Correlators from Amplitubes

Ross Glew

REVIEW 2 major objections 4 minor 36 references

Correlators from Amplitubes

T0 review · 2 major / 4 minor · reviewed 2026-08-06 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read For conformally coupled scalars, the paper derives a sign-free amplitube expansion of cosmological correlators in which only bipartite contracted graphs contribute.

desk verdict The bipartite selection rule is new and plausible, but the unproved graph identity in the step from (34) to (35) is the main gap and needs either a proof or a stated convention. read the letter →

arxiv 2507.07199 v2 pith:VGLWZ55W submitted 2025-07-09 hep-th gr-qc

Ross Glew This is my paper · ORCID

classification hep-thgr-qc

keywords cosmologicalcorrelatorswavefunctionoftheuniverseamplitubesconformallycoupledscalarsbipartitegraphsFeynmangraphexpansionpolytopes

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Observables in early-universe cosmology are spatial correlation functions, obtained from the wavefunction of the universe by the Born rule. The paper aims to express the graph-by-graph contribution to these correlators directly in terms of amplitubes, the same amplitude-like building blocks recently used for wavefunction coefficients. Its central result is that the correlator expansion is sign-free and sparser than the wavefunction expansion: a term survives only when a certain contracted graph is bipartite. If correct, this gives a combinatorial reason for a subtle simplicity of cosmological correlators and a concrete target for geometric interpretations.

What carries the argument

The engine is the amplitube $A_G = \sum_{\tau\in\Gamma_G}\prod_{t\in\tau} 1/H_t$, where the sum runs over tubings (maximal sets of pairwise compatible vertex-induced tubes) and $H_t$ is the energy sum associated with tube $t$. Starting from $\Psi_G = \sum_{I\subset E_G}(-1)^{|I|} A_{G\setminus I}$ and substituting into $\langle G\rangle = N\prod_{e}(2y_e)\sum_{I} 2^{\kappa_{G\setminus I}}\Psi_{G\setminus I}$, the derivation reduces to the combinatorial identity that $\sum_{J\subset I}(-1)^{|I\setminus J|}2^{\kappa_{G\setminus J}}$ equals twice the number of proper two-colorings of $G/\bar I$, namely $2\chi$. This identity converts an alternating sum into a bipartiteness test and is the mechanism behind every cancellation and sign removal in the examples.

What would settle it

Take $G$ to be a triangle and $I$ to be all three edges, so the contracted graph $G/\bar I$ is the triangle itself and is not bipartite; Eq. (35) demands that the coefficient of this amplitube vanish. The coefficient is the eight-term sum $\sum_{J\subset I}(-1)^{|I\setminus J|}2^{\kappa_{G\setminus J}}$, which can be evaluated directly by hand; a nonzero result would refute the formula, while a vanishing result supports it.

Watch

Extended reading notes

Core claim

On the paper's own terms: substituting the amplitube expansion of wavefunction coefficients into the known correlator formula produces, after a graph-theoretic simplification, the exact expression (35): the contribution of a Feynman graph $G$ is $2N\prod_{e\in E_G}(2y_e)$ times the sum over edge subsets $I$ of $\chi_{G/\bar I} A_{G\setminus I}$, where $\chi$ is 1 exactly when the graph obtained by contracting all edges outside $I$ is bipartite. Consequently every amplitube whose associated contracted graph fails bipartiteness drops out, and all surviving terms carry the same overall coefficient, so there are no relative minus signs. For tree graphs this means the correlator is the wavefunction expansion with all minus signs removed; at loop level, additional terms vanish entirely. The worked examples show these cancellations explicitly for the two-chain, three-chain, two-cycle, and a one-loop graph.

Load-bearing premise

The whole simplification rests on an unproved combinatorial identity, Eq. (34), justified only by 'it is not hard to convince oneself', which equates an alternating sum over edge subsets of powers of two with twice a two-coloring count; if that identity fails for some graph, the sign-free formula and its cancellations collapse.

Editorial extensions

If this is right

Correlator computations can be organized directly at the level of amplitubes, bypassing the sign-heavy wavefunction sum.
At tree level, the correlator is obtained from the wavefunction coefficient by deleting all minus signs, up to an overall normalization factor.
Loop-level contributions whose contracted graph is non-bipartite vanish in the correlator, so the expansion has fewer terms than the wavefunction expansion.
The bipartite criterion supplies a combinatorial origin for the subtle simplicity of cosmological correlators previously seen through dressing rules.
As the author notes speculatively, the appearance of bipartiteness invites a geometric interpretation and a possible connection to bipartite graphs in scattering amplitudes.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

The unproved identity in Eq. (34) resembles a deletion-contraction recursion and is likely provable through the chromatic or Tutte polynomial; such a proof would place the main formula on fully rigorous footing.
Because the simplification is purely graph-theoretic, the same sign-free formula should hold for any theory whose wavefunction coefficients admit an amplitube expansion, not only the conformally coupled scalar toy model treated here.
The bipartiteness criterion implies that odd cycles surviving contraction are the only source of cancellations; testing a one-loop graph built on a pentagon would probe whether the simplifications persist beyond the examples shown.
A positive-geometric realization might come from summing amplitubes only over tubings selected by a bipartiteness condition; the correlator's polytope would then be a parity-selected piece of the wavefunction's polytope.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Hi,

This short paper extends the amplitube expansion of wavefunction coefficients to full correlators. The main result, Eq. (35), is a compact bipartite selection rule: for each graph G, only edge subsets I whose contracted graph G/\bar I is bipartite survive, and all coefficients are positive. That is genuinely new and gives a combinatorial origin for the simplicity seen in recent work [27-29]. The examples (two-chain, three-chain, two-cycle, one-loop) check out and show the cancellations clearly. The writing is clear and the logic is easy to follow.

The soft spot is the step from Eq. (34) to Eq. (35). The coefficient sum over J is asserted to equal the number of proper 2-colorings of G/\bar I, with a terse 'not hard to convince oneself'. This is load-bearing: all the loop-level cancellations depend on it. In fact the identity is a q=2 evaluation of the chromatic polynomial, and it works for connected graphs, but only under a specific convention: loops created by contracting \bar I must be deleted, while loops not in \bar I survive. The paper never states this. For disconnected contracted graphs the coefficient would come out as 2^{c(G/\bar I)} times the bipartite indicator, so the formula as written needs G/\bar I connected or an explicit convention. The author should provide a proof or at least a citation for the identity. This is an easy fix, but the paper as it stands has a real gap at the central argument.

The reliance on Eq. (14) from the author's own prior paper is not a problem by itself; that result is published. But it does mean the new contribution rests on two imported formulas, and the only new ingredient is the identity above. The minor typos (e.g., Eq. (8) repeats x2) are cosmetic.

Verdict: I'd send it to a serious referee. The result is plausible, new, and potentially useful as an organizing principle for correlator computations. A referee should push for the missing proof or reference, and for a clear statement of graph-contraction conventions. If those are supplied, this is a nice paper for the cosmobootstrap audience.

Referee Report

2 major / 4 minor

Summary. The paper derives an amplitube expansion for the correlation function of conformally coupled scalars, starting from a known expression for the correlator as a sum over wavefunction coefficients and the author's earlier amplitube expansion of the wavefunction. The main claim is the closed formula ⟨G⟩ = 2N∏_{e∈E_G}(2y_e) \sum_{I⊂E_G} χ_{G/\bar I} A_{G\setminus I}, where χ_{G/\bar I} is 1 if the contracted graph is bipartite and 0 otherwise. The paper demonstrates this formula on several tree-level and one-loop examples, showing many cancellations and the absence of relative minus signs. The derivation hinges on an unproved combinatorial identity in the passage from Eq. (34) to Eq. (35); the paper states it is 'not hard to convince oneself' and calls it a well-known graph invariant, but supplies neither a proof nor a citation.

Significance. If the main formula is correct, it provides a remarkably simple combinatorial characterization of cosmological correlators, making contact with the recent 'subtle simplicity' and dressing-rule results of Refs. [27–29] and uncovering a potential link to bipartite structures in amplitudes. The formula is parameter-free and the worked examples are internally consistent; the claimed cancellations, including the loop-level vanishing terms, are nontrivial and are exactly the kind of simplification that justifies publication. The significance is, however, conditional on closing the proof gap in the single load-bearing combinatorial step, because all examples shown are special cases of that step.

major comments (2)

[§IV, Eqs. (34)–(36)] The step from Eq. (34) to Eq. (35) is the central load-bearing point of the paper, yet the identity C(I) = \sum_{J⊂I} (-1)^{|I|-|J|} 2^{κ_{G\setminus J}} = 2χ_{G/\bar I} is asserted with the comment that it is 'not hard to convince oneself' and is called a well-known graph invariant. No proof or citation is provided. This identity is not completely trivial: for connected graphs it follows from the q=2 evaluation of the chromatic polynomial of G/\bar I, but its validity depends on the convention for contracting edges that produce loops. The paper does not state this convention. I ask the author to supply a proof or an explicit reference for the identity and to spell out the loop-contraction convention.
[§IV, Eq. (35) and the worked examples] The formula with the prefactor 2N is stated without qualification, but the identity underlying it yields 2χ_{G/\bar I} only when G is connected; for disconnected graphs the coefficient would be 2^{c(G/\bar I)}χ_{G/\bar I}, where c(G/\bar I) is the number of connected components. All worked examples are connected graphs, so they do not probe this distinction. The paper should clarify whether the path sum is restricted to connected Feynman graphs or, if disconnected graphs are allowed, how the formula must be modified.

minor comments (4)

[§II.A, Eq. (8)] In the five-point example, the variable x2 is defined twice: 'x2 = |k3|' and then 'x2 = |k3| + |k4|'; the second assignment should presumably be x3, since the formula is meant to give three distinct vertex variables.
[§IV, Eq. (34)] The exponent in the factor (-1)^{I\setminus J} is written as a set difference rather than an integer; it should read (-1)^{|I|-|J|}.
[§I] There are several typographical errors in the text, for example 'At first site' should be 'At first sight' and 'probabliity distribution' should be 'probability distribution'; a careful proofreading pass is recommended.
[§IV, after Eq. (28)] The statement 'this statement is valid for all tree-level graphs' is asserted without proof at that point; since the later general formula (35) implies it, it would be cleaner to defer or briefly justify this claim.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the correlator amplitube formula is derived algebraically from two prior independent formulas plus a true but unproved graph identity.

full rationale

The paper's derivation is not circular. It starts from two explicitly cited independent inputs: the wavefunction-amplitube formula (14) (Eq. (14), citing [25]) and the known correlator-from-wavefunction formula (21) (Eq. (21), citing [24]). Substituting (14) into (21) yields Eq. (34), and passing from (34) to (35) rests on the combinatorial identity sum_{J⊆I} (-1)^{|I|-|J|} 2^{κ_{G\J}} = 2 χ_{G/\bar I}, which the paper asserts as 'not hard to convince oneself' (Section IV). This identity is a genuine graph-theoretic lemma (for connected G, the q=2 chromatic-polynomial evaluation of G/\bar I), not an equation true by definition or by the input formulas. No parameter is fitted, no quantity is defined in terms of the target result, and no conclusion is imported solely from a same-author citation. The self-citations [25] and [26] are load-bearing inputs, but they are prior parameter-free results with checkable content (the wavefunction expansion is tested in the paper's own worked examples against known coefficients), so they count as independent support rather than circularity. The paper's main weakness is the omitted proof of the combinatorial identity connecting (34) and (35); the manuscript neither proves it nor cites a source, and the contraction conventions (e.g., loop deletion) are left implicit. This is a rigor gap, not a circular step, and it does not raise the circularity score. Therefore the circularity score is 0.

Assumptions & free parameters 0 free parameters · 3 assumptions · 0 invented entities

No free parameters or invented entities are introduced. The central claim rests on two imported formulas, one self-cited, and on an unproved graph-counting identity that converts the summed expression into the bipartite selection rule.

assumptions (3)

domain assumption Eq. (21) correlator graph expansion in terms of wavefunction coefficients from [24].
The derivation starts from this formula, imported from Figueiredo and Vazao without proof or independent check in this paper.
domain assumption Eq. (14) wavefunction coefficients expand as alternating sums over amplitubes from [25].
This is a self-cited prior result by the same author; the paper assumes it holds for the graphs considered here.
ad hoc to paper The coefficient sum in Eq. (34) equals chi_{G/\bar I}, the bipartite indicator.
Asserted in Section IV with 'it is not hard to convince oneself'; no proof or citation is given, and it is the load-bearing step of the main result.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Correlators from Amplitubes." pith.science (2026). https://pith.science/paper/VGLWZ55W

@misc{pith2026250707199,
  author       = {Pith},
  title        = {Pith review of: Correlators from Amplitubes},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VGLWZ55W}},
  note         = {Machine review of arXiv:2507.07199}
}

read the original abstract

Recently, the wavefunction coefficients for conformally coupled scalars in an FRW cosmology have been presented as a sum over amplitude-like functions known as {\it amplitubes}. In this work we extend this analysis to full {\it correlation functions}. Remarkably, the amplitube expansion of the correlator exhibits many vanishing contributions that are otherwise present in the expansion of the wavefunction. Moreover, while the wavefunction coefficients suffer from relative minus signs between terms, the surviving terms in correlation function do not. These observations point to a hidden simplicity in the structure of the correlation function compared to that of the wavefunction coefficient.

Figures

Figures reproduced from arXiv: 2507.07199 by the authors.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

36 extracted references · 32 canonical work pages

[1]

intersect if t ∩ t′ ̸= ∅ and t ̸⊂ t′ and t′ ̸⊂ t,

work page
[2]

touch if t ∩ t′ = ∅ and t ∪ t′ ∈ TG,

work page
[3]

A tubing τ is then defined to be a maximal set of pair- wise compatible tubes of TG

are compatible if they do not intersect or touch. A tubing τ is then defined to be a maximal set of pair- wise compatible tubes of TG. The set of tubings will be denoted by Γ G. Finally, with this terminology introduced, the ampli- tube associated to the graph G is defined as AG = X τ ∈ΓG Y t∈τ 1 Ht . (13) For a more detailed discussion of these functions...

work page
[4]

Pimentel

Nima Arkani-Hamed, Daniel Baumann, Hayden Lee, and Guilherme L. Pimentel. The Cosmological Bootstrap: Inflationary Correlators from Symmetries and Singulari- ties. JHEP, 04:105, 2020. doi:10.1007/JHEP04(2020)105

doi:10.1007/jhep04(2020)105 2020
[5]

From locality and unitarity to cosmological correlators

Sadra Jazayeri, Enrico Pajer, and David Stefanyszyn. From locality and unitarity to cosmological correlators. JHEP, 10:065, 2021. doi:10.1007/JHEP10(2021)065

work page doi:10.1007/jhep10(2021)065 2021
[6]

Cosmological Polytopes and the Wavefunc- tion of the Universe

Nima Arkani-Hamed, Paolo Benincasa, and Alexander Postnikov. Cosmological Polytopes and the Wavefunc- tion of the Universe. 9 2017

work page 2017
[7]

Cosmological Polytopes and the Wave- functon of the Universe for Light States

Paolo Benincasa. Cosmological Polytopes and the Wave- functon of the Universe for Light States. 9 2019

work page 2019
[8]

The Geometry of Cosmological Correlators

Paolo Benincasa and Gabriele Dian. The Geometry of Cosmological Correlators. SciPost Phys. , 18:105, 2025. doi:10.21468/SciPostPhys.18.3.105

work page doi:10.21468/scipostphys.18.3.105 2025

Show all 36 references

[9]

Pimentel

Nima Arkani-Hamed, Daniel Baumann, Aaron Hillman, Austin Joyce, Hayden Lee, and Guilherme L. Pimentel. Differential Equations for Cosmological Correlators. 12 2023

2023
[10]

Kinematic Flow for Cosmological Loop Integrands

Daniel Baumann, Harry Goodhew, and Hayden Lee. Kinematic Flow for Cosmological Loop Integrands. 10 2024

2024
[11]

Pimentel, and Tom Westerdijk

Daniel Baumann, Harry Goodhew, Austin Joyce, Hay- den Lee, Guilherme L. Pimentel, and Tom Westerdijk. Geometry of Kinematic Flow. 4 2025

2025
[12]

Cosmology meets cohomology

Shounak De and Andrzej Pokraka. Cosmology meets cohomology. JHEP, 03:156, 2024. doi: 10.1007/JHEP03(2024)156

2024 doi
[13]

Differential equations and recursive solu- tions for cosmological amplitudes

Song He, Xuhang Jiang, Jiahao Liu, Qinglin Yang, and Yao-Qi Zhang. Differential equations and recursive solu- tions for cosmological amplitudes. JHEP, 01:001, 2025. doi:10.1007/JHEP01(2025)001

2025 doi
[14]

A physical basis for cosmological correlators from cuts

Shounak De and Andrzej Pokraka. A physical basis for cosmological correlators from cuts. JHEP, 03:040, 2025. doi:10.1007/JHEP03(2025)040

2025 doi
[15]

Grimm, Arno Hoefnagels, and Mick van Vliet

Thomas W. Grimm, Arno Hoefnagels, and Mick van Vliet. A Reduction Algorithm for Cosmological Corre- lators: Cuts, Contractions, and Complexity. 3 2025

2025
[16]

Canonical Differential Equations for Cosmology from Positive Geometries

Mattia Capuano, Livia Ferro, Tomasz Lukowski, and Alessandro Palazio. Canonical Differential Equations for Cosmology from Positive Geometries. 5 2025

2025
[17]

Cosmological correlators at the loop level

Zhehan Qin. Cosmological correlators at the loop level. JHEP, 03:051, 2025. doi:10.1007/JHEP03(2025)051

2025 doi
[18]

A Note on Kinematic Flow and Differential Equations for Two-Site One-Loop Graph in FR W Spacetime

Yanfeng Hang and Cong Shen. A Note on Kinematic Flow and Differential Equations for Two-Site One-Loop Graph in FR W Spacetime. 10 2024

2024
[19]

The cos- mological tree theorem

Santiago Agui Salcedo and Scott Melville. The cos- mological tree theorem. JHEP, 12:076, 2023. doi: 10.1007/JHEP12(2023)076

2023 doi
[20]

Cosmologi- cal Cutting Rules

Scott Melville and Enrico Pajer. Cosmologi- cal Cutting Rules. JHEP, 05:249, 2021. doi: 10.1007/JHEP05(2021)249

2021 doi
[21]

The Cosmological Optical Theorem

Harry Goodhew, Sadra Jazayeri, and Enrico Pajer. The Cosmological Optical Theorem. JCAP, 04:021, 2021. doi: 10.1088/1475-7516/2021/04/021

2021 doi
[22]

Cutting cosmological cor- relators

Harry Goodhew, Sadra Jazayeri, Mang Hei Gordon Lee, and Enrico Pajer. Cutting cosmological cor- relators. JCAP, 08:003, 2021. doi:10.1088/1475- 7516/2021/08/003

2021 doi
[23]

Hidden Zeros of the Cosmological Wavefunction

Shounak De, Shruti Paranjape, Andrzej Pokraka, Marcus Spradlin, and Anastasia Volovich. Hidden Zeros of the Cosmological Wavefunction. 3 2025

2025
[24]

Man- dal, Pierpaolo Mastrolia, and Francisco Vaz˜ ao

Paolo Benincasa, Giacomo Brunello, Manoj K. Man- dal, Pierpaolo Mastrolia, and Francisco Vaz˜ ao. One- loop corrections to the Bunch-Davies wave function of the universe. Phys. Rev. D , 111(8):085016, 2025. doi: 10.1103/PhysRevD.111.085016

2025 doi
[25]

Scattering Forms and the Positive Geometry of Kinematics, Color and the Worldsheet

Nima Arkani-Hamed, Yuntao Bai, Song He, and Gong- wang Yan. Scattering Forms and the Positive Geometry of Kinematics, Color and the Worldsheet. JHEP, 05:096,
[26]

Amplitubes: Graph Cosmohedra

Ross Glew and Tomasz Lukowski. Amplitubes: Graph Cosmohedra. 2 2025

2025
[27]

Cosmohedra

Nima Arkani-Hamed, Carolina Figueiredo, and Francisco Vaz˜ ao. Cosmohedra. 12 2024

2024
[28]

Correlator Polytopes

Carolina Figueiredo and Francisco Vaz˜ ao. Correlator Polytopes. 6 2025

2025
[29]

Wavefunction coefficients from amplitubes

Ross Glew. Wavefunction coefficients from amplitubes. JHEP, 07:064, 2025. doi:10.1007/JHEP07(2025)064

2025 doi
[30]

A realization of graph associahedra

Satyan L Devadoss. A realization of graph associahedra. Discrete Mathematics, 309(1):271–276, 2009

2009
[31]

The subtle simplicity of cosmological correlators

Chandramouli Chowdhury, Arthur Lipstein, Jiajie Mei, Ivo Sachs, and Pierre Vanhove. The subtle simplicity of cosmological correlators. JHEP, 03:007, 2025. doi: 10.1007/JHEP03(2025)007

2025 doi
[32]

Cosmological Dressing Rules

Chandramouli Chowdhury, Arthur Lipstein, Joe Mar- shall, Jiajie Mei, and Ivo Sachs. Cosmological Dressing Rules. 3 2025

2025
[33]

The in-out formal- ism for in-in correlators

Yaniv Donath and Enrico Pajer. The in-out formal- ism for in-in correlators. JHEP, 07:064, 2024. doi: 10.1007/JHEP07(2024)064

2024 doi
[35]

Bourjaily, Freddy Cac- hazo, Alexander B

Nima Arkani-Hamed, Jacob L. Bourjaily, Freddy Cac- hazo, Alexander B. Goncharov, Alexander Postnikov, and Jaroslav Trnka. Grassmannian Geometry of Scat- tering Amplitudes . Cambridge University Press, 4
[2016]

doi: 10.1017/CBO9781316091548

ISBN 978-1-107-08658-6, 978-1-316-57296-2. doi: 10.1017/CBO9781316091548

doi
[2018]

doi:10.1007/JHEP05(2018)096

2018 doi

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.

[1] [1]

intersect if t ∩ t′ ̸= ∅ and t ̸⊂ t′ and t′ ̸⊂ t,

work page

[2] [2]

touch if t ∩ t′ = ∅ and t ∪ t′ ∈ TG,

work page

[3] [3]

A tubing τ is then defined to be a maximal set of pair- wise compatible tubes of TG

are compatible if they do not intersect or touch. A tubing τ is then defined to be a maximal set of pair- wise compatible tubes of TG. The set of tubings will be denoted by Γ G. Finally, with this terminology introduced, the ampli- tube associated to the graph G is defined as AG = X τ ∈ΓG Y t∈τ 1 Ht . (13) For a more detailed discussion of these functions...

work page

[4] [4]

Pimentel

Nima Arkani-Hamed, Daniel Baumann, Hayden Lee, and Guilherme L. Pimentel. The Cosmological Bootstrap: Inflationary Correlators from Symmetries and Singulari- ties. JHEP, 04:105, 2020. doi:10.1007/JHEP04(2020)105

doi:10.1007/jhep04(2020)105 2020

[5] [5]

From locality and unitarity to cosmological correlators

Sadra Jazayeri, Enrico Pajer, and David Stefanyszyn. From locality and unitarity to cosmological correlators. JHEP, 10:065, 2021. doi:10.1007/JHEP10(2021)065

work page doi:10.1007/jhep10(2021)065 2021

[6] [6]

Cosmological Polytopes and the Wavefunc- tion of the Universe

Nima Arkani-Hamed, Paolo Benincasa, and Alexander Postnikov. Cosmological Polytopes and the Wavefunc- tion of the Universe. 9 2017

work page 2017

[7] [7]

Cosmological Polytopes and the Wave- functon of the Universe for Light States

Paolo Benincasa. Cosmological Polytopes and the Wave- functon of the Universe for Light States. 9 2019

work page 2019

[8] [8]

The Geometry of Cosmological Correlators

Paolo Benincasa and Gabriele Dian. The Geometry of Cosmological Correlators. SciPost Phys. , 18:105, 2025. doi:10.21468/SciPostPhys.18.3.105

work page doi:10.21468/scipostphys.18.3.105 2025