9701276v1 PDF

∗
Particle Physics Aspects of Modern Cosmology
Robert H. BRANDENBERGER
Department of Physics, Brown University
Providence, RI 02912, USA
arXiv:hep-ph/9701276v1 13 Jan 1997
Modern cosmology has created a tight link between particle physics

/ field theory and a wealth of new observational data on the structure
of the Universe. These lecture notes focus on some of the most impor-
tant aspects concerning the connection between theory and observations.
The lectures begin with an overview of some recent progress and prob-
lems in inflationary cosmology. In particular, a pedagogical discussion
of the theory of reheating is presented. The second topic is a survey
of the theory of cosmological perturbations, the cornerstone of modern
cosmology. The focus is on the gauge-invariant classical and quantum
theory of fluctuations. The third topic concerns the role of topological
defects in cosmology. Reviews of the cosmic string theory of galaxy for-
mation and of defect-mediated GUT and electroweak baryogenesis are
given.
BROWN-HET-1067
astro-ph/9701xxx
∗ Invited
lectures at the 15th Symposium on Theoretical Physics, Seoul National University,
August 22 - 28 1996; to be published in Field Theoretical Methods in Fundamental Physics,
ed. by Choonkyu Lee (Mineumsa Co. Ltd., Seoul, 1997).
1
I. INTRODUCTION AND OUTLINE
Most aspects of high energy physics beyond the standard model can only be tested
by going to energies far greater than those which present accelerators can provide.
Fortunately, the marriage between particle physics and cosmology has provided a way
to “experimentally” test the new theories of fundamental forces.
The key realization, discovered both in the context of the inflationary Universe
scenario [1] and of topological defects models [2] is that physics of the very early Uni-
verse may explain the origin of structure in the Universe. It now appears that a
rich set of data concerning the nonrandom distribution of matter on a wide range
of cosmological scales, and on the anisotropies in the cosmic microwave background
(CMB), may potentially be explained by high energy physics. In addition, studying
the consequences of particle physics models in the context of cosmology may lead
to severe constraints on new microscopic theories. Finally, particle physics and field
theory may provide explanations of some deep cosmological puzzles, e.g. why the
Universe at the present time appears so homogeneous, so close to being spatially flat,
and why it contains the observed small net baryon to entropy ratio.
In these lectures, I focus on three important aspects of modern cosmology. The
first concerns some fundamental problems of inflationary cosmology. In particular,
some recent progress in the understanding of “reheating” in inflation will be reviewed.
The second topic is the classical and quantum theory of cosmological perturba-
tions, the main tool of modern cosmology. A general relativistic and quantum me-
chanical analysis of the generation and evolution of linearized fluctuations is essential
in order to be able to accurately calculate the amplitude of density perturbations and
CMB anisotropies.
As a third topic, I discuss the role of topological defects in baryogenesis and as
possible sees for cosmological structure formation.
The specific outline is as follows:
1. Introduction and Outline
2. Lecture 1: Inflationary Universe: Progress and Problems

2.A Problems of Standard Cosmology
2.B Inflationary Universe Scenario
2.C Problems of Inflation
2.D Inflation and Nonsingular Cosmology
2.E Reheating in Inflationary Cosmology
2.F Summary
3. Lecture 2: Classical and Quantum Theory of Cosmological Perturba-

tions
3.A Basic Issues
3.B Newtonian Theory
2
3.C Relativistic Theory: Classical Analysis
3.D Relativistic Theory: Quantum Analysis
3.E Summary
4. Lecture 3: Topological Defects, Structure Formation and Baryogen-
esis
4.A Quantifying Data on Large-Scale Structure
4.B Topological Defects
4.C Formation of Defects in Cosmological Phase Transitions
4.D Evolution of Strings and Scaling
4.E Cosmic Strings and Structure Formation
4.F Specific Predictions
4.G Principles of Baryogenesis
4.H GUT Baryogenesis and Topological Defects
4.I Electroweak Baryogenesis and Topological Defects
4.J Summary
Unless otherwise specified, units in which h̄ = c = kB = 1 will be used. Distances
are expressed in Mpc (1pc ≃ 3.06 light years). Following the usual convention, h
indicates the expansion rate of the Universe in units of 100 km s−1 Mpc−1 , Ω = ρ/ρc
is the ratio of the energy density ρ to the critical density ρc (the density which yields
a spatially flat Universe), G is Newton’s constant and mpl is the Planck mass.
II. INFLATIONARY UNIVERSE: PROGRESS AND PROBLEMS
The hypothesis that the Universe underwent a period of exponential expansion at

very early times has become the most popular theory of the early Universe. Not only
does it solve some of the problems of standard big bang cosmology, but it also provides
a causal theory for the origin of inhomogeneities in the Universe which is predictive
and in reasonable agreement with current observational results. Nevertheless, there
are several problems of principle which merit further study.
A. Problems of Standard Cosmology
The standard big bang cosmology rests on three theoretical pillars: the cosmolog-
ical principle, Einstein’s general theory of relativity and a perfect fluid description of
matter.
The cosmological principle states that on large distance scales the Universe is
homogeneous. This implies that the metric of space-time can be written in Friedmann-
Robertson-Walker (FRW) form:
dr 2
" #
2 2
ds = a(t) 2
+ r 2 (dϑ2 + sin2 ϑdϕ2 ) , (1)
1 − kr
3
where the constant k determines the topology of the spatial sections. In the following,
we shall usually set k = 0, i.e. consider a spatially closed Universe. In this case, we
can without loss of generality take the scale factor a(t) to be equal to 1 at the present
time t0 , i.e. a(t0 ) = 1. The coordinates r, ϑ and ϕ are comoving spherical coordinates.
World lines with constant comoving coordinates are geodesics corresponding to par-
ticles at rest. If the Universe is expanding, i.e. a(t) is increasing, then the physical
distance ∆xp (t) between two points at rest with fixed comoving distance ∆xc grows:
∆xp = a(t)∆xc . (2)
The dynamics of an expanding Universe is determined by the Einstein equations,

which relate the expansion rate to the matter content, specifically to the energy
density ρ and pressure p. For a homogeneous and isotropic Universe, they reduce to
the Friedmann-Robertston-Walker (FRW) equations
2
ȧ k 8πG
− 2
= ρ (3)
a a 3
ä 4πG
=− (ρ + 3p) . (4)
a 3
These equations can be combined to yield the continuity equation (with Hubble con-
stant H = ȧ/a)
ρ̇ = −3H(ρ + p) . (5)
The third key assumption of standard cosmology is that matter is described by

an ideal gas with an equation of state
p = wρ . (6)
For cold matter, pressure is negligible and hence w = 0. From (5) it follows that
ρm (t) ∼ a−3 (t) , (7)
where ρm is the energy density in cold matter. For radiation we have w = 1/3 and
hence it follows from (5) that
ρr (t) ∼ a−4 (t) , (8)
ρr (t) being the energy density in radiation.

The three classic observational pillars of standard cosmology are Hubble’s law, the
existence and black body nature of the nearly isotropic CMB, and the abundances
of light elements (nucleosynthesis). These successes are discussed in detail in many
textbooks on cosmology, and will therefore not be reviewed here.
4
It is, however, important to recall two important aspects concerning the thermal
history of the early Universe. Since the energy density in radiation redshifts faster
than the matter energy density, it follows by working backwards in time from the
present data that although the energy density of the Universe is now mostly in cold
matter, it was initially dominated by radiation. The transition occurred at a time
denoted by teq , the “time of equal matter and radiation”. As will be discussed in
Section 3, teq is the time when perturbations can start to grow by gravitational
clustering. The second important time is trec , the “time of recombination” when
photons fell out of equilibrium (since ions and electrons had by then combined to
form electrically neutral atoms). The photons of the CMB have travelled without
scattering from trec . Their spatial distribution is predicted to be a black body since the
cosmological redshift preserves the black body nature of the initial spectrum (simply
redshifting the temperature) which was in turn determined by thermal equilibrium.
CMB anisotropies probe the density fluctuations at trec . Note that for the usual
values of the cosmological parameters, teq < trec .
Standard Big Bang cosmology is faced with several important problems. Only one
of these, the age problem, is a potential conflict with observations. The others which
I will focus on here – the homogeneity, flatness and formation of structure problems
(see e.g. [1]) – are questions which have no answer within the theory and are therefore
the main motivation for the new cosmological models which will be discussed in later
sections of these lecture notes.
The horizon problem is illustrated in Fig. 1. As is sketched, the comoving region
ℓp (trec ) over which the CMB is observed to be homogeneous to better than one part in
104 is much larger than the comoving forward light cone ℓf (trec ) at trec , which is the
maximal distance over which microphysical forces could have caused the homogeneity:
Zt0 1/3 !
trec

−1
ℓp (trec ) = dt a (t) ≃ 3 t0 1− (9)
t0
trec
trec
Z
2/3
ℓf (trec ) = dt a−1 (t) ≃ 3 t0 t1/3
rec . (10)
0
From the above equations it is obvious that ℓp (trec ) ≫ ℓf (trec ). Hence, standard
cosmology cannot explain the observed isotropy of the CMB.
5
FIG. 1. A space-time diagram (physical distance xp versus time t) illustrating the
homogeneity problem: the past light cone ℓp (t) at the time trec of last scattering is much
larger than the forward light cone ℓf (t) at trec .
In standard cosmology and in an expanding Universe, Ω = 1 is an unstable fixed

point. This can be seen as follows. For a spatially flat Universe (Ω = 1)
8πG
H2 = ρc , (11)
3
whereas for a nonflat Universe
8πG
H2 + ε T 2 = ρ, (12)
3
with
k
ε= . (13)
(aT )2
The quantity ε is proportional to s−2/3 , where s is the entropy density. Hence, in

standard cosmology, ε is constant. Combining (11) and (12) gives
ρ − ρc 3 εT 2
= ∼ T −2 . (14)
ρc 8πG ρc
Thus, as the temperature decreases, Ω − 1 increases. In fact, in order to explain the
present small value of Ω − 1 ∼ O(1), the initial energy density had to be extremely
close to critical density. For example, at T = 1015 GeV, (14) implies
ρ − ρc
∼ 10−50 . (15)
ρc
What is the origin of these fine tuned initial conditions? This is the flatness problem
of standard cosmology.
6
The third of the classic problems of standard cosmological model is the “forma-
tion of structure problem.” Observations indicate that galaxies and even clusters of
galaxies have nonrandom correlations on scales larger than 50 Mpc (see e.g. [3,4]).
This scale is comparable to the comoving horizon at teq . Thus, if the initial density
perturbations were produced much before teq , the correlations cannot be explained by
a causal mechanism. Gravity alone is, in general, too weak to build up correlations
on the scale of clusters after teq (see, however, the explosion scenario of [5]). Hence,
the two questions of what generates the primordial density perturbations and what
causes the observed correlations, do not have an answer in the context of standard
cosmology. This problem is illustrated by Fig. 2.
FIG. 2. A sketch (conformal separation vs. time) of the formation of structure problem:
the comoving separation dc between two clusters is larger than the forward light cone at
time teq .
There are other serious problems of standard cosmology, e.g. the age and the
cosmological constant problems. However, to date modern cosmology does not shed
any light on these problems, and I will therefore not address them here.
B. Inflationary Universe Scenario
The idea of inflation [1] is very simple (for some early reviews of inflation see e.g.
[6–9]). We assume there is a time interval beginning at ti and ending at tR (the
“reheating time”) during which the Universe is exponentially expanding, i.e.,
a(t) ∼ eHt , tǫ [ti , tR ] (16)
with constant Hubble expansion parameter H. Such a period is called “de Sitter”
or “inflationary.” The success of Big Bang nucleosynthesis sets an upper limit to the
time of reheating:
tR ≪ tN S , (17)
tN S being the time of nucleosynthesis.
7
FIG. 3. The phases of an inflationary Universe. The times ti and tR denote the beginning
and end of inflation, respectively. In some models of inflation, there is no initial radiation
dominated FRW period. Rather, the classical space-time emerges directly in an inflationary
state from some initial quantum gravity state.
The phases of an inflationary Universe are sketched in Fig. 3. Before the onset
of inflation there are no constraints on the state of the Universe. In some models a
classical space-time emerges immediately in an inflationary state, in others there is
an initial radiation dominated FRW period. Our sketch applies to the second case.
After tR , the Universe is very hot and dense, and the subsequent evolution is as
in standard cosmology. During the inflationary phase, the number density of any
particles initially in thermal equilibrium at t = ti decays exponentially. Hence, the
matter temperature Tm (t) also decays exponentially. At t = tR , all of the energy
which is responsible for inflation (see later) is released as thermal energy. This is a
nonadiabatic process during which the entropy increases by a large factor.
Fig. 4 is a sketch of how a period of inflation can solve the homogeneity problem.
∆t = tR −ti is the period of inflation. During inflation, the forward light cone increases
exponentially compared to a model without inflation, whereas the past light cone is
not affected for t ≥ tR . Hence, provided ∆t is sufficiently large, ℓf (tR ) will be greater
than ℓp (tR ).
8
FIG. 4. Sketch (physical coordinates vs. time) of the solution of the homogeneity
problem. During inflation, the forward light cone lf (t) is expanded exponentially when
measured in physical coordinates. Hence, it does not require many e-foldings of inflation in
order that lf (t) becomes larger than the past light cone at the time of last scattering. The
dashed line is the forward light cone without inflation.
Inflation also can solve the flatness problem [10,1] The key point is that the entropy
density s is no longer constant. As will be explained later, the temperatures at ti
and tR are essentially equal. Hence, the entropy increases during inflation by a factor
exp(3H∆t). Thus, ǫ decreases by a factor of exp(−2H∆t). Hence, (ρ − ρc )/ρ can
be of order 1 both at ti and at the present time. In fact, if inflation occurs at all,
then rather generically, the theory predicts that at the present time Ω = 1 to a high
accuracy (now Ω < 1 requires special initial conditions or rather special models [11] ).
Most importantly, inflation provides a mechanism which in a causal way generates
the primordial perturbations required for galaxies, clusters and even larger objects.
In inflationary Universe models, the Hubble radius (“apparent” horizon), 3t, and the
“actual” horizon (the forward light cone) do not coincide at late times. Provided that
the duration of inflation is sufficiently long, then (as sketched in Fig. 5) all scales
within our apparent horizon were inside the actual horizon since ti . Thus, it is in
principle possible to have a casual generation mechanism for perturbations [12–15] .
The generation of perturbations is supposed to be due to a causal microphysical
process. Such processes can only act coherently on length scales smaller than the
Hubble radius ℓH (t) where
ℓH (t) = H −1 (t) . (18)
A heuristic way to understand the meaning of ℓH (t) is to realize that it is the distance
9
which light (and hence the maximal distance any causal effects) can propagate in one
expansion time.
FIG. 5. A sketch (physical coordinates vs. time) of the solution of the formation of
structure problem. Provided that the period of inflation is sufficiently long, the separation
dc between two galaxy clusters is at all times smaller than the forward light cone. The
dashed line indicates the Hubble radius. Note that dc starts out smaller than the Hubble
radius, crosses it during the de Sitter period, and then reenters it at late times.
As will be discussed in Chapter 4, the density perturbations produced during

inflation are due to quantum fluctuations in the matter and gravitational fields [13,14] .
The amplitude of these inhomogeneities corresponds to a temperature TH
TH ∼ H , (19)
the Hawking temperature of the de Sitter phase. This implies that at all times t during
inflation, perturbations with a fixed physical wavelength ∼ H −1 will be produced.
Subsequently, the length of the waves is stretched with the expansion of space, and
soon becomes larger than the Hubble radius. The phases of the inhomogeneities
are random. Thus, the inflationary Universe scenario predicts perturbations on all
scales ranging from the comoving Hubble radius at the beginning of inflation to
the corresponding quantity at the time of reheating. In particular, provided that
inflation lasts sufficiently long, perturbations on scales of galaxies and beyond will be
generated. Note, however, that it is very dangerous to interpret de Sitter Hawking
10
radiation as thermal radiation. In fact, the equation of state of this “radiation” is
not thermal [16] .
Obviously, the key question is how to obtain inflation. From the FRW equations,
it follows that in order to get exponential increase of the scale factor, the equation of
state of matter must be
p = −ρ (20)
This is where the connection with particle physics comes in. The energy density and
pressure of a scalar quantum field ϕ are given by
1 2 1
ρ(ϕ) = ϕ̇ + (∇ϕ)2 + V (ϕ) (21)
2 2
1 2 1
p(ϕ) = ϕ̇ − (∇ϕ)2 − V (ϕ) . (22)
2 6
Thus, provided that at some initial time ti
ϕ̇(x, ti ) = ∇ϕ(xi ti ) = 0 (23)
and
V (ϕ(xi , ti )) > 0 , (24)
the equation of state of matter will be (20).

The next question is how to realize the required initial conditions (23) and to
maintain the key constraints
ϕ̇2 ≪ V (ϕ) , (∇ϕ)2 ≪ V (ϕ) (25)
for sufficiently long. Various ways of realizing these conditions were put forward, and
they gave rise to different models of inflation. I will focus on “old inflation,” “new
inflation”” and “chaotic inflation.” There are many other attempts at producing an
inflationary scenario, but there is as of now no convincing realization.
Old Inflation
The old inflationary Universe model [1,17] is based on a scalar field theory which
undergoes a first order phase transition. As a toy model, consider a scalar field
theory with the potential V (ϕ) of Figure 6. This potential has a metastable “false”
vacuum at ϕ = 0, whereas the lowest energy state (the “true” vacuum) is ϕ = a.
Finite temperature effects [18] lead to extra terms in the finite temperature effective
potential which are proportional to ϕ2 T 2 (the resulting finite temperature effective
potential is also depicted in Figure 6). Thus, at high temperatures, the energetically
preferred state is the false vacuum state. Note that this is only true if ϕ is in thermal
equilibrium with the other fields in the system.
11
FIG. 6. The finite temperature effective potential in a theory with a first order phase
transition.
For fairly general initial conditions, ϕ(x) is trapped in the metastable state ϕ = 0
as the Universe cools below the critical temperature Tc . As the Universe expands
further, all contributions to the energy-momentum tensor Tµν except for the contri-
bution
Tµν ∼ V (ϕ)gµν (26)
redshift. Hence, provided that the potential V (ϕ) is shifted upwards such that V (a) =
0, then the equation of state in the false vacuum approaches p = −ρ, and inflation
sets in. After a period Γ−1 , where Γ is the tunnelling rate, bubbles of ϕ = a begin
to nucleate [19] in a sea of false vacuum ϕ = 0. Inflation lasts until the false vacuum
decays. During inflation, the Hubble constant is given by
8πG
H2 = V (0) . (27)
3
Note that the condition V (a) = 0, which looks rather unnatural, is required to avoid
a large cosmological constant today (none of the present inflationary Universe models
manages to circumvent or solve the cosmological constant problem).
It was immediately realized that old inflation has a serious “graceful exit”
problem [1,20] . The bubbles nucleate after inflation with radius r ≪ 2tR and would
today be much smaller than our apparent horizon. Thus, unless bubbles percolate,
the model predicts extremely large inhomogeneities inside the Hubble radius, in con-
tradiction with the observed isotropy of the microwave background radiation.
For bubbles to percolate, a sufficiently large number must be produced so that
they collide and homogenize over a scale larger than the present Hubble radius. How-
ever, with exponential expansion, the volume between bubbles expands exponentially
whereas the volume inside bubbles expands only with a low power. This prevents
percolation.
New Inflation
12
Because of the graceful exit problem, old inflation never was considered to be a
viable cosmological model. However, soon after the seminal paper by Guth, Linde [21]
and independently Albrecht and Steinhardt [22] put forwards a modified scenario, the
New Inflationary Universe.
The starting point is a scalar field theory with a double well potential which
undergoes a second order phase transition (Fig. 7). V (ϕ) is symmetric and ϕ = 0 is
a local maximum of the zero temperature potential. Once again, it was argued that
finite temperature effects confine ϕ(x) to values near ϕ = 0 at temperatures T ≥ Tc .
For T < Tc , thermal fluctuations trigger the instability of ϕ(x) = 0 and ϕ(x) evolves
towards either of the global minima at ϕ = ±σ by the classical equation of motion
ϕ̈ + 3H ϕ̇ − a−2 ▽2 ϕ = −V ′ (ϕ) . (28)
Within a fluctuation region, ϕ(x) will be homogeneous. In such a region, we can

neglect the spatial gradient terms in Eq. (28). Then, from (21) and (22) we can read
off the induced equation of state. The condition for inflation is
ϕ̇2 ≪ V (ϕ) , (29)
i.e. slow rolling. Often, the “slow rolling” approximation is made to find solutions
of (28). This consists of dropping the ϕ̈ term.
FIG. 7. The finite temperature effective potential in a theory with a second order phase
transition.
There is no graceful exit problem in the new inflationary Universe. Since the
fluctuation domains are established before the onset of inflation, any boundary walls
will be inflated outside the present Hubble radius.
Let us, for the moment, return to the general features of the new inflationary
Universe scenario. At the time tc of the phase transition, ϕ(t) will start to move
from near ϕ = 0 towards either ±σ as described by the classical equation of motion,
i.e. (28). At or soon after tc , the energy-momentum tensor of the Universe will start
to be dominated by V (ϕ), and inflation will commence. ti shall denote the time of the
13
onset of inflation. Eventually, φ(t) will reach large values for which nonlinear effects
become important. The time at which this occurs is tB . For t > tB , ϕ(t) rapidly
accelerates, reaches ±σ, overshoots and starts oscillating about the global minimum
of V (ϕ). The amplitude of this oscillation is damped by the expansion of the Universe
and (predominantly) by the coupling of ϕ to other fields. At time tR , the energy in ϕ
drops below the energy of the thermal bath of particles produced during the period
of oscillation.
The evolution of ϕ(t) is sketched in Fig. 8. The time period between tB and tR
is called the reheating period and is usually short compared to the Hubble expansion
time. For t > tR , the Universe is again radiation dominated.
FIG. 8. Evolution of ϕ(t) and T (t) in the new inflationary Universe.
In order to obtain inflation, the potential V (ϕ) must be very flat near the false
vacuum at ϕ = 0. This can only be the case if all of the coupling constants appearing
in the potential are small. However, this implies that the ϕ cannot be in thermal
equilibrium at early times, which would be required to localize ϕ in the false vacuum.
In the absence of thermal equilibrium, the initial conditions for ϕ are only constrained
by requiring that the total energy density in ϕ not exceed the total energy density
of the Universe. Most of the phase space of these initial conditions lies at values of
|ϕ| >> σ. This leads to the “chaotic” inflation scenario [23] .
Chaotic Inflation
Consider a region in space where at the initial time ϕ(x) is very large, homo-
geneous and static. In this case, the energy-momentum tensor will be immediately
dominated by the large potential energy term and induce an equation of state p ≃ −ρ
which leads to inflation. Due to the large Hubble damping term in the scalar field
equation of motion, ϕ(x) will only roll very slowly towards ϕ = 0. The kinetic energy
contribution to Tµν will remain small, the spatial gradient contribution will be expo-
nentially suppressed due to the expansion of the Universe, and thus inflation persists.
14
Note that in contrast to old and new inflation, no initial thermal bath is required.
Note also that the precise form of V (ϕ) is irrelevant to the mechanism. In particular,
V (ϕ) need not be a double well potential. This is a significant advantage, since for
scalar fields other than Higgs fields used for spontaneous symmetry breaking, there
is no particle physics motivation for assuming a double well potential, and since the
inflaton (the field which gives rise to inflation) cannot be a conventional Higgs field
due to the severe fine tuning constraints.
The field and temperature evolution in a chaotic inflation model is similar to what
is depicted in Figure 8, except that ϕ is rolling towards the true vacuum at ϕ = σ
from the direction of large field values.
Chaotic inflation is a much more radical departure from standard cosmology than
old and new inflation. In the latter, the inflationary phase can be viewed as a short
phase of exponential expansion bounded at both ends by phases of radiation dom-
ination. In chaotic inflation, a piece of the Universe emerges with an inflationary
equation of state immediately after the quantum gravity (or string) epoch.
The chaotic inflationary Universe scenario has been developed in great detail (see
e.g. [24] for a recent review). One important addition is the inclusion of stochastic
noise [25] in the equation of motion for ϕ in order to take into account the effects of
quantum fluctuations. It can in fact be shown that for sufficiently large values of
|ϕ|, the stochastic force terms are more important than the classical relaxation force
V ′ (ϕ). There is equal probability for the quantum fluctuations to lead to an increase
or decrease of |ϕ|. Hence, in a substantial fraction of comoving volume, the field
ϕ will climb up the potential. This leads to the conclusion that chaotic inflation is
eternal. At all times, a large fraction of the physical space will be inflating. Another
consequence of including stochastic terms is that on large scales (much larger than
the present Hubble radius), the Universe will look extremely inhomogeneous.
C. Problems of Inflationary Cosmology
In spite of its great success at resolving some of the problems of standard cos-
mology and of providing a causal, predictive theory of structure formation, there are
several important unresolved conceptual problems in inflationary cosmology. I will
focus on three of these problems, the cosmological constant mystery, the fluctuation
problem, and the dynamics of reheating.
Cosmological Constant Problem
Since the cosmological constant acts as an effective energy density, its value is
bounded from above by the present energy density of the Universe. In Planck units,
the constraint on the effective cosmological constant Λef f is (see e.g. [26])
Λef f
4
≤ 10−122 . (30)
mpl
15
This constraint applies both to the bare cosmological constant and to any matter
contribution which acts as an effective cosmological constant.
The true vacuum value of the potential V (ϕ) acts as an effective cosmological
constant. Its value is not constrained by any particle physics requirements (in the
absence of special symmetries). The cosmological constant problem is thus even more
accute in inflationary cosmology than it usually is. The same unknown mechanism
which must act to shift the potential (see Figure 6) such that inflation occurs in the
false vacuum must also adjust the potential to vanish in the true vacuum.
Supersymmetric theories may provide a resolution of this problem, since unbro-
ken supersymmetry forces V (ϕ) = 0 in the supersymmetric vacuum. However, su-
persymmetry breaking will induce a nonvanishing V (ϕ) in the true vacuum after
supersymmetry breaking.
We may therefore be forced to look for realizations of inflation which do not make
use of scalar fields. There are several possibilities. It is possible to obtain inflation in
higher derivative gravity theories. In fact, the first model with exponential expansion
of the Universe was obtained [27] in an R2 gravity theory. The extra degrees of freedom
associated with the higher derivative terms act as scalar fields with a potential which
automatically vanishes in the true vacuum. For some recent work on higher derivative
gravity inflation see also [28].
Another way to obtain inflation is by making use of condensates (see [29] and
[30] for different approaches to this problem). An additional motivation for following
this route to inflation is that the symmetry breaking mechanisms observed in nature
(in condensed matter systems) are induced by the formation of condensates such as
Cooper pairs. Again, in a model of condensates there is no freedom to add a constant
to the effective potential.
The main problem of studying the possibility of obtaining inflation using conden-
sates is that the quantum effects which determine the theory are highly nonpertur-
bative. In particular, the effective potential written in terms of a condensate hϕi
does not correspond to a renormalizable theory and will in general [31] contain terms
of arbitrary power in hϕi. However (see [32]), one may make progress by assuming
certain general properties of the effective potential.
Let us [32] consider a theory in which at some time ti a condensate hϕi forms, i.e.
hϕi = 0 for t < ti and hϕi = 6 0 for t > ti . The expectation value of the Hamiltonian
H written in terms of the condensate hϕi contains terms of arbitrary powers of hϕi:
(−1)n n!an hϕin .

X
hHi = (31)
n
We summarize our ignorance of the nonperturbative physics in the assumption that

the resulting series is asymptotic, and in particular Borel summable, with coefficients
an ∝ 1. In this case, we can resum the series to obtain [32]
∞ f (t)dt
Z
hHi = e−1/t , (32)
0 t(tmpl + hϕi)
16
where the function f (t) is related to the coefficients an via
1 ∞
Z
an = dtf (t)t−n−2 e−1/t . (33)
n! 0
The expectation value of the Hamiltonian hHi can be interpreted as the effective
potential Vef f of this theory. The question is under which conditions this potential
gives rise to inflation. If we regard hϕi as a classical field (i.e. neglect the ultraviolet
and infrared divergences of the theory), then the dynamics of the model can be read
off directly from (32), with initial conditions for hϕi at the time ti close to hϕi = 0.
It is easy to check that rather generically, the conditions required to have slow rolling
of ϕ, namely
√
V ′ mpl << 48πV (34)
V ′′ m2pl << 24πV , (35)
are satisfied. However, since the potential decays only slowly at large values of hϕi
and since there is no true vacuum state at finite values of hϕi, the slow rolling con-
ditions are satisfied for all times. In this case, inflation would never end - an obvious
cosmological disaster.
However, hϕi is not a classical scalar field but the expectation value of a condensate
operator. Thus, we have to worry about diverging contributions to this expectation
value. In particular, in a theory with symmetry breaking there will often be massless
excitations which will give rise to infrared divergences. It is necessary to introduce
an infrared cutoff energy ε whose value is determined in the context of cosmology
by the Hubble expansion rate. Note in particular that this cutoff is time-dependent.
Effectively, we thus have a theory of two scalar fields hϕi and ε. In this case, the first
of the slow rolling conditions becomes (if ε is expressed in Planck units)
ε̇2 m2pl + ϕ̇2 << 2V . (36)
The infrared cutoff changes the form of the effective potential. We assume that
this change can be modelled by replacing hϕi by hϕi/ε. If we (following [33]) take
the infrared cutoff to be
H(0)
ε(t) = [1 − a(Ht)p ] , (37)
mpl
where 0 < a << 1 and p is an integer and the time at the beginning of the rolling has
been set to t = 0, then it can be shown [32] that an period of inflation with a graceful
exit is realized. After the condensate hϕi starts rolling at hϕi ∼ 0, inflation will
commence. As inflation proceeds, ε(t) will slowly grow and will eventually dominate
the energy functional, signaling an end of the inflationary period. From (37) it follows
that inflation lasts until a1/p Ht = 1.
17
This analysis demonstrates that it is in principle possible to obtain inflation from
condensates. However, the model must be studied in much more detail before we can
determine whether it gives a realization of inflation which is free of problems.
Fluctuation Problem
A generic problem for all realizations of inflation studied up to now concerns the
amplitude of the density perturbations which are induced by quantum fluctuations
during the period of exponential expansion. From the amplitude of CMB anisotropies
measured by COBE, and from the present amplitude of density inhomogeneities on
scales of clusters of galaxies, it follows that the amplitude of the mass fluctuations
δM /M on a length scale given by the comoving wavenumber k at the time tH (k)
when that scale crosses the Hubble radius in the FRW period is
δM
(k, tH (k)) ∝ 10−5 . (38)
M
The generation and evolution of fluctuations will be discussed in detail in Section
3. The perturbations arise during inflation as quantum excitations. Their amplitude
at the time ti (k) when the scale k leaves the Hubble radius during inflation is given
by
δM V ′ δϕ
(k, ti (k)) ≃ |t (k) , (39)
M ρ i
where δϕ is given by the amplitude of the quantum fluctuation of δϕ(k) (note that
this is a momentum space quantity). While the scale k is outside of the Hubble
radius, the fluctuation amplitude grows by general relativistic gravitational effects.
The amplitudes at ti (k) and tH (k) are related by
δM 1 δM
(K, tH (k)) ≃ |ti (k) (k, ti (k)) (40)
M 1 + p/ρ M
(see e.g. [34]). Combining (39) and (40) and working out the result for the potential
V (ϕ) = λϕ4 (41)

[35–37]
we obtain the result
δM
(K, tH (k)) ≃ 102 λ1/2 . (42)
M
Thus, in order to agree with the observed value (38), the coupling constant λ must
be extremely small:
λ ≤ 10−12 . (43)
It has been shown in [38] that the above conclusion is generic, at least for models in
which inflation is driven by a scalar field. In order that inflation does not produce a too
18
large amplitude of the spectrum of perturbations, a dimensionless number appearing
in the potential must be set to a very small value. A possible resolution of this
problem will be mentioned in the following subsection.
Reheating Problem
A question which has recently received a lot of attention and will be discussed in
greater detail in one of the following subsections is the issue of reheating in inflationary
cosmology. The question concerns the energy transfer between the inflaton and matter
fields which is supposed to take place at the end of inflation (see Fig. 8).
According to either new inflation or chaotic inflation, the dynamics of the inflaton
leads first to a transfer of energy from potential energy of the inflaton to kinetic
energy. After the period of slow rolling, the inflaton ϕ begins to oscillate about the
true minimum of V (ϕ). Quantum mechanically, the state of homogeneous oscillation
corresponds to a coherent state. Any coupling of ϕ to other fields (and even self
coupling terms of ϕ) will lead to a decay of this state. This corresponds to the particle
production. The produced particles will be relativistic, and thus at the conclusion of
the reheating period a radiation dominated Universe will emerge.
The key questions are by what mechanism and how fast the decay of the coherent
state takes place. It is important to determine the temperature of the produced par-
ticles at the end of the reheating period. The answers are relevant to many important
questions regarding the post-inflationary evolution. For example, it is important to
know whether the temperature after reheating is high enough to allow GUT baryogen-
esis and the production of GUT-scale topological defects. In supersymmetric models,
the answer determines the predicted abundance of gravitinos and other moduli fields.
Recently, there has been a complete change in our understanding of reheating.
This topic will be discussed in detail below.
D. Inflation and Nonsingular Cosmology
The question we wish to address in this subsection is whether it is possible to

construct a class of effective actions for gravity which have improved singularity prop-
erties and which predict inflation, with the constraint that they give the correct low
curvature limit. Since Planck scale physics will generate corrections to the Einstein
action, it is quite reasonable to consider higher derivative gravity models.
What follows is a summary of recent work [28] in which we have constructed an
effective action for gravity in which all solutions with sufficient symmetry are non-
singular. The theory is a higher derivative modification of the Einstein action, and
is obtained by a constructive procedure well motivated in analogy with the analysis
of point particle motion in special relativity. The resulting theory is asymptotically
free in a sense which will be specified below.
Our aim is to construct a theory with the property that the metric gµν approaches
DS
the de Sitter metric gµν , a metric with maximal symmetry which admits a geodesi-
19
cally complete and nonsingular extension, as the curvature R approaches the Planck
value Rpl . Here, R stands for any curvature invariant. Naturally, from our classical
considerations, Rpl is a free parameter. However, if our theory is connected with
Planck scale physics, we expect Rpl to be set by the Planck scale.
If successful, the above construction will have some very appealing consequences.
Consider, for example, a collapsing spatially homogeneous Universe. According to
Einstein’s theory, this Universe will collapse in finite proper time to a final “big
crunch” singularity (top left Penrose diagram of Figure 9). In our theory, however, the
Universe will approach a de Sitter model as the curvature increases. If the Universe is
closed, there will be a de Sitter bounce followed by re-expansion (bottom left Penrose
diagram in Figure 9). Similarly, in our theory spherically symmetric vacuum solutions
would be nonsingular, i.e., black holes would have no singularities in their centers.
The structure of a large black hole would be unchanged compared to what is predicted
by Einstein’s theory (top right, Figure 9) outside and even slightly inside the horizon,
since all curvature invariants are small in those regions. However, for r → 0 (where
r is the radial Schwarzschild coordinate), the solution changes and approaches a de
Sitter solution (bottom right, Figure 9). This would have interesting consequences
for the black hole information loss problem.
FIG. 9. Penrose diagrams for collapsing Universe (left) and black hole (right) in Ein-
stein’s theory (top) and in the nonsingular Universe (bottom). C, E, DS and H stand for
contracting phase, expanding phase, de Sitter phase and horizon, respectively, and wavy
lines indicate singularities.
20
To motivate our effective action construction, we turn to a well known analogy,
point particle motion in the theory of special relativity.
An Analogy
The transition from the Newtonian theory of point particle motion to the special
relativistic theory transforms a theory with no bound on the velocity into one in
which there is a limiting velocity, the speed of light c (in the following we use units
in which h̄ = c = 1). This transition can be obtained [28] by starting with the action
of a point particle with world line x(t):
Z
1
Sold = dt ẋ2 , (44)
2
introducing [39] a Lagrange multiplier field ϕwhich couples to ẋ2 , the quantity to be
made finite, and which has a potential V (ϕ). The new action is
1
Z
Snew = dt ẋ2 + ϕẋ2 − V (ϕ) . (45)
2
From the constraint equation
∂V
ẋ2 = , (46)
∂ϕ
it follows that ẋ2 is limited provided V (ϕ) increases no faster than linearly in ϕ for
large |ϕ|. The small ϕ asymptotics of V (ϕ) is determined by demanding that at low
velocities the correct Newtonian limit results:
V (ϕ) ∼ ϕ2 as |ϕ| → 0 , (47)

V (ϕ) ∼ ϕ as |ϕ| → ∞ . (48)
Choosing the simple interpolating potential
2ϕ2
V (ϕ) = , (49)
1 + 2ϕ
the Lagrange multiplier can be integrated out, resulting in the well-known action
1
Z √
Snew = dt 1 − ẋ2 (50)
2
for point particle motion in special relativity.
Construction
Our procedure for obtaining a nonsingular Universe theory [28] is based on gen-
eralizing the above Lagrange multiplier construction to gravity. Starting from the
21
Einstein action, we can introduce a Lagrange multiplier ϕ1 coupled to the Ricci scalar
R to obtain a theory with limited R:
Z
√
S = d4 x −g(R + ϕ1 R + V1 (ϕ1 )) , (51)
where the potential V1 (ϕ1 ) satisfies the asymptotic conditions (47) and (48).
However, this action is insufficient to obtain a nonsingular gravity theory. For ex-
ample, singular solutions of the Einstein equations with R = 0 are not effected at all.
The minimal requirements for a nonsingular theory is that all curvature invariants
remain bounded and the space-time manifold is geodesically complete. Implement-
ing the limiting curvature hypothesis [40] , these conditions can be reduced to more
manageable ones. First, we choose one curvature invariant I1 (gµν ) and demand that
it be explicitely bounded, i.e., |I1 | < I1pl , where I1pl is the Planck scale value of I1 .
In a second step, we demand that as I1 (gµν ) approaches I1pl , the metric gµν approach
DS
the de Sitter metric gµν , a definite nonsingular metric with maximal symmetry. In
this case, all curvature invariants are automatically bounded (they approach their de
Sitter values), and the space-time can be extended to be geodesically complete.
Our approach is to implement the second step of the above procedure by another
Lagrange multiplier construction [28] . We look for a curvature invariant I2 (gµν ) with
the property that
DS
I2 (gµν ) = 0 ⇔ gµν = gµν , (52)
introduce a second Lagrange multiplier field ϕ2 which couples to I2 and choose a
potential V2 (ϕ2 ) which forces I2 to zero at large |ϕ2 |:
Z
√
S = d4 x −g[R + ϕ1 I1 + V1 (ϕ1 ) + ϕ2 I2 + V2 (ϕ2 )] , (53)
with asymptotic conditions (47) and (48) for V1 (ϕ1 ) and conditions
V2 (ϕ2 ) ∼ const as |ϕ2 | → ∞ (54)
V2 (ϕ2 ) ∼ ϕ22 as |ϕ2 | → 0 , (55)
for V2 (ϕ2 ). The first constraint forces I2 to zero, the second is required in order to
obtain the correct low curvature limit.
These general conditions are reasonable, but not sufficient in order to obtain a
nonsingular theory. It must still be shown that all solutions are well behaved, i.e.,
that they asymptotically reach the regions |ϕ2 | → ∞ of phase space (or that they
can be controlled in some other way). This must be done for a specific realization of
the above general construction.
Specific Model
At the moment we are only able to find an invariant I2 which singles out de
Sitter space (by demanding I2 = 0) provided we assume that the metric has special
symmetries. The choice
22
I2 = (4Rµν Rµν − R2 + C 2 )1/2 , (56)
singles out the de Sitter metric among all homogeneous and isotropic metrics (in
which case adding C 2 , the Weyl tensor square, is superfluous), all homogeneous and
anisotropic metrics, and all radially symmetric metrics.
We choose the action [28,41]
" #
Z
4 √ 3 1/2
S = d x −g R + ϕ1 R − (ϕ2 + √ ϕ1 )I2 + V1 (ϕ1 ) + V2 (ϕ2 ) (57)
2
with
ϕ21
!
ln(1 + ϕ1 )
V1 (ϕ1 ) = 12 H02 1− (58)
1 + ϕ1 1 + ϕ1
√ ϕ22
V2 (ϕ2 ) = −2 3 H02 . (59)
1 + ϕ22
The general equations of motion resulting from this action are quite messy. How-
ever, when restricted to homogeneous and isotropic metrics of the form
ds2 = dt2 − a(t)2 (dx2 + dy 2 + dz 2 ) , (60)
the equations are fairly simple. With H = ȧ/a, the two ϕ1 and ϕ2 constraint equations
are
1 ′
H2 = V (61)
12 1
1
Ḣ = − √ V2′ , (62)
2 3
and the dynamical g00 equation becomes
1 √
3(1 − 2ϕ1 )H 2 + (V1 + V2 ) = 3H(ϕ̇2 + 3Hϕ2) . (63)
2
The phase space of all vacuum configurations is the half plane {(ϕ1 ≥ 0, ϕ2 )}. Equa-
tions (61) and (62) can be used to express H and Ḣ in terms of ϕ1 and ϕ2 . The
remaining dynamical equation (63) can then be recast as
√
" #
dϕ2 V ′′ 2
= − 1 ′ − 3ϕ2 + (1 − 2ϕ1 ) − ′ (V1 + V2 ) . (64)
dϕ1 4V2 V1
The solutions can be studied analytically in the asymptotic regions and numerically
throughout the entire phase space.
23
FIG. 10. Phase diagram of the homogeneous and isotropic solutions of the nonsingular
Universe. The asymptotic regions are labelled by A, B, C and D, flow lines are indicated
by arrows.
The resulting phase diagram of vacuum solutions is sketched in Fig. 10 (for

numerical results, see [41]). The point (ϕ1 , ϕ2 ) = (0, 0) corresponds to Minkowski
space-time M 4 , the regions |ϕ2 | → ∞ to de Sitter space. As shown, all solutions
either are periodic about M 4 or else they asymptotically approach de Sitter space.
Hence, all solutions are nonsingular. This conclusion remains unchanged if we add
spatial curvature to the model.
One of the most interesting properties of our theory is asymptotic freedom [41] ,
i.e., the coupling between matter and gravity goes to zero at high curvatures. It is
easy to add matter (e.g., dust or radiation) to our model by taking the combined
action
S = Sg + Sm , (65)
where Sg is the gravity action previously discussed, and Sm is the usual matter action
in an external background space-time metric.
We find [41] that in the asymptotic de Sitter regions, the trajectories of the solu-
tions in the (ϕ1 , ϕ2 ) plane are unchanged by adding matter. This applies, for example,
in a phase of de Sitter contraction when the matter energy density is increasing expo-
nentially but does not affect the metric. The physical reason for asymptotic freedom
is obvious: in the asymptotic regions of phase space, the space-time curvature ap-
proaches its maximal value and thus cannot be changed even by adding an arbitrary
high matter energy density.
Naturally, the phase space trajectories near (ϕ1 , ϕ2 ) = (0, 0) are strongly effected
by adding matter. In particular, M 4 ceases to be a stable fixed point of the evolution
equations.
24
Discussion
We have shown that a class of higher derivative extensions of the Einstein theory
exist for which many interesting solutions are nonsingular. Our class of models is very
special. Most higher derivative theories of gravity have, in fact, much worse singularity
properties than the Einstein theory. What is special about our class of theories is
that they are obtained using a well motivated Lagrange multiplier construction which
implements the limiting curvature hypothesis. We have shown that
i) all homogeneous and isotropic solutions are nonsingular [28,41]
ii) the two-dimensional black holes are nonsingular [42]
iii) nonsingular two-dimensional cosmologies exist [43] .
By construction, all solutions are de Sitter at high curvature. Thus, the theories
automatically have a period of inflation (driven by the gravity sector in analogy to
Starobinsky inflation [27] ) in the early Universe.
A very important property of our theories is asymptotic freedom. This means that
the coupling between matter and gravity goes to zero at high curvature, and might
lead to an automatic suppression mechanism for scalar fluctuations.
E. Reheating in Inflationary Cosmology
Reheating is an important stage in inflationary cosmology. It determines the

state of the Universe after inflation and has consequences for baryogenesis, defect
formation, and, as will be shown below, maybe even for the composition of the dark
matter of the Universe.
After slow rolling, the inflaton field begins to oscillate uniformly in space about
the true vacuum state. Quantum mechanically, this corresponds to a coherent state
of k = 0 inflaton particles. Due to interactions of the inflaton with itself and with
other fields, the coherent state will decay into quanta of elementary particles. This
corresponds to post-inflationary particle production.
Reheating is usually studied using simple scalar field toy models. The one we will
adopt here consists of two real scalar fields, the inflaton ϕ with Lagrangian
1 1
Lo = ∂µ ϕ∂ µ ϕ − λ(ϕ2 − σ 2 )2 (66)
2 4
interacting with a massless scalar field χ representing ordinary matter. The interac-
tion Lagrangian is taken to be
1 2 2 2
LI = g ϕ χ . (67)
2
Self interactions of χ are neglected.
By a change of variables
ϕ = ϕ̃ + σ , (68)
25
the interaction Lagrangian can be written as
1
LI = g 2 σ ϕ̃χ2 + g 2 ϕ̃2 χ2 . (69)
2
During the phase of coherent oscillations, the field ϕ̃ oscillates with a frequency
ω = mϕ = λ1/2 σ (70)
(neglecting the expansion of the Universe which can be taken into account as in
[44,45]).
Elementary Theory of Reheating
According to the elementary theory of reheating (see e.g. [46] and [47]), the decay
of the inflaton is calculated using first order perturbations theory. According to the
Feynman rules, the decay rate ΓB of ϕ (calculated assuming that the cubic coupling
term dominates) is given by
g2σ2
ΓB = . (71)
8πmφ
The decay leads to a decrease in the amplitude of ϕ (from now on we will drop
the tilde sign) which can be approximated by adding an extra damping term to the
equation of motion for ϕ:
ϕ̈ + 3H ϕ̇ + ΓB ϕ̇ = −V ′ (ϕ) . (72)
From the above equation it follows that as long as H > ΓB , particle production is
negligible. During the phase of coherent oscillation of ϕ, the energy density and hence
H are decreasing. Thus, eventually H = ΓB , and at that point reheating occurs (the
remaining energy density in ϕ is very quickly transferred to χ particles.
The temperature TR at the completion of reheating can be estimated by computing
the temperature of radiation corresponding to the value of H at which H = ΓB . From
the FRW equations it follows that
TR ∼ (ΓB mpl )1/2 . (73)
If we now use the “naturalness” constraint1
g2 ∼ λ (74)
1 Atone loop order, the cubic interaction term will contribute to λ by an amout ∆λ ∼ g2 . A
renormalized value of λ smaller than g2 needs to be finely tuned at each order in perturbation
theory, which is “unnatural”.
26
in conjunction with the constraint on the value of λ from (43), it follows that for
σ < mpl ,
TR < 1010 GeV . (75)
This would imply no GUT baryogenesis, no GUT-scale defect production, and no

gravitino problems in supersymmetric models with m3/2 > TR , where m3/2 is the
gravitino mass. As we shall see, these conclusions change radically if we adopt an
improved analysis of reheating.
Modern Theory of Reheating
However, as was first realized in [48], the above analysis misses an essential point.
To see this, we focus on the equation of motion for the matter field χ coupled to
the inflaton ϕ via the interaction Lagrangian LI of (69). Taking into account for the
moment only the cubic interaction term, the equation of motion becomes
∇ 2
χ̈ + 3H χ̇ − (( ) − m2χ − 2g 2σϕ)χ = 0 . (76)
a
Since the equation is linear in χ, the equations for the Fourier modes χk decouple:
χ̈k + 3H χ̇k + (kp2 + m2χ + 2g 2σϕ)χk = 0, (77)
where kp is the time-dependent physical wavenumber.

Let us for the moment neglect the expansion of the Universe. In this case, the
friction term in (77) drops out and kp is time-independent, and Equation (77) be-
comes a harmonic oscillator equation with a time-dependent mass determined by the
dynamics of ϕ. In the reheating phase, ϕ is undergoing oscillations. Thus, the mass in
(77) is varying periodically. In the mathematics literature, this equation is called the
Mathieu equation. It is well known that there is an instability. In physics, the effect
is known as parametric resonance (see e.g. [49]). At frequencies ωn corresponding
to half integer multiples of the frequency ω of the variation of the mass, i.e.
n
ωk2 = kp2 + m2χ = ( ω)2 n = 1, 2, ..., (78)
2
there are instability bands with widths ∆ωn . For values of ωk within the instability
band, the value of χk increases exponentially:
g 2σϕ0
χk ∼ eµt with µ ∼ , (79)
ω
with ϕ0 being the amplitude of the oscillation of ϕ. Since the widths of the instability
bands decrease as a power of the (small) coupling constant g 2 with increasing n, for
practical purposes only the lowest instability band is important. Its width is
1/2
∆ωk ∼ gσ 1/2 ϕ0 . (80)
27
Note, in particular, that there is no ultraviolet divergence in computing the total
energy transfer from the ϕ to the χ field due to parametric resonance.
It is easy to include the effects of the expansion of the Universe (see e.g. [48,44,45]).
The main effect is that the value of ωk becomes time-dependent. Thus, a mode
slowly enters and leaves the resonance bands. As a consequence, any mode lies in
the resonance band for only a finite time. This implies that the calculation of energy
transfer is perfectly well-behaved. No infinite time divergences arise.
It is now possible to estimate the rate of energy transfer, whose order of magnitude
is given by the phase space volume of the lowest instability band multiplied by the
rate of growth of the mode function χk . Using as an initial condition for χk the value
χk ∼ H given by the magnitude of the expected quantum fluctuations, we obtain
ω
ρ̇ ∼ µ( )2 ∆ωk Heµt . (81)
2
From (81) it follows that provided that the condition
µ∆t >> 1 (82)
is satisfied, where ∆t < H −1 is the time a mode spends in the instability band,
then the energy transfer will procede fast on the time scale of the expansion of the
Universe. In this case, there will be explosive particle production, and the energy
density in matter at the end of reheating will be given by the energy density at the
end of inflation.
The above is a summary of the main physics of the modern theory of reheating.
The actual analysis can be refined in many ways (see e.g. [44,45]). First of all, it
is easy to take the expansion of the Universe into account explicitly (by means of a
transformation of variables), to employ an exact solution of the background model
and to reduce the mode equation for χk to a Hill equation, an equation similar to the
Mathieu equation which also admits exponential instabilities.
The next improvement consists of treating the χ field quantum mechanically (keep-
ing ϕ as a classical background field). At this point, the techniques of quantum field
theory in a curved background can be applied. There is no need to impose artificial
classical initial conditions for χk . Instead, we may assume that χ starts in its initial
vacuum state (excitation of an initial thermal state has been studied in [50]), and the
Bogoliubov mode mixing technique (see e.g. [51]) can be used to compute the number
of particles at late times.
Using this improved analysis, we recover the result (81). Thus, provided that the
condition (82) is satisfied, reheating will be explosive. Working out the time ∆t that
a mode remains in the instability band for our model, expressing H in terms of ϕ0
and mpl , and ω in terms of σ, and using the naturalness relation g 2 ∼ λ, the condition
for explosive particle production becomes
ϕ0 mpl
>> 1 , (83)
σ2
28
which is satisfied for all chaotic inflation models with σ < mpl (recall that slow rolling
ends when ϕ ∼ mpl and that therefore the initial amplitude ϕ0 of oscillation is of the
order mpl ).
We conclude that rather generically, reheating in chaotic inflation models will be
explosive. This implies that the energy density after reheating will be approximately
equal to the energy density at the end of the slow rolling period. Therefore, as
suggested in [52,53] and [54], respectively, GUT scale defects may be produced after
reheating and GUT-scale baryogenesis scenarios may be realized, provided that the
GUT energy scale is lower than the energy scale at the end of slow rolling.
Note, however, that the state of χ after parametric resonance is not a thermal
state. The spectrum consists of high peaks in distinct wave bands. An important
question which remains to be studied is how this state thermalizes. For some inter-
esting work on this issue see [55]. As emphasized in [52] and [53], the large peaks
in the spectrum may lead to symmetry restoration and to the efficient production of
topological defects (for a differing view on this issue see [56,57]). Since the state after
explosive particle production is not a thermal state, it is useful to follow [44] and call
this process “preheating” instead of reheating.
A futher interesting conjecture which emerges from the parametric resonance anal-
ysis of preheating [44,45] is that the dark matter in the Universe may consist of remnant
coherent oscillations of the inflaton field. In fact, it can easily be checked from (83)
that the condition for efficient transfer of energy eventually breaks down when ϕ0
has decreased to a sufficiently small value. For the model considered here, an order
of magnitude calculation shows that the remnant oscillations may well contribute
significantly to the present value of Ω.
Note that the details of the analysis of preheating are quite model-dependent.
In fact [44] , in many models one does not get the kind of “narrow-band” resonance
discussed here, but “wide-band” resonance. In this case, the energy transfer is even
more efficient.
There has recently been a lot of work on various aspects of reheating (see e.g.
[58–61] for different approaches). Many important questions, e.g. concerning ther-
malization and back-reaction effects during and after preheating (or parametric res-
onance) remain to be fully analyzed.
F. Summary
The inflationary Universe is an attractive scenario for early Universe cosmology.

It can resolve some of the problems of standard cosmology, and in addition gives rise
to a predictive theory of structure formation (see e.g. [62] for a recent review).
However, important unsolved problems of principle remain. Rather generically,
the predicted amplitude of perturbations is too large (the spectral shape, however, is
in quite good agreement with the observations). The present realizations of inflation
based on scalar field also make the cosmological constant problem more accute. In
29
addition, there are no convincing particle-physics based realizations of inflation. Many
models of inflation resort to introducing a new matter sector. It is important to search
for a better connection between modern particle physics / field theory and inflation.
String cosmology and dilaton gravity (see e.g. the recent reviews in [63]) may provide
an interesting new approach to the unification of inflation and fundamental physics.
Recently, there has been much progress in the understanding of the energy transfer
at the end of inflation between the inflaton field and matter. It appears that resonance
phenomena such as parametric resonance play a crucial role. These new reheating
scenarios lead to a high reheating temperature, although much more work remains to
be done before one can reach a final conclusion on this issue.
III. CLASSICAL AND QUANTUM THEORY OF COSMOLOGICAL

PERTURBATIONS
In inflationary Universe and topological defect models of structure formation,

small amplitude seed perturbations are predicted to arise due to particle physics
effects in the very early Universe. They then grow by gravitational instability to
produce the cosmological structures we observe today. In order to be able to make
the connection between particle physics and observations, it is important to under-
stand the gravitational evolution of fluctuations. This section will introduce the basic
concepts of this topic.
As is evident from Figure 5 and from the discussion of inflation in the previous
section, general relativity and quantum mechanics both play a fundamental role in
the theory of perturbations. In inflationary Universe models, quantum effects seed
the fluctuations, and thus a quantum analysis of the generation of fluctuations is
essential. However, since the fluctuations are small, a linearized analysis is sufficient.
Since the scales on which we are interested in following the fluctuations are larger than
the Hubble radius for a long time interval, Newtonian gravity is obviously inadequate
to treat these perturbations, and general relativistic effects become essential.
In this section, we will first introduce some basic notation, then discuss the New-
tonian theory of linear fluctuations before turning to the full relativistic analysis.
A. Basic Issues
In this article we only discuss theories in which structures grow by gravitational

accretion. The basic mechanism is easy to understand. Consider first a flat space-
time background. A density perturbation with δρ > 0 leads to an excess gravitational
attractive force F acting on the surrounding matter. This force is proportional to δρ,
and will hence lead to exponential growth of the perturbation since
δ ρ̈ ∼ F ∼ δρ ⇒ δρ ∼ exp(αt) (84)
30
with a constant α which is proportional to Newton’s constant G.
In an expanding background space-time, the acceleration is damped by the ex-
pansion. If r(t) is the physical distance of a test particle from the perturbation, then
on a scale r
δρ
δ ρ̈ ∼ F ∼ , (85)
r 2 (t)
which results in power-law increase of δρ. The goal of this subsection is to discuss the
growth rates of inhomogeneities in more detail (see e.g. [64,65] for modern reviews).
Because of our assumption that all perturbations start out with a small amplitude,
we can linearize the equations for gravitational fluctuations. The analysis is then
greatly simplified by going to momentum space in which all modes δ(k) decouple.
We expand the fractional density contrast δ(x) as follows:
δρ(x) −3/2 1/2
Z
δ(x) = = (2π) V d3 k eik·x δ(k) , (86)
ρ
where V is a cutoff volume which disappears from all physical observables.
The “power spectrum” P (k) is defined by
P (k) =< |δ(k)|2 > , (87)
where the braces denote an ensemble average (in most structure formation models,
the generation of perturbations is a stochastic process, and hence observables can
only be calculated by averaging over the ensemble. For observations, the braces can
be viewed as an angular average).
The physical measure of mass fluctuations on a length scale λ is the r.m.s. mass
fluctuation δM/M(λ) on this scale. It is determined by the power spectrum in the
following way. We pick a center x0 of a sphere Bλ (x0 ) of radius λ and calculate
δM 2 1
Z
| | (x0 , λ) = | d3 xδ(x) |2 , (88)
M V (Bλ )
Bλ (x0 )
where V (Bλ ) is the volume of the sphere. Inserting the Fourier decomposition (86)
and taking the average value of this quantity over all x0 yields
!2
δM
Z
< (λ) >= d3 kWk (λ)|δ(k)|2 (89)
M
with a window function Wk (λ) with the following properties

≃ 1 k < kλ = 2π/λ

Wk (λ) (90)
≃ 0 k > kλ .
Therefore the r.m.s. mass perturbation on a scale λ becomes
31
δM
<| (λ)|2 >∼ kλ3 P (kλ) . (91)
M
If P (k) ∼ k n then n is called the index of the power spectrum. For n = 1 we get
the so-called Harrison-Zel’dovich scale invariant spectrum [66] .
Both inflationary Universe and topological defect models of structure formation
predict a roughly scale invariant spectrum. The distinguishing feature of this spec-
trum is that the r.m.s. mass perturbations are independent of the scale k when
measured at the time tH (k) when the associated wavelength is equal to the Hubble
radius, i.e., when the scale “enters” the Hubble radius. Let us derive this fact for the
scales entering during the matter dominated epoch. The time tH (k) is determined by
k −1 a(tH (k)) = tH (k) (92)
which leads to tH (k) ∼ k −3 . According to the linear theory of cosmological perturba-

tions discussed in the following subsection, the mass fluctuations increase as a(t) for
t > teq . Hence
!2/3
δM tH (k) δM
(k, tH (k)) = (k, t) ∼ const , (93)
M t M
since the first factor scales (from (92) as k −2 and – using (91) and inserting n = 1 –
the second as k 2 .
B. Newtonian Theory
The Newtonian theory of cosmological perturbations is an approximate analysis

which is valid on wavelengths λ much smaller than the Hubble radius t and for
negligible pressure p, i.e., p ≪ ρ. It is based on expanding the hydrodynamical
equations about a homogeneous background solution.
The starting points are the continuity, Euler and Poisson equations
ρ̇ + ∇(ρv) = 0 (94)
1
v̇ + (v · ∇)v = −∇φ − ∇p (95)
ρ
∇2 φ = 4πGρ (96)
for a fluid with energy density ρ, pressure p, velocity v and Newtonian gravitational
potential φ, written in terms of physical coordinates (t, r).
The transition to an expanding space is made by introducing comoving coordinates
x and peculiar velocity u = ẋ:
32
r = a(t)x (97)
v = ȧ(t)x + a(t)u . (98)
The first term on the right hand side of (98) is the expansion velocity.
The perturbation equations are obtained by linearizing Equations (94 - 96) about
a homogeneous background solution ρ = ρ̄(t), p = 0 and u = 0. The linearization
ansatz can be written
ρ(x, t) = ρ̄(t)(1 + δ(x, t)) |δ| << 1. (99)
If we consider adiabatic perturbations (no entropy density variations), then after some
algebra the linearized equations become
δ̇ + ∇ · u = 0 , (100)
ȧ
u̇ + 2 u = −a2 (∇δφ + c2s ∇δ) (101)
a
and
∇2 δφ = 4πGρ̄a2 δ , (102)
with the speed of sound cs given by

∂p
c2s = . (103)
∂ρ
The two first order equations (100) and (101) can be combined to yield a single second
order differential equation for δ. With the help of (102) this equation reads
c2s 2
δ̈ + 2H δ̇ − 4πGρ̄δ − ∇ δ=0 (104)
a2
which in momentum space becomes
c2s k 2
!
δ̈k + 2H δ̇k + − 4πGρ̄ δk = 0 . (105)
a2
Here, H(t) as usual denotes the expansion rate, and δk stands for δ(k).
Already a quick look at Equation (105) reveals the presence of a distinguished
scale for cosmological perturbations, the Jeans length
2π
λJ = (106)
kJ
33
with
!2
k 4πGρ̄
kJ2 = = . (107)
a c2s
On length scales larger than λJ , the spatial gradient term is negligible, and the term
linear in δ in (105) acts like a negative mass square quadratic potential with damping
due to the expansion of the Universe, in agreement with the intuitive analysis leading
to (refintu1) and (85). On length scales smaller than λJ , however, (105) becomes a
damped harmonic oscillator equation and perturbations on these scales decay.
For t > teq and for λ ≫ λJ , Equation (105) becomes
4 2
δ̈k + δ̇k − 2 δk = 0 (108)
3t 3t
and has the general solution
δk (t) = c1 t2/3 + c2 t−1 . (109)
This demonstrates that for t > teq and λ ≫ λJ , the dominant mode of perturbations
increases as a(t), a result we already used in the previous subsection (see (93)).
For λ ≪ λJ and t > teq , Equation (105) becomes
!2
k
δ̈k + 2H δ̇k + c2s δk = 0 , (110)
a
and has solutions corresponding to damped oscillations:

Z
δk (t) ∼ a−1/2 (t) exp{±ics k dt′ a(t′ )−1 } . (111)
As an important application of the Newtonian theory of cosmological perturba-

tions, let us compare sub-horizon scale fluctuations in a baryon-dominated Universe
(Ω = ΩB = 1) and in a CDM-dominated Universe with ΩCDM = 0.9 and Ω = 1. We
consider scales which enter the Hubble radius at about teq .
In the initial time interval teq < t < trec , the baryons are coupled to the photons.
Hence, the baryonic fluid has a large pressure pB
1
pB ≃ pr = ρr , (112)
3
and therefore the speed of sound is relativistic
!1/2 !1/2
pr 1 ρr
cs ≃ =√ . (113)
ρm 3 ρm
34
The value of cs slowly decreases in this time interval, attaining a value of about 1/10
at trec . From (107) it follows that the Jeans mass MJ , the mass inside a sphere of
radius λJ , increases until trec when it reaches its maximal value MJmax
4π
MJmax = MJ (trec ) = λJ (trec )3 ρ̄(trec ) ∼ 1017 (Ωh2 )−1/2 M⊙ . (114)
3
At the time of recombination, the baryons decouple from the radiation fluid.
Hence, the baryon pressure pB drops abruptly, as does the Jeans length (see (107)).
The remaining pressure pB is determined by the temperature and thus continues to
decrease as t increases. It can be shown that the Jeans mass continues to decrease
after trec , starting from a value
MJ− (trec ) ∼ 106(Ωh2 )−1/2 M⊙ (115)
(where the superscript “−” indicates the mass immediately after teq .
In contrast, CDM has negligible pressure throughout the period t > teq and hence
experiences no Jeans damping. A CDM perturbation which enters the Hubble radius
at teq with amplitude δi has an amplitude at trec given by
a(trec )
δkCDM (trec ) ≃ δi , (116)
a(teq )
whereas a perturbation with the same scale and initial amplitude in a baryon-
dominated Universe is damped
!−1/2
a(trec )
δkBDM (trec ) ≃ δi . (117)
a(teq )
In order for the perturbations to have the same amplitude today, the initial size of
the inhomogeneity must be much larger in a BDM-dominated Universe than in a
CDM-dominated one:
!3/2
z(teq )
δkBDM (teq ) ≃ δkCDM (teq ) . (118)
z(trec )
For Ω = 1 and h = 1/2 the enhancement factor is about 30.

In a CDM-dominated Universe the baryons experience Jeans damping, but after
trec the baryons quickly fall into the potential wells created by the CDM perturbations,
and hence the baryon perturbations are proportional to the CDM inhomogeneities.
The above considerations, coupled with information about CMB anisotropies, can
be used to rule out a model with Ω = ΩB = 1. The argument goes as follows. For
adiabatic fluctuations, the amplitude of CMB anisotropies on an angular scale ϑ is
determined by the value of δρ/ρ (strictly speaking, the relativistic potential Φ to be
discussed in the following subsection) on the corresponding length scale λ(ϑ) at teq :
35
δT 1 δρ
(ϑ) = (λ(ϑ), teq ) . (119)
T 3 ρ
On scales of clusters we know that (for Ω = 1 and h = 1/2)
!
δρ
(λ(ϑ), teq ) ≃ z(teq )−1 ≃ 10−4 , (120)
ρ CDM
using the fact that today on cluster scales δρ/ρ ≃ 1. The bounds on δT /T on small
angular scales are
δT
(ϑ) << 10−4 , (121)
T
consistent with the predictions for a CDM model, but inconsistent with those of a
Ω = ΩB = 1 model, according to which we would expect anisotropies of the order of
10−3 . This is yet another argument in support of the existence of nonbaryonic dark
matter.
To conclude this subsection, let us briefly discuss two further aspects related
to Newtonian perturbations. The first concerns matter inhomogeneities during the
radiation-dominated epoch. We consider matter fluctuations with cs = 0 in a smooth
relativistic background. In this case, Equation (105) becomes
δ̈k + 2H δ̇k − 4πGρ̄m δk = 0 , (122)
where ρ̄m denotes the average matter energy density. The Hubble expansion param-
eter obeys
8πG
H2 = (ρ̄m + ρ̄r ) , (123)
3
with ρ̄r the background radiation energy density. For t < teq , ρ̄m is negligible in both
(122) and (123), and (122) has the general solution
δk (t) = c1 log t + c2 . (124)
In particular, this result implies that CDM perturbations which enter the Hubble
radius before teq have an amplitude which grows only logarithmically in time until
teq .
Finally, we consider hot dark matter (HDM) fluctuations. Whereas CDM particles
are cold, i.e. their peculiar velocity is negligible for all times relevant for structure
formation, HDM particles have relativistic velocities at teq , i.e. v(teq ) ∼ 1. The prime
candidate for HDM is a 25h250 eV tau neutrino.
The new aspect of HDM is related to neutrino free streaming [67] . Because of
the large velocity of the dark matter particles, pure dark matter inhomogeneities are
washed out on all scales below the neutrino free streaming length λcj (t),
36
λcj (t) ∼ v(t)z(t)t , (125)
which is the comoving distance the particles move in one Hubble expansion time.
Since the neutrino velocity v(t) and the redshift z(t) both scale as a(t)−1 , the free
streaming length decreases as
λcj (t) ∼ t−1/3 (126)
after teq (before teq the radiation pressure dominates).

Hence, in an inflationary HDM model in which the fluctuations are dark matter
inhomogeneities, all perturbations on scales λ smaller than the maximal value of λcj (t)
are erased. The critical scale λmax
j is given by the value of λcj (t) at the time when the
neutrinos become non-relativistic, which is in turn determined by the neutrino mass
mν . The result is
mν
−2
λmax
j ≃ 30 Mpc , (127)
25eV
a scale much larger than the mean separation of galaxies and clusters. Since we ob-
serve galaxies outside of large-scale structures, this model is in blatant disagreement
with observations. However, theories in which the primordial perturbations are nona-
diabatic long-lived seeds (e.g. cosmic strings), may well be viable if the dark matter
is hot. As we shall see in Section 4, the cosmic string model in fact works well for
hot dark matter.
C. Relativistic Theory: Classical Analysis
On scales larger than the Hubble radius (λ > t) the Newtonian theory of cos-
mological perturbations obviously is inapplicable, and a general relativistic analysis
is needed. On these scales, matter is essentially frozen in comoving coordinates.
However, space-time fluctuations can still increase in amplitude.
In principle, it is straightforward to work out the general relativistic theory of
linear fluctuations [68] . We linearize the Einstein equations
Gµν = 8πGTµν (128)
(where Gµν is the Einstein tensor associated with the space-time metric gµν , and Tµν
is the energy-momentum tensor of matter) about an expanding FRW background
(0)
(gµν , ϕ(0) ):
(0)
gµν (x, t) = gµν (t) + hµν (x, t) (129)
ϕ(x, t) = ϕ(0) (t) + δϕ(x, t) (130)
and pick out the terms linear in hµν and δϕ to obtain
37
δGµν = 8πGδTµν . (131)
In the above, hµν is the perturbation in the metric and δϕ is the fluctuation of the
matter field ϕ. We have denoted all matter fields collectively by ϕ.
In practice, there are many complications which make this analysis highly nontriv-
ial. The first problem is “gauge invariance” [69] Imagine starting with a homogeneous
FRW cosmology and introducing new coordinates which mix x and t. In terms of the
new coordinates, the metric now looks inhomogeneous. The inhomogeneous piece of
the metric, however, must be a pure coordinate (or ”gauge”) artefact. Thus, when
analyzing relativistic perturbations, care must be taken to factor out effects due to
coordinate transformations.
FIG. 11. Sketch of how two choices of the mapping from the background space-time
manifold M0 to the physical manifold M induce two different coordinate systems on M.
The issue of gauge dependence is illustrated in Fig. 11. A coordinate system on

the physical inhomogeneous space-time manifold M can be viewed as a mapping D
of an unperturbed space-time M0 into M. A physical quantity Q is a geometrical
function defined on M. There is a corresponding physical quantity (0) Q defined on
M0 . In the coordinate system given by D, the perturbation δQ of Q at the space-time
point p ǫ M is
(0)
δQ(p) = Q(p) − Q (D −1 (p)) . (132)
However, in a second coordinate system D̃ the perturbation is given by

(0)
δ Q̃(p) = Q(p) − Q(D̃ −1 (p)) . (133)
The difference
38
∆Q(p) = δQ(p) − δ Q̃(p) (134)
is obviously a gauge artefact and carries no physical meaning.

There are various methods of dealing with gauge artefacts. The simplest and
most physical approach is to focus on gauge invariant variables, i.e., combinations
of the metric and matter perturbations which are invariant under linear coordinate
transformations.
The gauge invariant theory of cosmological perturbations is in principle straight-
forward, although technically rather tedious. In the following I will summarize the
main steps and refer the reader to [34] for the details and further references (see also
[70] for a pedagogical introduction and [71–78] for other approaches).
We consider perturbations about a spatially flat Friedmann-Robertson-Walker
metric
ds2 = a2 (η)(dη 2 − dx2 ) (135)
where η is conformal time (related to cosmic time t by a(η)dη = dt). A scalar metric
perturbation (see [79] for a precise definition) can be written in terms of four free
functions of space and time:
2φ −B,i

2
δgµν = a (η) . (136)
−B,i 2(ψδij + E,ij )
Scalar metric perturbations are the only perturbations which couple to energy density
and pressure.
The next step is to consider infinitesimal coordinate transformations
′
xµ = xµ + ξ µ (137)
which preserve the scalar nature of δgµν , and to calculate the induced transformations
of φ, ψ, B and E. Then we find invariant combinations to linear order. (Note that
there are in general no combinations which are invariant to all orders [80] .) After some
algebra, it follows that
Φ = φ + a−1 [(B − E ′ )a]′ (138)

a′
Ψ = ψ − (B − E ′ ) (139)
a
are two invariant combinations. In the above, a prime denotes differentiation with
respect to η.
There are various methods to derive the equations of motion for gauge invariant
variables. Perhaps the simplest way [34] is to consider the linearized Einstein equations
(131) and to write them out in the longitudinal gauge defined by
B=E=0 (140)
39
and in which Φ = φ and Ψ = ψ, to directly obtain gauge invariant equations.
For several types of matter, in particular for scalar field matter, the perturbation
of Tµν has the special property
δTji ∼ δji (141)
which imples Φ = Ψ. Hence, the scalar-type cosmological perturbations can in this

case be described by a single gauge invariant variable. The equation of motion takes
the form [37,81,75,82]
!2
k
ξ˙ = O Hξ (142)
aH
where
2 H −1 Φ̇ + Φ
ξ= + Φ. (143)
3 1+w
The variable w = p/ρ (with p and ρ background pressure and energy density
respectively) is a measure of the background equation of state. In particular, on
scales larger than the Hubble radius, the right hand side of (142) is negligible, and
hence ξ is constant.
The result that ξ˙ = 0 is a very powerful one. Let us first imagine that the equation
of state of matter is constant, i.e., w = const. In this case, ξ˙ = 0 implies
Φ(t) = const , (144)
i.e., this gauge invariant measure of perturbations remains constant outside the Hub-
ble radius.
Next, consider the evolution of Φ during a phase transition from an initial phase
with w = wi to a phase with w = wf . Long before and after the transition, Φ is
constant because of (144), and hence ξ˙ = 0 becomes
Φ
+ Φ = const , (145)
1+w
In order to make contact with matter perturbations and Newtonian intuition, it
is important to remark that, as a consequence of the Einstein constraint equations,
at Hubble radius crossing Φ is a measure of the fractional density fluctuations:
δρ
Φ(k, tH (k)) ∼ (k, tH (k)) . (146)
ρ
(Note that the latter quantity is approximately gauge invariant on scales smaller than
the Hubble radius).
40
D. Relativistic Theory: Quantum Analysis
The question of the origin of classical density perturbations from quantum fluc-
tuations in the de Sitter phase of an inflationary Universe is a rather subtle issue.
Starting from a homogeneous quantum state (e.g., the vacuum state in the FRW
coordinate frame at time ti , the beginning of inflation), a naive semiclassical anaylsis
would predict the absence of fluctuations since < ψ|Tµν (x)|ψ > is independent of
space.
However, as a simple thought experiment shows, such a naive analysis is inap-
propriate. Imagine a local gravitational mass detector D positioned close to a large
mass M which is suspended from a pole. The decay of an alpha particle will sever
the cord by which the mass is held to the pole and the mass will drop. According to
the semiclassical prescription
Gµν = 8πG < ψ|Tµν |ψ > , (147)
the metric (i.e., the mass measured) will slowly decrease. This is obviously not what
happens. The mass detector shows a signal which corresponds to one of the classical
trajectories which make up the state |ψ >, a trajectory corresponding to a sudden
drop in the gravitational force measured.
The origin of classical density perturbations as a consequence of quantum fluctua-
tions in a homogeneous state |ψ > can be analyzed along similar lines. The quantum
to classical transition is picking out [83–85] one of the typical classical trajectories
which make up the wave function of |ψ >. We can implement [86,87] the procedure as
follows: Define a classical scalar field
ϕcl (x, t) = ϕ0 (t) + δϕ(x, t) (148)
with vanishing spatial average of δϕ. The induced classical energy momentum tensor
cl
Tµν (x, t) which is the source for the metric is given by
cl
Tµν (x, t) = Tµν (ϕcl (x, t)) , (149)
where Tµν (ϕcl (x, t)) is defined as the canonical energy-momentum tensor of the clas-
cl
sical scalar field ϕcl (x, t). Unless δϕ vanishes, Tµν is inhomogeneous.
For applications to chaotic inflation, we take |ψ > to be a Gaussian state with
mean value ϕ0 (t)
< ψ|ϕ2 (x, t)|ψ >= ϕ20 (t) . (150)
Its width is taken to be the width of the vacuum state of the free scalar field theory
with mass determined by the curvature of V (ϕ) at ϕ0 . This state is used to define
the Fourier transform δ ϕ̃(k, t) by
|δ ϕ̃(k)|2 =< ψ| |ϕ̃(k)|2 |ψ > . (151)
41
The amplitude of δ ϕ̃(k) is identified with the width of the ground state wave function
of the harmonic osciallator ϕ̃(k). (Recall that each Fourier mode of a free scalar field
is a harmonic oscillator). Note that no divergences arise in the above construction.
By linearizing (149) about ϕ0 (t) we obtain the perturbation of the energy-
momentum tensor. In particular, the energy density fluctuation δ ρ̃(k) is given by
˙
δ ρ̃(k) = ϕ̇0 δ ϕ̃(k) + V ′ (ϕ0 )δ ϕ̃(k) . (152)
To obtain the initial amplitude (39) of δM/M, the above is to be evaluated at the
time ti (k).
The computation of the spectrum of density perturbations produced in the de
Sitter phase has been reduced to the evaluation of the expectation value (151). First,
we must specify the state |ψ >. (Recall that in non-Minkowski space-times there is
no uniquely defined vacuum state of a quantum field theory [51] ). We pick the FRW
frame of the pre-inflationary period. In this frame, the number density of particles
decreases exponentially. Hence we choose |ψ > to be the vacuum state in this frame
(see [88] for a discussion of other choices). ψ[ϕ̃(k), t], the wave functional of |ψ >,
can be calculated explicitly. It is basically the superposition of the ground state wave
functions for all oscillators
1
Z
ψ[ϕ̃(k), t] = N exp − (2π)−3 a3 (t) d3 kω(k, t)|ϕ̃(k)|2 . (153)
2
N is a normalization constant and ω(k, t) ∼ H at t = ti (k). Hence
δ ϕ̃(k, t) = (2π)3/2 a−3/2 ω(k, t)−1/2 ∼ (2π)3/2 k −3/2 H , t = ti (k) . (154)
Given the above determination of the intitial amplitude of density perturbations

at the time when they leave the Hubble radius during the de Sitter phase, and the
general relativistic analysis of the evolution of fluctuations discussed in the previous
subsection, it is easy to evaluate the r.m.s. inhomogeneities when they reenter the
Hubble radius at time tf (k).
First, we combine (152), (154), (39), (87) and (91) to obtain
!2 !2 !2
δM 3 V ′ (ϕ0 )δ ϕ̃(k, ti (k)) V ′ (ϕ0 )H
(k, ti (k)) ∼ k ∼ . (155)
M ρ0 ρ0
If the background scalar field is rolling slowly, then
V ′ (ϕ0 (ti (k))) = 3H|ϕ̇0 (ti (k))| . (156)
and
2
(1 + p/ρ)(ti (k)) ≃ ρ−1
0 ϕ̇0 (ti (k)) . (157)
Combining (155), (156), (157) and (40) we get
42
δM 3H 2 |ϕ̇0 (ti (k))| 3H 2
(k, tf (k)) ∼ = (158)
M ϕ̇20 (ti (k)) |ϕ̇0 (ti (k))|
This result can now be evaluated for specific models of inflation to find the conditions
on the particle physics parameters which give a value
δM
(k, tf (k)) ∼ 10−5 (159)
M
which is required if quantum fluctuations from inflation are to provide the seeds for
galaxy formation and agree with the CMB anisotropy limits.
For chaotic inflation with a potential
1
V (ϕ) = m2 ϕ2 , (160)
2
we can solve the slow rolling equations for the inflaton to obtain
δM m
(k, tf (k)) ∼ 10 (161)
M mpl
which implies that m ∼ 1013 GeV to agree with (159).

Similarly, for a quartic potential
1
V (ϕ) = λϕ4 (162)
4
we obtain
δM
(k, tf (k)) ∼ 10 · λ1/2 (163)
M
which requires λ ≤ 10−12 in order not to conflict with observations.
The conditions (161) and (163) require the presence of small parameters in the
particle physics model. It has been shown quite generally [38] that small parameters
are required if inflation is to solve the fluctuation problem.
I have chosen to present the analysis of fluctuations in inflationary cosmology in
two separate steps in order to highlight the crucial physics issues. Having done this,
it is possible to step back and construct a unified analysis of the quantum generation
and classical evolution of perturbations in an inflationary Universe (for a detailed
review see [34]).
The basic point is that at the linearized level, the equations describing both grav-
itational and matter perturbations can be quantized in a consistent way. The use of
gauge invariant variables makes the analysis both physically clear and computation-
ally simple.
The first step of this analysis is to consider the action for the linear perturbations
in a background homogeneous and isotropic Universe, and to express the result in
43
terms of gauge invariant variables describing the fluctuations. Focusing on the scalar
perturbations, it turns out that after a lot of algebra the action reduces to the action
of a single gauge invariant free scalar field with a time dependent mass [89,90] (the time
dependence relects the expansion of the background space-time). This result is not
suprising. Based on the study of classical cosmological perturbations, we know that
there is only one field degree of freedom for the scalar perturbations (for matter the-
ories which obey (141)). Since at the linearized level there are no mode interactions,
the action for this field must be that of a free scalar field.
The action thus has the same form as the action for a scalar matter field in a time
dependent gravitational or electromagnetic background, and we can use standard
methods to quantize this theory (see e.g. [51]). If we employ canonical quantization,
then the mode functions of the field operator obey the same classical equations as we
derived in the gauge-invariant analysis of relativistic perturbations.
The time dependence of the mass is relected in the nontrivial form of the solutions
of the mode equations. The mode equations have growing modes which correspond
to particle production or equivalently to the generation and amplification of fluctua-
tions. We can start the system off (e.g. at the beginning of inflation) in the vacuum
state (defined as a state with no particles with respect to a local comoving observer).
The state defined this way will not be the vacuum state from the point of view of
an observer at a later time. The Bogoliubov mode mixing technique can be used to
calculate the number density of perticles at a later time. In particular, expectation
values of field operators such as the power spectrum can be computed, and the re-
sults agree with those obtained using the heuristic analysis at the beginning of this
subsection.
E. Summary
To summarize the main results of the analysis of density fluctuations in inflationary

cosmology:
1. Quantum vacuum fluctuations in the de Sitter phase of an inflationary Universe

are the source of perturbations.
2. The quantum perturbations decohere on scales outside the Hubble radius and
can hence be treated classically.
3. As a consequence of the change in the background equation of state, the clas-

sical evolution outside the Hubble radius produces a large amplification of the
perturbations. In fact, unless the particle physics model contains very small
coupling constants, the predicted fluctuations are in excess of those allowed by
the bounds on cosmic microwave anisotropies.
44
4. The quantum generation and classical evolution of fluctuations can be treated
in a unified manner. The formalism is no more complicated that the study of a
free scalar field in a time dependent background.
5. Inflationary Universe models generically produce an approximately scale invari-

ant Harrison-Zel’dovich spectrum
δM
(k, tf (k)) ≃ const. (164)
M
It is not hard to construct models which give a different spectrum. All that is
required is a significant change in H during the period of inflation.
Once inside the Hubble radius, the evolution of the mass perturbations is influ-
enced by the damping effects discussed in Section 3.B, which in turn depend on the
composition of the dark matter.
On scales which enter the Hubble radius before teq , the perturbations can only
grow logarithmically in time between tf (k) and teq . This implies that (up to loga-
rithmic corrections), the mass perturbation spectrum is flat for wavelengths smaller
than λeq , the comoving Hubble radius at teq :
δM
(λ, t) ≃ const, t ≤ teq , λ < λeq , (165)
M
whereas on larger scales
δM
(λ, t) ∝ λ−2 . (166)
M
Equations (165) and (166) give the power spectrum in an Ω = 1 inflationary CDM
model. If the dark matter is hot, then neutrino free streaming cuts off the power
spectrum at λmax
J (see (127)).
IV. TOPOLOGICAL DEFECTS, STRUCTURE FORMATION AND

BARYOGENESIS
A. Quantifying Data on Large-Scale Structure
It is length scales corresponding to galaxies and larger which are of greatest inter-
est in cosmology when attempting to find an imprint of the primordial fluctuations
produced by particle physics. On these scales, gravitational effects are assumed to
be dominant, and the fluctuations are not too far from the linear regime. On smaller
scales, nonlinear gravitational and hydrodynamical effects determine the final state
and mask the initial perturbations.
45
To set the scales, consider the mean separation of galaxies, which is about 5h−1
Mpc [91] , and that of Abell clusters which is around 25h−1 Mpc [92] . The largest
coherent structures seen in current redshift surveys have a length of about 100h−1
Mpc [3,4] , the recent detections of CMB anisotropies probe the density field on length
scales of about 103 h−1 Mpc, and the present horizon corresponds to a distance of
about 3 · 103 h−1 Mpc.
Galaxies are gravitationally bound systems containing billions of stars. They are
non-randomly distributed in space. A quantitative measure of this non-randomness
is the “two-point correlation function” ξ2 (r) which gives the excess probability of
finding a galaxy at a distance r from a given galaxy:
n(r) − n0
ξ2 (r) =< > . (167)
n0
Here, n0 is the average number density of galaxies, and n(r) is the density of galaxies
a distance r from a given one. The pointed braces stand for ensemble averaging.
Recent observational results from a various galaxy redshift surveys yield reason-
able agreement [93] with a form
γ
r0

ξ2 (r) ≃ (168)
r
with scaling length r0 ≃ 5h−1 Mpc and power γ ≃ 1.8. A theory of structure
formation must explain both the amplitude and the slope of this correlation function.
On scales larger than galaxies, a better way to quantify structure is by means of
large-scale systematic redshift surveys. Such surveys have discovered coherent planar
structures and voids on scales of up to 100h−1 Mpc. Fig. 12 is a sketch of redshift
versus angle α in the sky for one 6o slice of the sky [3] . The second direction in
the sky has been projected onto the α − z plane. The most prominent feature is
the band of galaxies at a distance of about 100h−1 Mpc. This band also appears in
neighboring slices and is therefore presumably part of a planar density enhancement
of comoving planar size of at least (50 × 100) × h−2 Mpc2 . This structure is often
called the “Great Wall.” The challenge for theories of structure formation is not only
to explain the fact that galaxies are nonrandomly distributed, but also to predict
both the observed scale and topology of the galaxy distribution. Topological defect
models of structure formation attempt to address these questions.
Until 1992 there was little evidence for any convergence of the galaxy distribution
towards homogeneity. Each new survey led to the discovery of new coherent structures
in the Universe on a scale comparable to that of the survey. In 1996, results of a much
deeper redshift survey were published [4] which for the first time find no coherent
structures on the scale of the entire survey. In fact, so coherent structures on scales
larger than 100h−1 Mpc are seen. This is the first direct evidence for the cosmological
principle from optical surveys (the isotropy of the CMB has for a long time been a
strong point in its support).
46
FIG. 12. Results from the CFA redshift survey. Radial distance gives the redshift of
galaxies, the angular distance corresponds to right ascension. The results from several slices
of the sky (at different declinations) have been projected into the same cone.
In summary, a lot of data from optical and infrared galaxies are currently available,
and new data are being collected at a rapid rate. The observational constraints on
theories of structure formation are becoming tighter.
B. Toplogical Defects
According to particle physics theories, matter at high energies and temperatures

must be described in terms of fields. Gauge symmetries have proved to be extremely
useful in describing the standard model of particle physics, according to which at
high energies the laws of nature are invariant under a nonabelian group G of internal
symmetry transformations G = SU(3)c × SU(2)L × U(1)Y which at a temperature of
about 200 MeV is spontaneously broken down to G′ = SU(3)c × U(1). The subscript
on the SU(3) subgroup indicates that it is the color symmetry group of the strong
interactions, SU(2)L × U(1)Y is the Glashow-Weinberg-Salam (WS) model of weak
and electromagnetic interactions, the subscripts L and Y denoting left handedness
and hypercharge respectively. At low energies, the WS model spontaneously breaks
to the U(1) subgroup of electromagnetism.
Spontaneous symmetry breaking is induced by an order parameter ϕ taking on a
nontrivial expectation value < ϕ > below a certain temperture Tc . In some particle
physics models, ϕ is a fundamental scalar field in a nontrivial representation of the
gauge group G which is broken. However, ϕ could also be a fermion condensate, as
in the BCS theory of superconductivity.
47
Earlier we have seen that symmetry breaking phase transitions in gauge field
theories do not, in general, lead to inflation. In most models, the coupling constants
which arise in the effective potential for the scalar field ϕ driving the phase transition
are too large to generate a period of slow rolling which lasts more than one Hubble
time H −1 (t). Nevertheless, there are interesting remnants of the phase transition:
topological defects.
Consider a single component real scalar field with a typical symmetry breaking
potential
1
V (ϕ) = λ(ϕ2 − η 2 )2 (169)
4
Unless λ ≪ 1 there will be no inflation. The phase transition will take place on a
short time scale τ < H −1 , and will lead to correlation regions of radius ξ < t inside of
which ϕ is approximately constant, but outside of which ϕ ranges randomly over the
vacuum manifold M, the set of values of ϕ which minimizes V (ϕ) – in our example
ϕ = ±η. The correlation regions are separated by domain walls, regions in space
where ϕ leaves the vacuum manifold M and where, therefore, potential energy is
localized. Via the usual gravitational force, this energy density can act as a seed for
structure.
Topological defects are familiar from solid state and condensed matter systems.
Crystal defects, for example, form when water freezes or when a metal crystallizes [94] .
Point defects, line defects and planar defects are possible. Defects are also common
in liquid crystals [95] . They arise in a temperature quench from the disordered to the
ordered phase. Vortices in 4 He are analogs of global cosmic strings. Vortices and other
defects are also produced [96] during a quench below the critical temperature in 3 He.
Finally, vortex lines may play an important role in the theory of superconductivity [97] .
The analogies between defects in particle physics and condensed matter physics
are quite deep. Defects form for the same reason: the vacuum manifold is topologi-
cally nontrivial. The arguments [98] which say that in a theory which admits defects,
such defects will inevitably form, are applicable both in cosmology and in condensed
matter physics. Different, however, is the defect dynamics. The motion of defects
in condensed matter systems is friction-dominated, whereas the defects in cosmology
obey relativistic equations, second order in time derivatives, since they come from a
relativistic field theory.
After these general comments we turn to a classification of topological defects [98] .
We consider theories with an n-component order parameter ϕ and with a potential
n
energy function (free energy density) of the form (6.1) with ϕ2 = ϕ2i .
P
i=1
There are various types of local and global topological defects (regions of trapped
energy density) depending on the number n of components of ϕ (see e.g. [99] for a
comprehensive survey of topological defect models). The more rigorous mathematical
definition refers to the homotopy of M. The words “local” and “global” refer to
whether the symmetry which is broken is a gauge or global symmetry. In the case of
48
local symmetries, the topological defects have a well defined core outside of which ϕ
contains no energy density in spite of nonvanishing gradients ∇ϕ: the gauge fields Aµ
can absorb the gradient, i.e., Dµ ϕ = 0 when ∂µ ϕ 6= 0, where the covariant derivative
Dµ is defined by Dµ = ∂µ + ie Aµ , e being the gauge coupling constant. Global
topological defects, however, have long range density fields and forces.
Table 1 contains a list of topological defects with their topological characteristics.
A “v” marks acceptable theories, a “x” theories which are in conflict with observations
(for η ∼ 1016 GeV).
Theories with domain walls are ruled out [100] since a single domain wall stretching
across the Universe today would overclose the Universe. Local monopoles are also
ruled out [101] since they would overclose the Universe. Local textures are ineffective
at producing structures because there is no traped potential energy.
From now on we will focus on one type of defects: cosmic strings (see e.g.
[99,102,103] for recent reviews, and [104] for a classic review paper). These arise
in theories with a complex order parameter (n = 2). In this case the vacuum mani-
fold of the model is
M = S1 , (170)
which has nonvanishing first homotopy group:
Π1 (M) = Z 6= 1 . (171)
A cosmic string is a line of trapped energy density which arises whenever the field ϕ(x)
circles M along a closed path in space (e.g., along a circle). In this case, continuity
of ϕ implies that there must be a point with ϕ = 0 on any disk whose boundary is
the closed path. The points on different sheets connect up to form a line overdensity
of field energy.
To construct a field configuration with a string along the z axis [105] , take ϕ(x) to
cover M along a circle with radius r about the point (x, y) = (0, 0):
49
ϕ(r, ϑ) ≃ ηeiϑ , r ≫ η −1 . (172)
This configuration has winding number 1, i.e., it covers M exactly once. Maintaining
cylindrical symmetry, we can extend (172) to arbitrary r
ϕ(r, ϑ) = f (r)eiϑ , (173)
where f (0) = 0 and f (r) tends to η for large r. The width w can be found by
balancing potential and tension energy. The result is
w ∼ λ−1/2 η −1 . (174)
For local cosmic strings, i.e., strings arising due to the spontaneous breaking of
a gauge symmetry, the energy density decays exponentially for r ≫ w. In this case,
the energy µ per unit length of a string is finite and depends only on the symmetry
breaking scale η
µ ∼ η2 (175)
(independent of the coupling λ). The value of µ is the only free parameter in a cosmic
string model.
C. Formation of Defects in Cosmological Phase Transitions
The symmetry breaking phase transition takes place at T = Tc (called the critical
temperature). From condensed matter physics it is well known that in many cases
topological defects form during phase transitions, particularly if the transition rate
is fast on a scale compared to the system size. When cooling a metal, defects in
the crystal configuration will be frozen in; during a temperature quench of 4 He, thin
vortex tubes of the normal phase are trapped in the superfluid; and analogously in a
temperature quench of a superconductor, flux lines are trapped in a surrounding sea
of the superconducting Meissner phase.
In cosmology, the rate at which the phase transition proceeds is given by the
expansion rate of the Universe. Hence, topological defects will inevitably be produced
in a cosmological phase transition [98] , provided the underlying particle physics model
allows such defects.
The argument which ensures that in theories which admit topological or semitopo-
logical defects, such defects will be produced during a phase transition in the very
early Universe is called the Kibble mechanism [98] . To illustrate the physics, consider
a mechanical toy model, first introduced by Mazenko, Unruh and Wald [106] . Take
(see Fig. 13) a lattice of points on a flat table. At each point, a pencil is pivoted.
It is free to rotate and oscillate. The tips of nearest neighbor pencils are connected
with springs (to mimic the spatial gradient terms in the scalar field Lagrangean).
50
Newtonian gravity creates a potential energy V (ϕ) for each pencil (ϕ is the angle
relative to the vertical direction). V (ϕ) is minimized for |ϕ| = η (in our toy model
η = π/2). Hence, the Lagrangean of this pencil model is analogous to that of a scalar
field with symmetry breaking potential (169).
FIG. 13. The pencil model: the potential energy of a simple pencil has the same form
as that of scalar fields used for spontaneous symmetry breaking. The springs connecting
nearest neighbor pencils give rise to contributions to the energy which mimic spatial gradient
terms in field theory.
At high temperatures T ≫ Tc , all pencils undergo large amplitude high frequency

oscillations. However, by causality, the phases of oscillation of pencils with large
separation s are uncorrelated. For a system in thermal equilibrium, the length s
beyond which phases are random is the correlation length ξ(t). However, since the
system is quenched rapidly, there is a causality bound on ξ:
ξ(t) < t , (176)
where t is the causal horizon.
The critical temperature Tc is the temperature at which the thermal energy is
equal to the energy a pencil needs to jump from horizontal to vertical position. For
T < Tc , all pencils want to lie flat on the table. However, their orientations are
random beyond a distance ξ(t) determined by equating the free energy gained by
symmetry breaking (a volume effect) with the gradient energy lost (a surface effect).
As expected, ξ(T ) diverges at Tc . Very close to Tc , the thermal energy T is larger
than the volume energy gain Ecorr in a correlation volume. Hence, these domains
are unstable to thermal fluctuations. As T decreases, the thermal energy decreases
more rapidly than Ecorr . Below the “Ginsburg temperature” TG , there is insufficient
thermal energy to excite a correlation volume into the state ϕ = 0. Domains of size
ξ(tG ) ∼ λ−1 η −1 (177)
freeze out [98,107] . The boundaries between these domains become topological defects.
An improved version of this argument has recently been given by Zurek [108] (see also
[109]).
51
We conclude that in a theory in which a symmetry breaking phase transitions
satisfies the topological criteria for the existence of a fixed type of defect, a network of
such defects will form during the phase transition and will freeze out at the Ginsburg
temperature. The correlation length is initially given by (177), if the field ϕ is in
thermal equilibrium before the transition. Independent of this last assumption, the
causality bound implies that ξ(tG ) < tG .
For times t > tG the evolution of the network of defects may be complicated (as
for cosmic strings) or trivial (as for textures). In any case (see the caveats of [110]),
the causality bound persists at late times and states that even at late times, the mean
separation and length scale of defects is bounded by ξ(t) ≤ t.
Applied to cosmic strings, the Kibble mechanism implies that at the time of the
phase transition, a network of cosmic strings with typical step length ξ(tG ) will form.
According to numerical simulations [111] , about 80% of the initial energy is in infinite
strings (strings with curvature radius larger than the Hubble radius) and 20% in
closed loops.
D. Evolution of Strings and Scaling
FIG. 14. Formation of a loop by a self intersection of an infinite string. According to the
original cosmic string scenario, loops form with a radius R determined by the instantaneous
coherence length of the infinite string network.
The evolution of the cosmic string network for t > tG is complicated. The key
processes are loop production by intersections of infinite strings (see Fig. 14) and
loop shrinking by gravitational radiation. These two processes combine to create a
mechanism by which the infinite string network loses energy (and length as measured
in comoving coordinates). It can be shown (see e.g. [104]) that, as a consequence, the
correlation length of the string network is always proportional to its causality limit
ξ(t) ∼ t . (178)
Hence, the energy density ρ∞ (t) in long strings is a fixed fraction of the background
energy density ρc (t)
52
ρ∞ (t) ∼ µξ(t)−2 ∼ µt−2 (179)
or
ρ∞ (t)
∼ Gµ . (180)
ρc (t)
We conclude that the cosmic string network approaches a “scaling solution” in

which the statistical properties of the network are time independent if all distances
are scaled to the horizon distance.
E. Cosmic Strings and Structure Formation
The starting point of the structure formation scenario in the cosmic string theory
is the scaling solution for the cosmic string network, according to which at all times
t (in particular at teq , the time when perturbations can start to grow) there will be a
few long strings crossing each Hubble volume, plus a distribution of loops of radius
R ≪ t (see Fig. 15).
The cosmic string model admits three mechanisms for structure formation: loops,
filaments, and wakes. Cosmic string loops have the same time averaged field as a
point source with mass [112] M(R) = βRµ, R being the loop radius and β ∼ 2π.
Hence, loops will be seeds for spherical accretion of dust and radiation.
For loops with R ≤ teq , growth of perturbations in a model dominated by cold
dark matter starts at teq . Hence, the mass at the present time will be
M(R, t0 ) = z(teq )β Rµ . (181)
FIG. 15. Sketch of the scaling solution for the cosmic string network. The box corre-
sponds to one Hubble volume at arbitrary time t.
53
In the original cosmic string model [2,113] it was assumed that loops dominate over
wakes. However, according to the newer cosmic string evolution simulations [114] , most
of the energy in strings is in the long strings, and hence the loop accretion mechanism
is subdominant.
The second mechanism involves long strings moving with relativistic speed in
their normal plane which give rise to velocity perturbations in their wake [115] . The
mechanism is illustrated in Fig. 16: space normal to the string is a cone with deficit
angle [116]
α = 8πGµ . (182)
If the string is moving with normal velocity v through a bath of dark matter, a
velocity perturbation
δv = 4πGµvγ (183)
[with γ = (1 − v 2 )−1/2 ] towards the plane behind the string results. At times af-
ter teq , this induces planar overdensities, the most prominent (i.e., thickest at the
present time) and numerous of which were created at teq , the time of equal mat-
ter and radiation [117–119] . The corresponding planar dimensions are (in comoving
coordinates)
teq z(teq ) × teq z(teq )v ∼ (40 × 40v) Mpc2 . (184)
FIG. 16. Sketch of the mechanism by which a long straight cosmic string S moving
with velocity v in transverse direction through a plasma induces a velocity perturbation ∆v
towards the wake. Shown on the left is the deficit angle, in the center is a sketch of the
string moving in the plasma, and on the right is the sketch of how the plasma moves in the
frame in which the string is at rest.
The thickness d of these wakes can be calculated using the Zel’dovich

approximation [120] . The result is (for Gµ = 10−6 )
d ≃ Gµvγ(v)z(teq )2 teq ≃ 4v Mpc . (185)

Wakes arise if there is little small scale structure on the string. In this case, the
string tension equals the mass density, the string moves at relativistic speeds, and
there is no local gravitational attraction towards the string.
54
In contrast, if there is small scale structure on strings, then the string tension T is
smaller [121] than the mass per unit length µ , and thus there is a gravitational force
towards the string which gives rise to cylindrical accretion, producing filaments [122] .
Which of the mechanisms – filaments or wakes – dominates is determined by the
competition between the velocity induced by the Newtonian gravitational potential
of the strings and the velocity perturbation of the wake.
The cosmic string model predicts a scale-invariant spectrum of density pertur-
bations, exactly like inflationary Universe models but for a rather different reason.
Consider the r.m.s. mass fluctuations on a length scale 2πk −1 at the time tH (k) when
this scale enters the Hubble radius. From the cosmic string scaling solution it follows
that a fixed (i.e., tH (k) independent) number ṽ of strings of length of the order tH (k)
contribute to the mass excess δM(k, tH (k)). Thus
δM ṽµtH (k)
(k, tH (k)) ∼ −1 −2 ∼ ṽ Gµ . (186)
M G tH (k)t3H (k)
Note that the above argument predicting a scale invariant spectrum will hold for all
topological defect models which have a scaling solution, in particular also for global
monopoles and textures.
The amplitude of the r.m.s. mass fluctuations (equivalently: of the power spec-
trum) can be used to normalize Gµ. Since today on galaxy cluster scales
δM
(k, t0 ) ∼ 1 , (187)
M
the growth rate of fluctuations linear in a(t) yields
δM
(k, teq ) ∼ 10−4 , (188)
M
and therefore, using ṽ ∼ 10,
Gµ ∼ 10−5 . (189)
A similar value is obtained by normalizing the model to the COBE amplitude of CMB
anisotropies on large angular scales [123,124] (the normalizations from COBE and from
the power spectrum of density perturbations on large scales agree to within a factor
of 2). Thus, if cosmic strings are to be relevant for structure formation, they must
arise due to a symmetry breaking at energy scale η ≃ 1016 GeV. This scale happens to
be the scale of unification (GUT) of weak, strong and electromagnetic interactions.
It is tantalizing to speculate that cosmology is telling us that there indeed was new
physics at the GUT scale.
A big advantage of the cosmic string model over inflationary Universe models
is that HDM is a viable dark matter candidate. Cosmic string loops survive free
streaming, as discussed in Section 3.B, and can generate nonlinear structures on
55
galactic scales, as discussed in detail in [125,126]. Accretion of hot dark matter by a
string wake was studied in [119]. In this case, nonlinear perturbations develop only
late. At some time tnl , all scales up to a distance qmax from the wake center go
nonlinear. Here
qmax ∼ Gµvγ(v)z(teq )2 teq ∼ 4v Mpc , (190)
and it is the comoving thickness of the wake at tnl . Demanding that tnl corresponds
to a redshift greater than 1 leads to the constraint
Gµ > 5 · 10−7 . (191)
Note that in a cosmic string and hot dark matter model, wakes form nonlinear struc-
tures only very recently. Accretion onto loops and small scale structure on the long
strings provide two mechanisms which may lead to high redshift objects such as
quasars and high redshift galaxies. The first mechanism has recently been studied in
[127], the second in [128,129].
The power spectrum of density fluctuations in a cosmic string model with HDM
has recently been studied numerically by Mähönen [130] , based on previous work of
[131] (see also [132] for an earlier semi-analytical study). The spectral shape agrees
quite well with observations, and a bias factor of less than 2 is required to give the
best-fit amplitude for a COBE normalized model. Note, however, that the results
depend quite sensitively on the details of the string scaling solution which are at
present not well understood.
Due to lack of space, I will not discuss the global monopole [133] and global
texture [134] models of structure formation. There has been a lot of work on the
texture model, and the reader is referred to [136,135] for recent review articles.
F. Signatures for Strings
The cosmic string theory of structure formation makes several distinctive predic-
tions, both in terms of the galaxy distribution and in terms of CMB anisotropies. On
large scales (corresponding to the comoving Hubble radius at teq and larger, struc-
ture is predicted to be dominated either by planar [117–119] or filamentary [122] galaxy
concentrations. For models in which the strings have no local gravity, the resulting
nonlinear structures will look very different from the nonlinear structures in models
in which local gravity is the dominant force. As discovered and discussed recently in
[137], a baryon number excess is predicted in the nonlinear wakes. This may explain
the “cluster baryon crisis” [138] , the fact that the ratio of baryons to dark matter in
rich clusters is larger than what is compatible with the nucleosynthesis constraints in
a spatially flat Universe.
56
FIG. 17. Sketch of the Kaiser-Stebbins effect by which cosmic strings produce linear
discontinuities in the CMB. Photons γ passing on different sides of a moving string S
(velocity v) towards the observer O receive a relative Doppler shift due to the conical
nature of space perpendicular to the string (deficit angle α).
As described in the previous subsection, space perpendicular to a long straight

cosmic string is conical with deficit angle given by (182). Consider now CMB radiation
approaching an observer in a direction normal to the plane spanned by the string and
its velocity vector (see Fig. 17). Photons arriving at the observer having passed on
different sides of the string will obtain a relative Doppler shift which translates into
a temperature discontinuity of amplitude [139]
δT
= 4πGµvγ(v) , (192)
T
where v is the velocity of the string. Thus, the distinctive signature for cosmic strings
in the microwave sky are line discontinuities in T of the above magnitude.
Given ideal maps of the CMB sky it would be easy to detect strings. However,
real experiments have finite beam width. Taking into account averaging over a scale
corresponding to the beam width will smear out the discontinuities, and it turns out
to be surprisingly hard to distinguish the predictions of the cosmic string model from
that of inflation-based theories using quantitative statistics which are easy to evaluate
analytically, such as the kurtosis of the spatial gradient map of the CMB [140] . There
may be ways to distinguish between string and inflationary models by looking at the
angular power spectrum of CMB anisotropies. Work on this subject, however, is still
controversial [141–143] .
Global textures also produce distinctive non-Gaussian signatures [144] in CMB
maps. In fact, these signatures are more pronounced and on larger scales than the
signatures in the cosmic string model.
57
G. Principles of Baryogenesis
Baryogenesis is another area where particle physics and cosmology connect in a

very deep way. The goal is to explain the observed asymmetry between matter and
antimatter in the Universe. In particular, the objective is to be able to explain the
observed value of the net baryon to entropy ratio at the present time
∆nB
(t0 ) ∼ 10−10 (193)
s
starting from initial conditions in the very early Universe when this ratio vanishes.
Here, ∆nB is the net baryon number density and s the entropy density.
As pointed out by Sakharov [145] , three basic criteria must be satisfied in order to
have a chance at explaining the data:
1. The theory describing the microphysics must contain baryon number violating
processes.
2. These processes must be C and CP violating.
3. The baryon number violating processes must occur out of thermal equilibrium.
As was discovered in the 1970’s [146] , all three criteria can be satisfied in GUT
theories. In these models, baryon number violating processes are mediated by su-
perheavy Higgs and gauge particles. The baryon number violation is visible in the
Lagrangian, and occurs in perturbation theory (and is therefore in principle easy to
calculate). In addition to standard model CP violation, there are typically many
new sources of CP violation in the GUT sector. The third Sakharov condition can
also be realized: After the GUT symmetry-breaking phase transition, the superheavy
particles may fall out of thermal equilibrium. The out-of-equilibrium decay of these
particles can thus generate a nonvanishing baryon to entropy ratio.
The magnitude of the predicted nB /s depends on the asymmetry ε per decay, on
the coupling constant λ of the nB violating processes, and on the ratio nX /s of the
number density nX of superheavy Higgs and gauge particles to the number density
of photons, evaluated at the time td when the baryon number violating processes
fall out of thermal equilibrium, and assuming that this time occurs after the phase
transition. The quantity ε is proportional to the CP-violation parameter in the model.
In a GUT theory, this CP violation parameter can be large (order 1), whereas in the
standard electroweak theory it is given by the CP violating phases in the CKM mass
matrix and is very small. As shown in [146] it is easily possible to construct models
which give the right nB /s ratio after the GUT phase transition (for recent reviews of
baryogenesis see [147] and [148]).
58
H. GUT Baryogenesis and Topological Defects
The ratio nB /s, however, does not only depend on ε, but also on nX /s(td ). If the
temperature Td at the time td is greater than the mass mX of the superheavy particles,
then it follows from the thermal history in standard cosmology that nX ∼ s. However,
if Td < mX , then the number density of X particles is diluted exponentially in the
time interval between when T = mX and when T = Td . Thus, the predicted baryon
to entropy ratio is exponentially suppressed:
nB 1
∼ ∗ λ2 εe−mX /Td , (194)
s g
where g ∗ is the number of spin degrees of freedom in thermal equilibrium at the time
of the phase transition. In this case, the standard GUT baryogenesis mechanism is
ineffective.
However, topological defects may come to the rescue [149] . As we have seen ear-
lier in this section, topological defects will inevitably be produced in the symmetry
breaking GUT transition provided they are topologically allowed in that symmetry
breaking scheme. The topological defects provide an alternative mechanism of GUT
baryogenesis.
Inside of topological defects, the GUT symmetry is restored. In fact, the defects
can be viewed as solitonic configurations of X particles. The continuous decay of
defects at times after td provides an alternative way to generate a nonvanishing baryon
to entropy ratio. The defects constitute out of equilibrium configurations, and hence
their decay can produce a nonvanishing nB /s in the same way as the decay of free X
quanta.
The way to compute the resulting nB /s ratio is as follows: The defect scaling
solution gives the energy density in defects at all times. Taking the time derivative
of this density, and taking into account the expansion of the Universe, we obtain the
loss of energy attributed to defect decay. By energetics, we can estimate the number
of decays of individual quanta which the defect decay corresponds to. We can then
use the usual perturbative results to compute the resulting net baryon number.
Provided that mX < Td , then at the time when the baryon number violating
processes fall out of equilibrium (when we start generating a nonvanishing nB ) the
energy density in free X quanta is much larger than the defect density, and hence
the defect-driven baryogenesis mechanism is subdominant. However, if mX > Td ,
then as indicated in (194), the energy density in free quanta decays exponentially. In
constrast, the density in defects only decreases as a power of time, and hence soon
dominates baryogenesis.
One of the most important ingredients in the calculation is the time dependence
of ξ(t), the separation between defects. Immediately after the phase transition at the
time tf of the formation of the defect network, the separation is ξ(tf ) ∼ λ−1 η −1 . In
the time period immediately following, the time period of relevance for baryogenesis,
ξ(t) approaches the Hubble radius according to the equation [107]
59
t 5/4
ξ(t) ≃ ξ(tf )( ) . (195)
tf
Using this result to calculate the defect density, we obtain after some algebra
nB Td nB
|defect ∼ λ2 |0 , (196)
s η s
where nB /s|0 is the unsuppressed value of nB /s which can be obtained using the
standard GUT baryogenesis mechanism. We see from (196) that even for low values
of Td , the magnitude of nB /s which is obtained via the defect mechanism is only
suppressed by a power of Td . However, the maximum strength of the defect channel
is smaller than the maximum strength of the usual mechanism by a geometrical sup-
pression factor λ2 which expresses the fact that even at the time of defect formation,
the defect network only occupies a small volume.
I. Electroweak Baryogenesis and Topological Defects
It has been known for some time that there are baryon number violating processes
even in the standard electroweak theory. These processes are, however, nonpertur-
bative. They are connected with the t’Hooft anomaly [150] , which in turn is due to
the fact that the gauge theory vacuum is degenerate, and that the different degen-
erate vacuum states have different quantum numbers (Chern-Simons numbers). In
theories with fermions, this implies different baryon number. Configurations such as
sphalerons [151] which interpolate between two such vacuum states thus correspond to
baryon number violating processes.
As pointed out in [152], the anomalous baryon number violating processes are in
thermal equilibrium above the electroweak symmetry breaking scale. Therefore, any
net baryon to entropy ratio generated at a higher scale will be erased, unless this
ratio is protected by an additional quantum number such as a nonvanishing B − L
which is conserved by electroweak processes.
However, as first suggested in [153] and discussed in detail in many recent papers
(see [154] for reviews of the literature), it is possible to regenerate a nonvanishing
nB /s below the electroweak symmetry breaking scale. Since there are nB violating
processes and both C and CP violation in the standard model, Sakharov’s conditions
are satisfied provided that one can realize an out-of-equilibrium state after the phase
transition. Standard model CP violation is extremely weak. Thus, it appears neces-
sary to add some sector with extra CP violation to the standard model in order to
obtain an appreciable nB /s ratio. A simple possibility which has been invoked often
is to add a second Higgs doublet to the theory, with CP violating relative phases.
The standard way to obtain out-of-equilibrium baryon number violating processes
immediately after the electroweak phase transition is [154] to assume that the transi-
tion is strongly first order and proceeds by the nucleation of bubbles (note that these
are two assumptions, the second being stronger than the first!).
60
Bubbles are out-of-equilibrium configurations. Outside of the bubble (in the false
vacuum), the baryon number violating processes are unsuppressed, inside they are
exponentially suppressed. In the bubble wall, the Higgs fields have a nontrivial pro-
file, and hence (in models with additional CP violation in the Higgs sector) there is
enhanced CP violation in the bubble wall. In order to obtain net baryon production,
one may either use fermion scattering off bubble walls [155] (because of the CP viola-
tion in the scattering, this generates a lepton asymmetry outside the bubble which
converts via sphalerons to a baryon asymmetry) or sphaleron processes in the bubble
wall itself [156,157] . It has been shown that, using optimistic parameters (in particular
a large CP violating phase ∆θCP in the Higgs sector) it is possible to generate the
observed nB /s ratio. The resulting baryon to entropy ratio is of the order
nB mt
2
∼ αW (g ∗ )−1 ( )2 ∆θCP , (197)
s T
where αW refers to the electroweak interaction strength, g ∗ is the number of spin
degrees of freedom in thermal quilibrium at the time of the phase transition, and mt
is the top quark mass. The dependence on the top quark mass enters because net
baryogenesis only appears at the one-loop level.
However, analytical and numerical studies show that, for the large Higgs masses
which are indicated by the current experimental bounds, the electroweak phase tran-
sition will unlikely be sufficiently strongly first order to proceed by bubble nucleation.
In addition, there are some concerns as to whether it will proceed by bubble nucleation
at all (see e.g. [158]).
Once again, topological defects come to the rescue. In models which admit de-
fects, such defects will inevitably be produced in a phase transition independent of
its order. Moving topological defects can play the same role in baryogenesis as nucle-
ating bubbles. In the defect core, the electroweak symmetry is unbroken and hence
sphaleron processes are unsuppressed [159] . In the defect walls there is enhanced CP
violation for the same reason as in bubble walls. Hence, at a fixed point in space, a
nonvanishing baryon number will be produced when a topological defect passes by.
Defect-mediated electroweak baryogenesis has been worked out in detail in [160]
(see [161] for previous work) in the case of cosmic strings. The scenario is as follows:
at a particular point x in space, antibaryons are produced when the front side of the
defect passes by. While x is in the defect core, partial equilibration of nB takes place
via sphaleron processes. As the back side of the defect passes by, the same number of
baryons are produced as the number of antibaryons when the front side of the defect
passes by. Thus, at the end a positive number of baryons are left behind.
As in the case of defect-mediated GUT baryogenesis, the strength of defect-
mediated electroweak baryogenesis is suppressed by the ratio SF of the volume which
is passed by defects divided by the total volume, i.e.
nB nB
∼ SF |0 , (198)
s s
61
where (nB /s)|0 is the result of (197) obtained in the bubble nucleation mechanism.
A big caveat for defect-mediated electroweak baryogenesis is that the standard
electroweak theory does not admit topological defects. However, in a theory with
additional physics just above the electroweak scale it is possible to obtain defects
(see e.g. [162] for some specific models). The closer the scale η of the new physics is
to the electroweak scale ηEW , the larger the volume in defects and the more efficient
defect-mediated electroweak baryogenesis. Using the result of (195) for the separation
of defects, we obtain
ηEW 3/2
SF ∼ λ( ) . (199)
η
Obviously, the advantage of the defect-mediated baryongenesis scenario is that it
does not depend on the order and on the detailed dynamics of the electroweak phase
transition.
J. Summary
As we have seen, topological defects may play an important role in cosmology.

Defects are inevitably produced during symmetry breaking phase transitions in the
early Universe in all theories in which defects are topologically stable. Theories giving
rise to domain walls or local monopoles are ruled out by cosmological constraints.
Those producing cosmic strings, global monopoles and textures are quite attractive.
If the scale of symmetry breaking at which the defects are produced is about
16
10 GeV, then defects can act as the seeds for galaxy formation. Defect theories of
structure formation predict a roughly scale-invariant spectrum of density perturba-
tions, similar to inflation-based models. However, the phases in the density field are
distributed in a non-Gaussian manner. Thus, the predictions of defect models can
be distinguished from those of inflationary models. In addition, the predictions of
different defect models can be distinguished from eachother.
Focusing on the cosmic string theory of structure formation, we have seen that
the model gives rise to several distinctive signatures. The large-scale structure in the
Universe is predicted to be dominated by either planar or filamentary structures (de-
pending on whether there is small-scale structure on the strings), with a distinctive
scale given by the comoving Hubble radius at teq . There are many ways topolog-
ical measures with which one can quantify this prediction (see e.g. [163]). In the
CMB, the distinctive signature for strings is predicted to be line discontinuities in the
temperature maps.
Topological defects may also play a crucial role in baryogenesis. This applies both
to GUT and electroweak baryogenesis. The crucial point is that defects constitute out-
of-equilibrium configurations, and may therefore be the sites of net baryon production.
Acknowledgments
62
I wish to thank Professor Choonkyu Lee for inviting me to give these lectures
and all the organizers and participants for their wonderful hospitality in Korea. In
particular, I am grateful to Dr. Jai-chan Hwang for interesting discussions, and to
Bill Unruh for hosting me at the University of British Columbia where these lecture
notes were written up. I also would like to all of my research collaborators, on
whose work I have freely drawn. Partial financial support for the preparation of
this manuscript has been provided at Brown by the US Department of Energy under
Grant DE-FG0291ER40688,
[1] A. Guth, Phys. Rev. D23, 347 (1981).

[2] Ya.B. Zel’dovich, Mon. Not. R. astron. Soc. 192, 663 (1980);
A. Vilenkin, Phys. Rev. Lett. 46, 1169 (1981).
[3] V. de Lapparent, M. Geller and J. Huchra, Ap. J. (Lett) 302, L1 (1986).
[4] S. Landy, S. Shectman, H. Lin, R. Kirschner, A. Oemler and D. Tucker, Ap. J. (Lett.)
456, 1L (1996);
H. Lin, R. Kirshner, S. Shectman, S. Landy, A. Oemler, D. Tucker and P. Schechter,
Ap. J. (Lett.) 464, 60L (1996);
S. Shectman, S. Landy, A. Oemler, D. Tucker, H. Lin, R. Kirshner and P. Schechter,
Ap. J. (Suppl.) 470, 172S (1996).
[5] J. Ostriker and L. Cowie, Ap. J. (Lett.) 243, L127 (1981).
[6] A. Linde, ‘Particle Physics and Inflationary Cosmology’ (Harwood, Chur, 1990).
[7] S. Blau and A. Guth, ‘Inflationary Cosmology,’ in ‘300 Years of Gravitation’ ed. by
S. Hawking and W. Israel (Cambridge Univ. Press, Cambridge, 1987).
[8] K. Olive, Phys. Rep. 190, 307 (1990).
[9] R. Brandenberger, Rev. Mod. Phys. 57, 1 (1985).
[10] D. Kazanas, Ap. J. 241, L59 (1980).
[11] J. Gott III and T. Statler, Phys. Lett. 136B, 157 (1984);
M. Bucher, A. Goldhaber and N. Turok, Phys. Rev. D52, 3314 (1995);
A. Linde, Phys. Lett. B351, 99 (1995).
[12] W. Press, Phys. Scr. 21, 702 (1980).
[13] G. Chibisov and V. Mukhanov, ‘Galaxy Formation and Phonons,’ Lebedev Physical
Institute Preprint No. 162 (1980);
G. Chibisov and V. Mukhanov, Mon. Not. R. Astron. Soc. 200, 535 (1982).
[14] V. Lukash, Pis’ma Zh. Eksp. Teor. Fiz. 31, 631 (1980).
[15] K. Sato, Mon. Not. R. Astron. Soc. 195, 467 (1981).
[16] R. Brandenberger, Phys. Lett. 129B, 397 (1983).
[17] A. Guth and S.-H. Tye, Phys. Rev. Lett. 44, 631 (1980).
[18] D. Kirzhnits and A. Linde, Pis’ma Zh. Eksp. Teor. Fiz. 15, 745 (1972);
D. Kirzhnits and A. Linde, Zh. Eksp. Teor. Fiz. 67, 1263 (1974);
63
C. Bernard, Phys. Rev. D9, 3313 (1974);
L. Dolan and R. Jackiw, Phys. Rev. D9, 3320 (1974);
S. Weinberg, Phys. Rev. D9, 3357 (1974).
[19] S. Coleman, Phys. Rev. D15, 2929 (1977);
C. Callan and S. Coleman, Phys. Rev. D16, 1762 (1977);
M. Voloshin, Yu. Kobzarev and L. Okun, Sov. J. Nucl. Phys. 20, 644 (1975);
M. Stone, Phys. Rev. D14, 3568 (1976);
M. Stone, Phys. Lett. 67B, 186 (1977);
P. Frampton, Phys. Rev. Lett., 37, 1380 (1976);
S. Coleman, in ‘The Whys of Subnuclear Physics’ (Erice 1977), ed by A. Zichichi
(Plenum, New York, 1979).
[20] A. Guth and E. Weinberg, Nucl. Phys. B212, 321 (1983).
[21] A. Linde, Phys. Lett. 108B, 389 (1982).
[22] A. Albrecht and P. Steinhardt, Phys. Rev. Lett. 48, 1220 (1982).
[23] A. Linde, Phys. Lett. 129B, 177 (1983).
[24] A. Linde, D. Linde and A. Mezhlumian, Phys. Rev. D49, 1783 (1994);
A. Linde, ‘Lectures on Inflationary Cosmology’, Stanford preprint SU-ITP-94-36, hep-
th/9410082 (1994).
[25] A. Starobinsky, in ‘Current Trends in Field Theory, Quantum Gravity, and Strings’,
Lecture Notes in Physics, ed. by H. de Vega and N. Sanchez (Springer, Heidelberg,
1986).
[26] S. Weinberg, Rev. Mod. Phys. 61, 1 (1989);
S. Carroll, W. Press and E. Turner, Ann. Rev. Astron. Astrophys. 30, 499 (1992).
[27] A. Starobinsky, Phys. Lett. 91B, 99 (1980).
[28] V. Mukhanov and R. Brandenberger, Phys. Rev. Lett. 68, 1969 (1992).
[29] R. Ball and A. Matheson, Phys. Rev. D45, 2647 (1992).
[30] L. Parker and Y. Zhang, Phys. Rev. D47, 416 (1993).
[31] A. Zhitnitsky, “Effective Field Theories as Asymptotic Series: From QCD to Cosmol-
ogy”, hep-ph/9601348 (1996).
[32] R. Brandenberger and A. Zhitnitsky, “Can Asymptotic Series Resolve the Problems
of Inflation?”, Brown preprint BROWN-HET-1035, hep-ph/9604407 (1996).
[33] N. C. Tsamis and R. P. Woodard, Phys. Lett. B301, 351 (1993); Ann. Phys. 238, 1
(1995); “Quantum Gravity Slows Inflation”, hep-ph/9602315 (1996).
[34] V. Mukhanov, H. Feldman and R. Brandenberger, Phys. Rep. 215, 203 (1992).
[35] V. Mukhanov and G. Chibisov, JETP Lett. 33, 532 (1981).
[36] A. Guth and S.-Y. Pi, Phys. Rev. Lett. 49, 110 (1982);
S. Hawking, Phys. Lett. 115B, 295 (1982);
A. Starobinskii, Phys. Lett. 117B, 175 (1982);
V. Mukhanov, JETP Lett. 41, 493 (1985).
[37] J. Bardeen, P. Steinhardt and M. Turner, Phys. Rev. D28, 1809 (1983).
[38] F. Adams, K. Freese and A. Guth, Phys. Rev. D43, 965 (1991).
[39] B. Altshuler, Class. Quant. Grav. 7, 189 (1990).
64
[40] M. Markov, Pis’ma Zh. Eksp. Theor. Fiz. 36, 214 (1982);
M. Markov, Pis’ma Zh. Eksp. Theor. Fiz. 46, 342 (1987);
V. Ginsburg, V. Mukhanov and V. Frolov, Pis’ma Zh. Eksp. Theor. Fiz. 94, 3 (1988);
V. Frolov, M. Markov and V. Mukhanov, Phys. Rev. D41, 383 (1990).
[41] R. Brandenberger, V. Mukhanov and A. Sornborger, Phys. Rev. D48, 1629 (1993).
[42] M. Trodden, V. Mukhanov and R. Brandenberger, Phys. Lett. B316, 483 (1993).
[43] R. Moessner and M. Trodden, Phys. Rev. D51, 2801 (1995).
[44] L. Kofman, A. Linde and A. Starobinski, Phys. Rev. Lett. 73, 3195 (1994).
[45] Y. Shtanov, J. Traschen and R. Brandenberger, Phys. Rev. D51, 5438 (1995).
[46] A. Dolgov and A. Linde, Phys. Lett. 116B, 329 (1982).
[47] L. Abbott, E. Farhi and M. Wise, Phys. Lett. 117B, 29 (1982).
[48] J. Traschen and R. Brandenberger, Phys. Rev. D42, 2491 (1990).
[49] L. Landau and E. Lifshitz, ‘Mechanics’ (Pergamon, Oxford, 1960);
V. Arnold, ‘Mathematical Methods of Classical Mechanics’ (Springer, New York,
1978).
[50] M. Hotta, I. Joichi, S. Matsumoto and M. Yoshimura, “Quantum System under Peri-
odic Perturbation: Effect of Environment”, hep-ph/9608374 (1996).
[51] N. Birrell and P. Davies, ‘Quantum Fields in Curved Space’ (Cambridge Univ. Press,
Cambridge, 1982).
[52] L. Kofman, A. Linde and A. Starobinski, Phys. Rev. Lett. 76, 1011 (1996).
[53] I. Tkachev, Phys. Lett. B376, 35 (1996).
[54] E. Kolb, A. Linde and A. Riotto, Phys. Rev. Lett. 77, 4290 (1996).
[55] S. Khlebnikov and I. Tkkachev, Phys. Rev. Lett. 77, 219 (1996);
S. Khlebnikov and I. Tkachev, “The Universe after Inflation: the Wide Resonance
Case”, hep-ph/9608458 (1996).
[56] R. Allahverdi and B. Campbell, “Cosmological Reheating and Selfinteracting Final
State Bosons”, hep-ph/9606463 (1996).
[57] D. Boyanovsky, H. de Vega, R. Holman and J. Salgado, Phys. Rev. D54, 7570 (1996);
D. Boyanovsky, D. Cormier, H. de Vega, R. Holman, A. Singh and M. Srednicki,
“Preheating in FRW Universes”, hep-ph/9609527 (1996).
[58] M. Yoshimura, Prog. Theor. Phys. 94, 873 (1995);
H. Fujisaki, K. Kumekawa, M. Yamaguchi and M. Yoshimura, Phys. Rev. D53, 6805
(1996).
[59] D. Boyanovsky, H. de Vega, R. Holman, D.-S. Lee and A. Singh, Phys. Rev. D51,
4419 (1995);
D. Boyanovsky, M. D’Attanasio, H. de Vega, R. Holman and D.-S. Lee, Phys. Rev.
D52, 6805 (1995).
[60] D. Kaiser, Phys. Rev. D53, 1776 (1996).
[61] G. Anderson, A. Linde and A. Riotto, Phys. Rev. Lett. 77, 3716 (1996).
[62] A. Liddle and D. Lyth, Phys. Rep. 231, 1 (1993).
[63] M. Gasperini, “Status of String Cosmology: Phenomenological Aspects”, hep-
th/9509127 (1995);
65
G. Veneziano, “String Cosmology: Concepts and Consequences”, hep-th/9512091
(1995).
[64] G. Efstathiou, in ‘Physics of the Early Universe,’ proc. of the 1989 Scottish Univ.
Summer School in Physics, ed. by J. Peacock, A. Heavens and A. Davies (SUSSP
Publ., Edinburgh, 1990).
[65] T. Padmanabhan, ‘Structure Formation in the Universe’ (Cambridge Univ. Press,
Cambridge, 1993).
[66] E. Harrison, Phys. Rev. D1, 2726 (1970);
Ya.B. Zel’dovich, Mon. Not. R. astron. Soc. 160, 1p (1972).
[67] J. Bond, G. Efstathiou and J. Silk, Phys. Rev. Lett. 45, 1980 (1980);
J. Bond and A. Szalay, Ap. J. 274, 443 (1983).
[68] E. Lifshitz, Zh. Eksp. Teor. Fiz. 16, 587 (1946);
E. Lifshitz and I. Khalatnikov, Adv. Phys. 12, 185 (1963).
[69] W. Press and E. Vishniac, Ap. J. 239, 1 (1980).
[70] R. Brandenberger, H. Feldman, V. Mukhanov and T. Prokopec, ‘Gauge Invariant Cos-
mological Perturbations: Theory and Applications,’ publ. in “The Origin of Structure
in the Universe,” eds. E. Gunzig and P. Nardone (Kluwer, Dordrecht, 1993).
[71] J. Bardeen, Phys. Rev. D22, 1882 (1980).
[72] R. Brandenberger, R. Kahn and W. Press, Phys. Rev. D28, 1809 (1983).
[73] H. Kodama and M. Sasaki, Prog. Theor. Phys. Suppl. No. 78, 1 (1984).
[74] R. Durrer and N. Straumann, Helvet. Phys. Acta 61, 1027 (1988);
R. Durrer, Fund. Cosmic Phys. 15, 209 (1994), astro-ph/9311040.
[75] D. Lyth, Phys. Rev. D31, 1792 (1985);
D. Lyth and M. Mukherjee, Phys. Rev. D38, 485 (1988).
[76] J. Hwang and E. Vishniac, Ap. J. 353, 1 (1990).
[77] G.F.R. Ellis and M. Bruni, Phys. Rev. D40, 1804 (1989);
G.F.R. Ellis, J. Hwang and M. Bruni, Phys. Rev. D40, 1819 (1989).
[78] D. Salopek and J. Stewart, Phys. Rev. D51, 517 (1995).
[79] J. Stewart, Class. Quantum Grav. 7, 1169 (1990).
[80] J. Stewart and M. Walker, Proc. R. Soc. A341, 49 (1974).
[81] R. Brandenberger and R. Kahn, Phys. Rev. D28, 2172 (1984).
[82] R. Brandenberger, “Modern Cosmology and Structure Formation”, in ‘CP Violation
and the Limits of the Standard Model (TASI 94)’, ed. J. Donoghue (World Scientific,
Singapore, 1995), astro-ph/94 ;
R. Brandenberger, in ‘Physics of the Early Universe,’ proc. of the 1989 Scottish Univ.
Summer School in Physics, ed. by J. Peacock, A. Heavens and A. Davies (SUSSP
Publ., Edinburgh, 1990);
R. Brandenberger, ‘Lectures on Modern Cosmology and Structure Formation’, in ‘Par-
ticles and Fields’, ed. by O. Eboli and V. Ribelles (World Scientific, Singapore 1994).
[83] W. Zurek, Phys. Rev. D24, 1516 (1982);
W. Zurek, Phys. Rev. D26 1862 (1982).
[84] E. Joos and H. Zeh, Z. Phys. B59, 223 (1985);
66
H. Zeh, Phys. Lett. 116A, 9 (1986);
C. Kiefer, Class. Quantum Grav. 4, 1369 (1987);
T. Fukuyama and M. Morikawa, Phys. Rev. D39, 462 (1989);
J. Halliwell, Phys. Rev. D39, 2912 (1989);
T. Padmanabhan, Phys. Rev. D39, 2924 (1980);
W. Unruh and W. Zurek, Phys. Rev. D40, 1071 (1989);
E. Calzetta and F. Mazzitelli, Phys. Rev. D42, 4066 (1990);
S. Habib and R. Laflamme, Phys. Rev. D42, 4056 (1990);
H. Feldman and A. Kamenshchik, Class. Quantum Grav. 8, L65 (1991).
[85] M. Sakagami, Prog. Theor. Phys. 79, 443 (1988);
R. Brandenberger, R. Laflamme and M. Mijic, Mod. Phys. Lett. A5, 2311 (1990).
[86] J. Bardeen, unpublished (1984).
[87] R. Brandenberger, Nucl. Phys. B245, 328 (1984).
[88] R. Brandenberger and C. Hill, Phys. Lett. 179B, 30 (1986).
[89] V. Mukhanov, Zh. Eksp. Teor. Fiz. 94, 1 (1988).
[90] M. Sasaki, Prog. Theor. Phys. 76, 1036 (1986).
[91] M. Davis and J. Huchra, Ap. J. 254, 437 (1982).
[92] N. Bahcall and R. Soneira, Ap. J. 270, 20 (1983);
A. Klypin and A. Kopylov, Sov. Astr. Lett. 9, 41 (1983).
[93] M. Strauss et al., Ap. J. 385, 421 (1992).
[94] D. Mermin, Rev. Mod. Phys. 51, 591 (1979).
[95] P. de Gennes, ‘The Physics of Liquid Crystals’ (Clarendon Press, Oxford, 1974);
I. Chuang, R. Durrer, N. Turok and B. Yurke, Science 251, 1336 (1991);
M. Bowick, L. Chandar, E. Schiff and A. Srivastava, Science 263, 943 (1994).
[96] M. Salomaa and G. Volovik, Rev. Mod. Phys. 59, 533 (1987).
[97] A. Abrikosov, JETP 5, 1174 (1957).
[98] T.W.B. Kibble, J. Phys. A9, 1387 (1976).
[99] A. Vilenkin and E.P.S. Shellard, ‘Strings and Other Topological Defects’ (Cambridge
Univ. Press, Cambridge, 1994).
[100] Ya.B. Zel’dovich, I. Kobzarev and L. Okun, Zh. Eksp. Teor. Fiz. 67, 3 (1974).
[101] Ya.B. Zel’dovich and M. Khlopov, Phys. Lett. 79B, 239 (1978);
J. Preskill, Phys. Rev. Lett. 43, 1365 (1979).
[102] M. Hindmarsh and T.W.B. Kibble, Rept. Prog. Phys. 58, 477 (1995).
[103] R. Brandenberger, Int. J. Mod. Phys. A9, 2117 (1994).
[104] A. Vilenkin, Phys. Rep. 121, 263 (1985).
[105] H. Nielsen and P. Olesen, Nucl. Phys. B61, 45 (1973).
[106] G. Mazenko, W. Unruh and R. Wald, Phys. Rev. D31, 273 (1985).
[107] T.W.B. Kibble, Acta Physica Polonica B13, 723 (1982).
[108] W. Zurek, Acta Phys. Pol. B24, 1301 (1993);
W. Zurek, “Cosmological Experiments in Condensed Matter Systems”, cond-
mat/9607135 (1996), Phys. Rep. (in press).
[109] M. Hindmarsh, A.-C. Davis and R. Brandenberger, Phys. Rev. D49, 1944 (1994);
67
R. Brandenberger and A.-C. Davis, Phys. Lett. B332, 305 (1994).
[110] T.W.B. Kibble and E. Weinberg, Phys. Rev. D43, 3188 (1991).
[111] T. Vachaspati and A. Vilenkin, Phys. Rev. D30, 2036 (1984).
[112] N. Turok, Nucl. Phys. B242, 520 (1984).
[113] N. Turok and R. Brandenberger, Phys. Rev. D33, 2175 (1986);
A. Stebbins, Ap. J. (Lett.) 303, L21 (1986);
H. Sato, Prog. Theor. Phys. 75, 1342 (1986).
[114] D. Bennett and F. Bouchet, Phys. Rev. Lett. 60, 257 (1988);
B. Allen and E.P.S. Shellard, Phys. Rev. Lett. 64, 119 (1990);
A. Albrecht and N. Turok, Phys. Rev. D40, 973 (1989).
[115] J. Silk and A. Vilenkin, Phys. Rev. Lett. 53, 1700 (1984).
[116] A. Vilenkin, Phys. Rev. D23, 852 (1981);
J. Gott, Ap. J. 288, 422 (1985);
W. Hiscock, Phys. Rev. D31, 3288 (1985);
B. Linet, Gen. Rel. Grav. 17, 1109 (1985);
D. Garfinkle, Phys. Rev. D32, 1323 (1985);
R. Gregory, Phys. Rev. Lett. 59, 740 (1987).
[117] T. Vachaspati, Phys. Rev. Lett. 57, 1655 (1986).
[118] A. Stebbins, S. Veeraraghavan, R. Brandenberger, J. Silk and N. Turok, Ap. J. 322,
1 (1987).
[119] R. Brandenberger, L. Perivolaropoulos and A. Stebbins, Int. J. of Mod. Phys. A5,
1633 (1990);
L. Perivolarapoulos, R. Brandenberger and A. Stebbins, Phys. Rev. D41, 1764 (1990);
R. Brandenberger, Phys. Scripta T36, 114 (1991).
[120] Ya. B. Zeldovich, Astr. Astrophys. 5, 84 (1970).
[121] B. Carter, Phys. Rev. D41, 3869 (1990).
[122] T. Vachaspati and A. Vilenkin, Phys. Rev. Lett. 67, 1057 (1991);
T. Vachaspati, Phys. Rev. D45, 3487 (1992);
D. Vollick, Phys. Rev. D45, 1884 (1992).
[123] D. Bennett, A. Stebbins and F. Bouchet, Ap. J. (Lett.) 399, L5 (1992).
[124] L. Perivolaropoulos, Phys. Lett. 298B, 305 (1993).
[125] R. Brandenberger, N. Kaiser, D. Schramm and N. Turok, Phys. Rev. Lett. 59, 2371
(1987);
R. Brandenberger, N. Kaiser and N. Turok, Phys. Rev. D36, 2242 (1987).
[126] R. Scherrer, A. Melott and E. Bertschinger, Phys. Rev. Lett. 62, 379 (1989).
[127] R. Moessner and R. Brandenberger, Mon. Not. R. astr. Soc. 280, 797 (1996), astro-
ph/9510141.
[128] A. Aguirre and R. Brandenberger, Int. J. Mod. Phys. D4, 711 (1995), astro-
ph/9505031.
[129] V. Zanchin, J.A.S. Lima and R. Brandenberger, Phys. Rev. D54, 7219 (1996), astro-
ph/9607062.
[130] P. Mähönen, Ap. J. (Lett.) 459, L45 (1996).
68
[131] T. Hara and S. Miyoshi, Prog. Theor. Phys. 81, 1187 (1989);
T. Hara, S. Morioka and S. Miyoshi, Prog. Theor. Phys. 84, 867 (1990);
T. Hara et al., Ap. J. 428, 51 (1994).
[132] A. Albrecht and A. Stebbins, Phys. Rev. Lett. 69, 2615 (1992).
[133] M. Barriola and A. Vilenkin, Phys. Rev. Lett. 63, 341 (1989);
S. Rhie and D. Bennett, Phys. Rev. Lett. 65, 1709 (1990).
[134] N. Turok, Phys. Rev. Lett. 63, 2625 (1989).
[135] R. Durrer and Z.H. Zhou, Phys. Rev. D53, 5394 (1996).
[136] N. Turok, Phys. Scripta T36, 135 (1991).
[137] A. Sornborger, R. Brandenberger, B. Fryxell and K. Olson, “The Structure of Cosmic
String Wakes”, astro-ph/9608020 (1996), Ap. J. (in press).
[138] S. White, J. Navarro, A. Evrard and C. Frenk, Nature 366, 429 (1993).
[139] N. Kaiser and A. Stebbins, Nature 310, 391 (1984).
[140] R. Moessner, L. Perivolaropoulos and R. Brandenberger, Ap. J. 425, 365 (1994),
astro-ph/9310001.
[141] J. Magueijo, A. Albrecht, P. Ferreira and D. Coulson, Phys. Rev. D54, 3727 (1996),
astro-ph/9605047.
[142] W. Hu and M. White, Phys. Rev. Lett. 77, 1687 (1996), astro-ph/9602020.
[143] N. Turok, Phys. Rev. Lett. 77, 4138 (1996), astro-ph/9607109.
[144] N. Turok and D. Spergel, Phys. Rev. Lett. 64, 2736 (1990).
[145] A. Sakharov, Pisma Zh. Eksp. Teor. Fiz. 5, 32 (1967).
[146] M. Yoshimura, Phys. Rev. Lett. 41, 281 (1978);
A. Ignatiev, N. Krasnikov, V. Kuzmin and A. Tavkhelidze, Phys. Lett. 76B, 436
(1978);
S. Dimopoulos and L. Susskind, Phys. Rev. D18, 4500 (1978);
S. Weinberg, Phys. Rev. Lett. 42, 850 (1979);
D. Toussaint, S. Trieman, F. Wilczek and A. Zee, Phys. Rev. D19, 1036 (1979).
[147] A. Dolgov, Phys. Rep. 222, 309 (1992).
[148] V. Rubakov and M. Shaposhnikov, “Electroweak Baryon Number Nonconservation in
the Early Universe and in High-Energy Collisions”, hep-ph/9603208 (1996).
[149] R. Brandenberger, A.-C. Davis and M. Hindmarsh, Phys. Lett. B263, 239 (1991).
[150] G. t’Hooft, Phys. Rev. Lett. 37, 8 (1976).
[151] N. Manton, Phys. Rev. D28, 2019 (1983);
F. Klinkhamer and N. Manton, Phys. Rev. D30, 2212 (1984).
[152] V. Kuzmin, V. Rubakov and M. Shaposhnikov, Phys. Lett. B155, 36 (1985);
P. Arnold and L. McLerran, Phys. Rev. D36, 581 (1987).
[153] M. Shaposhnikov, JETP Lett. 44, 465 (1986);
M. Shaposhnikov, Nucl. Phys. B287, 757 (1987);
L. McLerran, Phys. Rev. Lett. 62, 1075 (1989).
[154] N. Turok, in ‘Perspectives on Higgs Physics’, ed. G. Kane (World Scientific, Singapore,
1992);
A. Cohen, D. Kaplan and A. Nelson, Ann. Rev. Nucl. Part. Sci. 43, 27 (1993).
69
[155] A. Cohen, D. Kaplan and A. Nelson, Phys. Lett. B245, 561 (1990);
A. Cohen, D. Kaplan and A. Nelson, Nucl. Phys. B349, 727 (1991);
A. Nelson, D. Kaplan and A. Cohen, Nucl. Phys. B373, 453 (1992);
M. Joyce, T. Prokopec and N. Turok, Phys. Lett. B338, 269 (1994);
M. Joyce, T. Prokopec and N. Turok, Phys. Rev. Lett. 75, 1695 (1995).
[156] N. Turok and T. Zadrozny, Phys. Rev. Lett. 65, 2331 (1990);
N. Turok and J. Zadrozny, Nucl. Phys. B358, 471 (1991);
L. McLerran, M. Shaposhnikov, N. Turok and M. Voloshin, Phys. Lett. B256, 451
(1991);
M. Dine, P. Huet, R. Singleton and L. Susskind, Phys. Lett. B257, 351 (1991).
[157] A. Cohen, D. Kaplan and A. Nelson, Phys. Lett. B263, 86 (1991);
M. Joyce, T. Prokopec and N. Turok, Phys. Rev. D53, 2958 (1996).
[158] M. Gleiser, Phys. Rev. Lett. 73, 3495 (1994);
J. Borrill and M. Gleiser, Phys. Rev. D51, 4121 (1995).
[159] W. Perkins, Nucl. Phys. B449, 265 (1995).
[160] R. Brandenberger, A.-C. Davis, T. Prokopec and M. Trodden, Phys. Rev. D53, 4257
(1996).
[161] R. Brandenberger, A.-C. Davis and M. Trodden, Phys. Lett. B332, 305 (1994);
R. Brandenberger and A.-C. Davis, Phys. Lett. B308, 79 (1993).
[162] M. Trodden, A.-C. Davis and R. Brandenberger, Phys. Lett. B349, 131 (1995).
[163] R. Brandenberger, D. Kaplan and S. Ramsey, “Some Statistics for Measuring Large
Scale Structure”, astro-ph/9310004 (1993);
A. Albrecht and J. Robinson, “A Statistic for Indentifying Cosmic String Wakes and
Other Sheet-Like Structures”, astro-ph/9505123.
70

9701276v1 PDF

Uploaded by

Copyright:

Available Formats

9701276v1 PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

9701276v1 PDF

Uploaded by

Copyright:

Available Formats

∗

Particle Physics Aspects of Modern Cosmology

Modern cosmology has created a tight link between particle physics

1. Introduction and Outline

2. Lecture 1: Inflationary Universe: Progress and Problems

3. Lecture 2: Classical and Quantum Theory of Cosmological Perturba-

II. INFLATIONARY UNIVERSE: PROGRESS AND PROBLEMS

The hypothesis that the Universe underwent a period of exponential expansion at

A. Problems of Standard Cosmology

∆xp = a(t)∆xc . (2)

The dynamics of an expanding Universe is determined by the Einstein equations,

The third key assumption of standard cosmology is that matter is described by

ρm (t) ∼ a−3 (t) , (7)

ρr (t) ∼ a−4 (t) , (8)

ρr (t) being the energy density in radiation.

In standard cosmology and in an expanding Universe, Ω = 1 is an unstable fixed

The quantity ε is proportional to s−2/3 , where s is the entropy density. Hence, in

B. Inflationary Universe Scenario

ℓH (t) = H −1 (t) . (18)

As will be discussed in Chapter 4, the density perturbations produced during

ϕ̇(x, ti ) = ∇ϕ(xi ti ) = 0 (23)

V (ϕ(xi , ti )) > 0 , (24)

the equation of state of matter will be (20).

ϕ̇2 ≪ V (ϕ) , (∇ϕ)2 ≪ V (ϕ) (25)

Tµν ∼ V (ϕ)gµν (26)

ϕ̈ + 3H ϕ̇ − a−2 ▽2 ϕ = −V ′ (ϕ) . (28)

Within a fluctuation region, ϕ(x) will be homogeneous. In such a region, we can

ϕ̇2 ≪ V (ϕ) , (29)

FIG. 8. Evolution of ϕ(t) and T (t) in the new inflationary Universe.

C. Problems of Inflationary Cosmology

(−1)n n!an hϕin .

We summarize our ignorance of the nonperturbative physics in the assumption that

V ′′ m2pl << 24πV , (35)

ε̇2 m2pl + ϕ̇2 << 2V . (36)

V (ϕ) = λϕ4 (41)

D. Inflation and Nonsingular Cosmology

The question we wish to address in this subsection is whether it is possible to

V (ϕ) ∼ ϕ2 as |ϕ| → 0 , (47)

Choosing the simple interpolating potential

ds2 = dt2 − a(t)2 (dx2 + dy 2 + dz 2 ) , (60)

The resulting phase diagram of vacuum solutions is sketched in Fig. 10 (for

E. Reheating in Inflationary Cosmology

Reheating is an important stage in inflationary cosmology. It determines the

TR ∼ (ΓB mpl )1/2 . (73)

If we now use the “naturalness” constraint1

TR < 1010 GeV . (75)

This would imply no GUT baryogenesis, no GUT-scale defect production, and no

χ̈k + 3H χ̇k + (kp2 + m2χ + 2g 2σϕ)χk = 0, (77)

where kp is the time-dependent physical wavenumber.

µ∆t >> 1 (82)

The inflationary Universe is an attractive scenario for early Universe cosmology.

III. CLASSICAL AND QUANTUM THEORY OF COSMOLOGICAL

In inflationary Universe and topological defect models of structure formation,

In this article we only discuss theories in which structures grow by gravitational

P (k) =< |δ(k)|2 > , (87)

with a window function Wk (λ) with the following properties

k −1 a(tH (k)) = tH (k) (92)

which leads to tH (k) ∼ k −3 . According to the linear theory of cosmological perturba-

The Newtonian theory of cosmological perturbations is an approximate analysis