum should also be bounded 
by (11/6) x 27rM 4 . It can be verified that are no other points at which the 
curvature behaves badly. The three constraints do have a set of solutions for 
a and (3. 

Let us now turn to a possible generalization of equation© to 

N ? - n 

J(r(z)) = II — IT ( 2 7) 

fc=l Z ~ h k 

In order to avoid singularities whenever z approaches any one of the a^'s or 
6fc's we need to insert a factor Y\ k \z — a^\~ 2ak \z — by\~ 2 ^ k in the ansatz for 
the metric and restrict each a k and /3k by 

a k > | fa > 1 (28) 

The requirement that should behave like \z\~ A whenever \z\ goes to infinity 
will then restrict these parameters by the linear relation 

N 1 

E("* + Pk) + - = 2 (29) 



9 



Clearly we can satisfy all these constraints only for AT = 1. Presumably the 
introduction of vorticity as in [2 j would allow us more freedom. However, at 
this stage it is not clear to us how to do this a simple physical way. 

The only way that we can keep r(z) as a solution of the simple equation 
J(t(z)) = (z — a) I (z — b) and yet introduce more than two branes is to modify 
the metric by inserting a factor like \z — a\~ 2a \z — b\~ 2f3 Yl k \z — a' k \~ 2ak \z — 
b' k \~ 2f3k , where a' k s and b' k s are different from a and b. In this case, by requiring 
that a' k s and /3' k s are smaller than unity, we can have a solution to all the 
constraints. 

3 Super symmetry 

Now we turn to the question of unbroken supersymmetries. Having set all 
the fields to zero apart from gravity and a pair of real scalars which we have 
assembled into a single complex field r, the only Killing spinor equations we 
need to examine are those of gravitino and the scalars. Although our argu- 
ment is valid for all ( ungauged) supergravities in D=6, to be concrete, in this 
section, we shall consider hypermatter scalars in ungauged (1,0) supergravi- 
ties. Such models are chiral and can be obtained from the compactification 
of the heterotic string on K3 JB] or the M theory on K3 x S 1 / Z2 |19j . They 
parameterize a quaternionic manifold of the form G/H x SU(2). 

The gravitino as well as the hyperino carry spinor indices of SU(2). Their 
supersymmetry variation in our bckground is given by 2 

= 9 m e r + l^ m T 45 e r + d m <f) a Q r a s e s (30) 

SijT = Vt r d m <p a T m e r (31) 

In these equations the spinor indices r, s... take values 1 and 2, the index m 
is tangent to the transverse space and takes the values 4 and 5. (fi a denote 
the two nonvanishing hyperscalars. uj m and Q in the gravitino equation 
denote the connections in the transverse space and the hyperscalar manifold 
respectively. Finally V" = V^(ia 3 ) rr + V 2 (ia 1 ) rr . We have denoted the 
components of the zweibein in the tangent space of the scalar manifold by 
V* and V 2 . All we need to know about them is that their only non vanishing 

2 Here we closely follow the notations of ^Hj an d |2()j . 



10 



components are V/ = V 2 2 . Finally we quote, without giving the details, that 
for our background the spin connections io m and Q are given by 



2 r 2 



l d z n 



+ d T lni]d z T — 



a 1 



2 z — a 



ff + 1/12 1 
2 z-6 



(32) 



d z 4> a Q 



i d z Ti 




(33) 



Q: 



4 r 2 



The condition <5-?/> rr = 0, can be solved provided we restrict the supersym- 
metry parameter e by 



This condition cuts the number of components of e by half. Upon im- 
posing this condition the gravitino equation Sipl^ = then reduces to a first 
order linear differential equation for e as a function of z and z which in gen- 
eral can be integrated. The solution will, however, be a mutivalued function 
of z. A full rotation around the branes will not bring e back to its original 
value. It has been suggested in |lj that by introducing a flat U(l) connection 
one can eliminate the multivaluedness by an Aharonov-Bohm phase. It will 
be interesting to find a supergravity model in six dimensions where this idea 
can be implemented without disturbing other nice features of our solutions. 

4 Conclusion 

In summary, we constructed nonsingular flat brane solutions in ungauged 
six dimensional supergravities with a transverse space of Euler number +2. 
Since our transverse space has no singularities or boundaries the value +2 for 
its Euler number implies that, topologically, it is a S 2 . Thus its volume will 
be finite, although it is not easy to evaluate it explicitly. The size of the space 
depends on an integration constant which is not determined at the classical 
level. It is thus a modulus. It will be interesting to see if temperature effects 
in a cosmological context can stabilize the value of this radius. This is known 
to happen in Kaluza Klein cosmology in the absence of branes [21] . Aspects 
of brane cosmology and the question of self tuning of cosmological constant 
in co-dimension two have been studied in |2"2"] . 



tr4(<T 3 ) rr + r 6 (t7i) ,T ]e r = o 



n 



Like the solutions in j3] the solutions presented in this paper also break 
~ of the supersymmetries. The local solution to the Killing spinor equa- 
tion cannot, however, be integrated to a single- valued global solution. It 
is suggested in [4J that by coupling the gravitino to a U(l) gauge field and 
allowing for multivalued Killing spinors one may recover single-valuedness. 
Such U(l) couplings do indeed exist in the gauged six- dimensional supergrav- 
ities. However, one difficulty, among others, in implementing this interesting 
idea is that in this case the scalar fields will have a nontrivial potential and 
their field equations cannot be solved by holomorphic configurations. 3 

Our construction cannot be considered a completely satisfactory solution 
to the cosmological constant problem, principally, due to the linear relation 
among the tensions that we derived in section 3. The introduction of vorticity 
along the lines of [2|, may improve our construction, although we do not have 
a simple physical way of implementing this idea in the present context. 



Acknowledgments 

We would like to thank Bobby Acharya for reminding us of the reference 
15J The work of V.P.N, was supported in part by the U.S. National Science 
Foundation grant number PHY-0244873. 



References 

[1] V. A. Rubakov and M. E. Shaposhnikov, Constant Phys. Lett. B 125, 

139 (1983); 

M. Visser, Phys. Lett. B 159, 22 (1985) |arXiv:hep-th/9910093| ; 
S. Randjbar-Daemi and C. Wetterich, Phys. Lett. B 166, 65 (1986); 
G. W. Gibbons and D . L. Wiltshire, Nucl. Phys. B 287, 717 (1987) 

|arXiv:hep-th/0109093| ; 

R. Sundrum, Phys. Rev. D 59, 085010 (1999) arXiv:hep-ph/9807348| ; 
N. Arkani-Hamed, S. Dimopoulos, N. Kaloper and R. Sundrum, Phys. 
Lett. B 480, 193 (2000) |arXiv:hep-th/0001197| ; 

S. Kac hru, M. B. Schulz and E. Silverstein, Phys. Rev. D 62, 045021 
(2000) |arXiv:hep-th/0001206| ; 

G. R. Dvali and G. Gabadadze, Phys. Rev. D 63, 065007 (2001) 
3 For an explicit form of the potential in (1,0) supergravity in six dimensions see |20| . 



12 



|arXiv:hep-th /0008054| ; 

J. E. Kim, B. Kyae and H. M. Lee, Phys. Rev. Lett. 86, 
4223 (2001) |arXiv:hep-th/0011118| , Nucl. Phys. B 613, 306 (2001) 
larXiv:hep-th/0101027| ~ 

C. De ffayet, G. R. Dvali and G . Gabadadze, Phys. Rev. D 65, 044023 
(2002) |arXiv:astro-ph/0105068| ; 

G. Dvali, G. Gabadadze and M. Shifman, Phys. Rev. D 67, 044020 (2003) 

jarXiv:hep-th/0202174| ; 

U. Ellwanger, JCAP 0311, 013 (2003) [arXiv:hep-th70304057 . 



[2] 
[3] 
[4] 
[5] 
[6] 
[7] 



S. Randjbar-Daemi and V. Rubakov, arXiv:hep-th/0407176 



H. M. Lee and A. Papazoglou, |arXiv:hep-th/0407208| 



A. Kehagias, arXiv:hep-th/0406025 



G. Dvali, S. Randjbar-Daemi and Ergin Segin unpublished. 

S. M. Carroll and M. M. Guica, |aTxiv:hep-th/0302067| 

I. Navarro, JCAP 0309, 004 (2003) |arXiv:hep-th/0302129| . 

S. Randjbar-Daemi, A. Salam and J. Strathdee, Nucl. Phys. B 214, 491 
(1983). 



[9] I. Navarro, Class. Quant. Grav. 20, 3603 (2003) arXiv:hep-th/0305014 



[10] A. Salam and E. Sezgin, Phys. Lett. B 147, 47 (1984). 

[11] S. Randjbar-Daemi, A. Salam, E. Sezgin and J. Strathdee, Phys. Lett. 
B 151 (1985) 351. 



[12] G. W. Gibbons, R. Guven and C. N. Pope, arXiv:hep-th/0307238 



[13] Y. Aghababaie, C. P. Burgess, S. L. Parameswaran and F. Quevedo, 
Nucl. Phys. B 680, 389 (2004) |arXiv:hep-th/0304256| . 

[14] Y. Aghababaie, C. P. Burgess, S. L. Parameswaran and F. Quevedo, 
JHEP 0309, 037 (2003) |arXiv:hep-th/0308064| . 

[15] B. R. Greene, A. D. Shapere, C. Vafa and S. T. Yau, Nucl. Phys. B 337 
(1990) 1. 



13 



[16] H. Nishino and E. Sezgin, Nucl. Phys. B 278 (1986) 353. 



[17] E.T. Wittakker and G.N. Watson, A Course of Modern Analysis, Cam- 
bridge University Press 1962, and, K. Chandrasekharan, Elliptic Func- 
tions, Springer- Verlag , 1985. 

[18] M. B. Green, J. H. Schwarz and P. C. West, Nucl. Phys. B 254 (1985) 
327. 

[19] A. Sen, Phys. Rev. D 53, 6725 (1996) [arXiv:hep-th/9602010| . 

[20] S. Randjbar-Daemi a nd E. Sezgin, Nucl. Phys. B 692 (2004) 346 
|arXiv:hep-th/0402217| . 

[21] S. Randjbar-Daemi, A. Salam and J. Strathdee, Phys. Lett. B 135, 388 
(1984). 

T. Bringmann a nd M. Eriksson, JCAP 0310, 006 (2003) 
|arXiv:astro-ph /0308498 ; T. Brin gmann, M. Eriksson and M. Gustafs- 
son, Phys. Rev. D 68, 063516 (2003) iarXiv:astro-ph/0303497| . 

[22] J. M. Clin e, J. Descheneau, M. G iovannini and J. Vinet, JHEP 0306, 
048 (2003) |arXiv:hep-th/0304147| ; 

J. Vinet and J.M. Cline, |hep-th/ 0406 141 



14