Controlled stochastic eternal inflation versus the refined swampland de Sitter conjecture: is the exclusion robust?

Brahma and Shandera (JHEP 11 (2019) 016) argued that single-field stochastic eternal inflation, when it is under perturbative control, is incompatible with the refined swampland de Sitter conjecture: perturbativity of the curvature perturbations requires the power spectrum amplitude to fall below $1/(144\,\eta_V^2)$ in terms of the potential slow-roll parameter $\eta_V$, while the onset of stochastic eternal inflation requires an amplitude of order one, and the two conditions can coexist only if the conjecture's constant $\tilde c$ is smaller than roughly $0.1$. The source paper explicitly left open whether multi-field dynamics, additional light or dissipative (warm inflation) degrees of freedom, couplings to heavy fields, non-perturbative stochastic regimes, or a numerically small $\tilde c$ evade this exclusion. The later literature shows the question is alive: D'Agnolo, Mangini, Rigo and Wang (Phys. Rev. D 110 (2024) 055007) constructed fluctuation-dominated hilltop eternal-inflation models compatible with both the distance and refined de Sitter conjectures, but their analysis treats the quantum kicks as Gaussian and does not enforce the perturbativity constraint that carries the Brahma-Shandera exclusion, and no later work has adjudicated this tension; Cohen, Green and Premkumar independently showed that the transition to eternal inflation is generically sensitive to non-Gaussian tails beyond perturbative control. Whether any controlled realization of stochastic eternal inflation consistent with the swampland criteria exists -- or a proof that none does -- remains open, as does the quantification of entropy contributed by modes redshifted into the effective theory during the fluctuation-dominated regime.

Background

Inflation is an early epoch of accelerated expansion, driven in its simplest form by a scalar field (the inflaton) rolling down a potential $V(\phi)$ with Hubble rate $H$ set by $3H^2M_p^2=V$, where $M_p$ is the reduced Planck mass. During inflation, quantum fluctuations of the inflaton of typical size $\langle\delta\phi\rangle_Q\simeq H/(2\pi)$ accumulate per Hubble time. When these quantum kicks compete with or dominate the classical roll, $H^2/(2\pi\dot\phi)\gtrsim 1$, some Hubble patches never leave the inflating region of the potential: inflation becomes self-reproducing, a regime called stochastic (or slow-roll) eternal inflation. In that regime the dimensionless power spectrum of curvature perturbations $\mathcal{P}_\zeta=(H^2/2\pi\dot\phi)^2$ is of order one or larger, instead of the observed $\sim 2\times 10^{-9}$. Eternal inflation underwrites the multiverse picture in which different vacua of a landscape (the ensemble of metastable vacua of a quantum-gravity theory such as string theory) are populated dynamically, and is often invoked for anthropic explanations of the cosmological constant.

The swampland program conjectures which low-energy effective field theories coupled to gravity cannot be completed into a theory of quantum gravity such as string theory. Two conjectures are relevant here. The distance conjecture bounds field excursions in effective theories coupled to gravity: super-Planckian displacements trigger a tower of light states and exit the effective description. The refined de Sitter (dS) swampland conjecture (Garg-Krishnan 2019; Ooguri-Palti-Shiu-Vafa 2019) states that any quasi-de Sitter solution sourced by a potential $V$ with slope $|V'|<c\,V/M_p$ (with $c>0$ of order one) must be unstable in the tachyonic sense, $V''\le-\tilde c\,V/M_p^2$ with $\tilde c>0$ of order one. In terms of the potential slow-roll parameters $\epsilon_V=M_p^2(V'/V)^2/2$ and $\eta_V=M_p^2V''/V$, the conjecture forces $|\eta_V|\gtrsim\tilde c=O(1)$ wherever slow roll occurs. Ooguri et al. derived this instability requirement from the compatibility of accelerated expansion with the covariant entropy bound (Bousso's conjecture that the entropy on any light-sheet -- a null hypersurface traced by light rays converging toward a surface -- is bounded by a quarter of that surface's area in Planck units) and the distance conjecture.

Kinney (Phys. Rev. Lett. 122 (2019) 081302) analyzed eternal hilltop inflation (inflation driven by a field perched near a maximum of its potential, where the classical roll is slow) using the standard Gaussian fluctuation amplitude and concluded it is marginally consistent with the refined dS conjecture. Brahma and Shandera (JHEP 11 (2019) 016) overturned this by enforcing a consistency requirement the earlier analysis missed: the perturbative expansion itself. Estimating the cubic (and higher) terms of the single-field action for the curvature perturbation $\zeta$, the requirement $S_n/S_2<1$ for all $n>2$ gives $\mathcal{P}_\zeta<1/(16[3(\eta_V-\epsilon_V)-\epsilon_V(\eta_V-\epsilon_V)-\epsilon_V\eta_V]^2)$, i.e. $\mathcal{P}_\zeta<1/(144\,\eta_V^2)$ when $|\eta_V|\gg\epsilon_V$. (Allowing the inflaton a non-trivial sound speed $c_s$ -- the propagation speed of its fluctuations, reduced when the inflaton couples to heavy fields that have been integrated out -- only strengthens the bound.) Stochastic eternal inflation needs $\mathcal{P}_\zeta\gtrsim 1$, so perturbative control requires $|\eta_V|\lesssim 1/12$, in tension with the conjecture's $|\eta_V|\gtrsim\tilde c=O(1)$ unless $\tilde c\lesssim O(1/10)$. They concluded that controlled single-field stochastic eternal inflation sits in the swampland, conditional on the distance conjecture, the covariant entropy bound, and perturbative control of the fluctuations.

The source paper states explicitly what remains unresolved: the multi-field case, additional light degrees of freedom, couplings to heavy fields, and warm inflation (inflation with dissipative particle production, where radiation and friction from a bath of other fields modify both the background roll and the fluctuations) are not covered by the single-field analysis; the entropy accounting may receive further contributions from modes redshifted into the low-energy theory during the fluctuation-dominated regime; non-perturbative stochastic regimes are unaddressed; and the loophole of a small $\tilde c$ is acknowledged but not resolved. Later work sharpened rather than settled these questions. Wang, Brandenberger and Heisenberg (Eur. Phys. J. C 80 (2020)) found parallel exclusions of hilltop stochastic eternal inflation from entropy bounds and a graceful-exit problem (the difficulty of ending the eternal regime and transitioning to the conventional hot big-bang evolution), while Blanco-Pillado, Deng and Vilenkin (JCAP 05 (2020) 014) showed that landscape-level eternal inflation survives in 'swampy' random potentials without flat hilltops, via tunneling rather than fluctuation domination. Cohen, Green and Premkumar (SciPost Phys. 14 (2023) 109; Phys. Rev. D 107 (2023) 083501) showed within a controlled de Sitter effective theory that the transition to eternal inflation is dominated by non-Gaussian tails of the fluctuation distribution and is often incalculable within the effective theory, making perturbativity a genuinely binding concern. Most strikingly, D'Agnolo, Mangini, Rigo and Wang (Phys. Rev. D 110 (2024) 055007) constructed tuned hilltop models with sub-Planckian excursions in which quantum-fluctuation-dominated eternal inflation occurs while both the distance and refined dS conjectures hold -- but their treatment models the kicks as Gaussian with width $H/2\pi$ per Hubble time and never checks the $S_n/S_2<1$ hierarchy, i.e. precisely the constraint that Brahma and Shandera showed fails in exactly this corner ($\mathcal{P}_\zeta\gtrsim 1$, $|\eta_V|\sim O(1)$). The 2024 constructions and the 2019 exclusion therefore coexist in the literature without adjudication.

Problem Statement

Determine whether stochastic eternal inflation -- the regime where quantum fluctuations of the inflaton dominate its classical roll, $H^2/(2\pi\dot\phi)\gtrsim 1$, equivalently $\mathcal{P}_\zeta\gtrsim 1$ -- admits a controlled realization that is consistent with the refined swampland de Sitter conjecture, or whether the incompatibility established by Brahma and Shandera for single-field, unit-sound-speed models is robust. Concretely, the question has the following jointly linked components, all bearing on the single objective of deciding whether controlled stochastic eternal inflation is compatible with quantum-gravity consistency criteria as encoded in the swampland conjectures:

1. Multi-field, extra-degree-of-freedom, and dissipative escape routes: the Brahma-Shandera perturbativity bound $\mathcal{P}_\zeta<1/(144\,\eta_V^2)$ (for $|\eta_V|\gg\epsilon_V$, with small sound speed only strengthening it) was derived only for single-field models. Establish whether multi-field dynamics, additional light fields, couplings to heavy fields, or warm-inflation dissipation can relax this bound so that a fluctuation-dominated regime with $\mathcal{P}_\zeta\gtrsim 1$ remains perturbatively calculable ($S_n/S_2<1$ for all $n>2$) while the potential satisfies $V''\le-\tilde c\,V/M_p^2$ with $\tilde c=O(1)$; or prove that no such relaxation is possible within a stated class of controlled effective theories.

2. Adjudication of existing candidate constructions: in particular, determine whether the D'Agnolo-Mangini-Rigo-Wang hilltop models (tuned initial conditions, $M_p^2|V''|/V=O(1)$, $\mathcal{P}_\zeta\gtrsim 1$) satisfy the perturbativity hierarchy in the fluctuation-dominated regime -- i.e. whether their Gaussian random-walk treatment is justified -- or whether, as the Brahma-Shandera estimate suggests, non-Gaussian corrections of order $S_3/S_2$ invalidate it.

3. Entropy accounting: quantify the contribution to the covariant entropy bound of modes redshifted into the low-energy effective theory during an enduring fluctuation-dominated regime, determining whether it only tightens the Brahma-Shandera bound and by how much.

4. Non-perturbative and small-$\tilde c$ loopholes: establish whether any controlled non-perturbative treatment of the stochastic regime could reconcile eternal inflation with the swampland criteria, and whether the refined de Sitter criteria could be relaxed to allow $\tilde c\lesssim O(0.1)$ -- small enough to reopen the window for perturbative eternal inflation -- the loophole the source acknowledges but does not resolve.

A complete answer must address all four components above and reach one of two terminal determinations: either it exhibits a controlled model (or class) realizing stochastic eternal inflation consistent with the refined de Sitter and distance conjectures, including a demonstrated control hierarchy (perturbative or non-perturbative), or it establishes the robustness of the Brahma-Shandera exclusion across the routes and candidate constructions of components 1, 2 and 4; in both cases the entropy accounting of component 3 is required, since its contribution can only tighten the Brahma-Shandera bound.

References

1. Suddhasattwa Brahma and Sarah Shandera, 'Stochastic eternal inflation is in the swampland', J. High Energ. Phys. 11 (2019) 016. DOI: 10.1007/JHEP11(2019)016; arXiv:1904.10979. https://arxiv.org/abs/1904.10979
2. Hirosi Ooguri, Eran Palti, Gary Shiu, and Cumrun Vafa, 'Distance and de Sitter Conjectures on the Swampland', Phys. Lett. B 788 (2019) 180. DOI: 10.1016/j.physletb.2018.11.018; arXiv:1810.05506. https://arxiv.org/abs/1810.05506
3. Sumit K. Garg and Chethan Krishnan, 'Bounds on Slow Roll and the de Sitter Swampland', J. High Energ. Phys. 11 (2019) 075. DOI: 10.1007/JHEP11(2019)075; arXiv:1807.05193. https://arxiv.org/abs/1807.05193
4. Raphael Bousso, 'A Covariant entropy conjecture', J. High Energ. Phys. 07 (1999) 004. DOI: 10.1088/1126-6708/1999/07/004; arXiv:hep-th/9905177. https://arxiv.org/abs/hep-th/9905177
5. William H. Kinney, 'Eternal Inflation and the Refined Swampland Conjecture', Phys. Rev. Lett. 122 (2019) 081302. DOI: 10.1103/PhysRevLett.122.081302; arXiv:1811.11698. https://arxiv.org/abs/1811.11698
6. Hiroki Matsui and Fuminobu Takahashi, 'Eternal Inflation and Swampland Conjectures', Phys. Rev. D 99 (2019) 023533. DOI: 10.1103/PhysRevD.99.023533; arXiv:1807.11938. https://arxiv.org/abs/1807.11938
7. Tom Rudelius, 'Conditions for (No) Eternal Inflation', J. Cosmol. Astropart. Phys. 08 (2019) 009. DOI: 10.1088/1475-7516/2019/08/009; arXiv:1905.05198. https://arxiv.org/abs/1905.05198
8. Ziwei Wang, Robert Brandenberger, and Lavinia Heisenberg, 'Eternal inflation, entropy bounds and the swampland', Eur. Phys. J. C 80 (2020) 864. DOI: 10.1140/epjc/s10052-020-8412-x; arXiv:1907.08943. https://arxiv.org/abs/1907.08943
9. Jose J. Blanco-Pillado, Heling Deng, and Alexander Vilenkin, 'Eternal Inflation in Swampy Landscapes', J. Cosmol. Astropart. Phys. 05 (2020) 014. DOI: 10.1088/1475-7516/2020/05/014; arXiv:1909.00068. https://arxiv.org/abs/1909.00068
10. Timothy Cohen, Daniel Green, and Akhil Premkumar, 'A tail of eternal inflation', SciPost Phys. 14 (2023) 109. DOI: 10.21468/SciPostPhys.14.5.109; arXiv:2111.09332. https://arxiv.org/abs/2111.09332
11. Timothy Cohen, Daniel Green, and Akhil Premkumar, 'Large deviations in the early Universe', Phys. Rev. D 107 (2023) 083501. DOI: 10.1103/PhysRevD.107.083501; arXiv:2212.02535. https://arxiv.org/abs/2212.02535
12. Georgios Fanaras and Alexander Vilenkin, 'Quantum cosmology, eternal inflation, and swampland conjectures', J. Cosmol. Astropart. Phys. 04 (2023) 034. DOI: 10.1088/1475-7516/2023/04/034; arXiv:2302.04962. https://arxiv.org/abs/2302.04962
13. Raffaele Tito D'Agnolo, Paolo Mangini, Gabriele Rigo, and Lian-Tao Wang, 'A multiverse outside of the swampland', Phys. Rev. D 110 (2024) 055007. DOI: 10.1103/PhysRevD.110.055007; arXiv:2404.13109. https://arxiv.org/abs/2404.13109
14. Arjun Berera and Jaime Calderón-Figueroa, 'Looking inside the Swampland from Warm Inflation: Dissipative Effects in De Sitter Expansion', Universe 9 (2023) 168. DOI: 10.3390/universe9040168. https://doi.org/10.3390/universe9040168
15. Vahid Kamali, Meysam Motaharfar, and Rudnei O. Ramos, 'Recent Developments in Warm Inflation', Universe 9 (2023) 124. DOI: 10.3390/universe9030124. https://doi.org/10.3390/universe9030124
16. Tom Banks, 'On the Limits of Effective Quantum Field Theory: Eternal Inflation, Landscapes, and Other Mythical Beasts', arXiv:1910.12817 (2019). https://arxiv.org/abs/1910.12817
17. Hiromasa Tajima and Yasusada Nambu, 'Stochastic inflation and entropy bound in de Sitter spacetime', Phys. Rev. D 111 (2025) 106009. DOI: 10.1103/PhysRevD.111.106009; arXiv:2405.10837. https://arxiv.org/abs/2405.10837