On adiabatic perturbations in the ekpyrotic scenario A. Linde,1 V. Mukhanov,2,3 and A. Vikman3 1Department of Physics, Stanford University, Stanford, CA 94305, USA 2Arnold-Sommerfeld-Center for Theoretical Physics, Department für Physik, Ludwig-Maximilians-Universität München, Theresienstr. 37, D-80333, Munich, Germany and 3CCPP, New York University, Meyer Hall of Physics, 4 Washington Place, New York, NY 10003,USA (Dated: January 6, 2010) 0 In a recent paper, Khoury and Steinhardt [1] proposed a way to generate adiabatic cosmological 1 perturbationswithanearlyflatspectruminacontractingUniverse. Toproducetheseperturbations 0 theyusedaregimeinwhichtheequationofstateexponentiallyrapidlychangedduringashorttime 2 interval. Leaving aside the singularity problem and the difficult question about the possibility to n transmittheseperturbationsfromacontractingUniversetotheexpandingphase,wewillshowthat a themethodsusedin[1]areinapplicableforthedescriptionofthecosmological evolutionandofthe J process of generation of perturbationsin this scenario. 6 ] I. INTRODUCTION wasahopethatthecosmologicalsingularityproblemwill h be solvedin the context of string theory, but despite the t - attempts of the best experts in string theory, this prob- p Inflationarytheoryprovidesasimplesolutiontomany lem remains unsolved. A recent attempt to solve this e cosmological problems [2–5]. It also provides a simple problem was made in the so-called ‘new ekpyrotic sce- h [ mechanism for the generationof adiabatic perturbations nario’[15–18]. This scenarioinvolvedmany exoticingre- of the metric with flat spectrum, which are responsible dients,suchasa violationofthe nullenergyconditionin 2 for large-scale structure formation and for the observed a model which combined the ekpyrotic scenario [8] with v CMB anisotropy [6, 7]. Many predictions of inflationary the ghost condensate theory [19]. However, this model 4 4 cosmologyare already confirmed by observations. There contained ghosts [20]. The negative sign of the higher 9 were many attempts to propose an alternative, equally derivative term in this scenario makes it hard to con- 0 compelling cosmological theory, but so far they have all structaghost-freeUVcompletionofthistheory[20]; see . been unsuccessful. also [21] for a discussion of another attempt to improve 2 1 the new ekpyrotic scenario. Even if one eventually suc- Nothing illustrates this statement better than the his- 9 ceeds in inventing a UV complete generalization of this tory of the development of the ekpyrotic/cyclic scenario 0 theory,suchageneralizationisexpectedto obeythe null [8]. This scenario was supposed to solve all cosmological : energy condition, in which case the new ekpyrotic sce- v problemswithoutusingastageofinflation. However,the i nario will not work anyway. X original version of this scenario did not work. The large mass and entropy of the universe remained unexplained, Leaving all of these unsolved problems aside, the au- r a insteadofsolvingthe homogeneityproblemthis scenario thorsoftheekpyroticscenariocontinueinvestigatinggen- onlymadeitworse,andinsteadofthebigbangexpected eration of cosmologicalperturbations in various versions in [8], there was a big crunch [9, 10]. of this scenario. In inflationary theory, these perturba- tionsaregeneratedafterthesingularity,butintheekpy- As a result, this scenario was replaced by the cyclic rotic scenario they are supposed to be generated in a scenario, which postulated the existence of an infinite collapsing universe before the singularity and preserved number of periods of expansion and contraction of the onthe waythroughthe singularity. Since the singularity universe[11]. However,whenthisscenariowasanalyzed, problemis not solved,one cannot really say much about taking into account the effect of particle production in it, unless one makes additionalassumptions. After a few the early universe, a very different cosmological regime years of debate, most of the experts agreed, following was found [12, 13]. In the latest version of this scenario, [24], that adiabatic perturbations with a flat spectrum the homogeneity problem is supposed to be solved by are not produced in the simplest one-field versions of the existence of an infinitely many stages of low-scale this theory, contrary to the original claims by the au- inflation, each of which should last at least 56 e-folds thors of the ekpyrotic/cyclic scenario [8, 11, 25]. There- [14]. fore more complicated versions of the ekpyrotic scenario The most difficult problem facing this scenario is the were proposed[26], in which the adiabatic perturbations problemof the cosmologicalsingularity. Originallythere are produced by a mechanism similar to the inflationary 2 curvaton mechanism [27]. The simplest versions of the equation of state w = p/ρ change tremendously during ekpyrotic scenario of this type were ruled out in [28]. the same time interval. Indeed, That is why it is important to examine the recent pa- H˙ = 1φ˙2 2 , (5) per by Khoury and Steinhardt [1] who have proposed a −2 ≈−c2t2 new mechanism of generation of adiabatic cosmological hence H˙ H2/c2 and H˙ c2H2. For the equa- perturbations with a nearly flat spectrum at the stage begin ≃ 0 end ≃ 0 tion of state we have prior to the stage of the ekpyrotic collapse. This is the main goal of our paper. In section II, we will examine 2 H˙ 1 1 t 2 t 2 1+w= begin c2 end . the unperturbed background cosmological solution de- −3H2 ≈ H2c2t2 ≈ c2 t ≈ t 0 (cid:18) (cid:19) (cid:18) (cid:19) scribing the model of Ref. [1]. As we will see, this solu- (6) tion requires investigation of the cosmological evolution At the beginning (1+w) c−2 1 and at the end at super-planckian values of curvature, when the meth- (1+w) c2 1. begin ≈ ≪ odsusedin[1]arenotvalid. Asimilarconclusionwillbe end ≈ ≫ reachedinsectionIII wherewewillshowthatthetheory According to [1], the stage described above during ofcosmologicalperturbationsusedin[1]isalsonotvalid. which the spectrum of perturbations is generated is fol- lowed by a subsequent scaling ekpyrotic stage, when the Hubble scale evolves as II. BACKGROUND MODEL tend H H . (7) 0 ≃ t (cid:18) (cid:19) The authors of [1] derive the following necessary condi- The modelrelies on investigationofa contractinguni- tion whichmustbesatisfiedifwewanttheperturbations verse in a theory with the scalar field potential generatedatthe pre-ekpyroticstagetobe ontheobserv- V =V 1 e−cφ , (1) able scales today (see equation (16) in [1]): 0 − where c 1. To simplify a(cid:0)compari(cid:1)sonof the results, we V 10−30 H , (8) 0 end ≫ ≤ | | willusethesamenotationsandunits(8πG=1)asin[1]. where p p Wewillstartwiththehomogeneousbackgroundsolution. Ignoring the evolution of the universe, the equation of tend H H , (9) motion for φ is end ≃ 0 t (cid:18) ekp(cid:19) φ¨+cV0e−cφ =0 . (2) is the Hubble constant not at t = tend, but at the end of the ekpyrotic stage, at t = t . Taking into account ekp It describes evolution of the field φ when t slowly in- that H √V we obtain from (8) that this necessary 0 0 creases from large negative t. Among other solutions, condition≃c−an be rewritten as this equation has the solution used in [1]: 1 t 10−60H−1 t 10−60 . (10) | ekp|≤ 0 | end|≃ c2H2 2 V 0 0 φ= ln ct . (3) c −r 2 ! HMe−re1 tim10e−i4s3es.xpOrenstshedeoitnheurnhitasndo,faPcclaonrdckingtimtoe[1t]P(se∼e P ∼ Thissolutiondescribesaslowlycontractinguniversewith alsobelow),theamplitudeofthegeneratedperturbations a nearly unchanging Hubble parameter H H < 0. is O(cH ). Since this amplitude should be O(10−5), the 0 0 ≈ One can check that during the pre-ekpyrotic stage, at ekpyrotic stage should end at t <t<t , where begin end t <10−50t . (11) ekp P | | 1 1 t V begin 0 In other words, we are dealing here with frequencies t , t = , H = , begin ≃ H0 end ≃ H0c2 c2 0 −r 3 whichare50ordersofmagnitudegreaterthanthePlanck (4) mass! This completely invalidates the use of the classi- this solutiondoes notreceivesignificantcorrectionseven cal equations of motion in the investigation of the back- if one takes into account the contraction of the universe ground solution in [1]. Moreover, any mechanism ad- and adds the term 3Hφ˙ to equation (2). dressingthesingularityproblem(ifthismechanismexists at all) should operate on this time scale. Duringthetimeintervalfromt tot (4)thescale begin end factor does not decrease significantly and therefore one This fact was not noticed in [1] because the authors can set it to be a 1. Note that the time variable t is concentrated on the energy density at the end of the ≃ negativeanditsmagnitudechangesfromalargenegative ekpyrotic stage, valueatthebeginningtoasmallnegativevalueattheend of this contracting stage. At the same time, the deriva- ρ H2 4 10100 , (12) tive of the Hubble constant H˙ and, correspondingly, the ≃ ≃ c4t2 ≃ c4 ekp 3 in Planck units. They studied two different regimes, where Φ is the gravitational potential induced by the c = 1028 and c = 1040. In both cases ρ 1, so perturbations of the scalar field δφ. Linear perturbation ≪ one could think that we are safely in the sub-Planckian theory is reliable only if Φ, as well as the energy density regime. However, at the ekpyrotic stage with c2 1 perturbations δT0/T0, are simultaneously much smaller one has p c2ρ ρ, so from the Einstein equa≫tion than unity. Othe0rwis0e the higher-order terms in pertur- H˙ = (ρ+∼p)/2 it≫follows that bation theory dominate over linear terms, and the per- − turbative expansion fails. 4 10100 p 2H˙ c2H2 , (13) ≈− ∼ ≃ c2t2 ≃ c2 Before going any further, let us illustrate the state- ekp mentsmadeaboveusinginflationaryperturbationsasan i.e. pressure p, as well as H˙ and the curvature R, are example. During inflation, the large-scale energy den- morethan20ordersofmagnitudegreaterthatthePlanck sity perturbations and perturbations of metric are small densityforbothofthevaluesoftheparametersdiscussed δT0/T0 2Φ<10−5. 0 0 ∼− in [1]. At the transition to the post-inflationary stage, the Thesituationwithhigher-ordercurvatureinvariantsis valueofΦgrowstoabout10−5andstabilizesatthislevel. even worse. For instance, MeanwhiletheenergydensityperturbationsδT0/T0con- 0 0 tinue to grow. Once these perturbations become large, 1 10150 R R;α H¨ . δT0/T0 O(1),galaxiesbegintoformandseparateform | ;α |∼ ∼ c2t3 ∼ c2 0 0 ∼ ekp thecosmologicalexpansion. Thesubsequentevolutionof q the large-scale structure of the universe, as well as the This means that they exceed the Planckian values by 94 orders of magnitude for c = 1028, and by 70 orders of related evolution of Φ, cannot be described by perturba- magnitude for c = 1040. If one considers even higher tion theory. However, this happens long after the end of inflation,whichiswhythestandardtheoryofgeneration invariants, the situation becomes more and more trou- of inflationary perturbations is reliable. blesome. This is a direct consequence of the abnormal smallnessofthe time atthe endofthe stageofthe ekpy- Ref. [1]didnotcontainanyinvestigationofthepertur- rosis (11), which happens in this scenario for all values bationsδT0/T0,whichisnecessarytoexaminevalidityof 0 0 of its parameters. theirresults. Aswewillsee,intheirmodel,theperturba- tionsδT0/T0arelargealreadyatthepre-ekpyroticstage, One caneasilycheckthat the time tend, when the pre- 0 0 and therefore one cannot apply perturbation theory for ekpyroticstageendsandthestageoftheekpyrosisbegins theinvestigationofthegenerationofperturbationsinthe inthisscenario,isalsomanyordersofmagnitudesmaller metric and in the energy density in this scenario. than the Planck time, for the values of the parameters used in [1]. Moreover,one can show that this conclusion To quantize scalar perturbations one introduces the is valid for all possible values of parameters compatible canonical variable [7] with the requirement cH 10−5 unless there is a long 0 ∼ stage of inflation after the ekpyrotic big crunch. v =a δφ+ 3(1+w)Φ , (15) Note that we are not talking here about hypotheti- (cid:16) p (cid:17) calprocesseswhicharesupposedtoleadtoabounceand which for every Fourier mode k satisfies the equation preventthecosmologicalsingularityinthisscenario. The pre-ekpyroticstage,aswellastheekpyroticstage,arede- z′′ scribed in [1] by standard methods of classical field the- vk′′+ k2− z vk =0 , (16) ory and general theory of relativity, which are not valid (cid:18) (cid:19) at t < t and at super-planckian values of curvature. P where prime means the derivative with respect to con- | | Our resultsshow thereforethatthe suggestedscenariois formal time η, and z = a 3(1+w). In the case under completely unreliable even as a background model. In consideration the next section we will show that even if one ignores all p of the problems discussed above, the calculation of the z′′ 2 , (17) spectrumofmetricperturbationsin[1]isalsounreliable. z ≃ t2 during the time interval t < t < t . During the begin end pre-ekpyrotic stage a 1, but the second derivative of a III. PERTURBATIONS ≈ can be large and hence before identifying conformal and physicaltimewehavetoperformalldifferentiationinour equations,andonly after thatit is safe to sett η. For The perturbed homogeneous flat universe in the ≈ the short-wavelengthperturbations with k t 1 conformal-Newtonian coordinate system is described by | |≫ the metric 1 1 v eikη eikt , (18) ds2 =a2(η) (1+2Φ)dη2 (1 2Φ)δikdxidxk , (14) k ≃ √k ≃ √k − − (cid:2) (cid:3) 4 wheretheamplitudeisfixedbythe requirementthatthe to express δφ in terms of Φ in (15), we obtain [7] perturbations are the minimal vacuum fluctuations. In the long-wave limit for k t 1 the solution of (16) is v 2 Φ˙ +HΦ | |≪ ζ = = +Φ, (28) z 3H(1+w) dη 1 t k v C z+C z , (19) k ≃ 1 2 z2 ≃ √k t where the dot denotes the derivative with respect to the Z physical time t = adη. By substituting (22) and using wherewehavefixedtheconstantofintegrationrequiring the backgroundequations of motion, according to which the continuity of the solution at the moment of “effec- R tive horizon” crossing, tk 1/k, and have neglected the H˙ 3 ≃ = (1+w), (29) decaying mode. It follows from here that for the long- H2 −2 wavelength perturbations the “conserved” ζ is equal we find v c t cH k k 0 ζ . (20) k ≡ z ≃ √k tb ≃ k3/2 ζ = 2 1 w˙ + u˙ + 1 Φ. (30) 3 2H(1+w)2 H(1+w)u 1+w! Thereforeat the end of the pre-ekpyroticstage the spec- trum It follows from here that ζ k3/2 cH (21) 2 t k ≃ 0 ζ Φ , (31) ≃−3t end isscale-invariantintherangeofwavelengths t <λ < end t . For cH 10−5 the amplitude of pe|rtur|bations for the perturbations with k t 1 during the pre- begin 0 | | ≃ | | ≪ is of the right order of magnitude. This result was the ekpyroticstage. Forderivingthisresultweused(25)and reason for the claim in [1] that the pre-ekpyrotic stage (26). Thus at the end of the pre-ekpyrotic stage ζ Φ ≃ canproduce the perturbations similarto those produced and hence the spectrum of the gravitational potential is by inflation. In this case the flat spectrum appears be- the same as that for ζ. cause of the very fast change of the equation of state in The next thing to do is to determine the amplitude a contracting universe. of the energy density perturbations and verify that they Letus find under whichconditionsthe resultobtained alsoremain small. Using the 0 0 Einstein equation, we − is valid at the end of the pre-ekpyrotic stage. With this find purpose we will calculate the gravitationalpotential and δT0 2k2 2 energy density perturbations. It is convenient to intro- 0 = Φ Φ˙ +HΦ = duce the variable u relatedto the gravitationalpotential (cid:18) T00 (cid:19)k −3H2a2 k− H (cid:16) k k(cid:17) as 2k2 w˙ 2u˙ +3(1+w) 2 Φ . (32) Φ=H(1+w)1/2u , (22) (cid:16)−3H2a2 − − H(1+w) − Hu(cid:17) k During the pre-ekpyrotic stage, this relation for pertur- and satisfying the equation bations with k >H reduces to 0 ′′ u′k′+(cid:18)k2− (11//zz) (cid:19)uk =0 . (23) (cid:18)δTT0000(cid:19)k ≃(cid:18)−32Hk22 + H2t(cid:19)Φk . (33) During the pre-ekpyrotic stage Thisequationimpliesthatattheendofthepre-ekpyrotic (1/z)′′ H0 stage, at t≃tend . (24) 1/z ≃− t δT0 0 2c2Φ (34) T0 ≃ Therefore, for perturbations with k2 t H , we have 0 0 | |≫| | for H <k cH and u e±ikt, (25) 0 ≪ 0 k ∝ δT0 k 2 while for k2 t H the solution is 0 c2Φ 2c2Φ , (35) | |≪| 0| T0 ≃ k cH ≫ k (cid:18) 0 (cid:19)k (cid:18) 0(cid:19) u C +C t . (26) k ≃ 1 2 for k cH0. ≫ Thus we see that for the observed value Φ 3 10−5 Using the 0 i Einstein equation, ∼ × − the perturbations of the energy density exceed unity un- 1 √3 less c . 102. Note that the large c approximation used Φ˙ +HΦ= φ˙δφ= H(1+w)1/2δφ , (27) in [1] requires c 1. Eq. (35) implies that even in the 2 2 ≫ 5 marginal case 1 c .102 the energy density perturba- H−1. In addition, according to (12), in this case one tionsexceedunit≪yatallscalessmallerthanO(10−2)H−1 wo0uldbeforcedtoconsidertheekpyroticstageendingat 0 forc.102.Forc&102 weexpectthatthelinearpertur- the density exceeding the Planck density by 100 orders bation theory breaks down at all interesting scales. Let of magnitude. usshowthatthisiswhatreallyhappensinthecaseunder consideration. From (27) we find that during the pre-ekpyrotic stage IV. CONCLUSIONS δφ cΦ , (36) ≃ forperturbationswithk t 1.Expansionoftheenergy As we have seen, the new versionof the ekpyrotic sce- | |≪ momentumtensoraroundthebackgroundcontainsterms nariorequirescalculationstobe performedatcurvatures of allordersin δφ. In particular,in the expansionof δT0 which are exponentially greater than the Planck curva- 0 we have ture. Moreover, the theory of generation of metric per- turbations used in this scenario is based on calculations 1 δT0 =V δφ+ V δφ2+... (37) whichareexponentially farawayfromthe domainofap- 0 ,φ 2 ,φφ plicability of this theory. These problems appear inde- pendently of the major problem of this scenario, which It is clear that linear perturbation theory is applicable is the problem of the cosmologicalsingularity. only if the quadratic term here is small compared to the linear term. Let us find when this may happen. Taking Each time when a new version of the ekpyrotic/cyclic into accountthat in the case under consideration V = ,φφ theory is proposed, the authors suggest that its predic- cV we find that ,φ tions be compared with predictions of inflationary cos- mology to check which of the theories is better with V δφ2 ,φφ =cδφ c2Φ . (38) respect to observations. In accordance with this tradi- V,φδφ ≃ tion, the authors of Ref. [1] say that their new theory “predictsnon-gaussianityandaspectrumofgravitational This ratio does not exceed unity only if 1 c . 102, waves that is observationally distinguishable from infla- ≪ inagreementwiththe conclusionwhichwejustobtained tion.” Themainproblemwiththissuggestionisthatone by a different method. Thus linear perturbation theory can test any theory only if it is internally consistent and is not applicable for c & 102, and it fails completely for only if it really makes definite predictions. c=1028 and for c=1040 studied in [1]. Note that this problem is much more severe than the potentially curable problem of anomalously large non- gaussianity discussed in [1]. The nongaussianity prob- lem appearsonly for extremelysmallwavelengthsk−1 > c−1H−1, where c can be as large as 1028 or 1040. Mean- Acknowledgments 0 while for c & 102 the problems discussed in our paper occur at all wavelengths. WearethankfultoIgnacySawickiforusefulcomments. Moreover, as we already mentioned, the spectrum is The work of A.L. was supported in part by NSF grant not satisfactory even in the marginal case 1 c . 102. PHY-0244728,bythe Alexander-von-HumboldtFounda- ≪ Eq. (35) implies, in particular, that in the limiting case tion, and by the FQXi grant RFP2-08-19. V.M. is sup- of c = O(1), when the large c approximation used in [1] ported by TRR 33 “The Dark Universe” and the Cluster breaks down, the spectrum of perturbations of density ofExcellence EXC 153“OriginandStructureofthe Uni- is blue with n 5, and its amplitude becomes greater verse”. A.V. is supported by the James Arthur Fellow- s ∼ thanunityonscalestwoordersofmagnitudesmallerthan ship. [1] J. Khoury and P. J. Steinhardt, “Adiabatic Ekpyrosis: lution To The Horizon And Flatness Problems,” Phys. Scale-Invariant Curvature Perturbations from a Single Rev. D 23, 347 (1981). ScalarFieldinaContractingUniverse,” arXiv:0910.2230 [4] A. D. Linde, “A New Inflationary Universe Scenario: A [hep-th]. Possible Solution Of The Horizon, Flatness, Homogene- [2] A.A.Starobinsky,“ANewTypeOfIsotropicCosmolog- ity,IsotropyAndPrimordialMonopoleProblems,” Phys. ical Models Without Singularity,” Phys. Lett. B 91, 99 Lett. B 108, 389 (1982). (1980). [5] A.D. Linde,“Chaotic Inflation,” Phys.Lett.B 129, 177 [3] A. H. Guth, “The Inflationary Universe: A Possible So- (1983). 6 [6] V.F. MukhanovandG. V.Chibisov, “QuantumFluctu- derivatives proposed in [23]. Ref. [23] suggested to use ation And ‘Nonsingular’ Universe,” JETP Lett. 33, 532 iterations, initially solving equations of the theory omit- (1981) [Pisma Zh.Eksp. Teor. Fiz. 33, 549 (1981)]. ting the higher derivative terms, and then substituting [7] V. F. Mukhanov, Physical Foundations of Cosmology, the solutions to the higher derivative terms. This pro- Cambridge University Press, 2005. posal works for a large class of theories where omission [8] J. Khoury, B. A. Ovrut, P. J. Steinhardt and N. Turok, of the higher-derivative terms does not lead to patholo- “Theekpyroticuniverse: Colliding branesand theorigin gies. However, it does not help in the case of the new of the hot big bang,” Phys. Rev. D 64, 123522 (2001) ekpyroticscenario because thisscenario, with thehigher [arXiv:hep-th/0103239]. derivativebeingomitted,exhibitsacatastrophicgradient [9] R. Kallosh, L. Kofman and A. D. Linde, “Py- instability [20]. rotechnic universe,” Phys. Rev. D 64, 123523 (2001) [22] P. Creminelli, G. D’Amico, J. Norena and F. Vernizzi, [arXiv:hep-th/0104073]. “The Effective Theory of Quintessence: the w<-1 Side [10] R. Kallosh, L. Kofman, A. D. Linde and A. A. Tseytlin, Unveiled,” JCAP 0902, 018 (2009) [arXiv:0811.0827 “BPS branes in cosmology,” Phys. Rev. D 64, 123524 [astro-ph]]. (2001) [arXiv:hep-th/0106241]. [23] S.Weinberg,“EffectiveFieldTheoryforInflation,” Phys. [11] P. J. Steinhardt and N. Turok, “Cosmic evolution in Rev. D 77, 123541 (2008) [arXiv:0804.4291 [hep-th]]. a cyclic universe,” Phys. Rev. D 65, 126003 (2002) [24] D. H. Lyth, “The primordial curvature perturbation [arXiv:hep-th/0111098]. in the ekpyrotic universe,” Phys. Lett. B 524, 1 [12] G. N. Felder, A. V. Frolov, L. Kofman and A. V. Linde, (2002) [arXiv:hep-ph/0106153]; D. H. Lyth, “The fail- “Cosmology with negative potentials,” Phys. Rev. D 66, ure of cosmological perturbation theory in the new 023507 (2002) [arXiv:hep-th/0202017]. ekpyrotic scenario,” Phys. Lett. B 526, 173 (2002) [13] A. Linde, “Inflationary theory versus ekpyrotic/cyclic [arXiv:hep-ph/0110007]. scenario,” in The future of theoretical physics and cos- [25] J. Khoury, B. A. Ovrut, P. J. Steinhardt and N. Turok, mology, (Cambridge Univ. Press, Cambridge 2002), p. “Density perturbations in the ekpyrotic scenario,” Phys. 801 [arXiv:hep-th/0205259]. Rev. D 66, 046005 (2002) [arXiv:hep-th/0109050]. [14] J.L.Lehners,P.J.SteinhardtandN.Turok,“TheReturn [26] J. L. Lehners, P. McFadden, N. Turok and P. J. Stein- of thePhoenix Universe,” arXiv:0910.0834 [hep-th]. hardt,“Generatingekpyroticcurvatureperturbationsbe- [15] P. Creminelli, M. A. Luty, A. Nicolis and L. Senatore, fore the big bang,” Phys. Rev. D 76, 103501 (2007) “Startingtheuniverse: Stableviolationofthenullenergy [arXiv:hep-th/0702153]; A. J. Tolley and D. H. Wes- condition and non-standard cosmologies,” JHEP 0612, ley, “Scale-invariance in expanding and contracting uni- 080 (2006) [arXiv:hep-th/0606090]. verses from two-field models,” JCAP 0705, 006 (2007) [16] E. I. Buchbinder, J. Khoury and B. A. Ovrut, “New [arXiv:hep-th/0703101]. Ekpyrotic Cosmology,” Phys. Rev. D 76, 123503 (2007) [27] A. D. Linde and V. Mukhanov, “Nongaussian isocur- [arXiv:hep-th/0702154]. vature perturbations from inflation,” Phys. Rev. D [17] P. Creminelli and L. Senatore, “A smooth bouncing cos- 56, 535 (1997) [arXiv:astro-ph/9610219]; K. Enqvist mologywithscaleinvariantspectrum,” JCAP0711,010 and M. S. Sloth, “Adiabatic CMB perturbations in (2007) [arXiv:hep-th/0702165]. pre big bang string cosmology,” Nucl. Phys. B 626, [18] E. I. Buchbinder, J. Khoury and B. A. Ovrut, “On the 395 (2002) [arXiv:hep-ph/0109214]; D. H. Lyth and Initial Conditions in New Ekpyrotic Cosmology,” JHEP D. Wands, “Generating the curvature perturbation 0711, 076 (2007) [arXiv:0706.3903 [hep-th]]. without an inflaton,” Phys. Lett. B 524, 5 (2002) [19] N. Arkani-Hamed, H. C. Cheng, M. A. Luty and [arXiv:hep-ph/0110002]; T. Moroi and T. Takahashi, S.Mukohyama,“Ghostcondensationandaconsistentin- “Effects of cosmological moduli fields on cosmic mi- frared modification of gravity,” JHEP 0405, 074 (2004) crowave background,” Phys. Lett. B 522, 215 (2001) [arXiv:hep-th/0312099]. [arXiv:hep-ph/0110096]. [20] R. Kallosh, J. U. Kang, A. D. Linde and V. Mukhanov, [28] K. Koyama, S. Mizuno, F. Vernizzi and D. Wands, “The New Ekpyrotic Ghost,” JCAP 0804, 018 (2008) “Non-Gaussianitiesfromekpyroticcollapsewithmultiple [arXiv:0712.2040 [hep-th]]. fields,” JCAP 0711, 024 (2007) [arXiv:0708.4321 [hep- [21] It was suggested in [22] that one might avoid the prob- th]]. lems of the new ekpyrotic scenario discussed in [20] by usinganapproachtotheeffectivefieldtheorywithhigher