Revisiting non-Gaussianity in non-attractor inflation models in the light of the cosmological soft theorem
RRevisiting non-Gaussianity in non-attractor inflation models in the light of thecosmological soft theorem
Teruaki Suyama, Yuichiro Tada, ∗ and Masahide Yamaguchi Department of Physics, Tokyo Institute of Technology,2-12-1 Ookayama, Meguro-ku, Tokyo 152-8551, Japan Department of Physics, Nagoya University, Nagoya 464-8602, Japan (Dated: February 17, 2021)We revisit the squeezed-limit non-Gaussianity in the single-field non-attractor inflationmodels from the viewpoint of the cosmological soft theorem. In the single-field attractormodels, inflaton’s trajectories with different initial conditions effectively converge into asingle trajectory in the phase space, and hence there is only one clock degree of freedom(DoF) in the scalar part. Its long-wavelength perturbations can be absorbed into the localcoordinate renormalization and lead to the so-called consistency relation between n - and( n + 1)-point functions. On the other hand, if the inflaton dynamics deviates from theattractor behavior, its long-wavelength perturbations cannot necessarily be absorbed andthe consistency relation is expected not to hold any longer. In this work, we derive a formulafor the squeezed bispectrum including the explicit correction to the consistency relation, asa proof of its violation in the non-attractor cases. First one must recall that non-attractorinflation needs to be followed by attractor inflation in a realistic case. Then, even if a specificnon-attractor phase is effectively governed by a single DoF of phase space (represented by theexact ultra-slow-roll limit) and followed by a single-DoF attractor phase, its transition phasenecessarily involves two DoF in dynamics and hence its long-wavelength perturbations cannotbe absorbed into the local coordinate renormalization. Thus, it can affect local physics, eventaking account of the so-called local observer effect , as shown by the fact that the bispectrumin the squeezed limit can go beyond the consistency relation. More concretely, the observedsqueezed bispectrum does not vanish in general for long-wavelength perturbations exitingthe horizon during a non-attractor phase. I. INTRODUCTION
The primordial curvature perturbations give us rich information on the dynamics of an inflationas well as serve as a seed of large scale structure formation. Their statistical features can beprobed even beyond the linear order recently. Then, the soft limit of the correlation functions ofthe primordial curvature perturbations becomes a powerful tool to probe the inflation dynamicsthrough clarifying the relation between their correlation functions. One of such examples is so-calledMaldacena’s consistency relation [1], which connects the squeezed limit of the bispectrum (( n + 1)-point functions) to the power spectra ( n -point functions). This relation holds true [2–4], as long as(i) there is only a single scalar field showing the attractor behavior (single clock inflation) and (ii)the vacuum is the Bunch-Davies one [5, 6]. In this paper, we focus on the former assumption (i). Itshould be noticed that, even if there is only a single scalar field, Maldacena’s consistency relationcan be violated. Such an example is ultra-slow-roll (USR) inflation (a non-attractor inflation) [7,8]. In USR inflation, its dynamics shows non-attractor-like behavior, depending solely on itsmomentum (velocity) thanks to the shift symmetry (though trajectories do not converge becauseof the dependence of an initial field value), and the comoving curvature perturbations can evolveeven after the horizon exit, which might lead to the violation of the consistency relation [9, 10].This feature is in sharp contrast to that of standard slow-roll (attractor) inflation, where the ∗ [email protected] a r X i v : . [ h e p - t h ] F e b momentum of a scalar field is determined by the field value, that is, trajectories with differentinitial conditions effectively converge into a single trajectory in the phase space and the curvatureperturbations become constants after the horizon exit. In this case, its long-wavelength curvatureperturbations can be renormalized into the redefinition of the local (background) coordinate, whichleads to the consistency relation.Recently, there appear some discrepancies among the literature [11–15] on the soft limit ofprimordial curvature perturbations for an inflation model including a non-attractor phase. USRinflation cannot last long enough to solve the horizon and flatness problems without fine-tuningof an initial condition and cannot fit the observed density perturbations, so that it needs to befollowed by standard slow-roll (SR) inflation. Then, the key question is whether Maldacena’s con-sistency relation for a long-wavelength mode exiting the horizon during an USR phase is finallyrecovered after an USR phase ends and the comoving curvature perturbations stop growing. Someliterature [11, 12] demonstrates that the final bispectrum in the squeezed limit depends on the fea-ture of the transition from an USR phase to a SR phase and can violate the consistency relation.On the other hand, another [15] claims that it finally reduces to Maldacena’s consistency relation.In particular, in Refs. [13–15], they extended Maldacena’s consistency relation to that includingthe time evolution of the curvature perturbations so as to discuss the case of USR inflation. Ac-cording to their formula, they claim that, once an USR phase ends and the comoving curvatureperturbations stop growing, it would finally recover Maldacena’s consistency relation. In this paper, we revisit this issue and point out that it is not enough to include the timedependence of the comoving curvature perturbations to discuss the above issue. In fact, one needs tofurther extend the soft-limit bispectrum formula given in Refs. [13–15] by including the momentumdependence of an inflaton in order to deal with an inflation model including a non-attractor phase.This is because, as we mentioned before, though the dynamics of an inflaton during both of USR andSR phases are governed by a single degree of freedom and its long-wavelength perturbations can beabsorbed into the local coordinate renormalization, two phase space degrees of freedom necessarilyplay an important role in the dynamics during the transition period, which leave imprints on alater SR phase. Thus, we need to take momentum dependence into account for the evolution of thecomoving curvature perturbations even after the renormalization of field-value dependence, whichcan finally lead to the violation of Maldacena’s consistency relation for an inflation model includinga non-attractor phase, reproducing the results in the literature successfully.We also mention the local observer effects [17–19]. As first noticed in Ref. [17], a local observercannot be the comoving observer due to the modulation of the long wavelength mode. In case ofstandard slow-roll and single-field inflation, this local observer effect exactly cancels the squeezedlimit of the bispectrum and hence the observed bispectrum in that limit vanishes except the sec-ondary effects. Then, the question arises what happens for USR inflation. Some literature [13–15]claims that the observed squeezed bispectrum universally vanishes irrespective of attractor or non-attractor inflation models. In this paper, we also try to address this question and will show that theobserved bispectrum in the squeezed limit for a long-wavelength mode exiting the horizon duringan USR phase does not vanish in general.The paper is organized as follows. In the next section, we derive the long-wavelength modulationon the short mode by including not only time dependence but also momentum dependence in orderto deal with a non-attractor inflation model (with a transition to a SR phase) as well as a pureattractor one. Then, we give the generalized single-field soft theorem. In Sec. III, we demonstratea generic method to evaluate the squeezed bispectrum of the curvature perturbations by takingaccount of the transition from an USR phase to a SR phase adequately. In Sec. IV, concrete Ref. [16] derives the same formula as Refs. [13–15]. However they explicitly restrict their discussion to theorieswith the exact shift symmetry, having no conflict with our results. The local observer effects on another soft limit called collapsed limit [20, 21] are discussed in Ref. [22]. examples are discussed to show how powerful our formula is and to reproduce the results obtainedin the literature. The final section is devoted to conclusions and discussion
II. CONSISTENCY RELATION OR NOT IN NON-ATTRACTOR CASES
Maldacena’s first computation of the cubic action reveals that the following consistency re-lation holds between the power spectrum P ζ and the squeezed bispectrum B ζ of the curvatureperturbation ζ in attractor single-field inflation [1]: B ζ ( q, k, k (cid:48) ) → q (cid:28) k,k (cid:48) (1 − n s ) P ζ ( k ) P ζ ( q ) , (1)where n s is the spectral index defined by n s − P ζ d log k . (2)It is now well understood as one example of the so-called cosmological soft theorem (see, e.g.,Refs. [23, 24]), which connects the squeezed ( n + 1)-point function with the n -point function.The ordinary cosmological soft theorem is a consequence of the fact that the long-wavelengthperturbation in the unique scalar degree of freedom (DoF) can be locally absorbed into the space-time coordinate renormalization in the single-field and attractor cases. It is easily understood inthe comoving gauge, where the inflaton field φ has no perturbation, φ ( t, x ) = φ ( t ) , (3)from the background value φ ( t ) and all perturbations are expressed in the metric form. Theseparate universe assumption requires that the spacetime geometry is approximated by the flatFLRW one in the long-wavelength limit asd s = − d t + ¯ a ( t ) d x + O (cid:18) k a H (cid:19) . (4)Here the local scale factor ¯ a = a e ζ L is modulated from the background a by the long-wavelengthcurvature perturbation ζ L . H = ˙ a/a is the Hubble parameter and k L is the corresponding wavenum-ber to the superhorizon perturbation ζ L . The general conservation of the curvature perturbationon a superhorizon scale ensures that ζ L is constant in the attractor case [25], and any constantrenormalization of the scale factor does not affect the local physics. The only difference is thenon-physical labeling between the physical length and the comoving coordinate. Due to the rescaleof ¯ a = a e ζ L , a specific physical length in the local patch is differently labeled to the comovingcoordinate by the factor e − ζ L from the background universe. It leads to an apparent modulationon the two-point function ξ ζ ( x ) = (cid:104) ζ ( y ) ζ ( y + x ) (cid:105) as¯ ξ ζ ( x ) = ξ ζ (cid:16) x e ζ L (cid:17) (cid:39) (1 + ζ L x i ∂ x i ) ξ ζ ( x ) , (5)at linear order in ζ L , where ¯ ξ ζ ( x ) denotes the (small-scale) two-point function modulated by thelong-wavelength perturbation. In Fourier space, one obtains¯ P ζ ( k ) (cid:39) (cid:20) − ζ L (cid:18) d log P ζ ( k )d log k + 3 (cid:19)(cid:21) P ζ ( k ) , (6)making use of integration by parts and the identity ∂ x i (cid:0) x i e − i k · x (cid:1) = ∂ k i (cid:0) k i e − i k · x (cid:1) . If there is noother bispectrum source, one reproduces the consistency relation (1) from this equation.Is it then possible to generalize this discussion to the non-attractor case? Beyond the attractorlimit, there can be two independent DoF, φ and ˙ φ , even in a single-field model, and thus it isexpected that the long-wavelength perturbation cannot be necessarily absorbed into the coordinaterenormalization any longer. It is in fact clarified as follows, extending the discussion in Ref. [15].Let us first explicitly show the “renormalization” procedure. In general, the spacetime metric witha long-wavelength perturbation can be written in the comoving gauge asd s = − e δ N L d t + a ( t )e ζ L (cid:0) d x + N L d t (cid:1) . (7)The linear constraint reads [1] δ N L ( t ) = 1 H ˙ ζ L ( t ) , N L ( t, x ) = 13 (cid:15) H x ˙ ζ L ( t ) , (8)in the long-wavelength limit. Here (cid:15) H = − ˙ H/H is the first slow-roll parameter. The inflaton fieldagain has no perturbation: φ ( t, x ) = φ ( t ). The background dynamics is dictated by ∂ φ ∂t + 3 H ∂φ ∂t + ∂∂φ V ( φ ) = 0 , M H = 12 (cid:18) ∂φ ∂t (cid:19) + V. (9)The interesting point is that, in this long-wavelength limit, the local dynamics can be understoodlike as a background one without perturbations. That is, one can take a coordinate transformation t → ¯ t = t + ξ ( t ) , x → ¯ x = e β ( t ) / x , (10)such that the metric shows a “background-like” formd s = − d¯ t + ¯ a d¯ x , ¯ a (¯ t ) = a ( t )e α ( t ) , (11)and the transformed inflaton ¯ φ (¯ t ) = φ ( t ) follows “background-like” equations of motion (EoM) ∂ ¯ φ∂ ¯ t + 3 ¯ H ∂ ¯ φ∂ ¯ t + ∂∂ ¯ φ V ( ¯ φ ) = 0 , M ¯ H = 12 (cid:18) ∂ ¯ φ∂ ¯ t (cid:19) + V, (12)with ¯ H = ( ∂ ¯ t ¯ a ) / ¯ a . Specifically one can take parameters as˙ ξ = 1 H ˙ ζ L , α + 13 β = ζ L , ˙ β = (cid:15) H ˙ ζ L , (13)and the above speculation is confirmed with use of the EoM for the curvature perturbation on asuperhorizon scale: dd t (cid:16) (cid:15) H a ˙ ζ L (cid:17) = 0 . (14)However it does not necessarily mean that the long-wavelength perturbation does not affect thelocal physics. As EoM (12) are second order in ¯ φ and first order in ¯ a (through ∂ ¯ t ¯ a = ¯ a ¯ H ), theirsolution is labeled by three independent DoF (cid:0) ¯ φ (¯ t ) , ¯ φ (cid:48) (¯ t ) , ¯ a (¯ t ) (cid:1) . Therefore, even if ¯ φ (¯ t ) is fixed tothe background value φ ( t ) and the scale factor is in itself irrelevant to physics, the local observercan distinguish the long-wavelength mode from the background variable via the time derivative¯ φ (cid:48) (¯ t ), which was not included in Ref. [15]. In fact the time derivative in the local patch is differentfrom the background one by ¯ φ (cid:48) (¯ t ) = (cid:18) ∂ ¯ t∂t (cid:19) − ∂ t ¯ φ (¯ t ) = (cid:16) ξ (cid:17) − ˙ φ , (15)even after the “renormalization” of the long-wavelength mode.Such a “renormalization” is nevertheless useful in order to evaluate the bispectrum in thesqueezed limit. The above discussion still holds even if one includes the short-wavelength pertur-bation ζ S in Eq. (7). Under the coordinate transformation (10), the short-wavelength curvatureperturbation transforms as a scalar [15], ζ S ( t, x ) = ¯ ζ (¯ t, ¯ x ) . (16)Therefore the short-wavelength mode accompanied by the long-wavelength perturbation is equiv-alent to the short-wavelength solution of the Mukhanov-Sasaki equation on the transformed back-ground (cid:0) ¯ φ (¯ t ) , ¯ φ (cid:48) (¯ t ) , ¯ a (¯ t ) (cid:1) without the long-wavelength mode. As ¯ φ (¯ t ) is fixed to φ ( t ), we label sucha solution by (cid:0) ¯ φ (cid:48) (¯ t ) , ¯ a (¯ t ) (cid:1) as ¯ ζ (¯ t, ¯ x ) = ζ [¯ t, ¯ x | ¯ φ (cid:48) (¯ t ) , ¯ a (¯ t )] to express this fact. Then, from Eq. (16),we have ζ S ( t, x ) = ζ (cid:2) ¯ t, ¯ x | ¯ φ (cid:48) (¯ t ) , ¯ a (¯ t ) (cid:3) . (17)Inclusion of ¯ φ (cid:48) -dependence is a new point of this paper. At linear order in ζ L , the explicit form ofbarred variables leads to ζ S ( t, x ) (cid:39) ζ (cid:34) t + ξ , (cid:18) β (cid:19) x (cid:12)(cid:12)(cid:12)(cid:12)(cid:12) (cid:32) − ˙ ξ − ¨ φ ˙ φ ξ (cid:33) ˙ φ ( t + ξ ) , (cid:0) α − Hξ (cid:1) a ( t + ξ ) (cid:35) . (18)We here note its dependence on the scale factor. The scale factor appears in EoM only through thecombination a − ∂ x , so that the solution satisfies a general relation ζ [ t, x | ˙ φ, a ] = ζ [ t, C x | ˙ φ, C − a ]for a constant C . Therefore, neglecting the time dependence of (1 + α − Hξ ), one goes further ζ S ( t, x ) (cid:39) ζ (cid:34) t + ξ , (cid:0) ζ L − Hξ (cid:1) x (cid:12)(cid:12)(cid:12)(cid:12)(cid:12) (cid:32) − ˙ ξ − ¨ φ ˙ φ ξ (cid:33) ˙ φ ( t + ξ ) , a ( t + ξ ) (cid:35) . (19)Therefore the short-wavelength mode is equivalent to the curvature perturbation solution onthe modified background momentum ˙ φ = (cid:16) − ˙ ξ − ¨ φ ˙ φ ξ (cid:17) ˙ φ at the shifted spacetime point (cid:0) t + ξ , (1 + ζ L − Hξ ) x (cid:1) . In terms of the Fourier-space two-point function, the long-wavelengthmodulation on the short mode is expressed by (cid:104) ζ S ( k ) ζ S ( k ) (cid:105) = (cid:104) ζ ( k ) ζ S ( k ) (cid:105) + ξ ( k L ) ˙ P ζ [ t, k S | ˙ φ ] − (cid:0) ζ L ( k L ) − Hξ ( k L ) (cid:1) ( n s − P ζ [ t, k S | ˙ φ ] − (cid:16) ˙ φ ˙ ξ ( k L ) + ¨ φ ξ ( k L ) (cid:17) ∂ ˙ φ P ζ [ t, k S | ˙ φ ] ˙ φ = ˙ φ , (20)where the first term denotes the fiducial two-point function without long-wavelength perturba-tions, and the ensemble average is performed for the short-wavelength modes under the fixedlong-wavelength mode ξ . Here we restore the spatial dependence of the long-wavelength modewith k + k + k L = and k S = k (cid:29) k L . The squeezed-limit bispectrum is given by its correlationwith the long-wavelength mode: (cid:104) ζ L ( k ) (cid:104) ζ S ζ S (cid:105) ( k , k ) (cid:105) = lim k → (2 π ) δ (3) ( k + k + k ) B ζ ( k , k , k ) . (21) Notice that ¯ t = const. hypersurface and the comoving hypersurface approximately coincide only on short scales. Note that the Hubble parameter is here understood as a function of φ and ˙ φ rather than the time derivative of thescale factor. In fact its time derivative ˙ α + (cid:15) H H ξ − H ˙ ξ is always slow-roll suppressed at most via Eqs. (13) and (14). Itsnegligibility is confirmed in a concrete example shown in Fig. 2. Thus one obtains the generalized single-field soft theorem in a formal expression aslim k → B ζ ( t ; k , k , k ) = (1 − n s ) P ζ ( t, k L ) P ζ ( t, k S )+ (cid:18)(cid:90) t ∂ t (cid:48) P ζ ( t, t (cid:48) ; k L ) H d t (cid:48) (cid:19)(cid:104) ˙ P ζ ( t, k S ) + H ( n s − P ζ ( t, k S ) (cid:105) − (cid:34) ˙ φ H ˙ P ζ ( t, k L ) + ¨ φ (cid:90) t ∂ t (cid:48) P ζ ( t, t (cid:48) ; k L ) H d t (cid:48) (cid:35) ∂ ˙ φ P ζ [ t, k S | ˙ φ ] ˙ φ = ˙ φ , (22)where P ζ ( t, t (cid:48) ; k ) denotes the unequal time two-point function: (cid:104) ζ ( t, k ) ζ ( t (cid:48) , k (cid:48) ) (cid:105) = (2 π ) δ (3) ( k + k (cid:48) ) P ζ ( t, t (cid:48) ; k ). The lower limit of the time integration of ˙ ξ should be chosen so that the resultant ξ properly satisfies its definition ¯ φ ( t + ξ ) = φ ( t ).The last term represents the ˙ φ correction to the generalized “consistency relation” proposed inRefs. [13–15] (see also Ref. [16] deriving the same formula but with a restriction to the exact shiftsymmetry). In a pure attractor case, the time-independence of ζ L recovers Maldacena’s consistencyrelation (1) from this formula. In the USR limit, | V (cid:48) | (cid:28) | H ˙ φ | , as an extreme example of non-attractor models, the curvature perturbation grows as ζ ∝ a and one finds that the last term againvanishes, making use of the background EoM ¨ φ + 3 H ˙ φ = 0. It could be understood to reflectthe fact that there is effectively only one DoF ˙ φ during the exact USR phase and the effect of sucha single DoF can be expressed by the spacetime coordinate renormalization. Making use of thetime-dependence ˙ P ζ = 6 HP ζ and the scale-invariance n s − f NL = 5 /
2, in terms of the non-linearity parameter f NL = 512 B ζ ( k L , k S , k S ) P ζ ( k L ) P ζ ( k S ) . (23)These results in the two extreme cases are consistent with the previous study [15]. However thelast correction does not vanish in general and can give a non-zero contribution to the squeezedbispectrum. In particular, once inflation experiences an USR phase, even if it is followed by a SRphase and ˙ φ quickly converges to the attractor value, the tiny difference in ˙ φ keeps retaining theinformation of perturbations during USR and yielding a non-vanishing bispectrum well after thetransition as we see in the rest of the paper.We here fix the field value φ and express the bispectrum correction in terms of the momentum-dependence of the power spectrum. We note that it is not necessary: one can instead fix themomentum and find the correction as the φ -dependence of the power spectrum to see the differencein the phase-space trajectory. In Appendix A, we concretely discuss this equality and successfullyobtain the same formula (45) we will derive in the next section. III. BISPECTRUM IN THE TRANSITION FROM ULTRA SLOW-ROLL TOSLOW-ROLL
We here focus on the models starting from an USR phase and followed by a SR phase. Mak-ing use of the modulated two-point function (20), we derive a general formula for the squeezedbispectrum well after the transition from USR to SR. Because perturbations exiting the horizonduring USR are known to be scale-invariant n s − P ζ → P ζ ’s dependence on it. The schematic figure 1 summarizes our setup.Let us first assume the situation that the USR condition is well satisfied before the inflatonreaches some value φ c , while the potential can start to deviate from the USR one (or even it can ϕ L ϕ c ϕ ϕ USR SR t L t c t ϕ ϕ δ ϕ L δ ϕ( t c ) δ ϕ( t ) δ ϕ c δ ϕ( t c ) δ ϕ( t ) FIG. 1. The schematic picture of our notation and setup to study the transition from the ultra slow-roll phase(gray region) to the slow-roll phase (white region). The inflaton φ evolves from left to right without loss ofgenerality. The inflaton dynamics is assumed to be well approximated by the USR solution before some point φ c . After φ c , the dynamics gradually or rapidly deviates from the USR one and it finally reaches the SRphase, where we evaluate the squeezed bispectrum at the time t . The blue line illustrates the unperturbedbackground trajectory φ , while the orange dashed line represents the perturbed (shifted) trajectory ¯ φ (27).Black dotted lines are the equal-time hypersurfaces. At the initial time t L , the long-wavelength perturbationshould be represented only by the field difference δφ L with the equated momenta: ¯ φ (cid:48) ( t L ) = ˙ φ ( t L ) (28). Thisfield difference nevertheless yields the momentum perturbation δ ˙ φ c at φ = φ c (32), which characterizes thelater perturbation. After φ = φ c , we allow a general potential form as long as it converges to a SR-type one.Converting δ ˙ φ c to the initial condition of the perturbation at t = t c , δφ ( t c ) and δ ˙ φ ( t c ) (35), we consider thesolution of the perturbed EoM (37) along such a potential. suddenly change) at that point. Then the chain rule leads to ∂P ζ ∂ ˙ φ = (cid:32) ∂ ˙ φ∂ ˙ φ c (cid:33) − ∂P ζ ∂ ˙ φ c , (24)where ˙ φ c is the momentum at φ = φ c . It is also useful to define the power of P ζ in ˙ φ c around thebackground trajectory by n := − ∂ log P ζ ∂ log ˙ φ c (cid:12)(cid:12)(cid:12)(cid:12) φ = φ , (25)so that Eq. (24) reads ∂P ζ ∂ ˙ φ = − n (cid:32) ∂ ˙ φ∂ ˙ φ c (cid:33) − P ζ ˙ φ c . (26)Then the remained non-trivial part is the ˙ φ c dependence of the momentum value ˙ φ at the evaluationtime.We then specifically consider the shifted trajectory ¯ φ ( t ) from the background solution φ ( t ).How should this ¯ φ ( t ) be defined? To utilize the formula (19), one finds that the shifted trajectory¯ φ ( t ) and the time shift ξ should satisfy¯ φ ( t + ξ ) = φ ( t ) , ¯ φ (cid:48) ( t ) = (cid:32) − ˙ ξ − ¨ φ ˙ φ ξ (cid:33) ˙ φ ( t ) . (27)In particular, the momentum shift vanishes, (cid:16) − ˙ ξ − ¨ φ ˙ φ ξ (cid:17) ˙ φ = ˙ φ , during USR as we mentionedin the second-to-last paragraph of the previous section. Therefore, at some initial time t L in theUSR phase, we only shift the field value and keep the momentum intact:¯ φ ( t L ) = φ ( t L ) + δφ L , ¯ φ (cid:48) ( t L ) = ˙ φ ( t L ) . (28)From this initial condition, we need derive the ˙ φ c modulation δ ˙ φ c := ¯ φ (cid:48) (¯ t c ) − ˙ φ ( t c ) , with ¯ φ (¯ t c ) = φ c , (29)and the perturbation at the evaluation time δφ ( t ) := ¯ φ ( t ) − φ ( t ) , δ ˙ φ ( t ) := ¯ φ (cid:48) ( t ) − ˙ φ ( t ) , (30)so that the chain-rule coefficient is given by (cid:32) ∂ ˙ φ ( t ) ∂ ˙ φ c (cid:33) − = δ ˙ φ c δ ˙ φ ( t ) (cid:12)(cid:12)(cid:12)(cid:12)(cid:12) δφ L → . (31)The numerator δ ˙ φ c can be easily obtained. With use of the USR EoM ¨ φ + 3 H c ˙ φ = 0 with aconstant Hubble parameter H c , one finds δ ˙ φ c = 3 H c δφ L . (32)The time shift at linear order, δt c := ¯ t c − t c (cid:39) − δφ L ˙ φ c , (33)is also useful. We then divide the shifted solution into the USR and SR ones as¯ φ ( t ) = (cid:40) ¯ φ USR ( t ) , for t < ¯ t c , ¯ φ SR ( t ) , for t ≥ ¯ t c . (34)The initial condition for the perturbation can be set at t c as follows at linear order in δφ L . δφ ( t c ) := ¯ φ SR ( t c ) − φ ( t c ) = ¯ φ SR (¯ t c − δt c ) − φ c (cid:39) − ¯ φ (cid:48) SR (¯ t c ) δt c (cid:39) − ˙ φ c δt c (cid:39) δφ L ,δ ˙ φ ( t c ) := ¯ φ (cid:48) SR ( t c ) − ˙ φ ( t c ) = ¯ φ (cid:48) SR (¯ t c − δt c ) − ˙ φ c (cid:39) δ ˙ φ c − ¯ φ (cid:48)(cid:48) SR (¯ t c ) δt c (cid:39) − V (cid:48) SR ( φ c )˙ φ c δφ L . (35)Once we express the background EoM in the SR phase by¨ φ = − H ( φ, ˙ φ ) ˙ φ − V ( φ ) =: F ( φ, ˙ φ ) , (36) Depending on δφ L , the shifted solution ¯ φ can be still in the USR phase at t c . To consider only the SR phase, wehowever extrapolate the SR solution ¯ φ SR to the time t c and set the perturbation initial condition with use of ¯ φ SR in that case. the perturbation EoM can be written as δ ¨ φ = F ˙ φ δ ˙ φ + F φ δφ, (37)where F ˙ φ = ∂ ˙ φ F and F φ = ∂ φ F . On the other hand, the time derivative of the backgroundEoM (36) reads ... φ = F ˙ φ ¨ φ + F φ ˙ φ. (38)Comparing them, one finds that u = ˙ φ is one solution of the perturbation EoM (37). Anotherindependent solution u can be found through the Wronskian W = u ˙ u − ˙ u u . The Wronskiansatisfies ˙ W = F ˙ φ W, (39)whose solution is obviously given by W = exp (cid:16)(cid:82) tt c F ˙ φ d t (cid:48) (cid:17) . u is then formally solved as u ( t ) = ˙ φ ( t ) (cid:90) tt c W ( t (cid:48) )˙ φ ( t (cid:48) ) d t (cid:48) . (40)Noting u ( t c ) = 0, the perturbation solution with the initial condition (35) is expressed by δφ ( t ) = δφ ( t c ) u ( t c ) (cid:18) u ( t ) − ˙ u ( t c )˙ u ( t c ) u ( t ) (cid:19) + δ ˙ φ ( t c )˙ u ( t c ) u ( t ) (cid:39) (cid:32) ˙ φ ( t )˙ φ c + 3 H c ˙ φ c ˙ φ ( t ) (cid:90) tt c W ˙ φ d t (cid:48) (cid:33) δφ L . (41)Therefore the chain-rule coefficient reads (cid:32) ∂ ˙ φ ( t ) ∂ ˙ φ c (cid:33) − = 3 H c (cid:32) ¨ φ ( t )˙ φ c + 3 H c ˙ φ c ¨ φ ( t ) (cid:90) tt c W ˙ φ d t (cid:48) + 3 H c ˙ φ c W ( t )˙ φ ( t ) (cid:33) − . (42)Given that F ˙ φ = − H + · · · where dots represent slow-roll corrections, the Wronskian W =exp (cid:16)(cid:82) tt c F ˙ φ d t (cid:48) (cid:17) quickly converges to zero for t (cid:29) t c . Hence, in the deep SR phase, one has (cid:32) ∂ ˙ φ∂ ˙ φ c (cid:33) − → H c ˙ φ c ¨ φ ( t ) Z c , Z c := 1 + 3 H c ˙ φ (cid:90) ∞ t c exp (cid:16)(cid:82) tt c F ˙ φ d t (cid:48) (cid:17) ˙ φ ( t ) d t . (43)Noting that ˙ ξ = ˙ ζ L /H quickly disappears soon after φ c and thus ξ can be approximated by thelast value ζ L /H c in the deep SR phase (correction is suppressed by the first slow-roll parameter),one finally finds a simple formula for the bispectrum correction as − (cid:16) ˙ φ ˙ ξ + ¨ φ ξ (cid:17) ∂ ˙ φ P ζ [ t, k S | ˙ φ ] ˙ φ = ˙ φ → nZ c ζ L P ζ ( k S ) . (44)Recalling that n s − P ζ → f NL → n Z c . (45)Therefore the non-zero squeezed bispectrum remains even after the USR phase unless n = 0.0 IV. EXAMPLES
Let us show several concrete consequences of our formula (45) in specific transition models inthis section.
A. Linear slope with sharp transition
We first consider a sharp transition from an exactly flat potential to a constant linear slope: V ( φ ) = V , for φ < φ c ,V (cid:18) − √ (cid:15) φ − φ c M Pl (cid:19) , for φ ≥ φ c . (46)The Hubble parameter is approximated to be constant, 3 M H (cid:39) V . The constant parameter (cid:15) corresponds to the first slow-roll parameter (cid:15) H in the deep SR phase. For this simple potential,one finds an analytic solution for the background as φ ( t ) = φ L + ˙ φ L H (cid:16) − e − H ( t − t L ) (cid:17) , for φ < φ c ,φ c − H (cid:16) π − ˙ φ L e − H ( t c − t L ) (cid:17)(cid:16) − e − H ( t − t c ) (cid:17) + π ( t − t c ) , for φ ≥ φ c , (47)with the φ c -crossing time t c = t L − H log (cid:20) − H ˙ φ L ( φ c − φ L ) (cid:21) , (48)the terminal velocity π = √ (cid:15) M Pl H, (49)and the initial condition φ ( t L ) = φ L , ˙ φ ( t L ) = ˙ φ L . (50)For a shifted trajectory ¯ φ ( t L ) = φ L + δφ L , ¯ φ (cid:48) ( t L ) = ˙ φ L , (51)its analytic expression is also easily obtained as¯ φ ( t ) = φ L + δφ L + ˙ φ L H (cid:16) − e − H ( t − t L ) (cid:17) , for ¯ φ < φ c ,φ c − H (cid:16) π − ˙ φ L e − H (¯ t c − t L ) (cid:17)(cid:16) − e − H ( t − ¯ t c ) (cid:17) + π ( t − ¯ t c ) , for ¯ φ ≥ φ c , (52)with ¯ t c = t L − H log (cid:20) − H ˙ φ L ( φ c − φ L − δφ L ) (cid:21) . (53)1From the condition ¯ φ ( t + ξ ) = φ ( t ), one further finds the solution for ξ at linear order in δφ L as ξ ( t ) (cid:39) − δφ L ˙ φ L e H ( t − t L ) , for t < t c − ˙ φ c + π (cid:0) − e − H ( t − t c ) (cid:1) ˙ φ c (cid:16) ˙ φ c e − H ( t − t c ) + π (cid:0) − e − H ( t − t c ) (cid:1)(cid:17) δφ L , for t ≥ t c . (54)As it is directly related with the curvature perturbation by ζ L = Hξ in the constant H approx-imation, this expression shows the exponential growth of ζ L during the USR phase as well as itsasymptotic value in the deep SR phase ζ L → − H δφ L ˙ φ c in the large transition limit π (cid:29) ˙ φ c . Further-more, this relation is applicable also for short-wavelength modes with a replacement δφ L → δφ S .One then finds P ζ ( k S ) → H ˙ φ P δφ ( k S ) , (55)which corresponds to the power of n = 2 defined in Eq. (25). Z c ’s integration can be also explicitlydone to obtain Z c = 1 + ˙ φ c π (cid:39) . (56)Therefore our formula (45) successfully reproduces the well-studied result by the rigorous in-informalism in the large transition limit [11, 12], f NL → . (57)We mention that the last term in Eq. (42) and the first term in Eq. (44) can be actually non-negligible in the linear slope case (vanishing V (cid:48)(cid:48) ) if the transition is not sufficiently large because¨ φ keeps decaying as well as W ( t ) and ˙ φ ˙ ξ in this case. If one keeps all relevant terms withoutany large transition approximation, our formula is modified as f NL → − δ n Z c , (58)with n →
21 + δ , Z c → − δ , (59)where δ = ˙ φ c /π . The non-linaerity parameter then reads f NL →
52 1(1 + δ ) , (60)which is exactly equivalent to the formula (2.45) in Ref. [11] with η V = 0. Our δ is equivalent to − /h of Ref. [11]. B. Linear slope with smooth transition
The previous linear slope (46) is somewhat unphysical because its first derivative is discontinuousat φ c . In this subsection, we smoothly connect the flat potential and the linear slope, referring tothe prescription in Ref. [12]. That is, we consider the potential V ( φ ) = V − √ (cid:15) φ − φ t + d φ log (cid:16) φ − φ t d φ (cid:17) M Pl , (61)so that its first derivative is continuous at φ t as V (cid:48) ( φ ) = − √ (cid:15) V M Pl (cid:18) φ − φ t d φ (cid:19) , (62)with the smoothness parameter d φ . Note that here φ c is not equivalent to φ t but we rather take φ c much before φ t so that the inflaton dynamics is well approximated by the USR solution until φ c to utilize our formula (45). In this case, the ˙ φ c -dependence n is not necessarily equal to two and Z c is not estimated as unity. Hence we numerically evaluate n and Z c : the power spectrum P ζ iscalculated with use of the δN formalism along various shifted background trajectories to find its˙ φ c ( ˙ φ at φ = φ c ) dependence (25), while Z c (43) is obtained by the numerical integration along theunperturbed trajectory.On the other hand, Passaglia, Hu, and Motohashi (PHM) in Ref. [12] modeled this transitionby the effective δ parameter characterized by the largeness and fastness of the transition. Theydetermined the transition regime as follows. First the end of transition is defined by φ ft := φ t + 2 d φ . (63)On the other hand, the beginning of the transition φ it is not simply given by φ t − d φ but insteaddetermined by the deviation from the USR solution as1 − ˙ φ USR ˙ φ (cid:12)(cid:12)(cid:12)(cid:12)(cid:12) φ it := 0 . − ˙ φ USR ˙ φ (cid:12)(cid:12)(cid:12)(cid:12)(cid:12) φ ft , (64)where φ USR is the USR solution ˙ φ USR ( t ) = ˙ φ L e − N ( t ) , (65)with the e-folding number N ( t ) = (cid:82) tt L H d t (cid:48) . The fastness of the transition is characterized by thee-folding number during the transition regime: d N := N (cid:12)(cid:12)(cid:12)(cid:12) φ ft − N (cid:12)(cid:12)(cid:12)(cid:12) φ it . (66)The largeness of the transition is defined by δ := (cid:118)(cid:117)(cid:117)(cid:116) (cid:15) V (cid:12)(cid:12) φ ft (cid:15) H (cid:12)(cid:12) φ it , (67) Our δ is equivalent to 1 /h of Ref. [12]. - - - δ f N L Our formulaPHM model d N = 5 × 10 -3 d N = 0.1 d N = 0.3 FIG. 2. The f NL prediction in our formula (45) (thick lines) and PHM model (69) (dotted lines), varyingthe fastness d N (66) (blue for d N = 5 × − , orange for d N = 0 .
1, and green for d N = 0 .
3) and the largeness δ (67) of the transition. It is also checked that our formula is well consistent with the full δN computationthough we do not explicitly plot their results because they cannot be distinguished from our formula’s resultsin this figure. with (cid:15) V = M (cid:18) V (cid:48) V (cid:19) , (cid:15) H = − ˙ HH = ˙ φ M H . (68)PHM then showed that the non-linearity parameter computed in the full in-in approach is wellapproximated by the same formula (60) for the sharp transition but with the effective δ parameteras f NL (cid:39)
52 1(1 + δ eff ) , δ eff = (cid:34)(cid:18) d N . (cid:19) + δ (cid:35) / . (69)In Fig. 2, we compare our formula (45) with this PHM model, varying the fastness d N and thelargeness δ . We also checked that our formula is well consistent with the full δN computation. C. Quadratic slope
We also consider a quadratic slope with a smooth transition: V ( φ ) = V , for φ < φ c ,V − m ( φ − φ c ) , for φ ≥ φ c , (70)with a constant Hubble parameter 3 M H (cid:39) V . The analytic solution for the background reads φ ( t ) = φ L + ˙ φ L H (cid:16) − e − H ( t − t L ) (cid:17) , for φ < φ c ,φ c + ˙ φ L C + − C − e − H ( t c − t L ) (cid:16) e C + ( t − t c ) − e C − ( t − t c ) (cid:17) , for φ ≥ φ c , (71)4where C ± = − H ± √ H + 4 m , t c = t L − H log (cid:20) − H ˙ φ L ( φ c − φ L ) (cid:21) . (72)The shifted trajectory is given by¯ φ ( t ) = φ L + δφ L + ˙ φ L H (cid:16) − e − H ( t − t L ) (cid:17) , for ¯ φ < φ c ,φ c + ˙ φ L C + − C − e − H (¯ t c − t L ) (cid:16) e C + ( t − ¯ t c ) − e C − ( t − ¯ t c ) (cid:17) , for ¯ φ ≥ φ c , (73)with ¯ t c = t L − H log (cid:20) − H ˙ φ L ( φ c − φ L − δφ L ) (cid:21) . (74)One finds the time shift ξ as ξ ( t ) (cid:39) − δφ L ˙ φ L e H ( t − t L ) , for t < t c ,δφ L ˙ φ c C + − C − e ( C + − C − )( t − t c ) C − − C + e ( C + − C − )( t − t c ) , for t ≥ t c , (75)at linear order in δφ L . With a replacement δφ L → δφ S , it again shows that the short-wavelengthcurvature perturbation converges to the constant P ζ ( k S ) → C − C H ˙ φ P δφ ( k S ) , (76)implying the power of n = 2. Z c is also obtained as Z c (cid:39) H m , (77)at the leading order in the slow-roll approximation m (cid:28) H . Therefore our formula reads f NL → m H . (78)One finds that the non-linearity is suppressed by the factor m / H in the slow-roll limit m (cid:28) H in this case, compared to the sharp transition case. Note that η V = − m / H in our case and thisresult is consistent with the formula (3.36) derived in Ref. [26] in the slow-roll limit | η V | (cid:28) V ( φ ) = V , for φ < φ c ,V (cid:18) − √ (cid:15) φ − φ c M Pl + η φ − φ c ) M (cid:19) , for φ ≥ φ c , (79)with constant parameters (cid:15) and η . Though we do not explicitly show its tedious expression, onefinds the analytic solution even for this potential. In the deep SR limit, the final result is obtainedas f NL → − s )(9 − δη − ( s − δη ) )2(3 + s )(3 − s + 2 δη ) , s = (cid:112) − η . (80)This result is consistent with the formula (2.55) in Ref. [11] derived with use of the δN formula upto O ( η ) corrections. Their difference mainly comes from the fact that they use the approximatedsolution (2.49) for φ while we employ the exact solution. Our result consistently reproduces Eq. (60)in the linear slope limit η → δ → ∞ .5 V. CONCLUSIONS AND DISCUSSION
In this paper, we have revisited the soft limit of the correlation functions of primordial curvatureperturbations in non-attractor inflation models because there are some discrepancies among theliterature. A non-attractor (USR) phase needs to be followed by an attractor (SR) phase because,otherwise, USR inflation cannot solve the horizon and flatness problems in general and cannotfit the observed density perturbations. Then, the question is whether Maldacena’s consistencyrelation holds true even for a long wavelength mode exiting the horizon during an USR phase, ornot. Some explicit computations [11, 12] show that it can violate, while the exact USR inflation,a representative of non-attractor models, can be understood by a generalized consistency relation(only during the USR phase though) [13–15].In order to address this question, we have extended the previous formula by including themomentum dependence of the inflaton, which is necessary to deal with a transition phase from anUSR one to a SR one adequately. It is true that single degree of freedom (of phase space) is enoughto describe the dynamics of an USR or a SR phase respectively, but two degrees of freedom arenecessary to describe the transition (and its later dynamics) properly. Then, we have obtained thegeneralized single-field soft theorem (22), which can deal with the effect of a transition phase aswell as an USR phase and/or a SR phase itself. This new formula clearly shows that Maldacena’sconsistency relation can be violated in general for long-wavelength perturbations exiting the horizonduring a non-attractor (USR) phase.Based on this new formula, we also discussed some examples of non-attractor (USR) inflationmodels followed by attractor (SR) phases with several types of transitions. Our formula successfullyreproduces the results based on the δN formula and also those obtained in the existing literature.Let us comment on the so-called local observer effect . The essence of this program is that thecomoving-coordinate correlators are contaminated by the long-wavelength metric perturbations, sothat the correlators measured by the physical time and length would be helpful to be compared withreal observational data. After the universe converges to an attractor behavior, the only modulationby long-wavelength perturbations is a scale shift due to the fluctuated scale factor ¯ a = a e ζ L asdiscussed at the beginning of Sec. II. The bispectrum is then corrected by − (1 − n s ) P ζ ( k L ) P ζ ( k S ) forlocal observers. Therefore, in case of an attractor (SR) single-field inflation where the (comoving)bispectrum is given by Maldacena’s consistency relation (1), this local observer effect exactlycancels the squeezed limit of the (physical) bispectrum except the secondary effects. It can be alsounderstood in terms of the backward e-foldings [19]. As the end of inflation gives a fixed physicalmeasure via the Hubble scale H − , the backward e-folding number from the end of inflation labelsthe physical scale up to the time variation of H . In the attractor and single-field case, an equalbackward e-folds hypersurface obviously defines one phase-space point. Thus there is no differencein each physically rescaled local patch in this case.In the USR limit, we indeed see that the long-wavelength perturbation can be renormalizedinto the local spacetime coordinate consistently with the literature [13–15]. In fact, the solution ξ = ζ L /H in the constant Hubble approximation kills the ( n s −
1) terms in our formula (20)and then the bispectrum evaluated at the time t − ξ in order to fix the physical time ¯ t vanishesduring the USR phase. However it does not necessarily mean the disappearance of the physicalsqueezed bispectrum after inflation. If the non-attractor phase is followed by an attractor universeas a standard scenario, the local oberver effect on the squeezed bispectrum is again given by − (1 − n s ) P ζ ( k L ) P ζ ( k S ), but it cancels only the first term in our formula (22) in fact. While the lastterm represents the transition effect as discussed in the main part of the paper, it is interesting tonote that the ( n s −
1) term also remains (though n s − → ξ . It is understood as follows. During the USR phase, the physical time ¯ t correspondswith the uniform- ˙ φ (and thus uniform-density) slice, but the uniform-density slice is given by the6comoving time t after the transition to the attractor phase. Then the time difference betweenthem, ξ , yields extra expansion, i.e., modulation in physical scale. In fact an equal backwarde-folds hypersurface from the end of inflation differently defines both φ and ˙ φ , depending on thephase-space trajectory, so that local observers differently see the local physics in each patch. Thatis why the ( n s −
1) term arises rather if the local observer effect is taken into account.One may wonder why the transition from an USR phase to a SR phase is important for the laterbehavior of long-wavelength perturbations, while the transition from a SR phase to an oscillatingphase after inflation is not. In order to answer this question, it should be noticed that, during a non-attractor (USR) phase, its dynamics is completely controlled solely by the momentum (velocity) ofan inflaton but its phase-space trajectory does not converge to a single trajectory because it stilldepends on an initial field value. This is the key reason why the transition from an USR phaseto a SR one can imprint the observed squeezed bispectrum, that is, local physics. On the otherhand, during an attractor (SR) phase, its phase-space trajectory effectively converges to a singletrajectory and hence the dynamics remains adiabatic in a later transition phase unless some chaoticbehavior exaggerating (extremely) small difference among (converging) trajectories is present atthe transition. Thus, such transition phase does not give important imprints.
ACKNOWLEDGMENTS
We are grateful to Yi-Fu Cai, Xingang Chen, Hayato Motohashi, Gonzalo A. Palma, SpyrosSypsas, and Dong-Gang Wang for helpful discussions T. S. was supported by the MEXT Grant-in-Aid for Scientific Research on Innovative Areas No. 17H06359, and No. 19K03864. Y. T. issupported by JSPS KAKENHI Grants No. JP18J01992 and No. JP19K14707. M. Y. is supportedin part by JSPS Grant-in-Aid for Scientific Research Numbers 18K18764 and JSPS Bilateral OpenPartnership Joint Research Projects.
Appendix A: Uniform- ˙ φ gauge In the main body, we start from the comoving gauge φ ( t, x ) = φ ( t ). The transformed field valueis then automatically fixed to the background one as ¯ φ (¯ t ) = φ ( t ), so that the bispectrum correctionis expressed by P ζ ’s dependence on the remained DoF, ˙ φ . However φ and ˙ φ are originally equal,independent DoF and it should be possible to fix the momentum as ¯ φ (cid:48) (¯ t ) = ˙ φ and express thebispectrum correction by the φ -dependence of the power spectrum. We clarify it in this appendix.Let us start without specifying the spacetime gauge first:d s = − e δ N L d t + a ( t )e ψ L (d x + N L d t ) , φ ( t, x ) = φ ( t ) + δφ L ( t ) . (A1)In the long-wavelength limit, the linear constraint reads δ N L ( t ) = − (cid:15) H Hδu L ( t ) + ˙ ψ L ( t ) H , N L ( t, x ) = 13 (cid:15) H x ˙ ζ L , (A2)with the comoving curvature perturbation ζ L = ψ L + Hδu L and the velocity perturbation δu L = − δφ L / ˙ φ . We then look for a coordinate transformation t → ¯ t = t + ξ ( t ) , x → ¯ x = e β ( t ) / x , φ ( t ) → ¯ φ (¯ t ) = φ ( t ) = φ ( t ) + δφ L ( t ) , (A3)such that the transformed momentum coincides with the background one as¯ φ (cid:48) (¯ t ) = (cid:16) ξ (cid:17) − (cid:16) ˙ φ + δ ˙ φ L (cid:17) = ˙ φ , (A4)7and ¯ φ (¯ t ) satisfies the “background-like” EoM ∂ ¯ φ∂ ¯ t + 3 ¯ H ∂ ¯ φ∂ ¯ t + ∂∂ ¯ φ V ( ¯ φ ) = 0 , M ¯ H = 12 (cid:18) ∂ ¯ φ∂ ¯ t (cid:19) + V ( ¯ φ ) , (A5)in the “background-like” spacetimed s = − d¯ t + ¯ a (¯ t ) d¯ x , ¯ a (¯ t ) = a ( t )e α ( t ) . (A6)First the vanishing metric perturbation condition (A6) fixes the transformation parameters ξ , α , and β as ˙ ξ = δ N L , α + 13 β = ψ L , ˙ β = (cid:15) H ˙ ζ L . (A7)The uniform- ˙ φ condition (A4) then reads δ ˙ φ L = (cid:15) H Hδφ L + ˙ φ H ˙ ψ L , ⇔ ψ L = ζ L + 1 Hη ˙ ζ L , (A8)with a slow-roll parameter η = ¨ φ / ( H ˙ φ ). As the right-hand side of the second equation isdeterministic gauge-independently, it can be understood as a gauge condition onto the spatialcurvature ψ L . Making use of EoM for ζ L in the long-wavelength limit (14),dd t (cid:16) (cid:15) H a ˙ ζ L (cid:17) = 0 , (A9)one can check that this gauge condition is in fact consistent with the “background-like” EoM (A5).Therefore, if one starts from the time slice satisfying the gauge condition (A8) (instead of thecomoving slice), one can indeed fix the momentum as ¯ φ (cid:48) (¯ t ) = ˙ φ (instead of the field value itself).In this gauge, the background trajectory is characterized by (cid:0) ¯ φ (¯ t ) , ¯ a (¯ t ) (cid:1) . Thus the short-wavelength curvature perturbation is equivalent to the solution on this modulated trajectory,similarly to Eq. (17), as ζ S ( t, x ) = ζ (cid:2) ¯ t, ¯ x | ¯ φ (¯ t ) , ¯ a (¯ t ) (cid:3) . (A10)At leading order in ζ L , it can be explicitly written as ζ S ( t, x ) (cid:39) ζ (cid:34) t + ξ , (cid:18) β (cid:19) x (cid:12)(cid:12)(cid:12)(cid:12)(cid:12) φ ( t + ξ ) − (cid:32) ξ − ˙ ζ L H η (cid:33) ˙ φ , (1 + α − Hξ ) a ( t + ξ ) (cid:35) . (A11)Neglecting the time dependence of (1 + α − Hξ ) again, it reads ζ S ( t, x ) (cid:39) ζ (cid:34) t + ξ , (cid:32) ζ L + ˙ ζ L Hη − Hξ (cid:33) x (cid:12)(cid:12)(cid:12)(cid:12)(cid:12) φ ( t + ξ ) − (cid:32) ξ − ˙ ζ L H η (cid:33) ˙ φ , a ( t + ξ ) (cid:35) . (A12)Hence the Fourier-space two-point function is given by (cid:104) ζ S ( k ) ζ S ( k ) (cid:105) = (cid:104) ζ ( k ) ζ ( k ) (cid:105) + ξ ( k L ) ˙ P ζ [ t, k S | φ ] − (cid:32) ζ L ( k L ) + ˙ ζ L ( k L ) Hη − Hξ ( k L ) (cid:33) ( n s − P ζ [ t, k S | φ ] − (cid:32) ξ ( k L ) − ˙ ζ L ( k L ) H η (cid:33) ˙ φ ∂ φ P ζ [ t, k S | φ ] φ = φ . (A13)8In terms of the bispectrum, one obtainslim k → B ζ ( t ; k , k , k ) = (1 − n s ) (cid:18) P ζ ( k L ) + 12 Hη ˙ P ζ ( k L ) (cid:19) P ζ ( k S ) − (cid:18)(cid:90) t η + η η ∂ t (cid:48) P ζ ( t, t (cid:48) ; k L ) H d t (cid:48) (cid:19)(cid:104) ˙ P ζ ( k S ) + H ( n s − P ζ ( k S ) (cid:105) + (cid:18) H η ˙ P ζ ( k L ) + (cid:90) t η + η η ∂ t (cid:48) P ζ ( t, t (cid:48) ; k L ) H d t (cid:48) (cid:19) ˙ φ ∂ φ P ζ [ t, k S | φ ] φ = φ , (A14)with use of the equation for ξ in this gauge,˙ ξ = − η + η η ˙ ζ L H , (A15)with η = ˙ η / ( Hη ). In the attractor limit, the conservation of ζ L again leads to Maldacena’sconsistency relation (1). We also note that, in the exact USR limit, n s − → , η → − , η → ∂ φ P ζ → P ζ →
0, the term of φ -dependence remains aslim k → B ζ ( k , k , k ) → (cid:18)(cid:90) t η + η η ∂ t (cid:48) P ζ ( t, t (cid:48) ; k L ) H d t (cid:48) (cid:19) ˙ φ ∂ φ P ζ [ t, k S | φ ] φ = φ , (A16)and can give a non-zero contribution.In the exact USR limit η → − ζ ’s growth, ˙ ζ = 3 Hζ , indicates that the gauge condition (A8)is nothing but the flat-slice rule ψ L = 0. As ξ = 0 in this limit, ¯ t is also understood as aflat-slice time coordinate. On the other hand, if the phase-space trajectory ( ¯ φ, ¯ φ (cid:48) ) converges tothe background one ( φ , ˙ φ ) as an attractor solution, both ¯ φ and ¯ φ (cid:48) (and thus the local energydensity ¯ ρ = ¯ φ (cid:48) + V ( ¯ φ )) simultaneously coincide with background values at the time ¯ t = t + ξ .Therefore, if inflation starts from an USR phase and then proceeds to a SR phase, ξ is a timedifference from an initial flat slice to a final uniform-density slice. It reads ζ L = Hξ according tothe δN formalism [25]. Making use of this and noting that the initial condition for ¯ φ is taken as¯ φ ( t L ) = φ ( t L ) + δφ L , ¯ φ (cid:48) ( t L ) = ˙ φ , (A17)also in this gauge similarly to the comoving case (28), one successfully reproduces our formula (45), f NL → n Z c , Z c = 1 + 3 H ˙ φ (cid:90) ∞ t c exp (cid:16)(cid:82) tt c F ˙ φ d t (cid:48) (cid:17) ˙ φ ( t ) d t , (A18)following the same procedure. [1] J. M. Maldacena, Non-Gaussian features of primordial fluctuations in single field inflationary models,JHEP , 013, arXiv:astro-ph/0210603 [astro-ph].[2] P. Creminelli and M. Zaldarriaga, Single field consistency relation for the 3-point function, JCAP ,006, arXiv:astro-ph/0407059 [astro-ph]. [3] C. Cheung, A. Fitzpatrick, J. Kaplan, and L. Senatore, On the consistency relation of the 3-pointfunction in single field inflation, JCAP , 021, arXiv:0709.0295 [hep-th].[4] L. Hui, A. Joyce, and S. S. C. Wong, Inflationary soft theorems revisited: A generalized consistencyrelation, JCAP , 060, arXiv:1811.05951 [hep-th].[5] I. Agullo and L. Parker, Non-gaussianities and the Stimulated creation of quanta in the inflationaryuniverse, Phys. Rev. D , 063526 (2011), arXiv:1010.5766 [astro-ph.CO].[6] L. Berezhiani and J. Khoury, On the Initial State and Consistency Relations, JCAP , 018,arXiv:1406.2689 [hep-th].[7] N. Tsamis and R. P. Woodard, Improved estimates of cosmological perturbations, Phys. Rev. D ,084005 (2004), arXiv:astro-ph/0307463.[8] W. H. Kinney, Horizon crossing and inflation with large eta, Phys. Rev. D , 023515 (2005), arXiv:gr-qc/0503017.[9] M. H. Namjoo, H. Firouzjahi, and M. Sasaki, Violation of non-Gaussianity consistency relation in asingle field inflationary model, EPL , 39001 (2013), arXiv:1210.3692 [astro-ph.CO].[10] J. Martin, H. Motohashi, and T. Suyama, Ultra Slow-Roll Inflation and the non-Gaussianity Consis-tency Relation, Phys. Rev. D , 023514 (2013), arXiv:1211.0083 [astro-ph.CO].[11] Y.-F. Cai, X. Chen, M. H. Namjoo, M. Sasaki, D.-G. Wang, and Z. Wang, Revisiting non-Gaussianityfrom non-attractor inflation models, JCAP , 012, arXiv:1712.09998 [astro-ph.CO].[12] S. Passaglia, W. Hu, and H. Motohashi, Primordial black holes and local non-Gaussianity in canonicalinflation, Phys. Rev. D , 043536 (2019), arXiv:1812.08243 [astro-ph.CO].[13] R. Bravo, S. Mooij, G. A. Palma, and B. Pradenas, Vanishing of local non-Gaussianity in canonicalsingle field inflation, JCAP (05), 025, arXiv:1711.05290 [astro-ph.CO].[14] R. Bravo, S. Mooij, G. A. Palma, and B. Pradenas, A generalized non-Gaussian consistency relationfor single field inflation, JCAP (05), 024, arXiv:1711.02680 [astro-ph.CO].[15] R. Bravo and G. A. Palma, Unifying attractor and non-attractor models of inflation under a single softtheorem, arXiv:2009.03369 [hep-th] (2020).[16] B. Finelli, G. Goon, E. Pajer, and L. Santoni, Soft Theorems For Shift-Symmetric Cosmologies, Phys.Rev. D , 063531 (2018), arXiv:1711.03737 [hep-th].[17] T. Tanaka and Y. Urakawa, Dominance of gauge artifact in the consistency relation for the primordialbispectrum, JCAP , 014, arXiv:1103.1251 [astro-ph.CO].[18] E. Pajer, F. Schmidt, and M. Zaldarriaga, The Observed Squeezed Limit of Cosmological Three-PointFunctions, Phys. Rev. D88 , 083502 (2013), arXiv:1305.0824 [astro-ph.CO].[19] Y. Tada and V. Vennin, Squeezed bispectrum in the δN formalism: local observer effect in field space,JCAP (02), 021, arXiv:1609.08876 [astro-ph.CO].[20] T. Suyama and M. Yamaguchi, Non-Gaussianity in the modulated reheating scenario, Phys. Rev. D77 ,023505 (2008), arXiv:0709.2545 [astro-ph].[21] T. Suyama, T. Takahashi, M. Yamaguchi, and S. Yokoyama, On Classification of Models of LargeLocal-Type Non-Gaussianity, JCAP , 030, arXiv:1009.1979 [astro-ph.CO].[22] T. Suyama, Y. Tada, and M. Yamaguchi, Local observer effect on the cosmological soft theorem, PTEP , 11 (2020), arXiv:2008.13364 [astro-ph.CO].[23] X. Chen, M.-x. Huang, and G. Shiu, The Inflationary Trispectrum for Models with Large Non-Gaussianities, Phys. Rev. D , 121301 (2006), arXiv:hep-th/0610235.[24] L. Senatore and M. Zaldarriaga, A note on the consistency condition of primordial fluctuations, J.Cosmology Astropart. Phys. , 001 (2012), arXiv:1203.6884 [astro-ph.CO].[25] D. H. Lyth, K. A. Malik, and M. Sasaki, A General proof of the conservation of the curvature pertur-bation, JCAP , 004, arXiv:astro-ph/0411220.[26] V. Atal and C. Germani, The role of non-gaussianities in Primordial Black Hole formation, Phys. DarkUniv.24