• Nenhum resultado encontrado

Braz. J. Phys. vol.31 número2

N/A
N/A
Protected

Academic year: 2018

Share "Braz. J. Phys. vol.31 número2"

Copied!
16
0
0

Texto

(1)

Frontiers of Inationary Cosmology

Robert H. Brandenberger

PhysisDepartment, BrownUniversity,

Providene,R.I.,USA

Reeivedon2February,2001

Afterabrief reviewthetheoryof osmologial perturbations,Ihighlight somereent progress in

theareaofreheatingininationaryosmology,fousinginpartiularonparametriampliationof

super-Hubbleosmologialutuations,andontheroleofnoiseintheresonanedynamis(yielding

a newproof of Andersonloalization). I thendisuss several important oneptual problems for

theurrentrealizationsofinationbasedonfundamentalsalarmatterelds,andreviewsomenew

approahesatsolvingtheseproblems.

I Introdution

Inationary osmology [1℄ has beome oneof the

or-nerstonesof modern osmology. Ination wastherst

theorywithinwhihitwaspossibletomakepreditions

aboutthestrutureoftheUniverseonlargesalesbased

onausalphysis. Thedevelopmentoftheinationary

Universesenariohasopenedupanewand extremely

promising avenue for onneting fundamental physis

withexperiment.

Afterabriefintrodutiontoinationaryosmology

(Setion II) and an overview of the theory of

osmo-logialperturbationsapplied to ination(Setion III),

I fous on reent improvements in our understanding

ofthetheoryofinationaryreheating(SetionIV),

fo-usinginpartiularonreentstudiesoftheparametri

ampliationofgravitationalutuationsandonthe

ef-fetsofnoiseontheresonanedynamis (whih yields

anewproofofAndersonloalization inonespatial

di-mension).

However, in spite of the remarkable suess of the

inationary Universe paradigm, there are several

se-riousoneptualproblemsforurrentmodelsin whih

inationisgeneratedbythepotentialenergyofasalar

matter eld. These problems are disussed in Setion

V.

SetionVIisasummaryofsomenewapproahesto

solvingtheabove-mentionedproblems. An attemptto

obtain ination from ondensatesis disussed, a

non-singular Universe onstrution making use of higher

derivativetermsinthegravitationalationisexplained,

and a framework for alulating the bak-reation of

osmologialperturbationsissummarized.

This short review fouses ona seleted numberof

topis at the forefront of inationary osmology. For

omprehensive reviews of ination, the reader is

re-tionarymodelbuildingintheontextof

supersymmet-rimodelsanbefoundin [6℄. Anextended andmore

pedagogialversionofthesenotesis[7℄. Notethat

Se-tions II, V and VI of this artile are idential to the

orrespondingsetionsin [8℄.

II Basis of Inationary

Cos-mology

MosturrentmodelsofinationarebasedonEinstein's

theoryofGeneralRelativitywithamattersouregiven

byasalareld'. Basedontheosmologialpriniple,

themetriofspae-timeonlargedistanesalesanbe

writteninFriedmann-Robertson-Walker(FRW)form:

ds 2

=dt 2

a(t) 2

dr 2

1 kr 2

+r 2

(d# 2

+sin 2

#d' 2

)

;

(1)

where the onstant k determines the topology of the

spatial setions. In the following, we shall set k =0,

i.e. onsideraspatially atUniverse. Inthisase, we

anwithoutlossofgeneralitytakethesalefatora(t)

tobeequalto 1atthe presenttime t

0

, i.e. a(t

0 )=1.

Theoordinates r;#and' areomovingspherial

o-ordinates.

ForahomogeneousandisotropiUniverseand

set-ting the osmologial onstant to zero, the Einstein

equationsredue totheFRWequations

_ a

a

2

= 8G

3

(2)

 a

a =

4G

3

(+3p); (3)

(2)

den-toyieldtheontinuityequation(withHubbleonstant

H =a=a)_

_

= 3H(+p): (4)

Theequationofstateofmatterisdesribedbya

num-berwdened by

p=w: (5)

Theideaof ination[1℄ isverysimple. Weassume

there is a time interval beginning at t

i

and ending at

t

R

(the\reheatingtime")duringwhihtheUniverseis

exponentiallyexpanding,i.e.,

a(t)e Ht

; t[t

i ;t

R

℄ (6)

withonstantHubbleexpansionparameterH. Suh a

period is alled\deSitter" or \inationary."The

su-essofBigBangnuleosynthesissetsanupperlimitto

thetimet

R

ofreheating,t

R t

NS ,t

NS

beingthetime

ofnuleosynthesis.

Duringtheinationary phase, thenumber density

of any partiles initially present at t = t

i

deays

ex-ponentially. At t = t

R

, allof the energy whih is

re-sponsibleforinationisreleased(seelater)asthermal

energy. This is a non-adiabati proess during whih

theentropyinreasesbyalargefator.

A period of ination an solve the homogeneity

problem of standardosmology,the reasonbeingthat

during ination the physial size of the forward light

oneexponentiallyexpandsandthusaneasilybeome

larger than thephysial size of thepast light one at

t

re

,thetimeoflastsattering,thusexplainingthenear

isotropyof theosmi mirowavebakground(CMB).

Inationalsosolvestheatness problem[9,1℄.

Most importantly, ination provides a mehanism

whihinaausalwaygeneratestheprimordial

pertur-bationsrequired for galaxies, lusters and even larger

objets. In inationary Universe models, the Hubble

radius(\apparent"horizon)andthe(\atual")horizon

(the forwardlight one)do notoinideat late times.

Provided that the duration of ination is suÆiently

long,then all saleswithin ourpresentapparent

hori-zonwereinsidethehorizonsinet

i

. Thus,itisin

prin-iple possible to have a ausal generation mehanism

forperturbations[10,11,12,13℄.

Aswill bedisussedinSetion III,thedensity

per-turbationsprodued during ination aredue to

quan-tum utuationsin thematterand gravitationalelds

[11,12℄. Theamplitudeoftheseinhomogeneities

orre-spondstoatemperatureT

H

H,theHawking

temper-ature ofthe deSitterphase. This leadsoneto expet

that atalltimes tduring ination,perturbationswith

a xed physial wavelength H 1

will be produed.

Subsequently,thelengthofthewavesisstrethedwith

theexpansionofspae, andsoonbeomesmuh larger

than the Hubble radius `

H

(t) = H 1

(t). The phases

of the inhomogeneities are random. Thus, the

ina-tionaryUniversesenariopreditsperturbationsonall

beginningofinationtotheorrespondingquantityat

thetime of reheating. Inpartiular, provided that

in-ationlastssuÆientlylong,perturbationsonsalesof

galaxiesandbeyondwill begenerated. Note, however,

thatitisverydangeroustointerpretdeSitterHawking

radiationasthermalradiation. Infat,theequationof

stateofthis\radiation"isnotthermal[14℄.

In most urrent models of ination, the

exponen-tialexpansionisdrivenbythepotentialenergydensity

V(')ofafundamental salarmattereld 'with

stan-dardation

S

m =

Z

d 4

x p

gL

m

(7)

L

m

(') = 1

2 D

'D

' V('); (8)

whereD

denotestheovariantderivative,andgisthe

determinantofthemetritensor. Theresulting

energy-momentum tensor yields the followingexpressions for

theenergydensityandthepressurep:

(') = 1

2 _ ' 2

+ 1

2 a

2

(r') 2

+V(') (9)

p(') = 1

2 _ ' 2

1

6 a

2

(r') 2

V('): (10)

It thus follows that if the salar eld is homogeneous

and stati,but the potentialenergy positive,then the

equation ofstatep= neessaryforexponential

in-ationresults(see(4)).

Mostoftheurrentrealizationsof potential-driven

inationarebasedonsatisfyingtheonditions

_ ' 2

;a 2

(r') 2

V('); (11)

viatheideaofslowrolling[15,16℄. Considerthe

equa-tionofmotionofthesalareld':



'+3H'_ a 2

5 2

'= V 0

('): (12)

Ifthesalareldstartsoutalmosthomogeneousandat

rest, iftheHubble dampingterm(the seondterm on

the l.h.s. of (12) islarge, and ifthe potential isquite

at (so that the term on the r.h.s. of (12) is small),

then'_ 2

mayremainsmallomparedtoV('), inwhih

ase exponential inationwill result. Note that if the

spatial gradienttermsareinitially negligible,theywill

remainnegligiblesinetheyredshift.

Chaotiination [17℄ is aprototypialinationary

senario. Considerasalareld'whihisveryweakly

oupled toitself and other elds. Inthis ase,'need

notbeinthermalequilibrium atthePlanktime,and

mostof thephasespaefor 'will orrespondto large

values of j'j (typially j'j m

pl

). Consider now a

region in spae where at the initial time '(x) is very

large, and approximately homogeneousand stati. In

(3)

imme-andindue anequationofstatep' whihleadsto

ination. DuetothelargeHubbledampingterminthe

salareldequationofmotion,'(x)will onlyrollvery

slowlytowards'=0 (we aremaking the assumption

that V(') hasaglobalminimumat anite valueof'

whih an then be hosen to be ' = 0). The kineti

energy ontribution to and pwill remain small, the

spatialgradientontributionwillbeexponentially

sup-pressedduetotheexpansionoftheUniverse,andthus

ination persists. Note that the preise form of V(')

isirrelevanttothemehanism.

It is diÆultto realizehaoti ination in

onven-tional supergravity models sine gravitational

orre-tions to the potential of salar elds typially render

thepotentialsteepforvaluesofj'joftheorderofm

pl

and larger. This prevents the slow rolling ondition

(11) from beingrealizable. Even if this onditionan

besatised,thereareonstraintsfromtheamplitudeof

produeddensityutuationswhiharemuhharderto

satisfy(seeSetionV).

Hybrid ination [18℄ is a solution to the

above-mentioned problem of haoti ination. Hybrid

ina-tion requires at least two salar elds to play an

im-portant role in the dynamis of the Universe. As a

toymodel,onsider thepotentialofatheorywithtwo

salarelds'and :

V('; ) = 1

4 (M

2 2

) 2

+ 1

2 m

2

' 2

+ 1

2

0

2

' 2

: (13)

For valuesofj'jlargerthan'

'

=

0

M 2

1=2

(14)

theminimumof is =0,whereasforsmallervalues

of 'thesymmetry ! is brokenand theground

state valueof j j tends to M. Theidea of hybrid

in-ation isthat ' isslowlyrollinglikethe inatoneld

in haotiination,but thattheenergy density ofthe

Universeisdominatedby . Inationterminates one

j'j dropsbelowtheritialvalue'

, atwhihpoint

startstomove.

Notethatinhybridination'

anbemuhsmaller

than m

pl

and hene ination without super-Plank

sale values of the elds is possible. It is possible to

implementhybridinationintheontextof

supergrav-ity(seee.g. [19℄).

Atthe present time thereare many realizationsof

potential-drivenination,butthereisnoanonial

the-ory. Alotofattentionisbeingdevotedtoimplementing

ination in the ontext of unied theories, the prime

andidatebeingsuperstringtheoryorM-theory. String

theory orM-theorylivein 10or11spae-time

dimen-sions,respetively. Whenompatiedto4spae-time

dimensions,thereexistmanymodulields,salarelds

whih desribe at diretions in the ompliated

va-devoted to attempts at implementing ination using

modulields(seee.g. [20℄andreferenestherein).

Reently,ithasbeensuggestedthatourspae-time

isabranein ahigher-dimensional spae-time(see[21℄

forthebasionstrution). Waysofobtainingination

onthebranearealsounderativeinvestigation(seee.g.

[22℄).

It should also not be forgotten that ination an

arisefromthepurelygravitationalsetorofthetheory,

as in the original model of Starobinsky [23℄ (see also

SetionVI), orthatitmayarisefrom kinetitermsin

aneetiveationasinpre-big-bangosmology[24℄or

ink-ination[25℄.

Theorieswith(almost)exponentialination

gener-ially predit an (almost) sale-invariant spetrum of

densityutuations,aswasrstrealizedin [10,11,12,

13℄andthenstudiedmorequantitativelyin[26,27,28℄.

ViatheSahs-Wolfeeet[29℄,thesedensity

perturba-tionsindue CMBanisotropieswithaspetrumwhih

isalsosale-invariantonlargeangularsales.

Theheuristipitureisasfollows. Iftheinationary

periodwhihlasts fromt

i tot

R

isalmostexponential,

thenthe physialeets whih areindependentof the

smalldeviationsfromexponentialexpansion(an

exam-pleofsomethingwhihdoesdependonthesedeviations

iseetsonnetedwiththeremnantradiationdensity

during ination) are time-translation-invariant. This

implies,for example, that quantum utuations at all

times have the same strength when measured on the

samephysiallengthsale.

Iftheinhomogeneitiesaresmall,theyandesribed

by lineartheory, whih implies that allFouriermodes

kevolveindependently. Theexponentialexpansion

in-ates the wavelength of any perturbation. Thus, the

wavelengthof perturbationsgenerated earlyin the

in-ationaryphaseonlengthsalessmallerthanthe

Hub-bleradiussoonbeomesequalto the(\exits")Hubble

radius (this happens at the time t

i

(k)) and ontinues

to inrease exponentially. After ination, the Hubble

radius inreasesas t while the physial wavelength of

autuation inreases only asa(t). Thus, eventually

the wavelength will ross the Hubble radius again (it

will \enter" the Hubble radius) at time t

f

(k). Thus,

it is possible for ination to generate utuations on

osmologialsalesbyausalphysis.

Anyphysial proesswhihobeysthesymmetryof

the inationary phase and whih generates

perturba-tionswill generateutuations ofequalstrength when

measuredwhentheyrosstheHubbleradius(see,

how-ever,Setion V.2):

ÆM

M (k;t

i

(k)) = onst (15)

(independent of k). Here,ÆM(k;t) denotes the r.m.s.

massutuationonalengthsalek 1

attimet.

(4)

sales(see, however,the omments at the end of

Se-tionIV.1). Therefore,themagnitudeof ÆM

M

anhange

onlybyafatorindependentofk,andheneitfollows

that

ÆM

M (k;t

f

(k)) = onst; (16)

whih is the denition of a sale-invariant spetrum

[30℄.

III Theory of Cosmologial

Per-turbations

On sales larger than the Hubble radius the

Newto-niantheoryof osmologialperturbationsis

inapplia-ble, and a general relativisti analysis is needed. On

these sales, matter is essentially frozen in omoving

oordinates. However,spae-timeutuationsanstill

inreaseinamplitude.

In priniple, it is straightforward to work out the

generalrelativistitheoryoflinearutuations[31℄. We

linearizetheEinsteinequations

G

=8GT

(17)

(where G

is the Einstein tensor assoiated with

the spae-time metri g

, and T

is the

energy-momentumtensorofmatter)aboutanexpandingFRW

bakground(g (0)

;'

(0)

):

g

(x ;t) = g (0)

(t)+h

(x;t) (18)

'(x ;t) = ' (0)

(t)+Æ'(x ;t) (19)

andpikoutthetermslinearinh

andÆ'toobtain

ÆG

= 8GÆT

: (20)

Intheabove,h

istheperturbationinthemetriand

Æ'istheutuationofthemattereld'.

In pratie, there are many ompliations whih

make this analysis highly nontrivial. The rst

prob-lem is \gaugeinvariane" [32℄. Imagine starting with

a homogeneousFRW osmology and introduing new

oordinates whih mix x and t. In terms of the new

oordinates,themetrinowlooksinhomogeneous. The

inhomogeneouspieeofthemetri,however,mustbea

pureoordinate(or"gauge")artifat. Thus,when

ana-lyzingrelativistiperturbations,aremustbetakento

fatorouteets duetooordinatetransformations.

Therearevariousmethodsofdealingwithgauge

ar-tifats. Thesimplest andmostphysialapproahisto

fousongaugeinvariantvariables,i.e.,ombinationsof

themetriand matter perturbations whih are

invari-antunder linearoordinatetransformations.

Thegaugeinvarianttheory ofosmologial

pertur-bationsisinpriniplestraightforward,although

tehni-ally rathertedious. InthefollowingI willsummarize

themainstepsandreferthereaderto[33℄forthedetails

and further referenes (see also [34℄ for apedagogial

introdutionand[35,36,37,38,39,40,41,42℄forother

approahes).

We onsider perturbations about a spatially at

Friedmann-Robertson-Walkermetri

ds 2

=a 2

()(d 2

dx 2

) (21)

where isonformaltime(relatedtoosmitimet by

a()d=dt). Atthelinearlevel,metriperturbations

an be deomposed into salar modes, vetor modes

andtensormodes(gravitationalwaves). Inthe

follow-ing,wewillfousonthesalarmodessinetheyarethe

onlyoneswhihoupletoenergydensityandpressure.

Asalarmetriperturbation(see[43℄forapreise

def-inition) an be written in termsof four free funtions

ofspaeandtime:

Æg

=a

2

()

2 B

;i

B

;i 2( Æ

ij +E

;ij )

: (22)

The next step is to onsider innitesimal

oordi-nate transformations whih preservethe salarnature

of Æg

, and to alulate the induedtransformations

of; ;B andE. Thenwendinvariantombinations

tolinearorder.(Notethatthereareingeneralno

om-binationswhih areinvariantto allorders[44℄.) After

somealgebra,itfollowsthat

= +a

1

[(B E 0

)a℄ 0

(23)

=

a 0

a

(B E 0

) (24)

aretwoinvariantombinations(aprimedenotes

dier-entiationwithrespetto).

Perhapsthe simplest way [33℄ to derive the

equa-tionsofmotionforgaugeinvariantvariablesisto

on-sider the linearized Einstein equations (20) and to

write them out in the longitudinal gauge dened by

B = E = 0, in whih = and = , to diretly

obtaingaugeinvariantequations.

Forseveraltypesofmatter, inpartiular forsalar

eldmatter,ÆT i

j Æ

i

j

whihimplies= . Hene,the

salar-typeosmologialperturbationsan inthis ase

be desribed by a single gauge invariant variable. In

theaseofasinglesalarmattereld',theperturbed

Einstein equationsanbeombinedtoyield



+ H 2

 '

_ '

_

+ k

2

a 2

+2 _

H 2H

 '

_ '

= 0: (25)

Forutuationswith saleslargerthantheHubble

ra-dius,thisequationofmotionanbewrittenintheform

ofanapproximate onservationlaw[28,45,39,46,47,

48℄

_ ' 2

_

= 0 (26)

where

= 2H

1

_

+

(5)

During the period of slow-rolling of the salar eld

(andalsoforsingleperfetuids),(26)beomessimply

_

=0.

If the equationof state of matter is onstant, i.e.,

w=onst, then _

=0impliesthat therelativisti

po-tential is time-independent on sales larger than the

Hubble radius, i.e. (t)=onst. Duringatransition

from an initial phase with w = w

i

to a phase with

w=w

f

,hanges. Inmanyases,agood

approxima-tiontothedynamisgivenby(26)is

1+w (t

i ) =

1+w (t

f

); (28)

To make ontat with late time matter

perturba-tions andwith theNewtonian intuition, itis usefulto

note that, asaonsequeneof theEinstein onstraint

equations,atthetimet

H

(k)whenamodekrossesthe

Hubbleradius,isameasureofthefrationaldensity

utuations:

(k;t

H (k))

Æ

(k;t

H

(k)): (29)

Theprimordialperturbationsinaninationary

os-mologyaregeneratedbyquantumutuations(seealso

[49, 50℄). Sine the sale of the utuations of

inter-est todaywaslargerthantheHubbleradiusforalong

time, it is ruial to onsider notjust matter

utua-tions, but alsothegravitationalutuationsdesribed

at a lassiallevelin the previous paragraphs. Thus,

the generation and evolution of osmologial

utua-tions in inationary osmology beomes a problem of

quantumgravity. However,duetothefatthatgravity

is an attrative fore, we know that the amplitudeof

the utuations had to have been extremely small in

theveryearlyUniverse. Hene,aperturbativeanalysis

willbewelljustied. Whatfollowsisaverybrief

sum-maryoftheuniedanalysisofthequantumgeneration

and evolution of perturbations in an inationary

Uni-verse(foradetailedreviewsee[33℄). Thebasipointis

that atthelinearizedlevel,thesystemof gravitational

andmatterperturbationsanbequantizedina

onsis-tent way. Theuse of gauge invariant variables makes

theanalysisbothphysiallylearandomputationally

simple. Dueto theEinstein onstraintequationwhih

ouples metri and matter utuations, there is only

onesalareld degreeof freedomto bequantized(see

[51℄ and[52℄fortheoriginalanalysis).

Therststepofthisanalysisistoexpandthe

grav-itational and matter ationsto quadrati order in the

utuationvariablesaboutalassialhomogeneousand

isotropibakgroundosmology. Fousingonthesalar

metrisetor,itturns outthatoneanexpressthe

re-sultingationforthequantumutuationsintermsof

a single gauge invariant variable whih is a

ombina-tion ofmetri andmatter perturbations, and that the

resultingationreduestotheationofasinglegauge

[52,51℄(the timedependenereetstheexpansionof

thebakgroundspae-time)Weanthususestandard

methods toquantize this theory. If weemploy

anon-ialquantization, thenthe modefuntions of theeld

operatorobeythesameequationsaswederivedinthe

gauge-invariantanalysis oflassialrelativisti

pertur-bations.

Thetimedependeneofthemassleadstoequations

whih havegrowingmodeswhih orrespondto

parti-le prodution or equivalently to the generation and

ampliationof utuations. Sine ination

exponen-tially dilutes the density of pre-existing matter, it is

reasonableto assume that the perturbations start o

(e.g. atthebeginningofination)in thevauumstate

(dened asastatewith nopartiles with respetto a

loal omoving observer). The state dened this way

will not be the vauum state from the point of view

of anobserverat alater time. TheBogoliubovmode

mixingtehniqueanbe usedto alulatethenumber

density of partiles at alater time. In partiular,

ex-petation values of eld operators suh as the power

spetrumanbeomputed.

Ifthebakgroundsalareldisrollingslowly,then

theresultingmassutuationsaregivenby

ÆM

M (k;t

f (k))

3H 2

j'_

0 (t

i (k))j

_ ' 2

0 (t

i (k))

= 3H

2

j'_

0 (t

i (k))j

(30)

Thisresultannowbeevaluatedforspei modelsof

inationto nd theonditionson thepartile physis

parameterswhih giveavalue

ÆM

M (k;t

f

(k))10 5

(31)

whihisrequiredifquantumutuationsfromination

aretoprovidetheseedsforgalaxyformationandagree

withtheCMBanisotropydata.

Forhaotiinationwithapotential

V(')= 1

2 m

2

' 2

; (32)

we an solve the slow rolling equations for the

ina-tonandobtaintherequirementm10 13

GeVtoagree

with(31). Similarly, fora quarti potential with

ou-plingonstant,theondition10 12

isrequiredin

ordernotto onit with observations. Thus, in both

examplesoneneedsaverysmallparameterinthe

par-tilephysismodel. It hasbeenshownquitegenerally

[53℄thatsmallparametersarerequiredifinationisto

solvetheutuationproblem.

Tosummarize, the main results of the analysis of

density utuationsin inationary osmology are: (1)

QuantumvauumutuationsinthedeSitterphaseof

aninationaryUniversearethesoureofperturbations.

(2)As aonsequeneofthehangein thebakground

(6)

ra-Infat,unlessthepartilephysismodelontainsvery

smallouplingonstants,thepreditedutuationsare

in exess of those allowed by the bounds on osmi

mirowaveanisotropies. (3) The quantum generation

and lassial evolution of utuations an be treated

in aunied manner. The formalism is no more

om-pliated that thestudy of afree salareld in atime

dependentbakground. (4)InationaryUniverse

mod-elsgeneriallyprodueanapproximatelysaleinvariant

Harrison-Zel'dovihspetrum(16).

IV Parametri Resonane and

Reheating

Reheatingisanimportantstageininationary

osmol-ogy. It determines the stateof the Universe after

in-ation and has onsequenes for baryogenesis, defet

formationandotheraspetsofosmology.

Afterslowrolling,theinatoneld beginsto

osil-late uniformly in spae aboutthe true vauum state.

Quantum mehanially,thisorrespondstoaoherent

stateofk=0inatonpartiles. Duetointerationsof

theinatonwithitselfandwithotherelds,the

oher-entstatewilldeayintoquantaofelementarypartiles.

This orrespondsto post-inationary partile

produ-tion.

Reheatingisusuallystudiedusingsimplesalareld

toymodels. Theonewewilladopthereonsistsoftwo

realsalarelds,theinaton'interatingwitha

mass-less salar eld representing ordinary matter. The

Lagrangianis

L = 1

2

'

' 1

2 m

2

' 2

+ 1

2

1

2 g

2

' 2

2

; (33)

withm10 13

GeV(seeSetionIIIforajustiationof

this hoie), and g 2

denotingthe interation oupling

onstant. Thebaremassandself interationsof are

negleted.

Intheelementary theory of reheating (seee.g. [54℄

and[55℄),thedeayoftheinatonwasalulatedusing

rstorderperturbationtheory. Thedeayrate

B of'

typially turns outto bemuh smallerthan the

Hub-bleexpansionrateat theend ofination(see[7℄ fora

workedexample). Thedeayleadstoadereaseinthe

amplitudeof 'whih anbe approximatedbyadding

an extra dampingterm to the equation of motion for

':



'+3H'_+

B _ ' = V

0

('): (34)

From the above equation it follows that as long as

H >

B

, partileprodution isnegligible. Duringthe

phase of oherent osillation of ', the energy density

andheneHaredereasing. Thus,eventuallyH =

B ,

and at that pointreheating ours (the remaining

en-ergydensityin'isveryquiklytransferredto

parti-les). However,whenthisours,thematter

tempera-tureis muh smallerthantheenergysale ofination

(reheatingisaslowproess). ThiswouldimplynoGUT

baryogenesisand noGUT-saledefetprodution. As

we shall see, these onlusions hange radially if we

adoptanimprovedanalysis ofreheating.

As was rst realized in [56℄, the above analysis

misses anessentialpoint. Tosee this,wefouson the

equation of motion for the matter eld . The

equa-tions for the Fourier modes

k

in the presene of a

oherentinatoneld osillatingwithamplitudeA,

'(t) = Aos(mt); (35)

is



k +3H_

k +(k

2

p +m

2

+

1

2 g

2

A 2

os(2mt))

k

= 0; (36)

where k

p

= k=a is the time-dependent physial

wavenumber, and m 2

=

1

2 A

2

(for other toy models a

similar equation is obtained, but with a dierent

re-lationship betweenthe massand theoeÆient of the

osillatingterm).

Letusforthemomentneglettheexpansionofthe

Universe. In this ase,the fritionterm in (36)drops

out,k

p

istime-independent,andEquation(36)beomes

aharmoniosillatorequationwithaperiodially

vary-ingmass. Inthe mathematisliterature,thisequation

is alled the Mathieu equation. It is wellknown that

there is an instability. In physis, the eet is known

asparametri resonane(seee.g. [57℄). At

frequen-ies !

n

orresponding to half integermultiples of the

frequeny! ofthevariationofthemass,i.e.

! 2

k =k

2

p +m

2

= (

n

2 !)

2

n=1;2;:::; (37)

there are instabilitybands with widths!

n

. For

val-ues of !

k

within the instabilityband, the value of

k

inreasesexponentially:

k e

t

with g

2

A 2

!

: (38)

Inmodelsof haoti inationA m

pl

. Hene,unless

g 2

isunnaturallysmall(atypialvalueisg 2

m=m

pl ),

itfollowsthatH. Theonstantisalledthe

Flo-quetexponent.

Sine the widths of the instability bands derease

as a power of the (small) oupling onstant g 2

with

inreasingn, forpratialpurposesonlythelowest

in-stabilityband isimportant. Itswidth is

!

k

gA: (39)

Note, in partiular, that there is no ultraviolet

diver-genein omputing thetotalenergy transferfrom the

'totheeld duetoparametriresonane[56℄.

Itis easyto inlude the eetsof theexpansionof

theUniverse (seee.g. [56,58, 59℄). Themaineet is

(7)

modeslowlyentersandleavestheresonanebands. As

aonsequene,anymodeliesintheresonanebandfor

onlyanitetime.

The rate of energy transfer is given by the phase

spae volume of thelowestinstability band multiplied

bytherateofgrowthofthemodefuntion

k

. Usingas

an initial onditionfor

k

the value

k

H given by

the magnitude of the expeted quantum utuations,

weobtain

_

(

!

2 )

2

!

k He

t

: (40)

Hene,theenergytransferwillproeedfastonthetime

sale of the expansion of the Universe. There will be

explosive partile prodution, and the energy density

inmatterattheendofreheatingwillbeapproximately

equaltotheenergydensityattheendofination.

Theaboveisasummaryofthemainphysisofthe

modern theory of reheating. The atual analysis an

berened in manyways(see e.g. [58,59, 60℄, and, in

the toymodel onsidered here, [61℄). First ofall, itis

easytotaketheexpansionoftheUniverseintoaount

expliitly (bymeans ofatransformation of variables),

to employ an exatsolution of the bakgroundmodel

andtoreduethemodeequationfor

k

toanequation

whihalsoadmitsexponentialinstabilities.

The next improvement onsists of treating the

eld quantum mehanially (keeping ' as a lassial

bakground eld). At this point, the tehniques of

quantum eld theory in a urved bakground an be

applied. There is no need to impose artiial

lassi-al initial onditions for

k

. Instead, wemay assume

that starts in its initial vauum state. The

Bogoli-ubovmodemixing tehniquean be used to ompute

thenumberof partilesatlatetimes.

Notethatthestateofafterparametriresonane

is notathermalstate. Thespetrumonsists ofhigh

peaks in distint wavebands. An important question

ishowthisstatethermalizes. Forsomereentprogress

onthisissuesee[62,63℄. Sinethestateafterexplosive

partile prodution is not athermal state, it is useful

tofollow[58℄andallthisproess\preheating"instead

of reheating.

Note that the details of theanalysis of preheating

are quite model-dependent. In fat [58, 60℄, in most

modelsonedoesnotgetthekindof\narrow-band"

res-onanedisussedhere,but\broad-band"resonane. In

this ase,theenergytransferisevenmoreeÆient.

Reently [64℄ it has been argued that parametri

resonanemayleadto resonantampliationof

super-Hubble-sale osmologialperturbations. Thepointis

that inthepreseneof anosillatinginatoneld, the

equation of motion (25) for the osmologial

pertur-bations ontainsaontribution to themass term(the

oeÆient of )whih is periodiallyvarying in time.

Hene, the equation takes on a similar form to the

Mathieu equationdisussed above(36). Insome

mod-with wavelength larger than the Hubble radius,

lead-ingtotheapparentampliationofsuper-Hubble-sale

modes. Suh aproess doesnot violateausality[48℄

sine it is driven by the inaton eld whih is

oher-entonsuper-Hubble salesat theend ofination asa

onsequeneof theausal dynamis of an inationary

Universe.

Theanalysis ofEquation(25)during theperiod of

reheatingis, however, ompliated by asingularityin

theoeÆientsofboth _

andattheturningpointsof

thesalarmattereld'. Thissingularitypersistswhen

usingthe`onservationlaw'form (26)oftheequation:

when '_ = 0, one annot immediately draw the

on-lusion that _ = 0. Note, also, that a large inrease

in thevalue of during reheatingis predited by the

usual theory of osmologialutuationswhih treats

thereheating period asa smoothhange in the

equa-tionofstatefrom that of nearlydeSitterto that ofa

radiation-dominatedUniverse(see(28)). Careful

anal-ysesforsimplesingle-eld[48,65℄modelsdemonstrated

thatthere isindeedno netgrowthofthephysial

am-plitude of gravitational utuations beyond what the

usualtheoryofosmologialperturbationspredits(see

also [66, 67℄ for earlierresults supporting this

onlu-sion). Thereisinreasingevidenethatthisonlusion

holdsin general formodels withpurely adiabati

per-turbations[68,69℄.

Intheaseofmultiplemattereldmodelsthereare

extratermsontherighthand side oftheequations of

motion(25)and(26)whihare notexponentially

sup-pressedonlengthsaleslargerthantheHubbleradius.

Thesetermsarerelatedtotheexisteneofisourvature

utuations.Asrstdemonstratedin[73℄,insuh

mod-els exponential inrease in the amplitude of during

reheatingisindeedpossible. However,inmanymodels

theperturbationmodeswhih anundergoparametri

ampliation during reheating are exponentially

sup-pressedduringination[70,71,72℄,andtheythushave

anegligibleeet onthenalamplitudeof. The

ri-terion for models (suh as the one proposed in [73℄)

to have exponential growthof the physial amplitude

of osmologial perturbations during ination is that

thereisanisourvature/entropymodewhihisnot

sup-pressedduringination[74℄. Theresultingexponential

ampliation of utuations renders these models

in-ompatiblewiththeobservationalonstraints,even

in-ludingbak-reationeets[75℄.

Sinetheexponentialpartileprodutionrate

dur-ingreheatingreliesonaparametriresonane

instabil-ity,it isreasonableto beonernedwhethertheeet

willsurviveinthepreseneofnoise. There arevarious

souresof noiseto be onerned about. Firstly, there

arethequantumorthermalutuationsintheinaton

itself,the utuationswhihin inationary osmology

arethesoureoftheobservedstrutureintheUniverse.

(8)

onsideredtheeetsofnoiseintheinatoneldonthe

dynamisof.

Weassumethatthedynamisoftheinatonis

de-sribed by periodi motion with superimposed noise

givenbyafuntion q(t;x),i.e.

'(t;x) = Aos(!t)+q(t;x): (41)

Forsimpliity, we havenegletedtheexpansionofthe

Universe. In the ase of spatially homogeneous noise

onsidered in [76℄, the dynamis of still deomposes

into independentFouriermodes whih obeythe

equa-tion



k +k

2

k +

m 2

+p(!t)+q(t)

k

=0; (42)

wherep(y)isafuntionwithperiod2. Thisequation

anbewrittenintheusualwayasahomogeneousrst

order22matrixdierentialequationwithaoeÆient

matrixontainingnoise.

Forderivingtheresultsmentionedbelow,itis

suÆ-ienttomakeertainstatistialassumptionsaboutq(t).

Weassume that thenoiseis drawn from somesample

spae(whihforhomogeneousnoiseanbetakento

be =C(<)), and that the noiseis ergodi, i.e. the

time average of the noise is equal to the expetation

valueof thenoiseoverthesamplespae. Inthisase,

itanbeshownthat thegeneralizedFloquetexponent

ofthesolutionsfor

k

is welldened. Toobtainmore

quantitative information about (q) it is neessaryto

makefurtherassumptionsaboutthenoise. Weassume

rstly that the noise q(t) is unorrelated in time on

sales larger than the time period T of the periodi

motiongivenbyp,andisidentiallydistributed forall

realizationsof thenoise. Seondly, weassumethat

re-stritingthenoiseq(t;)tothetimeinterval0t<T,

thenoisesampleswithinthesupportoftheprobability

measurellaneighborhood,in C(0;T),oftheorigin.

The main result whih was proved in [76℄ is that

the Floquet exponent (q) in the presene of noise is

stritlylargerthantheexponent(0)intheabseneof

noise

(q) > (0); (43)

whihdemonstratesthatthepreseneofnoiseleadsto

astritinreaseintherateofpartileprodution. The

proofwasbasedonanappliationofFurstenberg's

the-orem,atheoremonerningtheLyapunovexponentof

produtsofindependentidentiallydistributedrandom

matriesf

j

:j=1;:::;Ng.

Forinationaryreheating, theaboveresultimplies

thatnoiseintheinatoneliminatedthestabilitybands

of the system, and that all modes

k

grow

exponen-tially. Theresultwasextendedtoinhomogeneousnoise

in[77℄. Theanalysisismathematiallymuhmore

om-pliatedsinetheFouriermodesnolongerdeoupleand

theproblemisaprobleminthetheoryofpartial(rather

similar result to (43) an be derived, with the strit

inequalityreplaedby.

If we replae time derivatives by spatial

deriva-tives (denoted by a prime) in (42), we obtain the

time-independent Shrodinger equation for a

one-dimensional non-relativisti eletrongas in a periodi

potentialsubjetto aperiodinoise

~

E (x) = 00

+[V

p (x)+V

q

(x)℄ ; (44)

where V

p

is periodi, V

q

represents the random noise

ontributiontothepotential,and ~

Eisaonstant. Our

resultsimplythatin thepreseneofrandomnoise,the

periodi solutionsoftheequationwithperiodi

poten-tial (the Bloh waves) beome loalized, and that the

loalizationlengthdereasesasthenoiseamplitude

in-reases[78℄. Thus, weobtainanewproofofAnderson

loalization [79℄.

V Problemsof Inationary

Cos-mology

V.A FlutuationProblem

A generiproblem for allrealizations of

potential-driven ination studied up to now onerns the

am-plitudeofthedensityperturbationswhihareindued

byquantumutuationsduringtheperiodof

exponen-tial expansion [27, 28℄. From the amplitude of CMB

anisotropiesmeasuredbyCOBE,andfromthepresent

amplitudeofdensityinhomogeneitiesonsalesof

lus-ters of galaxies, it follows that the amplitude of the

massutuationsÆM=Monalengthsalegivenbythe

omoving wavenumberk at thetime t

f

(k) when that

salerossestheHubbleradiusintheFRWperiodisof

theorder10 5

.

However,as wasdisussed in detailin Setion III,

the present realizations of ination based on salar

quantum eld matter generially [53℄ predit a muh

largervalueoftheseutuations,unlessaparameterin

the salar eld potential takes on a very small value.

For example, as disussed at the end of Setion III,

in asingle eld haoti inationary model with

quar-tipotentialthemassutuationsgeneratedareofthe

order 10 2

1=2

. Thus, in order notto onit with

ob-servations,avalueofsmallerthan10 12

isrequired.

There have beenmany attempts to justify suh small

parameters based on spei partile physis models,

butnosingleonviningmodelhasemerged.

With the reentdisovery[64, 48℄ that long

wave-length gravitationalutuations may beamplied

ex-ponentiallyduring reheating,anewaspet ofthe

u-tuationproblemhasemerged. Allmodelsinwhihsuh

ampliationours(seee.g.[74℄foradisussionofthe

(9)

large,independentofthevalueoftheouplingonstant

[75℄.

V.B Trans-Plankian Problem

Inmanymodelsofination,inpartiularinhaoti

ination, theperiodofinationis solongthat

omov-ing sales of osmologialinterest todayorresponded

toaphysialwavelengthmuhsmallerthanthePlank

length at the beginning of ination. In extrapolating

the evolution of osmologial perturbations aording

tolineartheorytotheseveryearlytimes,weare

impli-itly making the assumptions that the theory remains

perturbative to arbitrarily high energies and that it

anbedesribedbylassialgeneralrelativity. Bothof

these assumptionsbreak down onsuper-Planksales.

Thus thequestionarisesasto whetherthepreditions

of the theory are robust against modiations of the

modelonsuper-Planksales.

Asimilarproblemours inblakholephysis[80℄.

The mixing between the modes falling towards the

blak hole from past innity and the Hawking

radia-tion modesemanating to future innity takesplae in

the extreme ultraviolet region, and ould be sensitive

to super-Plankphysis. However,in theaseofblak

holesithasbeenshownthatforsub-Plankwavelengths

atfutureinnity,thepreditionsdonothangeundera

lassofdrastimodiationsofthephysisdesribedby

hangesinthedispersionrelationofafreeeld[81,82℄.

Aswasreently[83,84℄disovered,theresultinthe

aseofinationaryosmologyisdierent: thespetrum

ofutuationsmaydependquitesensitivelyonthe

dis-persionrelationonsuper-Planksales. Ifwetakethe

initial state ofthe utuationsat thebeginningof

in-ation to be given by the state whih minimizes the

energy density, then for ertainof thedispersion

rela-tions onsidered in [82℄, the spetrum of utuations

hangesquiteradially. Theindexofthespetruman

hange(i.e. thespetrumisno longersale-invariant)

andosillationsin thespetrummaybeindued. Note

that forthe lass of dispersionrelations onsidered in

[81℄ thepreditionsarethestandardones.

Theaboveresultsmaybebadnewsforpeoplewho

would like to onsider salar-eld driven inationary

models asthe ultimate theory. However, the positive

interpretationoftheresultsisthatthespetrumof

u-tuations may provide awindowon super-Plank-sale

physis. The present observations an already be

in-terpreted in thesense [85℄ that thedispersionrelation

of theeetiveeld theorywhih emergesfrom string

theoryannotdiertoodrastiallyfromthedispersion

relationofafreesalareld.

V.C Singularity Problem

Salareld-drivenination doesnot eliminate

sin-gularities from osmology. Although the standard

downif matterhas anequationof statewith negative

pressure,asistheaseduringination, neverthelessit

anbeshownthat an initialsingularitypersists in

in-ationaryosmology[86℄. Thisimpliesthatthetheory

is inomplete. In partiular, the physial initial value

problemisnotdened.

V.DCosmologialConstant Problem

Sinetheosmologialonstantatsasaneetive

energydensity,itsvalueisboundedfrom abovebythe

presentenergydensityoftheUniverse. InPlankunits,

the onstraint on the eetive osmologial onstant

eff

is(seee.g. [87℄)

eff

m 4

pl 10

122

: (45)

This onstraintapplies both to the bare osmologial

onstantandtoanymatterontributionwhih atsas

aneetiveosmologialonstant.

Thetruevauumvalue(takenon,tobespei,at

'= 0) of the potential V(') ats asan eetive

os-mologialonstant. Itsvalueisnotonstrainedbyany

partilephysisrequirements(intheabseneofspeial

symmetries). Thustheremustbesomeasyetunknown

mehanismwhihanels(oratleastalmostompletely

anels)thegravitationaleets ofanyvauum

poten-tial energy of any salar eld. However, salar

eld-driveninationrelies onthealmostonstantpotential

energyV(')duringtheslow-rollingperiodating

grav-itationally. How an one be sure that the unknown

mehanism whih anels the gravitational eets of

V(0) does not also anel the gravitational eets of

V(') during the slow-rolling phase? This problem is

theAhilles heelof anysalareld-driven inationary

model.

Supersymmetry alone annot provide a resolution

of this problem. It is true that unbroken

supersym-metryforesV(')=0in thesupersymmetrivauum.

However, supersymmetry breaking will indue a

non-vanishingV(')in thetruevauum after

supersymme-trybreaking,andaosmologialonstantproblemofat

least60ordersofmagnituderemainsevenwiththe

low-est sale of supersymmetry breaking ompatible with

experiments.

VI New Avenues

In the light of the problems of potential-driven

ina-tiondisussed in theprevioussetions, it isof utmost

importane to study realizations of inationwhih do

notrequirefundamentalsalarelds,orompletelynew

avenuestowardsearlyUniverseosmologywhih,while

maintaining(someof)thesuessesofination,address

andresolvesomeofitsdiÆulties.

(10)

os-thepre-big-bangsenario[24℄,andthevaryingspeedof

lightpostulate[88,89℄. Thepre-big-bangsenariois a

modelinwhihtheUniversestartsinanemptyandat

dilaton-dominatedphasewhihleadstopole-law

ina-tion. A nie feature of this theory is that the

meha-nism of super-inationary expansionis ompletely

in-dependent of a potential and thus independentof the

osmologialonstantissue. The senario,however,is

onfronted with agraeful exit problem [90℄, and the

initialonditionsneedtobeveryspeial[91℄(see,

how-ever,thedisussionin [92℄).

It isalsoeasy torealizethat atheory inwhihthe

speedoflightismuhlargerintheearlyUniversethan

atthepresenttimeanleadtoasolutionofthehorizon

andatness problemsof standardosmologyandthus

an provide an alternative to ination for addressing

them. Forrealizationsofthissenariointheontextof

thebraneworldideasseee.g. [93,94,95℄.

String theory may lead to a natural resolution of

someof thepuzzles of inationaryosmology. This is

an area of ative researh. The reader is referred to

[20℄forareviewofreentstudiesofobtainingination

with moduli elds, and to [22℄ for attempts to obtain

ination withbranes. Below, three moreonventional

approahestoaddressingsomeoftheproblemsof

ina-tionwillbesummarized.

VI.A Inationfrom Condensates

Atthepresenttimethereisnodiretobservational

evidenefortheexisteneoffundamentalsalareldsin

nature(inspiteofthefatthatmostattrativeunied

theoriesofnaturerequiretheexisteneofsalareldsin

thelowenergyeetiveLagrangian). Salareldswere

initiallyintroduedtopartilephysistoyieldanorder

parameterforthesymmetrybreakingphasetransition.

Manyphasetransitionsexistin nature;however,inall

ases,theorderparameterisaondensate. Hene,itis

usefulto onsider thepossibility ofobtainingination

usingondensates,andinpartiulartoaskifthiswould

yieldadierentinationarysenario.

Theanalysisofatheorywithondensatesis

intrin-siallynon-perturbative. Theexpetation valueofthe

HamiltonianhHiofthetheoryontainstermswith

ar-bitrarily high powers of the expetation value h'i of

the ondensate. A reent study of the possibility of

obtaining ination in a theory with ondensates was

undertakenin[96℄(seealso[97℄forsomeearlierwork).

InsteadoftrunatingtheexpansionofhHiatsome

ar-bitraryorder,theassumptionwasmadethatthe

expan-sionofhHiinpowersofh'iisasymptotiand,

spei-ally,Borelsummable(ongeneralgroundsoneexpets

thattheexpansionwill beasymptoti-seee.g. [98℄)

hHi = 1

X

n=0 n!( 1)

n

a

n h'

n

i (46)

= Z

1

ds

f(s)

s(sm

pl +h'i)

e 1=s

: (47)

Therst linerepresentsthe originalseries, theseond

line the resummed series. The funtion f(s) is

deter-minedbytheoeÆientsa

n .

Theosmologialsenariois asfollows: the

expe-tation value h'i vanishes at times before the phase

transitionwhentheondensateforms. Afterwards,h'i

evolvesaording to the lassialequations of motion

withthepotentialgivenby(46)(assumingthatthe

ki-netitermassumesitsstandardform). Itaneasilybe

heked that the slow rolling onditions are satised.

However, the slow roll onditions remain satisedfor

all valuesof h'i, thus leading to agraeful exit

prob-lem-inationwill neverterminate.

However, orrelation funtions, in partiular h 2

i,

arein generalinfrareddivergentinthedeSitterphase

of an expanding Universe. Onemust therefore

intro-dueaphenomenologialutoparameter(t)intothe

vauum expetation value (VEV), and replae h'i by

h'i=. It is natural to take (t) H(t) (see e.g.

[99,100℄). Hene,thedynamialsystemonsistsoftwo

oupledfuntionsoftimeh'iand. Aarefulanalysis

showsthatagraefulexitfrominationourspreisely

ifhHitendstozerowhenh'itendsto largevalues.

Asisevident,thesenarioforinationinthis

om-posite eld model is very dierent from the standard

potential-driven inationary senario. It is

partiu-larly interesting that the graeful exit problem from

ination is linked to the osmologial onstant

prob-lem. Note that models of ination based on

ompos-itespresumablydonotsuerfromthetrans-Plankian

problem,thereasonbeingthattheeetiveeldtheory

whihdesribesthestronglyinteratingsystemis

time-translation-invariant during the de Sitter phase. The

physialpitureisthatmodeinterationsareessential,

andareresponsibleforgeneratingtheutuationsona

salekwhenthissaleleavestheHubbleradiusattime

t

i (k).

VI.BNonsingular Universe Constrution

Anotherpossibilityofobtaininginationisby

mak-ing useofamodiedgravitysetor. Morespeially,

we an add to the usual gravitational ation higher

derivative urvature terms. These terms beome

im-portant only at high urvatures. As realized a long

time ago [23℄, spei hoies of the higher

deriva-tive terms an lead to ination. It is well motivated

to onsider eetive gravitational ations with higher

derivativetermswhenstudyingthepropertiesof

spae-time at largeurvatures, sineanyeetive ationfor

lassialgravityobtainedfromstringtheory,quantum

gravity, or by integrating out matter elds, ontains

suhterms. Inourontext,theinterestingquestionis

whether oneanobtainaversionofination avoiding

someoftheproblemsdisussedintheprevioussetion,

speiallywhetheroneanobtainnonsingular

(11)

Most higher derivativegravitytheories have muh

worse singularity problems than Einstein's theory.

However, it is not unreasonable to expet that in the

fundamental theory of nature, be it string theory or

some other theory, the urvatureof spae-time is

lim-ited. In Refs. [101, 102℄ the hypothesis was made

that when the limiting urvature is reahed, the

ge-ometry must approah that of a maximally

symmet-ri spae-time, namely de Sitter spae. The question

now beomes whether it is possible to nd a lass of

higherderivativeeetiveationsforgravitywhihhave

thepropertythatat largeurvaturesthesolutions

ap-proah de Sitter spae. A nonsingular Universe

on-strutionwhihahievesthisgoalwasproposedinRefs.

[103,104℄. Itisbasedonaddingto theEinstein ation

apartiularombinationofquadratiinvariantsofthe

Riemanntensorhosensuhthattheinvariantvanishes

onlyindeSitterspae-times. Thisinvariantisoupled

totheEinsteinationviaaLagrangemultipliereldin

awaythattheLagrangemultiplieronstraintequation

fores the invariant to zero at high urvatures. Thus,

the metri beomes de Sitter and hene geodesially

omplete andexpliitlynonsingular.

Ifsuessful,theaboveonstrutionwillhavesome

veryappealingonsequenes. Consider,forexample,a

ollapsing spatially homogeneousUniverse. Aording

to Einstein'stheory,this Universewill ollapseina

-nitepropertimeto anal\bigrunh"singularity. In

the newtheory, however,the Universewillapproaha

de Sittermodelas theurvatureinreases. Ifthe

Uni-verseislosed,therewillbeadeSitterbounefollowed

by re-expansion. Similarly, spheriallysymmetri

va-uumsolutionsofthenewequationsofmotionwill

pre-sumablybenonsingular,i.e.,blakholeswouldhaveno

singularities in their enters. In two dimensions, this

onstrutionhasbeensuessfullyrealized[105℄.

ThenonsingularUniverseonstrutionof[103,104℄

anditsappliationstodilatonosmology[106,107℄are

reviewedinareentproeedingsartile[108℄. What

fol-lowsisjust averybriefsummaryofthepointsrelevant

to theproblemslistedin SetionV.

TheproedureforobtaininganonsingularUniverse

theory[103℄isbasedonaLagrangemultiplier

onstru-tion. Startingfrom theEinstein ation,onean

intro-due Lagrangemultiplierselds '

i

oupledto seleted

urvatureinvariantsI

i

,andwithpotentialsV

i ('

i )

ho-sen suh that at low urvature the theory redues to

Einstein'stheory,whereasathigh urvaturesthe

solu-tionsaremanifestlynonsingular. Theminimal

require-ments for anonsingular theory are that all urvature

invariants remain bounded and the spae-time

mani-fold isgeodesiallyomplete.

It is possible to ahieve this by a two-step

proe-dure. First, we hoose a urvature invariant I

1 (g ) (e.g. I 1

=R )anddemandthatitbeexpliitlybounded

by the '

1

onstraint equation. In a seond step, we

demand that as I (g )approahesits limiting value,

the metri g

approah the de Sitter metri g DS

, a

denite nonsingular metri with maximal symmetry.

Inthisase, allurvatureinvariantsare automatially

bounded(theyapproahtheirdeSittervalues),andthe

spae-timeanbeextendedtobegeodesiallyomplete.

Theseondstepanbeimplementedby[103℄hoosing

aurvatureinvariantI

2 (g

)withthepropertythat

I

2 (g

)=0 , g

=g

DS

; (48)

introduingaseondLagrangemultipliereld'

2 whih

ouples to I

2

, and hoosing a potential V

2 (' 2 ) whih foresI 2

tozeroatlargej'

2 j: S= Z d 4 x p

g[R+'

1 I 1 +V 1 (' 1 )+' 2 I 2 +V 2 (' 2 )℄; (49)

withasymptotionditions

V 1 (' 1 ) ' 1 asj' 1

j!1 (50)

V

2 ('

2

) onst asj'

2

j!1 (51)

V i (' i ) ' 2 i asj' i

j!0 i=1;2: (52)

TherstonstraintrendersR nite, theseond fores

I

2

to zero,andthethird isrequiredinorder toobtain

theorretlowurvaturelimit.

Theinvariant I 2 =(4R R R 2 +C 2 ) 1=2 ; (53)

singles out the de Sitter metri among all

homoge-neousandisotropi metris(in whihase addingC 2

,

the Weyl tensor square, is superuous), all

homoge-neousandanisotropimetris,andallradially

symmet-rimetris.

As a spei example onean onsider the ation

[103,104℄ S = Z d 4 x p g (54)

R+'

1 R (' 2 + 3 p 2 ' 1 )I 1=2 2 +V 1 (' 1 )+V

2 (' 2 ) with V 1 (' 1

) = 12H 2

0 '

2

1

1+'

1

1

ln(1+'

1 )

1+'

1 (55) V 2 (' 2

) = 2

p 3H 2 0 ' 2 2

1+' 2

2

: (56)

It anbe shown that allsolutions of the equations of

motion whih follow from this ation are nonsingular

[103, 104℄. Theyare either periodi aboutMinkowski

spae-time('

1 ;'

2

)=(0;0) orelse asymptotially

ap-proahdeSitterspae(j'

2

j!1).

Oneofthemostinterestingpropertiesofthistheory

isasymptotifreedom[104℄,i.e., theouplingbetween

matterandgravitygoesto zeroat high urvatures. It

(12)

eld) to thegravitationalation in the standardway.

Onendsthat intheasymptotideSitterregions,the

trajetoriesofthesolutionsprojetedontothe('

1 ;'

2 )

plane are unhanged by adding matter. This applies,

for example, in a phaseof de Sitter ontrationwhen

the matter energy density is inreasing exponentially

but does not aet the metri. The physial reason

for asymptoti freedom is obvious: in the asymptoti

regions of phase spae, the spae-time urvature

ap-proahesitsmaximalvalueandthusannotbehanged

evenbyadding anarbitrarilyhighmatter energy

den-sity. Hene,thereisthepossibilitythatthistheorywill

admitanaturalsuppressionmehanismfor

osmologi-alutuations. Ifthiswerethease,thenthesolution

ofthesingularityproblemwouldatthesametimehelp

resolvethe utuationproblem of potential-driven

in-ationaryosmology.

VI.C Bak-Reation of Cosmologial

Perturba-tions

Thelinear theory of osmologialperturbations in

inationaryosmologyiswellstudied. However,eets

beyondlinearorderhavereeivedverylittleattention.

Beyondlinearorder,perturbationsaneetthe

bak-groundin whih theypropagate, an eet well known

fromearlystudies[109℄ofgravitationalwaves. Aswill

besummarizedbelow,thebak-reationofosmologial

perturbations in an exponentiallyexpanding Universe

atslikeanegativeosmologialonstant,asrst

real-izedin theontext ofstudiesofgravitationalwavesin

deSitterspaein [110℄.

Gravitationalbak-reationofosmologial

pertur-bationsonernsitselfwiththeevolutionofspae-times

whih onsist of small utuations about a

symmet-riFriedmann-Robertson-Walkerspae-timewith

met-ri g (0)

. The goalis to study the evolution of spatial

averagesofobservablesintheperturbedspae-time. In

linear theory, suh averaged quantities evolve likethe

orrespondingvariablesin thebakgroundspae-time.

However, beyond linear theory perturbations havean

eet ontheaveragedquantities. Intheaseof

gravi-tationalwaves,thiseet iswell-known[109℄:

gravita-tionalwavesarryenergyandmomentumwhihaets

thebakgroundinwhihtheypropagate. Here,weshall

fousonsalarmetriperturbations.

Theideabehindtheanalysisofgravitational

bak-reation [111℄ is to expand the Einstein equations to

seond order in theperturbations, to assumethat the

rst order terms satisfy the equations of motion for

linearized osmologialperturbations[33℄(hene these

terms anel), to take the spatial average of the

re-mainingterms,andtoregardtheresultingequationsas

equations foranew homogeneousmetrig (0;br)

whih

inludestheeetoftheperturbationstoquadrati

or-der:

G

(g

(0;br)

) = 8G h

T (0)

+

i

(57)

where the eetive energy-momentum tensor

of

gravitational bak-reation ontains the terms

result-ing from spatial averaging of the seond order metri

andmatterperturbations:

=<T

(2)

1

8G G

(2)

>; (58)

wherepointedbraketsstandforspatialaveraging,and

thesupersriptsindiatetheorderin perturbations.

As formulated in (57) and (58), the bak-reation

problem is not independent of theoordinatizationof

spae-timeandheneisnotwelldened. Itispossible

totakeahomogeneousandisotropispae-time,hoose

dierent oordinates, and obtaina non-vanishing

.

This\gauge"problemisrelatedtothefatthatin the

abovepresription,thehypersurfaeoverwhihthe

av-erageistakendependsonthehoieofoordinates.

The key to resolving the gauge problem is to

re-alize that to seond order in perturbations, the

bak-ground variables hange. A gauge independent form

of the bak-reation equation (57) an hene be

de-rived [111℄ by dening bakground and perturbation

variables Q = Q (0)

+ÆQ whih do not hange

un-der linearoordinate transformations. Here,Q

repre-sentsolletivelybothmetriandmattervariables. The

gauge-invariantformofthebak-reationequationthen

looks formally idential to (57), exept that all

vari-ablesarereplaedbytheorrespondinggauge-invariant

ones. We will follow the notation of [33℄, and use as

gauge-invariantperturbationvariablestheBardeen

po-tentials [35℄ and whih in longitudinal gauge

o-inide withtheatualmetriperturbationsÆg

.

Cal-ulations hene simplify greatlyifwework diretlyin

longitudinal gauge. Thesealulationshavebeen

on-rmed[112℄byworkinginaompletelydierentgauge,

makinguseoftheovariantapproah.

In[113℄,theeetiveenergy-momentumtensor

of gravitational bak-reation was evaluated for long

wavelength utuations in aninationary Universe in

whih the matter responsible for ination is a salar

eld 'withthepotential

V(') = 1

2 m

2

' 2

: (59)

Sine in this model there is no anisotropi stress, the

perturbedmetriin longitudinal gaugeanbewritten

[33℄in termsof asinglegravitationalpotential

ds 2

=(1+2)dt 2

a(t) 2

(1 2)Æ

ij dx

i

dx j

; (60)

wherea(t)istheosmologialsalefator.

ItisnowstraightforwardtoomputeG (2)

andT

(2)

in terms of the bakground elds and themetri and

matter utuationsand Æ', Bytakingaveragesand

makinguseof(58),theeetiveenergy-momentum

(13)

Thegeneralexpressionsfortheeetiveenergy

den-sity (2)

= 0

0

andeetivepressurep (2)

= 1

3

i

i

involve

manyterms. However,theygreatlysimplifyifwe

on-sider perturbations with wavelength greater than the

Hubbleradius. Inthis ase,alltermsinvolvingspatial

gradientsarenegligible. Fromthetheoryoflinear

os-mologialperturbations(seee.g. [33℄)itfollowsthaton

saleslargerthantheHubbleradiusthetimederivative

of is also negligibleas longasthe equation of state

of thebakgrounddoesnothange. TheEinstein

on-straintequationsrelatethetwoperturbation variables

and Æ', enabling salar metri and matter

utua-tionstobedesribedintermsofasinglegauge-invariant

potential . During the slow-rollingperiod of the

in-ationaryUniverse, theonstraintequationtakesona

verysimpleformandimpliesthatandÆ'are

propor-tional. Theupshot of these onsiderationsis that

is proportionaltothetwopointfuntion < 2

>,with

a oeÆient tensor whih depends on the bakground

dynamis. Intheslow-rollingapproximationweobtain

[113℄

(2)

' 4V < 2

> (61)

and

p (2)

=

(2)

: (62)

Thisdemonstratesthattheeetiveenergy-momentum

tensor of long-wavelength osmologial perturbations

hasthesameformasanegativeosmologialonstant.

Note that during ination, the phase spae of

in-fraredmodesisgrowing. Hene,themagnitudeofj (2)

j

isalsoinreasing. Hene,thebak-reationmehanism

may lead to a dynamial relaxation mehanism for a

bare osmologial onstant driving ination [114℄. A

similar eet holds for pure gravityat two looporder

in thepreseneofabareosmologialonstant[110℄.

Theinterpretationof (2)

asaloaldensityhasbeen

ritiized, e.g. in [115℄. Instead of omputing

physi-al observables from a spatially averaged metri, one

should omputethe spatial average of physial

invari-antsorreted to quadratiorder in perturbation

the-ory. Work on this issue is in progress [116℄ (see also

[68℄).

VII Conlusions

Inationary osmology is an attrative senario. It

solvessomeproblems ofstandardosmologyandleads

tothepossibilityofaausaltheoryofstruture

forma-tion. The spei preditionsofaninationary model

ofstrutureformation,however,dependonthespei

realizationofination,whihmakestheideaofination

hardtoverifyorfalsify. Manymodelsofinationhave

beensuggested,but atthepresenttimenoneare

suÆ-ientlydistinguishedtoform a\standard"inationary

theory.

There is now a onsistent quantum theory of the

bationswhihdesribestheoriginofutuationsfrom

aninitialvauumstateoftheutuationmodesatthe

beginningofination,andwhihformsthebasisforthe

preisionalulationsofthepowerspetrumofdensity

utuations and of CMB anisotropieswhih allow

de-tailedomparisons with urrent and upoming

obser-vations.

As explained in Setion IV, a new theory of

in-ationary reheating (preheating) has been developed

based on parametri resonane. Preheating leads to

arapid energy transferbetween the inatoneld and

matterat the end of ination, withimportant

osmo-logialonsequenes. Reentdevelopmentsinthisarea

are the realization that long wavelength gravitational

utuations may beamplied exponentiallyin models

with anentropy perturbation mode whih is not

sup-pressed during ination, and the study of the eets

ofnoise in theinaton eld onthe resonaneproess,

leadingto theresultthat suhnoiseatually enhanes

theresonane(thisresult alsoleadsto anewproof of

Andersonloalization).

Note, however, that there are important

onep-tualproblems for salareld-driven inationary

mod-els. Foursuhproblemsdisussedin SetionVarethe

utuation problem, the trans-Plankian problem, the

singularityproblemandtheosmologial onstant

prob-lem,thelastofwhihistheAhillesheelofthese

ina-tionarymodels.

It maybethat aonviningrealizationof ination

willhavetowaitforanimprovementinour

understand-ingoffundamentalphysis. Somepromisingbut

inom-pleteavenueswhihaddresssomeoftheproblems

men-tionedaboveandwhihyieldinationaredisussedin

SetionVI.

Aknowledgements

I wish to thank Prof. J.A.S. Lima for the

invita-tiontopresentthisleture, andforhis wonderful

hos-pitalityduringmyvisitto Brazil. Ialsowish tothank

Prof. RihardMaKenzieforhospitalityatthe

Univer-sitede MontrealandProfs. JimCline andRobMyers

forhospitalityatMGillUniversityduringthetimethis

manusriptwasompleted. Theauthorissupportedin

partbytheUS Departmentof Energyunder Contrat

DE-FG02-91ER40688,TaskA.

Referenes

[1℄ A.Guth, Phys.Rev.D23,347(1981).

[2℄ A. Linde, `Partile Physis and Inationary

Cosmol-ogy'(Harwood,Chur,1990).

[3℄ S.BlauandA.Guth,`InationaryCosmology,'in`300

YearsofGravitation'ed.byS.HawkingandW.Israel

(14)

[5℄ A. Liddle and D. Lyth, `Cosmologial Ination and

Large-SaleStruture'(CambridgeUniv.Press,

Cam-bridge,2000).

[6℄ D.LythandA.Riotto, Phys.Rept.314,1(1999).

[7℄ R. Brandenberger, `InationaryCosmology: Progress

andProblems',hep-ph/9910410.

[8℄ R.Brandenberger,`StatusReviewofInationary

Cos-mology',hep-ph/0101119.

[9℄ D.Kazanas, Ap.J.241,L59(1980).

[10℄ W.Press, Phys.Sr.21,702(1980).

[11℄ G.ChibisovandV.Mukhanov,`GalaxyFormationand

Phonons,'LebedevPhysialInstitutePreprintNo.162

(1980);

G.ChibisovandV.Mukhanov, Mon.Not.R.Astron.

So.200,535(1982).

[12℄ V.Lukash, Pis'maZh.Eksp.Teor.Fiz.31,631(1980).

[13℄ K.Sato, Mon.Not.R.Astron.So.195,467(1981).

[14℄ R.Brandenberger, Phys.Lett.B129,397(1983).

[15℄ A.Linde, Phys.Lett.B108,389(1982).

[16℄ A. Albreht and P. Steinhardt, Phys.Rev. Lett.48,

1220(1982).

[17℄ A.Linde, Phys.Lett.B129,177(1983).

[18℄ A.Linde, Phys.Rev.D49,748(1994).

[19℄ P. Binetruy and G. Dvali, Phys. Lett. D 388, 241

(1996);

E.Halyo, Phys.Lett.B387,43(1996);

A. Linde and A. Riotto, Phys. Rev. D 56, R1844

(1997).

[20℄ T.Banks,`RemarksonMTheoretiCosmology',

hep-th/9906126.

[21℄ P. Horava and E. Witten, Nul. Phys. B 460, 506

(1996).

[22℄ G.DvaliandS.Tye, Phys.Lett.B450,72(1999);

N. Kaloper and A. Linde, Phys.Rev.D 59, 101303

(1999).

[23℄ A.Starobinski, Phys.Lett.B91,99(1980).

[24℄ M. Gasperini andG. Veneziano, Astropart. Phys.1,

317(1993).

[25℄ T.Damour and V. Mukhanov, Phys.Rev.Lett. 80,

3440(1998).

[26℄ V. Mukhanov and G.Chibisov, JETP Lett. 33,532

(1981).

[27℄ A.GuthandS.-Y.Pi, Phys.Rev.Lett.49,110(1982);

S.Hawking, Phys.Lett.B115,295(1982);

A.Starobinski, Phys.Lett.B117,175(1982);

V.Mukhanov, JETPLett.41,493(1985).

[28℄ J.Bardeen,P.SteinhardtandM.Turner, Phys.Rev.

D28,1809(1983).

[29℄ R.SahsandA.Wolfe, Ap.J.147,73(1967).

[30℄ E.Harrison, Phys.Rev.D1,2726 (1970);

Ya.B. Zel'dovih, Mon. Not.R. astron. So.160, 1p

[31℄ E.Lifshitz, Zh.Eksp.Teor.Fiz.16,587(1946);

E. Lifshitz and I. Khalatnikov, Adv. Phys. 12, 185

(1963).

[32℄ W.PressandE.Vishnia, Ap.J.239,1(1980).

[33℄ V. Mukhanov, H. Feldman and R. Brandenberger,

Phys.Rep.215,203(1992).

[34℄ R. Brandenberger, H. Feldman, V. Mukhanov and

T.Prokope,`GaugeInvariantCosmologial

Perturba-tions: TheoryandAppliations,'publ.in\TheOrigin

of Struture inthe Universe," eds. E. Gunzig and P.

Nardone(Kluwer,Dordreht,1993).

[35℄ J.Bardeen, Phys.Rev.D22,1882(1980).

[36℄ R.Brandenberger,R.KahnandW.Press, Phys.Rev.

D28,1809(1983).

[37℄ H.KodamaandM.Sasaki, Prog.Theor.Phys.Suppl.

No.78,1(1984).

[38℄ R.DurrerandN.Straumann, Helvet.Phys.Ata61,

1027(1988);

R.Durrer, Fund.CosmiPhys.15,209(1994),

astro-ph/9311040.

[39℄ D.Lyth, Phys.Rev.D31,1792(1985);

D. Lyth and M. Mukherjee, Phys. Rev. D 38, 485

(1988).

[40℄ J.HwangandE.Vishnia, Ap.J.353,1(1990).

[41℄ G.F.R. Ellis and M. Bruni, Phys.Rev. D 40, 1804

(1989);

G.F.R. Ellis, J. Hwang andM. Bruni, Phys.Rev.D

40,1819(1989).

[42℄ D. Salopek and J. Stewart, Phys. Rev. D 51, 517

(1995).

[43℄ J.Stewart, Class.QuantumGrav.7,1169(1990).

[44℄ J. Stewart and M. Walker, Pro. R.So. A 341, 49

(1974).

[45℄ R.BrandenbergerandR.Kahn, Phys.Rev.D28,2172

(1984).

[46℄ R.Brandenberger,\ModernCosmologyandStruture

Formation", in `CP Violation and the Limits of the

Standard Model (TASI94)', ed. J.Donoghue(World

Sienti,Singapore,1995),astro-ph/9411049;

R. Brandenberger,in`Physisof theEarly Universe,'

pro. of the 1989 Sottish Univ. Summer Shool in

Physis,ed.byJ.Peaok,A.HeavensandA.Davies

(SUSSPPubl.,Edinburgh,1990);

R. Brandenberger, `Letures on Modern Cosmology

andStrutureFormation',in`PartilesandFields',ed.

by O. Eboli and V. Ribelles (WorldSienti,

Singa-pore1994).

[47℄ J. Martin and D. Shwarz, Phys. Rev. D 57, 3302

(1998);

M.Gotz, Mon.Not.Roy.Astron.So.295,873(1998),

astro-ph/9704271.

[48℄ F.FinelliandR.Brandenberger, Phys.Rev.Lett.82,

1362(1999).

[49℄ J.Bardeen,unpublished(1984).

[50℄ R.Brandenberger, Nul.Phys.B245,328(1984).

Referências

Documentos relacionados

We onsider the saling behavior in the ritial domain of superondutors at zero external

trap and laser system being built in Rio, to trap and perform high resolution spetrosopy on.. old hydrogen, is presented along with a disussion on the tehniques and

An ultrastati spaetime admits a globally dened oordinate system in whih the omponents of the metri tensor are time-. independent and the onditions g00 = 1 and goi =

Four-jet Higgs boson andidate with a reonstruted Higgs boson mass of 112.9 GeV/ 2. All four jets

and magneti harges, being dierent from monopoles (magneti or eletri), the topologial ontent.. of monopoles, surfae terms, Dira strings and hiral struture in the inlusion of

the ultra high energy osmi rays [21℄. From

are not modied by nonommutativity sine they are linear and no Moyal produts are required. The propagators are the same as in the

on new phenomena whih are not the usual neutrino osillations indued by masses and mixing... These solutions are based on dierent assumptions: a) resonant spin-avor preession