r

II
'

5'6 l5?B' numter fvot.,rne

PHYSICSLETTERS

25 July1985

i
INFLATIONARY UNIVERSE GENERATED BY THE COMBINED ACTION OF A SCALARFIELD AND GRAVITATIONALVACUUM POLARIZATION L.A, KOFMAN
Ins|tute ol As!/ophysits and Ao,lospheric Phllrcs, Estonn,t AcoLlem\ ol S.lenfts. Ta/Iu. Torctete. 202444, USSR

A , D .L I N D E
P.N lzbedet Ph)sical Institae. Academ) of Scienrcs, Mas.ar, I179)1, USSR

and A.A. STAROBINSKY

Mascor', I 17134, USSR The lAndax lnstitute for Theoteticdl Ph)sics,Academr ol Sciences. Re.eived 25 Fbruary 1985

The chaolic innarionary unilerse scenano based on a Joint account o[ vacuum polarization and scalar field effects is compared10 thal in th separate c o n s r d e r e dT h e p r o b r b r h r ) f o r r h e u n i v e r s et o g o t h r o u g h t h e i n l l a l i o n a r y s t a g ei n c r e a s e s in the separale scenarios. s c n a n o sT . h e r o t a l e x p a n s i o nd u r i n g t h e i n f l a t r o n a r ys l a 8 er s e q u a l l o t h e p r o d u c t o f t h e e x p a n s i o n s

1. In the development of cosmologyduring the last fcw years,everincreasing hopes are related to models in which the universewas exponentially (or quasi-exponentially) expandingat very early stages of its evolution. There are two main types of models.The modelsof the lust type are connectedwith the account of the one{oop corections to the Einstein equationsand higher-derivative terms in the gravitationalaction [], 2l . The second-type models are basedon the investiof a slow evoluSationof cosnologicalconsequences tion of a scalarfield g in the exponentially expanding (inflationary) universe[3-5]. Below, thesemodels will be calledfor brevity type I and type II modelsrespectively.For an extensivereview of the history of the development and of thesemodels,their advantages difficulties and alsoof the objectiveswhich were pursuedby their creatorsseeref. [6] . For a long time the type I and type II models seemed to b competingand probably even mutually incompatible.The exponential expansionof the universewas thought of as a rather exotic phenomenon B.V. 0370-2693/85/S Publishers 03.30@ Elsevier Science (North-Holland Division) Physics Publishing

andit seemed unlikelythat both abovementioned couldwork simultaneously. Moreover, mechanisrns initial conditions seemed to be necessary for different y'] quan. of the scenarios the realization [1,3 . Since tum gravityis far from bein8completely elaborated it wasnot possible to solve the primordial andsince problem[7] in the lust models of the q?e monopole prefened originally to studythe I [1], mostscientists onewasthe type Il models[3,4] , andthe favourable univers scenario newinflationary [4] . MeanwNe, of the models of both wDes has since1982the status . significantlychanged wereessentially 2. FLst of all, the type II models modfied. It wasshownthat in orderto obtainsmall perturbations density afterinllation6plp - t0-" , whicharenecessary for galaxyformation, oneshould weaklyinteracting consider a theoryof an extremely field 9 [8-l l] . Such.a field usuallydoes not undergo phase transitions in the early anyhigh-temperature for a considerable universe since the timenecessary 361

\tolume 1578,nlmber 5,6

PHYSICSLETTERS

25 July 1985

modification of its initial yalue due to high-temperature effectsusually exceeds the ageof the universe[6. I 2] . Therefore it is extremely difficult (if posible at aI) to obtain a completely satisfactoryrealization of the new inflationary universescenario [4] , which is basedon the theory of high-temperature phasetransi1ions.Despitmany efforts, no consistentrealization of the new inJlationaryuniversescenariohasbeen susgested so far. Fortunately, this difficulty disappears in the chaotic inJlation scenario[5] , accordingto which the universewasinitially filled with somechaotic initial distribution of a scalar field 9. It turnsout that. in a wide of theories, the domainsof t}le universewith lgl -class > C M p l C = O ( I ) , M p = G r l 2 i s t h ep l a n c k m a s s le x pandedmore than e?0 times during the slow rolling of the field to the rninimum of its potential enery V(e). The specificproperty of the chaotic inflation scenario is that the expansionof the universeat the inflation_ ary stagers not preciselyexponential.The Flubbleparaneter H = d la [a(t is rhe sca.]e faclor of the uni_ verse] in this scenarioslowly changes during inflation, However,the variation ofll at the stageof inflation is srnallcomparedto the rate of the expansionitselfi.e. the condition

It is especially importantthat in the infinite (e.g.open)universe, thereshouldbe infnitely manv -C -l domains of a size/ > 2fl lo jusr l,pl) [5] wtrere before the beginning of the inllationarystage. Thesize of these domains at the presnt momentis larger than the sizeof the observable part of the universe. In this scenario inflationof the universe, from an exoticDhe. nomenon whichcouldoccur in a number of rathei specific theories only, becomes a naturalconsquenc of chaoticinitial mnditionsin the expanding universe, 3. Simultaneoudy, a similarmetamorphosis has takenplacewith the type I models. The frst model of this type Il] wasconstructed on the basis of theconformalanomaly of the tota.l vacuum energy-momentum tensor of massless (or light enough) quanturn fields, ct ,) = -(t l2l8or2)Ik f ikhciktn + kz(RikRik- In2; + rronl (2) /i = where R R ard rheconstants &1, k2, k3 depend on thenumber andtypes of thequantum fields contribut.
ing to (2). However,it was realizedsoon that to pro vide a zufficient duration of the inflationary cage one should asume Ora1 /r2 > 0, &3 < 0, l&3| ) &2, whictr ,..] is not satisfiedin typical cases (in particular, there exid- supergravitytheorieswhere ,t2 = 0). Therefore another versionof the type ll model wasconsidered [2] in which the locat term ]y'3R2lg,nM2 ntthU ( i\1p was added to the lagrangian of the gravitational field. This conespondsto the effective renormalization

lH)<Hz

(l)

when eq. (l) holdswill be calledthe quasideSitrer ones. According to the chaotic inflation scenario,the more the initial value ofg is, the greateris the inflation of the universe. The initial polential energyden. sity of the scalar field I(9) in this scenario can be arbi tmrily large.This circumstance has two imDortant consequences. Firstly,in contrastto other versions of the inflationary universescenado,the flatnessproblem can be solvedhere evenif tlle universewas sDatiallyclosed i n i t l a l l yl 6 l . l l r t u s r e m i n d I t 3 l t h a ri i t h e u n i v e r s e is spatially closed and its energy densiry at the planck time rp -,rll[' is far from the criticalone,then it recollapses duringa typicaltime t - rp, i.e. before the beginningof inflation, unlessinflaiion startswith I(p) >,,Yfl1. Secondly,the possibiliry for inflation to begir with V(,02 Ml makespossiblea realization of tie inJlationary scnariofor the type II modelsbased on the idea of quantum creation of the universe [6,14] . (Analogousrealizationfor the type I modelswas suggested in ref. [15] .) 362

rs.sarisfied_. Then Rf = lO{n o -3d26f . Thestages

of k3 . After that, the effectiye constant f? need not be proponional to the numberof Light quantun fields.Moregenerally, a locatterm/(R) canbeaddcd to the gravitationallagrangian, where/ is an arbitsry functionrcstricted by rhe conditionlimn_olf(R)lRl = 0 oniy (see ref. [16] in thisconnection). It should be mentonedthat the procedure of addi. tion of quadratic termsto the Einstiin cravitationr.l actionin general is not completely it harmlessince mayleadto the appearance of ghostparticles with l negative energy[17] . Fortunately, the introductiotr theR2 orl(R) term (in contraslto theCikbniklf, . term)does not leadto the appearance of gjrosrs Ilfl If the signof the R2 term is chosen correctly thn obtain only one new scalarparticle with positive gy andpositivemass$quared. It is this particlethst

\Iolume15?B,number 5,6

PH YSICS LETTERS

2 5J u l y1 9 8 5

wascalled scalaron in ref. [1] . Therefore, the hypothesisthat the scalaron rest massM is significantlv s.nall_ s-(10-4 l0 s)irl e r t h a nt h e P l a n c k m a[s M doesnot lead to any difficulties or inconsistencies. If M 4l{p andR,7r1^Rikln<M4 then the k, andk1 i n e q .t 2 l c a n b e n e g l e c t e T tenns d. h u s .r h e r y p eI neednot be based models part of conon th non-local formalanomalyfwe calJrhe k, andl2 rermsin eq. (2) nonlocal because riey camor be obtainedby variation of any localactionwith respect to the metricl . This conclusion completely eliminates the criticismof thc type I modelsexpressed in ref. [18] . It wasshownin ref. [2] that the modifiedtype I possesa prolonged models quasi-de Sitter stage. lnllation in these modelscanalsobe chaotic,i,e. the quantny // can be differenrin dil'ferenr poirrs of space, H =//(l.r). Therefore. the inflationaryuniverse scenario basedon a type I model with the R2 term (we shall continue to call this lerm vacuumpolarization for brevity) can be realizedover a wide rangeof initial conditions for the scalar curvature R, lfM < lOl5 GeV, then the monopoleproblemcan be also solvedin this scenario. This is the casebecause the phasetransition with SU(5) breakingtakesplace before the end of the inflationary stage [even if the unbrokenSU(5)phase waspresent at the very beginning of the expansion] ard the temperarurenever growsto I0l5 GeV after the end ofinfladon (the s c a l a r ol n iferime r- ul1u3 [tl for.MK l0rs Gev is largeenough,and therefore reheatingoccurs comparativelylate). The samereasonis responsible for the absence of the monopoleproblemin the type ll models basedon superSrayity[12] . 4. Thus, it appears now that scenarios of both Wpes do not needsignificartly differenriniriajcondirions' and, therefore, they can be realizedsimultaneouslv. So, we considerthe generalized model where the expansionof the universeis determinedby the combined inJluenceof a scalarfield 9 with a potential fq) pa Ut the vacuumpolarizarion describedby the lll'yR'zlghM 2 term jn the total lagrangiar(the numerical cofficient is chosenhere in sucha way that the sca.laron massis equal to M). Then the evolution of the flat Friedmann-Ro bertson-Walke r cosmoloqical model and of the scalarfield follows from the equ-ations

H2=(Bnt3M!)tib2 + v(ol -u-2Paii+anzfi-fi21,

$+lae+v'(d=0, H=ila, M4Mr, (3)

where the dot denotesdifferentiation with respectto the usual time r and the prime, differentiation with re_ spectto I lthe term in eq. (2) hasbeenneglected] . If the scalarfield changes slowly at the quasi" ) , I Q l p ,< H . w e c a nn e g l e c r . l . , | 2 , d eS i r r e r . s r a( gl e HH nd Ht in eqs.(3), and they rake the followinp form 6H'HlM' + Hz = (8rl3M i)v(\,

3Hb+ v'(0= o.

(4)

Eq. (4) can be integratedin the Beneral casby using g asan independent variable: aH r = 4^ Jt dv Gi'

dH2 | de= tM2 I v' (dl lH2 - 8rv(e)BM\l,

=,!n@lMr)2 n2 *rfu,i &)

j' pr, r)av, (s) "i *n *,(-u,

If the condition l&/,pl( 11is not satisfiedat the quasi(l), then the scalar de Sitter staBe field rapidlyvar.rishesand can be neglectedat all, and we retum to the inflationary regimegeneratedby the vacuumpolarization only [2] :

=lnzgr-t), M<H<Mp, 11
a(t)=orexpl-]iU21r,- 421,
(6)
where tl = const. is the time when the quaside Sitter stageends andal is the sizeof the universeat the end of inflation. An important result can be obtained from eqs.(5):
. t

na J!)=Sso, on
J

=-.-

Mi v'tv)

R" dp, r(9, ) * (3lMt)WA- H2). _;j

(7)

volume l5?B, number5,6

PHYSICS LETTERS

2 5I u l y1 9 8 5

Here t0 is the moment of the begiffIing of the quasi and,ag,H6,9gare the corlesponding de Sitter stage initial valuesof a,l/, 9. Comparingeq. (7) with the exscepresions obtained in refs. [2,5] for the separate (either a narioswhere only one of the two species scalarfield or vacuumpolarization) was taken into accountbut not boti, we arriveat the following simple rule: the total expansionof the universedudng the in(i.e.,a1ica)in the combined scenado, flationarystage asa function of g andfl, is equal to the considered scenarproduct of the total expansionin the spaJate io. This rule is rather generaland can be appliedalso to the inflationary scenariowith an ajbitrary number oniy grayitationally. of scalarfields i-nteracting l,et us further consider,foltowing ref. [5] , the theory with the potential r/(9) = i\p4, I < 1 Then eq.(7) and the lastof eqs.(5) take the form

= (i.e.,the Zeldovich of state)+! . limitingequation nearsingularity poladzation So,the effectof vacuum of staleto that equation the effective of is the change of the general so(the behaviour radiation of masdess polarization vacuum in the Pure lution nearsingularity one earlier in ref. [19]). However, wasconsidered case analonot think that theR2 term is completely should between them gousto radiation. The crucialdifference the moreSeneral becomes clearwhenoneconsiders of the For examPle, in the case anisotropic singularity. radiation becomes Bianchitypemodels, homogeneous polarizabut the vacuum nearsingularity, unimportant vacuum the clasicalone-paramelic tion transforms onewhere Kasner solutionto the newtwo-parametric only one the tfueeexponents 4" (a = I , 2, 3) satisfy equality, \-\ , Is\ \ s-.

H2 = -in(MlM)2 exp(-Mz l2\e2) x f exD(,,1.12 d(,r?). /2ro?)

J ' r

!ni=\r-+o).

?n")

t <1,q"<i 2 L o o ,w i t h
zst

...

s\

(this conditionguarantees that R2 <Rp1^Riklm

=6lMl)(ef; - 'c2) * {tlu\{aa tn(alai

r, <Mp. The ratio m/r/ is the Denote.ry = Mp lliG, in our combinedscnario. important parar-neter L-etus considerthe questionabout t}le origin of the quasi-deSitter stage. In the casewhen the Friedmann (the total energydensity model hasa flat three-space is equal to the critical one), all solutions,except t}Iose havingzero measure, reachsingularity.The most gneral solution of eqs.(3) near singularity(r -, 0) has the following form: a(t) - \/t ,

"t)

(at

The evolution of our combined model during exon Cl, pansion(in the direction of growing t) depends C2, Ca. Without going into details,we presnthere the that the Seneral resultsof our investigation.It appears solution reachesthe quasi-deSitter stage(6) ot (8) or (12) in the courseof time if one of the following three conditions is satisfied:

(i) lc3l>M3l2 ,

< M3t2' $) tc3t Q> {uf;tD^n0'ulr,,' | 2, c! 3 1uf, ^1. (iij\tcat < M3 |ul ^no, u 1
lc2l> MP.
( l0) inicoverthe mostPartof possible These inequalities probable Thismeans that it is higlrly tial conditions. modelto go throughthe inllation' for our combined duringits evolution. ary stage begins either If m ) M, the quasideSitterstage with the solution (8) whereboth the scalarfield and field w8s vacuumpolarizationwork or, if the scalar zufficiently snall initially, with the solution(6).I{
*r This regime for the theoty m292/2 without vacuufl pohtization terms wrs recently investiSated by Bctinski' Grishchuk, Khalatnikov .nd Zeldovich (to be publtuhed,

v =c tl\/l +c2 -',r, cltzx* ru2luil r ...

crt/r + ... H = tlzt -bM213u2rlci+

= t-t 1+nn21u!1cltscal\/;+ ..., R = -6(H+ 2H2)

R 2< R * R i k - t 4

(e)

(indepenwhere C1, C2,C3 arethe ubitrary constants to notehere It is worthwhile dentinitial conditions). the generthat,in the absence ofvacuumpolarization, (r) near singularity would behave as a rll3 al solution p of state whichcofiesponds to the effective equation 364

I 1

5,6 \'olume 15?B,number

PHYSICSLETTERS

25July1985

l/ on t for dle quasiie Sitter staSes of the Hubble parameter Fig. l. The dependence

),M tfuoughout the whole stage At the end of it, only vacuum polarization works, so the evolution is by the solution (6) also.This situatron cordescribed that ro the curveA i-nfig. I It is interesting responds regime "slow rolli-ng" the eYeI in was if the scalarfield in this case, l,i/,pl<fl duringthe quasideSitterstag of inflation the end to up it remainsin this regime (6) evenwhen l9l <,44p.lts evolutionduringthe stage hasthe form

- r)l}-r , e2= el{r +(4t,p?1M2)blrl(rr

92= const,n4cl - t) > 1.
(l l) m (M, tlte model can pass ln the oppositecase the fust quaside Sitter stages: tluough two successive (5), (8) the second and by eqs with 11)M, described works field with//(M,where only the scalar [5] ,
,L s = 9ne-^t, H = \/ ;r>'(et/Mp) ' Zmt

by a hot radiationdomhated isotroPicstagewhere ?"f - a-4 or by someanisotropic by vacsupported stage the that shown can be .It polarization uum [16,20] for of envelopes play the role curvesA and B in flg. I fig. l) (depicted dots in by variousevolution curves to different initial conditions. corresponding our resultsto the It is not difficult to generalize Cl , C2,Ca andH by assuming case non-homogeneous This corre' to be functions of spatialcoordinates. s?ondsto t}le caseof chaotic inflation. In Particular, of the resultsobtained after the proper generalization that the non-homogeneous in ref. [21] , it appears has ttre following form in the qynquaside Sitter stage chronous systemof reference: d s 2= d 1 2- t a p d x d d r B , d , P = 1 , 2 , 3 ,

exv(zp(t,') d,) r,u =a "u@

x [1 + oG2e-/rd)] , ( t3)
wherea^o(r) are arbirrary functions of tfuee spatial that the coordiniies restrictedby the onJy condition -I at the spatialgradientsshould be smallerthan l/ for the begirning of the stage(the sameis assumed of ofH also) and the dependence spatial dependence point of (6)' (8) each (5), in gaen by eqs. t is ,{1on space. 5. Thus, we have shown that the two mechanisrns

( l2) >ffp. leol B in fig. 1.De' to thecurve (12)corres?onds Thestage

pendingon initial conditions(9), there may or may not be a rathet short period of a power-lawexpansion with a(r') - t2l3 dominated by masive dust-like scalarons [1] betweentlre two quasi'deSitter stages. The solution can also pas through the only one quasi (12) in this case. de Sitter stage Let us finally mention the possibilitiesthat the (8), (6) or ( l2) were preceeded quaside Sitter regimes

36s

v o l u m el 5 ? 8 . n u m b ejr, 6

PHYSICS LETTERS q = J dtla(t)l we obrain; 6 o l o = ( 2 n \ 3 1 2[ 6 j p , i k r r s ^ 1 ^ 1

J-.-e \wPtPrk.

25 July1985

of inflation,which earlierwereconsidered separately, canhappily coexist with eachother and their ioint accounr results in the substanlial enJa_rgemenr oi the number of domainsin an inirially chaotic universe wfuch passthrough the inflationary stage. The initia.l state of such a universeeitier could be dominated by hot matter or it could be vacuumlike.W, candescribe the ensemble of initial conditions for the curvafure R, rhe-scalar field I and rhe temperalure f of matterin orJerentdomarns by rhe inilial probabiliryamplirude

=t^4*l pttzrz l((6plpii>lu2 , =(MIM) A(k) J5 h@lk r), M<m,

fi4) where af2rk 1= ( l _ 105 ) $n at thepresent time. These results atein an agreement with the estimates made in-refs. [6,27]andwith theresults obtained in ref. [28], but somewhat differfromtheresults obo...d (14)correspond l_:"1 [29]. Theperrurbatjons to thefollowing large-scale arisotropy of themiciowave background radiarion ; [2,30] A T . \ \ -f (a.,pl= IJ (AT/T)tny rn@. p). = J 'I--a; ^ l n " t^,^ '(k/kt). M) n,

0(R, e,r).

Differentclassical initial condltrons in an isotropic untverse correspond eltherto somepure stateof ry'in the case of a quantumcreationof the universe [14,15, 221 andin the case in which the tsotropic stateaD_ pears after the decayof a more complexanisotropic injtial state [20] , or to a mixed stareln the case of the ordinaryhot beginning. Another importantconclusion is that the inflationary srage produced by one of tle mechanisms can typically create conditions necessary for the other mecha_ (@r/T)i,,,= A2I tOon nrsmto comelnto play. In particular jf m t(t+ t), ) M and , ( ls) only scalar-frelds work injrially(ut rhatU ZnTr[1 = = 2nlaq at the presenr A f,= 3 x I F . t p o l > M p a r l h e b e g i n n i no rime. !(khor),k6o, gf r h eq r i a s i _ d Se r " r r e r Yh:t.: tollowstrotn direcl observations (usingthe conelasrage). lhe vacuunt polarization termsajways become tion function l2l) thatA ( l.5X t0-3. This,together -dynamically rhe regirne (6) [and rates place with eq ( t4), rrnpties ::8"Ii:.lt that eirher MlMp < l0-s ;i lrnalyl because,9 >o r / . I n t h eo p p o s i t e ( , 0 / , casem s l0-rr. rl the scaiar fiejJ waslessthan/1p inirjally,ir can bewould like to note also that in our scenano comegreater not _We than/ly'p due to the growth of its lons_ only adiabatic but alsothe isothermal density w a v ev a c u u nfll u c r u a l i o n s perturduring t h eq u - a s i [q.23.24] b,ations.are enerated,whjch may becomevery impor(6) produced oe )llter stag by rhe vacuum polariza_ tant at the late stages of the universe evoturion1:ij . tron providedH0 ) y/MM p in iai y . and rhen the sec_ ond.quaside Sitterstage (12) wi takeplace (a similar Rekrences mechanism wasusedin ref. [25] in order to solvethe proorem ol symmetrybreakjrgLnsupersymmetrjc phys. J1I A.A.Starobinsky, Lett.9lB (l980)99;in: granduntliedtheories). gavity, eds, euantum M.A.Markov andp. West (Plenum, Newyork, 1983) p. 103. 6. [t us finally say a few words about density pismaAsron.Zh,9 (l983) [2] A-A.Starobjnsky, per_ 5?9 turbations A5116n. 1",,. which aregenerated ISov. t0 (l984) 1351, in our modeldurin! phys. A.H. curh, I3l Rev.D23098i 341. the quasi-de Sitter stage.By using the expressions obphys. [4] A.D.Linde, Lett.l08B (1982) 389; tamedin_refs. [2,8_11] (seealsoref. [26]), it can be A. Atbrecht phys.Rev. andp,l. Steinhardt, Lett.48 si.r.ow1 (tne derailswill be given elsewhere) (1982\ 1220. that the amplitudeof adiabatic phys. perturbations [5] A.D.Linde, Lett.1298(l983) 177. generated in the comprog. phys.4? brnedscenariois determinedby the mechanism Rep. J6l A.D.Lindet fl984)925;ptenaly that talkat.lheXXII lntern.Conf.on HiSh oomrnatesat the end of the quasi_de energy physics Sitter slageand is ro bepublished in rie proc. given by the smallestof the amplitudes .-, ::efz_,-ly64l, which would zetoovtch l/l l_a-.ir. andM.yu.Xholopov, phys. Lett.?98 be gnerated (L978) in each of the two scenarios 239; sparatelv. phys. At the presentdusUike stage I.P.Preskill, Rv.Lett. 43 (19?9)1365. a(r) - ,Z/s - ,z lwnire 366

Volume15?8,numbe,5,6

PHYSICSLETTERS

25 July 1985 e and B.L. Hu, phys.Rev,D20

Letr. I t 78 (l 982) l ?5. .l [ 1 0 ] A . H .G u t ha n dS . . Y . P i ,p h y s . Rev. L e l r . 4 9( t 9 8 2 , M.S.Turner, phys. Rev, ]:^lardeen, D 2 80 9 8 3 ) 6 7 9 . . e[. l32B(t , h y sL f l 2 | A . D .L i n d e p 9 8 3 )J I ? . A.D. Linde, The very ll3l eariyuniuer,e. eds.U. Gibbon!, S. Halvkingand S. Siktos(CambridSe U.p., Cambridte, 1 9 8 3p ) .23s. It4l A.D. Linde,Zh. Eksp.Teor. Fiz. 87 (l984) 369;Letr. Nuovo Cimento 39(1984) 401 Z e t d o v r cp hi.5 ' m a J 1 5 lY a . B . A s t r o nZ . h .? 0 9 8 1 ) 5 ? 9 ; L.P.Grishchuk and ya.B. Zeldovich. n: euanrum sEuctureof space_time, eds.M. Duffand C. Islum Iltl p.I. Steinlurdt and

pr,rs. Lerr. llsBned2)29s. Ill fY l,*kr"e, yJ A-A. phys. Srarobinsky,

V.G, Gr''"adyan, A.A. Kochryan and S.C.Matinyan, preprinrEFt-?00(I5)-8a (t98a;. "^*"jl..phyr ,r^, ,t J:sr. '"'

.rl l li

i*;.!,ilr'iil;sahni

and AA sta'obin'tv' zr' ix"

')

tzzl r.r:.narueand S.W.HawkinS. phys.Re". O:a gg:) tI

pismaZh.Eksp. J2ll AA. Starobinsky, Teor. Fn.3?0 983) r-!Tp t*".tnr', Let. 3?19b3) 661. ,.., i5-

,,., f3,roj,o* .r,a trot r.rs.lurovich and A.A. Starobirsky. Eksp.Teor.Fi7. . Er? so (! 1 6 8 3[ S o v p . h y sJ lzlt e+l; . ,. , ! . 7 : 1 ? 7 9 t , h y sR e v . l l T J K . t s t e l i eP Dt6 ft977,95J. R . T a b b r i a n d p o l t o c k . [18] phys. [t.D. L e l r .l 2 5 B ( t 9 8 3 )
i19l T.V. Ruzmaikina and A.A. Ruzmaftin, Eksp.Tmr. Fz. s 7 ( r 9 6 9 )6 8 0 ;

,Cambridse. 1982) p.409.

t"lyo univ. preprinr urAp.e (l e84). ISll l;I:l"lf gmbridse univ. preprint ( 1984.). ljll i 1; T"**T: lrvJ t(.xrtut and R. Brandenberger, phys.
[30J P.l-8.feebles, Asuophys, J. 263 ( I 9E2)Ll;

l r68 fl9g21 33.5. , hys. l r r J A . D . L r n d ep L e r r .l 3 t B { 1 9 8 3 . , 3 3 0 . [26 J V.F..M^ukhanov and G.V. CtLiUisov, Sov.phys.JETPLerr. 33 (.1981 5)3 2 ;5 6 ( 1 9 8 2 ) 2s8

phys. .:.0 r.H.Fo,d. j3jl i ^",1.1: Rev. D26 e82) r 231. Llnoe,fhys. Letr. t.rl A.r-,.

,l
i
I

Letr. t4tB 09g4)

{
i

[ 3 1 ] A . D . L i r d e ,p i s ' m a Z h . E k s pI.e o r . F n . 4 0 ( ' g B 4 ) 4 9 6 .

and Ya B zerdovich' Ji.iTli"-T '|ti'??,,gevich

36'7