MCMC analysis of WMAP3 and SDSS data points to broken symmetry inflaton potentials and provides a lower bound on the tensor to scalar ratio. C. Destri(a), H. J. de Vega(b,c), and N. G. Sanchez(c) ∗ † ‡ (a) Dipartimento di Fisica G. Occhialini, Universit`a Milano-Bicocca Piazza della Scienza 3, 20126 Milano and INFN, sezione di Milano, via Celoria 16, Milano Italia (b) LPTHE, Laboratoire Associ´e au CNRS UMR 7589, Universit´e Pierre et Marie Curie (Paris VI) et Denis Diderot (Paris VII), Tour 24, 5 `eme. ´etage, 4, Place Jussieu, 75252 Paris, Cedex 05, France. (c) Observatoire de Paris, LERMA, Laboratoire Associ´e au CNRS UMR 8112, 61, Avenue de l’Observatoire, 75014 Paris, France. (Dated: February 5, 2008) We perform a MCMC (Monte Carlo Markov Chains) analysis of the available CMB and LSS data (including the three years WMAP data) with single field slow-roll new inflation and chaotic 8 inflation models. We do this within our approach to inflation as an effective field theory in the 0 Ginsburg-Landau spirit with fourth degree trinomial potentials in the inflaton field φ. We derive 0 explicit formulae and studyin detail thespectral index ns of theadiabatic fluctuations, theratio r 2 oftensortoscalar fluctuationsandtherunningindexdns/dlnk. Weusetheseanalyticformulas as n hardconstraintsonnsandrintheMCMCanalysis. Ouranalysisdiffersinthiscrucialaspectfrom a previous MCMC studies in the literature involving the WMAP3 data. Our results are as follow: J (i) The data strongly indicate the breaking (whether spontaneous or explicit) of the φ φ → − 9 symmetry of the inflaton potentials both for new and for chaotic inflation. (ii) Trinomial new inflationnaturallysatisfiesthisrequirementandprovidesanexcellentfittothedata. (iii)Trinomial 3 chaotic inflation produces the best fit in a very narrow corner of the parameter space. (iv) The v chaotic symmetric trinomial potential is almost certainly ruled out (at 95%CL). In trinomial 7 chaotic inflation the MCMC runs go towards a potential in the boundary of the parameter space 1 and which ressembles a spontaneously symmetry broken potential of new inflation. (v) The above 4 results and further physical analysis here lead us to conclude that new inflation gives the best 3 description of thedata. (vi) Wefind a lower bound for r within trinomial new inflation potentials: 0 r>0.016(95%CL)andr>0.049(68%CL). (vii)Thepreferrednewinflationtrinomialpotentialis 7 adoublewell,evenfunctionofthefieldwithamoderatequarticcouplingyieldingasmostprobable 0 values: ns 0.958, r 0.055. This value for r is within reach of forthcoming CMB observations. / ≃ ≃ h p - o r Contents t s a : I. Introduction and Results 2 v Xi II. The Inflaton Potential and the 1/N Slow Roll Expansion 5 r a III. Trinomial Chaotic Inflation: Spectral index, amplitude ratio, running index and limiting cases 8 A. The small asymmetry regime: 1<h 0. 9 B. The flat potential limit h 1−+ ≤ 10 →− C. The singular limit z =1 and then h 1+ yields the Harrison-Zeldovichspectrum 11 →− D. The high asymmetry h< 1 regime. 13 − IV. Trinomial New Inflation: Spectral index, amplitude ratio, running index and limiting cases 14 A. The shallow limit (weak coupling) y 0 16 → B. The steep limit (strong coupling) y 17 →∞ C. The extremely asymmetric limit h 17 | |→∞ D. Regions of n and r coveredby New Inflation and by Chaotic Inflation. 18 s V. Monte Carlo Markov Chains and data analysis with the trinomial inflation models 18 Electronicaddress: [email protected] ∗ Electronicaddress: [email protected] † Electronicaddress: [email protected] ‡ 2 A. MCMC results for new inflation. 20 B. MCMC results for chaotic inflation. 22 C. Conclusions. 23 References 24 I. INTRODUCTION AND RESULTS Inflation was introduced to solve several outstanding problems of the standard Big Bang model [1] and has now become an important part of the standard cosmology. At the same time, it provides a natural mechanism for the generation of scalar density fluctuations that seed large scale structure, thus explaining the origin of the tempera- ture anisotropies in the cosmic microwave background (CMB), as well as that of tensor perturbations (primordial gravitationalwaves) [2, 3, 4]. A distinct aspect of inflationary perturbations is that these are generated by quantum fluctuations of the scalar field(s) that drive inflation. After their wavelength becomes larger than the Hubble radius, these fluctuations are amplified and grow, becoming classical and decoupling from causal microphysical processes. Upon re-entering the horizon, during the matter era,these classicalperturbations seed the inhomogeneities which generate structure upon gravitationalcollapse[2, 3]. A greatdiversityofinflationarymodels predictfairly genericfeatures: agaussian,nearly scale invariant spectrum of (mostly) adiabatic scalar and tensor primordial fluctuations, making the inflationary paradigmfairlyrobust. Thegaussian,adiabaticandnearlyscaleinvariantspectrumofprimordialfluctuationsprovide anexcellentfittothehighlyprecisewealthofdataprovidedbytheWilkinsonMicrowaveAnisotropyProbe(WMAP)[5, 6]. Perhapsthemoststrikingvalidationofinflationasamechanismforgeneratingsuperhorizon (‘acausal’)fluctuations is the anticorrelation peak in the temperature-polarization (TE) angular power spectrum at l 150 corresponding ∼ to superhorizon scales [5]. The confirmation of many of the robust predictions of inflation by current high precision observations places inflationary cosmology on solid grounds. Amongst the wide variety of inflationary scenarios, single field slow roll models provide an appealing, simple and fairly generic description of inflation. Its simplest implementation is based on a scalar field (the inflaton) whose homogeneousexpectationvaluedrivesthedynamicsofthescalefactor,plussmallquantumfluctuations. Theinflaton potential is fairly flat during inflation. This flatness not only leads to a slowly varying Hubble parameter, hence ensuringa sufficient number ofe-folds,but alsoprovidesanexplanationfor the gaussianityofthe fluctuations aswell as for the (almost) scale invariance of their power spectrum. A flat potential precludes large non-linearities in the dynamics of the fluctuations of the scalar field. The current WMAP data seem to validate the simpler one-field slow roll scenario [5, 6]. Furthermore, because the potential is flat the scalar field is almost massless, and modes cross the horizon with an amplitude proportional to the Hubble parameter. This fact combined with a slowly varying Hubble parameter yields an almost scale invariant primordialpowerspectrum. Theslow-rollapproximationhasbeenrecentlycastasa1/N expansion[7],whereN 50 ∼ isthenumberofefoldsbeforetheendofinflationwhenmodesofcosmologicalrelevancetodayfirstcrossedtheHubble radius. The observational progress permits to start to discriminate among different inflationary models, placing stringent constraints on them. The upper bound on the ratio r of tensor to scalar fluctuations obtained by WMAP [5, 6] necessarily implies the presence of a mass term in the single field inflaton potential and therefore rules out the masslessmonomialφ4 potential[6, 7]. Hence,as minimalsingle fieldmodel, oneshouldconsidera sufficiently general quartic polynomial, that is the trinomial potential. Besides its simplicity, the trinomial potential is a physically well motivated potential for inflation in the context of the Ginsburg-Landau approach to effective field theories (see for example ref.[10]). This potential is rich enough to describe the physics of inflation and accurately reproduce the WMAP data [5, 6]. The slow-rollexpansion plus the WMAP data constraints the inflaton potential to have the form [7] V(φ)=N M4 w(χ), (1.1) where φ is the inflaton field, χ is a dimensionless, slowly varying field φ χ , (1.2) ≡ √N M Pl 3 w(χ) (1) and M is the energy scale of inflation which is determined by the amplitude of the scalar adiabatic ∼ O fluctuations [5] to be M 0.00319M =0.77 1016 GeV. (1.3) Pl ∼ × Following the spirit of the Ginsburg-Landautheory of phasetransitions, the simplest choice is a quartic trinomial for the inflaton potential [7, 8, 9]: 1 h y y w(χ)=w χ2+ χ3+ χ4 , (1.4) 0 ± 2 3 2 32 r where the coefficients w , h and y are dimensionless and of order one and the signs correspond to large field 0 ± and small field inflation, respectively (namely, chaotic inflation and new inflation, respectively). h describes how asymmetric is the potential, y is the dimensionless quartic coupling. Inserting eq.(1.4) in eq.(1.1) yields, m2 mg λ V(φ)=V φ2+ φ3+ φ4 , (1.5) 0 ± 2 3 4 where the mass m2 and the couplings g and λ are given by the following see-saw-like relations, M2 y M 2 y M 4 m= , g =h , λ= , V =N M4 w . (1.6) 0 0 M 2N M 8N M Pl r (cid:18) Pl(cid:19) (cid:18) Pl(cid:19) Notice that y (1) h guarantee that g (10 6) and λ (10 12) without any fine tuning as stressed in ref. − − ∼O ∼ ∼O ∼O [7]. Thatis,the smallnessofthe couplingshere is notaconsequenceof fine tunning but followdirectly fromthe form of the inflaton potential eq.(1.1) and the amplitude of the scalar fluctuations that fixes M [7]. During inflation the inflaton fieldχ slowlyruns fromits initialvalue till its finalvalue χ atthe absolute minima end of the potential w(χ). We have χ =0 for chaotic inflation while χ turns to be a function of the coupling y and end end the asymmetry h for new inflation. Inflation ends after a finite number of efolds provided w(χ )=w (χ )=0 end ′ end Enforcingthisconditionintheinflationarypotentialeq.(1.4)determinestheconstantw . Wehavew =0forchaotic 0 0 inflation while w turns to be a function of the coupling y and the asymmetry h for new inflation. χ can vary in the 0 interval (0, ) for chaotic inflation while χ is in the interval (0,χ ) for new inflation. end ∞ We derive explicit formulae and study in detail the spectral index n of the adiabatic fluctuations, the ratio r of s tensor to scalar fluctuations and the running index dn /dlnk both for trinomial chaotic inflation and trinomial new s inflation. The smallcoupling limit y 0 of eqs.(1.4)-(1.5) correspondsto the quadratic monomialpotentialwhile the strong → coupling limit y yields in chaotic inflation the massless quartic monomial potential. The extreme asymmetric limit h yie→lds∞a massive model without quadratic term. In such a limit the product h M2 must be kept fixed | |→∞ | | since it is determined by the amplitude of the scalar fluctuations. InthispaperweperformMonteCarloMarkovChains(MCMC)analysisofthecommonlyavailableCMBandLarge Scale Structure (LSS) data. For CMB we considered the three years WMAP data, which provide the dominating contribution, and also small scale data (ACBAR, CBI2, BOOMERANG03). For LSS we considered SDSS (Sloan Digital Sky Survey) [11]. We used the CosmoMC program [12] within the effective field theory of inflation. We used a largecollectionof parallelchains with a totalnumber of samples close to five million. For chaoticand new inflation we imposed as a hard constraint that the spectral index n of the adiabatic fluctuations and the ratio r are given s by the analytic formulas at order 1/N we derived for the trinomial inflaton potential. Our analysis differs in this crucial aspect from previous MCMC studies involving the WMAP3 data set [11]. As natural within inflation, we also included the inflationary consistency relation n = r/8 on the tensor spectral index. This constraint is in any T − case practically negligible. The details of the MCMC analysis are explained in Section V. Our approachis different to the inflationary flow equations [14] where the inflaton potential changes (that is, the model changes) asthe flowgoeson. We workwithagivenpotentialwithin the Ginsburg-Landau(GL) spirit,that is the trinomial potential. We investigate the physics of the chosen potential in the parameter space driven by the data throughthe Monte CarloMarkovChains. In our work,n and r arecomputed analyticallyto order1/N for the s 4 trinomialpotentials [eqs.(3.6)-(3.8) and(4.6)-(4.8)]. Since N 50,higher ordercorrectionsin 1/N areirrelevantand ∼ can be safely neglected. As shown in ref. [7], the slow roll expansion is in fact a systematic expansion in 1/N. WeallowedsevencosmologicalparameterstovaryinourMCMCruns: thescalarspectralindexn ,thetensor-scalar s ratior,thebaryonicmatterfractionω ,thedarkmatterfractionω ,theopticaldepthτ,theratioofthe(approximate) b c soundhorizontotheangulardiameterdistanceθandtheprimordialsuperhorizonpowerinthecurvatureperturbation at 0.05 Mpc−1, As. We allowedthe same seven parametersto vary in the MCMC runs for chaotic and new inflation. In the case of new inflation, since the characteristic banana–shaped allowed region in the (n ,r) plane [fig. 2] is s quite narrow and non–trivial, it is convenient to use the two independent variables z y χ2 and h in trinomial ≡ 8 inflationarysetup as MC parameters. That is, we used the analytic expressionswe found at order1/N, to express n s and r in terms of z and h. Concerningpriors,wekeptthesame,standardones,oftheΛCDM+rmodelforthefirstfiveparameters(ω , ω , τ, θ b c and A ), while we considered all the possibilities for z and h. s In the case of chaotic inflation we kept n and r as MC parameters, imposing as hard priors that they lay in s the region described by chaotic inflation [fig. 2]. This is technically convenient, since this region covers the major part of the probability support of n and r in the ΛCDM+r and the parametrization eqs.(3.6)-(3.8) in terms of the s parameters z and h becomes quite singular in the limit h 1. This is indeed the limit which allows to cover the → − region of highest likelihood. The priors on the other parameters where the same of the ΛCDM+r model and of new inflation. We did not marginalize over the SZ amplitude, and we did not include non-linear effects in the evolution of the matter spectrum. The relative corrections are in any case not significant [5], especially in the present context. InallourMCMCrunswekeepfixedthenumberofefoldsN sincehorizonexittilltheendofinflation. Thereasonis thatthemainphysicsthatdeterminesthevalueofN isnotcontainedintheavailabledatabutinvolvesthereheating era. Therefore, it is not reliable to fit N solely to the CMB and LSS data. The precise value of N is certainly near N = 50 [14, 15]. We take here the value N = 50 as a reference baseline value for numerical analysis, but from the explicit expressions obtained in the slow roll 1/N expansion we see that both n 1 and r scale as 1/N. Therefore, s − varying N produces a scale trasformation in the (n 1,r) plane, thus displacing the black and red curves in fig. 2 r − towards up and left or towards down and right. This produces however small quantitative changes in our bounds for r aswellasinthe mostprobablevaluesforr andn . MCMCsimulationswithvariableN andimposingthe trinomial s new inflation potential yielded N 50 as the most probable value. ∼ We plot in fig. 2 r vs. n both for chaotic and new inflation for fixed values of the asymmetry h and the s coupling y varying along the curves. We see that the regions of the trinomial new inflation and chaotic inflation are complementary in the (n ,r) plane. The red curves are those of chaotic inflation, while black curves are for new s inflation. Theleftmostblacklinecorrespondstonewinflationwithasymmetricpotentialh=0. Therightmostblack line describes the case of new inflation with an extreme asymmetric potential h 1. This last line is the border of | |≫ the region on the right described by chaotic inflation. Although chaotic inflation covers a much wider area than new inflation, this wide area is only a small corner of the parameter space (field z y χ2, asymmetry h) as shown by fig. ≡ 8 9. For the trinomial new inflation model we find a lower bound on r: r >0.016 (95%CL) , r >0.049 (68%CL) (new inflation). (1.7) while for n we find: s n >0.945 (95%CL) (newinflation). s The most probable values are (see fig. 6), n 0.956 , r 0.055 (newinflation). (1.8) s ≃ ≃ For trinomial new inflation there exists the theoretical upper limits: n 0.9615..., r 0.16 [9]. Thus, the most s ≤ ≤ probable value of n for trinomial new inflation eq.(1.8) is very close to its theoretical limiting value, and that of r is s not too far from it (see also fig. 6). The probability distribution for the asymmetry parameter h is peaked at h=0 with h <4.92with95%CL new inflation. | | 5 Thatis,wefindthatthemostprobabletrinomialnewinflationpotentialissymmetricandhasamoderatenonlinearity with the quartic coupling y 2 for h 0. This is the following potential: ≃ ≃ 2 y 8 w(χ)= χ2 . (1.9) 32 − y (cid:18) (cid:19) The χ χ symmetry is here broken since the absolute minimum of the potential is at χ=0. →− 6 For trinomial chaotic inflation, the chaotic symmetric trinomial potential h = 0 is almost certainly ruled out since h< 0.7 at 95 % confidence level (see fig. 9). − We see the maximun probability for strong asymmetry h< 0.95 and significant nonlinearity 4.207...<y <+ − ∞ That is, in chaotic inflation all three terms in the trinomial potential w(χ) do contribute. We have not introduced the running of the spectral index dn /dlnk in our MCMC fits since the running must s be very small of the order (N 2) 0.001 in slow-roll and for generic potentials [7]. We find that adding the new − O ∼ parameterdn /dlnk to the MCMC analysismakenegligible changesonthe fit ofn andr. All this suggeststhat the s s present data are not yet precise enough to allow a determination of dn /dlnk. s In summary, the data favour the breaking of the χ χ symmetry both for new and for chaotic inflation. →− Trinomial chaotic inflation produces its best fit to the data in a very narrow corner of the parameter space. The potential in chaotic inflation has a single minimum. But this minimum for the best fit potential happens to be in the boundary of the parameter space, precisely where the asymmetry of the potential is so large that arises an extra minimum to the potential. That is, the data are begging the potential to be a double well (that is, a new inflation potential). The MCMC runs go towards a potential in the boundary of the parameter space and maximal symmetry breaking. This limiting potential exhibits an inflexion point and a significative nonlinearity. This strongly suggest that the true potential may be outside the parameter space of chaotic inflation. On the contrary, starting with a double well, that is new inflation, the data are perfectly well fitted with small or even zero asymmetry. Inchaoticinflation,the MCMC minimizationleadsto alargenonlinearity(fromthe higher ordercubic andcuartic terms), making chaotic inflation not stable under higher order terms in the potential. Chaotic inflation thus results contrary to the Ginsburg-Landau spirit in which higher order terms of the potential must be smaller and smaller giving irrelevant corrections. On the contrary, new inflation results perfectly well fitted by the data with very small or zero cubic non-linearity, andthusnewinflationisstableunderhigherordertermsofthepotential,whichisperfectlywellwithintheGinsburg- Landau sprit. Indeed, one can fit the WMAP+LSS data both with trinomial new inflation as well as with trinomial chaotic inflation. However, the trinomial potential that gives the best fit for chaotic inflation is theoretically undesirable for the reasons explained above. Therefore, trinomial new inflation is the best model that describes the data. This implies that we do have a lower bound on r as given by eq.(1.7). Moreover,the best fit is obtained with the double well inflation potential eq.(1.9) yielding the values in eq.(1.8) for n and r. s Thispaperisorganizedasfollows: insectionIIwediscusstheslow-rollapproachasthe1/N expansion. Insections III and IV we present the trinomial chaotic inflation and the new trinomial inflation, respectively, displaying the analytic formulas for the spectral index n of the adiabatic fluctuations, the ratio r of tensor to scalar fluctuations s and the running index dn /dlnk. In section V we explain our MCMC analysis, we present our MCMC results for s new and chaotic trinomial inflation and we state our conclusions. II. THE INFLATON POTENTIAL AND THE 1/N SLOW ROLL EXPANSION The description of cosmological inflation is based on an isotropic and homogeneous geometry, which assuming flat spatial sections is determined by the invariant distance ds2 =dt2 a2(t)d~x2 . (2.1) − The scale factor obeys the Friedman equation ρ(t) 1 da H2(t)= , where H(t) , (2.2) 3M2 ≡ a(t) dt Pl 6 where ρ(t) is the total energy density and M =1/√8πG=2.4 1018GeV. Pl × In single field inflation the energy density is dominated by a homogeneous scalar condensate, the inflaton, whose dynamics is described by an effective Lagrangian φ˙2 ( φ)2 =a3(t) ∇ V(φ) . (2.3) L " 2 − 2a2(t) − # TheinflatonpotentialV(φ)isaslowlyvaryingfunctionofφinordertopermitaslow-rollsolutionfortheinflatonfield φ(t). We showed in ref.[7] that combining the slow roll expansion with the WMAP data yields an inflaton potential of the form V(φ)=N M4 w(χ), (2.4) where χ is a dimensionless, slowly varying field φ χ= , (2.5) √N M Pl w(χ) (1) , N 50 is the number of efolds since the cosmologicallyrelevant modes exited the horizontill the end ∼O ∼ of inflation and M is the energy scale of inflation The dynamics of the rescaled field χ exhibits the slow time evolution in terms of the stretched dimensionless time variable, tM2 H τ = , = (1). (2.6) M √N H≡ √N m O Pl The rescaled variables χ and τ change slowly with time. A large change in the field amplitude φ results in a small change in the χ amplitude, a change in φ M results in a χ change 1/√N. The form of the potential, eq.(2.4) Pl ∼ ∼ and the rescaleddimensionless inflaton field eq.(2.5) and time variable τ make manifestthe slow-rollexpansionas a consistent systematic expansion in powers of 1/N [7]. The inflaton mass around the minimum is given by a see-saw formula M2 m= 2.45 1013GeV. M ∼ × Pl The Hubble parameter when the cosmologically relevant modes exit the horizon is given by H =√N m 1.0 1014GeV=4.1m, H∼ × where we used that 1. As a result, m M and H M . Pl H∼ ≪ ≪ AGinsburg-Landaurealizationofthe inflationarypotentialthatfits the amplitude ofthe CMBanisotropyremark- ably well, reveals that the Hubble parameter, the inflaton mass and non-linear couplings are see-saw-like, namely powers of the ratio M/M multiplied by further powers of 1/N. Therefore, the smallness of the couplings is not Pl a result of fine tuning but a natural consequence of the form of the potential and the validity of the effective field theory description and slow roll. The quantum expansion in loops is therefore a double expansion on (H/M )2 and Pl 1/N. Notice that graviton corrections are also at least of order (H/M )2 because the amplitude of tensor modes is Pl of order H/M . We showed that the form of the potential which fits the WMAP data and is consistent with slow Pl roll eqs.(2.4)-(2.5) implies the small values for the inflaton self-couplings [7]. The equations of motion in terms of the dimensionless rescaled field χ and the slow time variable τ take the form, 2 1 1 dχ 2(τ)= +w(χ) , H 3"2N (cid:18)dτ(cid:19) # 1 d2χ dχ +3 +w (χ)=0 . (2.7) ′ N dτ2 H dτ The slow-rollapproximationfollows by neglecting the 1 terms in eqs.(2.7). Both w(χ) and (τ) are of order N0 for N H large N. Both equations make manifest the slow roll expansion as an expansion in 1/N. 7 Thenumberofe-foldsN[χ]sincethefieldχexitsthehorizontilltheendofinflation(whereχtakesthevalueχ ) end can be computed in close form from eqs. (2.7) in the slow-rollapproximation (neglecting 1/N corrections) N[χ] χend w(χ) = dχ 61. (2.8) N − w (χ) Zχ ′ Inflation ends after a finite number of efolds provided w(χ )=w (χ )=0 (2.9) end ′ end This condition will be enforced in the inflationary potentials. The amplitude of adiabatic scalar perturbations is expressed as [3, 4, 5, 7] N2 M 4 w3(χ) ∆(S) 2 = . (2.10) | kad| 12π2 (cid:18)MPl(cid:19) w′2(χ) The spectral index n , its running and the ratio of tensor to scalar fluctuations are expressed as s 2 3 w (χ) 2 w (χ) ′ ′′ n 1= + , s − −N w(χ) N w(χ) (cid:20) (cid:21) dn 2 w (χ)w (χ) 6 [w (χ)]4 8 [w (χ)]2 w (χ) s ′ ′′′ ′ ′ ′′ = + , dlnk −N2 w2(χ) − N2 w4(χ) N2 w3(χ) 8 w (χ) 2 ′ r = . (2.11) N w(χ) (cid:20) (cid:21) In eqs.(2.8)-(2.11) the field χ is computed at horizon exiting. We choose N[χ]=N =50. Since, w(χ) and w (χ) are of order one, we find from eq.(2.10) ′ M 2 2√3π ∆(S) 1.02 10 5 . (2.12) M ∼ N | kad|≃ × − (cid:18) Pl(cid:19) where we used N 50 and the WMAP value for ∆(S) = (4.67 0.27) 10 5 [5]. This fixes the scale of inflation ≃ | kad| ± × − to be M 3.19 10−3 MPL 0.77 1016GeV. ≃ × ≃ × This value pinpoints the scale of the potential during inflation to be at the GUT scale suggesting a deep connection between inflation and the physics at the GUT scale in cosmological space-time. We see that n 1 as well as the ratio r turn out to be of order 1/N. This nearly scale invariance is a natural s | − | property of inflation which is described by a quasi-de Sitter space-time geometry. This can be understood intuitively as follows: the geometry of the universe is scale invariant during de Sitter stage since the metric takes in conformal time the form 1 ds2 = (dη)2 (d~x)2 . (H η)2 − (cid:2) (cid:3) Therefore, the primordial power generated is scale invariant except for the fact that inflation is not eternal and lasts for N efolds. Hence, the primordialspectrumis scaleinvariantup to 1/N corrections. The Harrison-Zeldovichvalues n = 1, r = 0 and dn /dlnk = 0 correspond to a critical point as discussed in ref.[7]. This gaussian fixed point is s s not the inflationary model that reproduces the data but the inflation model hovers around it in the renormalization group sense with an almost scale invariant spectrum of scalar fluctuations during the slow roll stage. We analyze in the subsequent sections chaotic inflation and new inflation in its simple physical realizations within theGinzburg-Landauapproach(the trinomialpotential)[7,8,9]. WeperformwiththesemodelsMonteCarloMarkov chains analysis of the three years WMAP data and LSS data. 8 1 0.9 h = 0 0.8 h = −0.5 0.7 h = −0.8 0.6 ) χ ( w ) 0.5 8 y/ h = −1 ( 0.4 0.3 0.2 0.1 0 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 (y/8)1/2 χ FIG. 1: Trinomial Chaotic Inflation. We plot here the chaotic inflation trinomial potential [eq.(3.2) with positive quadratic term] y w(χ)vs. thefieldvariable√z=py/8χfordifferentvaluesoftheasymmetryparameterh,namely,h=0, 0.5, 0.8 8 − − and 1. Notice theinflection point at √z =1 when h= 1. − − III. TRINOMIAL CHAOTIC INFLATION: SPECTRAL INDEX, AMPLITUDE RATIO, RUNNING INDEX AND LIMITING CASES We consider now the trinomial potential with unbroken symmetry investigated in ref.[8]: m2 mg λ V(φ)= φ2+ φ3+ φ4 , (3.1) 2 3 4 where m2 >0 and g and λ are dimensionless couplings. The corresponding dimensionless potential w(χ) eqs.(1.1)-(1.2 ) takes the form 1 h y y w(χ)= χ2+ χ3+ χ4 , (3.2) 2 3 2 32 r where the quartic coupling y is dimensionless as well as the asymmetry parameter h. The couplings in eq.(3.1) and eq.(3.2) are related by eq.(1.6). Chaoticinflationisobtainedbychoosingtheinitialfieldχintheinterval(0,+ ). Theinflatonχslowlyrollsdown ∞ the slope of the potential from its initial value till the absolute minimum of the potential at the origin. Computing the number of efolds from eq.(2.8), we find the field χ at horizon crossing related to the couplings y and h. Without loss of generality, we choose h<0 and shall work with positive fields χ. The potential eq.(3.2) has extrema at χ=0 and χ given by, ± 8 χ = [ h i∆] , ∆ 1 h2 . (3.3) ± y − ± ≡ − r p That is, for h <1, w(χ) has only one realextremumat χ=0 while for h 1, w(χ) has three realextrema. There | | | |≥ is always a minimum at χ=0 since w (0)=1. At the non-zero extrema we have ′′ w (χ )= 2∆ (∆ ih) . ′′ ± − ± 9 We have for h >1, | | 8 χ = h h2 1 , and w (χ )=2 h2 1 h2 1 h . ′′ ± y − ± − ± − − ∓ r h p i p (cid:16)p (cid:17) Hence, for any h< 1, we have w (χ )>0 while w (χ )<0. Notice that χ >0 for h< 1. ′′ + ′′ Therefore, we hav−e chaotic inflation for positive field χ−in the regime h <1±using the infl−aton potential eq.(3.2). | | We can alsohave chaotic inflationwith the potential eq.(3.2) for negative field if h>3/√8, but for h >3/√8 the | | absolute minimum is no more at χ = 0 but at χ . Since w(χ )< 0 for h > 3/√8 we have to subtract in this case + + | | the value w(χ ) from w(χ) in order to enforce eq.(2.9). + We consider in subsections IIIA, IIIB and IIIC the regime 1 < h 0, χ 0. The case h < 1 is analyzed in − ≤ ≥ − subsection IIID. A. The small asymmetry regime: 1<h 0. − ≤ Inchaoticinflationtheinflatonfieldslowlyrollsdowntheslopeofthepotentialfromitsinitialvaluetilltheabsolute minimum of the potential at χ=0. It is convenient to define the field variable z by: y z χ2 . (3.4) ≡ 8 In terms of z the chaotic trinomial potential takes the form 4 4 1 w(χ)= z 1+ h√z+ z . y 3 2 (cid:18) (cid:19) When z . 1 we are in the quadratic regime where w(χ) is approximated by the χ2 term. For z & 1 we go to the non-linear regime in z and all three terms in w(χ) are of the same order of magnitude. Infig. 1,weplot y w(χ) asafunctionof√z forseveralvaluesofh 1. Weseethatthepotentialbecomesflatter 8 ≥− as h decreases. For h= 1, both w (χ) and w (χ) vanish at √z =1. The case h= 1 is singular since the inflaton ′ ′ − − gets stuck an infinite amount of time at the point √z =1. By inserting eq.(3.2) for w(χ) into eq.(2.8) for N[χ] and setting N[χ]=N we obtain the field χ or equivalently z , at horizon exit, in terms of the coupling y and the asymmetry parameter h, 4 4 4h 5 h+√z h y =z+ h√z+ 1 h2 log 1+2h√z+z 2h2 arctan arctan (3.5) 3 − 3 − 3∆ 2 − ∆ − ∆ (cid:18) (cid:19) (cid:18) (cid:19)(cid:20) (cid:18) (cid:19) (cid:18) (cid:19)(cid:21) (cid:0) (cid:1) This defines the field z as a monotonically increasing function of the coupling y for 0 < y, z < + . Recall that χ ∞ and z corresponds to the time of horizon exit. Weobtainfromeqs.(2.10),(2.11)and(3.2)thespectralindex,itsrunning,theratiorandtheamplitudeofadiabatic perturbations, 2 y (1+2h√z+z) 1+4h√z+3z n = 1 3 , (3.6) s − 2N z " 1+ 4h√z+ 1z 2 − 1+ 4h√z+ 1z# 3 2 3 2 (cid:0) (cid:1) dn y2 (1+2h√z+z) h√z+ 3z (1+2h√z+z)4 s = 2 3 + dlnk 2z2N2 "− 1+ 4h√z(cid:0)+ 1z 2 (cid:1) − 1+ 4h√z+ 1z 4 3 2 3 2 (cid:0) (cid:1) (cid:0) (cid:1) 2 (1+2h√z+z) (1+4h√z+3z) + 2 , (3.7) 1+ 4h√z+ 1z 3 # 3 2 (cid:0) (cid:1) 2 4y (1+2h√z+z) r = , (3.8) N z 1+ 4h√z+ 1z 2 3 2 (cid:0) (cid:1) 10 2N2 M 4 z2 1+ 4h√z+ 1z 3 ∆(S) 2 = 3 2 . (3.9) | kad| 3π2 (cid:18)MPl(cid:19) y2(cid:0)(1+2h√z+z)2(cid:1) In chaotic inflation, the limit z 0+ implies y 0+ (shallow limit), we have in this limit: → → y z→=0+ 2z+ (z32) , ns y→=0+ 1 2 , O − N r y→=0+ 8 , ∆(S) 2 y→=0+ N2 M 4 . (3.10) N | kad| 6π2 M (cid:18) Pl(cid:19) Theresultsinthey 0+ limitareindependentoftheasymmetryhandcoincidewiththoseforthepurelyquadratic monomial potential→1 χ2. 2 In the limit z + which implies y + (steep limit), we have for fixed h> 1, → ∞ → ∞ − 4 4 y z→=+∞z+ h√z+ 1 h2 logz+ (1), 3 − 3 O (cid:18) (cid:19) ns y→=+∞1 3 1+ 4 h + 1 4 h2 logz + 1 , − N 3 √z − 3 z O z (cid:20) (cid:18) (cid:19) (cid:18) (cid:19)(cid:21) r y→=+∞ 16 1+ 4 h + 1 4 h2 logz + 1 . N 3 √z − 3 z O z (cid:20) (cid:18) (cid:19) (cid:18) (cid:19)(cid:21) ∆(S) 2 y→=+∞ N2 M 4 z . (3.11) | kad| 12π2 M 2 (cid:18) Pl(cid:19) 1+ 4 h + 1 4 h2 logz + 1 3 √z − 3 z O z h (cid:0) (cid:1) (cid:0) (cid:1)i For h=0, n and r in the limit y + coincide with those of the purely quartic monomial potential 1 χ4: s → ∞ 2 3 16 n =1 , r = . (3.12) s − N N B. The flat potential limit h 1+ →− We consider chaotic inflation in the regime 1<h 0, χ 0. − ≤ ≥ At h = 1 the potential eq.(3.2) exhibits an inflexion point at χ 8. Namely, w (χ ) = w (χ ) = 0 while − 0 ≡ y ′ 0 ′′ 0 w(χ )=2/(3y)>0. That is, this happens at z = y χ2 =1. q 0 0 8 0 Therefore,forh> 1butveryclosetoh= 1,thefieldevolutionstronglyslowsdownnearthepointχ=χ . This 0 − − strong slow down shows up in the calculation of observables when the field χ at horizon crossing is χ > χ , namely, 0 for z > z = 1. For χ < χ , that is z < 1, the slow down of the field evolution will only appear if z 1. Therefore, 0 0 ≃ the limit h= 1 is singular since the inflaton field gets trapped at the point z =1. − Let us derive the h 1+ limit of y, n and r from eqs.(3.5)-(3.9) in the regimes z <1 and z >1, respectively. s →− When h 1+ we see from eq.(3.3) that ∆ 0 and the arguments of the two arctan in eq.(3.5) diverge. Hence, →− → the arctan tends to +π/2 or π/2. Depending on whether z < 1 or z > 1, the π/2’s terms cancel out or add, − respectively. The special case z =1 is investigated in the next subsection IIIC. In the case z <1 we get: 4 2 2 √z y =z √z log 1 √z + , h 1+ , z <1, (3.13) − 3 − 3 − 31 √z →− − (cid:0) (cid:1) 4 y (1 √z) (1 √z)(1 3√z) n = 1 3 − − − , h 1+ , z <1, (3.14) s − 2N z " 1 4 √z+ 1z 2 − 1 4 √z+ 1z # →− − 3 2 − 3 2 (cid:0) (cid:1) 4 4y (1 √z) r = − , h 1+ , z <1, N z 1 4 √z+ 1z 2 →− − 3 2 (cid:0) (cid:1)