Home

Search

Collections

Journals

About

Contact us

My IOPscience

Conservation of the nonlinear curvature perturbation in generic single-field inflation

This content has been downloaded from IOPscience. Please scroll down to see the full text.
2011 Class. Quantum Grav. 28 072001
(http://iopscience.iop.org/0264-9381/28/7/072001)
View the table of contents for this issue, or go to the journal homepage for more

Download details:
IP Address: 200.137.65.103
This content was downloaded on 21/08/2015 at 21:10

Please note that terms and conditions apply.

IOP PUBLISHING

CLASSICAL AND QUANTUM GRAVITY

Class. Quantum Grav. 28 (2011) 072001 (7pp)

doi:10.1088/0264-9381/28/7/072001

FAST TRACK COMMUNICATION

Conservation of the nonlinear curvature perturbation

in generic single-field inflation
Atsushi Naruko and Misao Sasaki
Yukawa Institute for Theoretical Physics, Kyoto University, Kyoto 606-8502, Japan
E-mail: naruko@yukawa.kyoto-u.ac.jp and misao@yukawa.kyoto-u.ac.jp

Received 24 January 2011, in final form 15 February 2011

Published 3 March 2011
Online at stacks.iop.org/CQG/28/072001
Abstract
It is known that the curvature perturbation on uniform energy density (or
comoving or uniform Hubble ) slices on superhorizon scales is conserved to full
nonlinear order if the pressure is only a function of the energy density (i.e. if the
perturbation is purely adiabatic), independent of the gravitational theory. Here,
we explicitly show that the same conservation holds for a universe dominated
by a single scalar field provided that the field is in an attractor regime, for
a very general class of scalar-field theories. However, we also show that if
the scalar-field equation contains a second time derivative of the metric, as
in the case of the Galileon (or kinetic braiding) theory, one has to invoke the
gravitational-field equations to show the conservation.
PACS number: 98.80.Cq
1. Introduction
The CMB anisotropy observation by WMAP clearly showed us that the primordial curvature
perturbations are nearly scale invariant and their statistics is Gaussian to very high accuracy
[1]. This is perfectly in agreement with the predictions of the standard, canonical singlefield, slow-roll inflation. Nevertheless, an indication was found in the CMB data as well as
in the large-scale structure data that there may be a detectable level of non-Gaussianity in
the curvature perturbation. Consequently, a lot of attention has been paid in recent years to
possible non-Gaussian features in the primordial curvature perturbation from inflation [2].
(See articles in a focus section in CQG [3] and references therein for recent developments.)
Apparently this direction of research involves nonlinear cosmological perturbations.
There are mainly two approaches to nonlinear cosmological perturbations. One is the
standard perturbative approach [47]. This is basically straightforward and can in principle
deal with most general situations as long as the perturbation expansion is applicable. But
the equations can become very much involved and quite often the physical transparency may
be lost.
The other is the gradient expansion approach [828]. In this approach, the field equations
are expanded in powers of spatial gradients. Hence this is applicable only to perturbations
0264-9381/11/072001+07$33.00 2011 IOP Publishing Ltd Printed in the UK & the USA

Class. Quantum Grav. 28 (2011) 072001

Fast Track Communication

on superhorizon scales. Nevertheless, it has a big advantage that the full nonlinear effects
are taken into account at each order of the gradient expansion. At leading order in gradient
expansion, it corresponds to the separate universe approach [29]. Namely the field equations
become ordinary differential equations with respect to time; hence, the physical quantities at
each spatial point (where each point corresponds to a Hubble horizon size region) evolve in
time independently from those at the rest of the space.
One of the most important results obtained in the gradient expansion approach is that
the full nonlinear curvature perturbation on uniform energy density (or comoving or uniform
Hubble) slices is conserved at leading order in gradient expansion if the pressure is only a
function of the energy density [27], or the perturbation is purely adiabatic. This is shown using
the energy conservation law, without using the Einstein equations. Thus, without solving the
field equations, one can predict the spectrum and the statistics of the curvature perturbation
at horizon re-entry during the late radiation or matter-dominated era once one knows these
properties of the curvature perturbation at horizon exit during inflation.
However, the assumption that the pressure is only a function of the energy density is
not rigorously true in the case of a scalar field. It is only approximately true in the limit of
the slow-roll inflation. In other words, the pressure and energy density perturbations are not
adiabatic in the sense of the standard fluid dynamics. Therefore, it is not completely clear
exactly under which condition the nonlinear curvature perturbation is conserved in the case of
a scalar field.
In this short paper, we focus on a universe dominated by a single scalar field and
explicitly show the conservation of the nonlinear curvature perturbation on comoving slices
( = (t)) at leading order in gradient expansion. We consider a very general theory
of a scalar field [3033], including a Galileon (or kinetic braiding) field which has been
attracting attention recently [3438]. For the gravitational part we assume Einstein gravity for
definiteness, but our discussion is applicable to any metric theory of gravity.
Assuming that the scalar-field dynamics is in an attractor regime so that the value of the
scalar field determines the dynamics completely, we find that the conservation of the nonlinear
curvature perturbation holds without using the gravitational-field equations just the same as
the fluid case, provided that the scalar-field equation contains only first time derivatives of the
metric. This condition is satisfied for a generic K-essence-type scalar field, including the case
of a canonical scalar, but not for a Galileon scalar because the Galileon field equation contains
second time derivatives of the metric. In the latter case, one has to invoke the gravitationalfield equations to see if the conservation still holds or not. In the case of Einstein gravity, the
conservation is shown to hold even for a Galileon scalar field.
2. Basic setup
We focus on the dynamics on superhorizon scales. We associate  with each spatial derivative.
So a quantity with nth spatial derivatives will be of O( n ).
We express the metric in the (3 + 1) form
ds 2 = g dx dx = 2 dt 2 + ij (dx i + i dt)(dx j + j dt),
(2.1)
i
where , and ij are the lapse function, the shift vector, and the spatial metric, respectively.
We choose the spatial coordinates such that i = O( 3 ). We further decompose the spatial
metric as
k
ij = a 2 (t) e2(t,x ) ij (t, x k ),
det ij = 1,
(2.2)
k

where a(t) e(t,x ) is the scale factor at each local point, while a(t) is the scale factor of a
fiducial homogeneous universe. When the gradient expansion is applied to an inflationary
2

Class. Quantum Grav. 28 (2011) 072001

Fast Track Communication

stage of the universe, it is known that we have t ij = O( 2 ) [23, 24, 27]. Then at leading
order in gradient expansion, we identify as the nonlinear curvature perturbation [27].
We consider a theory with the action
S = Sg + S ;


1
Sg = d4 x g 2 R,
2

4
S = d x g[W (X, ) G(X, )  ];

(2.3)
1
X g ,
2

where W and G are arbitrary functions of X and . For simplicity and definiteness, we
consider Einstein gravity, with 2 = 8 G, but our discussion below can be easily extended to
any metric theory of gravity. The energymomentum tensor of scalar field is given by
T = 2

S
= W,X + Wg G,X 
g
+ g G, g (G, + G, ).

(2.4)

Our theory includes a Galileon scalar if G = 0 [34, 35]. If we consider the case
G(X, ) = 0,

(2.5)

but retain a generic W , it corresponds to a K-essential scalar [30, 31], which includes a DBI
model [32, 33] as a special case. If we consider the case
G(X, ) = 0,

W (X, ) = X V (),

(2.6)

we recover the conventional canonical scalar theory.

3. Scalar-field equation
To derive the scalar-field equation, we take the variation of S with respect to :




S = d4 x g W,X X + W, (G,X X + G, )  G 


= d4 x g[(W, G,  ) (W,X G,X  )

+ (G,X X + G, ) ]


= d4 x g[(W, G,  ) + [(W,X G,X  ) ]

(G,X X + G, )],

(3.1)

where we have used the identity

Q X = Q g ( ) = (Q g ) + (Q g ),

(3.2)

and dropped the surface term after integration by parts. Thus, the field equation is given by

1
( g[(W,X G, ) G,X ( + X)]) + W, G,  = 0.
g

(3.3)

We introduce a unit timelike vector n orthonormal to the t = constant hypersurfaces:

n =

1
(1, 0, 0, 0).

(3.4)
3

Class. Quantum Grav. 28 (2011) 072001

Fast Track Communication

For this vector, the expansion K is given by1

1
1

K n ; = ( gn ) = .
g

(3.5)

where is the proper time of an observer with the 4-velocity n ,

d = (t, x k ) dt.

(3.6)

Since we concentrate on the superhorizon behaviour, we neglect the spatial derivatives

in (3.3) to obtain
1

( [(W,X G, ) G,X ( + X)]) + W, G,  = 0,

(3.7)

where
1

 = ( ) = 2 K ,

(3.8)

the above equation can be rewritten as

(W,X 2G, + 2G,X K )2 + ( G,X )K( )2 + G,X ( K + K 2 )( )2
+ [ (W,X G, ) + K(W,X 2G, )] W, = 0.

(3.9)

Here an important comment is in order. In the Galileon case, where G,X = 0, the scalar
field equation contains the second time derivative of the metric, K + K 2 = 2 / . As
we shall see below, the conservation of the nonlinear curvature perturbation can be proved
only if the field equation contains only first time derivatives of the metric. Thus, one has to
use the gravitational-field equations (the Einstein equations in the present case) to eliminate
the second or higher derivative terms.
The relevant components of the Einstein equations are
K 2 = 3 2 E,

K = 32 2 (E + W + g G, , ) = 32 2 (E + W G ), (3.10)

or
K + K 2 = 32 2 (E W + G ),

(3.11)

where E = T n n . Keeping only the time derivatives, E is given by

E=

1
T00 = [W,X G,X ( )]( )2 G W.
2

(3.12)

This gives
K + K 2 = 32 2 ([W,X G,X ( )]( )2 2W )



= 32 2 W,X + G,X K + G,X 2 ( )2 2W .

(3.13)

Eliminating K +K 2 in (3.9) by using equation (3.11), we obtain the scalar-field equation

involving only the first time derivatives of the metric,


(W,X 2G, + 2G,X K )2 + G,XX 2 + G,X K( )3


+ (W,XX G,X )2 + W,X G, ( )2 + K(W,X 2G, ) W,



+ 32 2 G,X ( )2 W,X + G,X K + G,X 2 ( )2 2W = 0.
(3.14)
1

Note the change of notation from [17], in which K is defined by the minus of the expansion; K = n ; .

Class. Quantum Grav. 28 (2011) 072001

Fast Track Communication

4. Conservation law
Quite generally a conservation law corresponds to an integral of motion. This implies that it
will be necessary for the scalar-field equation to be effectively first order in time derivatives
in order to derive a conservation law. In the present case, to show the conservation of the
nonlinear curvature perturbation, we assume that the system has evolved into an attractor stage
so that the time derivative of the scalar field has become a function of ,
= f ().

(4.1)

Note that this may be regarded as a generalization of the slow-roll case. In this regime, the
functions G and W become functions of only:




W = W (X, ) = W 12 f 2 , .
(4.2)
G = G(X, ) = G 12 f 2 , ,
We also assume f = 0. This implies that can be used to determine the time slicing if
desired.
We can rewrite equation (3.14) as
(W,X 2G, + 2G,X Kf )f, f + (G,XX f, f + G,X )Kf 3
+ [(W,XX G,X )f, f + W,X G, ]f 2 + K(W,X 2G, )f W,
+ 32 2 G,X f 2 [(W,X + G,X Kf + G,X f, f )f 2 2W ] = 0.
This equation can be arranged in the form A() + KB() = 0, or
A()
,
K =
B()
where A and B are given by

(4.3)

(4.4)

A() = (W,X 2G, )f, f + [(W,XX G,X )f, f + (W,X G, )]f 2 W,

+ 32 2 G,X f 2 [(W,X + G,X f, f )f 2 2W ],

(4.5)

B() = 2G,X f f, + (G,XX f, f + G,X )f + (W,X 2G, )f +

(4.6)

3 2
(G,X )2 f 5 .
2

Now we recall that K is expressed in terms of the metric components as

= (H + ),
K=

(4.7)

and = /t. Integrating K along the integral curve of n , that is, along x k =
where H a/a
constant, from ti to t, we obtain

 

 t
a(t)

k
k
+ (t, x ) (ti , x ) .
dt K = 3 ln
(4.8)
a(ti )
ti
On the other hand, the integral on the right-hand side of equation (4.4) gives
 t

A(  )
A()
dt 
d 
=
= F () F (i ).
B()
f (  )B(  )
ti
i

(4.9)

Therefore, combining equations (4.8) and (4.9), we find

 

a(t)
+ (t, x k ) (ti , x k ) = F () F (i ),
3 ln
(4.10)
a(ti )
for any t.
So far, we have not specified the time slicing. Now let us choose the uniform slicing,
( (t, x k ), x k ) = (t), or regard as a time coordinate. That is, we choose the comoving
slicing where n T i = 0. In this case, the equation (t) satisfies becomes identical to the one
5

Class. Quantum Grav. 28 (2011) 072001

Fast Track Communication

for the homogeneous and isotropic universe, and the fiducial scale factor a(t) can be chosen
to be the one for this homogeneous and isotropic universe.
Here it is worth mentioning another particular nature of the Galileon field. An explicit
expression for n T i at lowest order in the gradient expansion is
n T i = [(W,X + KG,X 2G, )i G,X i ( )].

(4.11)

Because of the presence of the term proportional to i ( ), it is clear that the uniform
slicing does not necessarily coincide with the comoving slicing in general. However, in the
present case, we have assumed that the system is in an attractor regime where has become a
function of alone, as given by equation (4.1). Therefore, we have i ( ) = i f = f, i .
That is, in the Galileon case, the uniform slicing coincides with the comoving slicing
provided that the system is in an attractor regime.
Then equation (4.10) implies


a(t)
= F () F (i ),
c (t, x k ) = c (ti , x k ),
(4.12)
3 ln
a(ti )
where c is evaluated on comoving slices. This is a proof of the conservation of the
nonlinear curvature perturbation on comoving slices. The key for the proof is the attractor
behaviour of the scalar field, equation (4.1).
5. Conclusion
We have shown that the nonlinear curvature perturbation on comoving slices is conserved
on superhorizon scales for a very general class of single-field inflation. It can be derived by
using only the scalar-field equation if it contains only first derivatives of the metric, while the
gravitational equations are necessary if it contains second or higher derivatives of the metric.
The key condition is that the scalar field is in an attractor regime so that can be taken as a
time coordinate.
Acknowledgments
This work was supported in part by Monbukagaku-sho Grant-in-Aid for the Global COE
programs, the Next Generation of Physics, Spun from Universality and Emergenceh at Kyoto
University. The work of MS was supported by JSPS Grant-in-Aid for Scientific Research (A)
no 21244033, and by Grant-in-Aid for Creative Scientific Research no 19GS0219. AN was
supported by Grant-in-Aid for JSPS Fellows no 21-1899.
References
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
6

Komatsu E et al (WMAP Collaboration) 2009 Astrophys. J. Suppl. 180 330 (arXiv:0803.0547)

Komatsu E and Spergel D N 2001 Phys. Rev. D 63 063002 (arXiv:astro-ph/0005036)
Sasaki M and Wands D 2010 Class. Quantum Grav. 27 120301
Maldacena J M 2003 J. High Energy Phys. JHEP05(2003)013 (arXiv:astro-ph/0210603)
Malik K A and Wands D 2004 Class. Quantum Grav. 21 L65 (arXiv:astro-ph/0307055)
Seery D and Lidsey J E 2005 J. Cosmol. Astropart. Phys. JCAP06(2005)003 (arXiv:astro-ph/0503692)
Malik K A 2007 J. Cosmol. Astropart. Phys. JCAP03(2007)004 (arXiv:astro-ph/0610864)
Lifshitz E M and Khalatnikov I M 1963 Adv. Phys. 12 185
Belinsky V A, Khalatnikov I M and Lifshitz E M 1982 Adv. Phys. 31 639
Starobinsky A A 1985 JETP Lett. 42 152
Bardeen J M 1980 Phys. Rev. D 22 1882
Salopek D S and Bond J R 1990 Phys. Rev. D 42 3936

Class. Quantum Grav. 28 (2011) 072001

[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22]
[23]
[24]
[25]
[26]
[27]
[28]
[29]
[30]
[31]
[32]
[33]
[34]
[35]
[36]
[37]
[38]

Fast Track Communication

Deruelle N and Langlois D 1995 Phys. Rev. D 52 2007 (arXiv:gr-qc/9411040)

Sasaki M and Stewart E D 1996 Prog. Theor. Phys. 95 71 (arXiv:astro-ph/9507001)
Nambu Y and Taruya A 1996 Class. Quantum Grav. 13 705 (arXiv:astro-ph/9411013)
Sasaki M and Tanaka T 1998 Prog. Theor. Phys. 99 763 (arXiv:gr-qc/9801017)
Shibata M and Sasaki M 1999 Phys. Rev. D 60 084002 (arXiv:gr-qc/9905064)
Lyth D H and Rodriguez Y 2005 Phys. Rev. D 71 123508 (arXiv:astro-ph/0502578)
Rigopoulos G I and Shellard E P S 2003 Phys. Rev. D 68 123518 (arXiv:astro-ph/0306620)
Rigopoulos G I and Shellard E P S 2005 J. Cosmol. Astropart. Phys. JCAP10(2005)006 (arXiv:astro-ph/
0405185)
Yokoyama S, Suyama T and Tanaka T 2007 J. Cosmol. Astropart. Phys. JCAP07(2007)013 (arXiv:0705.3178)
Yokoyama S, Suyama T and Tanaka T 2008 Phys. Rev. D 77 083511 (arXiv:0711.2920)
Tanaka Y and Sasaki M 2007 Prog. Theor. Phys. 117 633 (arXiv:gr-qc/0612191)
Tanaka Y and Sasaki M 2007 Prog. Theor. Phys. 118 455 (arXiv:0706.0678)
Takamizu Y-i and Mukohyama S 2009 J. Cosmol. Astropart. Phys. JCAP01(2009)013 (arXiv:0810.0746)
Takamizu Y-i, Mukohyama S, Sasaki M and Tanaka Y 2010 J. Cosmol. Astropart. Phys. JCAP06(2010)019
(arXiv:1004.1870)
Lyth D H, Malik K A and Sasaki M 2005 J. Cosmol. Astropart. Phys. JCAP05(2005)004 (arXiv:astro-ph/
0411220)
Afshordi N and Brandenberger R H 2001 Phys. Rev. D 63 123505 (arXiv:gr-qc/0011075)
Wands D, Malik K A, Lyth D H and Liddle A R 2000 Phys. Rev. D 62 043527 (arXiv:astro-ph/0003278)
Armendariz-Picon C, Damour T and Mukhanov V F 1999 Phys. Lett. B 458 209 (arXiv:hep-th/9904075)
Garriga J and Mukhanov V F 1999 Phys. Lett. B 458 219 (arXiv:hep-th/9904176)
Silverstein E and Tong D 2004 Phys. Rev. D 70 103505 (arXiv:hep-th/0310221)
Alishahiha M, Silverstein E and Tong D 2004 Phys. Rev. D 70 123505 (arXiv:hep-th/0404084)
Kobayashi T, Yamaguchi M and Yokoyama J 2010 Phys. Rev. Lett. 105 231302 (arXiv:1008.0603)
Nicolis A, Rattazzi R and Trincherini E 2009 Phys. Rev. D 79 064036 (arXiv:0811.2197)
Mizuno S and Koyama K 2010 Phys. Rev. D 82 103518 (arXiv:1009.0677)
Deffayet C, Pujolas O, Sawicki I and Vikman A 2010 J. Cosmol. Astropart. Phys. JCAP10(2010)026
(arXiv:1008.0048)
Kamada K, Kobayashi T, Yamaguchi M and Yokoyama J 2010 arXiv:1012.4238 [astro-ph.CO]