betaFeedback

WorksPhysics and Astronomy

Silent Heavy Fields: An Exclusion Rule for the Cosmological Collider

Abstract

The cosmological collider uses oscillatory signatures in primordial correlators to probe heavy fields during inflation. This probe has blind spots, as heavy fields can keep ticking in time while remaining silent in the correlators. In this letter, we derive an exclusion rule under which cosmological collider signals vanish. The rule covers two classes of diagrams where a de Sitter invariant heavy sector couples covariantly to a single external field at one vertex. We first prove it for exchanges mediated by the two-point function of the heavy sector to all loop orders, using a spectral decomposition followed by a field redefinition. The result then extends to one-loop polygons by matching their oscillatory contributions onto those of the first class. The approximate signals of previously studied processes that fall into these classes must therefore be revisited. As a nonvanishing example, we derive the leading signal of a triangle loop for the first time in a toy model motivated by chemical potentials.

Introduction

Heavy fields present during inflation oscillate with frequencies set by their masses. Through interactions with the light perturbations, their evolution from one interaction event to another can leave distinctive imprints on primordial correlators, taking the form of oscillations in the logarithm of momentum ratios. Such imprints provide the physical basis for cosmological collider (CC) physics [1, 2, 3, 4], giving us access to the heavy particle spectrum at the inflationary scale.

Over the past decade, extensive work on tree-level cosmological correlators has led to substantial theoretical progress [5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39], with the resulting predictions guiding most current observational searches for CC signals [40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50]. More recently, one-loop processes have attracted increasing attention [51, 52, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71] [53], as they offer richer phenomenology [72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85] and can even provide the leading contribution to CC signals in some scenarios [85, 86, 87, 88, 89]. This is particularly relevant when the particles involved must be produced in pairs, as is the case for fermions [68, 72, 73, 77, 82, 85, 86, 87, 90, 91] and particles carrying conserved charges [78, 92].

Existing studies of loop phenomenology often evaluate the CC signal approximately through a late-time expansion of the heavy propagators. However, the oscillating modes that appear in this expansion do not by themselves guarantee a nonzero CC signal. A particularly sharp illustration of this was recently found in a fermion bubble with a Yukawa coupling, which stays completely silent in the bispectrum once the fermion propagation respects de Sitter (dS) symmetry [68]. An analogous null was also found for a nonderivative scalar bubble in [93]. These examples raise a broader question: under what conditions do heavy fields that tick in time remain silent in the correlators?

In this letter, we derive an exclusion rule that answers this question for two broad classes of diagrams, in which a dS invariant heavy sector couples locally and covariantly to a single external field at one vertex. In the first class, the heavy sector connects this vertex to the rest of the diagram through its bulk two-point function. Using a spectral decomposition [94, 95, 96, 97], we express this two-point function as a superposition of tree exchanges, with the covariant coupling at this vertex mapped onto a quadratic mixing between the external field and the exchanged heavy mode. This mixing can be removed by a field redefinition [60, 68], so that the associated CC signal vanishes. The result holds irrespective of the interactions at the other end of the two-point function, which need not respect dS symmetry. The simplest realisation of this class is an elementary tree exchange, but the argument extends nonperturbatively to fully interacting composite two-point functions of the heavy sector.

In the second class, the heavy fields form a one-loop NN-gon, for which the absence of CC signals is not manifest. In the soft limit of the external field, the candidate nonanalytic contribution comes from the region in which the two internal propagators attached to its vertex carry soft loop momentum, while the remaining subgraph stays hard. Order by order in the soft expansion, we match this hard subgraph onto local operators at a single vertex [70], reducing the oscillatory contributions of the NN-gon to those of bubble diagrams of the first class. The exclusion rule therefore applies, and the CC signals naively expected from the oscillations of the internal heavy fields at late times are absent from the final correlator.

The exclusion rule has immediate phenomenological implications. Previously studied loop processes [73, 76, 79, 81, 82, 83] that fall into these two classes must have vanishing CC signals under the assumptions of the rule, in contrast to earlier approximate estimates. A nonzero signal therefore requires an ingredient that evades these assumptions, such as a noncovariant coupling to the external field or a heavy sector that breaks dS symmetry. Motivated by the helicity dependent Whittaker propagation induced by helical chemical potentials during inflation [17, 78, 92], we illustrate the latter possibility with a heavy scalar that shares this Whittaker propagation, while keeping its coupling to the external field covariant. Restoring realistic spin would mainly introduce the corresponding helicity and tensor structures without changing the underlying mechanism. We present the first analytical calculation of the leading CC signal from a Whittaker triangle, and find a signal that is generically nonzero and vanishes continuously as dS covariance is restored.

Notations and conventions. We work in the Poincaré patch of four dimensional dS space with metric gμν=a2(τ)ημνg_{\mu\nu}=a^2(\tau)\eta_{\mu\nu} and a(τ)=−1/(Hτ)a(\tau)=-1/(H\tau) for τ<0\tau<0. The external field ϕ\phi can actually be any light field coupled perturbatively to the heavy sector in our discussion. Observationally it corresponds to the massless inflaton fluctuation, although in explicit calculations we take it to be conformally coupled, as this simplifies the expressions without affecting our main discussion. Bold symbols denote comoving momentum with k=∣k∣k=|\mathbf{k}|. We consider the generic single soft limit kL/kS→0k_L/k_S\to0, taking the magnitude kLk_L of one external momentum to be small compared with a fixed hard scale kSk_S while the hard momentum approaches a fixed configuration consistently with momentum conservation, so that no other partial sum of external momentum becomes soft. By a CC signal we mean any term in this expansion that oscillates in log⁡(kL/kS)\log(k_L/k_S) with a nonzero frequency.

Vanishing CC signals

We now turn our attention to the central question of this letter, namely under what conditions heavy fields that tick in time remain silent in the correlators. Its answer is summarised at the outset in the following rule.

Exclusion rule. For diagrams with a single external field ϕkL\phi_{k_L} at one vertex, the CC signal generated by the heavy sector vanishes in the following two cases,

(i) diagrams with this vertex connected to the other vertex through a bulk two-point function of the heavy sector at arbitrary loop order, as in Figure 1(a), in any kinematic configuration,

Figure 1. Diagrams covered by the exclusion rule. (a) A bulk two-point exchange diagram with a single soft external leg ϕkL\phi_{k_L}. The blob represents an arbitrary connected dS covariant two-point function of the heavy sector. (b) One scalar principal series component of (a), represented by χν\chi_\nu exchange. (c) A general one-loop polygon with NN vertices. (d) The triangle contribution to the three-point correlator, with k3k_3 soft. In (c) and (d), red and blue lines denote soft and hard heavy propagators, respectively. The soft lines carry qq and q+kLq+k_L, with kL=k3k_L=k_3 in (d). In (c), τN\tau_N labels the soft vertex, while τ1\tau_1 and τN−1\tau_{N-1} label the hard endpoints of the propagators carrying qq and q+kLq+k_L, respectively. The region relevant to the CC signal has ∣τ1∣,∣τN−1∣≪∣τN∣|\tau_1|,|\tau_{N-1}|\ll|\tau_N|.

(ii) one-loop polygons, as in Figure 1(c), at every order in the generic single soft expansion,

provided that the heavy sector has dS covariant intrinsic dynamics and couples to this external field through a local dS covariant interaction, assuming the Bunch-Davies vacuum. Arbitrary local couplings are allowed at other vertices.1

In what follows, we refer to the vertex carrying ϕkL\phi_{k_L} as the soft vertex and to the remaining vertices as hard vertices. Let us first prove the rule for bulk two-point exchange, and then show how one-loop polygons reduce to this case by matching their oscillatory terms onto bubble diagrams.

Bulk two-point exchange

Consider the bulk two-point diagram in Figure 1(a). The local dS covariant interaction at the soft vertex contracts derivatives of ϕ\phi with tensor operators OAμ1⋯μr\mathcal{O}^{\mu_1\cdots\mu_r}_A, built from the heavy fields and their covariant derivatives. We can always integrate by parts to move all derivatives from ϕ\phi onto these operators,2 so that each term takes the form

SA=gA∫d4x−g OAμ1⋯μr∇μ1⋯∇μrϕ=gA∫d4x−g ϕOA,(1)\begin{aligned} S_A = g_A \int\mathrm{d}^4x\sqrt{-g}\,\mathcal{O}^{\mu_1\cdots\mu_r}_A\nabla_{\mu_1}\cdots\nabla_{\mu_r}\phi\\ &= g_A\int\mathrm{d}^4x\sqrt{-g}\,\phi\mathcal{O}_A, \tag*{(1)} \end{aligned}

where OA≡(−1)r∇μr⋯∇μ1OAμ1⋯μr\mathcal{O}_A\equiv(-1)^r\nabla_{\mu_r}\cdots\nabla_{\mu_1}\mathcal{O}^{\mu_1\cdots\mu_r}_A is a local covariant scalar operator. The heavy fields may be bosons or fermions of arbitrary spin. In all cases, the soft vertex reduces to a coupling between ϕ\phi and a local covariant scalar operator of the heavy sector.

The remaining external legs may couple through an otherwise arbitrary local interaction, which we write as

SB=gB∫d4x−g JB[{ϕ};τ]⋅OB.(2)S_B=g_B\int\mathrm{d}^4x\sqrt{-g}\,\mathcal{J}_B[\{\phi\};\tau]\cdot\mathcal{O}_B. \tag*{(2)}

Here OB\mathcal{O}_B is a local covariant operator of the heavy sector, while JB\mathcal{J}_B describes its otherwise arbitrary local coupling to the remaining external fields. The notation {ϕ}\{\phi\} includes these fields and their time and spatial derivatives, with any additional scale factors absorbed into the arbitrary time-dependent coefficients of JB\mathcal{J}_B. The dot denotes contraction over the heavy tensor components. At order gAgBg_Ag_B, the heavy dynamics enter only through the connected two-point function ⟨OA(X)OB(Y)⟩c\langle\mathcal{O}_A(X)\mathcal{O}_B(Y)\rangle_c. We keep this correlator fully nonperturbative, so the argument applies at arbitrary loop order within the heavy sector.

Since OA\mathcal{O}_A is a scalar operator, the dS spectral decomposition [94, 95, 96, 97] gives the principal series contribution to this correlator in the form

⟨OA(X)OB,μ1⋯μℓ(Y)⟩c=∫0∞dν ρAB(ν)Dμ1⋯μℓ(ν)(Y)G(ν)(X,Y)+⋯ .(3)\langle\mathcal{O}_A(X)\mathcal{O}_{B,\mu_1\cdots\mu_\ell}(Y)\rangle_c =\int_0^\infty\mathrm{d}\nu\,\rho_{AB}(\nu)\mathcal{D}^{(\nu)}_{\mu_1\cdots\mu_\ell}(Y)\mathcal{G}^{(\nu)}(X,Y)+\cdots. \tag*{(3)}

Here G(ν)\mathcal{G}^{(\nu)} is the propagator of a scalar field χν\chi_\nu with mass mνm_\nu, where mν2/H2=9/4+ν2m_\nu^2/H^2=9/4+\nu^2. The covariant differential operator Dμ1⋯μℓ(ν)\mathcal{D}^{(\nu)}_{\mu_1\cdots\mu_\ell} acts at YY and supplies the tensor structure of the heavy operator OB\mathcal{O}_B. The ellipsis denotes other contributions that do not carry the oscillatory signal of interest.3 Equation (3) holds for each Schwinger–Keldysh (SK) component, with the indices suppressed.

One-loop NN-gons

We now turn to one-loop NN-gons, whose heavy lines connect the soft vertex to several hard vertices. Let us consider the diagram in Figure 1(c), with the soft external momentum kLk_L attached to the vertex at τN\tau_N. We choose the loop momentum qq so that the two adjacent heavy propagators carry qq and q+kLq+k_L, with their hard endpoints at τ1\tau_1 and τN−1\tau_{N-1}, respectively. The NN-gon contribution TNT_N to the correlator takes the schematic form4

TN∼∑a1,…,aN=±[∏v=1N−1∫−∞0iav dτv (−τv)pveaviEvτv]×∫−∞0iaN dτN (−τN)pNeaNikLτN×Q.(4)T_N \sim\sum_{a_1,\ldots,a_N=\pm}\left[\prod_{v=1}^{N-1}\int_{-\infty}^{0} ia_v\,d\tau_v\,(-\tau_v)^{p_v}e^{a_v iE_v\tau_v}\right]\times\int_{-\infty}^{0} ia_N\,d\tau_N\,(-\tau_N)^{p_N}e^{a_N i k_L\tau_N}\times Q . \tag*{(4)}

where the loop integral is defined as

Q≡∫d3q(2π)3Da1aN(ν1)(q;τ1,τN)DaNaN−1(νN)(∣q+kL∣;τN,τN−1)×∏i=2N−1Dai−1ai(νi)(∣q−∑ℓ=1i−1kℓ∣;τi−1,τi).(5)Q \equiv\int\frac{d^3q}{(2\pi)^3}D_{a_1a_N}^{(\nu_1)}(q;\tau_1,\tau_N)D_{a_Na_{N-1}}^{(\nu_N)}(|q+k_L|;\tau_N,\tau_{N-1})\times\prod_{i=2}^{N-1}D_{a_{i-1}a_i}^{(\nu_i)}\left(\left|q-\sum_{\ell=1}^{i-1}k_\ell\right|;\tau_{i-1},\tau_i\right). \tag*{(5)}

Here ai=±1a_i=\pm1 label the SK branches, and DaiajD_{a_i a_j} denotes the bulk-to-bulk propagator of the field with mass parameter ν\nu. The quantities EiE_i and kik_i denote the sum of external momentum magnitudes and the total external momentum entering vertex ii, respectively. At the hard vertices i=1,…,N−1i=1,\ldots,N-1, we have Ei,ki≳kSE_i,k_i\gtrsim k_S, whereas EN=kN=kLE_N=k_N=k_L at the soft vertex. The integers pip_i account for all other time dependence.

To extract the CC signal, we first apply a directional cut to the two heavy propagators adjacent to the soft vertex. This replaces them by the Wightman components associated with the ordering τN<τ1,τN−1\tau_N<\tau_1,\tau_{N-1}, while the other terms do not contribute to the CC signal [70, 98, 99].

For the cut contribution, a CC signal can only come from the soft loop region q∼∣q+kL∣∼kL≪kSq\sim|q+k_L|\sim k_L\ll k_S. At fixed τ1\tau_1 and τN−1\tau_{N-1}, the corresponding mode arguments become small as kL→0k_L\to0. The two modes therefore admit superhorizon expansions at their hard endpoints, with oscillatory factors (−qτ1)±iν1(-q\tau_1)^{\pm i\nu_1} and (−∣q+kL∣τN−1)±iνN(-|q+k_L|\tau_{N-1})^{\pm i\nu_N}, respectively.

We can now Taylor expand the hard subgraph in the soft momentum qq and kLk_L. At leading order, its dependence on the soft loop momentum drops out, leaving only the two soft internal lines in the loop integral. The hard subgraph can then be represented by a local “pinched” operator connecting the hard external legs, whose coefficient is fixed by matching the hard time integrals. This reduces the NN-gon to a bubble diagram. At higher orders, the additional powers of soft momentum are reproduced by derivatives in the local operator, so the same matching procedure extends the reduction order by order in the soft expansion.5

Each matched bubble falls within case (i), for which we have already shown that the CC signal vanishes. The NN-gon CC signal therefore vanishes at every order in the soft expansion. Schematically, the argument can be summarised by the diagrammatic relation (11), and we put the detailed derivation in App. § 0.6.

[hard diagram]CC=pinch∑rCr[bubble]CC=∑r,νCrρr(ν)[soft line]CC=0.(6)\left[\text{hard diagram}\right]_{\mathrm{CC}} \mathrel{\overset{\mathrm{pinch}}{=}} \sum_r C_r\left[\text{bubble}\right]_{\mathrm{CC}} = \sum_{r,\nu}C_r\rho_r(\nu)\left[\text{soft line}\right]_{\mathrm{CC}} =0 . \tag*{(6)}

Applications of the rule

Over the past decade, CC physics has been used to explore a broad range of new physics scenarios including dark matter, Higgs inflation and leptogenesis, and many of the loop diagrams considered in these studies [73, 76, 79, 81, 82, 83] are covered by our exclusion rule. For example, the W/ZW/Z bubbles in the cosmological Higgs collider [76] fall into case (i), while the neutrino triangle proposed as a probe of leptogenesis [82] falls into case (ii), so that their oscillatory signals should vanish.

As such diagrams are difficult to evaluate analytically, earlier predictions were obtained by expanding the heavy field propagators at late times at both endpoints. Since the squeezed configuration does not validate such an expansion at the soft vertex, these predictions miss the contribution from the early time region. Previous results based on such expansions must therefore be revisited, including those used to calibrate the cosmological Higgs collider and to infer the parameters of leptogenesis. A nonzero signal in these processes requires an ingredient that evades the exclusion rule.

Beyond the exclusion rule

To generate a nonzero CC signal, at least one of the assumptions underlying the exclusion rule must be violated. A noncovariant soft vertex provides an immediate way to evade the rule. A nontrivial alternative is to keep the interaction local and dS covariant while breaking dS symmetry in the heavy sector, as we now illustrate with a chemical potential.

For concreteness, consider a heavy scalar σ\sigma coupled to ϕ\phi through the local covariant interaction

Sint=−λ2∫d4x−g ϕσ2.(7)S_{\mathrm{int}}=-\frac{\lambda}{2}\int d^4x\sqrt{-g}\,\phi\sigma^2 . \tag*{(7)}

The connected three-point function ⟨ϕϕϕ⟩\langle\phi\phi\phi\rangle receives a contribution at order λ3\lambda^3 from three insertions of this interaction, forming the triangle in Figure 1(d), the N=3N=3 case of Figure 1(c). Following our conventions, we take ϕ\phi to be conformally coupled with mϕ2=2H2m_\phi^2=2H^2 to simplify the calculation and avoid unnecessary subtleties.

Although a scalar does not admit a helical chemical potential, it can model the propagation of a spinning field with a chemical potential once the spin structures are omitted. These structures do not affect the oscillatory scaling of interest, so the scalar model allows us to show how the exclusion rule is evaded and carry out the calculation. For the mode function u(k,τ)u(k,\tau) of σ\sigma, we define the canonically normalised mode v(k,τ)=a(τ)u(k,τ)v(k,\tau)=a(\tau)u(k,\tau) and take

∂τ2v+(k2+2κkτ+ν2+1/4τ2)v=0,(8)\partial_\tau^2v+\left(k^2+\frac{2\kappa k}{\tau}+\frac{\nu^2+1/4}{\tau^2}\right)v=0 , \tag*{(8)}

where the term proportional to κ\kappa describes the effect of the chemical potential and breaks dS boosts, while κ=0\kappa=0 recovers the covariant mode equation. The solution is given by a Whittaker-WW function, and we give the ingredients needed for the calculation in App. § 0.7.

Triangle loops are notoriously difficult to evaluate analytically, and the presence of a chemical potential further complicates the calculation.

complicates the calculation. The pinching procedure [70] used in the proof above reduces the triangle to bubble integrals order by order in the soft expansion. These integrals can then be evaluated using Mellin-Barnes (MB) representations of the Whittaker propagators. To our knowledge, this provides the first analytical result for the CC signal from such a triangle with a chemical potential. Details of the calculation are given in App. § 0.7, while a more extensive treatment of loop CC signals with a chemical potential will be presented in a forthcoming companion paper.

Let us first define the dimensionless bispectrum BB by

⟨ϕϕϕ⟩′=τf3k3(k1+k2)4B(k1,k2,k3),(9)\langle\phi\phi\phi\rangle'=\frac{\tau_f^3k_3}{(k_1+k_2)^4}B(k_1,k_2,k_3) , \tag*{(9)}

where τf\tau_f denotes the end of inflation and the prime indicates that the momentum conservation delta function has been removed. Since each conformally coupled external mode scales as τ\tau at late times, the three-point function carries an overall factor of τf3\tau_f^3, which we have factored out in defining the bispectrum. The CC signal is given by the oscillatory part of B(k1,k2,k3)B(k_1,k_2,k_3) in the single soft limit k3=kL≪k1,k2k_3=k_L\ll k_1,k_2. At leading order, it can be formally expressed as

BCC≃A(κ,ν)(kLkS)2iν+c.c.,(10)B_{\mathrm{CC}}\simeq\mathcal{A}(\kappa,\nu)\left(\frac{k_L}{k_S}\right)^{2i\nu}+\mathrm{c.c.}, \tag*{(10)}

where kS=k1+k2k_S=k_1+k_2 and the complex coefficient A(κ,ν)\mathcal{A}(\kappa,\nu) determines the amplitude and phase of the signal. The closed form of A\mathcal{A} is lengthy, so we give it in App. § 0.7 and omit it here. By breaking dS symmetry in the heavy sector, the chemical potential evades the exclusion rule and produces the nonzero amplitudes shown in Figure 2 for different fixed masses. The rapid growth at large κ\kappa reflects the enhancement of each heavy propagator by a factor e2πκe^{2\pi\kappa}, giving an overall factor e6πκe^{6\pi\kappa} for the triangle. We have verified numerically that the amplitude follows the scaling κ−6e6πκ\kappa^{-6}e^{6\pi\kappa}, as indicated by the dashed curves. The limit of zero chemical potential provides a further nontrivial check of the result. As shown in the inset, the amplitude approaches zero linearly with κ\kappa. At κ=0\kappa=0, dS symmetry is restored and the exclusion rule requires A(0,ν)=0\mathcal{A}(0,\nu)=0, which we have verified explicitly. Further details can be found in App. § 0.7.

Figure 2. CC signals from a loop with a chemical potential term for different fixed masses. The dashed curves indicate the scaling κ−6e6πκ\kappa^{-6}e^{6\pi\kappa} at large κ\kappa. The inset resolves the behaviour at small κ\kappa, where the amplitude approaches zero linearly with the chemical potential.

Summary

The late-time oscillations of heavy field modes can be imprinted on correlators through interactions with the external field. One might therefore expect a nonzero coupling to these modes to be sufficient to generate oscillations in momentum ratios. Our results show that the corresponding CC signal can nevertheless be entirely absent.

In this letter, we have derived an exclusion rule that identifies when this silence occurs. The rule applies whenever a dS invariant heavy sector couples covariantly to a single external field at one vertex. For diagrams that connect this vertex to the other vertex through a two-point function of the heavy sector, we proved that the CC signal vanishes to all loop orders and in any kinematic configuration. We then extended the rule to one-loop polygons, for which the CC signal vanishes at every order in the soft expansion. In both classes, the result holds regardless of the interactions at other vertices.

We also showed how breaking dS symmetry in the heavy sector can evade the exclusion rule. In a scalar triangle model with Whittaker propagation motivated by helical chemical potentials, we obtained the first analytical result for the leading CC signal. The signal is generically nonzero and vanishes linearly as the chemical potential is switched off, recovering the null result required by the rule in the dS invariant limit.

Our results suggest several directions for further investigation. For instance, the vanishing of the triangle signal at zero chemical potential can be understood from the late-time boundary, since all interactions then respect dS symmetry and conformal invariance fixes the three-point function up to an overall coefficient, leaving no room for a CC signal. The exclusion rule goes beyond this case, covering higher-point correlators and arbitrary interactions at the hard vertices. In particular, a box diagram with dS covariant interactions contributes no CC signal to the trispectrum in the soft limit of one external momentum, although conformal invariance does not fix this four-point function. It would be interesting to understand these vanishing signals from the boundary perspective as well. Another natural direction is to extend the rule to higher-loop diagrams with arbitrary hard subgraphs attached to the soft vertex. In addition, spin deserves a detailed treatment for both the external and the heavy fields, which we will partly address in a forthcoming companion paper. Finally, it would be valuable to revisit the phenomenological predictions of previously studied loop processes in light of the exclusion rule.

Acknowledgements We thank Chengcheng Han, Enrico Pajer, Dong-Gang Wang, Yi Wang, and Zhong-Zhi Xianyu for useful discussions. S.A. is supported by the Japan Science and Technology Agency (JST) as part of Adopting Sustainable Partnerships for Innovative Research Ecosystem (ASPIRE), grant JPMJAP2318, and by JSPS KAKENHI Grant Number 26K07078. Z.Q. is supported by the China Scholarship Council (CSC). M.Y. is supported by IBS under the project code, IBS-R018-D3, and by JSPS Grant-in-Aid for Scientific Research Number JP23K20843. Y.Z. is supported by IBS under the project code, IBS-R018-D3. The authors used ChatGPT (OpenAI, GPT-5.6 Sol and GPT-6 Astra) and Claude (Anthropic, Fable 5.1) to assist in checking derivations, identifying possible inconsistencies, and improving the presentation of the manuscript. All AI-assisted outputs were independently checked by the authors, who take full responsibility for the content.

References

Proof of the exclusion rule for one-loop NN-gons

Throughout the appendices, we set H=1H = 1. For a scalar of mass parameter ν=m2−9/4\nu= \sqrt{m^2 - 9/4}, the normalised Bunch-Davies (BD) mode function can be chosen as

u(k,τ)=π2e−πν/2(−τ)3/2Hiν(1)(−kτ)=∑c=±1kicνfc(k,τ),(11)u(k,\tau)=\frac{\sqrt{\pi}}{2}e^{-\pi\nu/2}(-\tau)^{3/2}H_{i\nu}^{(1)}(-k\tau)=\sum_{c=\pm1} k^{ic\nu}f_c(k,\tau), \tag*{(11)}

where the functions f±f_{\pm} are

fc(k,τ)≡cπecπν/221+icνsinh⁡(πν)(−τ)3/2+icν∑n=0∞(−k2τ2/4)nn! Γ(n+1+icν),(12)f_c(k,\tau)\equiv\frac{c\sqrt{\pi}e^{c\pi\nu/2}}{2^{1+ic\nu}\sinh(\pi\nu)}(-\tau)^{3/2+ic\nu}\sum_{n=0}^{\infty}\frac{(-k^2\tau^2/4)^n}{n!\,\Gamma(n+1+ic\nu)}, \tag*{(12)}

which admit Taylor expansions in kk around the origin and obey fc∗=−ecπνf−cf_c^*=-e^{c\pi\nu}f_{-c}. The four SK bulk-to-bulk propagators are constructed from this mode function as follows:

D−+(k;τ1,τ2)=u(k,τ1)u∗(k,τ2),D+−(k;τ1,τ2)=u∗(k,τ1)u(k,τ2),D++(k;τ1,τ2)=θ(τ1−τ2)D−+(k;τ1,τ2)+θ(τ2−τ1)D+−(k;τ1,τ2),D−−(k;τ1,τ2)=θ(τ2−τ1)D−+(k;τ1,τ2)+θ(τ1−τ2)D+−(k;τ1,τ2).(13)\begin{aligned} D_{-+}(k;\tau_1,\tau_2)&=u(k,\tau_1)u^*(k,\tau_2),\\ D_{+-}(k;\tau_1,\tau_2)&=u^*(k,\tau_1)u(k,\tau_2),\\ D_{++}(k;\tau_1,\tau_2)&=\theta(\tau_1-\tau_2)D_{-+}(k;\tau_1,\tau_2)+\theta(\tau_2-\tau_1)D_{+-}(k;\tau_1,\tau_2),\\ D_{--}(k;\tau_1,\tau_2)&=\theta(\tau_2-\tau_1)D_{-+}(k;\tau_1,\tau_2)+\theta(\tau_1-\tau_2)D_{+-}(k;\tau_1,\tau_2). \tag*{(13)} \end{aligned}

For later convenience, we reproduce the integral expression for the NN-gon correlator in (4) and (5):

TN∼∑a1,…,aN=±[∏v=1N−1∫−∞0iavdτv(−τv)pveiavEvτv]∫−∞0iaNdτN(−τN)pNeiaNkLτNQ,(14)\mathcal{T}_N\sim\sum_{a_1,\ldots,a_N=\pm}\left[\prod_{v=1}^{N-1}\int_{-\infty}^{0}ia_vd\tau_v(-\tau_v)^{p_v}e^{ia_vE_v\tau_v}\right]\int_{-\infty}^{0}ia_Nd\tau_N(-\tau_N)^{p_N}e^{ia_Nk_L\tau_N}Q, \tag*{(14)}

where

Q≡∫d3q(2π)3Da1aN(ν1)(q;τ1,τN)DaNaN−1(νN)(∣q+kL∣;τN,τN−1)×∏i=2N−1Dai−1ai(νi)(∣q−∑ℓ=ijkℓ∣;τi−1,τi).(15)Q\equiv\int\frac{d^3q}{(2\pi)^3}D_{a_1a_N}^{(\nu_1)}(q;\tau_1,\tau_N)D_{a_Na_{N-1}}^{(\nu_N)}(|q+k_L|;\tau_N,\tau_{N-1})\times\prod_{i=2}^{N-1}D_{a_{i-1}a_i}^{(\nu_i)}\left(\left|q-\sum_{\ell=i}^{j}k_\ell\right|;\tau_{i-1},\tau_i\right). \tag*{(15)}

Here EvE_v and kvk_v are the total external energy and momentum entering vertex vv, and pvp_v encodes the schematic vertex time dependence. Overall couplings and fixed external normalisations are suppressed. We evaluate the loop integrals in dimensional regularisation, taking the physical limit d→3d\to3 after subtracting any UV divergence. Scaleless integrals vanish in this prescription.

We take the generic single soft limit kL/kS→0k_L/k_S\to0 defined in the main text, with kS>0k_S>0 fixed independently of kLk_L. The hard momenta approach a fixed configuration analytically while preserving momentum conservation, and no additional independent momentum channel becomes soft. More explicitly, both the hard vertex energies and the sums of external momenta entering consecutive vertices obey

Ev≳kS,v=1,…,N−1,(16)E_v\gtrsim k_S,\qquad v=1,\ldots,N-1, \tag*{(16)}
∣∑ℓ=ijkℓ∣≳kS,1≤i≤j≤N−1,(i,j)≠(1,N−1).(17)\left|\sum_{\ell=i}^{j}k_\ell\right|\gtrsim k_S,\qquad1\leq i\leq j\leq N-1,\qquad(i,j)\ne(1,N-1). \tag*{(17)}

In particular, i=ji=j gives kv≳kSk_v\gtrsim k_S at each hard vertex with v≠Nv\ne N. The excluded sum satisfies ∑ℓ=1N−1kℓ=−kL\sum_{\ell=1}^{N-1}k_\ell=-k_L and therefore represents the chosen soft channel.

Validity of the directional cut

CC signals can be identified as the discontinuity associated with the soft limit and can be extracted using appropriate cutting rules. For the single soft signal considered here, we apply a directional cut to the two propagators attached to the soft vertex. This cut was introduced for triangle diagrams in [70], and we now extend it to a general NN-gon.

Writing p≡q+kLp \equiv q + k_L and p≡∣p∣p \equiv\lvert p\rvert, each of the two propagators attached to τN\tau_N in (A8) splits into two parts:

Da1aN(ν1)(q;τ1,τN)=D−aN,aN(ν1)(q;τ1,τN)+δa1aNθ(τN−τ1)[DaN,−aN(ν1)(q;τ1,τN)−D−aN,aN(ν1)(q;τ1,τN)](18)D^{(\nu_1)}_{a_1a_N}(q;\tau_1,\tau_N)=D^{(\nu_1)}_{-a_N,a_N}(q;\tau_1,\tau_N) +\delta_{a_1a_N}\theta(\tau_N-\tau_1)\left[D^{(\nu_1)}_{a_N,-a_N}(q;\tau_1,\tau_N)-D^{(\nu_1)}_{-a_N,a_N}(q;\tau_1,\tau_N)\right] \tag*{(18)}
DaNaN−1(νN)(p;τN,τN−1)=DaN,−aN(νN)(p;τN,τN−1)+δaNaN−1θ(τN−τN−1)[D−aN,aN(νN)(p;τN,τN−1)−DaN,−aN(νN)(p;τN,τN−1)].(19)D^{(\nu_N)}_{a_Na_{N-1}}(p;\tau_N,\tau_{N-1})=D^{(\nu_N)}_{a_N,-a_N}(p;\tau_N,\tau_{N-1}) +\delta_{a_Na_{N-1}}\theta(\tau_N-\tau_{N-1})\left[D^{(\nu_N)}_{-a_N,a_N}(p;\tau_N,\tau_{N-1})-D^{(\nu_N)}_{a_N,-a_N}(p;\tau_N,\tau_{N-1})\right]. \tag*{(19)}

The first term in each of (A11) and (A12) is the cut part, while the second is the time-ordered part. The product of the two cut parts defines the cut contribution to (A8). Each of the other three time-ordered terms contains at least one step function multiplying a commutator D+−−D−+D_{+-}-D_{-+}. Below we show that these time-ordered terms admit Taylor expansions about kL=0k_L=0 and thus have no CC signal.

In particular, consider the time-ordered contribution TN,TO\mathcal{T}_{N,\mathrm{TO}} obtained by replacing DaNaN−1(νN)D^{(\nu_N)}_{a_Na_{N-1}} with its time-ordered part, namely the second term on the right-hand side of (A12), while keeping the other propagators unchanged. In the rest of this subsection, QTO\mathcal{Q}_{\mathrm{TO}} and PTO\mathcal{P}_{\mathrm{TO}} denote the loop integral and integrand of this contribution:

TN,TO=∑a1,…,aN=±δaNaN−1[∏v=1N−1∫−∞0iav dτv (−τv)pveiavEvτv]∫τN−10iaN dτN (−τN)pNeiaNkLτNQTO,(20)\mathcal{T}_{N,\mathrm{TO}}= \sum_{a_1,\ldots,a_N=\pm} \delta_{a_Na_{N-1}} \left[\prod_{v=1}^{N-1}\int_{-\infty}^{0} i a_v\,d\tau_v\,(-\tau_v)^{p_v}e^{i a_v E_v\tau_v}\right] \int_{\tau_{N-1}}^{0} i a_N\,d\tau_N\,(-\tau_N)^{p_N}e^{i a_N k_L\tau_N}\mathcal{Q}_{\mathrm{TO}}, \tag*{(20)}

where the step function has been absorbed into the lower limit, so the τN\tau_N interval is finite for fixed hard vertex times. The loop integral is

QTO≡∫d3q(2π)3PTO(q,kL),(21)\mathcal{Q}_{\mathrm{TO}}\equiv\int\frac{d^3q}{(2\pi)^3}\mathcal{P}_{\mathrm{TO}}(q,k_L), \tag*{(21)}

with

PTO(q,kL)≡Da1aN(ν1)(q;τ1,τN)×[D−aN,aN(νN)(p;τN,τN−1)−DaN,−aN(νN)(p;τN,τN−1)]×∏i=2N−1Dai−1ai(νi)(∣q−∑m=1i−1km∣;τi−1,τi).(22)\mathcal{P}_{\mathrm{TO}}(q,k_L)\equiv D^{(\nu_1)}_{a_1a_N}(q;\tau_1,\tau_N)\times\left[D^{(\nu_N)}_{-a_N,a_N}(p;\tau_N,\tau_{N-1})-D^{(\nu_N)}_{a_N,-a_N}(p;\tau_N,\tau_{N-1})\right] \times\prod_{i=2}^{N-1}D^{(\nu_i)}_{a_{i-1}a_i}\left(\left\lvert q-\sum_{m=1}^{i-1}k_m\right\rvert;\tau_{i-1},\tau_i\right). \tag*{(22)}

At fixed hard vertex times, the factor eiaNkLτNe^{i a_N k_L\tau_N} in (A13) admits a Taylor expansion in −kLτN-k_L\tau_N uniformly over the finite τN\tau_N interval and hence supplies only nonnegative integer powers of kLk_L. The remaining hard time integrations are performed term by term in this series, first in a common convergence domain and then by analytic continuation, preserving this integer power structure. Any nonanalytic dependence on kLk_L in this limit must therefore arise from the loop integral QTO\mathcal{Q}_{\mathrm{TO}}. We next analyse the analytic properties of QTO\mathcal{Q}_{\mathrm{TO}} in the soft limit kL→0k_L\to0 at fixed vertex times.

Following the method of regions, the asymptotic expansion as kL/kS→0k_L/k_S\to0 takes the form QTO∼QTO,1+QTO,2+QTO,3\mathcal{Q}_{\mathrm{TO}}\sim\mathcal{Q}_{\mathrm{TO},1}+\mathcal{Q}_{\mathrm{TO},2}+\mathcal{Q}_{\mathrm{TO},3}, where the three contributions correspond to the three momentum regions described below. Each contribution is defined by expanding PTO\mathcal{P}_{\mathrm{TO}} in the corresponding region and integrating term by term over all loop momentum space in dimensional regularisation.

(i) All internal lines hard. As all internal momentum magnitudes remain nonzero in this region, PTO\mathcal{P}_{\mathrm{TO}} admits an ordinary Taylor expansion in kLk_L:

PTO(q,kL)∼∑n=0∞1n![(kL⋅∇k)nPTO(q,k)]k=0,(23)\mathcal{P}_{\mathrm{TO}}(q,k_L)\sim\sum_{n=0}^{\infty}\frac{1}{n!}\left[(k_L\cdot\nabla_k)^n\mathcal{P}_{\mathrm{TO}}(q,k)\right]_{k=0}, \tag*{(23)}

so the integral over qq can also be expanded in kLk_L as a Taylor series

QTO,1∼∑n=0∞1n!∫d3q(2π)3[(kL⋅∇k)nPTO(q,k)]k=0,(24)\mathcal{Q}_{\mathrm{TO},1}\sim\sum_{n=0}^{\infty}\frac{1}{n!}\int\frac{d^3q}{(2\pi)^3}\left[(k_L\cdot\nabla_k)^n\mathcal{P}_{\mathrm{TO}}(q,k)\right]_{k=0}, \tag*{(24)}

where every order contains only integer powers of the soft momentum kLk_L.

(ii) One internal line soft. Since p−q=kLp-q=k_L, soft qq necessarily implies soft pp, and vice versa, so neither can be the isolated soft line. Let us then take the jj-th line with 2≤j≤N−12\le j\le N-1 to be soft and shift the loop momentum to

ℓ=q−∑i=1j−1ki,ℓ=∣ℓ∣,(25)\ell=q-\sum_{i=1}^{j-1}k_i,\qquad\ell=\lvert\ell\rvert, \tag*{(25)}

so the loop integrand becomes

PTO(ℓ+∑m=1j−1km,kL)=Daj−1,−aj(νj)(ℓ;τj−1,τj)Da1aN(ν1)(∣ℓ+∑m=1j−1km∣;τ1,τN)\mathcal{P}_{\mathrm{TO}}\left(\ell+\sum_{m=1}^{j-1}k_m,k_L\right)=D_{a_{j-1,-}a_j}^{(\nu_j)}\left(\ell;\tau_{j-1},\tau_j\right)D_{a_1a_N}^{(\nu_1)}\left(\left|\ell+\sum_{m=1}^{j-1}k_m\right|;\tau_1,\tau_N\right)
×[D−aN,aN(νN)(∣ℓ+kL+∑m=1j−1km∣;τN,τN−1)−DaN,−aN(νN)(∣ℓ+kL+∑m=1j−1km∣;τN,τN−1)]{}\times\left[D_{-a_N,a_N}^{(\nu_N)}\left(\left|\ell+k_L+\sum_{m=1}^{j-1}k_m\right|;\tau_N,\tau_{N-1}\right)-D_{a_N,-a_N}^{(\nu_N)}\left(\left|\ell+k_L+\sum_{m=1}^{j-1}k_m\right|;\tau_N,\tau_{N-1}\right)\right]
×∏i=2i≠jN−1Dai−1ai(νi)(∣ℓ+∑m=1j−1km−∑m=1i−1km∣;τi−1,τi).(26){}\times\prod_{\substack{i=2\\i\ne j}}^{N-1}D_{a_{i-1}a_i}^{(\nu_i)}\left(\left|\ell+\sum_{m=1}^{j-1}k_m-\sum_{m=1}^{i-1}k_m\right|;\tau_{i-1},\tau_i\right). \tag*{(26)}

All factors except the jj-th propagator remain hard and admit Taylor expansions in ℓ\ell, as in (A16), while the soft jj-th propagator has a Frobenius expansion, for example,

D+(νj)(ℓ;τj−1,τj)=−∑c=±1e−cπνjℓ2icνjfc(νj)(ℓ,τj−1)fc(νj)(ℓ,τj)D_{+}^{(\nu_j)}(\ell;\tau_{j-1},\tau_j)=-\sum_{c=\pm1}e^{-c\pi\nu_j}\ell^{2ic\nu_j}f_c^{(\nu_j)}(\ell,\tau_{j-1})f_c^{(\nu_j)}(\ell,\tau_j)
D+(νj)(ℓ;τj−1,τj)−∑c=±1ecπνjfc(νj)(ℓ,τj−1)f−c(νj)(ℓ,τj).(27)\phantom{D_{+}^{(\nu_j)}(\ell;\tau_{j-1},\tau_j)}-\sum_{c=\pm1}e^{c\pi\nu_j}f_{c}^{(\nu_j)}(\ell,\tau_{j-1})f_{-c}^{(\nu_j)}(\ell,\tau_j). \tag*{(27)}

Here fc(νj)f_c^{(\nu_j)} is given by (A2) with ν=νj\nu=\nu_j, and the superscript labels the mass parameter. All four SK propagators contain the same powers 1,ℓ±2iνj1,\ell^{\pm2i\nu_j} with coefficients analytic in ℓ2\ell^2. Combining this Frobenius expansion with the Taylor expansions of the hard factors gives

PTO(ℓ+∑m=1j−1km,kL)∼∑α∈{0,±2iνj}∑n=0∞ℓαPj,α(n)(ℓ,kL),(28)\mathcal{P}_{\mathrm{TO}}\left(\ell+\sum_{m=1}^{j-1}k_m,k_L\right)\sim\sum_{\alpha\in\{0,\pm2i\nu_j\}}\sum_{n=0}^{\infty}\ell^\alpha P_{j,\alpha}^{(n)}(\ell,k_L), \tag*{(28)}

where Pj,α(n)P_{j,\alpha}^{(n)} is a homogeneous polynomial of degree nn in ℓ\ell. Each term therefore gives a scaleless integral over all loop momentum space, which vanishes in dimensional regularisation. Thus this region contributes zero order by order:

QTO,2=∑j=2N−1∫d3ℓ(2π)3∑α∈{0,±2iνj}∑n=0∞ℓαPj,α(n)(ℓ,kL)=0.(29)Q_{\mathrm{TO},2}=\sum_{j=2}^{N-1}\int\frac{\mathrm{d}^3\ell}{(2\pi)^3}\sum_{\alpha\in\{0,\pm2i\nu_j\}}\sum_{n=0}^{\infty}\ell^\alpha P_{j,\alpha}^{(n)}(\ell,k_L)=0. \tag*{(29)}

(iii) Two internal lines soft. The only allowed pair is q,p=O(kL)q,p=\mathcal{O}(k_L), any other pair, or three or more soft lines, would violate (A10). Using pp as the loop variable gives

PTO(p−kL,kL)=Da1aN(ν1)(∣p−kL∣;τ1,τN)[D−aN,aN(νN)(p;τN,τN−1)−DaN,−aN(νN)(p;τN,τN−1)]\mathcal{P}_{\mathrm{TO}}(p-k_L,k_L)=D_{a_1a_N}^{(\nu_1)}\left(|p-k_L|;\tau_1,\tau_N\right)\left[D_{-a_N,a_N}^{(\nu_N)}(p;\tau_N,\tau_{N-1})-D_{a_N,-a_N}^{(\nu_N)}(p;\tau_N,\tau_{N-1})\right]
×∏i=2N−1Dai−1ai(νi)(∣p−kL−∑m=1i−1km∣;τi−1,τi).(30){}\times\prod_{i=2}^{N-1}D_{a_{i-1}a_i}^{(\nu_i)}\left(\left|p-k_L-\sum_{m=1}^{i-1}k_m\right|;\tau_{i-1},\tau_i\right). \tag*{(30)}

Using (A1), we obtain

D−aN,aN(νN)(p;τN,τN−1)−DaN,−aN(νN)(p;τN,τN−1)=−2aNsinh⁡(πνN)[f+(νN)(p,τN)f−(νN)(p,τN−1)−f−(νN)(p,τN)f+(νN)(p,τN−1)],(31)D_{-a_N,a_N}^{(\nu_N)}(p;\tau_N,\tau_{N-1})-D_{a_N,-a_N}^{(\nu_N)}(p;\tau_N,\tau_{N-1}) =-2a_N\sinh(\pi\nu_N)\left[f_+^{(\nu_N)}(p,\tau_N)f_-^{(\nu_N)}(p,\tau_{N-1})-f_-^{(\nu_N)}(p,\tau_N)f_+^{(\nu_N)}(p,\tau_{N-1})\right], \tag*{(31)}

which admits a Taylor expansion in p2p^2 at the soft limit. As in (A20), the qq-line propagator contains the powers qα=∣p−kL∣αq^\alpha=|p-k_L|^\alpha, with α∈{0,±2iν1}\alpha\in\{0,\pm2i\nu_1\}, multiplied by coefficients analytic in q2q^2. After factoring out these powers, all remaining factors admit joint Taylor expansions in both pp and kLk_L at fixed times. Combining these expansions gives

PTO(p−kL,kL)∼∑α∈{0,±2iν1}∑n=0∞∣p−kL∣αQα(n)(p,kL),(32)\mathcal{P}_{\mathrm{TO}}(p-k_L,k_L)\sim\sum_{\alpha\in\{0,\pm2i\nu_1\}}\sum_{n=0}^{\infty}|p-k_L|^\alpha Q_\alpha^{(n)}(p,k_L), \tag*{(32)}

where Qα(n)Q_{\alpha}^{(n)} is a homogeneous polynomial of total degree nn in p\mathbf{p} and kL\mathbf{k}_L. As in region (ii), this region contributes zero order by order:

QTO,3=∑α∈{0,±2iν1}∑n=0∞∫d3p(2π)3Qα(n)(p,kL)∣p−kL∣α=∑α∈{0,±2iν1}∑n=0∞∫d3q(2π)3Q~α(n)(q,kL)qα=0.(33)\begin{aligned} Q_{\mathrm{TO},3}=\sum_{\alpha\in\{0,\pm2i\nu_1\}}\sum_{n=0}^{\infty}\int\frac{\mathrm{d}^3p}{(2\pi)^3}Q_{\alpha}^{(n)}(\mathbf{p},\mathbf{k}_L)|\mathbf{p}-\mathbf{k}_L|^{\alpha} \\ &=\sum_{\alpha\in\{0,\pm2i\nu_1\}}\sum_{n=0}^{\infty}\int\frac{\mathrm{d}^3q}{(2\pi)^3}\widetilde{Q}_{\alpha}^{(n)}(\mathbf{q},\mathbf{k}_L)q^{\alpha}=0. \tag*{(33)} \end{aligned}

In the second line we shifted q=p−kL\mathbf{q}=\mathbf{p}-\mathbf{k}_L and defined Q~α(n)(q,kL)≡Qα(n)(q+kL,kL)\widetilde{Q}_{\alpha}^{(n)}(\mathbf{q},\mathbf{k}_L)\equiv Q_{\alpha}^{(n)}(\mathbf{q}+\mathbf{k}_L,\mathbf{k}_L). Since Q~α(n)\widetilde{Q}_{\alpha}^{(n)} remains polynomial, each monomial multiplied by qαq^{\alpha} gives a scaleless integral over q\mathbf{q}, which vanishes in dimensional regularisation. Therefore, QTO∼QTO,1+QTO,2+QTO,3Q_{\mathrm{TO}}\sim Q_{\mathrm{TO},1}+Q_{\mathrm{TO},2}+Q_{\mathrm{TO},3} contains only terms analytic in kL\mathbf{k}_L at fixed vertex times. Thus TN,TO\mathcal{T}_{N,\mathrm{TO}} carries no CC signal. The same argument applies to the other time-ordered terms, establishing the validity of the directional cut.

Matching to auxiliary bubbles

We now focus on the CC signal and apply the directional cut to the integrand in (A8):

Qcut≡∫d3q(2π)3Pcut,Q_{\mathrm{cut}}\equiv\int\frac{\mathrm{d}^3q}{(2\pi)^3}\mathcal{P}_{\mathrm{cut}},
Pcut=D−aN,aN(ν1)(q;τ1,τN)DaN,−aN(νN)(p;τN,τN−1)∏i=2N−1Dai−1ai(νi)(∣q−∑ℓ=1i−1kℓ∣;τi−1,τi).(34)\mathcal{P}_{\mathrm{cut}}=D_{-a_N,a_N}^{(\nu_1)}(q;\tau_1,\tau_N)D_{a_N,-a_N}^{(\nu_N)}(p;\tau_N,\tau_{N-1})\prod_{i=2}^{N-1}D_{a_{i-1}a_i}^{(\nu_i)}\left(\left|\mathbf{q}-\sum_{\ell=1}^{i-1}\mathbf{k}_{\ell}\right|;\tau_{i-1},\tau_i\right). \tag*{(34)}

The analysis of Sec. A 1 also applies to the cut contribution: the all hard region is analytic in the soft momentum, while an isolated soft line gives scaleless integrals order by order. Thus only the region p,q=O(kL)p,q=O(k_L) can contribute a CC signal.

We expand the dependence on the soft internal momenta at fixed hard kinematic parameters, retaining the hard external phases eiavEvτve^{ia_vE_v\tau_v} in the time integrals. The hard propagators are Taylor expanded in their soft momentum shifts, with their full dependence on the hard momenta and vertex times retained.

For definiteness, take aN=+a_N=+ and write uνju_{\nu_j} for the mode with mass parameter νj\nu_j. Using (A1) and (A2), we expand the two modes at τ1\tau_1 and τN−1\tau_{N-1} while keeping both modes at τN\tau_N exact. The cut loop integrand then has the double expansion

Pcut∣aN=+∼∑c1,c2=±∑m,n=0∞Pc1c2(m,n)(q,p;τ1,…,τN−1)×qic1ν1pic2νN×uν1∗(q,τN)uνN∗(p,τN).(35)\left.\mathcal{P}_{\mathrm{cut}}\right|_{a_N=+}\sim\sum_{c_1,c_2=\pm}\sum_{m,n=0}^{\infty}P_{c_1c_2}^{(m,n)}(\mathbf{q},\mathbf{p};\tau_1,\ldots,\tau_{N-1})\times q^{ic_1\nu_1}p^{ic_2\nu_N}\times u_{\nu_1}^{*}(q,\tau_N)u_{\nu_N}^{*}(p,\tau_N). \tag*{(35)}

Here Pc1c2(m,n)P_{c_1c_2}^{(m,n)} is homogeneous of degree mm in q\mathbf{q} and degree nn in p\mathbf{p}. Suppressing Cartesian indices on its coefficient tensors, we write

Pc1c2(m,n)(q,p;τ1,…,τN−1)≡cc1c2(m,n)(τ1,…,τN−1)qi1⋯qimpj1⋯pjn.(36)P_{c_1c_2}^{(m,n)}(\mathbf{q},\mathbf{p};\tau_1,\ldots,\tau_{N-1})\equiv c_{c_1c_2}^{(m,n)}(\tau_1,\ldots,\tau_{N-1})q_{i_1}\cdots q_{i_m}p_{j_1}\cdots p_{j_n}. \tag*{(36)}

The contraction of the suppressed indices of cc1c2(m,n)c_{c_1c_2}^{(m,n)} with the displayed momentum tensors is understood. The coefficients are symmetric within each group of mm and nn indices. Their hard time dependence is explicit, while their dependence on the hard kinematics and SK indices is suppressed. Using fc(ν)∗=−ecπνf−c(ν)f_c^{(\nu)*}=-e^{c\pi\nu}f_{-c}^{(\nu)} and relabelling the branches, the aN=−a_N=- expansion has the same coefficients c(m,n)c^{(m,n)}, with the soft mode pair replaced by e−π(c1ν1+c2νN)uν1(q,τN)uνN(p,τN)e^{-\pi(c_1\nu_1+c_2\nu_N)}u_{\nu_1}(q,\tau_N)u_{\nu_N}(p,\tau_N).

Denoting the cut contribution from the double soft region by TN,cut\mathcal{T}_{N,\mathrm{cut}}, we substitute (A28) and (A29) into (A7) and sum over aN=±a_N=\pm to obtain

TN,cut=∑a1,…,aN=±∑c1,c2=±∑m,n=0∞[∏v=1N−1∫−∞0iavdτv(−τv)pveiavEvτvcc1c2(m,n)(τ1,…,τN−1)]×[i∫−∞0dτN(−τN)pN∫d3q(2π)3qi1⋯qimpj1⋯pjnqic1ν1pic2νN\begin{aligned} \mathcal{T}_{N,\mathrm{cut}}=\sum_{a_1,\ldots,a_N=\pm}\sum_{c_1,c_2=\pm}\sum_{m,n=0}^{\infty}\left[\prod_{v=1}^{N-1}\int_{-\infty}^{0} i a_v d\tau_v(-\tau_v)^{p_v}e^{i a_v E_v\tau_v}c_{c_1c_2}^{(m,n)}(\tau_1,\ldots,\tau_{N-1})\right] \\ &\times\left[i\int_{-\infty}^{0}d\tau_N(-\tau_N)^{p_N}\int\frac{d^3q}{(2\pi)^3}q_{i_1}\cdots q_{i_m}p_{j_1}\cdots p_{j_n}q^{i c_1\nu_1}p^{i c_2\nu_N} \right. \end{aligned}
×(eikLτNuν1∗(q,τN)uνN∗(p,τN)−e−π(c1ν1+c2νN)e−ikLτNuν1(q,τN)uνN(p,τN))].(37)\left. \times\left(e^{i k_L\tau_N}u_{\nu_1}^{*}(q,\tau_N)u_{\nu_N}^{*}(p,\tau_N)-e^{-\pi(c_1\nu_1+c_2\nu_N)}e^{-i k_L\tau_N}u_{\nu_1}(q,\tau_N)u_{\nu_N}(p,\tau_N)\right)\right]. \tag*{(37)}

Rescaling q→kLqq\to k_Lq and τN→τN/kL\tau_N\to\tau_N/k_L shows that the loop integral together with the τN\tau_N integral in square brackets scales as kLm+n−pN−1+i(c1ν1+c2νN)k_L^{m+n-p_N-1+i(c_1\nu_1+c_2\nu_N)}. Thus (c1,c2)(c_1,c_2) fixes the oscillation frequency to be ∣c1ν1+c2νN∣|c_1\nu_1+c_2\nu_N|, while m+nm+n counts the additional integer powers of kLk_L.

To match (A30), consider the auxiliary bubble defined, up to overall normalisation and symmetry factors, by

T2(r,s)≡∑a1,aN=±∫−∞0ia1dτ1(−τ1)r+s−2eia1Eτ1∫−∞0iaNdτN(−τN)pNeiaNkLτN×∫d3q(2π)3qα1⋯qαrpβ1⋯pβs×Da1aN(ν1)(q;τ1,τN)DaNa1(νN)(p;τN,τ1).(38)\begin{aligned} \mathcal{T}_2^{(r,s)}\equiv\sum_{a_1,a_N=\pm}\int_{-\infty}^{0}i a_1d\tau_1(-\tau_1)^{r+s-2}e^{i a_1E\tau_1}\int_{-\infty}^{0}i a_Nd\tau_N(-\tau_N)^{p_N}e^{i a_Nk_L\tau_N} \\ &\times\int\frac{d^3q}{(2\pi)^3}q_{\alpha_1}\cdots q_{\alpha_r}p_{\beta_1}\cdots p_{\beta_s}\times D_{a_1a_N}^{(\nu_1)}(q;\tau_1,\tau_N)D_{a_Na_1}^{(\nu_N)}(p;\tau_N,\tau_1). \tag*{(38)} \end{aligned}

Here the spatial indices α1,⋯ ,αr,β1,⋯ ,βs\alpha_1,\cdots,\alpha_r,\beta_1,\cdots,\beta_s of T2(r,s)\mathcal{T}_2^{(r,s)} are suppressed. The auxiliary bubble is obtained by replacing the original hard subgraph and all external legs attached to it by a single local (but noncovariant) vertex −gϕ2O(r,s)\sqrt{-g}\phi^2\mathcal{O}^{(r,s)}. The two heavy propagators and the original soft vertex are retained. Here ϕ\phi is an auxiliary conformally coupled scalar whose two hard external legs carry momenta K1K_1 and K2K_2 satisfying

K1+K2=E≫kL,K1+K2=−kL,(39)K_1+K_2=E\gg k_L,\qquad K_1+K_2=-k_L, \tag*{(39)}

and the local noncovariant heavy field bilinear, with its spatial indices suppressed, is

O(r,s)=(−i)r+isar+s(τ1)(∂α1⋯∂αrσ1)(∂β1⋯∂βsσN).(40)\mathcal{O}^{(r,s)}=\frac{(-i)^{r+i s}}{a^{r+s}(\tau_1)}(\partial_{\alpha_1}\cdots\partial_{\alpha_r}\sigma_1)(\partial_{\beta_1}\cdots\partial_{\beta_s}\sigma_N). \tag*{(40)}

Similarly, the CC signal of T2(r,s)\mathcal{T}_2^{(r,s)} can be extracted using the directional cutting rule of Sec. A 1. Denoting its cut contribution from the double soft region by T2,cut(r,s)\mathcal{T}_{2,\mathrm{cut}}^{(r,s)}, we obtain

T2,cut(r,s)=∑a1=±∑m,n=0∞∑c1,c2=±∫−∞0ia1dτ1(−τ1)r+s−2eia1Eτ1dc1c2(r,s;m,n)(τ1)×[i∫−∞0dτN(−τN)pN∫d3q(2π)3qi1⋯qimpj1⋯pjnqic1ν1pic2νN\begin{aligned} \mathcal{T}_{2,\mathrm{cut}}^{(r,s)}=\sum_{a_1=\pm}\sum_{m,n=0}^{\infty}\sum_{c_1,c_2=\pm}\int_{-\infty}^{0}i a_1d\tau_1(-\tau_1)^{r+s-2}e^{i a_1E\tau_1}d_{c_1c_2}^{(r,s;m,n)}(\tau_1) \\ &\times\left[i\int_{-\infty}^{0}d\tau_N(-\tau_N)^{p_N}\int\frac{d^3q}{(2\pi)^3}q_{i_1}\cdots q_{i_m}p_{j_1}\cdots p_{j_n}q^{i c_1\nu_1}p^{i c_2\nu_N} \right. \end{aligned}
×(eikLτNuν1∗(q,τN)uνN∗(p,τN)−e−π(c1ν1+c2νN)e−ikLτNuν1(q,τN)uνN(p,τN))].(41)\left. \times\left(e^{i k_L\tau_N}u_{\nu_1}^{*}(q,\tau_N)u_{\nu_N}^{*}(p,\tau_N)-e^{-\pi(c_1\nu_1+c_2\nu_N)}e^{-i k_L\tau_N}u_{\nu_1}(q,\tau_N)u_{\nu_N}(p,\tau_N)\right)\right]. \tag*{(41)}

where, for fixed rr, ss, the coefficient tensors dc1c2(r,s;m,n)d_{c_1c_2}^{(r,s;m,n)} are defined by

qα1⋯qαrpβ1⋯pβsfc1(ν1)(q,τ1)fc2(νN)(p,τ1)≡∑m,n=0∞dc1c2(r,s;m,n)(τ1)qi1⋯qimpj1⋯pjn.(42)q_{\alpha_1}\cdots q_{\alpha_r}p_{\beta_1}\cdots p_{\beta_s}f_{c_1}^{(\nu_1)}(q,\tau_1)f_{c_2}^{(\nu_N)}(p,\tau_1)\equiv\sum_{m,n=0}^{\infty}d_{c_1c_2}^{(r,s;m,n)}(\tau_1)q_{i_1}\cdots q_{i_m}p_{j_1}\cdots p_{j_n}. \tag*{(42)}

For each nonzero frequency branch (c1,c2)(c_1,c_2), we determine gc1c2(r,s)g_{c_1c_2}^{(r,s)} by matching the hard time coefficients in (A30) and (A34) order by order in kLk_L, before evaluating the common loop and τN\tau_N integrals. In particular, matching the leading hard time coefficients fixes gc1c2(0,0)g_{c_1c_2}^{(0,0)} through

gc1c2(0,0)∑a1=±ia1∫−∞0dτ1(−τ1)−2eia1Eτ1dc1c2(0,0;0,0)(τ1)=∑a1,…,aN−1=±[∏v=1N−1∫−∞0iavdτv(−τv)pveiavEvτv]cc1c2(0,0)(τ1,…,τN−1).(43)\begin{aligned} g_{c_1c_2}^{(0,0)}\sum_{a_1=\pm}i a_1\int_{-\infty}^{0}d\tau_1(-\tau_1)^{-2}e^{i a_1E\tau_1}d_{c_1c_2}^{(0,0;0,0)}(\tau_1) \\ &=\sum_{a_1,\ldots,a_{N-1}=\pm}\left[\prod_{v=1}^{N-1}\int_{-\infty}^{0}i a_vd\tau_v(-\tau_v)^{p_v}e^{i a_vE_v\tau_v}\right]c_{c_1c_2}^{(0,0)}(\tau_1,\ldots,\tau_{N-1}). \tag*{(43)} \end{aligned}

(A36) matches the leading term in TN,cut\mathcal{T}_{N,\mathrm{cut}} carrying the oscillatory factor (kL/kS)i(c1ν1+c2νN)(k_L/k_S)^{i(c_1\nu_1+c_2\nu_N)} to the corresponding term in gc1c2(0,0)T2,cut(0,0)g_{c_1c_2}^{(0,0)}\mathcal{T}_{2,\mathrm{cut}}^{(0,0)}, generated by the hard vertex −ggc1c2(0,0)ϕ2O(0,0)\sqrt{-g}g_{c_1c_2}^{(0,0)}\phi_2\mathcal{O}^{(0,0)}. This auxiliary bubble satisfies case (i) of the exclusion rule, so its CC signal at this frequency vanishes. The matched term in TN,cut\mathcal{T}_{N,\mathrm{cut}} therefore also vanishes.

The same coefficient matching extends order by order in kLk_L. Denoting the corresponding terms at fixed (c1,c2)(c_1,c_2) by TN,cut;c1c2\mathcal{T}_{N,\mathrm{cut};c_1c_2} and T2,cut;c1c2(r,s)\mathcal{T}_{2,\mathrm{cut};c_1c_2}^{(r,s)}, we obtain

TN,cut;c1c2=∑r,s=0∞gc1c2(r,s)T2,cut;c1c2(r,s).(44)\mathcal{T}_{N,\mathrm{cut};c_1c_2}=\sum_{r,s=0}^{\infty}g_{c_1c_2}^{(r,s)}\mathcal{T}_{2,\mathrm{cut};c_1c_2}^{(r,s)}. \tag*{(44)}

Every auxiliary bubble T2(r,s)\mathcal{T}_2^{(r,s)} satisfies case (i) of the exclusion rule, so its CC signal vanishes. The matching relation in (A37) then implies that the NN-gon CC signal vanishes to all soft orders.

Triangle with a chemical potential

For products and ratios of Gamma functions in this section, we use the shorthand notations

Γ[a1,…,an]≡∏i=1nΓ(ai),Γ[a1,…,anb1,…,bm]≡∏i=1nΓ(ai)∏j=1mΓ(bj).(45)\Gamma[a_1,\ldots,a_n]\equiv\prod_{i=1}^{n}\Gamma(a_i),\qquad \Gamma\left[\begin{array}{c}a_1,\ldots,a_n\\b_1,\ldots,b_m\end{array}\right]\equiv\frac{\prod_{i=1}^{n}\Gamma(a_i)}{\prod_{j=1}^{m}\Gamma(b_j)}. \tag*{(45)}

For ν>0\nu>0, the normalised BD solution of (8) is

v(k,τ)=eπκ/22kW−iκ,iν(2ikτ),(46)v(k,\tau)=\frac{e^{\pi\kappa/2}}{\sqrt{2k}}W_{-i\kappa,i\nu}(2ik\tau), \tag*{(46)}

up to an irrelevant constant phase. The physical mode is then u(k,τ)=v(k,τ)/a(τ)u(k,\tau)=v(k,\tau)/a(\tau), with a(τ)=−1/τa(\tau)=-1/\tau. In particular, its leading late-time expansion can be written as

lim⁡−kτ→0u(k,τ)=e−iπ/4eπκ/2k3[C+(κ,ν)(−kτ)3/2+iν+C−(κ,ν)(−kτ)3/2−iν].(47)\lim_{-k\tau\to0}u(k,\tau)=e^{-i\pi/4}\frac{e^{\pi\kappa/2}}{\sqrt{k^3}}\left[C_+(\kappa,\nu)(-k\tau)^{3/2+i\nu}+C_-(\kappa,\nu)(-k\tau)^{3/2-i\nu}\right]. \tag*{(47)}

where the coefficients C±(κ,ν)C_\pm(\kappa,\nu) are given by

C+(κ,ν)=2iνeπν/2Γ(−2iν)Γ(12−iν+iκ),C−(κ,ν)=C+(κ,−ν).(48)C_+(\kappa,\nu)=\frac{2^{i\nu}e^{\pi\nu/2}\Gamma(-2i\nu)}{\Gamma\left(\frac{1}{2}-i\nu+i\kappa\right)},\qquad C_-(\kappa,\nu)=C_+(\kappa,-\nu). \tag*{(48)}

For κ=0\kappa=0, uu reduces to the Hankel mode in (A1) up to a constant phase. The SK propagators are given by (A3)–(A6) with this mode, and their dependence on (κ,ν)(\kappa,\nu) is suppressed. For the conformally coupled external scalar, the boundary propagator is

Ka(k,τ)=τfτ2keiakτ,a=±1,(49)K_a(k,\tau)=\frac{\tau_f\tau}{2k}e^{iak\tau},\qquad a=\pm1, \tag*{(49)}

to leading order as τf→0\tau_f\to0.

Three insertions of the interaction in (7) give

⟨ϕk1ϕk2ϕk3⟩τf′=λ3τf38k1k2k3∑a1,a2,a3=±1[∏ν=13∫−∞τf−iaν dτντν3eiaνkντν]∫d3q(2π)3Da1a3(q;τ1,τ3)Da3a2(∣q+k3∣;τ3,τ2)Da1a2(∣q−k1∣;τ1,τ2).(50)\left\langle\phi_{k_1}\phi_{k_2}\phi_{k_3}\right\rangle'_{\tau_f} =\frac{\lambda^3\tau_f^3}{8k_1k_2k_3} \sum_{a_1,a_2,a_3=\pm1} \left[\prod_{\nu=1}^{3}\int_{-\infty}^{\tau_f}\frac{-ia_\nu\,d\tau_\nu}{\tau_\nu^3}e^{ia_\nu k_\nu\tau_\nu}\right] \int\frac{d^3q}{(2\pi)^3}D_{a_1a_3}(q;\tau_1,\tau_3)D_{a_3a_2}(|q+k_3|;\tau_3,\tau_2)D_{a_1a_2}(|q-k_1|;\tau_1,\tau_2). \tag*{(50)}

where k1+k2+k3=0k_1+k_2+k_3=0 and the momentum convention agrees with Figure 1(d). It is straightforward to show that the directional cutting rule of Sec. A 1 also extracts the leading CC signal for Whittaker modes. In the squeezed limit k3=kL≪kS=k1+k2k_3=k_L\ll k_S=k_1+k_2, we repeat the steps leading to (A30) for $N=3,ablytyped using the Whittaker mode and its late-time expansion in (B2) and (B3). Retaining the leading terms m=n=0m=n=0 with equal sign branches, setting k1=k2=kS/2k_1=k_2=k_S/2 at this order, and using the normalisation in (9), we obtain

lim⁡kL/kS→0BCC=e2πκkS24kL2∑a1,a2=±1(−a1a2)∫−∞0dτ1dτ2(−τ1)−3/2+iν(−τ2)−3/2+iνDa1a2(kS/2;τ1,τ2)eikS(a1τ1+a2τ2)/2×[C+2(κ,ν)∫−∞0−dτ3τ3eikLτ3∫d3q(2π)3q−1/2+iν∣q+kL∣−1/2+iνWiκ,iν(−2iqτ3)Wiκ,iν(−2i∣q+kL∣τ3)+C+∗2(κ,−ν)∫−∞0−dτ3τ3e−ikLτ3∫d3q(2π)3q−1/2+iν∣q+kL∣−1/2+iνW−iκ,iν(2iqτ3)W−iκ,iν(2i∣q+kL∣τ3)]+c.c.(51)\begin{aligned} \lim_{k_L/k_S\to0} B_{\mathrm{CC}}={}&\frac{e^{2\pi\kappa}k_S^2}{4k_L^2}\sum_{a_1,a_2=\pm1}(-a_1a_2)\int_{-\infty}^{0}d\tau_1d\tau_2(-\tau_1)^{-3/2+i\nu}(-\tau_2)^{-3/2+i\nu}D_{a_1a_2}(k_S/2;\tau_1,\tau_2)e^{ik_S(a_1\tau_1+a_2\tau_2)/2}\\ &\times\left[C_+^2(\kappa,\nu)\int_{-\infty}^{0}\frac{-d\tau_3}{\tau_3}e^{ik_L\tau_3}\int\frac{d^3q}{(2\pi)^3}q^{-1/2+i\nu}|q+k_L|^{-1/2+i\nu}W_{i\kappa,i\nu}(-2iq\tau_3)W_{i\kappa,i\nu}(-2i|q+k_L|\tau_3)\right.\\ &\left.\qquad+C_+^{*2}(\kappa,-\nu)\int_{-\infty}^{0}\frac{-d\tau_3}{\tau_3}e^{-ik_L\tau_3}\int\frac{d^3q}{(2\pi)^3}q^{-1/2+i\nu}|q+k_L|^{-1/2+i\nu}W_{-i\kappa,i\nu}(2iq\tau_3)W_{-i\kappa,i\nu}(2i|q+k_L|\tau_3)\right]+\mathrm{c.c.} \tag*{(51)} \end{aligned}

The hard time factor in (B7) is a tree-level Whittaker two-point seed integral. Following [18], we define its dimensionless form and quote the corresponding closed-form result:

I(κ,ν)≡kS2+2iν∑a1,a2=±1(−a1a2)∫−∞0dτ1dτ2(−τ1)−3/2+iν(−τ2)−3/2+iνDa1a2(kS/2;τ1,τ2)eikS(a1τ1+a2τ2)/2=Γ(1+2iν)(ieiπν12+iν−iκ−ie−iπν12+iν+iκ)+eπκ(12+iν−iκ)(12+iν+iκ)Γ[2+2iν,2+2iν32+iν−iκ,32+iν+iκ].(52)\begin{aligned} \mathcal{I}(\kappa,\nu)&\equiv k_S^{2+2i\nu}\sum_{a_1,a_2=\pm1}(-a_1a_2)\int_{-\infty}^{0}d\tau_1d\tau_2(-\tau_1)^{-3/2+i\nu}(-\tau_2)^{-3/2+i\nu}D_{a_1a_2}(k_S/2;\tau_1,\tau_2)e^{ik_S(a_1\tau_1+a_2\tau_2)/2}\\ &=\Gamma(1+2i\nu)\left(\frac{ie^{i\pi\nu}}{\frac{1}{2}+i\nu-i\kappa}-\frac{ie^{-i\pi\nu}}{\frac{1}{2}+i\nu+i\kappa}\right)+\frac{e^{\pi\kappa}}{\left(\frac{1}{2}+i\nu-i\kappa\right)\left(\frac{1}{2}+i\nu+i\kappa\right)}\Gamma\left[\begin{matrix}2+2i\nu,2+2i\nu\\ \frac{3}{2}+i\nu-i\kappa,\frac{3}{2}+i\nu+i\kappa\end{matrix}\right]. \tag*{(52)} \end{aligned}

It remains to evaluate the two combined time and loop integrals in the square brackets of (B7). For later convenience, we define

J±(κ,ν)≡kL−2−2iν∫−∞0−dτ3τ3e±ikLτ3×∫d3q(2π)3q−1/2+iν∣q+kL∣−1/2+iνW±iκ,iν(∓2iqτ3)W±iκ,iν(∓2i∣q+kL∣τ3).(53)\begin{aligned} \mathcal{J}_{\pm}(\kappa,\nu)&\equiv k_L^{-2-2i\nu}\int_{-\infty}^{0}\frac{-d\tau_3}{\tau_3}e^{\pm ik_L\tau_3}\\ &\quad\times\int\frac{d^3q}{(2\pi)^3}q^{-1/2+i\nu}|q+k_L|^{-1/2+i\nu}W_{\pm i\kappa,i\nu}(\mp2iq\tau_3)W_{\pm i\kappa,i\nu}(\mp2i|q+k_L|\tau_3). \tag*{(53)} \end{aligned}

To evaluate this integral, we use the MB representation of the Whittaker functions:

W±iκ,iν(∓2ikτ)=∫−i∞+i∞ds2πie±iπ/4∓iπs/2(−2kτ)1/2−sK±(s),(54)W_{\pm i\kappa,i\nu}(\mp2ik\tau)=\int_{-i\infty}^{+i\infty}\frac{ds}{2\pi i}e^{\pm i\pi/4\mp i\pi s/2}(-2k\tau)^{1/2-s}\mathcal{K}_{\pm}(s), \tag*{(54)}
K±(s)≡Γ[s−iν,s+iν] 2F1[s−iν, s+iνs∓iκ+12;12],(55)\mathcal{K}_{\pm}(s)\equiv\Gamma[s-i\nu,s+i\nu]\,{}_2F_1\left[\begin{matrix}s-i\nu,\ s+i\nu\\s\mp i\kappa+\frac{1}{2}\end{matrix};\frac{1}{2}\right], \tag*{(55)}

where

2F1[a bc;z]≡1Γ(c)2F1[a bc;z],(56){}_2F_1\left[\begin{matrix}a\ b\\c\end{matrix};z\right]\equiv\frac{1}{\Gamma(c)}{}_2F_1\left[\begin{matrix}a\ b\\c\end{matrix};z\right], \tag*{(56)}

is the regularised hypergeometric function that is analytic for all indices a,b,ca,b,c at z=1/2z=1/2. Substituting the MB representation into (B9), we can perform both the time integral and the loop momentum integral analytically. Writing S≡s1+s2S\equiv s_1+s_2, the time integral gives

∫−∞0dτ3(−τ3)−Se±ikLτ3=∓ie±iπS/2Γ(1−S)kLS−1,(57)\int_{-\infty}^{0}d\tau_3(-\tau_3)^{-S}e^{\pm ik_L\tau_3}=\mp ie^{\pm i\pi S/2}\Gamma(1-S)k_L^{S-1}, \tag*{(57)}

while the loop momentum integral is

∫d3q(2π)3qiν−s1∣q+kL∣iν−s2=kL3+2iν−S(4π)3/2Γ[S−3−2iν2,3+iν−s12,3+iν−s22s1−iν2,s2−iν2,3+iν−S2].(58)\int\frac{d^3q}{(2\pi)^3}q^{i\nu-s_1}|q+k_L|^{i\nu-s_2}=\frac{k_L^{3+2i\nu-S}}{(4\pi)^{3/2}}\Gamma\left[\begin{matrix}\frac{S-3-2i\nu}{2},\frac{3+i\nu-s_1}{2},\frac{3+i\nu-s_2}{2}\\\frac{s_1-i\nu}{2},\frac{s_2-i\nu}{2},3+i\nu-\frac{S}{2}\end{matrix}\right]. \tag*{(58)}

Combining the two results, the dependence on kLk_L factorises completely, leaving a dimensionless double MB integral

J±(κ,ν)=1(4π)3/2∫−i∞+i∞ds1ds2(2πi)221−SK±(s1)K±(s2)Γ[1−S,S−3−2iν2,3+iν−s12,3+iν−s22s1−iν2,s2−iν2,3+iν−S2].(59)\mathcal{J}_{\pm}(\kappa,\nu)=\frac{1}{(4\pi)^{3/2}}\int_{-i\infty}^{+i\infty}\frac{ds_1ds_2}{(2\pi i)^2}2^{1-S}\mathcal{K}_{\pm}(s_1)\mathcal{K}_{\pm}(s_2)\Gamma\left[\begin{matrix}1-S,\frac{S-3-2i\nu}{2},\frac{3+i\nu-s_1}{2},\frac{3+i\nu-s_2}{2}\\\frac{s_1-i\nu}{2},\frac{s_2-i\nu}{2},3+i\nu-\frac{S}{2}\end{matrix}\right]. \tag*{(59)}

The remaining double MB integral can be evaluated analytically. Using the result in our companion work, we obtain the following representation in terms of Meijer-G and hypergeometric functions:

J±(κ,ν)=κΓ[−3−2iν,12+iν±iκ]{±i2−2−2iν±2iκ(1±2iκ)π3/2(12+iν±iκ)Γ[1+2iν2+iν±iκ,1∓2iκ]×[−4κ2 2F1[1−2−2iν∓2iκ−1∓2iκ;−1]+2(12+iν)2+2κ2−(12+iν∓iκ)2(1+2iν)−1∓2iκ]+2−1+2iνπΓ(12−iν∓iκ){[±2i(12+iν±iκ)(−3−2iν)−8κ]×G5,54,3[11232;52+iν+iκ,52+iν−iκ32+iν2+iν2−12−iν∓iκ;52+iν±iκ;1]+4κG5,54,3[1232;52+iν+iκ,52+iν−iκ32+iν2+iν3−12−iν∓iκ;52+iν±iκ;1]}}.(60)\begin{aligned} \mathcal{J}_{\pm}(\kappa,\nu)={}&\kappa\Gamma\left[-3-2i\nu,\frac{1}{2}+i\nu\pm i\kappa\right]\Bigg\{\frac{\pm i2^{-2-2i\nu\pm2i\kappa}(1\pm2i\kappa)}{\pi^{3/2}\left(\frac{1}{2}+i\nu\pm i\kappa\right)}\Gamma\left[\begin{matrix}1+2i\nu\\2+i\nu\pm i\kappa,1\mp2i\kappa\end{matrix}\right]\\ &\times\left[-4\kappa^2\,{}_2F_1\left[\begin{matrix}1-2-2i\nu\mp2i\kappa\\-1\mp2i\kappa\end{matrix};-1\right]+2\left(\frac{1}{2}+i\nu\right)^2+2\kappa^2-\frac{\left(\frac{1}{2}+i\nu\mp i\kappa\right)^2(1+2i\nu)}{-1\mp2i\kappa}\right]\\ &+\frac{2^{-1+2i\nu}}{\pi\Gamma\left(\frac{1}{2}-i\nu\mp i\kappa\right)} \left\{\left[\pm2i\left(\frac{1}{2}+i\nu\pm i\kappa\right)(-3-2i\nu)-8\kappa\right]\right.\\ &\left.\qquad\times G^{4,3}_{5,5}\left[\begin{matrix}1&1&2&\frac{3}{2};\frac{5}{2}+i\nu+i\kappa,\frac{5}{2}+i\nu-i\kappa\\\frac{3}{2}+i\nu&2+i\nu&2-\frac{1}{2}-i\nu\mp i\kappa;\frac{5}{2}+i\nu\pm i\kappa\end{matrix};1\right]\right.\\ &\left.\qquad+4\kappa G^{4,3}_{5,5}\left[\begin{matrix}1&2&\frac{3}{2};\frac{5}{2}+i\nu+i\kappa,\frac{5}{2}+i\nu-i\kappa\\\frac{3}{2}+i\nu&2+i\nu&3-\frac{1}{2}-i\nu\mp i\kappa;\frac{5}{2}+i\nu\pm i\kappa\end{matrix};1\right]\right\}\Bigg\}. \tag*{(60)} \end{aligned}

For numerical evaluation, a simpler representation follows by reducing the two Mellin integrals to a single ordinary integral:

J±(κ,ν)=−i26+2iνπ3Γ[2+2iν,2+2iν,−3−2iν52+iν∓iκ,52+iν∓iκ]∫0∞dzz[(1+4iz)3+2iν−(1−4iz)3+2iν]×(2F1[2+2iν212i52+iν∓iκ;122iz]−2F1[2+2iν212i52+iν∓iκ;12−2iz])2.(61)\begin{aligned} \mathcal{J}_{\pm}(\kappa,\nu)={}&\frac{-i}{2^{6+2i\nu}\pi^3}\Gamma\left[\begin{matrix}2+2i\nu,2+2i\nu,-3-2i\nu\\\frac{5}{2}+i\nu\mp i\kappa,\frac{5}{2}+i\nu\mp i\kappa\end{matrix}\right]\int_{0}^{\infty}\frac{dz}{z}\left[\left(1+\frac{4i}{z}\right)^{3+2i\nu}-\left(1-\frac{4i}{z}\right)^{3+2i\nu}\right]\\ &\times\left( {}_2F_1\left[\begin{matrix}2+2i\nu&2&1&2i\\\frac{5}{2}+i\nu\mp i\kappa;&\frac{1}{2}&\frac{2i}{z}\end{matrix}\right]-{}_2F_1\left[\begin{matrix}2+2i\nu&2&1&2i\\\frac{5}{2}+i\nu\mp i\kappa;&\frac{1}{2}&-\frac{2i}{z}\end{matrix}\right]\right)^2. \tag*{(61)} \end{aligned}

Combining (B7), (B8), and (B9), we finally obtain the leading CC signal

lim⁡kL/kS→0BCC=A(κ,ν)(kLkS)2iν+c.c.,(62)\lim_{k_L/k_S\to0}B_{\rm CC}=\mathcal{A}(\kappa,\nu)\left(\frac{k_L}{k_S}\right)^{2i\nu}+\mathrm{c.c.}, \tag*{(62)}
A(κ,ν)≡e2πκ4I(κ,ν)[C+2(κ,ν)J+(κ,ν)+C+∗2(κ,−ν)J−(κ,ν)].(63)\mathcal{A}(\kappa,\nu)\equiv\frac{e^{2\pi\kappa}}{4}\mathcal{I}(\kappa,\nu)\left[\mathcal{C}_{+}^{2}(\kappa,\nu)\mathcal{J}_{+}(\kappa,\nu)+\mathcal{C}_{+}^{*2}(\kappa,-\nu)\mathcal{J}_{-}(\kappa,\nu)\right]. \tag*{(63)}

The characteristic nonanalytic scaling (kL/kS)2iν\left(k_L/k_S\right)^{2i\nu} exhibits the expected CC oscillation with frequency 2ν2\nu. In particular, (B16) gives

A(0,ν)=0.(64)\mathcal{A}(0,\nu)=0. \tag*{(64)}

Thus the candidate massive late-time scaling remains present in the individual heavy field propagators, while its nonanalytic coefficient vanishes when dS covariance is restored, providing an explicit one-loop realisation of the exclusion rule discussed in the main text.

References

  1. [1]X. Chen and Y. Wang, “Quasi-Single Field Inflation and Non-Gaussianities,” JCAP 1004 (2010) 027, arXiv:0911.3380 [hep-th].DOI
  2. [2]D. Baumann and D. Green, “Signatures of Supersymmetry from the Early Universe,” Phys. Rev. D85 (2012) 103520, arXiv:1109.0292 [hep-th].DOI
  3. [3]T. Noumi, M. Yamaguchi, and D. Yokoyama, “Effective field theory approach to quasi-single field inflation and effects of heavy fields,” JHEP 06 (2013) 051, arXiv:1211.1624 [hep-th].DOI
  4. [4]N. Arkani-Hamed and J. Maldacena, “Cosmological Collider Physics,” arXiv:1503.08043 [hep-th].
  5. [5]N. Arkani-Hamed, D. Baumann, H. Lee, and G. L. Pimentel, “The Cosmological Bootstrap: Inflationary Correlators from Symmetries and Singularities,” JHEP 04 (2020) 105, arXiv:1811.00024 [hep-th].
  6. [6]D. Baumann, C. Duaso Pueyo, A. Joyce, H. Lee, and G. L. Pimentel, “The cosmological bootstrap: weight-shifting operators and scalar seeds,” JHEP 12 (2020) 204, arXiv:1910.14051 [hep-th].DOI
  7. [7]C. Sleight, “A Mellin Space Approach to Cosmological Correlators,” JHEP 01 (2020) 090, arXiv:1906.12302 [hep-th].
  8. [8]C. Sleight and M. Taronna, “Bootstrapping Inflationary Correlators in Mellin Space,” JHEP 02 (2020) 098, arXiv:1907.01143 [hep-th].
  9. [9]C. Sleight and M. Taronna, “From AdS to dS exchanges: Spectral representation, Mellin amplitudes, and crossing,” Phys. Rev. D 104 no. 8, (2021) L081902, arXiv:2007.09993 [hep-th].
  10. [10]E. Pajer, “Building a Boostless Bootstrap for the Bispectrum,” JCAP 01 (2021) 023, arXiv:2010.12818 [hep-th].DOI
  11. [11]S. Jazayeri, E. Pajer, and D. Stefanyszyn, “From locality and unitarity to cosmological correlators,” JHEP 10 (2021) 065, arXiv:2103.08649 [hep-th].DOI
  12. [12]H. Goodhew, S. Jazayeri, and E. Pajer, “The Cosmological Optical Theorem,” JCAP 04 (2021) 021, arXiv:2009.02898 [hep-th].DOI
  13. [13]S. Melville and E. Pajer, “Cosmological Cutting Rules,” JHEP 05 (2021) 249, arXiv:2103.09832 [hep-th].DOI
  14. [14]G. L. Pimentel and D.-G. Wang, “Boostless cosmological collider bootstrap,” JHEP 10 (2022) 177, arXiv:2205.00013 [hep-th].DOI
  15. [15]S. Jazayeri and S. Renaux-Petel, “Cosmological bootstrap in slow motion,” JHEP 12 (2022) 137, arXiv:2205.10340 [hep-th].DOI
  16. [16]D.-G. Wang, G. L. Pimentel, and A. Achúcarro, “Bootstrapping multi-field inflation: non-Gaussianities from light scalars revisited,” JCAP 05 (2023) 043, arXiv:2212.14035 [astro-ph.CO].DOI
  17. [17]L. Qin and Z.-Z. Xianyu, “Helical inflation correlators: partial Mellin-Barnes and bootstrap equations,” JHEP 04 (2023) 059, arXiv:2208.13790 [hep-th].DOI
  18. [18]Z. Qin and Z.-Z. Xianyu, “Closed-form formulae for inflation correlators,” JHEP 07 (2023) 001, arXiv:2301.07047 [hep-th].DOI
  19. [19]S. De and A. Pokraka, “Cosmology meets cosmology,” JHEP 03 (2024) 156, arXiv:2308.03753 [hep-th].DOI
  20. [20]N. Arkani-Hamed, D. Baumann, A. Hillman, A. Joyce, H. Lee, and G. L. Pimentel, “Kinematic Flow and the Emergence of Time,” Phys. Rev. Lett. 135 no. 3, (2025) 031602, arXiv:2312.05300 [hep-th].
  21. [21]D. Stefanyszyn, X. Tong, and Y. Zhu, “Cosmological correlators through the looking glass: reality, parity, and factorisation,” JHEP 05 (2024) 196, arXiv:2309.07769 [hep-th].DOI
  22. [22]S. Aoki, L. Pinol, F. Sano, M. Yamaguchi, and Y. Zhu, “Cosmological correlators with double massive exchanges: bootstrap equation and phenomenology,” JHEP 09 (2024) 176, arXiv:2404.09547 [hep-th].arxiv.org/abs/2404.09547
  23. [23]H. Liu and Z.-Z. Xianyu, “Massive inflationary amplitudes: differential equations and complete solutions for general trees,” JHEP 09 (2025) 183, arXiv:2412.07843 [hep-th].arxiv.org/abs/2412.07843
  24. [24]Z. Qin, S. Renaux-Petel, X. Tong, D. Werth, and Y. Zhu, “The exact and approximate tales of boost-breaking cosmological correlators,” JCAP 09 (2025) 058, arXiv:2506.01555 [hep-th].arxiv.org/abs/2506.01555
  25. [25]D. Werth, “Spectral representation of cosmological correlators,” JHEP 12 (2024) 017, arXiv:2409.02072 [hep-th].arxiv.org/abs/2409.02072
  26. [26]J. Chen, B. Feng, and Y.-X. Tao, “Multivariate hypergeometric solutions of cosmological (dS) correlators by d log-form differential equations,” JHEP 03 (2025) 075, arXiv:2411.03088 [hep-th].DOI
  27. [27]D. Stefanyszyn, X. Tong, and Y. Zhu, “There and Back Again: Mapping and Factorizing Cosmological Observables,” Phys. Rev. Lett. 133 no. 22, (2024) 221501, arXiv:2406.00099 [hep-th].DOI
  28. [28]G. L. Pimentel and C. Yang, “Strongly coupled sectors in inflation: gapless theories and unparticles,” JHEP 04 (2026) 146, arXiv:2503.17840 [hep-th].DOI
  29. [29]T. Cheung and D. Stefanyszyn, “Massive spinning fields during inflation: Feynman rules and correlator comparison,” JCAP 05 (2026) 022, arXiv:2509.08888 [hep-th].DOI
  30. [30]C. Chowdhury, A. Lipstein, J. Marshall, J. Mei, and I. Sachs, “Cosmological dressing rules,” JHEP 03 (2026) 076, arXiv:2503.10598 [hep-th].arxiv.org/abs/2503.10598
  31. [31]S. Jazayeri, X. Tong, and Y. Zhu, “Every wrinkle carries a memory: an integro-differential bootstrap for features in cosmological correlators,” JHEP 06 (2026) 022, arXiv:2511.00152 [hep-th].arxiv.org/abs/2511.00152
  32. [32]T. Cheung, C. Duaso-Pueyo, D. Stefanyszyn, and D.-G. Wang, “General Spin-1 Dynamics During Inflation: Microcausality Constraints and Cosmological Collider Signals,” arXiv:2610.10693 [hep-th].arxiv.org/abs/2610.10693
  33. [33]J. Huenupi, C. Muñoz, G. A. Palma, and S. Sypsas, “Pushing the Primordial Frontier: Exact Linear Solutions in Multifield Inflation,” arXiv:2606.18248 [astro-ph.CO].arxiv.org/abs/2606.18248
  34. [34]M. Arundine and G. L. Pimentel, “Cosmological Collider in the Grassmannian,” arXiv:2605.21581 [hep-th].arxiv.org/abs/2605.21581
  35. [35]M. C. Ferreira and F. T. Falciano, “Bootstrapping a Higher-Derivative Cosmological Collider: Borel-Laplace Resummation and Stokes Phenomena,”arxiv.org/abs/2610.42020
  36. [36]N. Belrhalil, A. Poisson, S. Renaux-Petel, and D. Werth, “De Sitter Momentum Space,” arXiv:2601.15228 [hep-th].
  37. [37]N. Belrhalil, A. Poisson, and S. Renaux-Petel, “Laplace Space for Cosmological Correlators,” arXiv:2606.27309 [hep-th].
  38. [38]J. Gräfe and D. Werth, “All Tree-Level Massive Cosmological Correlators via Spectral Gluing,” arXiv:2607.18223 [hep-th].arxiv.org/abs/2607.18223
  39. [39]N. Arkani-Hamed, P. Benincasa, and A. Postnikov, “Cosmological Polytopes and the Wavefunction of the Universe,” arXiv:1709.02813 [hep-th].
  40. [40]G. Cabass, O. H. E. Philcox, M. M. Ivanov, K. Akitsu, S.-F. Chen, M. Simonović, and M. Zaldarriaga, “BOSS constraints on massive particles during inflation: The cosmological collider in action,” Phys. Rev. Lett. 135 no. 6, (2025) 063510, arXiv:2404.01894 [astro-ph.CO].arxiv.org/abs/2404.01894
  41. [41]S. Goldstein, O. H. E. Philcox, J. C. Hill, and L. Hui, “Intermediate mass-range particles from small scales: Nonperturbative techniques for cosmological collider physics from large-scale structure surveys,” Phys. Rev. D 110 no. 8, (2024) 083516, arXiv:2407.08731 [astro-ph.CO].arxiv.org/abs/2407.08731
  42. [42]W. Sohn, D.-G. Wang, J. R. Fergusson, and E. P. S. Shellard, “Searching for cosmological collider in the Planck CMB data,” JCAP 09 (2024) 016, arXiv:2404.07203 [astro-ph.CO].DOI
  43. [43]P. Suman, D.-G. Wang, W. Sohn, J. R. Fergusson, and E. P. S. Shellard, “How Significant are Cosmological Collider Signals in the Planck Data?,” arXiv:2511.17500 [astro-ph.CO].arxiv.org/abs/2511.17500
  44. [44]P. Suman, D.-G. Wang, W. Sohn, J. R. Fergusson, and E. P. S. Shellard, “Searching for Cosmological Collider in the Planck CMB Data II: collider templates and Modal analysis,” arXiv:2512.22085 [astro-ph.CO].arxiv.org/abs/2512.22085
  45. [45]O. H. E. Philcox, “Searching for inflationary physics with the CMB trispectrum. III. Constraints from Planck,” Phys. Rev. D 111 no. 12, (2025) 123534, arXiv:2502.06931 [astro-ph.CO].arxiv.org/abs/2502.06931
  46. [46]D. Anbajagane and H. Lee, “Primordial physics in the nonlinear Universe: Mapping cosmological collider models to weak-lensing observables,” JCAP 05 (2026) 076, arXiv:2509.02693 [astro-ph.CO].arxiv.org/abs/2509.02693
  47. [47]D. Green, J. Han, and B. Wallisch, “Extending the Cosmological Collider: New Scaling Regimes and Constraints from BOSS,” arXiv:2602.12232 [astro-ph.CO].arxiv.org/abs/2602.12232
  48. [48]S. Kumar, Q. Lu, Z.-Z. Xianyu, and Y. Zhang, “Cosmological Collider Searches beyond the Hubble Scale with Planck Data,” arXiv:2603.15728 [hep-ph].arxiv.org/abs/2603.15728
  49. [49]S. Kumar, Q. Lu, Z.-Z. Xianyu, and Y. Zhang, “Scalars at the Cosmological Collider: Full Shapes of Tree Diagrams and Bispectrum Searches using Planck Data,” arXiv:2604.07434 [hep-ph].arxiv.org/abs/2604.07434
  50. [50]O. H. E. Philcox, “Dissecting the Scalar Cosmological Collider with the Cosmic Microwave Background,” arXiv:2607.18369 [astro-ph.CO].arxiv.org/abs/2607.18369
  51. [51]Z.-Z. Xianyu and H. Zhang, “Bootstrapping one-loop inflation correlators with the spectral decomposition,” JHEP 04 (2023) 103, arXiv:2211.03810 [hep-th].DOI
  52. [52]S. A. Salcedo, M. H. G. Lee, S. Melville, and E. E. Pajer, “The Analytic Wavefunction,” JHEP 06 (2023) 020, arXiv:2212.08009 [hep-th]. M. H. G. Lee, C. McCulloch, and E. Pajer, “Leading loops in cosmological correlators,” JHEP 11 (2023) 038, arXiv:2305.11228 [hep-th].
  53. [54]Z. Qin and Z.-Z. Xianyu, “Inflation correlators at the one-loop order: nonanalyticity, factorization, cutting rule, and OPE,” JHEP 09 (2023) 116, arXiv:2304.13295 [hep-th].arxiv.org/abs/2304.13295
  54. [55]S. Céspedes, A.-C. Davis, and D.-G. Wang, “On the IR divergences in de Sitter space: loops, resummation and the semi-classical wavefunction,” JHEP 04 (2024) 004, arXiv:2311.17990 [hep-th].DOI
  55. [56]S. Agui Salcedo and S. Melville, “The cosmological tree theorem,” JHEP 12 (2023) 076, arXiv:2308.00680 [hep-th].arxiv.org/abs/2308.00680
  56. [57]Z. Qin, “Cosmological correlators at the loop level,” JHEP 03 (2025) 051, arXiv:2411.13636 [hep-th].arxiv.org/abs/2411.13636
  57. [58]S. Bhowmick, M. H. G. Lee, D. Ghosh, and F. Ullah, “Singularities in cosmological loop correlators,” JHEP 02 (2026) 116, arXiv:2503.21880 [hep-th].arxiv.org/abs/2503.21880
  58. [59]D. Jain, E. Pajer, and X. Tong, “Unitary and Analytic Renormalisation of Cosmological Correlators,” JHEP 03 (2026) 225, arXiv:2509.02696 [hep-th].arxiv.org/abs/2509.02696
  59. [60]S. Céspedes, Z. Qin, and D.-G. Wang, “λφ4 as an effective theory in de Sitter,” JHEP 05 (2026) 143, arXiv:2510.25826 [hep-th].arxiv.org/abs/2510.25826
  60. [61]D. Baumann, H. Goodhew, A. Joyce, H. Lee, G. L. Pimentel, and T. Westerdijk, “Geometry of kinematic flow,” JHEP 05 (2026) 211, arXiv:2504.14890 [hep-th].
  61. [62]G. L. Pimentel and T. Westerdijk, “On Cosmological Correlators at One Loop,” arXiv:2601.00952 [hep-th].
  62. [63]J. Gräfe and I. Sachs, “Split representations and bubble resummation for massive de Sitter correlators,” JHEP 07 (2026) 074, arXiv:2602.09977 [hep-th].arxiv.org/abs/2602.09977
  63. [64]A. Farren, C. McCulloch, E. Pajer, and X. Tong, ““All-Loop Renormalization and the Phase of the de Sitter Wavefunction,” arXiv:2603.08794 [hep-th].
  64. [65]T. Westerdijk and C. Yang, “Bananas are unparticles: differential equations and cosmological bootstrap,” JHEP 09 (2025) 089, arXiv:2503.08775 [hep-th].DOI
  65. [66]D. Baumann, A. Joyce, H. Lee, and K. Salehi Vaziri, “Differential Equations for Massive Correlators,” arXiv:2604.08658 [hep-th].
  66. [67]J. Chen, B. Feng, Z. Qin, and Y.-X. Tao, “Loop integrals in de Sitter spacetime: The parity-split IBP system and dlog-form differential equations,” arXiv:2604.14549 [hep-th].
  67. [68]S. Aoki, Z. Qin, M. Yamaguchi, and Y. Zhu, “Fermionic Bubble Loop in Cosmological Collider Revisited: Exact signals from spectral and Mellin-Barnes methods,” arXiv:2605.28054 [hep-th].
  68. [69]H. Liu, Z. Qin, J. Wu, Z.-Z. Xianyu, and H. Zhang, “On-Shell Bootstrap of Loop Inflation Correlators with Spectral Dispersion,” arXiv:2606.02686 [hep-th].
  69. [70]Z. Qin and Z.-Z. Xianyu, “Cosmological Collider Signals From a Triangle Loop,” arXiv:2609.09288 [hep-th].
  70. [71]M. Borinsky, A. Herderschee, and Q. Lu, “On the Numerical Integration of One-Loop Cosmological Collider Signals,” arXiv:2609.10673 [hep-ph].
  71. [72]X. Chen, Y. Wang, and Z.-Z. Xianyu, “Standard Model Background of the Cosmological Collider,” Phys. Rev. Lett. 118 no. 26, (2017) 261302, arXiv:1610.06597 [hep-th].DOI
  72. [73]X. Chen, Y. Wang, and Z.-Z. Xianyu, “Standard Model Mass Spectrum in Inflationary Universe,” JHEP 04 (2017) 058, arXiv:1612.08122 [hep-th].DOI
  73. [74]W. Z. Chua, Q. Ding, Y. Wang, and S. Zhou, “Imprints of Schwinger Effect on Primordial Spectra,” JHEP 04 (2019) 066, arXiv:1810.09815 [hep-th].arxiv.org/abs/1810.09815
  74. [75]L. Li, T. Nakama, C. M. Sou, Y. Wang, and S. Zhou, “Gravitational Production of Superheavy Dark Matter and Associated Cosmological Signatures,” JHEP 07 (2019) 067, arXiv:1903.08842 [astro-ph.CO].DOI
  75. [76]S. Lu, Y. Wang, and Z.-Z. Xianyu, “A Cosmological Higgs Collider,” JHEP 02 (2020) 011, arXiv:1907.07390 [hep-th].DOI
  76. [77]A. Hook, J. Huang, and D. Racco, “Minimal signatures of the Standard Model in non-Gaussianities,” Phys. Rev. D 101 no. 2, (2020) 023519, arXiv:1908.00019 [hep-ph].DOI
  77. [78]L.-T. Wang and Z.-Z. Xianyu, “In Search of Large Signals at the Cosmological Collider,” JHEP 02 (2020) 044, arXiv:1910.12876 [hep-ph].DOI
  78. [79]L. Li, S. Lu, Y. Wang, and S. Zhou, “Cosmological Signatures of Superheavy Dark Matter,” JHEP 07 (2020) 231, arXiv:2002.01131 [hep-ph].DOI
  79. [80]L.-T. Wang and Z.-Z. Xianyu, “Gauge Boson Signals at the Cosmological Collider,” JHEP 11 (2020) 082, arXiv:2004.02887 [hep-th].DOI
  80. [81]S. Aoki and M. Yamaguchi, “Disentangling mass spectra of multiple fields in cosmological collider,” JHEP 04 (2021) 127, arXiv:2012.13667 [hep-th].DOI
  81. [82]Y. Cui and Z.-Z. Xianyu, “Probing Leptogenesis with the Cosmological Collider,” Phys. Rev. Lett. 129 no. 11, (2022) 111301, arXiv:2112.10793 [hep-ph].DOI
  82. [83]Y. Ema and S. Verner, “Cosmological collider signatures of Higgs-R2 inflation,” JCAP 04 (2024) 039, arXiv:2309.10841 [hep-ph].arxiv.org/abs/2309.10841
  83. [84]A. Bodas, E. Broadberry, R. Sundrum, and Z. Xu, “Charged loops at the cosmological collider with chemical potential,” JHEP 01 (2026) 083, arXiv:2507.22978 [hep-ph].DOI
  84. [85]J. You, L. Song, C. Han, H.-J. He, X. Chen, and Z.-Z. Xianyu, “Cosmological Collider Signatures from Right-Handed Neutrino Loop,” arXiv:2605.21419 [hep-ph].
  85. [86]X. Chen, Y. Wang, and Z.-Z. Xianyu, “Neutrino Signatures in Primordial Non-Gaussianities,” JHEP 09 (2018) 022, arXiv:1805.02656 [hep-ph].DOI
  86. [87]A. Hook, J. Huang, and D. Racco, “Searches for other vacua. Part II. A new Higgsstory at the cosmological collider.” JHEP 01 (2020) 105, arXiv:1907.10624 [hep-ph].DOI
  87. [88]X. Tong, Y. Wang, C. Zhang, and Y. Zhu, “BCS in the sky: signatures of inflationary fermion condensation,” JCAP 04 (2024) 022, arXiv:2304.09428 [hep-th].DOI
  88. [89]S. Aoki and A. Strumia, “Cosmological collider signals of modular spontaneous CP breaking,” arXiv:2604.05548 [hep-ph].
  89. [90]J. You, L. Song, H.-J. He, and C. Han, “Cosmological non-Gaussianity from the neutrino seesaw mechanism,” Phys. Rev. D 112 no. 8, (2025) 083555, arXiv:2412.16033 [hep-ph].DOI
  90. [91]C. Han, H.-J. He, L. Song, and J. You, “Cosmological signatures of neutrino seesaw mechanism,” Phys. Rev. D 112 no. 8, (2025) L081309, arXiv:2412.21045 [hep-ph].DOI
  91. [92]C. M. Sou, X. Tong, and Y. Wang, “Chemical-potential-assisted particle production in FRW spacetimes,” JHEP 06 (2021) 129, arXiv:2104.08772 [hep-th].DOI
  92. [93]S. Aoki and M. Yamaguchi, “Hubble Susceptibility: Beyond Effective-Mass Spectroscopy in the Cosmological Collider,” arXiv:2610.26014 [hep-th].
  93. [94]D. Marolf and I. A. Morrison, “The IR stability of de Sitter: Loop corrections to scalar propagators,” Phys. Rev. D 82 (2010) 105032, arXiv:1006.0035 [gr-qc].
  94. [95]L. Di Pietro, V. Gorbenko, and S. Komatsu, “Analyticity and unitarity for cosmological correlators,” JHEP 03 (2022) 023, arXiv:2108.01695 [hep-th].
  95. [96]C. Sleight and M. Taronna, “From dS to AdS and back,” JHEP 12 (2021) 074, arXiv:2109.02725 [hep-th].DOI
  96. [97]M. Loparco, J. Penedones, K. Salehi Vaziri, and Z. Sun, “The Källén-Lehmann representation in de Sitter spacetime,” JHEP 12 (2023) 159, arXiv:2306.00090 [hep-th].DOI
  97. [98]X. Tong, Y. Wang, and Y. Zhu, “Cutting rule for cosmological collider signals: a bulk evolution perspective,” JHEP 03 (2022) 181, arXiv:2112.03448 [hep-th].DOI
  98. [99]Y. Ema and K. Mukaida, “Cutting rule for in-in correlators and cosmological collider,” JHEP 12 (2024) 194, arXiv:2409.07521 [hep-th].arxiv.org/abs/2409.07521

Paper details

vlnc:W_6hndspj7
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

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

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents