This action might not be possible to undo. Are you sure you want to continue?
r
X
i
v
:
0
9
0
1
.
0
2
6
5
v
1
[
h
e
p

t
h
]
2
J
a
n
2
0
0
9
Annu. Rev. Nuc. Part. Sci. 2009 59
Advances in Inﬂation in String Theory
Daniel Baumann
Department of Physics, Harvard University, Cambridge, MA 02138
Liam McAllister
Department of Physics, Cornell University, Ithaca, NY 14853
Key Words Inﬂation, String Theory, Cosmology
Abstract We provide a pedagogical overview of inﬂation in string theory. Our theme is
the sensitivity of inﬂation to Planckscale physics, which we argue provides both the primary
motivation and the central theoretical challenge for the subject. We illustrate these issues
through two case studies of inﬂationary scenarios in string theory: warped Dbrane inﬂation
and axion monodromy inﬂation. Finally, we indicate how future observations can test scenarios
of inﬂation in string theory.
CONTENTS
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Inﬂation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
Motivation for Inﬂation in String Theory . . . . . . . . . . . . . . . . . . . . . . . . . 6
Organization of this Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
Inﬂation in String Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1
Annu. Rev. Nuc. Part. Sci. 2009 59
Inﬂation in Eﬀective Field Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
The Eta Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
From String Compactiﬁcations to the Inﬂaton Action . . . . . . . . . . . . . . . . . . . 13
Case Study of SmallField Inﬂation: Warped Dbrane Inﬂation . . . . . . . . . . . 16
D3branes in Warped Throat Geometries . . . . . . . . . . . . . . . . . . . . . . . . . . 16
The D3brane Potential . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
Supergravity Analysis of the D3brane Potential . . . . . . . . . . . . . . . . . . . . . . 20
Gauge Theory Interpretation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
Summary and Perspective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
LargeField Inﬂation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
The Lyth Bound . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
SuperPlanckian Fields and Flat Potentials . . . . . . . . . . . . . . . . . . . . . . . . 33
Case Study of LargeField Inﬂation: Axion Monodromy . . . . . . . . . . . . . . . 35
Axions in String Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
Axion Inﬂation in String Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
Compactiﬁcation Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
Summary and Perspective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
Theoretical Prospects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
Observational Prospects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
2
Annu. Rev. Nuc. Part. Sci. 2009 59
1 Introduction
Recent advances in observational cosmology have brought us closer to a funda
mental understanding of the origin of structure in the universe. Observations
of variations in the cosmic microwave background (CMB) temperature and of
the spatial distribution of galaxies in the sky have yielded a consistent picture
in which gravitational instability drives primordial ﬂuctuations to condense into
largescale structures, such as our own galaxy. Moreover, quantum ﬁeld theory
and general relativity provide an elegant microphysical mechanism, inﬂation, for
generating these primordial perturbations during an early period of accelerated
expansion. The classical dynamics of this inﬂationary era (1, 2, 3) explains the
largescale homogeneity, isotropy and ﬂatness of the universe, while quantum
ﬂuctuations during inﬂation lead to small inhomogeneities. The general prop
erties of the spectrum of inﬂationary inhomogeneities were predicted long ago
(4, 5, 6, 7, 8, 9) and are in beautiful agreement with recent CMB observations,
e.g. by the Wilkinson Microwave Anisotropy Probe (WMAP) (10).
Although inﬂation is remarkably successful as a phenomenological model for
the dynamics of the very early universe, a detailed understanding of the physical
origin of the inﬂationary expansion has remained elusive. In this review we will
highlight speciﬁc aspects of inﬂation that depend sensitively on the ultraviolet
(UV) completion of quantum ﬁeld theory and gravity, i.e. on the ﬁeld content
and interactions at energies approaching the Planck scale. Such issues are most
naturally addressed in a theory of Planckscale physics, for which string theory
is the bestdeveloped candidate. This motivates understanding the physics of
inﬂation in string theory.
3
Annu. Rev. Nuc. Part. Sci. 2009 59
1.1 Inﬂation
Inﬂation may be deﬁned as a period of exponential expansion of space,
ds
2
= −dt
2
+e
2Ht
dx
2
, where ǫ ≡ −
˙
H
H
2
< 1 , (1)
which arises if the universe is dominated by a form of stressenergy that sources
a nearlyconstant Hubble parameter H. This requirement for accelerated expan
sion can be fulﬁlled by a range of qualitatively diﬀerent mechanisms with varied
theoretical motivations (11, 12, 13). For concreteness, we restrict ourselves to the
simple case of singleﬁeld slowroll inﬂation, where the inﬂationary dynamics is
described by a single order parameter φ (a fundamental scalar ﬁeld or a compos
ite ﬁeld) with canonical kinetic term
1
2
(∂
µ
φ)
2
and potential energy density V (φ).
Prolonged accelerated expansion then occurs if the slowroll parameters are small
ǫ ≃
M
2
pl
2
_
V
′
V
_
2
≪1 , η = M
2
pl
¸
¸
¸
¸
V
′′
V
¸
¸
¸
¸
≪1 , (2)
where the primes denote derivatives with respect to the inﬂaton ﬁeld φ.
In addition to smoothing the universe on large scales, inﬂation stretches quan
tum ﬂuctuations of light degrees of freedom (m ≪ H), creating a spectrum of
small perturbations in the observed CMB temperature and polarization. These
perturbations are the key to structure formation and to tests of inﬂation, so we
pause to explain them; see e.g. Ref. (12, 13) for a more detailed discussion.
In inﬂationary scenarios with a single inﬂaton ﬁeld, the light degrees of free
dom during inﬂation are the inﬂaton itself and the two polarization modes of
the graviton. Fluctuations of the inﬂaton lead to perturbations of the time at
which inﬂation ends, and hence source perturbations in the energy density af
ter inﬂation. These ﬂuctuations are visible to us as anisotropies in the CMB
temperature. Both the inﬂaton (scalar) ﬂuctuations and the graviton (tensor)
4
Annu. Rev. Nuc. Part. Sci. 2009 59
ﬂuctuations source polarization of the CMB photons. Ref. (14, 15) made the im
portant observation that the spin2 polarization ﬁeld of the CMB photons may
be decomposed into two distinct scalar or spin0 modes:
• Emode: this curlfree mode is characterized by polarization vectors that
are radial around cold spots and tangential around hot spots on the sky, and
is generated by both scalar and tensor perturbations. Emode polarization
and its crosscorrelations with the CMB temperature ﬂuctuations were ﬁrst
detected by DASI (16) and have recently been mapped out in greater detail
by WMAP (17).
• Bmode: this divergencefree mode is characterized by polarization vectors
with vorticity around any point on the sky. Primordial Bmodes can only
be produced by gravitational waves and are therefore considered an unam
biguous signature of inﬂationary tensor perturbations. The energy scale of
inﬂation determines the amplitude of tensor perturbations, and hence the
Bmode amplitude, so that a detection would ﬁx the inﬂationary energy
scale. The search for primordial Bmodes is a key eﬀort of observational
cosmology (18).
The nature of the inﬂationary epoch is imprinted on the sky in the temperature
and polarization anisotropies of the CMB. In slowroll inﬂation, small ǫ and η
ensure that the spectra of scalar and tensor ﬂuctuations are nearly scaleinvariant.
The shapes of the primordial perturbation spectra are therefore intimately tied
to the inﬂationary background dynamics as dictated by the shape of the inﬂaton
potential V (φ).
CMB observations have improved dramatically in the past decade, and near
future experiments will almost certainly continue this trend. Temperature aniso
5
Annu. Rev. Nuc. Part. Sci. 2009 59
tropies have been measured at the cosmic variance limit over a large range of
scales, and experimentalists are now preparing for precision measurements of the
polarization of the CMB (18). A detection of inﬂationary tensor perturbations
via their unique Bmode signature would be especially interesting, as their am
plitude relates directly to the energy scale of inﬂation. This provides a unique
opportunity to probe physics at energies near the GUT scale, far out of reach of
terrestrial collider experiments.
1.2 Motivation for Inﬂation in String Theory
Besides the intellectual satisfaction of providing a microscopic description of the
inﬂationary process, there are more detailed motivations for studying inﬂation in
the context of string theory. While inﬂation is frustratingly eﬀective at making
most signatures of highenergy physics unobservable, e.g. by exponentially dilut
ing any preexisting density of GUTscale relics, the duration and the details of
inﬂation are nevertheless sensitive to certain aspects of Planckscale physics. In
the remainder of this review we ﬂesh this out in more detail, but we now brieﬂy
preview two examples of the UV sensitivity of inﬂation that will be central to
our discussion.
1.2.1 Flatness of the Inflaton Potential From a topdown perspec
tive the ﬂatness of the inﬂaton potential in Planck units, as quantiﬁed by the
slowroll conditions, Eq. (2), is a nontrivial constraint. As we will show in §2,
small (Plancksuppressed) corrections to the potential often induce important
corrections to the curvature of the potential, ∆η ∼ O(1). To assess whether inﬂa
tion can nevertheless occur requires detailed information about Plancksuppressed
corrections to the inﬂaton potential. This requires either phenomenological as
6
Annu. Rev. Nuc. Part. Sci. 2009 59
sumptions, or preferably microphysical knowledge, about physics at the Planck
scale. String theory equips us to compute such corrections from ﬁrst principles,
and in §3 we give an example of such an analysis for the case of warped Dbrane
inﬂation (19).
1.2.2 Inflationary Gravitational Waves As we explain in §4, the UV
sensitivity of inﬂation is particularly strong in models with observable gravita
tional waves. A large gravitational wave signal from inﬂation is associated with
a high energy scale for the inﬂaton potential and a superPlanckian variation of
the inﬂaton ﬁeld, ∆φ ≫ M
pl
, between the time when CMB ﬂuctuations were
created and the end of inﬂation. In §4 we explain why theoretical control of the
shape of the potential over a superPlanckian range requires certain assumptions
about the UV structure of the theory. Models of largeﬁeld inﬂation are therefore
most naturally studied in a UVcomplete theory, such as string theory. In §5
we present the ﬁrst controlled examples of largeﬁeld inﬂation in string theory
(20, 21).
Given the exciting possibility of measuring the gravitational wave signature
of inﬂation in the polarization of the CMB, the issue of controlled largeﬁeld
inﬂation is of both theoretical and experimental relevance (12).
1.3 Organization of this Review
Inﬂation in string theory is the subject of close to 1,000 papers, and space con
siderations prevent us from presenting a truly comprehensive review that sum
marizes and assesses each important class of models. (Some representative con
tributions to the subject include (22); we refer the reader to the reviews (23) for
a more complete list of references.) Instead, our goal is an exposition of what
7
Annu. Rev. Nuc. Part. Sci. 2009 59
we believe to be the primary theme of the subject: the sensitivity of inﬂation
to Planckscale physics. As we will explain, this is the twoway connection by
which string theory can clarify inﬂationary modelbuilding, and cosmological ex
periments can constrain string theory models. To illustrate this idea in depth,
we focus on two case studies: warped Dbrane inﬂation and axion monodromy
inﬂation. These two scenarios are instructive examples from the general classes
of smallﬁeld and largeﬁeld inﬂation, respectively.
2 Inﬂation in String Theory
2.1 Inﬂation in Eﬀective Field Theory
As a phenomenon in quantum ﬁeld theory coupled to general relativity, inﬂation
does not appear to be natural. In particular, the set of Lagrangians suitable for
inﬂation is a minute subset of the set of all possible Lagrangians. Moreover, in
wide classes of models, inﬂation emerges only for rather special initial conditions,
e.g. initial conﬁgurations with tiny kinetic energy, in the case of smallﬁeld sce
narios. Although one would hope to explore and quantify the naturalness both
of inﬂationary Lagrangians and of inﬂationary initial conditions, the question of
initial conditions appears inextricable from the active yet incomplete program
of understanding measures in eternal inﬂation (24). (However, see e.g. (25) for
recent eﬀorts to quantify or to ameliorate the ﬁnetuning of initial conditions.)
In this review we will focus on the question of how (un)natural it is to have a
Lagrangian suitable for inﬂation.
For a single inﬂaton ﬁeld with a canonical kinetic term, the necessary conditions
for inﬂation can be stated in terms of the inﬂaton potential. Inﬂation requires a
potential that is quite ﬂat in Planck units (see Eq. (2)), and as we now argue,
8
Annu. Rev. Nuc. Part. Sci. 2009 59
this condition is sensitive to Planckscale physics. Let us recall that the presence
of some form of new physics at the Planck scale is required in order to render
gravitongraviton scattering sensible, just as unitarity of WW scattering requires
new physics at the TeV scale. Although we know that new degrees of freedom
must emerge, we cannot say whether the physics of the Planck scale is a ﬁnite
theory of quantum gravity, such as string theory, or is instead simply an eﬀective
theory for some unimagined physics at yet higher scales. However, the structure
of the Planckscale theory has meaningful – and, in very favorable cases, testable
– consequences for the form of the inﬂaton potential.
As usual, the eﬀects of highscale physics above some cutoﬀ Λ are eﬃciently
described by the coeﬃcients of operators in the lowenergy eﬀective theory. In
tegrating out particles of mass M ≥ Λ gives rise to operators of the form
O
δ
M
δ−4
, (3)
where δ denotes the mass dimension of the operator.
Sensitivity to such operators is commonplace in particle physics: for example,
bounds on ﬂavorchanging processes place limits on physics above the TeV scale,
and lower bounds on the proton lifetime even allow us to constrain GUTscale
operators that would mediate proton decay. However, particle physics consid
erations alone do not often reach beyond operators of dimension δ = 6, nor go
beyond M ∼ M
GUT
. (Scenarios of gravitymediated supersymmetry breaking
are one exception.) Equivalently, Planckscale processes, and operators of very
high dimension, are irrelevant for most of particle physics: they decouple from
lowenergy phenomena.
In inﬂation, however, the ﬂatness of the potential in Planck units introduces
9
Annu. Rev. Nuc. Part. Sci. 2009 59
sensitivity to δ ≤ 6 Plancksuppressed operators, such as
O
6
M
2
pl
. (4)
As we explain in §2.2, an understanding of such operators is required to address
the smallness of the eta parameter, i.e. to ensure that the theory supports at least
60 efolds of inﬂationary expansion. This sensitivity to dimensionsix Planck
suppressed operators is therefore common to all models of inﬂation.
For largeﬁeld models of inﬂation the UV sensitivity of the inﬂaton action is
dramatically enhanced. As we discuss in §4, in this important class of inﬂation
ary models the potential becomes sensitive to an inﬁnite series of operators of
arbitrary dimension.
2.2 The Eta Problem
In the absence of any speciﬁc symmetries protecting the inﬂaton potential, con
tributions to the Lagrangian of the general form
O
6
M
2
pl
=
O
4
M
2
pl
φ
2
(5)
are allowed. If the dimensionfour operator O
4
has a vacuum expectation value
(vev) comparable to the inﬂationary energy density, O
4
∼ V , then this term
corrects the inﬂaton mass by order H, or equivalently corrects the eta parameter
by order one, leading to an important problem for inﬂationary modelbuilding.
Let us reiterate that contributions of this form may be thought of as arising from
integrating out Planckscale degrees of freedom. In this section we discuss this
socalled eta problem in eﬀective ﬁeld theory, §2.2.1, and illustrate the problem
in a supergravity example, §2.2.2.
10
Annu. Rev. Nuc. Part. Sci. 2009 59
2.2.1 Radiative Instability of the Inflaton Mass In a generic ef
fective theory with cutoﬀ Λ, the mass of a scalar ﬁeld runs to the cutoﬀ scale
unless it is protected by some symmetry. Since the cutoﬀ for an eﬀective theory
of inﬂation is at least the Hubble scale, Λ ≥ H, this implies that a small inﬂaton
mass (m
φ
≪H) is radiatively unstable. Equivalently, the eta parameter receives
radiative corrections,
∆η =
∆m
2
φ
3H
2
≥ 1 , (6)
preventing prolonged inﬂation.
The diﬃculty here is analogous to the Higgs hierarchy problem, but super
symmetry does not suﬃce to stabilize the inﬂaton mass: the inﬂationary energy
necessarily breaks supersymmetry, and the resulting splittings in supermultiplets
are of order H, so that supersymmetry does not protect a small inﬂaton mass
m
φ
≪H.
In §5 we discuss the natural proposal to protect the inﬂaton potential via a
shift symmetry φ → φ + const., which is equivalent to identifying the inﬂaton
with a pseudoNambuGoldstoneboson. In the absence of such a symmetry the
eta problem seems to imply the necessity of ﬁnetuning the inﬂationary action in
order to get inﬂation.
2.2.2 Supergravity Example An important instance of the eta problem
arises in locallysupersymmetric theories, i.e. in supergravity (26). This case
is relevant for many string theory models of inﬂation because fourdimensional
supergravity is the lowenergy eﬀective theory of supersymmetric string compact
iﬁcations.
In N = 1 supergravity, a key term in the scalar potential is the Fterm poten
11
Annu. Rev. Nuc. Part. Sci. 2009 59
tial,
V
F
= e
K/M
2
pl
_
K
ϕ¯ ϕ
D
ϕ
WD
ϕ
W −
3
M
2
pl
W
2
_
, (7)
where K(ϕ, ¯ ϕ) and W(ϕ) are the K¨ ahler potential and the superpotential, re
spectively; ϕ is a complex scalar ﬁeld which is taken to be the inﬂaton; and we
have deﬁned D
ϕ
W ≡ ∂
ϕ
W + M
−2
pl
(∂
ϕ
K)W. For simplicity of presentation, we
have assumed that there are no other light degrees of freedom, but generalizing
our expressions to include other ﬁelds is straightforward.
The K¨ ahler potential determines the inﬂaton kinetic term, −K
,ϕ¯ ϕ
∂ϕ∂ ¯ ϕ, while
the superpotential determines the interactions. To derive the inﬂaton mass, we
expand K around some chosen origin, which we denote by ϕ ≡ 0 without loss of
generality, i.e. K(ϕ, ¯ ϕ) = K
0
+ K
,ϕ¯ ϕ

0
ϕ¯ ϕ + · · · . The inﬂationary Lagrangian
then becomes
L ≈ −K
,ϕ¯ ϕ
∂ϕ∂ ¯ ϕ −V
0
_
1 + K
,ϕ¯ ϕ

0
ϕ¯ ϕ
M
2
pl
+. . .
_
(8)
≡ −∂φ∂
¯
φ −V
0
_
1 +
φ
¯
φ
M
2
pl
_
+. . . , (9)
where we have deﬁned the canonical inﬂaton ﬁeld φ
¯
φ ≈ K
ϕ¯ ϕ

0
ϕ¯ ϕ and V
0
≡
V
F

ϕ=0
. We have retained the leading correction to the potential originating in
the expansion of e
K/M
2
pl
in Eq. (7), which could plausibly be called a universal cor
rection in Fterm scenarios. The omitted terms, some of which can be of the same
order as the terms we keep, arise from expanding
_
K
ϕ¯ ϕ
D
ϕ
WD
ϕ
W −
3
M
2
pl
W
2
_
in Eq. (7) and clearly depend on the modeldependent structure of the K¨ ahler
potential and the superpotential.
The result is of the form of Eq. (4) with
O
6
= V
0
φ
¯
φ (10)
12
Annu. Rev. Nuc. Part. Sci. 2009 59
and implies a large modelindependent contribution to the eta parameter
∆η = 1 , (11)
as well as a modeldependent contribution which is typically of the same order. It
is therefore clear that in an inﬂationary scenario driven by an Fterm potential,
eta will generically be of order unity.
Under what circumstances can inﬂation still occur, in a model based on a su
persymmetric Lagrangian? One obvious possibility is that the modeldependent
contributions to eta approximately cancel the modelindependent contribution,
so that the smallness of the inﬂaton mass is a result of ﬁnetuning. In the case
study of §3 we will provide a concrete example in which the structure of all rele
vant contributions to eta can be computed, so that one can sensibly pursue such
a ﬁnetuning argument.
Clearly, it would be far more satisfying to exhibit a mechanism that removes
the eta problem by ensuring that ∆η ≪ 1. This requires either that the F
term potential is negligible, or that the inﬂaton does not appear in the Fterm
potential. The ﬁrst case does not often arise, because Fterm potentials play an
important role in presentlyunderstood models for stabilization of the compact
dimensions of string theory (27, 28). However, in §5 we will present a scenario
in which the inﬂaton is an axion and does not appear in the K¨ ahler potential,
or in the Fterm potential, to any order in perturbation theory. This evades the
particular incarnation of the eta problem that we have described above.
2.3 From String Compactiﬁcations to the Inﬂaton Action
2.3.1 Elements of String Compactifications It is a famous fact that
the quantum theory of strings is naturally deﬁned in more than four spacetime
13
Annu. Rev. Nuc. Part. Sci. 2009 59
dimensions, with fourdimensional physics emerging upon compactiﬁcation of the
additional spatial dimensions. For concreteness, we will focus on compactiﬁca
tions of the critical tendimensional type IIB string theory on sixdimensional
CalabiYau spaces (to be precise, our compactiﬁcations will only be conformal to
spaces that are wellapproximated by orientifolds of CalabiYau manifolds, but
we will not need this ﬁne point.)
The vast number of distinct compactiﬁcations in this class are distinguished by
their topology, geometry, and discrete data such as quantized ﬂuxes and wrapped
Dbranes. A central task in string theory modelbuilding is to understand in detail
how the tendimensional sources determine the fourdimensional eﬀective theory.
If we denote the tendimensional compactiﬁcation data by C, the procedure in
question may be written schematically as
S
10
[C] → S
4
. (12)
Distinct compactiﬁcation data C give rise to a multitude of fourdimensional
eﬀective theories S
4
with varied ﬁeld content, kinetic terms, scalar potentials,
and symmetry properties. By understanding the space of possible data C and the
nature of the map in Eq. (12), we can hope to identify, and perhaps even classify,
compactiﬁcations that give rise to interesting fourdimensional physics.
2.3.2 The Effective Inflaton Action For our purposes, the most im
portant degrees of freedom of the eﬀective theory are fourdimensional scalar
ﬁelds. Scalar ﬁelds known as moduli arise from deformations of the compactiﬁ
cation manifold, typically numbering in the hundreds for the CalabiYau spaces
under consideration, and from the positions, orientations, and gauge ﬁeld conﬁg
urations of any Dbranes. From given compactiﬁcation data one can compute the
kinetic terms and scalar potentials of the moduli; in turn, the expectation values
14
Annu. Rev. Nuc. Part. Sci. 2009 59
of the moduli determine the parameters of the fourdimensional eﬀective the
ory. In the presence of generic tendimensional sources of stressenergy, such as
Dbranes and quantized ﬂuxes, there is an energy cost for deforming the compact
iﬁcation, and many (though not always all) of the moduli ﬁelds become massive
(28).
It is useful to divide the scalar ﬁelds arising in S
4
into a set of light ﬁelds φ, ψ
with masses below the Hubble scale (m
φ
, m
ψ
≪ H) and a set of heavy ﬁelds χ
with masses much greater than the Hubble scale (m
χ
≫ H). Here one of the
light ﬁelds, denoted φ, has been identiﬁed as the inﬂaton candidate.
To understand whether successful inﬂation can occur, one must understand all
the scalar ﬁelds, both heavy and light. First, suﬃciently massive moduli ﬁelds are
eﬀectively frozen during inﬂation, and one should integrate them out to obtain
an eﬀective action for the light ﬁelds only,
S
4
(φ, ψ, χ) → S
4,eﬀ
(φ, ψ) . (13)
Integrating out these heavy modes generically induces contributions to the po
tential of the putative inﬂaton: that is, moduli stabilization contributes to the
eta problem. This is completely analogous to the appearance of corrections from
higherdimension operators in our discussion of eﬀective ﬁeld theory in §2.1.
Next, if scalar ﬁelds in addition to the inﬂaton are light during inﬂation, they
typically have important eﬀects on the dynamics, and one should study the evo
lution of all ﬁelds ψ with masses m
ψ
≪ H. Moreover, even if the resulting
multiﬁeld inﬂationary dynamics is suitable, light degrees of freedom can create
problems for latetime cosmology. Light scalars absorb energy during inﬂation
and, if they persist after inﬂation, they can release this energy during or after
Big Bang nucleosynthesis, spoiling the successful predictions of the light ele
15
Annu. Rev. Nuc. Part. Sci. 2009 59
ment abundances. Moreover, light moduli would be problematic in the present
universe, as they mediate ﬁfth forces of gravitational strength. To avoid these
latetime problems, it suﬃces to ensure that m
ψ
≫ 30 TeV, as in this case the
moduli decay before Big Bang nucleosynthesis. A simplifying assumption that is
occasionally invoked is that all ﬁelds aside from the inﬂaton should have m ≫H,
but this is not required on physical grounds: it serves only to arrange that the
eﬀective theory during inﬂation has only a single degree of freedom.
3 Case Study of SmallField Inﬂation: Warped Dbrane Inﬂation
In string theory models of inﬂation the operators contributing to the inﬂaton
potential can be enumerated, and in principle even their coeﬃcients can be com
puted in terms of given compactiﬁcation data. To illustrate these issues, it is use
ful to examine a concrete model in detail. In the following we therefore present
a case study of a comparatively wellunderstood model of smallﬁeld inﬂation,
warped Dbrane inﬂation.
3.1 D3branes in Warped Throat Geometries
In this scenario inﬂation is driven by the motion of a D3brane in a warped throat
region of a stabilized compact space (29). To preserve fourdimensional Lorentz
(or de Sitter) invariance, the D3brane ﬁlls our fourdimensional spacetime and
is pointlike in the extra dimensions (see Figure 1). The global compactiﬁcation
is assumed to be a warped product of fourdimensional spacetime (with metric
g
µν
) and a conformallyCalabiYau space,
ds
2
= e
2A(y)
g
µν
dx
µ
dx
ν
+e
−2A(y)
g
mn
dy
m
dy
n
, (14)
16
Annu. Rev. Nuc. Part. Sci. 2009 59
with g
mn
a CalabiYau metric that can be approximated in some region by a cone
over a ﬁvedimensional Einstein manifold X
5
,
g
mn
dy
m
dy
n
= dr
2
+r
2
ds
2
X
5
. (15)
A canonical example of such a throat region is the KlebanovStrassler (KS) ge
ometry (30), for which X
5
is the
_
SU(2) × SU(2)
_
/U(1) coset space T
1,1
, and
the wouldbe conical singularity at the tip of the throat, r = 0, is smoothed by
the presence of appropriate ﬂuxes. The tip of the throat is therefore located at a
ﬁnite radial coordinate r
IR
, while at r = r
UV
the throat is glued into an unwarped
bulk geometry. In the relevant regime r
IR
≪ r < r
UV
the warp factor may be
written as (31)
e
−4A(r)
=
R
4
r
4
ln
r
r
IR
, R
4
≡
81
8
(g
s
Mα
′
)
2
, (16)
where
ln
r
UV
r
IR
≈
2πK
3g
s
M
. (17)
Here, M and K are integers specifying the ﬂux background (30, 32).
Warping sourced by ﬂuxes is commonplace in modern compactiﬁcations, and
there has been much progress in understanding the stabilization of the moduli of
such a compactiﬁcation (28). Positing a stabilized compactiﬁcation containing a
KS throat therefore seems reasonable given present knowledge.
Inﬂation proceeds as a D3brane moves radially inward in the throat region,
towards an antiD3brane that is naturally situated at the tip of the throat. The
inﬂaton kinetic term is determined by the DiracBornInfeld (DBI) action for
a probe D3brane, and leads to an identiﬁcation of the canonical inﬂaton ﬁeld
with a multiple of the radial coordinate, φ
2
≡ T
3
r
2
. Here, T
3
≡
_
(2π)
3
g
s
α
′2
¸
−1
is the D3brane tension, with g
s
the string coupling and 2πα
′
the inverse string
17
Annu. Rev. Nuc. Part. Sci. 2009 59
tension. The exit from inﬂation occurs when open strings stretched between the
approaching pair become tachyonic and condense, annihilating the branes.
warped throat
r
D3
D3
Ψ
bulk
Figure 1: D3brane inﬂation in a warped throat geometry. The D3branes are
spacetimeﬁlling in four dimensions and therefore pointlike in the extra dimen
sions. The circle stands for the base manifold X
5
with angular coordinates Ψ.
The brane moves in the radial direction r. At r
UV
the throat attaches to a com
pact CalabiYau space. AntiD3branes minimize their energy at the tip of the
throat, r
IR
.
In this simpliﬁed picture, inﬂation is driven by the extremely weak (warping
suppressed) Coulomb interaction of the braneantibrane pair (29). The true story,
however, is more complex, as moduli stabilization introduces new terms in the
inﬂaton potential which typically overwhelm the Coulomb term and drive more
complicated dynamics (29, 33, 34, 35, 36, 19). This pattern is precisely what we
anticipated in our eﬀective ﬁeld theory discussion: integrating out moduli ﬁelds
can be expected to induce important corrections to the potential.
3.2 The D3brane Potential
An important correction induced by moduli stabilization is the inﬂaton mass term
arising from the supergravity Fterm potential, §2.2.2. In a vacuum stabilized
18
Annu. Rev. Nuc. Part. Sci. 2009 59
by an Fterm potential, i.e. by superpotential terms involving the moduli, one
ﬁnds the mass term H
2
0
φ
2
=
1
3
V
0
(φ
⋆
)
φ
2
M
2
pl
(29), where φ
⋆
is an arbitrary reference
value for the inﬂaton ﬁeld and the parameter H
0
should not be confused with the
presentday Hubble constant.
However, one expects additional contributions to the potential from a variety
of other sources, such as additional eﬀects in the compactiﬁcation that break
supersymmetry (19). Let us deﬁne ∆V (φ) to encapsulate all contributions to the
potential aside from the Coulomb interaction V
0
(φ) and the mass term H
2
0
φ
2
;
then the total potential and the associated contributions to the eta parameter
may be written as
V (φ) = V
0
(φ) + H
2
0
φ
2
+ ∆V (φ) (18)
η(φ) = η
0
+
2
3
+ ∆η(φ) = ? (19)
where η
0
≪ 1 because the Coulomb interaction is very weak. (More generally,
V
0
(φ) can be deﬁned to be all terms in V (φ) with negligible contributions to
η. Besides the braneantibrane Coulomb interaction, this can include any other
sources of nearlyconstant energy, e.g. bulk contributions to the cosmological
constant.)
Clearly, η can only be small if ∆V can cancel the mass term in Eq. (18).
We must therefore enumerate all relevant contributions to ∆V , and attempt
to understand the circumstances under which an approximate cancellation can
occur. Note that identifying a subset of contributions to ∆V while remaining
ignorant of others is insuﬃcient.
Warped Dbrane inﬂation has received a signiﬁcant amount of theoretical at
tention in part because of its high degree of computability. Quite generally, if
we had access to the full data of an explicit, stabilized compactiﬁcation with
19
Annu. Rev. Nuc. Part. Sci. 2009 59
small curvatures and weak string coupling, we would in principle be able to com
pute the potential of a Dbrane inﬂaton to any desired accuracy, by performing
a careful dimensional reduction. This is not possible at present for a generic
compact CalabiYau, for two reasons: for general CalabiYau spaces hardly any
metric data is available, and examples with entirely explicit moduli stabilization
are rare.
However, a suﬃciently long throat is wellapproximated by a noncompact throat
geometry (i.e., a throat of inﬁnite length), for which the CalabiYau metric can
often be found, as in the important example of the KlebanovStrassler solution
(30), which is entirely explicit and everywhere smooth. Having complete met
ric data greatly facilitates the study of probe Dbrane dynamics, at least at the
level of an unstabilized compactiﬁcation. Furthermore, we will now explain how
the eﬀects of moduli stabilization and of the ﬁnite length of the throat can be
incorporated systematically. The method involves examining perturbations to
the supergravity solution that describes the throat in which the D3brane moves.
For concreteness we will work with the example of a KS throat, but the method
is far more general. Our treatment will allow us to give explicit expressions for
the correction terms ∆V in Eq. (18), and hence to extract the characteristics of
inﬂation in the presence of moduli stabilization.
3.3 Supergravity Analysis of the D3brane Potential
3.3.1 Perturbations to the Geometry Above we gave the explicit so
lution for the noncompact warped throat region. We now describe a systematic
way to estimate the leading corrections to the throat solution as perturbations to
the geometry and ﬂuxes at large r (near r
UV
in Figure 1). We then consider the
20
Annu. Rev. Nuc. Part. Sci. 2009 59
eﬀect of these perturbations on the potential for a D3brane at a location well
inside the throat.
Before we can explain the idea underlying this approach, we need a few facts
about the coupling of D3branes to the background ﬁelds of type IIB supergravity,
which is the lowenergy limit of type IIB string theory. Type IIB supergravity
contains a metric, which in the background takes the form of Eq. (14), as well as
various pform ﬁelds. Importantly, the fourdimensional potential V as a function
of the D3brane position in the extra dimensions is only aﬀected by a very speciﬁc
combination of background ﬁelds:
V = T
3
_
e
4A
−α
_
≡ T
3
Φ
−
, (20)
where α is the potential for the ﬁveform ﬁeld
F
5
= (1 +⋆
10
)
_
dα(y) ∧ dx
0
∧ dx
1
∧ dx
2
∧ dx
3
_
, (21)
and ⋆
10
is the tendimensional Hodge star. The D3brane potential is therefore
dictated by the proﬁle of Φ
−
≡ e
4A
− α. Furthermore, Φ
−
vanishes in the un
perturbed KS throat, and more generally in the class of ﬂux compactiﬁcations
considered by Giddings, Kachru, and Polchinski (GKP) (32). Finally, the Ein
stein equations and ﬁveform Bianchi identity imply that perturbations of Φ
−
around such backgrounds satisfy, at the linear level, the sixdimensional Laplace
equation (32, 19),
∇
2
Φ
−
= 0 . (22)
In a general compactiﬁcation, little is known about the solution for Φ
−
, and
one can therefore draw few general conclusions about the D3brane potential.
However, in a noncompact throat geometry, one can often solve the Laplace
equation. Moreover, the potential for a D3brane in a throat that is glued into
21
Annu. Rev. Nuc. Part. Sci. 2009 59
an arbitrary compact CalabiYau must be expressible in terms of a solution for
Φ
−
in the throat, i.e. in terms of a superposition of harmonic functions.
As this is the crucial idea, let us repeat: no matter how complicated the Calabi
Yau to which the throat is attached, it must be possible to express the D3brane
potential via some Φ
−
proﬁle in the throat, as Φ
−
is the only supergravity ﬁeld
that sources a D3brane potential. Moreover, for a suﬃciently long but ﬁnite
throat, the Φ
−
proﬁle is given to an excellent approximation by a solution of the
Laplace equation in the noncompact throat geometry. (Corrections are expected
to be exponentially small when the throat is long.) Thus, the structure of the
inﬂaton potential is dictated by the structure of solutions of the Laplace equation
in the noncompact throat.
3.3.2 Harmonic Analysis Let us therefore solve the Laplace equation (22)
in the KS background. We denote the eigenfunctions of the angular Laplacian
on the base manifold X
5
= T
1,1
by Y
LM
(Ψ) (37), where the multiindices L ≡
(J
1
, J
2
, R), M ≡ (m
1
, m
2
) label SU(2) ×SU(2) ×U(1) quantum numbers under
the corresponding isometries of T
1,1
. We may then express the solution of the
Laplace equation (22) as the following expansion
Φ
−
(r, Ψ) =
L
f
L
(Ψ)
_
r
r
UV
_
∆(L)
, (23)
where
f
L
(Ψ) ≡
M
Φ
LM
Y
LM
(Ψ) +c.c. , (24)
and Φ
LM
are constant coeﬃcients. The quantities ∆(L) are related to the eigen
values of the angular Laplacian
∆(L) ≡ −2 +
_
6[J
1
(J
1
+ 1) +J
2
(J
2
+ 2) −R
2
/8] + 4 . (25)
22
Annu. Rev. Nuc. Part. Sci. 2009 59
To determine the leading perturbations to the brane potential we are interested
in the lowest eigenvalues, since via Eq. (23) these correspond to the perturbations
that diminish most slowly towards the tip of the throat. Incorporating the group
theoretic selection rules that restrict the allowed quantum numbers (37), one ﬁnds
that the smallest eigenvalues corresponding to nontrivial perturbations are
∆ =
3
2
for (J
1
, J
2
, R) = (1/2, 1/2, 1) , (26)
∆ = 2 for (J
1
, J
2
, R) = (1, 0, 0), (0, 1, 0) . (27)
For simplicity, we now assume that a single mode dominates the expansion in
Eq. (23),
Φ
−
≈ f
L
(Ψ)
_
r
r
UV
_
∆(L)
. (28)
(If more than one angular mode is relevant during inﬂation, then the dynamics
is signiﬁcantly more complicated than what is described below.) To isolate the
radial dynamics, we ﬁrst minimize the potential in the angular directions. When
the angular coordinates have relaxed to their minima, the potential reduces to an
eﬀective singleﬁeld potential for the radial direction r. In the singleperturbation
case of Eq. (28), the Φ
−
perturbation then always leads to a repulsive force,
i.e. the eﬀect of the perturbation is to push the brane out of the throat. The
proof is straightforward: any nonconstant spherical harmonic is orthogonal to
the constant (L = 0) harmonic, and hence any nontrivial harmonic necessarily
attains both positive and negative values. Therefore, there always exists an
angular location Ψ
⋆
where f
L
(Ψ
⋆
) is negative. It follows that at ﬁxed radial
location, the D3brane potential induced by the term in Eq. (28) is minimized
at an angular location where the contribution of Eq. (28) to the radial potential
is negative. This contribution to the potential is minimized at r → ∞. The
23
Annu. Rev. Nuc. Part. Sci. 2009 59
potential induced by any individual perturbation of the form of Eq. (28) therefore
produces a radiallyexpulsive force. This is fortunate, since only a repulsive force
allows cancellation with the mass term in Eq. (18) to alleviate the eta problem.
3.3.3 Phenomenological Implications If only one angular mode dom
inates the UV perturbation of the throat geometry, then the radial D3brane
potential is
V (φ) = V
0
(φ) +M
2
pl
H
2
0
_
_
_
φ
M
pl
_
2
−a
∆
_
φ
M
pl
_
∆
_
_
. (29)
where
a
∆
≡ c
∆
_
M
pl
φ
UV
_
∆
, and φ ∝ r . (30)
The magnitudes of the coeﬃcients a
∆
, c
∆
are highly modeldependent and were
estimated in Ref. (19). The above classiﬁcation of the leading perturbations to
the inﬂaton potential via the eigenvalues of the angular Laplacian hence leads to
two cases with distinct phenomenology:
1. Fractional Case
In a general compactiﬁcation, the dominant perturbation corresponds to the
smallest possible eigenvalue, ∆ =
3
2
. By the repulsivity argument we just
gave, this gives a negative contribution to the potential in Eq. (18), ∆V ∝
−φ
3/2
. The dynamics during inﬂation is then governed by the following
phenomenological potential
V (φ) = V
0
(φ) +M
2
pl
H
2
0
_
_
_
φ
M
pl
_
2
−a
3/2
_
φ
M
pl
_
3/2
_
_
. (31)
For a potential of this form, the eta parameter can be ﬁnetuned to be small
locally, near an approximate inﬂection point. This inﬂection point model
is phenomenologically identical to the explicit model of Dbrane inﬂation
24
Annu. Rev. Nuc. Part. Sci. 2009 59
(33, 34, 35, 36) in which a modulistabilizing D7brane stack descends into
the throat region while wrapping a suitable fourcycle.
2. Quadratic Case
Although the ∆ =
3
2
perturbation is generically dominant, it may be for
bidden by a discrete symmetry, i.e. by an unbroken global symmetry of the
full compact manifold (19). In this case, the leading correction comes from
the ∆ = 2 perturbation, ∆V ∝ −φ
2
. The relevant phenomenological model
is then
V (φ) = V
0
(φ) +βH
2
0
φ
2
, (32)
where the parameter β allows a nearly continuous tuning of the inﬂaton
mass. The maximallytuned case β = 0 was ﬁrst analyzed in Appendix D
of Ref. (29) and the phenomenology for general β was discussed in Ref. (38).
(We should note that in the limit β ≪ 1, corrections from perturbations
with ∆ > 2 can be important.) As β →1, the potential becomes steep, but
inﬂation may still occur via the DBI eﬀect (39). The parameter β may even
be negative, pushing the brane out of the throat and allowing a realization
of IR DBI inﬂation (40).
The above summarizes the phenomenology of warped Dbrane inﬂation under
the simplifying assumption that a single angular mode dominates in the Φ
−
perturbation of Eq. (28). In general, more than one Lmode may be important
in Eq. (28). In that case we expect the angular dynamics of the brane to be
signiﬁcantly more complicated, with potentially important consequences for the
eﬀective singleﬁeld potential.
25
Annu. Rev. Nuc. Part. Sci. 2009 59
3.4 Gauge Theory Interpretation
3.4.1 Gauge/Gravity Duality In the previous section we have shown
how to compute, in supergravity, the leading contributions to the inﬂaton poten
tial induced by moduli stabilization and by the coupling of the throat region to
the compact space. We will now present a very instructive dual description of
this analysis.
The celebrated AdS/CFT correspondence (41) is a duality between type IIB
string theory on Antide Sitter (AdS) (or asymptotically, approximately AdS)
spaces and conformal ﬁeld theories (CFTs) on their boundaries. An important
class of dual pairs consists of warped throat geometries and N = 1 supersymmetric
ﬁeld theories. The system of interest to us, a D3brane moving in a warped
throat solution of type IIB supergravity, therefore admits a dual description in
terms of an approximate CFT. The corresponding N = 1 supersymmetric gauge
theory is approximately conformal over a large range of energy scales, and then
eventually conﬁnes in the infrared. The gradual deviations from conformality
manifest themselves on the gravity side as logarithmic corrections to the warp
factor e
2A
, Eq. (16).
On the gravity side of the correspondence, we were interested in nonnormal
izable perturbations of the ﬁeld Φ
−
. In AdS/CFT, nonnormalizable perturba
tions of supergravity ﬁelds correspond to perturbations of the CFT Lagrange
density by irrelevant operators,
∆L = M
4−δ
UV
O
δ
, (33)
where M
UV
is the UV cutoﬀ of the gauge theory and O
δ
is an operator of mass
dimension δ ≥ 4. One advantage of this dual description is that the contribu
26
Annu. Rev. Nuc. Part. Sci. 2009 59
tions to the eta parameter are now manifestly organized in terms of operator
perturbations, precisely as in our general eﬀective ﬁeld theory treatment in §2,
cf. Eq. (3).
We will now outline how the eta problem may be studied on the gauge theory
side of the AdS/CFT correspondence, by classifying irrelevant perturbations to
the gauge theory. For a more complete description we refer the interested reader
to Ref. (19), while readers less interested in these details may skip to §3.5 without
loss of continuity.
3.4.2 Perturbations of the Gauge Theory The conﬁguration space
of a probe D3brane in a KS throat corresponds to a portion of the Coulomb
branch of the dual CFT (i.e. to a portion of the gauge theory moduli space in
which the expectation values of D3brane collective coordinates reduce the rank of
the nonabelian part of the gauge group, but do not reduce the total rank.) Thus,
to understand the potential on this conﬁguration space, we are interested in the
potential on the Coulomb branch of the CFT. Such a potential can be generated
if the CFT Lagrangian is perturbed by operators composed of the scalar ﬁelds
that parameterize the Coulomb branch.
In particular, we are interested in perturbations that do not explicitly break
supersymmetry and that incorporate the eﬀects of bulk CalabiYau ﬁelds. The
leading such terms are of the form of Eq. (33), with
O
δ
≡
_
d
4
θ X
†
X O
∆
, (34)
where X is a bulk moduli ﬁeld. This term, being an integral over superspace, is
allowed in a supersymmetric Lagrangian, but will break supersymmetry sponta
neously if X obtains an Fterm vacuum expectation value. Notice that O
δ
is a
composite operator, containing both bulk and CFT ﬁelds, with total dimension
27
Annu. Rev. Nuc. Part. Sci. 2009 59
δ = 4 + ∆. Such a perturbation yields a term in the D3brane potential of the
form (19)
∆V ∝ φ
∆
. (35)
In the following we will identify the CFT operators O
∆
that correspond to per
turbations of Φ
−
and hence induce a D3brane potential; the operator dimensions
∆ will then dictate the structure of possible terms in the D3brane potential.
3.4.3 Classification of Operators To enumerate the lowestdimension
contributing operators, we must give a few more details of the structure of the
gauge theory. (More background on the gauge theory dual to the KS throat may
be found in Ref. (42).) The approximate CFT that is dual to the KS throat is an
SU(N+M) ×SU(N) gauge theory with bifundamental ﬁelds A
i
, B
i
(i, j = 1, 2).
These ﬁelds parameterize the Coulomb branch and, in particular, contain the data
specifying the D3brane position. The singletrace operators built out of the ﬁelds
A
i
, B
i
and their complex conjugates are labeled by their SU(2)
A
×SU(2)
B
×U(1)
R
quantum numbers (J
1
, J
2
, R); this symmetry group corresponds to the isometries
of the base manifold T
1,1
. Using the AdS/CFT correspondence, the dimensions
of these operators are given by Eq. (25). In fact, the leading contributions to the
D3brane potential involve either chiral operators whose dimensions are dictated
by their U(1)
R
charges, or operators related by supersymmetry to the Noether
currents of the global symmetries, and in either case the dimensions are protected
and could be computed directly in the gauge theory. However, this will not be
true of the operators that induce subleading corrections.
Chiral operators. For J
1
= J
2
= R/2, one has chiral operators of the form
O
∆
= Tr
_
A
(i
1
B
(j
1
A
i
2
B
j
2
. . . A
i
R
)
B
j
R
)
_
+c.c. (36)
28
Annu. Rev. Nuc. Part. Sci. 2009 59
The dimensions of these operators, ∆ = 3R/2, are ﬁxed by N = 1 superconformal
invariance. The lowestdimension such operators are
O
3/2
= Tr (A
i
B
j
) +c.c. , (37)
which have {J
1
, J
2
, R} = {
1
2
,
1
2
, 1}. These chiral operators have ∆ = 3/2, and in
generic situations they contribute the leading term in the inﬂaton potential via
Eq. (35).
Nonchiral operators. There are a number of operators which have the next
lowest dimension, ∆ = 2. For example, there are operators with {J
1
, J
2
, R} =
{1, 0, 0}:
O
2
= Tr
_
A
1
¯
A
2
_
, Tr
_
A
2
¯
A
1
_
,
1
√
2
Tr
_
A
1
¯
A
1
−A
2
¯
A
2
_
, (38)
and the corresponding {J
1
, J
2
, R} = {0, 1, 0} operators made out of the ﬁelds B
j
.
These nonchiral operators are in the same supermultiplets as SU(2) × SU(2)
global symmetry currents, and so their dimension is exactly 2.
Table 1: AdS/CFT Dictionary for Warped Dbrane Inﬂation.
Gravity Side Gauge Theory Side
∆L ∆V = T
3
Φ
−
∇
2
Φ
−
= 0
∆V = −
_
φ
φ
UV
_
∆
∆L =
_
d
4
θ X
†
X O
∆
O
∆
= Tr
_
A
(i
1
B
(j
1
. . . A
i
R
)
B
j
R
)
_
∆V = −
_
φ
φ
UV
_
∆
∆ eigenvalue of Laplacian operator dimension
φ radial location energy scale
φ
UV
maximal UV radius UV cutoﬀ
29
Annu. Rev. Nuc. Part. Sci. 2009 59
In the event that a discrete symmetry preserved by the full string compactiﬁ
cation forbids the chiral operators O
3/2
, the leading contribution to the inﬂaton
potential comes from O
2
. This operator classiﬁcation precisely matches our su
pergravity analysis in §3.3, and the correspondence is summarized in Table 1.
Finally, note that the contributing composite operators O
δ
have dimensions 11/2
and 6, in precisely the range that we argued on general grounds in §2 could yield
orderunity contributions to the eta parameter.
3.5 Summary and Perspective
In §2 we explained how the eta problem is sensitive to dimensionsix Planck
suppressed operators. In eﬀective ﬁeld theory models of inﬂation one can of course
always assume a solution to the eta problem by a cancellation of the contributing
correction terms; in other words, one can postulate that a ﬂat potential V (φ)
arises after an approximate cancellation among dimensionsix Plancksuppressed
corrections. In string theory models of inﬂation, to follow this path would be to
abdicate the opportunity to use Plancksuppressed contributions as a (limited)
window onto string theory. Moreover, once φ is identiﬁed with a physical degree
of freedom of a string compactiﬁcation, the precise form of the potential is in
principle fully speciﬁed by the remaining data of the compactiﬁcation. (Mixing
conjecture into the analysis at this stage would eﬀectively transform a ‘string
derived’ scenario into a ‘stringinspired’ scenario; the latter may be interesting
as a cosmological model, but will not contribute to our understanding of string
theory.) Thus, overcoming the eta problem becomes a detailed computational
question. One can in principle compute the full potential from ﬁrst principles,
and in practice one can often classify corrections to the leadingorder potential.
30
Annu. Rev. Nuc. Part. Sci. 2009 59
In this section, we have enumerated the leading corrections for warped Dbrane
inﬂation and showed that an accidental cancellation (or ﬁnetuning) allows small
eta over a limited range of inﬂaton values. This gives a nontrivial existence proof
for inﬂationary solutions in warped throat models with D3branes.
4 LargeField Inﬂation
The UV sensitivity of inﬂation described in §2 is vastly increased in the special
case of largeﬁeld models, i.e. scenarios in which the inﬂaton traverses a distance
in ﬁeld space larger than the Planck mass. This class is particularly interesting
because it includes every inﬂationary model that yields a detectablylarge pri
mordial gravitational wave signal (43), as we now review.
4.1 The Lyth Bound
In singleﬁeld slowroll inﬂation the power spectrum of tensor ﬂuctuations is
P
t
=
2
π
2
_
H
M
pl
_
2
, (39)
where H is the Hubble expansion rate. During inﬂation H is approximately
constant, the background spacetime is nearly de Sitter and quantum ﬂuctuations
in any light ﬁeld such as the metric scale with H. The power spectrum of scalar
ﬂuctuations is
P
s
=
_
H
2π
_
2
_
H
˙
φ
_
2
. (40)
The ﬁrst factor in Eq. (40) represents the power spectrum of the inﬂaton ﬂuctu
ations (arising from quantum ﬂuctuations in de Sitter space), while the second
factor comes from the conversion of ﬂuctuations of the inﬂaton into ﬂuctuations
of the spatial 3curvature. On scales smaller than the physical horizon, spatial
31
Annu. Rev. Nuc. Part. Sci. 2009 59
curvature ﬂuctuations relate to the observed ﬂuctuations in the matter density
and in the CMB temperature. The ratio between the tensor and scalar ﬂuctuation
amplitudes is
r ≡
P
t
P
s
= 8
_
˙
φ
HM
pl
_
2
. (41)
If we deﬁne dN ≡ Hdt, then we may use Eq. (41) to write the ﬁeld variation
between the end of inﬂation, N
end
, and the time when ﬂuctuations on CMB scales
were generated, N
cmb
, as the following integral
∆φ
M
pl
=
_
N
end
N
cmb
dN
_
r(N)
8
_
1/2
. (42)
Since the tensortoscalar ratio r(N) is nearly constant during slowroll inﬂation,
one can derive the following important relation, originally due to Lyth (43):
∆φ
M
pl
≃ O(1)
_
r
⋆
0.01
_
1/2
, (43)
where r
⋆
is the value of the tensortoscalar ratio on CMB scales, r
⋆
≡ r(N
cmb
).
In any model with r
⋆
> 0.01 one must therefore ensure that ǫ, η ≪ 1 over a
superPlanckian range ∆φ > M
pl
.
This result implies two necessary conditions for largeﬁeld inﬂation:
i) an obvious requirement is that large ﬁeld ranges are kinematically allowed,
i.e. that the scalar ﬁeld space (in canonical units) has diameter > M
pl
.
This is nontrivial, as in typical string compactiﬁcations many ﬁelds are not
permitted such large excursions. (D3brane inﬂation in warped throats,
§3, is one class of examples where the kinematic requirement for large ﬁeld
ranges cannot be fulﬁlled (44).)
ii) the ﬂatness of the inﬂaton potential needs to be controlled dynamically over
a superPlanckian ﬁeld range. We discuss this challenge in eﬀective ﬁeld
theory in §4.2 and in string theory in §5.
32
Annu. Rev. Nuc. Part. Sci. 2009 59
4.2 SuperPlanckian Fields and Flat Potentials
To begin, let us consider superPlanckian ﬁeld excursions in the context of Wilso
nian eﬀective ﬁeld theory.
4.2.1 No Shift Symmetry In the absence of any special symmetries, the
potential in largeﬁeld inﬂation becomes sensitive to an inﬁnite series of Planck
suppressed operators. The physical interpretation of these terms is as follows: as
the inﬂaton expectation value changes, any other ﬁelds χ to which the inﬂaton
couples experience changes in mass, selfcoupling, etc. In particular, any ﬁeld
coupled with at least gravitational strength to the inﬂaton experiences signiﬁ
cant changes when the inﬂaton undergoes a superPlanckian excursion. These
variations of the χ masses and couplings in turn feed back into changes of the
inﬂaton potential and therefore threaten to spoil the delicate ﬂatness required
for inﬂation. Note that this applies not just to the light (m ≪ H) degrees of
freedom, but even to ﬁelds with masses near the Planck scale: integrating out
Planckscale degrees of freedom generically (i.e., for couplings of order unity)
introduces Plancksuppressed operators in the eﬀective action. For nearly all
questions in particle physics, such operators are negligible, but in inﬂation they
play an important role.
The particular operators which appear are determined, as always, by the sym
metries of the lowenergy action. As an example, imposing only the symmetry
φ →−φ on the inﬂaton leads to the following eﬀective action:
L
eﬀ
(φ) = −
1
2
(∂φ)
2
−
1
2
m
2
φ
2
−
1
4
λφ
4
−
∞
p=1
_
λ
p
φ
4
+ν
p
(∂φ)
2
_
_
φ
M
pl
_
2p
+. . . (44)
Unless the UV theory enjoys further symmetries, one expects that the coeﬃcients
λ
p
and ν
p
are of order unity. Thus, whenever φ traverses a distance of order
33
Annu. Rev. Nuc. Part. Sci. 2009 59
M
pl
in a direction that is not protected by a suitably powerful symmetry, the
eﬀective Lagrangian receives substantial corrections from an inﬁnite series of
higherdimension operators. In order to have inﬂation, the potential should of
course be approximately ﬂat over a superPlanckian range. If this is to arise
by accident or by ﬁnetuning, it requires a conspiracy among inﬁnitely many
coeﬃcients, which has been termed ‘functional ﬁnetuning’ (compare this to the
eta problem, §2.2, which only requires tuning of one mass parameter).
4.2.2 Shift Symmetry There is a sensible way to control this inﬁnite series
of corrections: one can invoke an approximate symmetry that forbids the inﬂaton
from coupling to other ﬁelds in any way that would spoil the structure of the
inﬂaton potential. Such a shift symmetry,
φ →φ +const (45)
protects the inﬂaton potential in a natural way. (Proposals using shift symmetries
to protect the potential in string inﬂation include (45, 46, 21).)
In the case with a shift symmetry, the action of chaotic inﬂation (47)
L
eﬀ
(φ) = −
1
2
(∂φ)
2
−λ
p
φ
p
, (46)
with small coeﬃcient λ
p
is technically natural. However, because we require
that this symmetry protects the inﬂaton even from couplings to Planckscale
degrees of freedom, it is essential that the symmetry should be approximately
respected by the Planckscale theory – in other words, the proposed symmetry of
the lowenergy eﬀective action should admit a UVcompletion. Hence, largeﬁeld
inﬂation should be formulated in a theory that has access to information about
approximate symmetries at the Planck scale. Let us remark that in eﬀective ﬁeld
theory in general, UVcompletion of an assumed lowenergy symmetry is rarely
34
Annu. Rev. Nuc. Part. Sci. 2009 59
an urgent question. The present situation is diﬀerent because we do not know
whether all reasonable eﬀective actions can in fact arise as lowenergy limits of
string theory, and indeed it has been conjectured that many eﬀective theories do
not admit UVcompletion in string theory (48, 49, 50). Therefore, it is important
to verify that any proposed symmetry of Planckscale physics can be realized in
string theory.
To construct an inﬂationary model with detectable gravitational waves, we are
therefore interested in ﬁnding, in string theory, a conﬁguration that has both
a large kinematic range, and a potential protected by a shift symmetry that is
approximately preserved by the full string theory.
5 Case Study of LargeField Inﬂation: Axion Monodromy
We now turn to our second case study, an example of largeﬁeld inﬂation in string
theory. As we have just discussed, the particular challenge in these models is the
need to control an inﬁnite series of contributions to the inﬂaton potential, arising
from couplings of the inﬂaton to degrees of freedom with masses near the Planck
scale. Direct enumeration and ﬁnetuning of such terms (as in the smallﬁeld
example in §3) is manifestly impractical, and it appears essential to develop a
symmetry argument controlling or forbidding these terms.
An inﬂuential proposal in this direction is Natural Inﬂation (51), in which a
pseudoNambuGoldstone boson (i.e., an axion) is the inﬂaton. At the perturba
tive level, the axion a enjoys a continuous shift symmetry a → a + const which
is broken by nonperturbative eﬀects to a discrete symmetry a → a + 2π. The
nonperturbative eﬀects generate a periodic potential
V (φ) =
Λ
4
2
_
1 −cos
_
φ
f
__
+. . . , (47)
35
Annu. Rev. Nuc. Part. Sci. 2009 59
where Λ is a dynamicallygenerated scale, f is known as the axion decay constant,
φ ≡ af, and the omitted terms are higher harmonics. The reader can easily verify
from Eq. (2) that if the omitted terms are negligible and f > M
pl
, this potential
can drive prolonged inﬂation.
As explained above, an important question, in any proposed eﬀective theory in
which a superPlanckian ﬁeld range is protected by a shift symmetry, is whether
this structure can be UVcompleted. We should therefore search in string theory
for an axion with decay constant f > M
pl
.
5.1 Axions in String Theory
5.1.1 Axions from pForms Axions are plentiful in string compactiﬁca
tions, arising from pform gauge potentials integrated on pcycles of the compact
space. For example, in type IIB string theory, there are axions b
i
= 2π
_
Σ
i
B aris
ing from integrating the NeveuSchwarz (NS) twoform B over twocycles Σ
i
, as
well as axions c
i
= 2π
_
Σ
i
C arising from the RamondRamond (RR) twoform C.
In the absence of additional ingredients such as ﬂuxes and spaceﬁlling wrapped
branes, the potential for these axions is classically ﬂat and has a continuous shift
symmetry which originates in the gauge invariance of the tendimensional ac
tion. Instanton eﬀects break this symmetry to a discrete subgroup, b
i
→b
i
+ 2π
(c
i
→c
i
+2π). This leads to a periodic contribution to the axion potential whose
periodicity we will now estimate. We will ﬁnd that the axion decay constants are
smaller than M
pl
in known, computable limits of string theory (52, 53). Readers
less familiar with string compactiﬁcations can accept this assertion and skip to
§5.2 without loss of continuity.
36
Annu. Rev. Nuc. Part. Sci. 2009 59
5.1.2 Axion Decay Constants in String Theory Let ω
i
be a basis for
H
2
(X, Z), the space of twoforms on the compact space X, with
_
Σ
i
ω
j
= α
′
δ
j
i
.
The NS twoform potential B may be expanded as
B =
1
2π
i
b
i
(x) ω
i
, (48)
with x the fourdimensional spacetime coordinate. The axion decay constant can
be inferred from the normalization of the axion kinetic term, which in this case
descends from the tendimensional term
1
(2π)
7
g
2
s
α
′4
_
d
10
x
1
2
dB
2
⊃
1
2
_
d
4
x
√
−g γ
ij
(∂
µ
b
i
∂
µ
b
j
) , (49)
where
γ
ij
≡
1
6(2π)
9
g
2
s
α
′4
_
X
ω
i
∧ ⋆ ω
j
(50)
and ⋆ is the sixdimensional Hodge star. By performing the integral over the
internal space X and diagonalizing the ﬁeld space metric as γ
ij
→f
2
i
δ
ij
, one can
extract the axion decay constant f
i
.
It is too early to draw universal conclusions, but a body of evidence suggests
that the resulting axion periodicities are always smaller than M
pl
in computable
limits of string theory (52, 53). As this will be essential for our arguments, we
will illustrate this result in a simple example. Suppose that the compactiﬁcation
is isotropic, with typical lengthscale L and volume L
6
. Then using
α
′
M
2
pl
=
2
(2π)
7
L
6
g
2
s
α
′3
(51)
we ﬁnd from Eq. (50) that
f
2
≈ M
2
pl
α
′2
6(2π)
2
L
4
. (52)
In controlled compactiﬁcations we require L ≫
√
α
′
, so that f ≪ M
pl
. Qualita
tively similar conclusions apply in much more general conﬁgurations (52, 53).
37
Annu. Rev. Nuc. Part. Sci. 2009 59
5.2 Axion Inﬂation in String Theory
The above result would seem to imply that Natural Inﬂation from a single axion
ﬁeld cannot be realized in known string compactiﬁcations: string theory provides
many axions, but none of these has a suﬃciently large ﬁeld range. However, there
are at least two reasonable proposals to circumvent this obstacle.
5.2.1 Nflation The ﬁrst suggestion was that a collective excitation of
many hundreds of axions could have an eﬀective ﬁeld range large enough for in
ﬂation (46,54). This ‘Nﬂation’ proposal is a speciﬁc example of assisted inﬂation
(55), but, importantly, one in which symmetry helps to protect the axion potential
from corrections that would impede inﬂation. Although promising, this scenario
still awaits a proof of principle demonstration, as the presence of a large number
of light ﬁelds leads to a problematic renormalization of the Newton constant, and
hence to an eﬀectively reduced ﬁeld range. For recent studies of Nﬂation see
(56, 57).
5.2.2 Axion Monodromy We will instead describe an elementary mecha
nism, monodromy, which allows inﬂation to persist through multiple circuits of a
single periodic axion ﬁeld. A system is said to undergo monodromy if, upon trans
port around a closed loop in the (naive) conﬁguration space, the system reaches
a new conﬁguration. A spiral staircase is a canonical example: the naive conﬁgu
ration space is described by the angular coordinate, but the system changes upon
transport by 2π. (In fact, we will ﬁnd that this simple model gives an excellent
description of the potential in axion monodromy inﬂation.) The idea of using
monodromy to achieve controlled largeﬁeld inﬂation in string theory was ﬁrst
proposed by Silverstein and Westphal (20), who discussed a model involving a
D4brane wound inside a nilmanifold. In this section we will focus instead on the
38
Annu. Rev. Nuc. Part. Sci. 2009 59
subsequent axion monodromy proposal of Ref. (21), where a monodromy arises
in the fourdimensional potential energy upon transport around a circle in the
ﬁeld space parameterized by an axion.
Monodromies of this sort are possible in a variety of compactiﬁcations, but
we will focus on a single concrete example. Consider type IIB string theory on
a CalabiYau orientifold, i.e. a quotient of a CalabiYau manifold by a discrete
symmetry that includes worldsheet orientation reversal and a geometric involu
tion. Speciﬁcally, we will suppose that the involution has ﬁxed points and ﬁxed
fourcycles, known as O3planes and O7planes, respectively. If in addition the
compactiﬁcation includes a D5brane that wraps a suitable twocycle Σ and ﬁlls
spacetime, then the axion b = 2π
_
Σ
B can exhibit monodromy in the potential
energy. (Similarly, a wrapped NS5brane produces monodromy for the axion
c = 2π
_
Σ
C.) In other words, a D5brane wrapping Σ carries a potential energy
that is not a periodic function of the axion, as the shift symmetry of the axion
action is broken by the presence of the wrapped brane; in fact, the potential
energy increases without bound as b increases.
In the D5brane case, the relevant potential comes from the DiracBornInfeld
action for the wrapped Dbrane,
S
DBI
=
1
(2π)
5
g
s
α
′3
_
M
4
×Σ
d
6
ξ
_
det(G+B) (53)
=
1
(2π)
6
g
s
α
′2
_
M
4
d
4
x
√
−g
_
(2π)
2
ℓ
4
Σ
+b
2
, (54)
where ℓ
Σ
is the size of the twocycle Σ in string units. The brane energy, Eq. (54),
is clearly not invariant under the shift symmetry b → b + 2π, although this is a
symmetry of the corresponding compactiﬁcation without the wrapped D5brane.
Thus, the DBI action leads directly to monodromy for b. Moreover, when b ≫ℓ
2
Σ
,
the potential is asymptotically linear in the canonicallynormalized ﬁeld ϕ
b
∝ b.
39
Annu. Rev. Nuc. Part. Sci. 2009 59
Before we give more details of the eﬀects of compactiﬁcation on the axion
potential, let us qualitatively summarize the inﬂationary dynamics in this model.
One begins with a D5brane wrapping a curve Σ, upon which
_
Σ
B is taken to
be large. In other words, the axion b has a large initial vev. Inﬂation proceeds
by the reduction of this vev, until ﬁnally
_
Σ
B = 0 and the D5brane is nearly
‘empty’, i.e. has little worldvolume ﬂux. During this process the D5brane does
not move, nor do any of the closedstring moduli shift appreciably. For small
axion vevs, the asymptotically linear potential we have described is inaccurate,
and the curvature of the potential becomes nonnegligible; see Eq. (54). At this
stage, the axion begins to oscillate around its origin. Couplings between the axion
and other degrees of freedom, either closed string modes or open string modes,
drain energy from the inﬂaton oscillations. If a suﬃcient fraction of this energy
is eventually transmitted to visiblesector degrees of freedom – which may reside,
for example, on a stack of Dbranes elsewhere in the compactiﬁcation – then the
hot Big Bang begins. The details of reheating depend strongly on the form of
the couplings between the Standard Model degrees of freedom and the inﬂaton,
and this is an important open question, both in this model and in string inﬂation
more generally. (For representative work on reheating after string inﬂation, see
(58).)
5.3 Compactiﬁcation Considerations
Having explained the essential idea of axion monodromy inﬂation, we must still
ensure that the proposed inﬂationary mechanism is compatible with moduli sta
bilization and can be realized in a consistent compactiﬁcation. An immediate
concern is whether there are additional contributions to the potential, beyond
40
Annu. Rev. Nuc. Part. Sci. 2009 59
the linear term identiﬁed above, that could have important eﬀects during inﬂa
tion. As we have emphasized throughout this review, one expects that in the
absence of a symmetry protecting the inﬂaton potential, generic corrections due
to moduli stabilization will contribute ∆η ∼ O(1). It is therefore essential to
verify that the continuous shift symmetry which protects the inﬂaton potential
is preserved to an appropriate degree by the stabilized compactiﬁcation. For the
special case of moduli stabilization in which nonperturbative eﬀects play a role,
ensuring that the shift symmetry is not spoiled can be quite subtle. We will now
explain this point, but readers less interested in the details can skip to §5.4.
5.3.1 Axion Shift Symmetries in String Theory We ﬁrst observe that
a continuous shift symmetry b →b +const forbids all nonderivative terms in the
eﬀective action for b, but does not constrain terms involving only the spacetime
derivative ∂
µ
b. Therefore, the shift symmetry is unbroken to the extent that all
nonderivative terms are constrained to vanish.
We now check this criterion in the example of interest by recalling the classic
DineSeiberg treatment (59) of axion shift symmetries in string theory. Dine
and Seiberg proved that to any order in perturbation theory (in the absence of
Dbranes), the eﬀective action for the axion b can only be a function of ∂
µ
b,
i.e. b has a shift symmetry. To show this, they observed that the zeromomentum
coupling of b (corresponding to nonderivative terms) is a total derivative on the
worldsheet, and hence vanishes when the worldsheet has no boundary and wraps
a topologically trivial cycle in spacetime.
Their argument proceeds as follows. The twoform B couples to the worldsheet
as (60)
i
2πα
′
_
Σ
d
2
ξ ǫ
αβ
∂
α
X
µ
∂
β
X
ν
B
µν
(X) , (55)
41
Annu. Rev. Nuc. Part. Sci. 2009 59
where X
µ
are the spacetime coordinates and ξ
α
are twodimensional string world
sheet coordinates; i.e. Eq. (55) is the pullback of B onto the worldsheet. If B
is imagined to be a constant in spacetime, then the above coupling is a total
derivative on the worldsheet. Equivalently, the zeromomentum portion of the
axion eﬀective action in spacetime arises from a totalderivative term on the
string worldsheet. (In general backgrounds, B is not constant, but it is the cou
pling of the constant portion of B that governs zeromomentum terms in the
fourdimensional eﬀective action.)
Hence, if the string worldsheet has no boundary and is topologically trivial,
the zeromomentum coupling of the axion b must vanish, and the axion therefore
cannot have any nonderivative couplings. Thus, as long as the worldsheet has no
boundary, the axion has no nonderivative couplings to any order in sigmamodel
perturbation theory (i.e., the perturbation theory of the quantum ﬁeld theory
living on the string worldsheet, whose coupling constant is the inverse space
time curvature in units of α
′
), because worldsheets wrapping nontrivial curves in
the tendimensional spacetime contribute only nonperturbatively, as worldsheet
instantons.
However, closed string worldsheets can develop boundaries in the presence of
Dbranes, on which the strings can break and end. Therefore, in a compacti
ﬁcation without Dbranes, the shift symmetry of b is preserved to all orders in
perturbation theory, while in a compactiﬁcation containing Dbranes, the shift
symmetry can be violated.
5.3.2 The Eta Problem for b In the present setting, we have deliberately
invoked D5branes, in order to produce a monodromy in the potential. However,
provided that this potential, which we identiﬁed with the inﬂaton potential, is the
42
Annu. Rev. Nuc. Part. Sci. 2009 59
leading eﬀect breaking the shift symmetry, the resulting structure is technically
natural.
Although this sounds promising, in the case of the b axion there is in fact an
additional ingredient which also breaks the axion shift symmetry. K¨ ahler moduli
stabilization is accomplished, in a wellstudied class of models (27, 28), by the
inclusion of nonperturbative eﬀects, e.g. from Euclidean D3branes (D3brane
instantons). Such eﬀects can circumvent the DineSeiberg argument given above,
because Euclidean Dbranes are nonperturbative eﬀects and provide boundaries
for string worldsheets.
We will now sketch the speciﬁc diﬃculty presented by Euclidean D3branes,
referring the interested reader to Ref. (21) for details. Supposing for simplic
ity that there is only one K¨ ahler modulus, T, the superpotential is of the form
W = W
0
+Aexp(−2π T), where the exponential term is the Euclidean D3brane
contribution. At the energy scales in question, W
0
and A are constants depending
on the stabilized values of the complex structure moduli and of the dilaton. Fur
thermore, the K¨ ahler potential takes the form (61) K = −3M
2
pl
ln(T +
¯
T −d b
2
),
with d a constant depending on the intersection numbers of the compactiﬁca
tion and on the stabilized value of the dilaton. Although a shift of b can be
compensated in the K¨ ahler potential by a shift of T +
¯
T, the superpotential is
then not invariant. Clearly, the continuous shift symmetry is broken by the non
perturbative superpotential term generated by Euclidean D3branes. Euclidean
D3branes therefore make important contributions to the potential of b, and in
fact generate an eta problem. One can easily verify (21) that this is precisely
analogous to the eta problem in D3brane inﬂation.
43
Annu. Rev. Nuc. Part. Sci. 2009 59
5.3.3 Flat Potential for c The situation may seem discouraging, be
cause even a shift symmetry that was valid to all orders in perturbation theory
has turned out to be inadequate to protect the inﬂaton potential! However, we
will now ﬁnd an even more potent symmetry in the case of the c axion.
Although the NS axion b and the RR axion c have many shared features, a
crucial distinction is that c couples to D1branes, via the electric coupling
_
Σ
C , (56)
but does not couple directly to D3branes (or to D3brane instantons) that carry
vanishing D1brane charge. Thus, if the moduli are stabilized exclusively by
instantons to which c does not couple, even nonperturbative moduli stabilization
will not violate the shift symmetry of c. We refer to Ref. (21) for a description
of compactiﬁcations in which this mechanism is operative.
These considerations suggest the following scenario. Instead of a wrapped D5
brane introducing a potential for a b axion, we consider a wrapped NS5brane
that provides a potential for a c axion. Even in the presence of nonperturbative
stabilization of the K¨ ahler moduli, such an axion can enjoy the protection of a
shift symmetry over a superPlanckian range. The corresponding inﬂationary
scenario is natural in the technical sense.
5.4 Summary and Perspective
In §4 we showed that an observable gravitational wave signal correlates with the
inﬂaton ﬁeld moving over a superPlanckian distance during inﬂation. Eﬀective
ﬁeld theory models of largeﬁeld inﬂation then require a shift symmetry to protect
the ﬂatness of the potential over a superPlanckian range. It has therefore become
an important question whether such shift symmetries arise in string theory and
44
Annu. Rev. Nuc. Part. Sci. 2009 59
can be used to realize largeﬁeld inﬂation.
In this section, we argued that the ﬁrst examples of shift symmetries in string
theory that protect the potential over a superPlanckian range are becoming
available. We explained the dual role of the monodromy: i) it results in a large
kinematic ﬁeld range ∆φ > M
pl
by allowing a small fundamental domain to be
traversed repeatedly, and ii) in combination with the shift symmetry it controls
corrections to the potential over a superPlanckian range. The shift symmetry,
only weakly broken by V , controls corrections ∆V within a fundamental domain,
and furthermore relates corrections in one fundamental domain to those in any
other. Monodromy therefore eﬀectively reduces a largeﬁeld problem to a small
ﬁeld problem (20).
Although more work is required to understand these models and the compact
iﬁcations in which they arise, monodromy appears to be a robust and rather
promising mechanism for realizing largeﬁeld inﬂation, and hence an observable
gravitational wave signal, in string theory.
6 Outlook
6.1 Theoretical Prospects
As we hope this review has illustrated, theoretical progress in recent years has
been dramatic. A decade ago, only a few proposals for connecting string theory to
cosmology were available, and the problem of stabilizing the moduli had not been
addressed. We now have a wide array of inﬂationary models motivated by string
theory, and the beststudied examples among these incorporate some information
about moduli stabilization. Moreover, a few mechanisms for inﬂation in string
theory have been shown to be robust, persisting after full moduli stabilization
45
Annu. Rev. Nuc. Part. Sci. 2009 59
with all relevant corrections included.
Aside from demonstrating that inﬂation is possible in string theory, what has
been accomplished? In our view the primary use of explicit models of inﬂation
in string theory is as test cases, or toy models, for the sensitivity of inﬂation
to quantum gravity. On the theoretical front, these models have underlined the
importance of the eta problem in general ﬁeld theory realizations of inﬂation;
they have led to mechanisms for inﬂation that might seem unnatural in ﬁeld
theory, but are apparently natural in string theory; and they have sharpened our
understanding of the implications of a detection of primordial tensor modes.
It is of course diﬃcult to predict the direction of future theoretical progress,
not least because unforeseen fundamental advances in string theory can be ex
pected to enlarge the toolkit of inﬂationary modelbuilders. However, it is safe to
anticipate further gradual progress in moduli stabilization, including the appear
ance of additional explicit examples with all moduli stabilized; entirely explicit
models of inﬂation in such compactiﬁcations will undoubtedly follow. At present,
few successful models exist in Mtheory or in heterotic string theory (however,
see e.g. (62) et seq.), and under mild assumptions, inﬂation can be shown to be
impossible in certain classes of type IIA compactiﬁcations (63, 64, 65). It would
be surprising if it turned out that inﬂation is much more natural in one weakly
coupled limit of string theory than in the rest, and the present disparity can be
attributed in part to the diﬀerences among the modulistabilizing tools presently
available in the various limits. Clearly, it would be useful to understand how
inﬂation can arise in more diverse string vacua.
The inﬂationary models now available in string theory are subject to stringent
theoretical constraints arising from consistency requirements (e.g., tadpole can
46
Annu. Rev. Nuc. Part. Sci. 2009 59
cellation) and from the need for some degree of computability. In turn, these
limitations lead to correlations among the cosmological observables, i.e. to pre
dictions. Some of these constraints will undoubtedly disappear as we learn to
explore more general string compactiﬁcations. However, one can hope that some
constraints may remain, so that the set of inﬂationary eﬀective actions derived
from string theory would be a proper subset of the set of inﬂationary eﬀective
actions in a general quantum ﬁeld theory. Establishing such a proposition would
require a far more comprehensive understanding of string compactiﬁcations than
is available at present.
6.2 Observational Prospects
The theoretical aspects of inﬂation described in this review are interesting largely
because they can be tested experimentally using present and future cosmological
data. In order to describe this connection, we will very brieﬂy review recent
achievements and nearfuture prospects in observational cosmology.
6.2.1 Present and Future Observations Observations of the cosmic
microwave background anisotropies, of the distribution of galaxies on the sky,
and of the redshiftluminosity relations of type Ia supernovae have transformed
cosmology into an exact science. This has revealed a strange universe ﬁlled
with 73% dark energy, 23% dark matter, and only 4% baryons. In addition,
we now have a ﬁrm qualitative understanding of the formation of largescale
structures, like the cosmic web of galaxies, through the gravitational instability
of small primordial ﬂuctuations. The perturbation spectrum that forms the seeds
of these structures is found to be nearly (but not exactly) scaleinvariant, nearly
Gaussian, and adiabatic (10), precisely as predicted by the simplest models of
47
Annu. Rev. Nuc. Part. Sci. 2009 59
inﬂation (11, 12).
Future observations will dramatically extend our knowledge of the primordial
ﬂuctuations, probe further details of the inﬂationary paradigm, and allow us to
constrain or exclude a considerable fraction of the proposed scenarios for inﬂa
tion. The Planck satellite will measure the temperature anisotropies of the CMB
with unprecedented accuracy over a large range of scales; in combination with
smallscale CMB experiments (e.g. ACT (66) and SPT (67)) this will provide
crucial information on deviations of the scalar spectrum from scaleinvariance,
Gaussianity and adiabaticity.
CMB polarization experiments from the ground (e.g. Clover (68), QUIET (69),
and BICEP (70)) and from balloons (e.g. EBEX (71) and SPIDER (72)) promise
to provide the ﬁrst signiﬁcant constraints on inﬂationary tensor perturbations. A
planned CMB polarization satellite (CMBPol (12,18,73)) would be designed to be
sensitive enough to detect Bmodes down to a tensortoscalar ratio of r = 0.01,
thereby including all models of largeﬁeld inﬂation (∆φ > M
pl
).
6.2.2 UV Physics in the Sky? The most dramatic conﬁrmation of inﬂa
tion would come from a detection of Bmode polarization, which would establish
the energy scale of inﬂation and would indicate that the inﬂaton traversed a
superPlanckian distance. As we have argued in this review, superPlanckian dis
placements are a key instance in which the inﬂaton eﬀective action is particularly
sensitive to the physics of the Planck scale. As a concrete example of the discrim
inatory power of tensor perturbations, any detection of primordial gravitational
waves would exclude the warped D3brane inﬂation scenario of §3 (44), while an
upper bound r < 0.07 (or a detection with r ≫ 0.07) would exclude the axion
monodromy scenario of §5 (21).
48
Annu. Rev. Nuc. Part. Sci. 2009 59
A further opportunity arises because singleﬁeld slowroll inﬂation predicts null
results for many cosmological observables, as the primordial scalar ﬂuctuations
are predicted to be scaleinvariant, Gaussian and adiabatic to a high degree. A de
tection of nonGaussianity, isocurvature ﬂuctuations or a large scaledependence
(running) would therefore rule out singleﬁeld slowroll inﬂation. Inﬂationary ef
fective actions that do allow for a signiﬁcant nonGaussianity, nonadiabaticity or
scaledependence often require higherderivative interactions and/or more than
one light ﬁeld, and such actions arise rather naturally in string theory. Although
we have focused in this review on the sensitivity of the inﬂaton potential to
Planckscale physics, the inﬂaton kinetic term is equally UVsensitive, and string
theory provides a promising framework for understanding the higherderivative
interactions that can produce signiﬁcant nonGaussianity (39, 74).
Finally, CMB temperature and polarization anisotropies induced by relic cos
mic strings or other topological defects provide probes of the physics of the end
of inﬂation or of the postinﬂationary era. Cosmic strings are automatically pro
duced at the end of braneantibrane inﬂation (75, 76), and the stability and phe
nomenological properties of the resulting cosmic string network are determined
by the properties of the warped geometry. Detecting cosmic superstrings via
lensing or through their characteristic bursts of gravitational waves is an exciting
prospect.
6.3 Conclusions
Recent work by many authors has led to the emergence of robust mechanisms
for inﬂation in string theory. The primary motivations for these works are the
sensitivity of inﬂationary eﬀective actions to the ultraviolet completion of gravity,
49
Annu. Rev. Nuc. Part. Sci. 2009 59
and the prospect of empirical tests using precision cosmological data. String
theory models of inﬂation have now achieved a reasonable level of theoretical
control and are genuinely falsiﬁable by observational data. Indeed, many string
inﬂation models are already signiﬁcantly constrained by the current data (77). A
more diﬃcult question is how cosmological observations might possibly provide
evidence in favor of a string theory model of inﬂation. Present observations and
present theoretical considerations do not oblige us to expect an eventual positive
result. However, if we are fortunate enough to detect evidence for string theory
in the sky, this will most plausibly arise through a distinctive signature that is
unnatural, albeit presumably possible, in ﬁeld theory models. Perhaps the best
hope would be a striking correlation of many observables.
The theoretical community will eagerly await the coming generation of exper
imental results (12), in the hope of extracting further clues about the physical
properties of the early universe.
Acknowledgements.
We are grateful to Shamit Kachru, Igor Klebanov, Andrei Linde, Enrico Pajer,
Hiranya Peiris, David Poland, Jesse Thaler, and Eva Silverstein for useful discus
sions and comments on the draft. LM also thanks K. Narayan, Angel Uranga,
and Marco Zagermann for discussions and correspondence on related topics.
DB thanks the theory group at Cornell and Trident Caf´e for their hospitality
while some of the work on this review was performed. LM thanks the theory
groups at Harvard and IMSc, Chennai, as well as the organizers of the 2008 Indian
Strings Meeting, for their hospitality while this review was being prepared.
The research of DB is supported by the David and Lucile Packard Foundation
50
Annu. Rev. Nuc. Part. Sci. 2009 59
and the Alfred P. Sloan Foundation. The research of LM is supported by NSF
grant PHY0355005.
Literature Cited
1. A. H. Guth, Phys. Rev. D 23, 347 (1981).
2. A. D. Linde, Phys. Lett. B 108, 389 (1982).
3. A. Albrecht and P. J. Steinhardt, Phys. Rev. Lett. 48, 1220 (1982).
4. V. F. Mukhanov and G. V. Chibisov, JETP Lett. 33, 532 (1981).
5. V. F. Mukhanov, JETP Lett. 41, 493 (1985).
6. S. W. Hawking, Phys. Lett. B 115, 295 (1982).
7. A. A. Starobinsky, Phys. Lett. B 117, 175 (1982).
8. A. H. Guth and S. Y. Pi, Phys. Rev. Lett. 49, 1110 (1982).
9. J. M. Bardeen, P. J. Steinhardt and M. S. Turner, Phys. Rev. D 28, 679
(1983).
10. E. Komatsu et al. [WMAP Collaboration], arXiv:0803.0547 [astroph].
11. D. H. Lyth and A. Riotto, Phys. Rept. 314, 1 (1999) [arXiv:hepph/9807278].
12. D. Baumann et al., arXiv:0811.3919 [astroph].
13. D. Baumann, “TASI Lectures on Inﬂation,” in preparation.
14. M. Zaldarriaga and U. Seljak, Phys. Rev. D 55, 1830 (1997)
[arXiv:astroph/9609170].
15. M. Kamionkowski, A. Kosowsky and A. Stebbins, Phys. Rev. D 55, 7368
(1997) [arXiv:astroph/9611125].
16. J. Kovac, E. M. Leitch, C. Pryke, J. E. Carlstrom, N. W. Halverson and
W. L. Holzapfel, Nature 420, 772 (2002) [arXiv:astroph/0209478].
17. A. Kogut et al. [WMAP Collaboration], Astrophys. J. Suppl. 148, 161 (2003)
51
Annu. Rev. Nuc. Part. Sci. 2009 59
[arXiv:astroph/0302213].
18. D. Baumann et al., arXiv:0811.3911 [astroph].
19. D. Baumann, A. Dymarsky, S. Kachru, I. R. Klebanov and L. McAllister,
arXiv:0808.2811 [hepth].
20. E. Silverstein and A. Westphal, arXiv:0803.3085 [hepth].
21. L. McAllister, E. Silverstein and A. Westphal, arXiv:0808.0706 [hepth].
22. P. Binetruy and M. K. Gaillard, Phys. Rev. D 34, 3069 (1986).
T. Banks, M. Berkooz, S. H. Shenker, G. W. Moore and P. J. Steinhardt,
Phys. Rev. D 52, 3548 (1995) [arXiv:hepth/9503114].
G. R. Dvali and S. H. H. Tye, Phys. Lett. B 450, 72 (1999)
[arXiv:hepph/9812483].
S. H. S. Alexander, Phys. Rev. D 65, 023507 (2002) [arXiv:hepth/0105032].
G. R. Dvali, Q. Shaﬁ and S. Solganik, arXiv:hepth/0105203.
G. Shiu and S. H. H. Tye, Phys. Lett. B 516, 421 (2001)
[arXiv:hepth/0106274].
C. Herdeiro, S. Hirano and R. Kallosh, JHEP 0112, 027 (2001)
[arXiv:hepth/0110271].
C. P. Burgess, P. Martineau, F. Quevedo, G. Rajesh and R. J. Zhang, JHEP
0203, 052 (2002) [arXiv:hepth/0111025].
J. GarciaBellido, R. Rabadan and F. Zamora, JHEP 0201, 036 (2002)
[arXiv:hepth/0112147].
R. Blumenhagen, B. Kors, D. Lust and T. Ott, Nucl. Phys. B 641, 235 (2002)
[arXiv:hepth/0202124].
K. Dasgupta, C. Herdeiro, S. Hirano and R. Kallosh, Phys. Rev. D 65, 126002
(2002) [arXiv:hepth/0203019].
52
Annu. Rev. Nuc. Part. Sci. 2009 59
N. T. Jones, H. Stoica and S. H. H. Tye, JHEP 0207, 051 (2002)
[arXiv:hepth/0203163].
D. Choudhury, D. Ghoshal, D. P. Jatkar and S. Panda, JCAP 0307, 009
(2003) [arXiv:hepth/0305104].
H. Firouzjahi and S. H. H. Tye, Phys. Lett. B 584, 147 (2004)
[arXiv:hepth/0312020].
C. P. Burgess, J. M. Cline, H. Stoica and F. Quevedo, JHEP 0409, 033
(2004) [arXiv:hepth/0403119].
O. DeWolfe, S. Kachru and H. L. Verlinde, JHEP 0405, 017 (2004)
[arXiv:hepth/0403123].
N. Iizuka and S. P. Trivedi, Phys. Rev. D 70, 043519 (2004)
[arXiv:hepth/0403203].
K. Dasgupta, J. P. Hsu, R. Kallosh, A. Linde and M. Zagermann, JHEP
0408, 030 (2004) [arXiv:hepth/0405247].
J. J. BlancoPillado et al., JHEP 0411, 063 (2004) [arXiv:hepth/0406230].
D. Cremades, F. Quevedo and A. Sinha, JHEP 0510, 106 (2005)
[arXiv:hepth/0505252].
J. M. Cline and H. Stoica, Phys. Rev. D 72, 126004 (2005)
[arXiv:hepth/0508029].
J. P. Conlon and F. Quevedo, JHEP 0601, 146 (2006)
[arXiv:hepth/0509012].
A. Avgoustidis, D. Cremades and F. Quevedo, arXiv:hepth/0606031.
X. Chen, S. Sarangi, S. H. Henry Tye and J. Xu, JCAP 0611, 015 (2006)
[arXiv:hepth/0608082].
C. P. Burgess, J. M. Cline, K. Dasgupta and H. Firouzjahi, JHEP 0703, 027
53
Annu. Rev. Nuc. Part. Sci. 2009 59
(2007) [arXiv:hepth/0610320].
J. R. Bond, L. Kofman, S. Prokushkin and P. M. Vaudrevange, Phys. Rev.
D 75, 123511 (2007) [arXiv:hepth/0612197].
N. Barnaby, T. Biswas and J. M. Cline, JHEP 0704, 056 (2007)
[arXiv:hepth/0612230].
B. Dutta, J. Kumar and L. Leblond, arXiv:hepth/0703278.
T. Kobayashi, S. Mukohyama and S. Kinoshita, arXiv:0708.4285 [hepth].
M. Becker, L. Leblond and S. E. Shandera, arXiv:0709.1170 [hepth].
E. Pajer, JCAP 0804, 031 (2008) [arXiv:0802.2916 [hepth]].
M. Haack, R. Kallosh, A. Krause, A. Linde, D. Lust and M. Zagermann,
Nucl. Phys. B 806, 103 (2009) [arXiv:0804.3961 [hepth]].
J. P. Conlon, R. Kallosh, A. Linde and F. Quevedo, JCAP 0809, 011 (2008)
[arXiv:0806.0809 [hepth]].
H. Y. Chen, J. O. Gong and G. Shiu, JHEP 0809, 011 (2008)
[arXiv:0807.1927 [hepth]].
M. Cicoli, C. P. Burgess and F. Quevedo, arXiv:0808.0691 [hepth].
A. Ali, R. Chingangbam, S. Panda and M. Sami, arXiv:0809.4941 [hepth].
C. P. Burgess, J. M. Cline and M. Postma, arXiv:0811.1503 [hepth].
P. J. Steinhardt and D. Wesley, arXiv:0811.1614 [hepth].
H. Y. Chen and J. O. Gong, arXiv:0812.4649 [hepth].
23. A. Linde, eConf C040802, L024 (2004) [arXiv:hepth/0503195];
S. H. Henry Tye, arXiv:hepth/0610221;
J. M. Cline, arXiv:hepth/0612129;
R. Kallosh, arXiv:hepth/0702059;
L. McAllister and E. Silverstein, Gen. Rel. Grav. 40, 565 (2008)
54
Annu. Rev. Nuc. Part. Sci. 2009 59
[arXiv:0710.2951 [hepth]].
24. A. Vilenkin, Phys. Rev. D 27, 2848 (1983).
A. D. Linde, Phys. Lett. B 175, 395 (1986).
A. D. Linde and A. Mezhlumian, Phys. Lett. B 307, 25 (1993)
[arXiv:grqc/9304015].
A. D. Linde, D. A. Linde and A. Mezhlumian, Phys. Rev. D 49, 1783 (1994)
[arXiv:grqc/9306035].
J. GarciaBellido, A. D. Linde and D. A. Linde, Phys. Rev. D 50, 730 (1994)
[arXiv:astroph/9312039].
A. Vilenkin, Phys. Rev. Lett. 74, 846 (1995) [arXiv:grqc/9406010].
J. Garriga, D. SchwartzPerlov, A. Vilenkin and S. Winitzki, JCAP 0601,
017 (2006) [arXiv:hepth/0509184].
R. Bousso, Phys. Rev. Lett. 97, 191302 (2006) [arXiv:hepth/0605263].
A. Linde, JCAP 0706, 017 (2007) [arXiv:0705.1160 [hepth]].
R. Bousso, B. Freivogel and I. S. Yang, arXiv:0808.3770 [hepth].
A. De Simone, A. H. Guth, A. Linde, M. Noorbala, M. P. Salem and
A. Vilenkin, arXiv:0808.3778 [hepth].
J. Garriga and A. Vilenkin, arXiv:0809.4257 [hepth].
S. Winitzki, Phys. Rev. D 78, 123518 (2008) [arXiv:0810.1517 [grqc]].
A. Linde, V. Vanchurin and S. Winitzki, arXiv:0812.0005 [hepth].
25. N. Itzhaki and E. D. Kovetz, JHEP 0710, 054 (2007) [arXiv:0708.2798 [hep
th]]. JHEPA,0710,054;
B. Underwood, Phys. Rev. D 78, 023509 (2008) [arXiv:0802.2117 [hepth]].
S. Bird, H. V. Peiris and R. Easther, Phys. Rev. D 78, 083518 (2008)
[arXiv:0807.3745 [astroph]].
55
Annu. Rev. Nuc. Part. Sci. 2009 59
L. Hoi and J. M. Cline, arXiv:0810.1303 [hepth].
26. E. J. Copeland, A. R. Liddle, D. H. Lyth, E. D. Stewart and D. Wands, Phys.
Rev. D 49, 6410 (1994) [arXiv:astroph/9401011].
27. S. Kachru, R. Kallosh, A. Linde and S. P. Trivedi, Phys. Rev. D 68, 046005
(2003) [arXiv:hepth/0301240].
28. M. R. Douglas and S. Kachru, Rev. Mod. Phys. 79, 733 (2007)
[arXiv:hepth/0610102].
29. S. Kachru, R. Kallosh, A. Linde, J. M. Maldacena, L. P. McAllister and
S. P. Trivedi, JCAP 0310, 013 (2003) [arXiv:hepth/0308055].
30. I. R. Klebanov and M. J. Strassler, JHEP 0008, 052 (2000)
[arXiv:hepth/0007191].
31. I. R. Klebanov and A. A. Tseytlin, Nucl. Phys. B 578, 123 (2000)
[arXiv:hepth/0002159].
32. S. B. Giddings, S. Kachru and J. Polchinski, Phys. Rev. D 66, 106006 (2002)
[arXiv:hepth/0105097].
33. D. Baumann, A. Dymarsky, I. R. Klebanov, J. M. Maldacena, L. P. McAl
lister and A. Murugan, JHEP 0611, 031 (2006) [arXiv:hepth/0607050].
34. D. Baumann, A. Dymarsky, I. R. Klebanov, L. McAllister and P. J. Stein
hardt, Phys. Rev. Lett. 99, 141601 (2007) [arXiv:0705.3837 [hepth]].
35. D. Baumann, A. Dymarsky, I. R. Klebanov and L. McAllister, JCAP 0801,
024 (2008) [arXiv:0706.0360 [hepth]].
36. A. Krause and E. Pajer, JCAP 0807, 023 (2008) [arXiv:0705.4682 [hepth]].
37. A. Ceresole, G. Dall’Agata, R. D’Auria and S. Ferrara, Phys. Rev. D 61,
066001 (2000) [arXiv:hepth/9905226].
38. H. Firouzjahi and S. H. Tye, JCAP 0503, 009 (2005) [arXiv:hepth/0501099].
56
Annu. Rev. Nuc. Part. Sci. 2009 59
39. E. Silverstein and D. Tong, Phys. Rev. D 70, 103505 (2004)
[arXiv:hepth/0310221].
40. X. Chen, JHEP 0508, 045 (2005) [arXiv:hepth/0501184].
41. J. M. Maldacena, Adv. Theor. Math. Phys. 2, 231 (1998) [Int. J. Theor.
Phys. 38, 1113 (1999)] [arXiv:hepth/9711200];
S. S. Gubser, I. R. Klebanov and A. M. Polyakov, Phys. Lett. B 428, 105
(1998) [arXiv:hepth/9802109];
E. Witten, Adv. Theor. Math. Phys. 2, 253 (1998) [arXiv:hepth/9802150].
42. C. P. Herzog, I. R. Klebanov and P. Ouyang, arXiv:hepth/0108101.
43. D. H. Lyth, Phys. Rev. Lett. 78, 1861 (1997) [arXiv:hepph/9606387].
44. D. Baumann and L. McAllister, Phys. Rev. D 75, 123508 (2007)
[arXiv:hepth/0610285].
45. J. P. Hsu, R. Kallosh and S. Prokushkin, JCAP 0312, 009 (2003)
[arXiv:hepth/0311077].
J. P. Hsu and R. Kallosh, JHEP 0404, 042 (2004) [arXiv:hepth/0402047].
46. S. Dimopoulos, S. Kachru, J. McGreevy and J. G. Wacker, JCAP 0808, 003
(2008) [arXiv:hepth/0507205].
47. A. D. Linde, Phys. Lett. B 129, 177 (1983).
48. C. Vafa, arXiv:hepth/0509212.
49. H. Ooguri and C. Vafa, Nucl. Phys. B 766, 21 (2007) [arXiv:hepth/0605264].
50. A. Adams, N. ArkaniHamed, S. Dubovsky, A. Nicolis and R. Rattazzi, JHEP
0610, 014 (2006) [arXiv:hepth/0602178].
51. K. Freese, J. A. Frieman and A. V. Olinto, Phys. Rev. Lett. 65, 3233 (1990).
52. T. Banks, M. Dine, P. J. Fox and E. Gorbatov, JCAP 0306, 001 (2003)
[arXiv:hepth/0303252].
57
Annu. Rev. Nuc. Part. Sci. 2009 59
53. P. Svrcek and E. Witten, JHEP 0606, 051 (2006) [arXiv:hepth/0605206].
54. R. Easther and L. McAllister, JCAP 0605, 018 (2006)
[arXiv:hepth/0512102].
55. A. R. Liddle, A. Mazumdar and F. E. Schunck, Phys. Rev. D 58, 061301
(1998) [arXiv:astroph/9804177].
56. R. Kallosh, N. Sivanandam and M. Soroush, Phys. Rev. D 77, 043501 (2008)
[arXiv:0710.3429 [hepth]].
57. T. W. Grimm, Phys. Rev. D 77, 126007 (2008) [arXiv:0710.3883 [hepth]].
58. N. Barnaby, C. P. Burgess and J. M. Cline, JCAP 0504, 007 (2005)
[arXiv:hepth/0412040].
L. Kofman and P. Yi, Phys. Rev. D 72, 106001 (2005)
[arXiv:hepth/0507257].
A. R. Frey, A. Mazumdar and R. C. Myers, Phys. Rev. D 73, 026003 (2006)
[arXiv:hepth/0508139].
D. Chialva, G. Shiu and B. Underwood, JHEP 0601, 014 (2006)
[arXiv:hepth/0508229].
X. Chen and S. H. Tye, JCAP 0606, 011 (2006) [arXiv:hepth/0602136].
D. R. Green, Phys. Rev. D 76, 103504 (2007) [arXiv:0707.3832 [hepth]].
A. Berndsen, J. M. Cline and H. Stoica, Phys. Rev. D 77, 123522 (2008)
[arXiv:0710.1299 [hepth]].
R. H. Brandenberger, K. Dasgupta and A. C. Davis, Phys. Rev. D 78, 083502
(2008) [arXiv:0801.3674 [hepth]].
A. Buchel and L. Kofman, Phys. Rev. D 78, 086002 (2008) [arXiv:0804.0584
[hepth]].
R. H. Brandenberger, A. Knauf and L. C. Lorenz, JHEP 0810, 110 (2008)
58
Annu. Rev. Nuc. Part. Sci. 2009 59
[arXiv:0808.3936 [hepth]].
59. M. Dine and N. Seiberg, Phys. Rev. Lett. 57, 2625 (1986).
60. J. Polchinski, “String Theory. Vol. 1: An Introduction to the Bosonic String,”
Cambridge, UK: Univ. Pr. (1998) 402 p
61. T. W. Grimm and J. Louis, Nucl. Phys. B 699, 387 (2004)
[arXiv:hepth/0403067].
62. K. Becker, M. Becker and A. Krause, Nucl. Phys. B 715, 349 (2005)
[arXiv:hepth/0501130].
63. M. P. Hertzberg, S. Kachru, W. Taylor and M. Tegmark, JHEP 0712, 095
(2007) [arXiv:0711.2512 [hepth]].
64. R. Flauger, S. Paban, D. Robbins and T. Wrase, arXiv:0812.3886 [hepth].
65. C. Caviezel, P. K¨ orber, S. Kors, D. L¨ ust, T. Wrase and M. Zagermann,
arXiv:0812.3551 [hepth].
66. A. Kosowsky [the ACT Collaboration], New Astron. Rev. 47, 939 (2003)
[arXiv:astroph/0402234].
67. J. E. Ruhl et al. [The SPT Collaboration], Proc. SPIE Int. Soc. Opt. Eng.
5498, 11 (2004) [arXiv:astroph/0411122].
68. A. C. Taylor [the Clover Collaboration], New Astron. Rev. 50, 993 (2006)
[arXiv:astroph/0610716].
69. D. Samtleben [the QUIET Collaboration], Nuovo Cim. 122B, 1353 (2007)
[arXiv:0802.2657 [astroph]].
70. K. W. Yoon et al. [the BICEP Collaboration], arXiv:astroph/0606278.
71. P. Oxley et al. [the EBEX Collaboration], Proc. SPIE Int. Soc. Opt. Eng.
5543, 320 (2004) [arXiv:astroph/0501111].
72. C. J. MacTavish et al. [the SPIDER Collaboration], arXiv:0710.0375 [astro
59
Annu. Rev. Nuc. Part. Sci. 2009 59
ph].
73. J. Bock et al., arXiv:0805.4207 [astroph].
74. M. Alishahiha, E. Silverstein and D. Tong, Phys. Rev. D 70, 123505 (2004)
[arXiv:hepth/0404084].
75. S. Sarangi and S. H. H. Tye, Phys. Lett. B 536, 185 (2002)
[arXiv:hepth/0204074].
76. E. J. Copeland, R. C. Myers and J. Polchinski, Comptes Rendus Physique 5
(2004) 1021.
77. S. E. Shandera and S. H. Tye, JCAP 0605, 007 (2006)
[arXiv:hepth/0601099].
R. Bean, S. E. Shandera, S. H. Henry Tye and J. Xu, JCAP 0705, 004 (2007)
[arXiv:hepth/0702107].
H. V. Peiris, D. Baumann, B. Friedman and A. Cooray, Phys. Rev. D 76,
103517 (2007) [arXiv:0706.1240 [astroph]].
R. Bean, X. Chen, H. V. Peiris and J. Xu, Phys. Rev. D 77, 023527 (2008)
[arXiv:0710.1812 [hepth]].
60
Annu. Rev. Nuc. Part. Sci. 2009 59 Inﬂation in Eﬀective Field Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The Eta Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . From String Compactiﬁcations to the Inﬂaton Action . . . . . . . . . . . . . . . . . . . 8 10 13
Case Study of SmallField Inﬂation: Warped Dbrane Inﬂation . . . . . . . . . . .
D3branes in Warped Throat Geometries . . . . . . . . . . . . . . . . . . . . . . . . . . The D3brane Potential . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Supergravity Analysis of the D3brane Potential . . . . . . . . . . . . . . . . . . . . . . Gauge Theory Interpretation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
16
16 18 20 26 30
Summary and Perspective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
LargeField Inﬂation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
The Lyth Bound . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . SuperPlanckian Fields and Flat Potentials . . . . . . . . . . . . . . . . . . . . . . . .
31
31 33
Case Study of LargeField Inﬂation: Axion Monodromy . . . . . . . . . . . . . . .
Axions in String Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Axion Inﬂation in String Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Compactiﬁcation Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Summary and Perspective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
35
36 38 40 44
Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Theoretical Prospects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Observational Prospects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
45
45 47 49
2
Annu. Rev. Nuc. Part. Sci. 2009 59
1
Introduction
Recent advances in observational cosmology have brought us closer to a fundamental understanding of the origin of structure in the universe. Observations of variations in the cosmic microwave background (CMB) temperature and of the spatial distribution of galaxies in the sky have yielded a consistent picture in which gravitational instability drives primordial ﬂuctuations to condense into largescale structures, such as our own galaxy. Moreover, quantum ﬁeld theory and general relativity provide an elegant microphysical mechanism, inﬂation, for generating these primordial perturbations during an early period of accelerated expansion. The classical dynamics of this inﬂationary era (1, 2, 3) explains the largescale homogeneity, isotropy and ﬂatness of the universe, while quantum ﬂuctuations during inﬂation lead to small inhomogeneities. The general properties of the spectrum of inﬂationary inhomogeneities were predicted long ago (4, 5, 6, 7, 8, 9) and are in beautiful agreement with recent CMB observations, e.g. by the Wilkinson Microwave Anisotropy Probe (WMAP) (10). Although inﬂation is remarkably successful as a phenomenological model for the dynamics of the very early universe, a detailed understanding of the physical origin of the inﬂationary expansion has remained elusive. In this review we will highlight speciﬁc aspects of inﬂation that depend sensitively on the ultraviolet (UV) completion of quantum ﬁeld theory and gravity, i.e. on the ﬁeld content and interactions at energies approaching the Planck scale. Such issues are most naturally addressed in a theory of Planckscale physics, for which string theory is the bestdeveloped candidate. This motivates understanding the physics of inﬂation in string theory.
3
Both the inﬂaton (scalar) ﬂuctuations and the graviton (tensor) 4 . 2009 59 1. This requirement for accelerated expansion can be fulﬁlled by a range of qualitatively diﬀerent mechanisms with varied theoretical motivations (11. (12. where the inﬂationary dynamics is described by a single order parameter φ (a fundamental scalar ﬁeld or a compos1 ite ﬁeld) with canonical kinetic term 2 (∂µ φ)2 and potential energy density V (φ). Prolonged accelerated expansion then occurs if the slowroll parameters are small 2 Mpl ǫ≃ 2 V′ V 2 ≪ 1. Ref. inﬂation stretches quantum ﬂuctuations of light degrees of freedom (m ≪ H). Sci. so we pause to explain them. For concreteness. Part. Rev. These ﬂuctuations are visible to us as anisotropies in the CMB temperature. Fluctuations of the inﬂaton lead to perturbations of the time at which inﬂation ends. we restrict ourselves to the simple case of singleﬁeld slowroll inﬂation.Annu. see e. ds2 = −dt2 + e2Ht dx2 . In addition to smoothing the universe on large scales. V (2) where the primes denote derivatives with respect to the inﬂaton ﬁeld φ. creating a spectrum of small perturbations in the observed CMB temperature and polarization.g.1 Inﬂation Inﬂation may be deﬁned as a period of exponential expansion of space. These perturbations are the key to structure formation and to tests of inﬂation. where ǫ ≡ − ˙ H < 1. Nuc. 13) for a more detailed discussion. 12. 2 η = Mpl V ′′ ≪ 1. 13). In inﬂationary scenarios with a single inﬂaton ﬁeld. H2 (1) which arises if the universe is dominated by a form of stressenergy that sources a nearlyconstant Hubble parameter H. and hence source perturbations in the energy density after inﬂation. the light degrees of freedom during inﬂation are the inﬂaton itself and the two polarization modes of the graviton.
The energy scale of inﬂation determines the amplitude of tensor perturbations. The shapes of the primordial perturbation spectra are therefore intimately tied to the inﬂationary background dynamics as dictated by the shape of the inﬂaton potential V (φ). and nearfuture experiments will almost certainly continue this trend. The nature of the inﬂationary epoch is imprinted on the sky in the temperature and polarization anisotropies of the CMB. Sci. The search for primordial Bmodes is a key eﬀort of observational cosmology (18). and is generated by both scalar and tensor perturbations. • Bmode: this divergencefree mode is characterized by polarization vectors with vorticity around any point on the sky. 15) made the important observation that the spin2 polarization ﬁeld of the CMB photons may be decomposed into two distinct scalar or spin0 modes: • Emode: this curlfree mode is characterized by polarization vectors that are radial around cold spots and tangential around hot spots on the sky. Temperature aniso5 . Emode polarization and its crosscorrelations with the CMB temperature ﬂuctuations were ﬁrst detected by DASI (16) and have recently been mapped out in greater detail by WMAP (17). In slowroll inﬂation. small ǫ and η ensure that the spectra of scalar and tensor ﬂuctuations are nearly scaleinvariant. so that a detection would ﬁx the inﬂationary energy scale. Primordial Bmodes can only be produced by gravitational waves and are therefore considered an unambiguous signature of inﬂationary tensor perturbations. Nuc. Rev.Annu. 2009 59 ﬂuctuations source polarization of the CMB photons. Ref. and hence the Bmode amplitude. CMB observations have improved dramatically in the past decade. Part. (14.
2009 59 tropies have been measured at the cosmic variance limit over a large range of scales.g. e. (2). far out of reach of terrestrial collider experiments. ∆η ∼ O(1). This requires either phenomenological as 6 . Part. by exponentially diluting any preexisting density of GUTscale relics.Annu. Sci. the duration and the details of inﬂation are nevertheless sensitive to certain aspects of Planckscale physics. To assess whether inﬂation can nevertheless occur requires detailed information about Plancksuppressed corrections to the inﬂaton potential.1 Flatness of the Inflaton Potential From a topdown perspec tive the ﬂatness of the inﬂaton potential in Planck units. but we now brieﬂy preview two examples of the UV sensitivity of inﬂation that will be central to our discussion. This provides a unique opportunity to probe physics at energies near the GUT scale. Eq. as quantiﬁed by the slowroll conditions. A detection of inﬂationary tensor perturbations via their unique Bmode signature would be especially interesting. Rev. Nuc. there are more detailed motivations for studying inﬂation in the context of string theory. 1. 1. As we will show in §2. small (Plancksuppressed) corrections to the potential often induce important corrections to the curvature of the potential. is a nontrivial constraint.2 Motivation for Inﬂation in String Theory Besides the intellectual satisfaction of providing a microscopic description of the inﬂationary process. and experimentalists are now preparing for precision measurements of the polarization of the CMB (18). While inﬂation is frustratingly eﬀective at making most signatures of highenergy physics unobservable. In the remainder of this review we ﬂesh this out in more detail.2. as their amplitude relates directly to the energy scale of inﬂation.
1. the UV sensitivity of inﬂation is particularly strong in models with observable gravitational waves.Annu. and space considerations prevent us from presenting a truly comprehensive review that summarizes and assesses each important class of models. our goal is an exposition of what 7 . the issue of controlled largeﬁeld inﬂation is of both theoretical and experimental relevance (12).2 Inflationary Gravitational Waves As we explain in §4. Sci. about physics at the Planck scale. ∆φ ≫ Mpl . Part. (Some representative contributions to the subject include (22). 2009 59 sumptions. In §5 we present the ﬁrst controlled examples of largeﬁeld inﬂation in string theory (20. or preferably microphysical knowledge. A large gravitational wave signal from inﬂation is associated with a high energy scale for the inﬂaton potential and a superPlanckian variation of the inﬂaton ﬁeld.000 papers. Nuc.2. and in §3 we give an example of such an analysis for the case of warped Dbrane inﬂation (19). 21). such as string theory. Models of largeﬁeld inﬂation are therefore most naturally studied in a UVcomplete theory.) Instead. Rev. between the time when CMB ﬂuctuations were created and the end of inﬂation. 1.3 Organization of this Review Inﬂation in string theory is the subject of close to 1. we refer the reader to the reviews (23) for a more complete list of references. In §4 we explain why theoretical control of the shape of the potential over a superPlanckian range requires certain assumptions about the UV structure of the theory. String theory equips us to compute such corrections from ﬁrst principles. Given the exciting possibility of measuring the gravitational wave signature of inﬂation in the polarization of the CMB.
Annu.g. (However. in wide classes of models. To illustrate this idea in depth. Moreover. e. Inﬂation requires a potential that is quite ﬂat in Planck units (see Eq. the set of Lagrangians suitable for inﬂation is a minute subset of the set of all possible Lagrangians. Part. we focus on two case studies: warped Dbrane inﬂation and axion monodromy inﬂation. inﬂation emerges only for rather special initial conditions. These two scenarios are instructive examples from the general classes of smallﬁeld and largeﬁeld inﬂation. Nuc. initial conﬁgurations with tiny kinetic energy. Rev. Sci. respectively. In particular. 8 . (25) for recent eﬀorts to quantify or to ameliorate the ﬁnetuning of initial conditions. the question of initial conditions appears inextricable from the active yet incomplete program of understanding measures in eternal inﬂation (24). For a single inﬂaton ﬁeld with a canonical kinetic term. and as we now argue.) In this review we will focus on the question of how (un)natural it is to have a Lagrangian suitable for inﬂation. As we will explain. the necessary conditions for inﬂation can be stated in terms of the inﬂaton potential. (2)). 2 2. 2009 59 we believe to be the primary theme of the subject: the sensitivity of inﬂation to Planckscale physics.g. see e. in the case of smallﬁeld scenarios. this is the twoway connection by which string theory can clarify inﬂationary modelbuilding.1 Inﬂation in String Theory Inﬂation in Eﬀective Field Theory As a phenomenon in quantum ﬁeld theory coupled to general relativity. and cosmological experiments can constrain string theory models. Although one would hope to explore and quantify the naturalness both of inﬂationary Lagrangians and of inﬂationary initial conditions. inﬂation does not appear to be natural.
the ﬂatness of the potential in Planck units introduces (3) 9 . In inﬂation. As usual. Planckscale processes. the eﬀects of highscale physics above some cutoﬀ Λ are eﬃciently described by the coeﬃcients of operators in the lowenergy eﬀective theory.) Equivalently. M δ−4 where δ denotes the mass dimension of the operator. are irrelevant for most of particle physics: they decouple from lowenergy phenomena. testable – consequences for the form of the inﬂaton potential. 2009 59 this condition is sensitive to Planckscale physics. Part. Sensitivity to such operators is commonplace in particle physics: for example. Integrating out particles of mass M ≥ Λ gives rise to operators of the form Oδ . Sci. nor go beyond M ∼ MGUT . Nuc. the structure of the Planckscale theory has meaningful – and. we cannot say whether the physics of the Planck scale is a ﬁnite theory of quantum gravity. just as unitarity of W W scattering requires new physics at the TeV scale. in very favorable cases. Rev. However. however. and lower bounds on the proton lifetime even allow us to constrain GUTscale operators that would mediate proton decay. Although we know that new degrees of freedom must emerge. (Scenarios of gravitymediated supersymmetry breaking are one exception. However. and operators of very high dimension. bounds on ﬂavorchanging processes place limits on physics above the TeV scale. such as string theory. Let us recall that the presence of some form of new physics at the Planck scale is required in order to render gravitongraviton scattering sensible.Annu. or is instead simply an eﬀective theory for some unimagined physics at yet higher scales. particle physics considerations alone do not often reach beyond operators of dimension δ = 6.
§2. to ensure that the theory supports at least 60 efolds of inﬂationary expansion. or equivalently corrects the eta parameter by order one. If the dimensionfour operator O4 has a vacuum expectation value (vev) comparable to the inﬂationary energy density. Part. contributions to the Lagrangian of the general form O6 O4 2 2 = M2 φ Mpl pl (5) are allowed. leading to an important problem for inﬂationary modelbuilding. Rev. then this term corrects the inﬂaton mass by order H. 2. As we discuss in §4. Let us reiterate that contributions of this form may be thought of as arising from integrating out Planckscale degrees of freedom. Mpl (4) As we explain in §2. In this section we discuss this socalled eta problem in eﬀective ﬁeld theory. O4 ∼ V . This sensitivity to dimensionsix Plancksuppressed operators is therefore common to all models of inﬂation. and illustrate the problem in a supergravity example. an understanding of such operators is required to address the smallness of the eta parameter. such as O6 2 .2.2. in this important class of inﬂationary models the potential becomes sensitive to an inﬁnite series of operators of arbitrary dimension. 10 . Nuc. Sci.2.e. i.2. 2009 59 sensitivity to δ ≤ 6 Plancksuppressed operators.2 The Eta Problem In the absence of any speciﬁc symmetries protecting the inﬂaton potential.Annu. §2. For largeﬁeld models of inﬂation the UV sensitivity of the inﬂaton action is dramatically enhanced.1.
Since the cutoﬀ for an eﬀective theory of inﬂation is at least the Hubble scale. In §5 we discuss the natural proposal to protect the inﬂaton potential via a shift symmetry φ → φ + const. In N = 1 supergravity. i..2 Supergravity Example An important instance of the eta problem ∆m2 φ ≥ 1. This case is relevant for many string theory models of inﬂation because fourdimensional supergravity is the lowenergy eﬀective theory of supersymmetric string compactiﬁcations.2.2. in supergravity (26). a key term in the scalar potential is the Fterm poten 11 . Rev.e. Equivalently. and the resulting splittings in supermultiplets are of order H. the eta parameter receives radiative corrections. Nuc. Sci. 2. which is equivalent to identifying the inﬂaton with a pseudoNambuGoldstoneboson. but supersymmetry does not suﬃce to stabilize the inﬂaton mass: the inﬂationary energy necessarily breaks supersymmetry. ∆η = preventing prolonged inﬂation. this implies that a small inﬂaton mass (mφ ≪ H) is radiatively unstable. 3H 2 (6) arises in locallysupersymmetric theories. In the absence of such a symmetry the eta problem seems to imply the necessity of ﬁnetuning the inﬂationary action in order to get inﬂation. Λ ≥ H. Part. The diﬃculty here is analogous to the Higgs hierarchy problem. 2009 59 2. the mass of a scalar ﬁeld runs to the cutoﬀ scale unless it is protected by some symmetry.1 Radiative Instability of the Inflaton Mass In a generic ef fective theory with cutoﬀ Λ.Annu. so that supersymmetry does not protect a small inﬂaton mass mφ ≪ H.
but generalizing our expressions to include other ﬁelds is straightforward. K(ϕ. ϕ) and W (ϕ) are the K¨hler potential and the superpotential. −K. 2 W  Mpl (7) where K(ϕ. . The K¨hler potential determines the inﬂaton kinetic term. while a ¯ ¯ the superpotential determines the interactions. we have assumed that there are no other light degrees of freedom..e. ϕ) = K0 + K. 2009 59 tial. (7) and clearly depend on the modeldependent structure of the K¨hler a potential and the superpotential. The omitted terms.ϕϕ ∂ϕ∂ ϕ. Nuc. Sci. (7). ¯ VF = eK/Mpl K ϕϕ Dϕ W Dϕ W − 2 3 2 . The result is of the form of Eq. i.. re¯ a spectively.ϕϕ 0 ϕϕ + · · · . which we denote by ϕ ≡ 0 without loss of generality. Rev. ϕ is a complex scalar ﬁeld which is taken to be the inﬂaton. We have retained the leading correction to the potential originating in the expansion of eK/Mpl in Eq.ϕϕ ∂ϕ∂ ϕ − V0 1 + K. Mpl ϕϕ ¯ 2 + . and we −2 have deﬁned Dϕ W ≡ ∂ϕ W + Mpl (∂ϕ K)W . some of which can be of the same ¯ order as the terms we keep.. To derive the inﬂaton mass. For simplicity of presentation.. (4) with ¯ O6 = V0 φφ (10) 12 . arise from expanding K ϕϕ Dϕ W Dϕ W − 3 2 2 W  Mpl 2 in Eq. The inﬂationary Lagrangian ¯ ¯ ¯ then becomes L ≈ −K.Annu. Part. Mpl (8) (9) ¯ where we have deﬁned the canonical inﬂaton ﬁeld φφ ≈ Kϕϕ 0 ϕϕ and V0 ≡ ¯ ¯ VF ϕ=0 .ϕϕ 0 ¯ ¯ ¯ ¯ ≡ −∂φ∂ φ − V0 1 + ¯ φφ 2 + . we expand K around some chosen origin. which could plausibly be called a universal correction in Fterm scenarios.
Nuc. The ﬁrst case does not often arise. so that one can sensibly pursue such a ﬁnetuning argument.1 From String Compactiﬁcations to the Inﬂaton Action Elements of String Compactifications It is a famous fact that the quantum theory of strings is naturally deﬁned in more than four spacetime 13 . This requires either that the Fterm potential is negligible. so that the smallness of the inﬂaton mass is a result of ﬁnetuning. to any order in perturbation theory. Rev.3 2.3. eta will generically be of order unity. or that the inﬂaton does not appear in the Fterm potential. (11) as well as a modeldependent contribution which is typically of the same order. 28). a or in the Fterm potential. However. Sci. 2. Clearly. because Fterm potentials play an important role in presentlyunderstood models for stabilization of the compact dimensions of string theory (27. It is therefore clear that in an inﬂationary scenario driven by an Fterm potential. Part. In the case study of §3 we will provide a concrete example in which the structure of all relevant contributions to eta can be computed. This evades the particular incarnation of the eta problem that we have described above. in §5 we will present a scenario in which the inﬂaton is an axion and does not appear in the K¨hler potential. 2009 59 and implies a large modelindependent contribution to the eta parameter ∆η = 1 . it would be far more satisfying to exhibit a mechanism that removes the eta problem by ensuring that ∆η ≪ 1. Under what circumstances can inﬂation still occur. in a model based on a supersymmetric Lagrangian? One obvious possibility is that the modeldependent contributions to eta approximately cancel the modelindependent contribution.Annu.
2009 59 dimensions.3. the procedure in question may be written schematically as S10 [C] → S4 . we will focus on compactiﬁcations of the critical tendimensional type IIB string theory on sixdimensional CalabiYau spaces (to be precise.2 The Effective Inflaton Action For our purposes. 2. By understanding the space of possible data C and the nature of the map in Eq. compactiﬁcations that give rise to interesting fourdimensional physics. we can hope to identify. and gauge ﬁeld conﬁgurations of any Dbranes. in turn. Rev. and discrete data such as quantized ﬂuxes and wrapped Dbranes.Annu. Part. A central task in string theory modelbuilding is to understand in detail how the tendimensional sources determine the fourdimensional eﬀective theory. geometry. For concreteness. scalar potentials. kinetic terms. and from the positions. orientations. with fourdimensional physics emerging upon compactiﬁcation of the additional spatial dimensions. but we will not need this ﬁne point. and perhaps even classify. Sci. Scalar ﬁelds known as moduli arise from deformations of the compactiﬁcation manifold. the expectation values 14 . our compactiﬁcations will only be conformal to spaces that are wellapproximated by orientifolds of CalabiYau manifolds. Nuc. If we denote the tendimensional compactiﬁcation data by C. (12) Distinct compactiﬁcation data C give rise to a multitude of fourdimensional eﬀective theories S4 with varied ﬁeld content. the most im portant degrees of freedom of the eﬀective theory are fourdimensional scalar ﬁelds. typically numbering in the hundreds for the CalabiYau spaces under consideration. and symmetry properties.) The vast number of distinct compactiﬁcations in this class are distinguished by their topology. (12). From given compactiﬁcation data one can compute the kinetic terms and scalar potentials of the moduli.
and many (though not always all) of the moduli ﬁelds become massive (28). This is completely analogous to the appearance of corrections from higherdimension operators in our discussion of eﬀective ﬁeld theory in §2. Nuc. It is useful to divide the scalar ﬁelds arising in S4 into a set of light ﬁelds φ. one must understand all the scalar ﬁelds. Rev.eﬀ (φ.Annu. denoted φ. they typically have important eﬀects on the dynamics. if scalar ﬁelds in addition to the inﬂaton are light during inﬂation. they can release this energy during or after Big Bang nucleosynthesis. moduli stabilization contributes to the eta problem. spoiling the successful predictions of the light ele15 . Here one of the light ﬁelds. ψ with masses below the Hubble scale (mφ . suﬃciently massive moduli ﬁelds are eﬀectively frozen during inﬂation. Sci. χ) → S4. S4 (φ. mψ ≪ H) and a set of heavy ﬁelds χ with masses much greater than the Hubble scale (mχ ≫ H). 2009 59 of the moduli determine the parameters of the fourdimensional eﬀective theory. light degrees of freedom can create problems for latetime cosmology. ψ. and one should study the evolution of all ﬁelds ψ with masses mψ ≪ H. if they persist after inﬂation. both heavy and light. Next.1. ψ) . Light scalars absorb energy during inﬂation and. has been identiﬁed as the inﬂaton candidate. such as Dbranes and quantized ﬂuxes. Moreover. Part. In the presence of generic tendimensional sources of stressenergy. First. there is an energy cost for deforming the compactiﬁcation. To understand whether successful inﬂation can occur. and one should integrate them out to obtain an eﬀective action for the light ﬁelds only. (13) Integrating out these heavy modes generically induces contributions to the potential of the putative inﬂaton: that is. even if the resulting multiﬁeld inﬂationary dynamics is suitable.
as they mediate ﬁfth forces of gravitational strength. To illustrate these issues. the D3brane ﬁlls our fourdimensional spacetime and is pointlike in the extra dimensions (see Figure 1). warped Dbrane inﬂation. The global compactiﬁcation is assumed to be a warped product of fourdimensional spacetime (with metric gµν ) and a conformallyCalabiYau space. but this is not required on physical grounds: it serves only to arrange that the eﬀective theory during inﬂation has only a single degree of freedom. Sci. as in this case the moduli decay before Big Bang nucleosynthesis. 2009 59 ment abundances. it suﬃces to ensure that mψ ≫ 30 TeV. (14) 16 . In the following we therefore present a case study of a comparatively wellunderstood model of smallﬁeld inﬂation. and in principle even their coeﬃcients can be computed in terms of given compactiﬁcation data. ds2 = e2A(y) gµν dxµ dxν + e−2A(y) gmn dy m dy n . it is useful to examine a concrete model in detail. Moreover. A simplifying assumption that is occasionally invoked is that all ﬁelds aside from the inﬂaton should have m ≫ H.1 D3branes in Warped Throat Geometries In this scenario inﬂation is driven by the motion of a D3brane in a warped throat region of a stabilized compact space (29).Annu. Rev. 3 Case Study of SmallField Inﬂation: Warped Dbrane Inﬂation In string theory models of inﬂation the operators contributing to the inﬂaton potential can be enumerated. To preserve fourdimensional Lorentz (or de Sitter) invariance. light moduli would be problematic in the present universe. 3. To avoid these latetime problems. Nuc. Part.
Annu. Rev. Nuc. Part. Sci. 2009 59
with gmn a CalabiYau metric that can be approximated in some region by a cone over a ﬁvedimensional Einstein manifold X5 , gmn dy m dy n = dr 2 + r 2 ds2 5 . X (15)
A canonical example of such a throat region is the KlebanovStrassler (KS) geometry (30), for which X5 is the SU (2) × SU (2) /U (1) coset space T 1,1 , and the wouldbe conical singularity at the tip of the throat, r = 0, is smoothed by the presence of appropriate ﬂuxes. The tip of the throat is therefore located at a ﬁnite radial coordinate rIR , while at r = rUV the throat is glued into an unwarped bulk geometry. In the relevant regime rIR ≪ r < rUV the warp factor may be written as (31) e−4A(r) = where ln rUV 2πK ≈ . rIR 3gs M (17) R4 r ln , 4 r rIR R4 ≡ 81 (gs M α′ )2 , 8 (16)
Here, M and K are integers specifying the ﬂux background (30, 32). Warping sourced by ﬂuxes is commonplace in modern compactiﬁcations, and there has been much progress in understanding the stabilization of the moduli of such a compactiﬁcation (28). Positing a stabilized compactiﬁcation containing a KS throat therefore seems reasonable given present knowledge. Inﬂation proceeds as a D3brane moves radially inward in the throat region, towards an antiD3brane that is naturally situated at the tip of the throat. The inﬂaton kinetic term is determined by the DiracBornInfeld (DBI) action for a probe D3brane, and leads to an identiﬁcation of the canonical inﬂaton ﬁeld with a multiple of the radial coordinate, φ2 ≡ T3 r 2 . Here, T3 ≡ (2π)3 gs α′2
−1
is the D3brane tension, with gs the string coupling and 2πα′ the inverse string 17
Annu. Rev. Nuc. Part. Sci. 2009 59
tension. The exit from inﬂation occurs when open strings stretched between the approaching pair become tachyonic and condense, annihilating the branes.
Ψ
D3
D3
r bulk
warped throat
Figure 1: D3brane inﬂation in a warped throat geometry. The D3branes are spacetimeﬁlling in four dimensions and therefore pointlike in the extra dimensions. The circle stands for the base manifold X5 with angular coordinates Ψ. The brane moves in the radial direction r. At rUV the throat attaches to a compact CalabiYau space. AntiD3branes minimize their energy at the tip of the throat, rIR . In this simpliﬁed picture, inﬂation is driven by the extremely weak (warpingsuppressed) Coulomb interaction of the braneantibrane pair (29). The true story, however, is more complex, as moduli stabilization introduces new terms in the inﬂaton potential which typically overwhelm the Coulomb term and drive more complicated dynamics (29, 33, 34, 35, 36, 19). This pattern is precisely what we anticipated in our eﬀective ﬁeld theory discussion: integrating out moduli ﬁelds can be expected to induce important corrections to the potential.
3.2
The D3brane Potential
An important correction induced by moduli stabilization is the inﬂaton mass term arising from the supergravity Fterm potential, §2.2.2. In a vacuum stabilized 18
Annu. Rev. Nuc. Part. Sci. 2009 59
by an Fterm potential, i.e. by superpotential terms involving the moduli, one
φ 1 2 ﬁnds the mass term H0 φ2 = 3 V0 (φ⋆ ) M 2 (29), where φ⋆ is an arbitrary reference
pl 2
value for the inﬂaton ﬁeld and the parameter H0 should not be confused with the presentday Hubble constant. However, one expects additional contributions to the potential from a variety of other sources, such as additional eﬀects in the compactiﬁcation that break supersymmetry (19). Let us deﬁne ∆V (φ) to encapsulate all contributions to the
2 potential aside from the Coulomb interaction V0 (φ) and the mass term H0 φ2 ;
then the total potential and the associated contributions to the eta parameter may be written as
2 V (φ) = V0 (φ) + H0 φ2 + ∆V (φ)
(18) = ? (19)
η(φ) =
η0
+
2 3
+ ∆η(φ)
where η0 ≪ 1 because the Coulomb interaction is very weak. (More generally, V0 (φ) can be deﬁned to be all terms in V (φ) with negligible contributions to η. Besides the braneantibrane Coulomb interaction, this can include any other sources of nearlyconstant energy, e.g. bulk contributions to the cosmological constant.) Clearly, η can only be small if ∆V can cancel the mass term in Eq. (18). We must therefore enumerate all relevant contributions to ∆V , and attempt to understand the circumstances under which an approximate cancellation can occur. Note that identifying a subset of contributions to ∆V while remaining ignorant of others is insuﬃcient. Warped Dbrane inﬂation has received a signiﬁcant amount of theoretical attention in part because of its high degree of computability. Quite generally, if we had access to the full data of an explicit, stabilized compactiﬁcation with 19
at least at the level of an unstabilized compactiﬁcation.3 3. Our treatment will allow us to give explicit expressions for the correction terms ∆V in Eq. We then consider the 20 . for two reasons: for general CalabiYau spaces hardly any metric data is available.. 2009 59 small curvatures and weak string coupling. we would in principle be able to compute the potential of a Dbrane inﬂaton to any desired accuracy.e. However.1 Supergravity Analysis of the D3brane Potential Perturbations to the Geometry Above we gave the explicit so lution for the noncompact warped throat region. The method involves examining perturbations to the supergravity solution that describes the throat in which the D3brane moves. a suﬃciently long throat is wellapproximated by a noncompact throat geometry (i. Nuc. for which the CalabiYau metric can often be found. Part. (18). a throat of inﬁnite length). Having complete metric data greatly facilitates the study of probe Dbrane dynamics. Furthermore. and hence to extract the characteristics of inﬂation in the presence of moduli stabilization. 3. For concreteness we will work with the example of a KS throat. We now describe a systematic way to estimate the leading corrections to the throat solution as perturbations to the geometry and ﬂuxes at large r (near rUV in Figure 1). by performing a careful dimensional reduction. Rev. we will now explain how the eﬀects of moduli stabilization and of the ﬁnite length of the throat can be incorporated systematically. but the method is far more general.3. This is not possible at present for a generic compact CalabiYau. as in the important example of the KlebanovStrassler solution (30). and examples with entirely explicit moduli stabilization are rare. Sci. which is entirely explicit and everywhere smooth.Annu.
Φ− vanishes in the unperturbed KS throat. (14). Finally. and one can therefore draw few general conclusions about the D3brane potential. Part. Type IIB supergravity contains a metric. the Einstein equations and ﬁveform Bianchi identity imply that perturbations of Φ− around such backgrounds satisfy. which is the lowenergy limit of type IIB string theory. as well as various pform ﬁelds. and more generally in the class of ﬂux compactiﬁcations considered by Giddings. one can often solve the Laplace equation. and Polchinski (GKP) (32). in a noncompact throat geometry. 2009 59 eﬀect of these perturbations on the potential for a D3brane at a location well inside the throat. the potential for a D3brane in a throat that is glued into 21 . Sci. (21) (20) and ⋆10 is the tendimensional Hodge star.Annu. Importantly. Moreover. the sixdimensional Laplace equation (32. at the linear level. we need a few facts about the coupling of D3branes to the background ﬁelds of type IIB supergravity. little is known about the solution for Φ− . (22) In a general compactiﬁcation. 19). where α is the potential for the ﬁveform ﬁeld F5 = (1 + ⋆10 ) dα(y) ∧ dx0 ∧ dx1 ∧ dx2 ∧ dx3 . Rev. Kachru. ∇2 Φ − = 0 . the fourdimensional potential V as a function of the D3brane position in the extra dimensions is only aﬀected by a very speciﬁc combination of background ﬁelds: V = T3 e4A − α ≡ T3 Φ− . Nuc. The D3brane potential is therefore dictated by the proﬁle of Φ− ≡ e4A − α. Before we can explain the idea underlying this approach. which in the background takes the form of Eq. Furthermore. However.
it must be possible to express the D3brane potential via some Φ− proﬁle in the throat. Part. Rev.1 . Sci. R). Moreover. i.) Thus. The quantities ∆(L) are related to the eigenvalues of the angular Laplacian ∆(L) ≡ −2 + 6[J1 (J1 + 1) + J2 (J2 + 2) − R2 /8] + 4 .c. (25) 22 . as Φ− is the only supergravity ﬁeld that sources a D3brane potential. the structure of the inﬂaton potential is dictated by the structure of solutions of the Laplace equation in the noncompact throat.3. the Φ− proﬁle is given to an excellent approximation by a solution of the Laplace equation in the noncompact throat geometry. Ψ) = L fL (Ψ) r rUV ∆(L) .1 by YLM (Ψ) (37). Nuc. (23) where fL (Ψ) ≡ ΦLM YLM (Ψ) + c. 2009 59 an arbitrary compact CalabiYau must be expressible in terms of a solution for Φ− in the throat. J2 . (Corrections are expected to be exponentially small when the throat is long. . M ≡ (m1 .Annu. let us repeat: no matter how complicated the CalabiYau to which the throat is attached. As this is the crucial idea. in terms of a superposition of harmonic functions. for a suﬃciently long but ﬁnite throat. We denote the eigenfunctions of the angular Laplacian on the base manifold X5 = T 1. where the multiindices L ≡ (J1 . M (24) and ΦLM are constant coeﬃcients.e. 3.2 Harmonic Analysis Let us therefore solve the Laplace equation (22) in the KS background. We may then express the solution of the Laplace equation (22) as the following expansion Φ− (r. m2 ) label SU (2) × SU (2) × U (1) quantum numbers under the corresponding isometries of T 1.
1/2. then the dynamics is signiﬁcantly more complicated than what is described below. It follows that at ﬁxed radial location. we ﬁrst minimize the potential in the angular directions. (28) to the radial potential is negative. (28). 0. J2 . (28) is minimized at an angular location where the contribution of Eq. Nuc. 2009 59 To determine the leading perturbations to the brane potential we are interested in the lowest eigenvalues. the potential reduces to an eﬀective singleﬁeld potential for the radial direction r. 1) . 0) . (0. This contribution to the potential is minimized at r → ∞.e. i. Part. 0). J2 . one ﬁnds that the smallest eigenvalues corresponding to nontrivial perturbations are ∆= 3 2 for (J1 . (28) (If more than one angular mode is relevant during inﬂation.) To isolate the radial dynamics. Therefore. The 23 . The proof is straightforward: any nonconstant spherical harmonic is orthogonal to the constant (L = 0) harmonic. there always exists an angular location Ψ⋆ where fL (Ψ⋆ ) is negative. and hence any nontrivial harmonic necessarily attains both positive and negative values. R) = (1/2. (23) these correspond to the perturbations that diminish most slowly towards the tip of the throat. Incorporating the grouptheoretic selection rules that restrict the allowed quantum numbers (37). Φ− ≈ fL (Ψ) r rUV ∆(L) . Rev. When the angular coordinates have relaxed to their minima. 1.Annu. (26) (27) ∆=2 For simplicity. In the singleperturbation case of Eq. (23). R) = (1. Sci. the Φ− perturbation then always leads to a repulsive force. we now assume that a single mode dominates the expansion in Eq. the D3brane potential induced by the term in Eq. for (J1 . since via Eq. the eﬀect of the perturbation is to push the brane out of the throat.
The above classiﬁcation of the leading perturbations to the inﬂaton potential via the eigenvalues of the angular Laplacian hence leads to two cases with distinct phenomenology: 1. the dominant perturbation corresponds to the 3 smallest possible eigenvalue. the eta parameter can be ﬁnetuned to be small locally. 2009 59 potential induced by any individual perturbation of the form of Eq. c∆ are highly modeldependent and were estimated in Ref. ∆ = 2 . The dynamics during inﬂation is then governed by the following phenomenological potential φ 2 2 V (φ) = V0 (φ) + Mpl H0 Mpl 2 − a3/2 φ Mpl 3/2 . near an approximate inﬂection point. this gives a negative contribution to the potential in Eq. This is fortunate. Nuc. (28) therefore produces a radiallyexpulsive force. Sci. (19). then the radial D3brane potential is 2 2 V (φ) = V0 (φ) + Mpl H0 φ Mpl 2 − a∆ φ Mpl ∆ . and φ ∝ r.3 Phenomenological Implications If only one angular mode dom inates the UV perturbation of the throat geometry. (18). ∆V ∝ −φ3/2 . 3. Fractional Case In a general compactiﬁcation. This inﬂection point model is phenomenologically identical to the explicit model of Dbrane inﬂation 24 . Rev. Part. (29) where a∆ ≡ c∆ Mpl φUV ∆ . since only a repulsive force allows cancellation with the mass term in Eq. By the repulsivity argument we just gave.3.Annu. (31) For a potential of this form. (30) The magnitudes of the coeﬃcients a∆ . (18) to alleviate the eta problem.
the potential becomes steep. it may be for bidden by a discrete symmetry. 25 . 34. Quadratic Case Although the ∆ = 3 2 perturbation is generically dominant. The parameter β may even be negative. In this case. 35. The relevant phenomenological model is then 2 V (φ) = V0 (φ) + βH0 φ2 . by an unbroken global symmetry of the full compact manifold (19). In that case we expect the angular dynamics of the brane to be signiﬁcantly more complicated.) As β → 1. Rev. but inﬂation may still occur via the DBI eﬀect (39). (28). 2009 59 (33. (32) where the parameter β allows a nearly continuous tuning of the inﬂaton mass. pushing the brane out of the throat and allowing a realization of IR DBI inﬂation (40). Sci. (38). the leading correction comes from the ∆ = 2 perturbation. (29) and the phenomenology for general β was discussed in Ref. 36) in which a modulistabilizing D7brane stack descends into the throat region while wrapping a suitable fourcycle. with potentially important consequences for the eﬀective singleﬁeld potential. In general. Nuc. (28). corrections from perturbations with ∆ > 2 can be important. The above summarizes the phenomenology of warped Dbrane inﬂation under the simplifying assumption that a single angular mode dominates in the Φ− perturbation of Eq. ∆V ∝ −φ2 . 2.Annu. Part. The maximallytuned case β = 0 was ﬁrst analyzed in Appendix D of Ref.e. i. more than one Lmode may be important in Eq. (We should note that in the limit β ≪ 1.
Rev. The corresponding N = 1 supersymmetric gauge theory is approximately conformal over a large range of energy scales. One advantage of this dual description is that the contribu26 . Nuc. 4−δ ∆L = MUV Oδ .Annu. The celebrated AdS/CFT correspondence (41) is a duality between type IIB string theory on Antide Sitter (AdS) (or asymptotically. On the gravity side of the correspondence. (16). The system of interest to us. approximately AdS) spaces and conformal ﬁeld theories (CFTs) on their boundaries.4. We will now present a very instructive dual description of this analysis. in supergravity. we were interested in nonnormalizable perturbations of the ﬁeld Φ− . nonnormalizable perturbations of supergravity ﬁelds correspond to perturbations of the CFT Lagrange density by irrelevant operators.4 3. the leading contributions to the inﬂaton potential induced by moduli stabilization and by the coupling of the throat region to the compact space.1 Gauge Theory Interpretation Gauge/Gravity Duality In the previous section we have shown how to compute. a D3brane moving in a warped throat solution of type IIB supergravity. Part. Eq. An important class of dual pairs consists of warped throat geometries and N = 1 supersymmetric ﬁeld theories. In AdS/CFT. (33) where MUV is the UV cutoﬀ of the gauge theory and Oδ is an operator of mass dimension δ ≥ 4. The gradual deviations from conformality manifest themselves on the gravity side as logarithmic corrections to the warp factor e2A . Sci. 2009 59 3. and then eventually conﬁnes in the infrared. therefore admits a dual description in terms of an approximate CFT.
with total dimension 27 . Part. Notice that Oδ is a composite operator. but will break supersymmetry spontaneously if X obtains an Fterm vacuum expectation value. with Oδ ≡ d4 θ X † X O∆ . while readers less interested in these details may skip to §3. (34) where X is a bulk moduli ﬁeld.4. Nuc. This term.2 Perturbations of the Gauge Theory The conﬁguration space of a probe D3brane in a KS throat corresponds to a portion of the Coulomb branch of the dual CFT (i. but do not reduce the total rank. Sci. 2009 59 tions to the eta parameter are now manifestly organized in terms of operator perturbations. In particular. We will now outline how the eta problem may be studied on the gauge theory side of the AdS/CFT correspondence. we are interested in the potential on the Coulomb branch of the CFT. cf. is allowed in a supersymmetric Lagrangian. precisely as in our general eﬀective ﬁeld theory treatment in §2. The leading such terms are of the form of Eq. (19). For a more complete description we refer the interested reader to Ref. containing both bulk and CFT ﬁelds. being an integral over superspace.) Thus. Rev.5 without loss of continuity. we are interested in perturbations that do not explicitly break supersymmetry and that incorporate the eﬀects of bulk CalabiYau ﬁelds. to understand the potential on this conﬁguration space. Such a potential can be generated if the CFT Lagrangian is perturbed by operators composed of the scalar ﬁelds that parameterize the Coulomb branch.e. by classifying irrelevant perturbations to the gauge theory. Eq. to a portion of the gauge theory moduli space in which the expectation values of D3brane collective coordinates reduce the rank of the nonabelian part of the gauge group.Annu. (3). 3. (33).
(36) 28 . the leading contributions to the D3brane potential involve either chiral operators whose dimensions are dictated by their U (1)R charges. j = 1. J2 . Chiral operators. Part. However. . These ﬁelds parameterize the Coulomb branch and. Rev. (35) In the following we will identify the CFT operators O∆ that correspond to perturbations of Φ− and hence induce a D3brane potential.1 . In fact. the operator dimensions ∆ will then dictate the structure of possible terms in the D3brane potential. (42). 2).4. contain the data specifying the D3brane position. this will not be true of the operators that induce subleading corrections. Bi (i. R). 2009 59 δ = 4 + ∆. Bi and their complex conjugates are labeled by their SU (2)A ×SU (2)B ×U (1)R quantum numbers (J1 . Using the AdS/CFT correspondence. Such a perturbation yields a term in the D3brane potential of the form (19) ∆V ∝ φ∆ . one has chiral operators of the form O∆ = Tr A(i1 B(j1 Ai2 Bj2 . and in either case the dimensions are protected and could be computed directly in the gauge theory. Sci. For J1 = J2 = R/2. (25).3 Classification of Operators To enumerate the lowestdimension contributing operators.c. Nuc.) The approximate CFT that is dual to the KS throat is an SU (N + M ) × SU (N ) gauge theory with bifundamental ﬁelds Ai . 3. the dimensions of these operators are given by Eq. The singletrace operators built out of the ﬁelds Ai . we must give a few more details of the structure of the gauge theory. or operators related by supersymmetry to the Noether currents of the global symmetries. this symmetry group corresponds to the isometries of the base manifold T 1. in particular. AiR ) BjR ) + c. (More background on the gauge theory dual to the KS throat may be found in Ref.Annu. .
Annu. R} = { 1 .c. Rev. . and in 2 generic situations they contribute the leading term in the inﬂaton potential via Eq. These nonchiral operators are in the same supermultiplets as SU (2) × SU (2) global symmetry currents. are ﬁxed by N = 1 superconformal invariance. J2 . Part. 1 ¯ ¯ √ Tr A1 A1 − A2 A2 . 0}: O2 = ¯ Tr A1 A2 . Nonchiral operators. These chiral operators have ∆ = 3/2. . J2 . The lowestdimension such operators are O3/2 = Tr (Ai Bj ) + c. 1}. . J2 . (35). Sci. Gravity Side ∆L ∆V = T3 Φ− ∇2 Φ − = 0 ∆V = − ∆ φ φUV ∆ φ φUV Gauge Theory Side ∆L = d4 θ X † X O∆ O∆ = Tr A(i1 B(j1 . 1. Nuc. 0. 0} operators made out of the ﬁelds Bj . R} = {1. R} = {0. 2 (38) and the corresponding {J1 . there are operators with {J1 . 2009 59 The dimensions of these operators. For example. ∆ = 2. 2 . and so their dimension is exactly 2. (37) 1 which have {J1 . Table 1: AdS/CFT Dictionary for Warped Dbrane Inﬂation. ∆ = 3R/2. ¯ Tr A2 A1 . AiR ) BjR ) ∆V = − ∆ φ φUV eigenvalue of Laplacian radial location maximal UV radius operator dimension energy scale UV cutoﬀ 29 . There are a number of operators which have the next lowest dimension.
the precise form of the potential is in principle fully speciﬁed by the remaining data of the compactiﬁcation. Sci. overcoming the eta problem becomes a detailed computational question. one can postulate that a ﬂat potential V (φ) arises after an approximate cancellation among dimensionsix Plancksuppressed corrections. This operator classiﬁcation precisely matches our supergravity analysis in §3. 2009 59 In the event that a discrete symmetry preserved by the full string compactiﬁcation forbids the chiral operators O3/2 .Annu. Rev. the leading contribution to the inﬂaton potential comes from O2 . 3. note that the contributing composite operators Oδ have dimensions 11/2 and 6. Finally. once φ is identiﬁed with a physical degree of freedom of a string compactiﬁcation. In eﬀective ﬁeld theory models of inﬂation one can of course always assume a solution to the eta problem by a cancellation of the contributing correction terms. to follow this path would be to abdicate the opportunity to use Plancksuppressed contributions as a (limited) window onto string theory. but will not contribute to our understanding of string theory. the latter may be interesting as a cosmological model. and in practice one can often classify corrections to the leadingorder potential. and the correspondence is summarized in Table 1.5 Summary and Perspective In §2 we explained how the eta problem is sensitive to dimensionsix Planck suppressed operators. Nuc. Moreover. in other words. in precisely the range that we argued on general grounds in §2 could yield orderunity contributions to the eta parameter. Part.) Thus. (Mixing conjecture into the analysis at this stage would eﬀectively transform a ‘stringderived’ scenario into a ‘stringinspired’ scenario.3. One can in principle compute the full potential from ﬁrst principles. 30 . In string theory models of inﬂation.
During inﬂation H is approximately constant. 4 LargeField Inﬂation The UV sensitivity of inﬂation described in §2 is vastly increased in the special case of largeﬁeld models. as we now review. 2009 59 In this section. Rev.e. we have enumerated the leading corrections for warped Dbrane inﬂation and showed that an accidental cancellation (or ﬁnetuning) allows small eta over a limited range of inﬂaton values. i. Nuc. while the second factor comes from the conversion of ﬂuctuations of the inﬂaton into ﬂuctuations of the spatial 3curvature. spatial 31 . Part. (40) represents the power spectrum of the inﬂaton ﬂuctuations (arising from quantum ﬂuctuations in de Sitter space). On scales smaller than the physical horizon. (39) where H is the Hubble expansion rate.Annu. This class is particularly interesting because it includes every inﬂationary model that yields a detectablylarge primordial gravitational wave signal (43). the background spacetime is nearly de Sitter and quantum ﬂuctuations in any light ﬁeld such as the metric scale with H. (40) The ﬁrst factor in Eq. scenarios in which the inﬂaton traverses a distance in ﬁeld space larger than the Planck mass. 4. The power spectrum of scalar ﬂuctuations is Ps = H 2π 2 H ˙ φ 2 .1 The Lyth Bound In singleﬁeld slowroll inﬂation the power spectrum of tensor ﬂuctuations is 2 Pt = 2 π H Mpl 2 . Sci. This gives a nontrivial existence proof for inﬂationary solutions in warped throat models with D3branes.
as in typical string compactiﬁcations many ﬁelds are not permitted such large excursions. This result implies two necessary conditions for largeﬁeld inﬂation: i) an obvious requirement is that large ﬁeld ranges are kinematically allowed.01 1/2 . is one class of examples where the kinematic requirement for large ﬁeld ranges cannot be fulﬁlled (44). Sci. 32 . as the following integral ∆φ = Mpl Nend Ncmb dN r(N ) 8 1/2 . In any model with r⋆ > 0. §3. one can derive the following important relation. (D3brane inﬂation in warped throats. originally due to Lyth (43): r⋆ ∆φ ≃ O(1) Mpl 0. i. Part. η ≪ 1 over a superPlanckian range ∆φ > Mpl . r⋆ ≡ r(Ncmb ). 2009 59 curvature ﬂuctuations relate to the observed ﬂuctuations in the matter density and in the CMB temperature. (42) Since the tensortoscalar ratio r(N ) is nearly constant during slowroll inﬂation.e. Nuc. (41) to write the ﬁeld variation between the end of inﬂation.2 and in string theory in §5.01 one must therefore ensure that ǫ. then we may use Eq. Ncmb .Annu. and the time when ﬂuctuations on CMB scales were generated. that the scalar ﬁeld space (in canonical units) has diameter > Mpl . The ratio between the tensor and scalar ﬂuctuation amplitudes is Pt r≡ =8 Ps ˙ φ HMpl 2 . (41) If we deﬁne dN ≡ Hdt. (43) where r⋆ is the value of the tensortoscalar ratio on CMB scales. Nend .) ii) the ﬂatness of the inﬂaton potential needs to be controlled dynamically over a superPlanckian ﬁeld range. This is nontrivial. Rev. We discuss this challenge in eﬀective ﬁeld theory in §4.
selfcoupling. . as always. Thus. but in inﬂation they play an important role. Part. such operators are negligible. etc. The particular operators which appear are determined. For nearly all questions in particle physics. whenever φ traverses a distance of order 33 . In particular. any other ﬁelds χ to which the inﬂaton couples experience changes in mass. 2009 59 4.1 No Shift Symmetry In the absence of any special symmetries. As an example. for couplings of order unity) introduces Plancksuppressed operators in the eﬀective action.e. Sci.. The physical interpretation of these terms is as follows: as the inﬂaton expectation value changes. (44) Unless the UV theory enjoys further symmetries. Rev. by the symmetries of the lowenergy action. Note that this applies not just to the light (m ≪ H) degrees of freedom. any ﬁeld coupled with at least gravitational strength to the inﬂaton experiences signiﬁcant changes when the inﬂaton undergoes a superPlanckian excursion. let us consider superPlanckian ﬁeld excursions in the context of Wilsonian eﬀective ﬁeld theory. but even to ﬁelds with masses near the Planck scale: integrating out Planckscale degrees of freedom generically (i. Nuc. . imposing only the symmetry φ → −φ on the inﬂaton leads to the following eﬀective action: ∞ 1 1 1 Leﬀ (φ) = − (∂φ)2 − m2 φ2 − λφ4 − λp φ4 + νp (∂φ)2 2 2 4 p=1 φ Mpl 2p + .Annu. one expects that the coeﬃcients λp and νp are of order unity. the potential in largeﬁeld inﬂation becomes sensitive to an inﬁnite series of Plancksuppressed operators.2.2 SuperPlanckian Fields and Flat Potentials To begin. These variations of the χ masses and couplings in turn feed back into changes of the inﬂaton potential and therefore threaten to spoil the delicate ﬂatness required for inﬂation. 4.
because we require that this symmetry protects the inﬂaton even from couplings to Planckscale degrees of freedom.2. the eﬀective Lagrangian receives substantial corrections from an inﬁnite series of higherdimension operators. §2. Sci. Let us remark that in eﬀective ﬁeld theory in general. which has been termed ‘functional ﬁnetuning’ (compare this to the eta problem. the potential should of course be approximately ﬂat over a superPlanckian range. UVcompletion of an assumed lowenergy symmetry is rarely 34 .) In the case with a shift symmetry. In order to have inﬂation. largeﬁeld inﬂation should be formulated in a theory that has access to information about approximate symmetries at the Planck scale. 46.2. However. Rev. the action of chaotic inﬂation (47) 1 Leﬀ (φ) = − (∂φ)2 − λp φp . 4. Hence. Nuc. the proposed symmetry of the lowenergy eﬀective action should admit a UVcompletion. 2009 59 Mpl in a direction that is not protected by a suitably powerful symmetry. φ → φ + const (45) protects the inﬂaton potential in a natural way. 2 (46) with small coeﬃcient λp is technically natural. Such a shift symmetry. Part. (Proposals using shift symmetries to protect the potential in string inﬂation include (45. If this is to arise by accident or by ﬁnetuning. 21).2 Shift Symmetry There is a sensible way to control this inﬁnite series of corrections: one can invoke an approximate symmetry that forbids the inﬂaton from coupling to other ﬁelds in any way that would spoil the structure of the inﬂaton potential. which only requires tuning of one mass parameter).Annu. it is essential that the symmetry should be approximately respected by the Planckscale theory – in other words. it requires a conspiracy among inﬁnitely many coeﬃcients.
Part. and it appears essential to develop a symmetry argument controlling or forbidding these terms.e.Annu. and indeed it has been conjectured that many eﬀective theories do not admit UVcompletion in string theory (48. it is important to verify that any proposed symmetry of Planckscale physics can be realized in string theory. As we have just discussed. Rev. an axion) is the inﬂaton. arising from couplings of the inﬂaton to degrees of freedom with masses near the Planck scale. in which a pseudoNambuGoldstone boson (i. The present situation is diﬀerent because we do not know whether all reasonable eﬀective actions can in fact arise as lowenergy limits of string theory. in string theory. we are therefore interested in ﬁnding. a conﬁguration that has both a large kinematic range. and a potential protected by a shift symmetry that is approximately preserved by the full string theory. Sci. 50). 5 Case Study of LargeField Inﬂation: Axion Monodromy We now turn to our second case study. (47) . Nuc. . At the perturbative level.. an example of largeﬁeld inﬂation in string theory. An inﬂuential proposal in this direction is Natural Inﬂation (51).. The nonperturbative eﬀects generate a periodic potential V (φ) = Λ4 φ 1 − cos 2 f 35 + . 49.. 2009 59 an urgent question. Direct enumeration and ﬁnetuning of such terms (as in the smallﬁeld example in §3) is manifestly impractical. To construct an inﬂationary model with detectable gravitational waves. the axion a enjoys a continuous shift symmetry a → a + const which is broken by nonperturbative eﬀects to a discrete symmetry a → a + 2π. Therefore. the particular challenge in these models is the need to control an inﬁnite series of contributions to the inﬂaton potential.
in any proposed eﬀective theory in which a superPlanckian ﬁeld range is protected by a shift symmetry. arising from pform gauge potentials integrated on pcycles of the compact space. In the absence of additional ingredients such as ﬂuxes and spaceﬁlling wrapped branes. computable limits of string theory (52. Part. The reader can easily verify from Eq. this potential can drive prolonged inﬂation. This leads to a periodic contribution to the axion potential whose periodicity we will now estimate.1 Axions in String Theory Axions from pForms Axions are plentiful in string compactiﬁca tions.Annu. Instanton eﬀects break this symmetry to a discrete subgroup. (2) that if the omitted terms are negligible and f > Mpl . As explained above.1 5.1. Rev. 53). 5. as well as axions ci = 2π Σi C arising from the RamondRamond (RR) twoform C.2 without loss of continuity. Readers less familiar with string compactiﬁcations can accept this assertion and skip to §5. and the omitted terms are higher harmonics. in type IIB string theory. Nuc. an important question. bi → bi + 2π (ci → ci + 2π). is whether this structure can be UVcompleted. the potential for these axions is classically ﬂat and has a continuous shift symmetry which originates in the gauge invariance of the tendimensional action. Sci. We will ﬁnd that the axion decay constants are smaller than Mpl in known. 36 . φ ≡ af . For example. We should therefore search in string theory for an axion with decay constant f > Mpl . 2009 59 where Λ is a dynamicallygenerated scale. there are axions bi = 2π Σi B aris ing from integrating the NeveuSchwarz (NS) twoform B over twocycles Σi . f is known as the axion decay constant.
the space of twoforms on the compact space X. (48) with x the fourdimensional spacetime coordinate. 37 . Suppose that the compactiﬁcation is isotropic. Qualita tively similar conclusions apply in much more general conﬁgurations (52. Part. Rev. we will illustrate this result in a simple example. 6(2π)2 L4 √ (52) In controlled compactiﬁcations we require L ≫ α′ . i ω j = α′ δi j . As this will be essential for our arguments. (49) ωi ∧ ⋆ ωj (50) and ⋆ is the sixdimensional Hodge star.2 Axion Decay Constants in String Theory Let ω i be a basis for Σi H 2 (X. By performing the integral over the internal space X and diagonalizing the ﬁeld space metric as γ ij → fi2 δij . Sci. 53). 53). with The NS twoform potential B may be expanded as B= 1 2π bi (x) ω i . which in this case descends from the tendimensional term 1 2 (2π)7 gs α′4 where γ ij ≡ 1 2 6(2π)9 gs α′4 X 1 d10 x dB2 2 ⊃ 1 2 √ d4 x −g γ ij (∂ µ bi ∂µ bj ) . The axion decay constant can be inferred from the normalization of the axion kinetic term.Annu. 2009 59 5. (50) that 2 f 2 ≈ Mpl α′2 . It is too early to draw universal conclusions. with typical lengthscale L and volume L6 . one can extract the axion decay constant fi .1. but a body of evidence suggests that the resulting axion periodicities are always smaller than Mpl in computable limits of string theory (52. Z). Nuc. Then using 2 α′ Mpl = 2 L6 2 (2π)7 gs α′3 (51) we ﬁnd from Eq. so that f ≪ Mpl .
but none of these has a suﬃciently large ﬁeld range.2 Axion Inﬂation in String Theory The above result would seem to imply that Natural Inﬂation from a single axion ﬁeld cannot be realized in known string compactiﬁcations: string theory provides many axions. Part. 5. but. upon transport around a closed loop in the (naive) conﬁguration space.) The idea of using monodromy to achieve controlled largeﬁeld inﬂation in string theory was ﬁrst proposed by Silverstein and Westphal (20). importantly. Rev.54).Annu. as the presence of a large number of light ﬁelds leads to a problematic renormalization of the Newton constant. This ‘Nﬂation’ proposal is a speciﬁc example of assisted inﬂation (55). A spiral staircase is a canonical example: the naive conﬁguration space is described by the angular coordinate. 57).2 Axion Monodromy We will instead describe an elementary mecha nism. but the system changes upon transport by 2π. In this section we will focus instead on the 38 . However. Nuc.2. there are at least two reasonable proposals to circumvent this obstacle. 2009 59 5. Although promising. we will ﬁnd that this simple model gives an excellent description of the potential in axion monodromy inﬂation. and hence to an eﬀectively reduced ﬁeld range. A system is said to undergo monodromy if. this scenario still awaits a proof of principle demonstration. who discussed a model involving a D4brane wound inside a nilmanifold. the system reaches a new conﬁguration. monodromy. For recent studies of Nﬂation see (56. one in which symmetry helps to protect the axion potential from corrections that would impede inﬂation.2. 5. (In fact. which allows inﬂation to persist through multiple circuits of a single periodic axion ﬁeld.1 Nflation The ﬁrst suggestion was that a collective excitation of many hundreds of axions could have an eﬀective ﬁeld range large enough for inﬂation (46. Sci.
a wrapped NS5brane produces monodromy for the axion c = 2π Σ C. then the axion b = 2π ΣB can exhibit monodromy in the potential energy. Consider type IIB string theory on a CalabiYau orientifold. 39 . as the shift symmetry of the axion action is broken by the presence of the wrapped brane. Eq. Part. Moreover. Monodromies of this sort are possible in a variety of compactiﬁcations. (21). The brane energy. the relevant potential comes from the DiracBornInfeld action for the wrapped Dbrane. when b ≫ ℓ2 . If in addition the compactiﬁcation includes a D5brane that wraps a suitable twocycle Σ and ﬁlls spacetime. Thus. In the D5brane case. a D5brane wrapping Σ carries a potential energy that is not a periodic function of the axion. we will suppose that the involution has ﬁxed points and ﬁxed fourcycles. i. (Similarly. the potential energy increases without bound as b increases.) In other words. Sci.e. known as O3planes and O7planes. where a monodromy arises in the fourdimensional potential energy upon transport around a circle in the ﬁeld space parameterized by an axion. a quotient of a CalabiYau manifold by a discrete symmetry that includes worldsheet orientation reversal and a geometric involution. (54). is clearly not invariant under the shift symmetry b → b + 2π.Annu. Σ (53) (54) M4 √ d4 x −g where ℓΣ is the size of the twocycle Σ in string units. respectively. 2009 59 subsequent axion monodromy proposal of Ref. Speciﬁcally. in fact. Rev. Nuc. but we will focus on a single concrete example. Σ the potential is asymptotically linear in the canonicallynormalized ﬁeld ϕb ∝ b. the DBI action leads directly to monodromy for b. although this is a symmetry of the corresponding compactiﬁcation without the wrapped D5brane. SDBI = = 1 (2π)5 gs α′3 1 6 g α′2 (2π) s d6 ξ M4 ×Σ det(G + B) (2π)2 ℓ4 + b2 .
e. both in this model and in string inﬂation more generally. upon which ΣB is taken to be large.3 Compactiﬁcation Considerations Having explained the essential idea of axion monodromy inﬂation. for example. see Eq. Nuc. The details of reheating depend strongly on the form of the couplings between the Standard Model degrees of freedom and the inﬂaton. (For representative work on reheating after string inﬂation. Sci. has little worldvolume ﬂux. nor do any of the closedstring moduli shift appreciably. see (58). For small axion vevs. and this is an important open question. beyond 40 . Inﬂation proceeds by the reduction of this vev. At this stage. on a stack of Dbranes elsewhere in the compactiﬁcation – then the hot Big Bang begins. until ﬁnally Σ B = 0 and the D5brane is nearly ‘empty’. An immediate concern is whether there are additional contributions to the potential. the axion b has a large initial vev.Annu. and the curvature of the potential becomes nonnegligible. If a suﬃcient fraction of this energy is eventually transmitted to visiblesector degrees of freedom – which may reside. During this process the D5brane does not move. 2009 59 Before we give more details of the eﬀects of compactiﬁcation on the axion potential. Rev.) 5. we must still ensure that the proposed inﬂationary mechanism is compatible with moduli stabilization and can be realized in a consistent compactiﬁcation. i. Couplings between the axion and other degrees of freedom. Part. In other words. the asymptotically linear potential we have described is inaccurate. the axion begins to oscillate around its origin. One begins with a D5brane wrapping a curve Σ. (54). drain energy from the inﬂaton oscillations. either closed string modes or open string modes. let us qualitatively summarize the inﬂationary dynamics in this model.
Annu. To show this. that could have important eﬀects during inﬂation. but does not constrain terms involving only the spacetime derivative ∂µ b. We will now explain this point. It is therefore essential to verify that the continuous shift symmetry which protects the inﬂaton potential is preserved to an appropriate degree by the stabilized compactiﬁcation. Rev. one expects that in the absence of a symmetry protecting the inﬂaton potential. but readers less interested in the details can skip to §5. 2009 59 the linear term identiﬁed above. As we have emphasized throughout this review. We now check this criterion in the example of interest by recalling the classic DineSeiberg treatment (59) of axion shift symmetries in string theory.4.1 Axion Shift Symmetries in String Theory We ﬁrst observe that a continuous shift symmetry b → b + const forbids all nonderivative terms in the eﬀective action for b. Part. i. ensuring that the shift symmetry is not spoiled can be quite subtle. and hence vanishes when the worldsheet has no boundary and wraps a topologically trivial cycle in spacetime. 5. Sci. The twoform B couples to the worldsheet as (60) i 2πα′ Σ d2 ξ ǫαβ ∂α X µ ∂β X ν Bµν (X) .3. they observed that the zeromomentum coupling of b (corresponding to nonderivative terms) is a total derivative on the worldsheet.e. (55) 41 . the eﬀective action for the axion b can only be a function of ∂µ b. the shift symmetry is unbroken to the extent that all nonderivative terms are constrained to vanish. Dine and Seiberg proved that to any order in perturbation theory (in the absence of Dbranes). Their argument proceeds as follows. generic corrections due to moduli stabilization will contribute ∆η ∼ O(1). Therefore. Nuc. For the special case of moduli stabilization in which nonperturbative eﬀects play a role. b has a shift symmetry.
in order to produce a monodromy in the potential.e. However.Annu. which we identiﬁed with the inﬂaton potential. if the string worldsheet has no boundary and is topologically trivial. Equivalently. whose coupling constant is the inverse spacetime curvature in units of α′ ). we have deliberately invoked D5branes. in a compactiﬁcation without Dbranes. then the above coupling is a total derivative on the worldsheet. If B is imagined to be a constant in spacetime. as worldsheet instantons.. provided that this potential.) Hence. B is not constant. the zeromomentum coupling of the axion b must vanish. as long as the worldsheet has no boundary. the shift symmetry of b is preserved to all orders in perturbation theory. but it is the coupling of the constant portion of B that governs zeromomentum terms in the fourdimensional eﬀective action.2 The Eta Problem for b In the present setting. the zeromomentum portion of the axion eﬀective action in spacetime arises from a totalderivative term on the string worldsheet. However. and the axion therefore cannot have any nonderivative couplings. is the 42 . the shift symmetry can be violated. Sci.3. Eq. i. the perturbation theory of the quantum ﬁeld theory living on the string worldsheet. Rev. 5. 2009 59 where X µ are the spacetime coordinates and ξ α are twodimensional string worldsheet coordinates. (In general backgrounds. while in a compactiﬁcation containing Dbranes. the axion has no nonderivative couplings to any order in sigmamodel perturbation theory (i. (55) is the pullback of B onto the worldsheet. Nuc. Part. Thus. on which the strings can break and end. because worldsheets wrapping nontrivial curves in the tendimensional spacetime contribute only nonperturbatively. closed string worldsheets can develop boundaries in the presence of Dbranes. Therefore.e.
the resulting structure is technically natural. Although a shift of b can be ¯ compensated in the K¨hler potential by a shift of T + T . Supposing for simplicity that there is only one K¨hler modulus. Rev. the continuous shift symmetry is broken by the nonperturbative superpotential term generated by Euclidean D3branes.Annu. Fur2 ¯ thermore.g. because Euclidean Dbranes are nonperturbative eﬀects and provide boundaries for string worldsheets. the superpotential is of the form a W = W0 + A exp(−2π T ). the K¨hler potential takes the form (61) K = −3Mpl ln(T + T − d b2 ). Clearly. K¨hler moduli a stabilization is accomplished. by the inclusion of nonperturbative eﬀects. Such eﬀects can circumvent the DineSeiberg argument given above. e. We will now sketch the speciﬁc diﬃculty presented by Euclidean D3branes. in a wellstudied class of models (27. a with d a constant depending on the intersection numbers of the compactiﬁcation and on the stabilized value of the dilaton. 43 . W0 and A are constants depending on the stabilized values of the complex structure moduli and of the dilaton. (21) for details. 2009 59 leading eﬀect breaking the shift symmetry. 28). T . the superpotential is a then not invariant. Part. Euclidean D3branes therefore make important contributions to the potential of b. Although this sounds promising. One can easily verify (21) that this is precisely analogous to the eta problem in D3brane inﬂation. where the exponential term is the Euclidean D3brane contribution. in the case of the b axion there is in fact an additional ingredient which also breaks the axion shift symmetry. and in fact generate an eta problem. referring the interested reader to Ref. At the energy scales in question. Nuc. from Euclidean D3branes (D3brane instantons). Sci.
be cause even a shift symmetry that was valid to all orders in perturbation theory has turned out to be inadequate to protect the inﬂaton potential! However. Eﬀective ﬁeld theory models of largeﬁeld inﬂation then require a shift symmetry to protect the ﬂatness of the potential over a superPlanckian range. we will now ﬁnd an even more potent symmetry in the case of the c axion. Σ (56) but does not couple directly to D3branes (or to D3brane instantons) that carry vanishing D1brane charge. Part. Thus. 5. We refer to Ref. via the electric coupling C. such an axion can enjoy the protection of a a shift symmetry over a superPlanckian range.3. These considerations suggest the following scenario. The corresponding inﬂationary scenario is natural in the technical sense. Although the NS axion b and the RR axion c have many shared features. Even in the presence of nonperturbative stabilization of the K¨hler moduli.4 Summary and Perspective In §4 we showed that an observable gravitational wave signal correlates with the inﬂaton ﬁeld moving over a superPlanckian distance during inﬂation. Nuc. if the moduli are stabilized exclusively by instantons to which c does not couple. (21) for a description of compactiﬁcations in which this mechanism is operative. It has therefore become an important question whether such shift symmetries arise in string theory and 44 .Annu. Sci. we consider a wrapped NS5brane that provides a potential for a c axion. a crucial distinction is that c couples to D1branes. Rev. Instead of a wrapped D5brane introducing a potential for a b axion. even nonperturbative moduli stabilization will not violate the shift symmetry of c. 2009 59 5.3 Flat Potential for c The situation may seem discouraging.
persisting after full moduli stabilization 45 . we argued that the ﬁrst examples of shift symmetries in string theory that protect the potential over a superPlanckian range are becoming available. 6 6. The shift symmetry. Nuc. and the problem of stabilizing the moduli had not been addressed.1 Outlook Theoretical Prospects As we hope this review has illustrated. 2009 59 can be used to realize largeﬁeld inﬂation. only a few proposals for connecting string theory to cosmology were available. a few mechanisms for inﬂation in string theory have been shown to be robust. only weakly broken by V . controls corrections ∆V within a fundamental domain.Annu. and hence an observable gravitational wave signal. A decade ago. Monodromy therefore eﬀectively reduces a largeﬁeld problem to a smallﬁeld problem (20). and the beststudied examples among these incorporate some information about moduli stabilization. in string theory. theoretical progress in recent years has been dramatic. We now have a wide array of inﬂationary models motivated by string theory. Moreover. Part. Rev. Sci. and ii) in combination with the shift symmetry it controls corrections to the potential over a superPlanckian range. In this section. Although more work is required to understand these models and the compactiﬁcations in which they arise. We explained the dual role of the monodromy: i) it results in a large kinematic ﬁeld range ∆φ > Mpl by allowing a small fundamental domain to be traversed repeatedly. and furthermore relates corrections in one fundamental domain to those in any other. monodromy appears to be a robust and rather promising mechanism for realizing largeﬁeld inﬂation.
it would be useful to understand how inﬂation can arise in more diverse string vacua.). inﬂation can be shown to be impossible in certain classes of type IIA compactiﬁcations (63. tadpole can 46 .g. and under mild assumptions. The inﬂationary models now available in string theory are subject to stringent theoretical constraints arising from consistency requirements (e. On the theoretical front. It would be surprising if it turned out that inﬂation is much more natural in one weaklycoupled limit of string theory than in the rest. It is of course diﬃcult to predict the direction of future theoretical progress. for the sensitivity of inﬂation to quantum gravity. Part.. it is safe to anticipate further gradual progress in moduli stabilization. entirely explicit models of inﬂation in such compactiﬁcations will undoubtedly follow. or toy models.Annu. At present. few successful models exist in Mtheory or in heterotic string theory (however. they have led to mechanisms for inﬂation that might seem unnatural in ﬁeld theory. 64. including the appearance of additional explicit examples with all moduli stabilized. Nuc. Sci. Clearly. and they have sharpened our understanding of the implications of a detection of primordial tensor modes. 2009 59 with all relevant corrections included.g. see e. and the present disparity can be attributed in part to the diﬀerences among the modulistabilizing tools presently available in the various limits. what has been accomplished? In our view the primary use of explicit models of inﬂation in string theory is as test cases. not least because unforeseen fundamental advances in string theory can be expected to enlarge the toolkit of inﬂationary modelbuilders. these models have underlined the importance of the eta problem in general ﬁeld theory realizations of inﬂation. Aside from demonstrating that inﬂation is possible in string theory. Rev. (62) et seq. 65). However. but are apparently natural in string theory.
Establishing such a proposition would require a far more comprehensive understanding of string compactiﬁcations than is available at present. This has revealed a strange universe ﬁlled with 73% dark energy. In order to describe this connection. 2009 59 cellation) and from the need for some degree of computability. 6. to predictions. one can hope that some constraints may remain. Some of these constraints will undoubtedly disappear as we learn to explore more general string compactiﬁcations. we will very brieﬂy review recent achievements and nearfuture prospects in observational cosmology. these limitations lead to correlations among the cosmological observables. 23% dark matter. and adiabatic (10).e. However. through the gravitational instability of small primordial ﬂuctuations. we now have a ﬁrm qualitative understanding of the formation of largescale structures. In addition. Sci. i. like the cosmic web of galaxies. so that the set of inﬂationary eﬀective actions derived from string theory would be a proper subset of the set of inﬂationary eﬀective actions in a general quantum ﬁeld theory. Nuc. nearly Gaussian. Part.2 Observational Prospects The theoretical aspects of inﬂation described in this review are interesting largely because they can be tested experimentally using present and future cosmological data. of the distribution of galaxies on the sky. 6.1 Present and Future Observations Observations of the cosmic microwave background anisotropies. and of the redshiftluminosity relations of type Ia supernovae have transformed cosmology into an exact science.2. and only 4% baryons. The perturbation spectrum that forms the seeds of these structures is found to be nearly (but not exactly) scaleinvariant.Annu. precisely as predicted by the simplest models of 47 . In turn. Rev.
A planned CMB polarization satellite (CMBPol (12. Nuc. Part. As a concrete example of the discriminatory power of tensor perturbations. CMB polarization experiments from the ground (e. in combination with smallscale CMB experiments (e. 6. while an upper bound r < 0. probe further details of the inﬂationary paradigm. The Planck satellite will measure the temperature anisotropies of the CMB with unprecedented accuracy over a large range of scales. any detection of primordial gravitational waves would exclude the warped D3brane inﬂation scenario of §3 (44). Clover (68). superPlanckian displacements are a key instance in which the inﬂaton eﬀective action is particularly sensitive to the physics of the Planck scale. 2009 59 inﬂation (11. EBEX (71) and SPIDER (72)) promise to provide the ﬁrst signiﬁcant constraints on inﬂationary tensor perturbations.g. Gaussianity and adiabaticity.18.2. QUIET (69). Future observations will dramatically extend our knowledge of the primordial ﬂuctuations. ACT (66) and SPT (67)) this will provide crucial information on deviations of the scalar spectrum from scaleinvariance.2 UV Physics in the Sky? The most dramatic conﬁrmation of inﬂa tion would come from a detection of Bmode polarization.Annu. 12).01.07 (or a detection with r ≫ 0. As we have argued in this review. and allow us to constrain or exclude a considerable fraction of the proposed scenarios for inﬂation. and BICEP (70)) and from balloons (e.73)) would be designed to be sensitive enough to detect Bmodes down to a tensortoscalar ratio of r = 0. which would establish the energy scale of inﬂation and would indicate that the inﬂaton traversed a superPlanckian distance.g.g. Sci. 48 . Rev. thereby including all models of largeﬁeld inﬂation (∆φ > Mpl ).07) would exclude the axion monodromy scenario of §5 (21).
Although we have focused in this review on the sensitivity of the inﬂaton potential to Planckscale physics. The primary motivations for these works are the sensitivity of inﬂationary eﬀective actions to the ultraviolet completion of gravity. 74). Rev. and string theory provides a promising framework for understanding the higherderivative interactions that can produce signiﬁcant nonGaussianity (39.3 Conclusions Recent work by many authors has led to the emergence of robust mechanisms for inﬂation in string theory. as the primordial scalar ﬂuctuations are predicted to be scaleinvariant. isocurvature ﬂuctuations or a large scaledependence (running) would therefore rule out singleﬁeld slowroll inﬂation. Nuc. the inﬂaton kinetic term is equally UVsensitive. Detecting cosmic superstrings via lensing or through their characteristic bursts of gravitational waves is an exciting prospect. A detection of nonGaussianity. 2009 59 A further opportunity arises because singleﬁeld slowroll inﬂation predicts null results for many cosmological observables. Finally. and the stability and phenomenological properties of the resulting cosmic string network are determined by the properties of the warped geometry.Annu. 49 . Inﬂationary effective actions that do allow for a signiﬁcant nonGaussianity. Cosmic strings are automatically produced at the end of braneantibrane inﬂation (75. 76). CMB temperature and polarization anisotropies induced by relic cosmic strings or other topological defects provide probes of the physics of the end of inﬂation or of the postinﬂationary era. nonadiabaticity or scaledependence often require higherderivative interactions and/or more than one light ﬁeld. Gaussian and adiabatic to a high degree. Sci. and such actions arise rather naturally in string theory. Part. 6.
Present observations and present theoretical considerations do not oblige us to expect an eventual positive result. Enrico Pajer. Sci. However. Hiranya Peiris. Part. for their hospitality while this review was being prepared. The research of DB is supported by the David and Lucile Packard Foundation 50 . Jesse Thaler. The theoretical community will eagerly await the coming generation of experimental results (12). Angel Uranga. Nuc. in ﬁeld theory models. DB thanks the theory group at Cornell and Trident Caf´ for their hospitality e while some of the work on this review was performed. Rev. Indeed. LM also thanks K. if we are fortunate enough to detect evidence for string theory in the sky. Narayan. 2009 59 and the prospect of empirical tests using precision cosmological data. many string inﬂation models are already signiﬁcantly constrained by the current data (77). A more diﬃcult question is how cosmological observations might possibly provide evidence in favor of a string theory model of inﬂation. this will most plausibly arise through a distinctive signature that is unnatural. David Poland. in the hope of extracting further clues about the physical properties of the early universe. as well as the organizers of the 2008 Indian Strings Meeting.Annu. We are grateful to Shamit Kachru. Igor Klebanov. Perhaps the best hope would be a striking correlation of many observables. Andrei Linde. Chennai. albeit presumably possible. Acknowledgements. String theory models of inﬂation have now achieved a reasonable level of theoretical control and are genuinely falsiﬁable by observational data. and Eva Silverstein for useful discussions and comments on the draft. and Marco Zagermann for discussions and correspondence on related topics. LM thanks the theory groups at Harvard and IMSc.
11. S. arXiv:0811. Mukhanov. F. F. 175 (1982). Rept. D 23.” in preparation. 15. Steinhardt and M. Lett. H. W. 493 (1985). Komatsu et al. Starobinsky. M. Linde. Lett. Riotto. 314. M. Kamionkowski. A. [WMAP Collaboration]. Hawking. C. 49. J.3919 [astroph]. 7368 (1997) [arXiv:astroph/9611125]. A. 295 (1982). Lett. Nuc. D 55. B 115. Rev. Phys. Bardeen. Zaldarriaga and U. 1 (1999) [arXiv:hepph/9807278].. Leitch. Astrophys.Annu. Halverson and W. Rev. D. Steinhardt. Sci. [arXiv:astroph/9609170]. Rev. D. D. 12. A. 33. 14. Phys. Sloan Foundation. Lyth and A. 1830 (1997) 51 . The research of LM is supported by NSF grant PHY0355005. 148. Turner. 16. 9. Baumann et al. 10. 1220 (1982). A. 2009 59 and the Alfred P. A. Kogut et al. Albrecht and P. Seljak. Y. J. Phys. Nature 420. D. JETP Lett. V. Baumann. J. 347 (1981). 41. Rev. Chibisov. Literature Cited 1. 161 (2003) Phys. Rev. 13. D 28. arXiv:0803. Part. D 55. Rev. Suppl. W. 8. 7. Pi. Stebbins. JETP Lett. Phys. J. B 108. Guth. A. 17. 1110 (1982). Kosowsky and A. Phys. Holzapfel. 6. 389 (1982). Pryke. “TASI Lectures on Inﬂation. H. Mukhanov and G. Phys. Rev. E. M. S. V. Kovac. L. M. Lett.0547 [astroph]. 679 (1983). A. B 117. A. J. Phys. V. E. 48. 532 (1981). Phys. [WMAP Collaboration]. P. Lett. Guth and S. 3. Phys. 2. H. 5. 772 (2002) [arXiv:astroph/0209478]. Carlstrom. N. E. J. 4.
Hirano and R. P. Dasgupta. I. Gaillard. K. Silverstein and A. C. arXiv:hepth/0105203. 126002 (2002) [arXiv:hepth/0203019]. S. H. F. Binetruy and M. 036 (2002) [arXiv:hepth/0112147]. 20. H. Dvali. 3548 (1995) [arXiv:hepth/9503114]. Hirano and R. H. H.3911 [astroph]. S. Kallosh. Shenker. J. S. K. Baumann et al. Shaﬁ and S. Dvali and S. Phys. Rabadan and F. Kors. B 450. D. S. Sci. Westphal.2811 [hepth]. arXiv:0808. Nucl. Part. W. JHEP 0203. Steinhardt. G. L. Rajesh and R. Banks. R. Phys. Rev. McAllister. D. D 65. H. 052 (2002) [arXiv:hepth/0111025]. J. B. Quevedo. Shiu and S. Burgess. arXiv:0808. Q. Phys. 235 (2002) [arXiv:hepth/0202124]. G. Martineau. Klebanov and L. GarciaBellido. Rev. C. 023507 (2002) [arXiv:hepth/0105032].0706 [hepth]. 21. H. E. JHEP 0112. Lett. 18. McAllister. D 34. Baumann. Tye. A. Phys. Herdeiro. Moore and P. Blumenhagen. T. Kallosh. C. 22. E. R. S. Phys. D 52. Lust and T. Nuc.. JHEP 0201. Silverstein and A. 3069 (1986). Alexander. J. 027 (2001) [arXiv:hepth/0110271]. Herdeiro. Westphal. [arXiv:hepph/9812483].3085 [hepth]. P. R. R. Rev. Phys. Tye. arXiv:0811. B 641. Phys. 52 . Berkooz. D. Zamora. Zhang. Dymarsky. 2009 59 [arXiv:astroph/0302213]. arXiv:0803. 421 (2001) 72 (1999) [arXiv:hepth/0106274]. M. Lett. Rev. G. 19. R. G. P.Annu. Ott. S. Rev. Solganik. D 65. G. Kachru. B 516.
H. J. Jones. DeWolfe.. K. 146 (2006) Phys. Quevedo. Henry Tye and J. B 584. S. Hsu. JHEP 0405. Tye. Lett. P. [arXiv:hepth/0508029]. Choudhury. 030 (2004) [arXiv:hepth/0405247]. Cremades and F. Rev. O. Panda. JHEP 0601. S. P. Firouzjahi. Sinha. 2009 59 N. T. Firouzjahi and S. Trivedi. JHEP 0409. 051 (2002) [arXiv:hepth/0203163]. Phys. Stoica and F. N. Burgess. L. Zagermann. D. H. M. arXiv:hepth/0606031. D. H. M. Avgoustidis. H. 015 (2006) [arXiv:hepth/0608082]. Conlon and F. Ghoshal. J. R. X. D. Rev. P. Xu. Burgess. 063 (2004) [arXiv:hepth/0406230]. Kallosh. Cline and H. P. JHEP 0411. BlancoPillado et al. K. C. D. 033 (2004) [arXiv:hepth/0403119]. J. Sarangi. 017 (2004) [arXiv:hepth/0403123]. Quevedo and A. JHEP 0510. A. D. H. JCAP 0307. D 72. Stoica and S. Linde and M. A. Quevedo. H. Cremades. Rev. Nuc. Jatkar and S. JHEP 0703. H. C. Cline. Stoica. S. JHEP 0408. J. 126004 (2005) Phys. Iizuka and S. Dasgupta and H. Quevedo. Kachru and H. JHEP 0207. JCAP 0611.Annu. M. Chen. D 70. 027 53 . Cline. P. 106 (2005) [arXiv:hepth/0505252]. P. Tye. J. Part. 043519 (2004) [arXiv:hepth/0509012]. J. 009 (2003) [arXiv:hepth/0305104]. 147 (2004) [arXiv:hepth/0312020]. Dasgupta. Verlinde. [arXiv:hepth/0403203]. Sci. J. F. H.
Kobayashi. Conlon. Kallosh. Rev. J.3961 [hepth]]. Leblond and S. P. A. S. 056 (2007) [arXiv:hepth/0612230]. Sami. arXiv:0808. Gong. Postma. Linde. M. Haack. Vaudrevange. J.1503 [hepth]. Krause. Nuc. M. arXiv:0811. Mukohyama and S.1170 [hepth]. 103 (2009) [arXiv:0804. Cline. D. S. JHEP 0704. Henry Tye. 2009 59 (2007) [arXiv:hepth/0610320]. J. S. B 806. A. 40. Phys. Panda and M. 011 (2008) [arXiv:0807. E. E. O. arXiv:hepth/0610221. eConf C040802. Rev. arXiv:0709. Shiu. H. Wesley. 123511 (2007) [arXiv:hepth/0612197].1614 [hepth].4649 [hepth].4941 [hepth]. Nucl. R. Rel. Grav. Kallosh. O. Cline. Steinhardt and D.0809 [hepth]]. Kumar and L. Prokushkin and P. Cline and M. Part. R. Phys. arXiv:0809. Kallosh.4285 [hepth]. M. M. D 75. N. L. H. Linde. T. P. arXiv:0708. Sci. Chen. Gen. J. Becker. Y. A. Quevedo. Cicoli. arXiv:0811. R. arXiv:hepth/0703278. M. L. Kofman. Biswas and J.2916 [hepth]]. Zagermann.1927 [hepth]]. L024 (2004) [arXiv:hepth/0503195]. Dutta. H. R. JCAP 0809. Burgess and F. Gong and G. 031 (2008) [arXiv:0802. R. 23.0691 [hepth]. A. C. J. JCAP 0804. 565 (2008) 54 . Barnaby. Bond. T. arXiv:hepth/0702059. S. C. Lust and M. B. A. McAllister and E. JHEP 0809.Annu. Quevedo. M. Ali. Shandera. P. Chingangbam. Pajer. Linde and F. Burgess. Kinoshita. arXiv:0812. M. L. arXiv:hepth/0612129. Leblond. J. Chen and J. 011 (2008) [arXiv:0806. J. Silverstein. Y. P.
Linde. B 307. J. Vilenkin and S. Rev.4257 [hepth]. D 27. Phys. R. Rev. B. D 78. Mezhlumian. GarciaBellido. JCAP 0601. Phys. 2009 59 [arXiv:0710. B. Phys. A. Linde and A. Linde. Rev. Vilenkin. Phys. D 78. Phys. Vilenkin.0005 [hepth]. JCAP 0706. De Simone. Phys. D. 395 (1986). Rev. Vanchurin and S. 55 .2798 [hepth]].3778 [hepth].2951 [hepth]]. Phys. B 175. Bousso. A. A. J. 97. 191302 (2006) [arXiv:hepth/0605263]. 730 (1994) [arXiv:astroph/9312039]. Lett. SchwartzPerlov. Underwood. A. Part. Winitzki. Lett. Nuc. arXiv:0808. arXiv:0812. arXiv:0808. A. D. Linde.3770 [hepth]. Rev. D 78. 846 (1995) [arXiv:grqc/9406010]. Rev. A.2117 [hepth]]. D. Bird. 123518 (2008) [arXiv:0810. A.1160 [hepth]]. Linde. Vilenkin. M. D. R. P.0710. S. Lett. 74. D. Linde and A. Salem and A. Garriga and A. 25. Mezhlumian. A. N.Annu. A. Bousso. D 50. A. Guth. Linde and D. Phys. A. Rev. Itzhaki and E. Winitzki. Noorbala. A. Phys. 1783 (1994) [arXiv:grqc/9306035]. A. 25 (1993) [arXiv:grqc/9304015]. Easther. H. S. D 49. Peiris and R. Rev. A. Winitzki. H. D. 2848 (1983). Yang. Phys. M. Kovetz. 017 (2007) [arXiv:0705. 24. V. Vilenkin. 083518 (2008) [arXiv:0807. D. Rev. Linde. 017 (2006) [arXiv:hepth/0509184]. JHEP 0710. JHEPA. arXiv:0809. 054 (2007) [arXiv:0708. Garriga. J. Freivogel and I. S.3745 [astroph]].1517 [grqc]]. 023509 (2008) [arXiv:0802.054. Lett. Linde. Sci. V.
Baumann. L. Phys. 123 (2000) [arXiv:hepth/0002159]. Strassler. Dall’Agata. Copeland. Nuc. Krause and E. Giddings. 32. A. Dymarsky. 29. McAllister and S. 024 (2008) [arXiv:0706. Kachru. A. 106006 (2002) [arXiv:hepth/0105097]. A. P. Trivedi. 37. R. A. A. D. JHEP 0008. JCAP 0503.1303 [hepth]. 052 (2000) [arXiv:hepth/0007191]. JHEP 0611. S. 733 (2007) [arXiv:hepth/0610102]. 066001 (2000) [arXiv:hepth/9905226]. Phys. McAllister and P. S. J. Kachru. 031 (2006) [arXiv:hepth/0607050]. Phys. McAllister and A. P. Klebanov and A. R. Nucl. D. D 68. 33. M. Phys. D 66. L. arXiv:0810. 31. Rev. D’Auria and S. J. 27. 141601 (2007) [arXiv:0705. Steinhardt. R.4682 [hepth]]. I. 36. R. H. Dymarsky. Wands. Firouzjahi and S. Klebanov. I. Part. Sci. Trivedi. 009 (2005) [arXiv:hepth/0501099]. Kallosh. J. Linde and S. Hoi and J. J. Klebanov. I. Stewart and D. B. 6410 (1994) [arXiv:astroph/9401011]. 35. J. 38. R. Klebanov and L. P. S. JCAP 0801. Rev. E. Lyth. D 49. Phys. Baumann. 79. Rev. Baumann.Annu. 99. 56 .0360 [hepth]]. Mod. D. L. Linde. D 61. Klebanov and M.3837 [hepth]]. Lett. 023 (2008) [arXiv:0705. Liddle. S. M. I. Phys. Douglas and S. A. Cline. M. I. Ferrara. Phys. M. JCAP 0807. 26. B 578. R. P. Rev. JCAP 0310. G. 013 (2003) [arXiv:hepth/0308055]. Pajer. 28. 30. A. R. E. Polchinski. Maldacena. Rev. Ceresole. A. 34. H. Murugan. Kallosh. Kachru. Tye. Kachru and J. D. R. R. Dymarsky. Rev. H. R. McAllister. Rev. D. Tseytlin. A. 046005 (2003) [arXiv:hepth/0301240]. 2009 59 L. Maldacena.
P. E. R. 045 (2005) [arXiv:hepth/0501184]. S. 52. 2. Lett. arXiv:hepth/0509212. N. ArkaniHamed. 45. G. Ooguri and C. A. Lett. J. Sci. V. J. McGreevy and J. Frieman and A. 2. M. 009 (2003) [arXiv:hepth/0311077]. S. 48. Lett. 65. JHEP 0404. 1861 (1997) [arXiv:hepph/9606387]. C. Phys. R. P. 123508 (2007) [arXiv:hepth/0610285]. Wacker. Kallosh and S. C. JCAP 0312. 57 . Ouyang. Phys. arXiv:hepth/0108101. Vafa. B 129. Rattazzi. Lett. Vafa. Prokushkin. Kallosh. 78. JHEP 0508. J. Klebanov and P. Witten. Dine. JCAP 0808. B 428. 51. Klebanov and A. Linde. 231 (1998) [Int. 103505 (2004) [arXiv:hepth/0310221]. S. J. 50. D 70. I. Phys. Theor. Kachru. Part. 1113 (1999)] [arXiv:hepth/9711200]. Nucl. Tong. Gubser. 105 (1998) [arXiv:hepth/9802109]. 001 (2003) [arXiv:hepth/0303252]. Dubovsky. M. Baumann and L. M. 014 (2006) [arXiv:hepth/0602178]. 3233 (1990). J. Phys. Dimopoulos. H. A. J. E. Silverstein and D. S. S. P. 042 (2004) [arXiv:hepth/0402047]. Rev. Rev. H. P. Phys. Maldacena. Gorbatov. Herzog. 47. Lyth. Theor. Polyakov. Phys. Nicolis and R. Nuc. Adams. A. 253 (1998) [arXiv:hepth/9802150]. T. Phys. 46. K. Phys. A. Chen. J. 42. McAllister. 38. 40. 2009 59 39. R. D. 44.Annu. Phys. JHEP 0610. 49. Olinto. 43. Hsu. X. Math. Rev. Hsu and R. Adv. D. Rev. 177 (1983). Rev. Banks. JCAP 0306. Phys. B 766. 21 (2007) [arXiv:hepth/0605264]. Adv. 41. 003 (2008) [arXiv:hepth/0507205]. Fox and E. Math. D 75. I. Freese. D. Theor.
D 77. Easther and L. Lorenz. D. D 72. Barnaby. T. Phys. Kofman. Cline and H. Rev. R. Sivanandam and M. D 77. 123522 (2008) [arXiv:0710. Stoica. M. Rev.Annu. H. Mazumdar and F. 126007 (2008) [arXiv:0710. Chen and S. Buchel and L. Rev. 57. Phys. N. C. McAllister. Green. Sci. Phys. JHEP 0810. Rev. W. Phys. P. Kofman and P. D 78. Phys. Rev. Phys.3883 [hepth]]. 083502 (2008) [arXiv:0801. Witten. 110 (2008) 58 . Kallosh. P. Brandenberger. Underwood. Davis. A. 018 (2006) [arXiv:hepth/0512102]. R. Phys. Cline. Myers. D 78. J. 55. Frey. R. Burgess and J.3429 [hepth]]. 026003 (2006) [arXiv:hepth/0508139]. D 76. 103504 (2007) [arXiv:0707. G. Dasgupta and A. X. Berndsen. A. A. JHEP 0606. A. Yi. 106001 (2005) [arXiv:hepth/0507257]. Chialva. Phys. D. E.1299 [hepth]].3832 [hepth]]. 2009 59 53. D 58. Rev. Nuc. Phys. Mazumdar and R. 007 (2005) [arXiv:hepth/0412040]. JCAP 0605. Rev. C. R. 54. Rev. 043501 (2008) [arXiv:0710. H. A. C. R. Grimm. N. K. Svrcek and E. A. Rev. Brandenberger. D 77. 56. 051 (2006) [arXiv:hepth/0605206]. Shiu and B. 086002 (2008) [arXiv:0804. Part. JCAP 0504. Soroush. L. 014 (2006) [arXiv:hepth/0508229]. Tye. 011 (2006) [arXiv:hepth/0602136]. Knauf and L. M. 58. H.3674 [hepth]]. C. R. Rev. JCAP 0606. Schunck. JHEP 0601. 061301 (1998) [arXiv:astroph/9804177].0584 [hepth]]. D 73. Liddle. R. A.
2657 [astroph]]. J. 50. P. D. 11 (2004) [arXiv:astroph/0411122]. C. Hertzberg. 68. 59. K¨rber. Phys. 095 (2007) [arXiv:0711. 64. 5543. A. 387 (2004) 59 . Samtleben [the QUIET Collaboration]. 67. Rev. 993 (2006) [arXiv:astroph/0610716]. arXiv:0710. Caviezel.3886 [hepth]. R. Soc. Tegmark. 47. Lett. Phys. Krause.Annu. Becker. 71. Kors. UK: Univ. 2625 (1986). Seiberg.3551 [hepth]. W. K. Taylor and M.3936 [hepth]]. D. [arXiv:hepth/0403067]. 72. New Astron.” Cambridge. Rev. S. Yoon et al. P. Oxley et al. [the SPIDER Collaboration]. C. “String Theory. Taylor [the Clover Collaboration]. 66. 5498. B 699. Grimm and J. Ruhl et al.0375 [astroNucl. (1998) 402 p 61. D. Zagermann. Flauger. Wrase and M. 1353 (2007) [arXiv:0802. JHEP 0712. 349 (2005) [arXiv:hepth/0501130]. Polchinski. arXiv:0812. 320 (2004) [arXiv:astroph/0501111]. Nucl. Rev. Dine and N. W. Nuovo Cim. SPIE Int. 69. S. Becker and A. 62. S. Sci. arXiv:astroph/0606278. B 715. 60. J. M. Vol. Pr. C.2512 [hepth]]. Kosowsky [the ACT Collaboration]. M. Kachru. [The SPT Collaboration]. K. Proc. T. Eng. MacTavish et al. 65. Proc. M. P. New Astron. Rev. o u arXiv:0812. Part. 939 (2003) [arXiv:astroph/0402234]. J. 122B. Paban. Robbins and T. E. W. 2009 59 [arXiv:0808. 70. Wrase. Eng. Louis. Soc. 63. 1: An Introduction to the Bosonic String. SPIE Int. A. Opt. Nuc. Phys. [the BICEP Collaboration]. T. L¨st. 57. [the EBEX Collaboration]. Opt.
R. 004 (2007) [arXiv:hepth/0702107]. S. C. Part. 74. J. 75. S. E. Lett. Phys. Rev. Tye. H. 77. R.1240 [astroph]]. Tong. E. Bean. 73. E. 185 (2002) [arXiv:hepth/0601099]. Rev. Henry Tye and J. V. Bean. H. J. Phys. Friedman and A. S. 007 (2006) Phys. Copeland. E. JCAP 0605. D 77. Phys. Xu. D. H. M.1812 [hepth]]. D 76. Bock et al. Tye.Annu. R. B.4207 [astroph]. Cooray. H. 60 . Myers and J. Shandera. V. 76. Sarangi and S. Nuc.. Comptes Rendus Physique 5 (2004) 1021. 2009 59 ph]. Rev. 103517 (2007) [arXiv:0706. Peiris and J. Silverstein and D. H. JCAP 0705. Xu. Chen. Rev. [arXiv:hepth/0204074]. D 70. 023527 (2008) [arXiv:0710. Alishahiha. S. Shandera and S. X. Sci. H. arXiv:0805. 123505 (2004) [arXiv:hepth/0404084]. Peiris. Polchinski. Baumann. B 536.