betaFeedback

WorksPhysics and Astronomy

Gravitational waves and the exit from de Sitter in Inflation without an Inflaton

Abstract

In Inflation without an Inflaton (IWI), second-order de Sitter (dS) gravitons generate primordial curvature as the Coulomb-type potential of their gradient stress. Each pair contributes (p/k)2(p/k)^2, yielding exponential enhancement over the contributing e-folds. With exact Bunch-Davies evolution, this solution satisfies all four scalar second-order Einstein equations in pure dS. The retarded transverse-traceless solution instead has a UV kernel of order unity, (k/p)2(k/p)^2 below the scalar kernel, and a one-loop correction Δh,22/Δh2≃0.04 Δh2Δ^2_{h,2}/Δ^2_h\simeq0.04\,Δ^2_h. Its source acts before internal freeze-out, but oscillatory cancellations prevent tensor enhancement. We evolve the curvature through a smooth exit into radiation domination. Conserved curvature resides on the vacuum-decay hypersurface. A global depletion clock holds each scale on zero-shear time for NwN_{\rm w} e-folds after horizon exit before local release, giving all scales equal enhancement. Energy-momentum conservation feeds the depleted vacuum energy into a non-vacuum sector, modeled as radiation. Deep in radiation domination, Δζ2=cwΔh4eNw/20Δ^2_ζ=c_wΔ_h^4e^{N_{\rm w}}/20, with clock-dependent cwc_w, ns−1=−4εn_s-1=-4ε, and nT=−2εn_T=-2ε. The spectrum is adiabatic and Gaussian; super-horizon ζζ is conserved to better than 10−410^{-4}, and changing the adiabatic exit fluid alters it by less than 0.2%0.2\%. A direct radiation exit requires 31≲Nw≲5031\lesssim N_{\rm w}\lesssim50, or 3×10−6≲r≲0.0363\times10^{-6}\lesssim r\lesssim0.036. At interferometer frequencies, ΩGWh2≃2×10−21Ω_{\rm GW}h^2\simeq2\times10^{-21}–3×10−173\times10^{-17}, accessible to BBO for r≳0.01r\gtrsim0.01 and ultimate DECIGO for r≳10−5r\gtrsim10^{-5}; CMB B-mode searches probe r≳10−3r\gtrsim10^{-3}. Above kende−Nwk_{\rm end}e^{-N_{\rm w}}, the scalar spectrum steepens to k−1k^{-1}, suppressing μμ-distortions for sharp release when Nw≳45N_{\rm w}\gtrsim45.

Introduction

In standard inflation, a scalar field seeds the curvature perturbation through its quantum fluctuations and ends inflation through its classical roll, while the tensor modes are the vacuum fluctuations of the graviton. The Inflation without an Inflaton (IWI) framework, recently developed in Refs. [1–4], removes the scalar field. De Sitter (dS) space is described as a coherent state of gravitons [5, 6], the vacuum energy ends through the quantum break of that state, and the curvature perturbation is produced by the gravitons themselves at second order. Hence, there is no first-order scalar source. For tensor perturbations in a dS background, we consider the Bunch-Davies state of the graviton, with ⟨χij⟩=0\langle\chi_{ij}\rangle=0 and ⟨χijχkl⟩≠0\langle\chi_{ij}\chi_{kl}\rangle\ne0, and the curvature perturbation is the second-order response of the metric to it, δ(1)Gνμ[g(2)]=−δ(2)Gνμ[χ,χ]\delta^{(1)}G^\mu_{\nu}[g^{(2)}]=-\delta^{(2)}G^\mu_{\nu}[\chi,\chi], where g(2)g^{(2)} is the second-order metric. At this order, the Einstein equations fix the scalar potential through a constraint, and the potential is the Coulomb-type potential of the gravitons’ gradient stress, ϕ∼∇−2(∂χ ∂χ)\phi\sim\nabla^{-2}(\partial\chi\,\partial\chi). A pair of gravitons with momenta p1,p2\mathbf{p}_1,\mathbf{p}_2 and total momentum k\mathbf{k} contributes an amount that grows as (p/k)2(p/k)^2, once both modes have frozen. Summing over pairs up to a maximum ratio xmax⁡=eNobsx_{\max}=e^{N_{\mathrm{obs}}} gives As∝Δh4eNobsA_s\propto\Delta_h^4e^{N_{\mathrm{obs}}} [2], where NobsN_{\mathrm{obs}} is the number of observed e-folds and Δh2\Delta_h^2 is the tensor power spectrum. Ref. [2] takes the same ratio xmax⁡x_{\max} for every scale, and interprets the fixed ratio as a nonlinear coupling between modes, either between phases of the graviton condensate or through interactions between particles and gravitons. A slow drift of the Hubble rate, which can be tied to the depletion of the coherent state [4, 6, 7, 8], gives the small red tilt ns−1=−4ϵn_s-1=-4\epsilon [4].

If a pair of gravitons can produce an enhanced scalar, the same second-order equations might also produce an enhanced second-order tensor, with a tensor-to-scalar ratio rr of order unity that would rule the scenario out. We find that this does not happen. The TT sector obeys a wave equation rather than an elliptic scalar constraint, and the oscillatory Bunch-Davies evolution suppresses the retarded tensor response without requiring the source to vanish before internal freeze-out, leaving a small second-order correction that is finite in the ultraviolet.

We then study what happens to the scalar when dS ends. The second-order response is computed in pure GR+Λ\Lambda, whereas the Hubble rate drifts and the vacuum energy eventually decays. Such a vacuum energy cannot be conserved by itself, and the Bianchi identity forces it to feed a non-vacuum sector. In pure dS there is no matter whose density or velocity could define the usual conserved curvature ζ\zeta, and what IWI computes is the curvature of the zero-shear slices. After the exit, the conserved ζ\zeta of the radiation era is the curvature of the hypersurface on which the vacuum energy decays [9, 10], so the clock that ends de Sitter decides what survives (§ 1.3 C). In standard inflation, this clock is the inflaton, and ζ\zeta is the number of e-folds each patch accumulates before the field reaches its final value [11, 12, 13].

We show that in IWI the enhanced scalar survives, together with the observed tilt, if the vacuum decays on a global clock that keeps each comoving scale on zero-shear time for a fixed number of e-folds after the scale leaves the horizon. After that, the clocks keep their readings and tick with local proper time. Since the window is the same for every scale, the spectrum is scale invariant, and the drift of the Hubble rate across the window supplies the tilt. Such a clock is necessarily non-local; it is carried by the depleting vacuum energy, and we model the sector that receives the energy as a passive sink. For adiabatic sinks, we find that a change in the equation of state moves the result by less than 0.2 per cent, and the local entropy and anisotropic responses we have tested rescale the amplitude, but do not produce the window. Any microscopic completion has to produce the window through the response of the vacuum and the sector it feeds taken together.

We also clarify the relation with Refs. [1, 2]. The solution ψ2=0\psi_2=0, ϕ2=Fχ/4\phi_2=\mathcal{F}_\chi/4 of Ref. [1] is the late-time limit of the exact Bunch-Davies solution, which satisfies all four scalar Einstein equations in pure GR+Λ\Lambda (Appendix B). The second-order equations of Ref. [1] in pure GR+Λ\Lambda are consistent without additional matter. An apparent over-constraint, which would make a fluid term necessary for consistency [2], arises only when time derivatives are set to zero before the late-time limit is taken. The fluid of Ref. [1], which is the averaged stress of the sub-horizon gravitons, is therefore not needed for consistency. In the present construction, the drift and the exit come instead from the depleting vacuum and the sector it feeds, and the physical curvature is that of the surface on which the vacuum decays. Our Einstein tensor reproduces the source kernels of Ref. [2] and, to 2 per cent, its amplitude (Sec. II). The window used here is a dynamical realisation of the fixed-ratio prescription of Ref. [2], with NwN_w in place of NobsN_{\rm obs}, and it has to respect the end of dS (§ 1.3 C).

In section II we present the setup of the second-order calculation. Section III studies the second-order tensors, while section IV describes the exit, the role of the depletion sector and the evolution of the curvature through the transition. In section V we discuss the observational consequences, and we conclude in section VI. The technical material is collected in the Appendixes.

The graviton-induced curvature

Second-order metric and normalisation

We write the metric to second order as

ds2=a2(η){−(1+ψ)dη2+2∂iB dxidη+[(1−ϕ)δij+2∂i∂jE+χij+hij(2)]dxidxj}.(1)ds^2=a^2(\eta)\left\{-(1+\psi)d\eta^2+2\partial_iB\,dx^i d\eta+\left[(1-\phi)\delta_{ij}+2\partial_i\partial_jE+\chi_{ij}+h_{ij}^{(2)}\right]dx^i dx^j\right\}. \tag*{(1)}

Here, χij\chi_{ij} is the first-order TT graviton, hij(2)h_{ij}^{(2)} is the second-order TT tensor, and ϕ,ψ,B,E\phi,\psi,B,E are second-order scalars. There are no first-order scalars. In the Poisson gauge, B=E=0B=E=0, the IWI potentials ϕIWI=ϕ/2\phi_{\rm IWI}=\phi/2.

In de Sitter (dS), the scale factor reads a(η)=−1/(Hη)a(\eta)=-1/(H\eta), where HH is the energy scale of inflation. The gravitons are the Bunch-Davies vacuum fluctuations, and every second-order field below is a response to them. More precisely, the tensor modes are χij∝eijT(pη)\chi_{ij}\propto e_{ij}\mathcal{T}(p\eta) with T(x)=(1+ix)e−ix\mathcal{T}(x)=(1+ix)e^{-ix}, normalised so that T→1\mathcal{T}\to1 outside the horizon, and the tensor power per logarithmic interval is Δh2=(16/π)(H/mPl)2\Delta_h^2=(16/\pi)(H/m_{\rm Pl})^2, with mPlm_{\rm Pl} the Planck mass. We use the canonical normalisation ⟨χijχij⟩′=Ph\langle\chi_{ij}\chi_{ij}\rangle'=P_h, with polarisation tensors satisfying e:e∗=2e:e^*=2 and a power Ph/4P_h/4 per helicity. Refs. [1, 2, 3] use Ph/2P_h/2 per helicity, which multiplies every scalar power spectrum by a factor of 4; hence, at fixed rr their window NwN_w, introduced below, is shorter by ln⁡4\ln4.

We compute the mixed Einstein tensor G νμG^\mu_{\ \nu} exactly to second order, with a first-order field made of two plane waves of generic momenta and polarisations,

χij=Eij(1)f1(η)eip1⋅x+Eij(2)f2(η)eip2⋅x.(2)\chi_{ij}=E_{ij}^{(1)}f_1(\eta)e^{i\mathbf{p}_1\cdot\mathbf{x}}+E_{ij}^{(2)}f_2(\eta)e^{i\mathbf{p}_2\cdot\mathbf{x}}. \tag*{(2)}

where the pa\mathbf{p}_a are the internal momenta, Eij(a)=eijλa(p^a)E_{ij}^{(a)} = e_{ij}^{\lambda_a}(\hat{\mathbf{p}}_a) are the TT tensors encoding the polarisation and amplitude of each helicity λa\lambda_a, while fa(η)≡T(paη)=(1+ipaη)e−ipaηf_a(\eta) \equiv\mathcal{T}(p_a\eta) = (1 + ip_a\eta)e^{-ip_a\eta} are the Bunch-Davies mode functions, with pa=∣pa∣p_a = \lvert\mathbf{p}_a\rvert and a=1,2a = 1, 2.

The second-order fields sit at k=p1+p2\mathbf{k} = \mathbf{p}_1+\mathbf{p}_2, and we read their equations off the bilinear cross term. We denote as QμνQ^\mu{}_\nu the graviton-bilinear part of a2Gμνa^2G^\mu{}_\nu. The result reproduces the second-order source kernels of Ref. [2] to 2×10−142\times10^{-14}, so the two describe the same second-order system. Moreover, its tensor sector reproduces Maldacena’s graviton bispectrum exactly, in shape, polarisation structure and amplitude [14]; its TT source agrees with that of Ref. [15] to 6×10−156\times10^{-15}; and under xμ→xμ+ξμx^\mu\to x^\mu+\xi^\mu the full four-scalar operator changes by exactly −LξGˉμν-\mathcal{L}_{\xi}\bar{G}^\mu{}_\nu. The second-order equations are listed in Appendix A.

The zero-shear potential and its enhancement

In vacuum, the scalar sector is algebraic at every instant. Combining the 0000 and 0i0i equations gives a comoving Poisson equation, and the traceless part of the ijij equation gives the anisotropic-stress relation,

k2ϕ=−Q00+3iHk^⋅Q0/k,(3)k^2\phi=-Q^0{}_0+3i\mathcal{H}\hat{\mathbf{k}}\cdot\mathbf{Q}^0/k, \tag*{(3)}
ϕ−ψ=T≡3kikjQijk4−Qllk2(B=E=0).(4)\phi-\psi=T\equiv\frac{3k_i k_j Q^i{}_j}{k^4}-\frac{Q^l{}_l}{k^2}\quad(B=E=0). \tag*{(4)}

These constraints determine the inhomogeneous scalar response at k≠0k\ne0. The exactly homogeneous mode is governed instead by the spatially averaged Einstein equations, as discussed in Appendix G.

At late times ψ→0\psi\to0 and ϕ\phi tends to Fχ/4\mathcal{F}_{\chi}/4, where Fχ\mathcal{F}_{\chi} is the inverse-Laplacian functional of the frozen gravitons given in Ref. [1]. The late-time limit is thus the solution ψ2=0\psi_2=0, ϕ2=Fχ/4\phi_2=\mathcal{F}_{\chi}/4 identified there. The exact solution satisfies all four scalar Einstein equations (see Appendix B). Once the internal modes have frozen, we get Q∼p2χχQ\sim p^2\chi\chi and the inverse Laplacians give a kernel ∝(p/k)2\propto(p/k)^2, so the curvature of the zero-shear slices,

ζP=−12ϕ−112χijχij,(5)\zeta_{\mathrm{P}}=-\frac{1}{2}\phi-\frac{1}{12}\chi_{ij}\chi_{ij}, \tag*{(5)}

grows as (p/k)2(p/k)^2 per pair 1. In pure dS, ζP\zeta_{\mathrm{P}} is a well-defined geometric quantity, but it is not by itself the observable, since there is no matter whose density could define uniform-density slices (see the discussion in section IV C).

Writing a second-order field XX as

X(k)=∑λ1λ2∫dp1 KXhλ1(p1)hλ2(k−p1),(6)X(\mathbf{k})=\sum_{\lambda_1\lambda_2}\int\mathrm{d}\mathbf{p}_1\,K_X h_{\lambda_1}(\mathbf{p}_1)h_{\lambda_2}(\mathbf{k}-\mathbf{p}_1), \tag*{(6)}

with hλh_\lambda the helicity amplitudes and KXK_X the symmetrised pair kernel, its power spectrum is

ΔX2(k)=Δh464∫0∞dx gX(x),\Delta_X^2(k)=\frac{\Delta_h^4}{64}\int_0^\infty\mathrm{d}x\,g_X(x),
gX=∫∣1−x∣1+xdy ∑λ1λ2∣KX∣2x2y2,(7)g_X=\int_{\lvert1-x\rvert}^{1+x}\mathrm{d}y\,\frac{\sum_{\lambda_1\lambda_2}\lvert K_X\rvert^2}{x^2y^2}, \tag*{(7)}

where x=p1/kx=p_1/k, y=p2/k=∣k−p1∣/ky=p_2/k=\lvert\mathbf{k}-\mathbf{p}_1\rvert/k are dimensionless variables, and d3p=2πk3y dx dy\mathrm{d}^3p=2\pi k^3y\,\mathrm{d}x\,\mathrm{d}y. The normalisation of (7) follows from Wick’s theorem with two contractions. For the zero-shear curvature ζ\zeta, we obtain gζ→16/5g_{\zeta}\to16/5 at large xx (see Figure 1). Every pair contributes equally, and

Figure 1. We plot gX(x)g_X(x), which is the helicity-summed, yy-integrated squared kernel of Eq. (7), as a function x=p1/kx=p_1/k. For the zero-shear curvature ζP\zeta_P, gζg_\zeta tends to a constant (solid blue line), so its power grows with the number of contributing graviton pairs. The constant tensor mode SFF/k2S^{FF}/k^2 behaves the same way, so gTg_T tends to a constant value (dashed orange line). On the other hand, the retarded Bunch-Davies tensor decays as gT→4915x−4g_T\to\frac{49}{15}x^{-4} (solid green line), and its loop converges.

Δζ2=120Δh4xmax⁡,(8)\Delta_{\zeta}^2=\frac{1}{20}\Delta_h^4 x_{\max}, \tag*{(8)}

where xmax⁡x_{\max} is the largest xx that contributes 2. The ultraviolet dominates because at fixed time the graviton stress is a local bilinear dominated by the shortest frozen modes, so on large scales it behaves as white noise and ζP\zeta_{\mathrm{P}} is its Poisson potential. How large xmax⁡x_{\max} is, and whether it is the same for every kk, is decided by the exit from dS, as we will discuss in section IV.

Tensor modes are not enhanced

The second-order tensor equation

The TT projection of the second-order ijij equation gives

hij(2)′′+2Hhij(2)′+k2hij(2)=Sij,h_{ij}^{(2)''}+2\mathcal{H}h_{ij}^{(2)'}+k^2h_{ij}^{(2)}=S_{ij},
Sij≡−2Λij,lmQml=SijFFf1f2+Sijddf1′f2′,(9)S_{ij}\equiv-2\Lambda_{ij,lm}Q^m{}_l=S_{ij}^{FF}f_1f_2+S_{ij}^{dd}f'_1f'_2, \tag*{(9)}

with Λij,lm\Lambda_{ij,lm} the TT projector, which removes exactly the trace and longitudinal parts of QijQ_{ij} that carry the ∇−2\nabla^{-2} of (4). The tensors SFFS^{FF} and SddS^{dd} are constant, built from momenta and polarisations alone, and the source does not contain H\mathcal{H}-dependent terms, with all the time dependence of the source in the two mode-function products. The f1′f2′f'_1f'_2 term, Sdd=Λ(E(1)E(2)+E(2)E(1))S^{dd}=\Lambda(E^{(1)}E^{(2)}+E^{(2)}E^{(1)}), is a pure field redefinition, since in exponential variables, gij=a2(eγ)ijg_{ij}=a^2(e^\gamma)_{ij}, the source is purely gradient, as in Maldacena’s cubic action [14].

The causal solution

The retarded Green function of the operator in Eq. (9) is

Gk(η,η′)=T(kη′)T∗(kη)−T(kη)T∗(kη′)2ik3ηη′(10)G_k(\eta,\eta')=\frac{\mathcal{T}(k\eta')\mathcal{T}^{*}(k\eta)-\mathcal{T}(k\eta)\mathcal{T}^{*}(k\eta')}{2ik^3\eta\eta'} \tag*{(10)}

For Bunch-Davies gravitons, fa(η)≡T(paη)f_a(\eta)\equiv\mathcal{T}(p_a\eta), and the solution of Eq. (9) that vanishes in the far past, at late time, is

h(2)(k,0)=SFFJFF+SddJdd,h^{(2)}(\mathbf{k},0)=S^{FF}J_{FF}+S^{dd}J_{dd},
JFF=−P2+2p1p2−k2(P2−k2)2,Jdd=2p12p22(P2−k2)2,(11)J_{FF}=-\frac{P^2+2p_1p_2-k^2}{(P^2-k^2)^2},\qquad J_{dd}=\frac{2p_1^2p_2^2}{(P^2-k^2)^2}, \tag*{(11)}

with P=p1+p2P=p_1+p_2 (Appendix C). A direct numerical integration along a rotated contour reproduces these closed forms to 10−1010^{-10}. For p1,2≫kp_{1,2}\gg k, they tend to JFF→−3/(8p2)J_{FF}\to-3/(8p^2) and Jdd→1/8J_{dd}\to1/8, so the response kernel is of order one, a factor (k/p)2(k/p)^2 below the zero-shear scalar.

This absence of the enhancement comes from oscillatory cancellations across the full retarded Bunch-Davies integral. Although the TT source is nonzero before internal freeze-out, the combined contributions from the subhorizon and frozen regimes yield an O(1)O(1) tensor response, so h(2)∼χχh^{(2)}\sim\chi\chi. The fact that the scalar acquires the (p/k)2(p/k)^2 enhancement follows because Eqs. (3) and (4) are constraints that hold at every instant, whereas Eq. (9) is a wave equation whose response is built up over time. To see this more clearly, suppose instead that the source were treated as frozen from the moment the external mode crosses the horizon, i.e. ∣η∣=1/k|\eta|=1/k, even though the internal modes are still oscillating at that time. The solution would then be h(2)≃(1−sin⁡1)SFF/k2h^{(2)}\simeq(1-\sin1)S^{FF}/k^2, with constant mode SFF/k2S^{FF}/k^2; both carry the same (p/k)2(p/k)^2 as the scalar that is removed by the oscillatory evolution.

Power spectrum

Inserting Eq. (11) into Eq. (7) gives gT→4915x−4g_T\to\frac{49}{15}x^{-4} (see Figure 1), so the loop converges and depends neither on the ultraviolet cutoff nor on the duration of dS. Over the hard region p1,p2>kp_1,p_2>k, we find ∫gT dx=2.33\int g_T\,dx=2.33, dominated by

x≲4x\lesssim4, and giving a second-order tensor power

Δh,22Δh2∣hard=2.3364Δh2=0.036 Δh2≲3×10−12.(12)\left.\frac{\Delta_{h,2}^2}{\Delta_h^2}\right|_{\mathrm{hard}}=\frac{2.33}{64}\Delta_h^2=0.036\,\Delta_h^2\lesssim3\times10^{-12}. \tag*{(12)}

Considering the constant mode SFF/k2S^{FF}/k^2 would instead give gT→256/15g_T\to256/15 (see Figure 1) and Δh,22=163As\Delta_{h,2}^2=\frac{16}{3}A_s, i.e. r≃5r\simeq5, and the solution with the source switched on at k∣η∣=1k|\eta|=1 is smaller by (1−sin⁡1)2≃0.025(1-\sin1)^2\simeq0.025 and gives r≃0.13r\simeq0.13, still above the bound r<0.036r<0.036 [16]. Only the full retarded Bunch-Davies evolution brings the tensor down to Eq. (12).

The soft and collinear parts of ⟨h(2)h(2)⟩\langle h^{(2)}h^{(2)}\rangle contain the usual phase-accumulation poles at P=kP=k, which are not specific to IWI. We expect them to cancel against ⟨χh(3)⟩\langle\chi h^{(3)}\rangle in the complete one-loop result, as the same poles cancel in the tree-level bispectrum once the operator ordering is consistent. We have not computed ⟨χh(3)⟩\langle\chi h^{(3)}\rangle; power counting bounds the remainder by O(Δh2ln⁡(kend/k))O(\Delta_h^2\ln(k_{\mathrm{end}}/k)).

With no first-order scalars, hij(2)h_{ij}^{(2)} is invariant under second-order gauge transformations, since second-order generators only shift ϕ,ψ,B,E\phi,\psi,B,E and the vectors. The tensors therefore do not depend on the slicing on which dS ends, and super-horizon tensors cross the transition unchanged. The tree-level tensor bispectrum is exactly that of GR, with fNL+++≃1.69f_{\mathrm{NL}}^{+++}\simeq1.69 and feq+++≃0.02f_{\mathrm{eq}}^{+++}\simeq0.02 in the helicity basis, which corresponds to a fractional non-Gaussianity fΔh/2≲10−5f\Delta_h/2\lesssim10^{-5}. Once the enhanced scalar exists, it sources tensors in the radiation era only at order As2A_s^2 [17, 18, 19, 20, 21], and tensor-induced tensors in the radiation era are of order Δ42\Delta_4^2 [15]. Both are negligible as we will discuss in § 1.4 C.

Momentum window versus temporal support

The scalar mechanism of § 1.3 restricts the pairs that count to p≲keNwp \lesssim k e^{N_w}. Imposing the same momentum window on the tensor loop changes nothing: with the retarded Bunch-Davies kernel the part of the loop beyond xmax⁡=eNwx_{\max}=e^{N_w} is a fraction 0.47xmax⁡−30.47x_{\max}^{-3} of Eq. (12), below 10−3410^{-34} for any Nw≳26N_w \gtrsim26. The momentum window is therefore irrelevant to the tensors.

In fact, the tensor result depends only on the time over which the TT source acts. The tensor is small because the internal gravitons oscillate and average out, while the external mode is already outside the horizon. A source acting from the external crossing would instead give a constant gTg_T and rr in the range 0.10.1–55 (Sec. III C). Our tensor result therefore holds provided the depletion acts as a scalar clock and adds no long-range, correlated spin-2 response that anticipates or replaces the Bunch-Davies evolution of the TT source.

In the model of § 1.3 this condition is met automatically. The clock perturbation is isotropic, δT νμ=−δρ δ νμ\delta T^\mu_{\ \nu}=-\delta\rho\,\delta^\mu_{\ \nu}, and the radiation sink is a perfect fluid whose TT stress starts at fourth order. A free-streaming sink would also carry a first-order TT stress that damps the tensor legs, as neutrinos do in the radiation era [22], but with an energy fraction below 10−210^{-2} the effect is below a per cent. A scalar anisotropic stress of the sink, on the other hand, has no TT projection and enters only the scalar sector. The scalar perturbations that the clock leaves behind source tensors only at second order in themselves, at the scalar-induced level As2A_s^2 (see § 1.4 C), and cross terms between the enhanced scalar and the gravitons correct the tensor power at relative order AsA_s. A microscopic depletion carrying long-range spin-2 correlations would be a new source of gravitational waves, and the tensor calculation would have to be redone with it included. Within the present model, no such source is present, and the gravitational waves of IWI are the first-order vacuum gravitons, with a tensor tilt of nT=−2ϵn_T=-2\epsilon, where ϵ\epsilon parametrizes the slow drift in the Hubble rate.

The exit from de Sitter

What the exit clock has to do

The vacuum energy converts into radiation on a hypersurface Σ\Sigma defined by some clock. For super-horizon modes, the junction conditions for a transition from w=−1w=-1 to w=1/3w=1/3 [9, 10] give

ζ+=ζP+δNΣ,(13)\zeta_+ = \zeta_P + \delta N_\Sigma, \tag*{(13)}

where δNΣ=H δtΣ\delta N_\Sigma=\mathcal{H}\,\delta t_\Sigma is the delay of Σ\Sigma relative to the zero-shear slice. The value ζ+\zeta_+ is the curvature of Σ\Sigma itself, and it is conserved afterwards because the radiation era starts as a single adiabatic fluid. A smooth transition computed without junction conditions gives the same result (see § 1.3 D).

We split the exit time of each patch into its proper time and an offset, ζ+=ζτ+HΔt\zeta_+=\zeta_\tau+H\Delta t. Considering the proper-time part, we find that it is local and bounded. Along the comoving geodesics of dS, the vacuum Einstein equations give exactly Rμνuμuν=−ΛR_{\mu\nu}u^\mu u^\nu=-\Lambda, so Raychaudhuri’s equation reads θ˙=−θ2/3−σijσij+Λ\dot\theta=-\theta^2/3-\sigma_{ij}\sigma^{ij}+\Lambda. Writing θ=3H+δθ\theta=3H+\delta\theta, the local number of e-folds at fixed proper time is δNV=−(6H)−1∫σ2dt\delta N_V=-(6H)^{-1}\int\sigma^2dt. For a graviton pair σ2=14γ˙ijγ˙ij\sigma^2=\frac{1}{4}\dot\gamma_{ij}\dot\gamma_{ij} and with f′(η)=p2ηe−ipηf'(\eta)=p^2\eta e^{-ip\eta}, one obtains

δNV=−12E(1):E(2)p12p22(p1+p2)4,(14)\delta N_V=-\frac{1}{2}E^{(1)}:E^{(2)}\frac{p_1^2p_2^2}{(p_1+p_2)^4}, \tag*{(14)}

which tends to a constant for p≫kp\gg k. The curvature ζτ\zeta_\tau of the constant-proper-time slice differs from this volume term by an anisotropic piece, −k2E/3-k^2E/3 in synchronous gauge, which is also local and bounded. For (x,y)=(10,10.4)(x,y)=(10,10.4) in exact dS δNV=−0.062\delta N_V=-0.062 and ζτ=−0.138\zeta_\tau=-0.138.

The whole enhancement is thus in the offset Δt\Delta t, and a clock that carries it has to follow zero-shear time. Such a clock is non-local, because any clock built from local physics ends dS on a surface whose curvature is the bounded local term. In standard inflation, the inflaton plays this role, since its fluctuation δϕ\delta\phi, generated at horizon exit and carried locally, makes different patches reach the end at different times.

Finally, the clock must also give a scale-invariant spectrum. By Eq. (8), the spectrum is set by how many pairs each scale collects, and a scale collects pairs for as long as its patches stay synchronised on zero-shear time. For the amplitude to be the same for every scale, every scale must collect pairs for the same number of e-folds, counted from its own horizon exit.

A zero-shear clock with a finite window

We have seen in § 1.3 A the requirements on the clock, i.e. a non-local, zero-shear clock with a per-scale window. Here, we take a global clock that holds each comoving scale on zero-shear time while the scale is less than NwN_w e-folds outside the horizon, ln⁡(H/k)<Nw\ln(\mathcal{H}/k)<N_w. After that, the clocks keep their readings and tick with local proper time, and dS ends when they reach a common value. Equivalently, the synchronising influence reaches a fixed physical distance of eNwe^{N_w} Hubble radii at every moment (see Figure 2), which corresponds to 101310^{13}–102210^{22} Hubble radii across the allowed range. We postulate this clock and identify it with the break of the dS coherent state as a whole [6, 7]. The depletion fixes the common reading at which dS ends, and the coherence of the state across super-horizon distances fixes the window. This is a property of the state as a whole and is distinct from the coherence of individual graviton pairs, which become classical within 9–23 e-folds of horizon exit [8].

Figure 2. We show the picture of the mechanism by plotting the comoving scale as a function of the number of e-folds NN. A comoving scale 1/k1/k (solid orange line) leaves the Hubble radius (solid black line) and remains synchronised on zero-shear time within the synchronisation reach eNw/aHe^{N_w}/aH (dashed blue line), where NwN_w is the window’s number of e-folds. The moment of horizon crossing is shown with circles, while the release happens NwN_w e-folds later and it is shown with a square. Internal modes 1/p1/p that freeze before this release are recorded in the clock offset (solid green line). Internal modes that freeze later contribute only a bounded local term (solid gray lines). The dotted gray line shows the end of the de Sitter phase.

Let ξ\xi be the time shift from the zero-shear slices to the clock slices, η→η+ξ\eta\to\eta+ \xi, so that the clock slices have curvature ζP−Hξ\zeta_{\mathrm{P}} - H\xi and δNΣ=−Hξ\delta N_{\Sigma} = -H\xi in Eq. (13). Then, the clock is defined by

synchronised: ξ=0\xi= 0; \qquad free-running: ξ′=12ψ−Hξ\xi' = \frac{1}{2}\psi- H\xi.

{#eq:15}

The free-running condition is that the clock reads proper time along the zero-shear normals, ψ~=ψ−2(ξ′+Hξ)=0\widetilde{\psi} = \psi- 2(\xi' + H\xi) = 0. We spread the release smoothly over ΔNr\Delta N_r e-folds around ln⁡(H/k)=Nw\ln(\mathcal{H}/k) = N_w. Since the vacuum energy is uniform on the clock slices, in Poisson gauge it carries δρv=ρˉv′ξ\delta\rho_v = \bar{\rho}_v'\xi and δpv=−δρv\delta p_v = -\delta\rho_v.

Considering a single pair, it builds its zero-shear offset while its internal modes freeze, at H≃p\mathcal{H} \simeq p. If the scale kk is still synchronised at that time, then the clock carries the whole offset forward. On the other hand, if it is not synchronised, the free-running clock integrates the lapse, which cancels the offset and leaves only the local term. Only pairs with p≲keNwp \lesssim k e^{N_w} therefore contribute. By the dilation symmetry of dS, the kernel depends on xx, yy and NwN_w but not on kk, so every scale receives the same enhancement. This requires each clock to keep, after the release, the offset it has acquired instead of resetting to local time.

Clock, sink and the physical curvature

The calculation of § 1.1 is pure GR+Λ\Lambda, whereas in the cosmology we want to describe the Hubble rate drifts, the vacuum depletes, and dS ends. A vacuum component Tv νμ=−ρvδνμT^\mu_{v\ \nu} = -\rho_v\delta^\mu_{\nu} satisfies ∇μTv νμ=−∂νρv\nabla_\mu T^\mu_{v\ \nu} = -\partial_\nu\rho_v identically, so if ρv\rho_v changes the Bianchi identity requires a second sector with

∇μTeff νμ=∂νρv=Jν.(15)\nabla_\mu T^\mu_{\mathrm{eff}\ \nu} = \partial_\nu\rho_v = J_\nu. \tag*{(15)}

Within GR, the minimal content of the model is therefore a depleting vacuum together with a non-vacuum sector that absorbs its energy. Because pv=−ρvp_v = -\rho_v, the transfer is not a free function, but it is fixed by the slices on which ρv\rho_v is uniform, with no momentum transfer in their frame. A different JνJ_\nu would require the depleting component to depart from pv=−ρvp_v = -\rho_v at the perturbative level, for instance by carrying momentum or anisotropic stress. The sink is a product of the depletion, analogous to the reheating products of standard inflation but produced continuously at order ϵ\epsilon. It is the non-vacuum enthalpy that any drift of HH requires, since −2H˙=ρ+p-2\dot{H} = \rho+ p, and in the background it plays the role that ϕ˙ 2\dot{\phi}^{\,2} plays in standard inflation, but, unlike the inflaton, it is not the source of the curvature perturbation. We model the sink as a passive fluid.3 The homogeneous quadratic stress of the Bunch-Davies gravitons is distinct from this depletion-generated sector. Its finite-window contribution is of order Δh2\Delta_h^2, far below the background ϵ\epsilon required by the observed tilt. The corresponding homogeneous backreaction and its renormalized interpretation are examined in Appendix G.

The clock and the sink play different roles. A slicing can be built from any background scalar XX with Xˉ′≠0\bar{X}' \ne0, and its curvature is

ζX=ζP−HδXXˉ′.(16)\zeta_X = \zeta_{\mathrm{P}} - H\frac{\delta X}{\bar{X}'} . \tag*{(16)}

For a perfect fluid with no exchange ρˉ′=−3H(ρ+p)\bar{\rho}' = -3\mathcal{H}(\rho+p), and this reduces to the familiar ζP+δρ/[3(ρ+p)]\zeta_{\mathrm{P}}+\delta\rho/[3(\rho+p)]. Pure Λ\Lambda has ρˉ′=0\bar{\rho}' = 0 and defines no clock, which is why ζP\zeta_{\mathrm{P}} alone is not observable. The depleting vacuum, on the other hand, has ρˉv′≠0\bar{\rho}_v' \ne0, so its uniform slices define one even though ρv+pv=0\rho_v+p_v=0. In our construction, these are the clock slices of Eq. (15), with δρv=ρˉv′ξ\delta\rho_v=\bar{\rho}_v'\xi. This identifies the slices on which the clock is read, but it does not give the law by which they are synchronised, released, and keep their memory, which we describe in Sec. IV B, and the post-exit curvature depends on that law. The sink cannot take over the clock’s role. After freeze-out, its uniform-density slices converge to the zero-shear ones, as do those of the total density (Appendix E), with an offset larger by a factor 2/ϵ2/\epsilon, although both decay as p/aHp/aH. Reading the curvature on the sink’s slices would therefore record every pair up to kendk_{\mathrm{end}} and give Δζ2∝kend/k\Delta_\zeta^2 \propto k_{\mathrm{end}}/k.

After the exit the matter is an ordinary fluid, and the curvature on uniform total density,

ζ=−12ϕ−112χijχij+δρ3(ρ+p)(17)\zeta=-\frac{1}{2}\phi-\frac{1}{12}\chi_{ij}\chi_{ij}+\frac{\delta\rho}{3(\rho+p)} \tag*{(17)}

is defined and conserved. This is the quantity to which AsA_s refers. It equals the curvature of the clock surface, hence ζP\zeta_P for a zero-shear exit, and it follows from the clock dynamics, the transfer, and the matching (see Appendix E). During quasi-dS, while the clock is synchronised and once the internal modes have frozen, the density term δρ/[3(ρ+p)]\delta\rho/[3(\rho+p)] in Eq. (17) reduces to the velocity term Hv=−ψ/6Hv=-\psi/6 and decays as p/aHp/aH. After a zero-shear exit, the uniform-density curvature equals ζP\zeta_P to about 10−310^{-3}, and with the window clock the post-exit value does not depend on when the exit happens, to 10−610^{-6}.

We consider radiation as a natural sink. With q≡−ρ˙v>0q\equiv-\dot{\rho}_v>0,

ρ˙r+4Hρr=q,ϵ≡−H˙H2=32ρ+pρ=2ρrρ(18)\dot{\rho}_r+4H\rho_r=q,\qquad\epsilon\equiv-\frac{\dot{H}}{H^2}=\frac{3}{2}\frac{\rho+p}{\rho}=\frac{2\rho_r}{\rho} \tag*{(18)}

so a steady drift has ρr=ϵρ/2\rho_r=\epsilon\rho/2. For a sink with energy fraction Ωe\Omega_e and equation of state wew_e the same relation reads ϵ=32Ωe(1+we)\epsilon=\frac{3}{2}\Omega_e(1+w_e). Acceleration ends at ϵ=1\epsilon=1, so ρr=ρ/2\rho_r=\rho/2, which requires the transfer to grow to q∼Hρq\sim H\rho. A steady depletion does not get there by itself. In fact, for q=gHρvq=gH\rho_v with constant g<2g<2 the background approaches a fixed point with ϵ=g/2\epsilon=g/2 for any sink with we>−1/3w_e>-1/3, and inflation continues. The exit therefore needs a change in the rate or in the state, which we impose on the clock slices over ΔN=0.3\Delta N=0.3. If the radiation is replaced by a fluid with constant ww and rest-frame sound speed cs2c_s^2, the velocity term becomes Hv=−ψ/(4+6cs2)Hv=-\psi/(4+6c_s^2), independent of ww. For adiabatic sinks (cs2=wc_s^2=w) and for cs2>wc_s^2>w the post-exit ∣ζ∣|\zeta| then changes by at most 0.16%0.16\% for all the pairs we tested, up to u=30u=30 with u≡p/(keNw)u\equiv p/(k e^{N_w}), and the same bound holds for every closure when the pair freezes before or during the release (u≤3u\leq3). For cs2<wc_s^2<w the post-release tail is numerically ill-conditioned in our formulation (see Appendix E).

We also tested a non-passive sink, adding to the radiation either a non-adiabatic pressure δpnad=λ(ρ+p)BV\delta p_{\mathrm{nad}}=\lambda(\rho+p)B_V or a scalar anisotropic stress Πe=λ(ρ+p)BK\Pi_e=\lambda(\rho+p)B_K, active while the vacuum depletes and through the exit. Here, BVB_V is the local volume bilinear of the graviton pair and BKB_K is a kinetic bilinear that vanishes once the internal modes freeze (see more details in Appendix E). We find that the non-adiabatic pressure changes the post-exit ζ\zeta by 0.011λ0.011\lambda, a local term of order χχ\chi\chi whose share of the enhanced curvature falls as x−2x^{-2} (2×10−32\times10^{-3} at x=3x=3 to 2×10−62\times10^{-6} at x=100x=100). The anisotropic stress, entering the traceless constraint through ∇−2\nabla^{-2}, carries the same (p/k)2(p/k)^2 as the gravitons’ own stress. For λ=1\lambda=1 it lowers ∣ζ∣|\zeta| by 4.84.8–4.9%4.9\% for all xx, and by 4.3%4.3\% for ϵ=0.003\epsilon=0.003. The change is linear in λ\lambda; it rescales cwc_w and moves the window by 0.1λ0.1\lambda e-folds. A persistent stress ∝χχ\propto\chi\chi, as in a purely elastic medium, would instead grow as (aH/k)2(aH/k)^2 and dominate the spectrum, which excludes a response of that form. A viscoelastic response, whose memory of the strain relaxes on a finite time, as well as a spatially non-local response lie outside these tests. None of the responses we tested selects pairs. A response that is local in space cannot distinguish p≲keNwp\lesssim k e^{N_w} from p≳keNwp\gtrsim k e^{N_w} without correlations over eNwe^{N_w} Hubble radii. A produced sector that implemented the window would carry exactly the non-locality that we assign to the clock, since it would be a realisation of the clock, with its density, entropy, stress and memory entering its dynamics. Also, a response of the sink correlated with the gravitons does not double count their stress, because QμνQ^{\mu\nu} is the Bunch-Davies fluctuation, with no occupied quanta, while the sink is made of the quanta that the depletion produces. Within the passive closure, then, the sink adds no free function to the observables: its enthalpy is fixed by the tilt (Sec. IV E), and the other responses we tested either drop out or rescale cwc_w. Clearly, these tests cover the responses listed here, not every medium the depletion could produce.

The window is distinct from the end of dS. The condition p≲keNwp\lesssim k e^{N_w} is a fixed ratio between internal and external scales, i.e. xw=eNwx_w=e^{N_w}, whereas the requirement that the internal modes freeze before the end, xend=kend/kx_{\mathrm{end}}=k_{\mathrm{end}}/k, depends on kk. Notice that a fixed ratio is the prescription of Ref. [2], which takes xmax⁡=eNobs=kend/H0x_{\max}=e^{N_{\mathrm{obs}}}=k_{\mathrm{end}}/H_0 for every kk. Our clock realises that prescription dynamically and adds the transfer to the post-exit curvature. However, xmax⁡=eNobsx_{\max}=e^{N_{\mathrm{obs}}} needs internal modes up to keNobs>kendk e^{N_{\mathrm{obs}}}>k_{\mathrm{end}} for every k>H0k>H_0, and these modes never leave the horizon before the end. With both conditions, the effective count is xeff=min⁡(eNw,kend/k)x_{\mathrm{eff}}=\min(e^{N_w},k_{\mathrm{end}}/k), and it is the same for every scale only below kb=kende−Nwk_b=k_{\mathrm{end}}e^{-N_w}, beyond which it produces the k−1k^{-1} tail (see Sec. IV E). Hence, NwN_w has to be shorter than NobsN_{\mathrm{obs}} by at least ln⁡(k/H0)\ln(k/H_0), plus half the release width (Sec. V). For the smallest scale that must be scale invariant, this implies about 0.50.5 e-folds at the pivot and 88 at k=1 Mpc−1k=1\,\mathrm{Mpc}^{-1}. For Nw=NobsN_w=N_{\mathrm{obs}} the break would sit at kb=H0k_b=H_0. In our realisation, the clock selects the pairs through its perturbation δρv(k)=ρv′ζ(k)\delta\rho_v(\mathbf{k})=\rho'_v\zeta(\mathbf{k}), an isotropic spin-0 response whose pair kernel sets the ratio ∣ζ∣/∣ζP∣|\zeta|/|\zeta_P| of Figure 3. This relation identifies the clock slices, but does not give the microscopic equation for ζ\zeta, which may involve the produced sector and the vacuum. We therefore state the selection as a requirement on any microscopic completion: the response of the coupled system of depleting vacuum and produced sector has to reproduce the fixed ratio p/k≲eNwp/k\lesssim e^{N_w}, and the density, entropy, stress and memory of the produced medium may all take part in it. The window makes the spectrum scale invariant, and the tilt comes from the drift of HH across it.

Curvature after the exit and ratio against normalized momentum

Figure 3. Left: Curvature ∣ζ∣|\zeta| after the exit for one pair (y=x+0.4y=x+0.4, positive helicities), with windows xc=eNw=10x_{\mathrm{c}}=e^{N_{\mathrm{w}}}=10, 30 and 100 released over ΔNr=2\Delta N_{\mathrm{r}}=2 e-folds. Each curve tracks the never-released zero-shear value (solid black line) up to its window x≃xcx\simeq x_{\mathrm{c}} (dotted lines) and falls steeply beyond. The gray dashed line is a clock released inside the horizon. Right: Ratio ∣ζ∣/∣ζP∣|\zeta|/|\zeta_{\mathrm{P}}| against u=x/xc=p/(keNw)u=x/x_{\mathrm{c}}=p/(k e^{N_{\mathrm{w}}}), for three release widths. Curves with different xcx_{\mathrm{c}} at the same uu agree to better than half a per cent. The dashed black line is the schematic classical-pair limit, in which only pairs frozen before the release are recorded.

Solving through the transition

We integrate the exact second-order equations for single graviton pairs from inside the horizon, at k∣η∣=5k|\eta| = 5 with Bunch-Davies modes, through the dS phase and a smooth exit, and into the radiation era. The background is quasi-dS, with a vacuum energy ρv∝e−2ϵN\rho_v \propto e^{-2\epsilon N} that decays slowly into the radiation sink of Eq. (18) with ϵ=0.0088\epsilon= 0.0088, and at the end converts completely over ΔN=0.3\Delta N = 0.3 e-folds on the clock slices. We evolve the scalar sector in Poisson gauge, Eqs. (26)–(29), with the vacuum perturbation above and the clock of Eq. (15), while the gravitons evolve on the exact background. After the exit we evaluate ζ\zeta on uniform-density slices, Eq. (17), which for radiation is ζ=−12ϕ−112χχ+16δρr/ρr\zeta= -\frac{1}{2}\phi- \frac{1}{12}\chi\chi+ \frac{1}{6}\delta\rho_r/\rho_r. The details are in Appendix D.

A clock that is never released reproduces a direct Poisson-gauge evolution to 10−1110^{-11}. A clock released deep inside the horizon approaches the proper-time value, ∣ζ∣=0.125|\zeta| = 0.125 for release at p∣η∣=546p|\eta| = 546 with x=10x = 10, compared with ∣ζ∣=0.129|\zeta| = 0.129 from a synchronous-gauge evolution on the same background. Moving the exit from 3 to 10 e-folds after the release changes ζ\zeta by less than 10−610^{-6}.

Figure 3 shows the result for the curvature ∣ζ∣|\zeta|. We see that for u≪1u \ll1 the curvature is the zero-shear one, while for u≫1u \gg1 it falls steeply, as u−3u^{-3} between u=10u = 10 and u=30u = 30 for ΔNr=2\Delta N_r = 2. At fixed uu, the ratio depends on xcx_c: at u=10u = 10 and ΔNr=2\Delta N_r = 2, for example, (x,xc)=(100,10)(x, x_c) = (100, 10) and (300,30)(300, 30) give 0.253 and 0.254. Near u∼1u \sim1–10, the kernel is enhanced, because pairs that cross the horizon during the release hand their partly formed offset to the clock. The effect shrinks as the release slows down. At u=10u = 10 the ∣ζ∣/∣ζP∣|\zeta|/|\zeta_P| is 2.8, 0.77 and 0.25 for ΔNr=0.5\Delta N_r = 0.5, 1 and 2, while at u=30u = 30 we get ∣ζ∣/∣ζP∣=0.09|\zeta|/|\zeta_P| = 0.09, 0.02 and 0.008, respectively.

Spectrum and tilt

Since the kernel is the same for every kk,

Δζ2(k)=cw20Δh4(keNw)eNw,(19)\Delta_\zeta^2(k) = \frac{c_w}{20}\Delta_h^4(k e^{N_w})e^{N_w}, \tag*{(19)}

with

cw≡516e−Nw∫0∞gζrec dx,(20)c_w \equiv\frac{5}{16}e^{-N_w}\int_0^\infty g_\zeta^{\mathrm{rec}}\,dx, \tag*{(20)}

where gζrecg_\zeta^{\mathrm{rec}} is Eq. (7) evaluated with the curvature that each pair leaves after the exit, so that gζrec→gζf2g_\zeta^{\mathrm{rec}} \to g_\zeta f^2 for x≫1x \gg1. The smallest value, cw=1.17c_w = 1.17, is obtained from the full loop over all configurations and helicities for a sharp release, in which each pair is recorded with the zero-shear value it has when the window closes. For pairs that freeze close to the release, this exceeds their late-time value. A release spread over a finite number of e-folds also records part of the offset of pairs that cross the horizon during the release, and from the positive-helicity slice of the dynamical runs we estimate cw≈7.8c_w \approx7.8, 18 and 71 for ΔNr=2\Delta N_r = 2, 1 and 0.5, respectively. A release much faster than an e-fold would also record the sub-horizon, vacuum part of the zero-shear offset, so we regard ΔNr≳1\Delta N_r \gtrsim1 as the physical range. Only the combination cweNwc_w e^{N_w} enters the amplitude, so the release profile moves the lower edge of the allowed window by only ln⁡(cw/1.17)≲4\ln(c_w/1.17) \lesssim4.

The dominant pairs have p≃(1–10) keNwp \simeq(1\text{–}10)\,k e^{N_w}, a factor that is absorbed into cwc_w. They freeze a fixed number of e-folds after kk exits, with Δh2∝H2∝p−2ϵ\Delta_h^2 \propto H^2 \propto p^{-2\epsilon}, and since the scalar carries four tensor legs, Δζ2∝H4(p)\Delta_\zeta^2 \propto H^4(p), giving the consistency relation

ns−1=2nT(21)n_s - 1 = 2n_T \tag*{(21)}

Then, the scalar and tensor tilts read

ns−1=−4ϵ,nT=−2ϵ,(22)n_s - 1 = -4\epsilon,\qquad n_T = -2\epsilon, \tag*{(22)}

respectively. The running αs=−4 dϵ/dln⁡k=O(ϵ2)\alpha_s = -4\,d\epsilon/d\ln k = \mathcal{O}(\epsilon^2) is found to be αs=+8ϵ2≃6×10−4\alpha_s = +8\epsilon^2 \simeq6 \times10^{-4} if ϵ\epsilon evolves as in Ref. [4]. The observed $n_s = 0.965 [23] fixes ϵ=0.0088\epsilon= 0.0088. Since ϵ=32Ω(1+we)\epsilon= \frac{3}{2}\Omega(1+w_e), the tilt fixes the enthalpy fraction of the sink, Ωe(1+we)≃(1−ns)/6\Omega_e(1+w_e) \simeq(1-n_s)/6, and not wew_e or cs2c_s^2 separately. For example, for radiation we obtain Ωr≃0.0044\Omega_r \simeq0.0044. The relation ns−1=2nTn_s - 1 = 2n_T also assumes that the dimensionless response does not depend on scale. Indeed, writing Δζ2=Δh4(k)I(k)\Delta_\zeta^2 = \Delta_h^4(k)\mathcal{I}(k), a scale-dependent response gives ns−1=2nT+dln⁡I/dln⁡kn_s - 1 = 2n_T + d\ln\mathcal{I}/d\ln k. The window sets the enhancement and makes it scale-independent, whereas a count running to the end for every scale would give ns−1≃−1n_s - 1 \simeq-1. Therefore, the red tilt comes entirely from the drift of HH across the window. Modes that exit less than NwN_w e-folds before the end have fewer pairs available, Δh4=kend/k\Delta_h^4 = k_{\mathrm{end}}/k, so beyond kb=kende−Nwk_b = k_{\mathrm{end}}e^{-N_w} the spectrum falls as k−1k^{-1}.

Because all patches end on the same clock, the radiation era starts with a single adiabatic mode and no isocurvature. The curvature is a sum over ∼eNw\sim e^{N_w} independent pairs, and power counting of the one-loop bispectrum gives fNL∼Nw/(Δh2e2Nw)≲10−14f_{\mathrm{NL}} \sim N_w/(\Delta_h^2 e^{2N_w}) \lesssim10^{-14} at the lower edge of the window, the same scaling as in Ref. [3].

Conservation through the radiation era

After the exit, the matter is a single radiation fluid. The frozen gravitons stay frozen until they re-enter, and then behave as radiation with a locally determined stress. The non-adiabatic pressure vanishes, so ζ\zeta on uniform density is conserved on super-horizon scales [22, 24]. Figure 4 shows this for pairs inside (u=0.1,0.3,1)(u = 0.1, 0.3, 1) and outside (u=10u=10) the window. From the exit until the external mode approaches the horizon, ζ\zeta stays constant between 1×10−61\times10^{-6} and 6×10−56\times10^{-5} while k/aH<1/30k/aH<1/30, including across the re-entry of the internal modes. The rise near k/aH∼0.1k/aH\sim0.1 is the usual (k/aH)2(k/aH)^2 gradient term of the adiabatic mode, and after that ζ\zeta sets the amplitude of the acoustic oscillations in the standard way.

Observational consequences

The allowed window

Several numbers of e-folds enter here, and they have to be kept apart. The duration of dS, NtotN_{\mathrm{tot}}, runs from the exit of the first mode to the end and is set by the lifetime of the depleting state 4. Here, NtotN_{\mathrm{tot}} only has to exceed NobsN_{\mathrm{obs}}. In Ref. [2], the observable number of e-folds is taken to be Nobs=ln⁡(kend/H0)N_{\mathrm{obs}}=\ln(k_{\mathrm{end}}/H_0), the separation between the largest observable scale and the last mode to exit. N(k)=ln⁡(kend/k)N(k)=\ln(k_{\mathrm{end}}/k), with N∗=N(k∗)N_* = N(k_*), is the time that scale kk spends outside the horizon before the end. NwN_{\mathrm{w}} is the part of that time over which kk collects pairs, and it sets the amplitude through eNwe^{N_{\mathrm{w}}}. Finally, ΔNr\Delta N_{\mathrm{r}} and ΔN=0.3\Delta N=0.3 are the widths of the release and of the exit. Only NwN_{\mathrm{w}} is a parameter of the clock. Because N(k)=ln⁡(kend/k)N(k)=\ln(k_{\mathrm{end}}/k) depends only on the comoving scales, it is fixed by HendH_{\mathrm{end}} and by what follows the end of dS, and lengthening the dS phase raises NtotN_{\mathrm{tot}}, but leaves N(k)N(k) unchanged.

We take IWI’s natural exit, directly into radiation at ρend=3M2Hend2\rho_{\mathrm{end}}=3M^2H_{\mathrm{end}}^2 with MM the reduced Planck mass. The number of e-folds between the pivot k∗=0.05 Mpc−1k_*=0.05\,\mathrm{Mpc}^{-1} and the end is then fixed by HendH_{\mathrm{end}}. Because of the drift, the Hubble rate at the end is lower than at the pivot, Hend=H∗e−N∗H_{\mathrm{end}}=H_*e^{-N_*}, and the legs that set the amplitude at the pivot freeze with H=H∗e−NwH=H_*e^{-N_{\mathrm{w}}}. For each NwN_{\mathrm{w}}, Eq. (19) with

As=2.1×10−9A_s = 2.1 \times10^{-9} then fixes H∗H_*, the tensor-to-scalar ratio r=Δh2(k∗)/Asr = \Delta_h^2(k_*)/A_s and N∗N_*. Since rr falls as NwN_w grows, the window is bounded from below by the BICEP/Keck limit r0.05<0.036r_{0.05} < 0.036 [16], where k∗=0.05 Mpc−1k_* = 0.05\,\mathrm{Mpc}^{-1}.

From above, every scale that has to be scale invariant needs its window to close before the end. The release of scale kk is complete Nw+ΔNr/2N_w + \Delta N_r/2 e-folds after it exits. With the exit placed one e-fold after the centre of the release, for ΔNr=2\Delta N_r = 2 and 1, the post-exit curvature of pairs with u=0.3u = 0.3–30 agrees with that of a late exit to 0.2%, and ∫f2du\int f^2du to 0.3%, although the pairs with u≳3u \gtrsim3 are still inside the horizon at the exit. After the release, the clock reads local proper time, so the offset it has recorded is complete and the remaining pairs do not have to freeze before the end. An exit during the release instead records more of the zero-shear offset, f(3)=2.08f(3) = 2.08 instead of 1.39 for an exit at the centre of a release with ΔNr=2\Delta N_r = 2, so the scales whose release is under way at the end form a transition of width ΔNr\Delta N_r around kbk_b before the k−1k^{-1} tail. Requiring scale invariance down to k=1 Mpc−1k = 1\,\mathrm{Mpc}^{-1} therefore means Nw+ΔNr/2≤N(1 Mpc−1)N_w + \Delta N_r/2 \leq N(1\,\mathrm{Mpc}^{-1}). For the sharp release, cw=1.17c_w = 1.17, this gives

30.5≤Nw≤50.3,2.6×10−6≤r≤0.036.(23)30.5 \leq N_w \leq50.3,\qquad2.6 \times10^{-6} \leq r \leq0.036 . \tag*{(23)}

For cw=7.8c_w = 7.8 (ΔNr=2\Delta N_r = 2) the range is 28.6≤Nw≤49.228.6 \leq N_w \leq49.2. Across the window Nobs=N∗+5.4N_{\mathrm{obs}} = N_* + 5.4 runs from 61.1 to 58.7, so NwN_w is between 8 and 31 e-folds shorter than NobsN_{\mathrm{obs}}.

These bounds hold for the direct radiation exit. Under this exit, the literal prescription of Ref. [2], Nw=NobsN_w = N_{\mathrm{obs}}, has no room, because for every k>H0k > H_0 it would need pairs that exit after the end. A history with a phase of different equation of state between the end of dS and the radiation era changes N(k)N(k), and with it the upper bound. That would be a different realization, which needs its own curvature matching, so we do not consider it here. Table I lists the results for a direct radiation exit. The reheating temperature is Trh∼(0.4T_{\mathrm{rh}} \sim(0.4–4.5)×1015 GeV4.5) \times10^{15}\,\mathrm{GeV}, and the end of dS corresponds to fend≃(1.1f_{\mathrm{end}} \simeq(1.1–12)×107 Hz12) \times10^7\,\mathrm{Hz} today.

NwN_wrrH∗ [GeV]H_* \ [\mathrm{GeV}]N∗N_*kb [Mpc−1]k_b \ [\mathrm{Mpc}^{-1}]108μ10^8\mu
30.53.6×10−23.6 \times10^{-2}4.7×10134.7 \times10^{13}55.74.2×1094.2 \times10^91.65
321.8×10−21.8 \times10^{-2}3.3×10133.3 \times10^{13}55.58.3×1088.3 \times10^81.65
354.2×10−34.2 \times10^{-3}1.6×10131.6 \times10^{13}55.22.9×1072.9 \times10^71.65
403.8×10−43.8 \times10^{-4}4.8×10124.8 \times10^{12}54.61.1×1051.1 \times10^51.65
438.8×10−58.8 \times10^{-5}2.3×10122.3 \times10^{12}54.23.7×1033.7 \times10^31.61
453.4×10−53.4 \times10^{-5}1.4×10121.4 \times10^{12}54.03.9×1023.9 \times10^21.23
487.9×10−67.9 \times10^{-6}7.0×10117.0 \times10^{11}53.6140.24
50.32.6×10−62.6 \times10^{-6}4.0×10114.0 \times10^{11}53.31.00.02

Table I. Direct radiation exit with the Hubble drift included, for cw=1.17c_w = 1.17, As=2.1×10−9A_s = 2.1 \times10^{-9} and ns=0.965n_s = 0.965. We report: the window NwN_w; the tensor-to-scalar ratio rr at the pivot k∗=0.05 Mpc−1k_* = 0.05\,\mathrm{Mpc}^{-1}; the Hubble rate H∗H_* at k∗k_*; the e-folds from the pivot to the end, N∗=ln⁡(kend/k∗)N_* = \ln(k_{\mathrm{end}}/k_*); the small-scale break kb=kende−Nwk_b = k_{\mathrm{end}}e^{-N_w}; and the chemical-potential distortion μ\mu for a sharp break. Without the break, μ=1.65×10−8\mu= 1.65 \times10^{-8} throughout.

Tensor-to-scalar ratio and consistency relation

The tensor-to-scalar rr falls by a factor e−(1−4ϵ)/2e^{-(1-4\epsilon)/2} per unit of NwN_w (Figure 5, right). B-mode programmes that target r∼10−3r \sim10^{-3} [25, 26] probe Nw≲38N_w \lesssim38 for cw=1.17c_w = 1.17, or Nw≲36N_w \lesssim36 for cw=7.8c_w = 7.8. Whatever the value of rr, the tensor tilt in IWI is nT=(ns−1)/2≃−0.018n_T = (n_s - 1)/2 \simeq-0.018, which differs from the single-field relation5 nT=−r/8n_T = -r/8.

Figure 5. Left: Scalar spectrum Δζ2(k)/As\Delta_\zeta^2(k)/A_s for Nw=30,40,45,50N_w=30,40,45,50. The spectrum is flat, with ns−1=−4ϵn_s-1=-4\epsilon, through the CMB and large-scale-structure range (gray band), and steepens to k−1k^{-1} beyond kbk_b. The orange band marks the μ\mu-distortion scales. Right: Tensor-to-scalar ratio rr against NwN_w for cw=1.17c_w=1.17 (solid line) and cw=7.8c_w=7.8 (dashed line). The upper shaded region is excluded because r>0.036r>0.036, while the right-hand one is excluded because the window would reach the end of de Sitter for k<1 Mpc−1k<1\,\mathrm{Mpc}^{-1}. The dotted line is the LiteBIRD/CMB-S4 reach. Circles are the scrambling-time windows of Eq. (24). The top axis gives μ\mu-distortion for cw=1.17c_w=1.17.

A natural candidate for the window is the dS scrambling time, t∗=H−1ln⁡SdSt_* = H^{-1}\ln S_{\mathrm{dS}}, with SdS=πmPl2/H2S_{\mathrm{dS}} = \pi m_{\mathrm{Pl}}^2/H^2, which is the time over which the degrees of freedom of a horizon scramble information [27, 28, 29]. Identifying it with the window assumes that regions which have left each other’s horizon stay coherent for that long. Setting Nw=ln⁡SdS+δN_w = \ln S_{\mathrm{dS}} + \delta, evaluated when the window closes, gives Δh2eNw=16eδ\Delta_h^2 e^{N_w} = 16e^\delta exactly and

r=54cwe−δe2ϵNw,ns−1=nT=−2ϵ,(24)r = \frac{5}{4c_w}e^{-\delta}e^{2\epsilon N_w},\qquad n_s - 1 = n_T = -2\epsilon, \tag*{(24)}

where the factor e2ϵNwe^{2\epsilon N_w} accounts for the drift between the pivot and its legs. The ratio then no longer depends on HH. The tilt changes as well, because eNw∝H−2e^{N_w} \propto H^{-2} when the window closes. With ϵ=0.0088\epsilon= 0.0088 the bound r<0.036r < 0.036 requires δ≥4.5\delta\geq4.5 for cw=1.17c_w = 1.17 and δ≥2.5\delta\geq2.5 for cw=7.8c_w = 7.8, so the window has to exceed the scrambling time by a few e-folds. For δ\delta up to 7, the window is Nw≃30N_w \simeq30–37 for cw=1.17c_w = 1.17 and rr lies between 3×10−33 \times10^{-3} and the current limit, within reach of the next generation of B-mode experiments. Such a window would violate the consistency relation ns−1=2nTn_s - 1 = 2n_T of (21) and Ref. [4], and it needs a drift twice as large. A measurement of nTn_T therefore separates the two cases.

The gravitational-wave background today

The stochastic background is dominated by the first-order tensors. For modes that re-enter during radiation domination [30], the stochastic gravitational-wave background today is

ΩGWh2(f)=Ωrh224g∗(Tf)g∗0(gs0gs(Tf))4/3Δh2(k∗)(ff∗)nT,(25)\Omega_{\mathrm{GW}}h^2(f) = \frac{\Omega_r h^2}{24}\frac{g_*(T_f)}{g_{*0}}\left(\frac{g_{s0}}{g_s(T_f)}\right)^{4/3}\Delta_h^2(k_*)\left(\frac{f}{f_*}\right)^{n_T}, \tag*{(25)}

up to fendf_{\mathrm{end}}, with Ωrh2=4.18×10−5\Omega_rh^2 = 4.18 \times10^{-5} and TfT_f the temperature at re-entry.

Figure 6 shows the result for ΩGWh2(f)\Omega_{\mathrm{GW}}h^2(f) across the window. At 0.3 Hz, ΩGWh2\Omega_{\mathrm{GW}}h^2 runs from 2.7×10−172.7 \times10^{-17} at the lower edge to 1.9×10−211.9 \times10^{-21} at the upper edge. We compare it with approximate power-law-integrated sensitivities [31] for four years of data at signal-to-noise ratio 10 (see more details in Appendix F), assuming that astrophysical foregrounds can be subtracted. BBO [32] detects the background for r≳0.011r \gtrsim0.011, that is Nw≲33N_w \lesssim33 for cw=1.17c_w=1.17. The baseline DECIGO [33] would need r≳0.08r \gtrsim0.08, which is already excluded, while LISA [34] and ground-based interferometers [35] are far from the signal. An ultimate DECIGO, with Ωhh2∼10−20\Omega_{\mathrm{h}}h^2 \sim10^{-20} [36], reaches r≳1.3×10−5r \gtrsim1.3 \times10^{-5} or Nw≲47N_w \lesssim47. Expressed in terms of rr, these reaches depend neither on the normalisation convention nor on cwc_w.

Figure 6. Gravitational-wave background today, ΩGWh2\Omega_{\rm GW}h^2, against frequency ff, across the allowed window (solid lines, labelled by NwN_w and rr), with the scalar-induced background for Nw=30.5N_w=30.5 (dotted red line). Black curves are approximate power-law-integrated sensitivities (SNR = 10, four years) for LISA, BBO and DECIGO. Bars mark representative levels for an ultimate DECIGO and for ET/CE. The dotted gray line is the integrated ΔNeff<0.3\Delta N_{\rm eff}<0.3 bound and the gray band the CMB scales. Each spectrum ends at fendf_{\rm end}, while the steps are the changes in g∗.g_*.

The scalar-induced gravitational wave background is 0.82As2Ωrh20.82 A_s^2\Omega_r h^2 times the g∗g_* factor of Eq. (25), about 6×10−236 \times10^{-23} on scales k<kbk<k_b, and it falls steeply beyond. It was computed at second order in the Einstein-de Sitter and radiation eras in Refs. [17, 18] and revisited in Refs. [19, 20, 21]. Tensor-induced tensors are of order Δh4Ωr≲10−25\Delta_h^4\Omega_r \lesssim10^{-25}. The integrated energy density is at most ∫ΩGWh2 dln⁡f≃3×10−15\int\Omega_{\mathrm{GW}}h^2\,\mathrm{d}\ln f \simeq3 \times10^{-15}, reached at the lower edge, nine orders of magnitude below the ΔNeff\Delta N_{\mathrm{eff}} bound. Therefore, both contributions are negligible throughout the window.

Spectral distortions and small scales

The break at kbk_b is a distinctive feature of the mechanism. For Nw≲42N_w \lesssim42 it lies beyond the scales that produce μ\mu-distortions, and μ\mu keeps its standard value, 1.65×10−81.65 \times10^{-8} for these parameters [37]. For larger windows the break moves into the μ\mu era, 31.6≲k/Mpc−1≲540031.6 \lesssim k/\mathrm{Mpc}^{-1} \lesssim5400, and μ\mu drops to 1.6×10−81.6 \times10^{-8} at Nw=43N_w=43, 1.2×10−81.2 \times10^{-8} at Nw=45N_w=45, 2.4×10−92.4 \times10^{-9} at Nw=48N_w=48 and 2×10−102 \times10^{-10} at the upper edge of the window. A measured μ\mu below the standard value, with no detected B-modes, would therefore point to a long window. The same steepening suppresses the small-scale power relevant for primordial black holes and for the induced background at high frequencies, as these values assume a sharp break. For a release of finite width, the scales whose release is under way at the end receive more than the plateau value (Sec. V), so the spectrum can rise before it falls; we have not computed the shape of this transition, and for Nw≳43N_w \gtrsim43 the value of μ\mu is uncertain in both directions.

Discussion

We have shown that the tensor modes of Inflation without an Inflaton (IWI) are not enhanced at second order, and that the enhanced graviton-induced curvature can be carried into the radiation era with the observed tilt.

The tensors are not enhanced because the transverse-traceless (TT) sector obeys a wave equation whose retarded Bunch-Davies response is suppressed by oscillatory cancellations across the full time integral, whereas the scalar potential is fixed by an elliptic constraint that responds at once to the stress of all frozen gravitons. The gravitational waves of IWI are then the vacuum gravitons, Gaussian to a part in 10510^5, with tensor tilt nT=−2ϵn_T=-2\epsilon, as long as the depletion adds no long-range spin-2 response of its own (Sec. III D).

For the scalar, the curvature that survives is that of the hypersurface on which de Sitter (dS) ends, and its enhanced part is the offset between this hypersurface and local proper time, which the clock carries by following zero-shear time. In the δN\delta N picture, the inflaton is both the source of this offset and the clock that reads it. In IWI the two roles can be distinguished, although their dynamics need not be independent. The Bunch-Davies gravitons provide the source through the second-order constraints, and in our construction the depleting vacuum provides a clock that reads it within a finite reach, p/k≲eNwp/k\lesssim e^{N_w}, which makes the enhancement scale independent. This is the fixed-ratio prescription of Ref. [2], realised in time and made compatible with the end of dS. The non-vacuum sector fed by the depletion is the sink required by energy-momentum conservation, i.e. the counterpart of the reheating products. For adiabatic sinks, we find that it changes the amplitude by less than 0.2 per cent. The radiation era then begins with a single adiabatic and Gaussian mode, conserved on super-horizon scales, with amplitude cwΔh4eNw/20c_w\Delta_h^4e^{N_w}/20 and tilt ns−1=−4ϵn_s-1=-4\epsilon, which steepens beyond kende−Nwk_{\rm end}e^{-N_w}. On the other hand, Refs. [4, 8] read the spectrum at horizon crossing, where the populated pairs have loop momenta of order the external scale. The result for the scalar tilt ns−1=−4ϵn_s-1=-4\epsilon is the same, but which of the two readings carries the observed amplitude is to be investigated.

Several results follow directly from the Einstein equations: the sourcing of the scalar by graviton pairs, the (p/k)2(p/k)^2 enhancement of the zero-shear curvature, the consistency of the pure-Λ\Lambda solution with all four scalar second-order equations, the absence of tensor enhancement in the hard-region loop with Bunch-Davies sources and the need for a sink once the vacuum depletes. We have checked numerically that the post-exit uniform-density curvature is conserved for the radiation sink, and that it is insensitive to the sink for adiabatic sinks and for cs2>wc_s^2>w. However, for cs2<wc_s^2<w our runs did not converge, so we do not draw any conclusion from them. The local entropy and anisotropic responses we tested either leave ζ\zeta unchanged up to an unenhanced local term or rescale cwc_w by about 5 per cent for a response of relative strength λ\lambda, while a persistent stress proportional to the strain is excluded. We assume the global zero-shear clock, which we identify with the break of the dS coherent state as a whole and which synchronises regions up to 102210^{22} Hubble radii apart. We also assume the value of the window, with a few e-folds beyond the scrambling time as a natural candidate; the memory of the clock after the release, which sets cwc_w but moves the lower edge of the window by only a few e-folds; and the end of dS, which a steady depletion does not produce. The transfer of the enhanced curvature into the radiation era therefore follows, given the hypotheses of the adopted clock: synchronisation on zero-shear time for NwN_w e-folds, release, and memory afterwards. How the system composed of the vacuum and produced sector realises them remains open to further investigation.

Moreover, within the passive closure, we have seen that the equation of state of the sink matters little. The main open problem is to derive from the dynamics of the coherent state how the depletion synchronises each scale with zero-shear time for a fixed NwN_w, how it realises the finite reach p/k≲eNwp/k \lesssim e^{N_w} and why it keeps its memory afterwards, and it has to do so through the response of the vacuum and the produced sector together. The same calculation should fix the transfer JνJ_\nu and the perturbative properties of the sink, including any density, entropy, stress and memory response, decide how the produced sector takes part in the selection, and establish whether the depletion produces a long-range spin-2 response, which would be a new source of primordial gravitational waves. We are studying these questions in ongoing work with G. Dvali on horizon memory and a state-resolved quantum-break clock [38, 39], which separates the coherent occupation of the dS state from the activity of its microscopic channels. Its connection to the p/kp/k selection remains to be established.

A natural candidate for the microscopic identity of the clock is suggested by the operator-algebraic program of Gómez [40, 41, 42, 43]. There, a clock is not optional: the local observables of a dS patch form a type III1_1 factor whose reparametrization-invariant subalgebra is trivial, so gauge-invariant observables exist only after the algebra is extended by a clock and the operators are clock dressed (the crossed product), and the ϵ=0\epsilon= 0 limit of the gauge-invariant algebra is argued to be algebraically nonexistent [43]. Gómez’s model-independent clock for the planar patch [42] has, one by one, the properties postulated in § 1.3 B. It is a two-mode system, one mode inside and one beyond the horizon, whose state, i.e. the EPR squeezed pair generated by the de Sitter Bogolyubov transformation, is coherent across super-horizon distances: the non-locality the window requires is the entanglement of the clock state. It is a spin-0 number-phase system, [N^,Φ^]=−i[\hat{N},\hat{\Phi}] = -i, so it adds no long-range spin-2 response (Sec. III D). Its energy is ⟨Hc⟩=ϵH2a−2\langle H_c\rangle= \epsilon H^2 a^{-2}, which is the enthalpy of the sink, so the background bookkeeping is already the clock’s. And the gravitational bound on clock precision gives the dressing a maximal reach [42], of the kind a finite window needs, with ln⁡SdS\ln S_{\mathrm{dS}} its natural scale, i.e. the scrambling-time candidate of Eq. (24). What the algebraic clock does not yet provide is the value of NwN_w and the release-and-memory law of § 1.3 B; these are precisely the properties the state-resolved quantum-break clock is meant to derive in [39]. If the identification holds, the window clock postulated here is not an addition to the scenario but the crossed-product dressing that any dS observable requires.

Several predictions are independent of these details. The tensor tilt is directly connected to the scalar one via nT=(ns−1)/2n_T = (n_s - 1)/2, or ns−1n_s - 1 for a scrambling-time window, whatever the value of rr, so a measurement of nTn_T both tests the mechanism and distinguishes the two cases. The tensor-to-scalar rr ratio lies between 3×10−63 \times10^{-6} and the current limit, and the scalar spectrum has a small-scale break whose position is tied to rr through NwN_w. Finally, the gravitational-wave background is that of the vacuum gravitons, within reach of BBO for r≳0.01r \gtrsim0.01 and of an ultimate DECIGO for r≳10−5r \gtrsim10^{-5}.

Acknowledgments

The work of RJ is supported by the Simons Foundation. Funding for the work of MT and RJ was partially provided by project PID2022-141125NB-I00, and grant CEX2024-001451-M funded by MICIU/AEI/10.13039/501100011033. DB and SM acknowledge support from the COSMOS network (www.cosmosnet.it) through ASI (Italian Space Agency) Grants 2016-24-H.0, 2016-24-H.1-2018 and 2020-9-HH.0. This work is supported in part by the MUR Departments of Excellence grant “Quantum Frontiers” of the Physics and Astronomy Department of Padova, Italy. MA acknowledges the hospitality of the Institute of Cosmos Sciences of the University of Barcelona (ICCUB), where this work was carried out. The project that gave rise to these results received the support of a fellowship from “la Caixa” Foundation (ID 100010434), awarded to MT with code LCF/BQ/D124/12070012.

AI usage: Claude Code was used to assist with coding, computations, and graphic generation. The ideas, inferences, and results all originate in the (human) brains of the authors. Claude and ChatGPT were used to check citations, spelling, grammar, inconsistencies and also to enhance the flow of the text and its logic. Handwritten notes were transcribed to Latex using these tools. The original text was written by the authors, who are responsible for the results, findings, and conclusions of this article.

References

Second-order equations

We take k=kz^\mathbf{k}=k\hat{\mathbf{z}} and denote as QνμQ^\mu_\nu the graviton-bilinear part of a2Gνμa^2G^\mu_\nu. At the order of the bilinear cross term, the scalar equations in an arbitrary gauge are as follows (8πG=18\pi G=1):

k2ϕ+3Hϕ′+3H2ψ−2Hk2(B−E′)+Q00=−a2δρ,(26)k^2\phi+3\mathcal{H}\phi'+3\mathcal{H}^2\psi-2\mathcal{H}k^2(B-E')+Q^0_0=-a^2\delta\rho, \tag*{(26)}
ϕ′+Hψ+ik⋅Q0k=−a2(ρ+p)v,(27)\phi'+\mathcal{H}\psi+i\frac{\mathbf{k}\cdot\mathbf{Q}^0}{k}=-a^2(\rho+p)v, \tag*{(27)}
ϕ−ψ−2[(B−E′)′+2H(B−E′)]=T,(28)\phi-\psi-2\left[(B-E')'+2\mathcal{H}(B-E')\right]=T, \tag*{(28)}
ϕ′′+2Hϕ′+Hψ′+(2a′′a−H2)ψ+Qzz=a2δp,(29)\phi''+2\mathcal{H}\phi'+\mathcal{H}\psi'+\left(2\frac{a''}{a}-\mathcal{H}^2\right)\psi+Q^z_z=a^2\delta p, \tag*{(29)}

with TT defined in Eq. (4). We have checked symbolically, for a generic η(η)\eta(\eta), that under xμ→xμ+ξμx^\mu\to x^\mu+\xi^\mu the operator transforms by exactly −LξGνμ-\mathcal{L}_{\xi}G^\mu_\nu, and that its graviton-bilinear parts are the same in every gauge.

In Poisson gauge, the dynamical equation is Eq. (29)}, with $\psi=\phi-T$ and $\delta\rho$ taken from Eq. (A1)`. For the window clock δρ=13δρr−δρv\delta\rho=\frac{1}{3}\delta\rho_r-\delta\rho_v, with δρv=ρˉv′ξ\delta\rho_v=\bar{\rho}'_v\xi, and the curvature on uniform total density is

ζ=−12ϕ−112χ+δρ3(ρ+p).\zeta=-\frac{1}{2}\phi-\frac{1}{12}\chi+\frac{\delta\rho}{3(\rho+p)}.

Consistency of the pure-Λ\Lambda solution

In pure GR+Λ\Lambda we have δρ=v=δp=Π=0\delta\rho=v=\delta p=\Pi=0. We take ϕ\phi from the comoving Poisson equation (3), and $\psi=\phi-T from Eq. (4), with exact Bunch-Davies legs, and evaluate on this solution the full second-order operator a2Gμνa^{2}G_{\mu\nu}, metric and graviton parts together. The derivatives ϕ′\phi', ϕ′′\phi'', ψ′\psi' and ψ′′\psi'' are those of the functions themselves. We compute them by fourth-order finite differences with step 10−20∣η∣10^{-20}|\eta| in 80-digit arithmetic. Table II gives, for each pair, the largest residual of the four scalar projections (00, k^⋅0i\hat{k}\cdot0i, trace, and longitudinal ijij) over seven times in k∣η∣∈[10−6,5]k|\eta|\in[10^{-6},5], each divided by the larger of its metric and graviton parts. All four vanish to the precision of the differencing, ≤5×10−25\leq5\times10^{-25}. The frozen form of the same solution ϕ=Fχ/4\phi=\mathcal{F}\chi/4, with ψ=ϕ′=ϕ′′=0\psi=\phi'=\phi''=0, instead fails the 00, trace and longitudinal equations at order one (see Table II). Its 0i0i residual is the quantity mm defined below, which vanishes as k∣η∣k|\eta|, but 3Hm3\mathcal{H}m, which appears in the 00 equation, stays finite.

pair (x,y)(x,y), helicitiesexact all fourfrozen ansatz 00frozen ansatz tracefrozen ansatz long.
(10,10.4),++(10,10.4),++2×10−282\times10^{-28}0.700.701.00
(3,3.4),+−(3,3.4),+-5×10−255\times10^{-25}1.011.011.00
(0.7,0.8),++(0.7,0.8),++5×10−285\times10^{-28}0.510.511.00
(0.98,0.23),−+(0.98,0.23),-+3×10−283\times10^{-28}1.281.281.00
(1.05,0.15),−−(1.05,0.15),--2×10−272\times10^{-27}1.751.751.00

Table II. Relative residuals of the four scalar Einstein equations in pure GR+Λ\Lambda. Exact: the Bunch-Davies solution evaluated with its own time derivatives. We find that all four residuals are comparable, so we quote their maximum over seven times in k∣η∣∈[10−6,5]k|\eta|\in[10^{-6},5]. Frozen ansatz: ϕ=Fχ/4\phi=\mathcal{F}\chi/4, ψ=ϕ′=ϕ′′=0\psi=\phi'=\phi''=0, with exact Bunch-Davies legs in QQ, at k∣η∣=10−6k|\eta|=10^{-6}. The residuals differ by equation, so we list the 00, the trace, and the longitudinal residual (the 0i0i residual vanishes as k∣η∣k|\eta| and is omitted). The last two pairs are non-planar random configurations.

The reason the frozen ansatz fails is that three combinations stay finite at late times although the frozen limit sets each of them to zero. Each is a power of H∝1/∣η∣\mathcal{H}\propto1/|\eta| multiplying quantities that vanish, or tend to constants, in conformal time. The first is the graviton term Hχikχki′\mathcal{H}\chi^{ik}\chi'_{ki}, which tends to χik∇2χki\chi^{ik}\nabla^{2}\chi_{ki} because Hf′→−p2f\mathcal{H}f'\to-p^{2}f for Bunch-Davies legs. It appears only in Q00Q^{0}_{0}, since QllQ^{l}_{l}, TT and QzzQ^{z}_{z} have no part proportional to H\mathcal{H}. The other two come from the 00 and the 0i0i equation. Writing m≡ik^⋅Q0/km\equiv i\hat{k}\cdot Q^{0}/k, which tends to m0ηm_{0}\eta, the 00 equation gives 3H(ϕ′+Hψ)=−3Hm→3m03\mathcal{H}(\phi'+\mathcal{H}\psi)=-3\mathcal{H}m\to3m_{0}. Since 2a′′/a−H2=2H′+H22a''/a-\mathcal{H}^{2}=2\mathcal{H}'+\mathcal{H}^{2} in exact dS, it also gives ϕ′′+2Hϕ′+Hψ′+(2H′+H2)ψ=−a−2(a2m)′→m0\phi''+2\mathcal{H}\phi'+\mathcal{H}\psi'+(2\mathcal{H}'+\mathcal{H}^{2})\psi=-a^{-2}(a^{2}m)'\to m_{0} in the longitudinal equation. Keeping all three, the late-time system reads

k2ϕ=−Q00−3m0,ϕ=T,Qzz=−m0,Q00=Qll.(30)k^{2}\phi=-Q^{0}_{0}-3m_{0},\qquad\phi=T,\qquad Q^{z}_{z}=-m_{0},\qquad Q^{0}_{0}=Q^{l}_{l}. \tag*{(30)}

The last two relations express momentum and energy conservation of the frozen graviton stress, which the Bianchi identity guarantees, while the last says that frozen tensor gradients act as a w=−1/3w=-1/3 component.

For (x,y)=(10,10.4)(x,y)=(10,10.4) with positive helicities, at k∣η∣=10−4k|\eta|=10^{-4}, we find k2ϕ=107.50k^{2}\phi=107.50 and Q00=−362.48Q^{0}_{0}=-362.48, of which −414.64-414.64 comes from the Hχχ′\mathcal{H}\chi\chi' term. The 0i0i equation then gives 3m0=254.983m_{0}=254.98, the longitudinal component is Qzz=−84.99Q^{z}_{z}=-84.99 and the trace is Qll=−362.48Q^{l}_{l}=-362.48. These are real parts, and the imaginary parts are about 0.1%0.1\%. If instead χ′=ψ=ϕ′=ϕ′′=0\chi'=\psi=\phi'=\phi''=0 is imposed before the limit is taken, the three finite combinations are lost: the 00 equation gives k2ϕ=−52.2k^{2}\phi=-52.2, the trace equation k2ϕ=−Qll=+362.5k^{2}\phi=-Q^{l}_{l}=+362.5, the traceless equation still gives k2ϕ=k2T=107.5k^{2}\phi=k^{2}T=107.5, and the longitudinal equation demands Qzz=0Q^{z}_{z}=0, although it should give Qzz=−84.99Q^{z}_{z}=-84.99. This is the apparent overconstraint that motivated the effective fluid of [1]: the second-order equations of [1], including the equation for ϕ2\phi_{2}, are correct, and their large-scale solution ϕ2=Fχ/4\phi_{2}=\mathcal{F}_{\chi}/4, ψ2=0\psi_{2}=0 is the late-time limit of the exact solution. The overconstraint is an artifact of the order in which the limits are taken.

Tensor time integrals

The late-time tensor is

h(2)(k,0)=∫−∞0Gk(0,η′)S(η′) dη′,h^{(2)}(\mathbf{k},0)=\int_{-\infty}^{0}G_{k}(0,\eta')S(\eta')\,\mathrm{d}\eta',

with the retarded Green function of Eq. (10) at η=0\eta=0,

Gk(0,η′)=kη′cos⁡kη′−sin⁡kη′k3η′2→∣η′∣3for k∣η′∣≪1.(31)G_{k}(0,\eta')=\frac{k\eta'\cos k\eta'-\sin k\eta'}{k^{3}\eta'^{2}}\to\frac{|\eta'|}{3}\qquad\text{for }k|\eta'|\ll1. \tag*{(31)}

For Bunch-Davies legs, the two source structures carry f1f2f_{1}f_{2} and f1′f2′f_{1}'f_{2}', each of which is e−iPη′e^{-iP\eta'} times a polynomial, with P=p1+p2P=p_{1}+p_{2}. The integrals are then elementary once the contour is tilted into the damping direction, so that e−iPη′e^{-iP\eta'} provides the convergence factor. The integrals carry a short-distance regulator δ\delta (the lower limit of ∣η′∣|\eta'|): the 1/δ1/\delta and ln⁡δ\ln\delta pieces of a cutoff regularisation cancel and leave the real closed forms of Eq. (11). For k→0k\to0, JFF→−(P2+2p1p2)/P4J_{FF}\to-(P^{2}+2p_{1}p_{2})/P^{4} and Jdd→2p12p22/P4J_{dd}\to2p_{1}^{2}p_{2}^{2}/P^{4}, while for p1,p2≫kp_{1},p_{2}\gg k we obtain JFF→−3/(8p2)J_{FF}\to-3/(8p^{2}) and Jdd→1/8J_{dd}\to1/8 (see Sec. III B). We checked the closed forms against numerical integration along η′=−s(1+0.4i)\eta'=-s(1+0.4i), s∈(0,∞)s\in(0,\infty), finding agreement to 10−1010^{-10} for (p1,p2)=(0.7,0.8),(3.1,2.6),(20.3,20.1)(p_{1},p_{2})=(0.7,0.8),(3.1,2.6),(20.3,20.1) and (0.3,1.25)(0.3,1.25) in units of kk.

Numerical scheme for the window clock

We use N=ln⁡aN=\ln a as the time variable and evolve the vacuum and radiation densities, the two graviton modes, the Poisson-gauge potential ϕ\phi with ϕ′\phi', and the clock offset ξ\xi, with an eighth-order Runge-Kutta integrator at relative tolerance 10−1010^{-10}. The release is a quintic smooth-step in ln⁡(H/k)\ln(\mathcal{H}/k) of width ΔNr\Delta N_{r} centered on NwN_{w}. While clock is synchronised, the condition ξ=0\xi= 0 of Eq. (15) is enforced softly, by driving ξ\xi to zero at rate 20H20H. The exit is a smoothstep in NN of width 0.3, placed a few e-folds after both the release and the freezing of the internal pair. We have checked that the results are robust to where the exit is placed. Moving it to one e-fold after the centre of the release, with xc=eNw=10x_c = e^{N_w} = 10 and u=x/xc=0.3u = x/x_c = 0.3–30, changes ∣ζ∣|\zeta| by at most 0.2 per cent for ΔNr=2\Delta N_r = 2 and 1 (§ 1.4).

An unreleased clock reproduces the direct Poisson-gauge evolution to 10−1110^{-11}. A clock released deep inside the horizon approaches the synchronous-gauge result, ∣ζ∣=0.129|\zeta| = 0.129 for x=10x = 10, releasing progressively deeper, at p∣η∣=74p|\eta| = 74, 201 and 546, gives ∣ζ∣=0.62|\zeta| = 0.62, 0.15 and 0.125, converging to that value from below. The ratio to the zero-shear value depends on xx and xcx_c only through uu, to better than half a per cent. After the exit, ζ\zeta is constant to 6×10−56 \times10^{-5} while k/aH<1/30k/aH < 1/30 (Figure 4). These runs use a single pair configuration, y=x+0.4y = x + 0.4 with positive helicities, but the dilation argument of § 1.3 B holds for all configurations.

Relative change of curvature after the exit

Figure 4. Relative change of ζ\zeta after the exit, 104(∣ζ∣/∣ζexit∣−1)10^{4}(|\zeta|/|\zeta_{\mathrm{exit}}|-1), against k/aHk/aH in the radiation era, for four pairs inside (u=0.1u=0.1, 0.3, 1) and outside (u=10u=10) the window. Dotted vertical lines mark the re-entry of the internal modes, p=aHp=aH. On super-horizon scales, the curvature ζ\zeta is conserved to better than 10−410^{-4}.

Uniform density versus zero shear

Eliminating ϕ′+Hψ\phi' + \mathcal{H}\psi between Eqs. (26) and (27) gives the exact decomposition

ζ=ζP+Hv+δρc3(ρ+p)(32)\zeta= \zeta_{\mathrm{P}} + \mathcal{H}v + \frac{\delta\rho_c}{3(\rho+ p)} \tag*{(32)}
a2δρc=−(k2ϕ+Q00−3iHk^⋅Q0/k)a^2\delta\rho_c = -\left(k^2\phi+ Q^0_0 - 3i\mathcal{H}\hat{\mathbf{k}}\cdot\mathbf{Q}^0/k\right)

where vv is the velocity of the sink and δρc\delta\rho_c its comoving density, which is suppressed by (k/aH)2(k/aH)^2.

The left panel of Figure 7 shows ∣δρ/3(ρ+p)∣/∣ζP∣|\delta\rho/3(\rho+ p)|/|\zeta_{\mathrm{P}}| during quasi-dS for four pairs, with a zero-shear exit, ϵ=0.0088\epsilon= 0.0088 and the clock synchronised throughout. Equation (32) holds to 10−1310^{-13}, and the velocity term dominates, with the comoving density below 3×10−33 \times10^{-3} of it from one e-fold after freeze-out and below 10−410^{-4} after five e-folds. A few e-folds after freeze-out, we find numerically Hv=−0.1670ψ\mathcal{H}v = -0.1670\psi for all four pairs. This value follows from the Euler equation of the sink, which for radiation injected at rest in steady state reads v′+4Hv=ψ−δp/(ρ+p)v' + 4\mathcal{H}v = \psi- \delta p/(\rho+ p), the term 4Hv4\mathcal{H}v being the drag of the injected energy. With δp/(ρ+p)=δρ/[3(ρ+p)]≃Hv\delta p/(\rho+ p) = \delta\rho/[3(\rho+ p)] \simeq\mathcal{H}v and ψ∝η\psi\propto\eta it gives Hv=−ψ/6\mathcal{H}v = -\psi/6 up to corrections of order ϵ\epsilon. Since ∣ψ∣≃(2|\psi| \simeq(2–2.5)(p/aH)∣ζP∣2.5)(p/aH)|\zeta_{\mathrm{P}}|, the correction to ζP\zeta_{\mathrm{P}} is (0.33(0.33–0.42)(p/aH)∣ζP∣0.42)(p/aH)|\zeta_{\mathrm{P}}| and carries no factor 1/ϵ1/\epsilon. Dropping ϕ′\phi' in Eq. (26) would instead have given ∣ψ∣/2ϵ|\psi|/2\epsilon, about 330 times larger (dashed curves in Figure 7, left).

Figure 7. Left: We show the density contribution ∣δρ/3(ρ+p)∣/∣ζP∣|\delta\rho/3(\rho+p)|/|\zeta_{\mathrm{P}}| in quasi-de Sitter (ϵ=0.0088\epsilon=0.0088) against ln⁡(aH/p)\ln(aH/p), the e-folds since the internal pair froze, for four pairs (solid lines overlapping). Dashed lines show the estimate ∣ψ∣/2ϵ|\psi|/2\epsilon that dropping ϕ′\phi' in Eq. (26) would give. The dotted line shows 0.42 p/aH0.42\,p/aH. Right: We show the x=30x=30 pair through a zero-shear exit and into the radiation era, against N−NexitN-N_{\mathrm{exit}}. The split between ζP/ζ\zeta_{\mathrm{P}}/\zeta (blue) and the density term [δρ/3(ρ+p)]/ζ[\delta\rho/3(\rho+p)]/\zeta (red) changes when the internal modes re-enter (p=aHp=aH, orange line), while the uniform-density ζ/ζexit\zeta/\zeta_{\mathrm{exit}} (black dashed) stays constant. The lapse ψ/ζ\psi/\zeta (green) goes from ≃0\simeq0 while the modes are frozen to the adiabatic value as they re-enter. Dotted horizontal lines mark 2/32/3, 1/31/3 and −4/3-4/3, i.e. the adiabatic-mode values the curves relax to.

After a zero-shear exit, the uniform-density curvature equals the late dS ζP\zeta_{\mathrm{P}} to 1.4×10−41.4 \times10^{-4}, 4.5×10−44.5 \times10^{-4} and 1.3×10−31.3 \times10^{-3} for x=3x = 3, 10 and 30, respectively, whose internal modes froze 8.5, 7.3 and 6.2 e-folds before the exit. Both quantities approach the frozen value as p/aHp/aH. With the window clock released before the exit, moving the exit from 3 to 6 to 10 e-folds after the release changes ∣ζ∣|\zeta| by less than 10−610^{-6} for (x,xc)=(10,30)(x,x_c) = (10,30), (3,30)(3,30) and (30,10)(30,10), respectively, because the free-running clock integrates the decaying lapse.

In the radiation era, we see from the right panel of Figure 7 that the split changes when the internal gravitons re-enter. While they are frozen, ψ≃0\psi\simeq0 and δρ≃0\delta\rho\simeq0. Once they re-enter, their anisotropic stress disappears, and the potentials relax to the standard adiabatic mode, ϕ≃ψ≃−43ζ\phi\simeq\psi\simeq-\frac{4}{3}\zeta, with δρ/[3(ρ+p)]≃13ζ\delta\rho/[3(\rho+ p)] \simeq\frac{1}{3}\zeta and δp→25ζ\delta p \rightarrow\frac{2}{5}\zeta. The uniform-density ζ\zeta does not change, to 7×10−67 \times10^{-6} at k/aH=1/30k/aH = 1/30. This requires the graviton-volume term −χχ/12-\chi\chi/12: when the internal modes of the x=3x = 3 pair re-enter (k/aH=1/3k/aH = 1/3) ζ\zeta drifts by 2% without it and by 7×10−47 \times10^{-4} with it.

To test the dependence on the sink, we replaced the radiation by a fluid with constant ww and rest-frame sound speed cs2c_s^2, for which δp=cs2δρ+(cs2−w)ρˉ′v\delta p = c_s^2\delta\rho+ (c_s^2 - w)\bar{\rho}'v [44], with δρ=δρ\delta\rho= \delta\rho and ρˉ′\bar{\rho}' including the exchange. We kept Jν=∂νρvJ_\nu= \partial_\nu\rho_v and converted the vacuum into the same fluid at the exit. The velocity term becomes Hv=−ψ/(4+6cs2)\mathcal{H}v = -\psi/(4 + 6c_s^2), independent of ww. Table III gives the fractional change of ∣ζ∣|\zeta| after the exit relative to radiation, for positive helicities. After a zero-shear exit, the changes of ±1.5×10−3\pm1.5 \times10^{-3} come from the change in the background fraction of the sink, 2ϵ/[3(1+w)]2\epsilon/[3(1+w)] at fixed ϵ\epsilon, and the effect of cs2c_s^2 is only 10−610^{-6}. The window ratio f=∣ζreleased∣/∣ζneverreleased∣f = |\zeta_{\mathrm{releas ed}}|/|\zeta_{\mathrm{never released}}| at u=1/3u = 1/3 lies between 1.0152 and 1.0160 for all closures.

wwcs2c_s^2δρ3(ρ+p)ψ\frac{\delta\rho}{3(\rho+p)\psi}zero shearu=13u=\frac{1}{3}u=3u=3u=10u=10
00−0.251-0.251+1.5+1.5+1.1+1.1+0.7+0.7+0.8+0.8
11−0.100-0.100−1.5-1.5−1.1-1.1−0.7-0.7−0.8-0.8
1/31/31−0.100-0.100+0.001+0.001+0.37+0.37+0.8+0.8+0.8+0.8
01/31/3−0.168-0.168+1.5+1.5+1.5+1.5+1.5+1.5+1.5+1.5
1/31/30.05−0.233-0.233−0.001-0.001−0.35-0.35−0.65-0.65n.c.
11/31/3−0.167-0.167−1.5-1.5−1.5-1.5−0.8-0.8n.c.

Table III. Effect of changing the sink, for sinks with equation of state ww and sound speed cs2c_s^2 (first two columns), for positive helicities. The third column reports the density term δρ/[3(ρ+p)ψ]\delta\rho/[3(\rho+p)\psi], measured three e-folds after freeze-out, and it equals to Hv/ψ\mathcal{H}v/\psi to 5×10−45\times10^{-4}. The last four columns show how much the post-exit ζ\zeta shifts when radiation is replaced by that sink. Each entry is the fractional change from the radiation result (w=cs2=1/3w=c_s^2=1/3), in units of 10−310^{-3}. Considering different exits and pairs: zero shear is a clock never released (x=10x=10), while the u=13,3,10u=\frac{1}{3},3,10 columns are the released windows, with u=p/(keNw)u=p/(k e^{N_w}), at (x,xc)=(10,30)(x,x_c)=(10,30), (30,10)(30,10) and (100,10)(100,10), respectively. The u=30u=30 run, (300,10)(300,10), matches the u=10u=10 column for the first four sinks. n.c.: not converged (see text).

When cs2≠wc_s^2 \ne w the pressure contains (cs2−w)ρˉ′v(c_s^2 - w)\bar{\rho}'v, and vv has to be inferred from Eq. (27) by dividing by a2(ρ+p)≃2ϵH2a^2(\rho+ p) \simeq2\epsilon\mathcal{H}^2. While the internal modes are inside the horizon, this division amplifies the integration error. For cs2>wc_s^2 > w, and for u≤3u \le3, the entries of Table III do not change when the tolerance is tightened from 10−1010^{-10} to 10−1210^{-12}. For cs2<wc_s^2 < w and u≥10u \ge10 they do change. For (w,cs2)=(1,1/3)(w,c_s^2) = (1,1/3) at u=10u = 10 the change is −1.1%-1.1\% at 10−1210^{-12} and −0.8%-0.8\% at 10−1310^{-13}, and at u=30u = 30 it has not converged. These closures need a formulation that evolves the density and velocity of the sink explicitly.

For the responses of a non-passive sink, we kept radiation and added either δpnad=λ(ρ+p)Bρv/ρ\delta p_{\mathrm{nad}} = \lambda(\rho+ p)B\rho_v/\rho to the pressure or Πe=λ(ρ+p)Bρv/ρ\Pi_e = \lambda(\rho+ p)B\rho_v/\rho to the traceless equation, ϕ−ψ=T=2a2Πe/k2\phi- \psi= T = 2a^2\Pi_e/k^2, with the corresponding 23a2Πe\frac{2}{3}a^2\Pi_e in Eq. (29). The factor ρv/ρ\rho_v/\rho switches the response off after the exit. We used the volume bilinear BV=E(1):Ef1f2/6\mathcal{B}_V = E^{(1)} : E_{f_1f_2}/6, which is the mean pair part of χχ/12\chi\chi/12, and the kinetic bilinear BK=E(1):E(2)(p^1p^2)\mathcal{B}_K = E^{(1)} : E^{(2)}(\hat{p}_1\hat{p}_2), which decays as (p/aH)2(p/aH)^2 after freeze-out. The runs use y=x+0.4y = x + 0.4 with positive helicities and a zero-shear exit 7.6 e-folds after the external mode leaves the horizon. Both changes are linear in λ\lambda (λ=1/2\lambda= 1/2 halves them). The pressure term gives the same Δζ=0.011λ\Delta\zeta= 0.011\lambda for ϵ=0.003\epsilon= 0.003 and 0.0088, which shows that it is generated during the exit, where the sink is no longer a small fraction. With BV\mathcal{B}_V in the anisotropic stress, the elastic case, the post-exit ∣ζ∣|\zeta| is 5×1065 \times10^6 for λ=1\lambda= 1 at every xx, and of order (aH/k)2(aH/k)^2 at the exit.

Gravitational-wave spectrum and sensitivities

Eq. (25) uses a smooth fit to the Standard Model g∗(T)g_*(T) and gs(T)g_s(T) [45] and neglects neutrino damping, which matters only below ∼10−10 Hz\sim10^{-10}\,\mathrm{Hz}. The frequency is f=k/2πa0f=k/2\pi a_0 with a0a_0 the scale factor today, and fend=kend/2πa0f_{\mathrm{end}}=k_{\mathrm{end}}/2\pi a_0 follows from the radiation exit. The sensitivity curves are power-law integrated [31], for SNR =10=10 and four years of data. For LISA we use the sky-averaged noise of Ref. [46] in autocorrelation, and for BBO and DECIGO the effective noise fits of Ref. [47] for a cross-correlated pair. The minima of the curves are h2Ω≃8×10−14h^2\Omega\simeq8\times10^{-14} for LISA, 9×10−189\times10^{-18} for BBO and 6×10−176\times10^{-17} for DECIGO. They are indicative only, and a change in sensitivity by a factor FF moves the BBO reach by ≃2.1ln⁡F\simeq2.1\ln F in NwN_w. The μ\mu-distortion is computed with

μ≃2.2∫Δζ2(k)[e−k/5400−e−(k/31.6)2]dln⁡k(33)\mu\simeq2.2\int\Delta_\zeta^2(k)\left[e^{-k/5400}-e^{-(k/31.6)^2}\right]\mathrm{d}\ln k \tag*{(33)}

with kk in Mpc−1\mathrm{Mpc}^{-1} [37].

The homogeneous scalar mode and graviton backreaction

The scalar response to quadratic graviton fluctuations and the homogeneous backreaction belong to distinct sectors of the second-order Einstein equations. For nonzero external momentum, the scalar potential obeys the elliptic constraint

k2ϕ(k)=−Q00(k)+3iHk^⋅Q0(k)k,k≠0,(34)k^2\phi(\mathbf{k})=-Q_0^0(\mathbf{k})+3i\mathcal{H}\frac{\hat{\mathbf{k}}\cdot\mathbf{Q}^0(\mathbf{k})}{k},\qquad k\ne0, \tag*{(34)}

where Q νμQ^{\mu}_{\ \nu} is the quadratic graviton contribution to a2G νμa^{2}G^{\mu}_{\ \nu}. For internal momenta p≫kp \gg k, the inverse Laplacian generates the (p/k)2(p/k)^{2} enhancement. The scalar evolution equation containing ϕ′′\phi^{\prime\prime} introduces no independent propagating degree of freedom: the Bianchi identities relate it to the constraints once the graviton equations hold. By contrast, the TT sector evolves through a retarded Green function and its source can contain p2χ2p^{2}\chi^{2} terms, but the TT projection removes the longitudinal and trace structures responsible for the scalar enhancement. Oscillatory cancellations across the full retarded Bunch-Davies integral, including the sub-horizon and frozen regimes, leave an order-one tensor kernel.

a. The homogeneous mode. Separating the quadratic source into its mean and fluctuation, δQ νμ=Q νμ−⟨Q νμ⟩\delta Q^{\mu}_{\ \nu}=Q^{\mu}_{\ \nu}-\langle Q^{\mu}_{\ \nu}\rangle, statistical homogeneity gives ⟨ϕ(k)⟩=0\langle\phi(\mathbf{k})\rangle=0 for k≠0k\ne0. The connected scalar fluctuations are sourced by δQ νμ\delta Q^{\mu}_{\ \nu}, whereas the mean belongs to the homogeneous Einstein equations. The exact k=0k=0 mode cannot be obtained from Eq. (34), since the inverse Laplacian is undefined there and the scalar momentum constraint involves division by kk. For the homogeneous metric

ds2=a2[−(1+ψ0)dη2+(1−ϕ0)δijdxidxj],ds^{2}=a^{2}\left[-(1+\psi_{0})d\eta^{2}+(1-\phi_{0})\delta_{ij}dx^{i}dx^{j}\right],

the undivided temporal and spatial equations give

3Hϕ0′+3H2ψ0+⟨Q 00⟩=0,3\mathcal{H}\phi_{0}^{\prime}+3\mathcal{H}^{2}\psi_{0}+\langle Q^{0}_{\ 0}\rangle=0,
ϕ0′′+2Hϕ0′+Hψ0′+(2a′′a−H2)ψ0+13⟨Q ii⟩=0.(35)\phi_{0}^{\prime\prime}+2\mathcal{H}\phi_{0}^{\prime}+\mathcal{H}\psi_{0}^{\prime}+\left(2\frac{a^{\prime\prime}}{a}-\mathcal{H}^{2}\right)\psi_{0}+\frac{1}{3}\langle Q^{i}_{\ i}\rangle=0. \tag*{(35)}

There is no independent homogeneous scalar momentum constraint, and compatibility follows from the Bianchi identities. The correction to the expansion rate is

δHH=−ϕ0′+Hψ02H=⟨Q 00⟩6H2=12ρGWρΛ,(36)\frac{\delta H}{H}=-\frac{\phi_{0}^{\prime}+\mathcal{H}\psi_{0}}{2\mathcal{H}}=\frac{\langle Q^{0}_{\ 0}\rangle}{6\mathcal{H}^{2}}=\frac{1}{2}\frac{\rho_{\mathrm{GW}}}{\rho_{\Lambda}}, \tag*{(36)}

where 8πGρGW=⟨Q 00⟩/a28\pi G\rho_{\mathrm{GW}}=\langle Q^{0}_{\ 0}\rangle/a^{2} and ρΛ=3H2/(8πG)\rho_{\Lambda}=3H^{2}/(8\pi G). The homogeneous backreaction is therefore sourced by the averaged quadratic stress, not the enhanced inhomogeneous scalar response.

b. The quadratic graviton stress. Expanding the ADM Hamiltonian constraint, 2Gnn=(3)R+K2−KijKij2G_{nn}={}^{(3)}R+K^{2}-K_{ij}K^{ij}, to quadratic order in the TT perturbation χij\chi_{ij} and spatially averaging gives

⟨Q 00⟩=H⟨χijχij′⟩+18⟨χij′χij′⟩+18⟨∂ℓχij∂ℓχij⟩.(37)\langle Q^{0}_{\ 0}\rangle=\mathcal{H}\langle\chi_{ij}\chi_{ij}^{\prime}\rangle+\frac{1}{8}\langle\chi_{ij}^{\prime}\chi_{ij}^{\prime}\rangle+\frac{1}{8}\langle\partial_{\ell}\chi_{ij}\partial_{\ell}\chi_{ij}\rangle. \tag*{(37)}

The mixed term remains finite at late times despite χij′→0\chi_{ij}^{\prime}\to0 mode by mode, and must therefore be retained before taking the super-horizon limit. Using the exact Bunch-Davies mode function T(pη)=(1+ipη)e−ipη\mathcal{T}(p\eta)=(1+ip\eta)e^{-ip\eta} and x=p/(aH)x=p/(aH), the bare energy density follows from Eq. (37), while the pressure is obtained from ρ˙p+3H(ρp+pp)=0\dot{\rho}_{p}+3H(\rho_{p}+p_{p})=0, applied to each shell at fixed comoving momentum (see also [4]):

8πGdρGWdln⁡p=Δh2H28x2(2x2−7),8\pi G\frac{d\rho_{\mathrm{GW}}}{d\ln p}=\frac{\Delta_{h}^{2}H^{2}}{8}x^{2}(2x^{2}-7),
8πGdpGWdln⁡p=Δh2H224x2(2x2+7).(38)8\pi G\frac{dp_{\mathrm{GW}}}{d\ln p}=\frac{\Delta_{h}^{2}H^{2}}{24}x^{2}(2x^{2}+7). \tag*{(38)}

The bare effective energy density is negative for x2<7/2x^{2}<7/2 because of the mixed expansion term in Eq. (37), reflecting the non-positive-definite character of the gravitational effective stress at long wavelengths. In the homogeneous gauge ψ0=0\psi_{0}=0, the constraint also yields, for each comoving shell,

dϕ0dln⁡p=Δh248(x4−7x2)+Cp.(39)\frac{d\phi_{0}}{d\ln p}=\frac{\Delta_{h}^{2}}{48}(x^{4}-7x^{2})+C_{p}. \tag*{(39)}

The constant CpC_{p} fixes the homogeneous background convention, and requiring the shell contribution to vanish as x→0x\to0 sets Cp=0C_{p}=0. Thus, ϕ0\phi_{0} is not a gauge-independent observable, and absorbing the homogeneous correction into the renormalized background defines the convention ⟨ϕ⟩ren=0\langle\phi\rangle_{\mathrm{ren}}=0, while retaining the physical backreaction in the background geometry.

c. Magnitude and interpretation of the backreaction. The homogeneous expansion obeys −H˙/H2=4πG(ρeff+peff)/H2-\dot{H}/H^{2}=4\pi G(\rho_{\mathrm{eff}}+p_{\mathrm{eff}})/H^{2}. To characterize a finite population of Bunch-Davies gravitons, define

E(X)≡4πGH2∫0Xdxxd(ρGW+pGW)dln⁡p∣p=aHx.(40)\mathcal{E}(X)\equiv\frac{4\pi G}{H^{2}}\int_{0}^{X}\frac{dx}{x}\left.\frac{d(\rho_{\mathrm{GW}}+p_{\mathrm{GW}})}{d\ln p}\right|_{p=aHx}. \tag*{(40)}

For a scale-invariant tensor spectrum in exact dS, E(X)=Δh2(2X4−7X2)/48\mathcal{E}(X)=\Delta_{h}^{2}(2X^{4}-7X^{2})/48, so E(1)=−5Δh2/48\mathcal{E}(1)=-5\Delta_{h}^{2}/48 (see [4]). With Δh2=rAs\Delta_{h}^{2}=rA_{s} and the tensor-amplitude range considered here, $6\times10^{-16}\lesssim|\mathcal{E}(1)|\lesssim8\times10^{-12}$, far below the ϵ≃0.0088\epsilon\simeq0.0088 associated with the observed scalar tilt. The finite-window quantity E(X)\mathcal{E}(X) is not the renormalized slow-roll parameter. At fixed XX, the moving cutoff p=aHXp=aHX introduces a flux of modes, so the selected sector is not separately conserved, unlike the individual fixed-comoving-momentum shells. The full backreaction instead requires the renormalized homogeneous stress and gravitational counterterms. A dS-invariant renormalized vacuum contribution satisfies ρren=−pren\rho_{\mathrm{ren}}=-p_{\mathrm{ren}} and shifts the expansion scale without generating a nonzero ϵ\epsilon, so the Gaussian graviton stress is distinct from the depletion-generated sector whose enthalpy determines ϵ\epsilon. Although linear tensors are gauge-invariant, the homogeneous second-order geometry retains a background-definition freedom, and its contributions to ϕ\phi and ζP=−ϕ/2−χijχij/12 can be absorbed into the renormalized background. In pure dS,\zeta_{\mathrm{P}}=-\phi/2-\chi_{ij}\chi_{ij}/12\ \text{can be absorbed into the renormalized background. In pure dS,}\zeta_{\mathrm{P}}$ is not the physical uniform-density curvature, which requires a non-vacuum clock. The scalar spectrum is determined by connected fluctuations at k≠0k\ne0.

References

  1. [1]D. Bertacca, R. Jimenez, S. Matarrese, and A. Ricciardone, Phys. Rev. Res. 7, L032010 (2025), arXiv:2412.14265 [astro-ph.CO].arxiv.org/abs/2412.14265
  2. [2]M. Taforetti, M. Abdelaziz, D. Bertacca, R. Jimenez, S. Matarrese, and A. Ricciardone, Phys. Rev. D 113, 094011 (2011). 023553 (2026), arXiv:2511.11808 [astro-ph.CO].
  3. [3]M. Abdelaziz, M. Traforetti, D. Bertacca, R. Jimenez, S. Matarrese, and A. Ricciardone, Phys. Rev. D 114, 023517 (2026), arXiv:2603.15036 [astro-ph.CO].
  4. [4]R. Jimenez, D. Bertacca, M. Abdelaziz, S. Matarrese, A. Ricciardone, and M. Traforetti, (2026), arXiv:2610.20378 [astro-ph.CO].arxiv.org/abs/2610.20378
  5. [5]G. Dvali and C. Gomez, JCAP 01, 023, arXiv:1312.4795 [hep-th].DOI
  6. [6]G. Dvali, C. Gomez, and S. Zell, JCAP 06, 028, arXiv:1701.08776 [hep-th].DOI
  7. [7]C. Gomez and R. Jimenez, Phys. Dark Univ. 36, 101035 (2022), arXiv:2111.05380 [hep-th].arxiv.org/abs/2111.05380
  8. [8]D. Bertacca, R. Jimenez, and S. Matarrese, in preparation (2026).
  9. [9]W. Israel, Nuovo Cim. B 44S10, 1 (1966), [Erratum: Nuovo Cim.B 48, 463 (1967)].
  10. [10]N. Deruelle and V. F. Mukhanov, Phys. Rev. D 52, 5549 (1995), arXiv:gr-qc/9503050.DOI
  11. [11]A. A. Starobinsky, JETP Lett. 42, 152 (1985).
  12. [12]M. Sasaki and E. D. Stewart, Prog. Theor. Phys. 95, 71 (1996), arXiv:astro-ph/9507001.DOI
  13. [13]D. H. Lyth, K. A. Malik, and M. Sasaki, JCAP 05, 004, arXiv:astro-ph/0411220.
  14. [14]J. M. Maldacena, JHEP 05, 013, arXiv:astro-ph/0210603.DOI
  15. [15]F.-Y. Chen, J.-Z. Zhou, D. Wu, Z.-C. Li, and P.-Y. Wu, (2025), arXiv:2507.22688 [astro-ph.CO].arxiv.org/abs/2507.22688
  16. [16]P. A. R. Ade et al. (BICEP, Keck), Phys. Rev. Lett. 127, 151301 (2021), arXiv:2110.00483 [astro-ph.CO].DOI
  17. [17]S. Matarrese, S. Mollerach, and M. Bruni, Phys. Rev. D 58, 043504 (1998), arXiv:astro-ph/9707278.DOI
  18. [18]S. Mollerach, D. Harari, and S. Matarrese, Phys. Rev. D 69, 063002 (2004), arXiv:astro-ph/0310711.DOI
  19. [19]K. N. Ananda, C. Clarkson, and D. Wands, Phys. Rev. D 75, 123518 (2007), arXiv:gr-qc/0612013.DOI
  20. [20]D. Baumann, P. J. Steinhardt, K. Takahashi, and K. Ichiki, Phys. Rev. D 76, 084019 (2007), arXiv:hep-th/0703290.DOI
  21. [21]K. Kohri and T. Terada, Phys. Rev. D 97, 123532 (2018), arXiv:1804.08577 [gr-qc].DOI
  22. [22]S. Weinberg, Phys. Rev. D 69, 023503 (2004), arXiv:astro-ph/0306304.DOI
  23. [23]Y. Akrami et al. (Planck), Astron. Astrophys. 641, A10 (2020), arXiv:1807.06211 [astro-ph.CO].DOI
  24. [24]D. Wands, K. A. Malik, D. H. Lyth, and A. R. Liddle, Phys. Rev. D 62, 043527 (2000), arXiv:astro-ph/0003278.DOI
  25. [25]E. Allys et al. (LiteBIRD), PTEP 2023, 042F01 (2023), arXiv:2202.02773 [astro-ph.IM].DOI
  26. [26]K. N. Abazajian et al. (CMB-S4), CMB-S4 Science Book, First Edition (2016) arXiv:1610.02743 [astro-ph.CO].arxiv.org/abs/1610.02743
  27. [27]Y. Sekino and L. Susskind, JHEP 10, 065, arXiv:0808.2096 [hep-th].DOI
  28. [28]L. Susskind, (2011), arXiv:1101.6048 [hep-th].arxiv.org/abs/1101.6048
  29. [29]L. Alsma and G. Shiu, JHEP 05, 152, arXiv:2002.01326 [hep-th].
  30. [30]Y. Watanabe and E. Komatsu, Phys. Rev. D 73, 123515 (2006), arXiv:astro-ph/0604176.DOI
  31. [31]E. Thrane and J. D. Romano, Phys. Rev. D 88, 124032 (2013), arXiv:1310.5300 [astro-ph.IM].DOI
  32. [32]J. Crowder and N. J. Cornish, Phys. Rev. D 72, 083005 (2005), arXiv:gr-qc/0506015.DOI
  33. [33]S. Kawamura et al., Classical and Quantum Gravity 28,
  34. [34]P. Amaro-Seoane et al. (LISA), (2017), arXiv:1702.00786 [astro-ph.IM].
  35. [35]M. Punturo et al., Class. Quant. Grav. 27, 194002 (2010).
  36. [36]H. Kudoh, A. Taruya, T. Hiramatsu, and Y. Himemoto, Phys. Rev. D 73, 064006 (2006), arXiv:gr-qc/0511145.DOI
  37. [37]J. Chluba, A. L. Erickcek, and I. Ben-Dayan, Astrophys. J. 758, 76 (2012), arXiv:1203.2681 [astro-ph.CO].DOI
  38. [38]D. Bertacca, G. Dvali, and R. Jimenez, in preparation (2026).
  39. [39]D. Bertacca, G. Dvali, and R. Jimenez, in preparation (2026).
  40. [40]C. Gomez, (2022), arXiv:2207.06704 [hep-th].arxiv.org/abs/2207.06704
  41. [41]C. Gomez, (2023), arXiv:2302.14747 [hep-th].arxiv.org/abs/2302.14747
  42. [42]C. Gomez, (2023), arXiv:2304.11845 [hep-th].arxiv.org/abs/2304.11845
  43. [43]C. Gomez, Int. J. Mod. Phys. A 40, 2540013 (2025), arXiv:2407.20671 [hep-th].arxiv.org/abs/2407.20671
  44. [44]W. Hu, Astrophys. J. 506, 485 (1998), arXiv:astro-ph/9801234.DOI
  45. [45]K. Saikawa and S. Shirai, JCAP 05, 035, arXiv:1803.01038 [hep-ph].DOI
  46. [46]T. Robson, N. J. Cornish, and C. Liu, Class. Quant. Grav. 36, 105011 (2019), arXiv:1803.01944 [astro-ph.HE].DOI
  47. [47]K. Yagi and N. Seto, Phys. Rev. D 83, 044011 (2011), [Erratum: Phys.Rev.D 95, 109901 (2017)], arXiv:1101.3940 [astro-ph.CO].DOI

Paper details

vlnc:W_c47fg6sh
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 3

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Claude Code, ChatGPT

View statement
Looking for related papers…
Contents