J

ournal of Cosmology and Astroparticle Physics

An IOP and SISSA journal

The trispectrum in the multi-brid ination

JCAP05(2009)005

Qing-Guo Huang
School of Physics, Korea Institute for Advanced Study, 207-43, Cheongryangri-Dong, Dongdaemun-Gu, Seoul 130-722, Korea E-mail: huangqg@kias.re.kr Received March 15, 2009 Accepted April 20, 2009 Published May 6, 2009

Abstract. The trispectrum is at least as important as the bispectrum and its size can be characterized by two parameters NL and gNL . In this short paper, we focus on the Multi-brid ination, in particular the two-brid ination model in arXiv.0805.0974, and nd that NL is always positive and roughly equals to ( 6 fNL )2 for the low scale ination, but gNL can be 5 negative or positive and its order of magnitude can be the same as that of NL or even larger. Keywords: ination, non-gaussianity, cosmological perturbation theory ArXiv ePrint: 0903.1542

c 2009 IOP Publishing Ltd and SISSA

doi:10.1088/1475-7516/2009/05/005

Contents
1 Introduction 2 The trispectrum in the multi-brid ination model 2.1 g1 = g2 2.2 g1 = g2 3 Conclusions 1 2 5 6 7

JCAP05(2009)005

Introduction

The single-eld slow-roll ination predicts a roughly Gaussian distribution of the primordial power spectrum [1]. However in the general case with a large number of elds, a large deviation from Gaussian distribution is expected. A well-understood ansatz of non-Gaussianity has a local shape. Working in the framework of Fourier transformation of , the primordial power spectrum P is dened by (k1 )(k2 ) = (2)3 P (k1 )3 (k1 + k2 ), and the primordial bispectrum and trispectrum are dened by (k1 )(k2 )(k3 ) = (2)3 B (k1 , k2 , k3 )3 (k1 + k2 + k3 ), (k1 )(k2 )(k3 )(k4 ) = (2)3 T (k1 , k2 , k3 , k4 )3 (k1 + k2 + k3 + k4 ). The bispectrum and trispectrum are respectively related to the power spectrum by 6 fNL [P (k1 )P (k2 ) + 2 perms], 5 T (k1 , k2 , k3 , k4 ) = NL [P (k13 )P (k3 )P (k4 ) + 11 perms] 54 + gNL [P (k2 )P (k3 )P (k4 ) + 3 perms], 25 B (k1 , k2 , k3 ) = (1.4) (1.2) (1.3) (1.1)

(1.5)

where fNL , NL and gNL are the non-Gaussianity parameters which measure the size of the non-Gaussianity. In the N formalism [2], the curvature perturbation in the multi-eld ination model can be expanded as = N = N (i + i ) N (i ) 1 1 = N,i i + N,ij i j + N,ijk i j k + , 2 6 where N,i = N , i N,ij = 2N , i j N,ijk = 3N , i j k

(1.6)

(1.7)

and the repeated sub-indices are summed over. Therefore the non-Gaussianity parameters are given by, [3], fNL = NL gNL 5 N,ij N,i N,j , 6 (N,l N,l )2 N,ij N,ik N,j N,k , = (N,l N,l )3 25 N,ijk N,i N,j N,k = . 54 (N,l N,l )3 (1.8) (1.9) (1.10)

From the Cauchy-Schwarz inequality, the NL has a lower bound, [4], NL 6 fNL 5
2

JCAP05(2009)005

(1.11)

The above inequality is saturated in the single eld case, or the vector N,i is an eigenvector of the matrix N,ij in the case with multi elds. So NL is expected to be large if fNL 1. Since N,ijk is quite model-dependent, gNL can be negative or positive, and its order of magnitude can be large or small. WMAP 5yr data [5] implies
local 9 < fNL < 111

(1.12)

at 2 level. Even though the Gaussian distribution is still consistent with data, the allowed local negative part of fNL has been cut from the WMAP 3yr data signicantly. For example, if NL > 560, it can also be detected by Planck [6]. In [6] the authors also pointed out that the trispectrum of the Planck data is more sensitive to primordial non-Gaussianity than the bispectrum for |fNL | 50. The trispectrum is at least as important as the bispectrum. As we know, a large local-shape fNL can be achieved in the multi-brid ination. In this short paper, we investigate the trispectrum in the special two-brid ination model in [7]. One can discuss more complicated setups, such as [8]. However, we will see that the phenomenology in this simple model is rich enough. In section 2, we calculate the trispectrum in this twobrid ination model and discuss its property in detail. Some discussions are given in section 3.

The trispectrum in the multi-brid ination model

In this section, we consider the two-brid ination in [7] with the potential V (1 , 2 ) = V0 exp (1 1 + 2 2 ) , (2.1)

where 1 and 2 are two free parameters in the unit of Mp = 1. The eective mass for each inaton is mi = 3i H where H is the Hubble parameter during ination. Here the inatons are assumed to slowly roll down their potentials and the slow-roll conditions require i 1 for i = 1, 2. The dynamics of these two inatons is governed by di i , dN
2 2 gi 2 2 + i i=1

(i = 1, 2),

(2.2)

where dN = Hdt. In [7], V0 is prompted to be V0 = 1 2 2 2 4

2

(2.3)

Therefore the ination ends when i = i,f which is given by

2 2 gi 2 = 2 . i,f i=1

(2.4)

For convenience, we parametrize i,f as 1,f = cos , g1 2,f = sin . g2 (2.5)

Without loss of generality, we assume i i > 0 (for i = 1, 2). The number of e-folds before the end of ination is given by N= 1 ln 2 e21 /1 + e22 /2 e21,f /1 + e22,f /2 . (2.6)

JCAP05(2009)005

Because the surface where the ination ends in the eld space is not exactly the surface of constant energy density, the author in [7] introduced a correction to the number of e-folds before the end of ination as Nc = 1 2 cos + sin . 4 g1 g2 (2.7)

However, this correction term can be neglected if i 1. So we ignore it in our paper. In order to simplify the notation, we dene some new parameters as g =
2 2 g1 cos2 + g2 sin2 ,

(2.8) (2.9)

1 = cos , We expand N to the third order, N =

2 = sin .

g2 g2 (sin 1 cos 2 )2 g1 cos 1 + g2 sin 2 + 1 2 gc+ 2g3 c3 +

3 3 g1 g2 c (sin 1 cos 2 )3 , 2 2 g4 c5 +

(2.10)

where c = c+ g2 g1 cos sin sin cos , g g g1 g2 = cos cos + sin sin . g g (2.11) (2.12)

Assume that the scalar eld uctuations 1 and 2 are Gaussian and non-correlated, i j = H 2
2

ij ,

(2.13)

where H denotes the Hubble parameter at the time of Hubble exit. Hence the amplitude of the primordial power spectrum is P = 1 2 c2 + H 2
2

(2.14)

Since the amplitude of the tensor perturbation is still given by PT = 8 the tensor-scalar ratio becomes r = PT /P = 82 c2 . + We can also easily calculate the spectral index, namely ns = 1 2 . (2.17) (2.16) H 2
2

(2.15)

JCAP05(2009)005

The spectral index and tensor scalar ratio can be measured by the experiments. Once they are xed, the parameters and c+ in this two-brid ination model are xed as well. WMAP 5yr data [5] implies ns = 0.96+0.014 . (2.18) 0.013 For ns = 0.96, = 0.2. On the other hand, we can prove that c2 1 which implies + r 8(1 ns ), where the equality is satised when g1 sin cos = g2 cos sin . (2.20) (2.19)

Eq. (2.19) can be taken as a consistency relation for this two-brid ination model. For ns = 0.96, r 0.32 which is consistent with WMAP 5yr data (r < 0.20). From eqs. (1.4) and (1.5), the non-Gaussiniaty parameters are given by fNL = NL
2 2 5g1 g2 c2 , 6g3 c+ 4 4 2 g1 g2 c2 , = 2 g6 c2 + 3 3 252 g1 g2 c c3 , 18 2 g4 c2 +

(2.21) (2.22) (2.23)

gNL = where c=

g1 g2 cos sin sin cos . g g 36 1 2 f , 25 c2 NL 2g2 c 2 f . = g1 g2 c NL

(2.24)

We see that the second order non-Gaussianity parameters are related to fNL by NL = gNL Since and c2 1, c2 1 and thus + c2 = 1 c2 + NL 6 fNL 5
2

(2.25) (2.26)

(2.27)

(2.28)

The inequality is saturated when g1 cos cos = g2 sin sin . (2.29)

c2 < 0.625 for ns = 0.96 and r < 0.20. If r 1, c+ 0 and then NL + 2.1 g1 = g2 c = c = sin( ), Therefore fNL = NL 1 c+ , c+ 36 f2 , = 25(1 c2 ) NL + 5g 6 c+ = cos( ).

2 Keeping fNL xed, NL can be much larger than fNL if c+ 1. However, in the limit of c+ 1, fNL should approach to 0. Here we want to remind the readers that both and c+ can be xed by the spectral index and the tensor-scalar ratio. Considering r , (2.30) c2 = + 8(1 ns ) 36 2 25 fNL .

JCAP05(2009)005

From eq. (2.8), g1 = g2 = g in this case. Now c , c, and c+ are simplied to be (2.31)

(2.32) (2.33) (2.34)

2 gNL = 2fNL .

2 In this special case, gNL is positive and is related to fNL by 2fNL . If r 1, c+ 0 and then 36 2 NL 25 fNL which is roughly the same order of magnitude as gNL , and

fNL

5g . 6c+ 4 . 12

(2.35)

Because = 0 during ination, the Hubble parameter is related to by

2 H =
1

(2.36)

Considering H = 1.1 104 r 2 , is related to r by

= 1.95 102 4 r 4 . If r 1, the non-Gaussianity parameter fNL takes the form fNL g r 121(1 ns ) r
1 4

(2.37)

(2.38)

On the other hand, is required to be large compared to the Hubble parameter during ination for the eld to work as a water-fall eld. So we have (r/) 4 Therefore fNL
1

177.

(2.39) (2.40)

g 2.1 104 (1 ns ) . r

For ns = 0.96, fNL 810g/r. 2 To summarize, if g1 = g2 , fNL is positive and can be very large, and gNL = 2fNL . In 2 this case, NL 36 fNL and the equality is roughly satised if r 1. 25

2.2

g1 = g2

2 If g1 = g2 , we nd that gNL = 2fNL which must be positive. In this subsection, we focus on the case of g1 = g2 and invistigate whether gNL can be negative, but the absolute value is still large. Since there is a symmetry between 1 and 2 , we assume g1 < 1, (2.41) = g2

without loss of generality. Keeping fNL xed, a large absolute value of gNL might be obtained if c = 0 which implies = 0 = tan1 ( tan ). (2.42)

JCAP05(2009)005

On the other hand, fNL should be 0 when c = 0. So we consider that slightly deviates from 0 , namely = 0 + , (2.43) with 0 . For 1, we have g1 , 2 c , sin 2 c , 2 c+ 1, g cos2 + 2 sin2 .

(2.44) (2.45) (2.46) (2.47)

where = (2.48)

In this limit, c+ 1 and then the tensor-scalar ratio is slightly larger than the WMAP bound if ns = 0.96. Here we ignore this constrant rst and illustrate an important observation. Now we have 5g2 7 2 , (2.49) fNL 6 3 and 36 2 2 f , (2.50) 25 4 2 NL sin 2 2 gNL fNL . (2.51) 4 Since both and sin 2 are positive, gNL can be negative if > 0. In order to get the precise results and go beyond the above approximation, we should adopt the numerical calculation. For example, inputting ns = 0.96, r = 0.22, = 0.1, we have NL
2 NL /fNL 4.6,

(2.52)

and g2 2 , gNL /fNL 10; (2.53) g2 2 = 1.38, = 0.1 fNL = 1.12 , gNL /fNL 4.92. (2.54) So there is still a big room for a large fNL . We see that a negative gNL with large absolute value can be obtained in the case of g1 = g2 . = 0.79, = 0.466 fNL = 0.03

Conclusions

The non-Gaussianity is expected to be large in the multi-brid ination. In the simple twobrid ination model [7], NL must be positive and roughly equals to ( 6 fNL )2 due to the upper 5 bound on the ination scale (r < 0.2), but gNL can be negative or positive and its order of magnitude can be the same as that of NL or even larger. In [9] the authors consider a more complicated case where NL can be much larger than ( 6 fNL )2 , but it seems unlikely for the 5 case with red-tilted power spectrum and large positive fNL . As we know, multi-brid ination model and curvaton model are the only cases that can be analyzed systematically (over a fairly wide range of the model parameters) and are capable local of yielding fairly large local-shape fNL . In the usual single curvaton model, the curvature 6 perturbation is assumed to be generated by curvaton and then we have NL = ( 5 fNL )2 . The size of gNL depends on the curvaton potential. If the relevant part of curvaton potential takes the exactly quadratic form, we have gNL = 10 fNL . If a non-quadratic correction becomes 3 visible compared to the quadratic term, gNL in the curvaton model can has a large deviation from the relation gNL = 10 fNL and the sign of gNL depends on the sign of the non-quadratic 3 term which can be positive or negative as well. The order of magnitude of gNL for the curvaton 2 model with non-quadratic correction can be roughly the same as fNL which takes the same order of NL . See [10] in detail. If the curvaton potential is dominated by the non-quadratic 2 term, NL gNL fNL [11]. On the other hand, in the mixed curvaton scenario [11, 12] where the total curvature perturbation is still mainly produced by the inaton, not curvaton, NL is enhanced by a large factor 1/ with respect to ( 6 fNL )2 , where measures the size of curva5 ture perturbation caused by curvaton compared to the total curvature perturbation including the contribution from inaton. So NL can be much larger than ( 6 fNL )2 in the mixed cur5 vaton scenario. This is dierent from the multi-brid ination. In the mixed curvaton model, cv cv whether gNL is enhanced or suppressed depends on how gNL is related to fNL , where the superscript cv denotes the non-Gaussianity generated by curvaton in the usual curvaton model. cv cv For example, if gNL fNL , gNL fNL which is expected to be very small and undetectable, cv cv but if gNL (fNL )2 , gNL (fNL )2 / which is enhanced and may be detectable. If we can not only detect fNL , but also get some restricted constraints on NL and gNL in the forthcoming experiments, it is possible to distinguish multi-brid ination model from curvaton model. Recently there are many papers concerning on the brispectrum and the trispectrum [13, 14]. Here we want to stress that the trispectrum might be as important as the bispectrum and we encourage more theorists and experimenters to pay more attentions to the trispectrum in the near future.

JCAP05(2009)005

Acknowledgments
We would like to thank M. Sasaki for very helpful discussions.

References
[1] J.M. Maldacena, Non-gaussian features of primordial uctuations in single eld inationary models, JHEP 05 (2003) 013 [astro-ph/0210603] [SPIRES]. [2] A.A. Starobinsky, Multicomponent de Sitter (inationary) stages and the generation of perturbations, JETP Lett. 42 (1985) 152 [Pisma Zh. Eksp. Teor. Fiz. 42 (1985) 124] [SPIRES];

M. Sasaki and E.D. Stewart, A general analytic formula for the spectral index of the density perturbations produced during ination, Prog. Theor. Phys. 95 (1996) 71 [astro-ph/9507001] [SPIRES]; M. Sasaki and T. Tanaka, Super-horizon scale dynamics of multi-scalar ination, Prog. Theor. Phys. 99 (1998) 763 [gr-qc/9801017] [SPIRES]; D.H. Lyth, K.A. Malik and M. Sasaki, A general proof of the conservation of the curvature perturbation, JCAP 05 (2005) 004 [astro-ph/0411220] [SPIRES]; D.H. Lyth and Y. Rodriguez, The inationary prediction for primordial non-gaussianity, Phys. Rev. Lett. 95 (2005) 121302 [astro-ph/0504045] [SPIRES]. [3] L. Alabidi and D.H. Lyth, Ination models and observation, JCAP 05 (2006) 016 [astro-ph/0510441] [SPIRES]; C.T. Byrnes, M. Sasaki and D. Wands, The primordial trispectrum from ination, Phys. Rev. D 74 (2006) 123519 [astro-ph/0611075] [SPIRES]. [4] T. Suyama and M. Yamaguchi, Non-gaussianity in the modulated reheating scenario, Phys. Rev. D 77 (2008) 023505 [arXiv:0709.2545] [SPIRES]. [5] WMAP collaboration, E. Komatsu et al., Five-year Wilkinson Microwave Anisotropy Probe (WMAP) observations: cosmological interpretation, Astrophys. J. Suppl. 180 (2009) 330 [arXiv:0803.0547] [SPIRES]. [6] N. Kogo and E. Komatsu, Angular trispectrum of CMB temperature anisotropy from primordial non-gaussianity with the full radiation transfer function, Phys. Rev. D 73 (2006) 083007 [astro-ph/0602099] [SPIRES]. [7] M. Sasaki, Multi-brid ination and non-gaussianity, Prog. Theor. Phys. 120 (2008) 159 [arXiv:0805.0974] [SPIRES]. [8] A. Naruko and M. Sasaki, Large non-gaussianity from multi-brid ination, Prog. Theor. Phys. 121 (2009) 193 [arXiv:0807.0180] [SPIRES]; H.-Y. Chen, J.-O. Gong and G. Shiu, Systematics of multi-eld eects at the end of warped brane ination, JHEP 09 (2008) 011 [arXiv:0807.1927] [SPIRES]; S.H. Henry Tye, J. Xu and Y. Zhang, Multi-eld ination with a random potential, JCAP 04 (2009) 018 [arXiv:0812.1944] [SPIRES]. [9] C.T. Byrnes, K.-Y. Choi and L.M.H. Hall, Large non-gaussianity from two-component hybrid ination, JCAP 02 (2009) 017 [arXiv:0812.0807] [SPIRES]; Conditions for large non-Gaussianity in two-eld slow-roll ination, JCAP 10 (2008) 008 [arXiv:0807.1101] [SPIRES]. [10] M. Sasaki, J. Valiviita and D. Wands, Non-gaussianity of the primordial perturbation in the curvaton model, Phys. Rev. D 74 (2006) 103003 [astro-ph/0607627] [SPIRES]; K. Enqvist and T. Takahashi, Signatures of non-gaussianity in the curvaton model, JCAP 09 (2008) 012 [arXiv:0807.3069] [SPIRES]; Q.-G. Huang and Y. Wang, Curvaton dynamics and the non-linearity parameters in curvaton model, JCAP 09 (2008) 025 [arXiv:0808.1168] [SPIRES]; M. Kawasaki, K. Nakayama and F. Takahashi, Hilltop non-gaussianity, JCAP 01 (2009) 026 [arXiv:0810.1585] [SPIRES]; P. Chingangbam and Q.-G. Huang, The curvature perturbation in the axion-type curvaton model, JCAP 04 (2009) 031 [arXiv:0902.2619] [SPIRES]. [11] Q.-G. Huang, Curvaton with polynomial potential, JCAP 11 (2008) 005 [arXiv:0808.1793] [SPIRES]. [12] Q.-G. Huang, N-vaton, JCAP 09 (2008) 017 [arXiv:0807.1567] [SPIRES]. [13] Q.-G. Huang, Large non-gaussianity implication for curvaton scenario, Phys. Lett. B 669 (2008) 260 [arXiv:0801.0467] [SPIRES]; T. Matsuda, Modulated ination, Phys. Lett. B 665 (2008) 338 [arXiv:0801.2648] [SPIRES];

JCAP05(2009)005

M. Li, T. Wang and Y. Wang, General single eld ination with large positive non-gaussianity, JCAP 03 (2008) 028 [arXiv:0801.0040] [SPIRES]; K. Ichikawa, T. Suyama, T. Takahashi and M. Yamaguchi, Non-gaussianity, spectral index and tensor modes in mixed inaton and curvaton models, Phys. Rev. D 78 (2008) 023513 [arXiv:0802.4138] [SPIRES]; T. Suyama and F. Takahashi, Non-gaussianity from symmetry, JCAP 09 (2008) 007 [arXiv:0804.0425] [SPIRES]; X. Gao, Primordial non-gaussianities of general multiple eld ination, JCAP 06 (2008) 029 [arXiv:0804.1055] [SPIRES]; M. Beltran, Isocurvature, non-gaussianity and the curvaton model, Phys. Rev. D 78 (2008) 023530 [arXiv:0804.1097] [SPIRES]; B. Chen, Y. Wang and W. Xue, Inationary nongaussianity from thermal uctuations, JCAP 05 (2008) 014 [arXiv:0712.2345] [SPIRES]; M. Li, C. Lin, T. Wang and Y. Wang, Non-gaussianity, isocurvature perturbation, gravitational waves and a no-go theorem for isocurvaton, arXiv:0805.1299 [SPIRES]; S. Li, Y.-F. Cai and Y.-S. Piao, DBI-curvaton, Phys. Lett. B 671 (2009) 423 [arXiv:0806.2363] [SPIRES]; Q.-G. Huang, Spectral index in curvaton scenario, Phys. Rev. D 78 (2008) 043515 [arXiv:0807.0050] [SPIRES]; M. Li and Y. Wang, Consistency relations for non-gaussianity, JCAP 09 (2008) 018 [arXiv:0807.3058] [SPIRES]; M. Li and C. Lin, Reconstruction of the isocurvaton scenario, arXiv:0807.4352 [SPIRES]; M. Kawasaki, K. Nakayama, T. Sekiguchi, T. Suyama and F. Takahashi, Non-gaussianity from isocurvature perturbations, JCAP 11 (2008) 019 [arXiv:0808.0009] [SPIRES]; M. Kawasaki, K. Nakayama and F. Takahashi, Non-gaussianity from baryon asymmetry, JCAP 01 (2009) 002 [arXiv:0809.2242] [SPIRES]; K. Dimopoulos, M. Karciauskas, D.H. Lyth and Y. Rodriguez, Statistical anisotropy of the curvature perturbation from vector eld perturbations, arXiv:0809.1055 [SPIRES]; T. Moroi and T. Takahashi, Non-gaussianity and baryonic isocurvature uctuations in the curvaton scenario, Phys. Lett. B 671 (2009) 339 [arXiv:0810.0189] [SPIRES]; M. Kawasaki, K. Nakayama, T. Sekiguchi, T. Suyama and F. Takahashi, A general analysis of non-gaussianity from isocurvature perturbations, JCAP 01 (2009) 042 [arXiv:0810.0208] [SPIRES]; T. Matsuda, Non-standard kinetic term as a natural source of non-gaussianity, JHEP 10 (2008) 089 [arXiv:0810.3291] [SPIRES]; C. Hikage, K. Koyama, T. Matsubara, T. Takahashi and M. Yamaguchi, Limits on isocurvature perturbations from non-gaussianity in WMAP temperature anisotropy, arXiv:0812.3500 [SPIRES]; P.D. Meerburg, J.P. van der Schaar and P.S. Corasaniti, Signatures of initial state modications on bispectrum statistics, arXiv:0901.4044 [SPIRES]; H.R.S. Cogollo, Y. Rodriguez and C.A. Valenzuela-Toledo, On the issue of the series convergence and loop corrections in the generation of observable primordial non-gaussianity in slow-roll ination. Part I: the bispectrum, JCAP 08 (2008) 029 [arXiv:0806.1546] [SPIRES]; X. Ji and T. Wang, Curvature and entropy perturbations in generalized gravity, arXiv:0903.0379 [SPIRES]; M.-X. Huang, G. Shiu and B. Underwood, Multield DBI ination and non-gaussianities, Phys. Rev. D 77 (2008) 023511 [arXiv:0709.3299] [SPIRES]; D. Langlois and S. Renaux-Petel, Perturbations in generalized multi-eld ination, JCAP 04 (2008) 017 [arXiv:0801.1085] [SPIRES]; D. Langlois, S. Renaux-Petel, D.A. Steer and T. Tanaka, Primordial uctuations and non-gaussianities in multi-eld DBI ination, Phys. Rev. Lett. 101 (2008) 061301 [arXiv:0804.3139] [SPIRES];

JCAP05(2009)005

D. Langlois, S. Renaux-Petel, D.A. Steer and T. Tanaka, Primordial perturbations and non-gaussianities in DBI and general multi-eld ination, Phys. Rev. D 78 (2008) 063523 [arXiv:0806.0336] [SPIRES]; F. Arroja, S. Mizuno and K. Koyama, Non-gaussianity from the bispectrum in general multiple eld ination, JCAP 08 (2008) 015 [arXiv:0806.0619] [SPIRES]. [14] C.T. Byrnes, K. Koyama, M. Sasaki and D. Wands, Diagrammatic approach to non-gaussianity from ination, JCAP 11 (2007) 027 [arXiv:0705.4096] [SPIRES]; Y. Rodriguez and C.A. Valenzuela-Toledo, On the issue of the series convergence and loop corrections in the generation of observable primordial non-gaussianity in slow-roll ination. Part II: the trispectrum, arXiv:0811.4092 [SPIRES].

JCAP05(2009)005

10