{"id":"23714e00-8d55-4e5c-a66c-d34d119d72a5","arxiv_id":"2509.02684","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A complete map of singularities and local series expansions for the hypergeometric building blocks of tree-level cosmological correlators.","lead":"Cosmological correlators, the observables of the early universe, contain hard nested integrals that resist analytic computation. This paper maps every singularity of their building blocks, called family trees, and gives series expansions around each one.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Completeness of the singularity classification depends on an unproven extension of Landau analysis to integrands with branch points; a direct pinch analysis of the meromorphic MB integral would settle it.","rationale":"The reader flagged the same primary weakness. I agree it is the most load-bearing: every later result (series at singularities, factorization) presupposes that the singularity list is complete. The paper provides independent evidence—explicit series—for the singularities it does list, but not for the absence of others. The branch-point assertion in Sec. 3 is the only argument for completeness, and it is a genuine gap because standard Landau analysis assumes meromorphic integrands. The proposed MB pinch test bypasses the branch-point issue entirely, since the MB integrand is meromorphic in the integration variables, making the singularity analysis rigorous and self-contained. I also note the convergence concern raised by the reader: the paper leaves the exact convergence boundary to future work. I consider it secondary because the series are Horn-type and the examples converge in finite domains; a nonzero radius is plausible but unproven. Hence the verdict should remain CONDITIONAL: the central claims are likely correct but require the completeness proof (or the MB pinch check) and a convergence statement before being taken at face value.","tokens_in":48166,"tokens_out":18355,"duration_ms":172953,"concrete_test":"Perform a self-contained Landau/pinch analysis on the MB representation (23), which is meromorphic in the Mellin variables s_j for fixed ω (all singularities in s are poles of Γ factors). For a nontrivial tree such as [[1(2)(34)]] or the five-site tree of Fig. 3, enumerate all pinch and endpoint conditions as the ω ratios vary, including limits to 0 and ∞, and verify that the resulting set of hypersurfaces in ω-space is exactly {Σ_{i∈G} ω_i = 0 or ∞ for G a root-bearing connected subgraph}. If any additional hypersurface appears, Result 1 is incomplete; if the lists match, the branch-point assertion is confirmed and the completeness concern is settled.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Result 1 (\"a family tree has a singularity iff the sum of energies of a root-bearing subgraph goes to zero or infinity\") is the foundation for every series in the paper. Its proof in Sec. 3 applies Landau analysis to the Fourier energy integral (29). That integrand is not meromorphic in the integration variables: each factor (iϵ_i)^{-q_i}/Γ(1-q_i) has branch points at ϵ_i=0 and ϵ_i=∞. The paper asserts these branch points are \"inert\" and \"do not move with external variables,\" so \"the Landau analysis is still applicable,\" but gives no proof and only refers to a modified analysis in [77]. If a branch point can participate in a pinch with a moving pole at a non-root-bearing partial-energy condition, the \"only if\" direction fails: there would be singularities not on the list, and the claimed completeness—and with it the exhaustiveness of the series in (36), (68), and the factorization theorem (70)—would be invalid. The found singularities are each supported by explicit convergent series, but the absence of others is supported only by this asserted Landau step.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the analytic structure of 'family trees', the multivariate hypergeometric functions that arise as canonical building blocks of tree-level cosmological correlators after the family-tree decomposition. It claims a complete characterization of singularities in energy space: a family tree is singular exactly when the sum of energies of a root-bearing connected subgraph tends to zero or infinity (Result 1, Sec. 3). For each such singularity it derives hypergeometric series expansions: Eq. (36) for infinite partial energies, Eq. (68) for zero partial energies, with the small total-energy expansion (47) as a special case. It also proves a factorization theorem (Eq. (70)) stating that the singular part at a zero partial-energy limit factorizes into subgraphs to all orders, and shows that regularized family trees are entire functions of the twist parameters (Sec. 7). The technical engine is a Mellin-Barnes representation of the time integrals, evaluated by a pole-collecting algorithm whose equivalence to the conic-hull method is proved in App. D; the results are cross-checked on many examples in App. E. As a byproduct, the paper supplies a proof of the previously unproved large total-energy series (34).","tokens_in":48303,"tokens_out":7250,"duration_ms":73572,"significance":"If the completeness claim holds, this is a substantial contribution to the analytical theory of cosmological correlators: it converts the singularity structure of a large class of hypergeometric functions into explicit, parameter-free local series, which can serve as boundary data for dispersive bootstrap and numerical evaluation. The derivations are unusually explicit, with a systematic pole-collecting algorithm, a proof of equivalence to the conic-hull method, and many fully worked examples in App. E. The main caveat is that the 'only if' direction of the singularity classification rests on an unproven extension of Landau analysis to an integrand with branch points; this is a correctness-risk point that the authors should repair or explicitly qualify. Conditional on that repair, the paper's results would justify the advertised applications.","major_comments":[{"comment":"The completeness ('only if') direction of the singularity classification is obtained by applying Landau analysis to the energy integral (29), whose integrand contains branch points at ϵ_i = 0 and ϵ_i = ∞ in addition to the moving poles. The manuscript asserts that these branch points are 'inert' and 'do not move with external variables', and therefore that 'the Landau analysis is still applicable', but no proof or precise supporting theorem is given; the reference to a modified analysis in [77] is not developed. Because the exhaustiveness of the series in (36), (68), and the factorization theorem (70) depends on this classification, the 'only if' direction is currently unproven. Please provide a direct pinch analysis of the Mellin-Barnes integral, or quote a theorem covering integrands with fixed algebraic branch points, or explicitly restrict Result 1 to the directions that are proven.","section":"Sec. 3, Result 1 (Eq. (29))"},{"comment":"The paper defines 'formally convergent' series and states that the exact convergence boundary is left to future work (Sec. 2.3), yet Results 2–4 claim exact series representations with finite convergent domains. The conic-hull equivalence in App. D establishes formal convergence only; it does not prove that the resulting multivariate series converges in any open neighborhood of the singular point. Since the local-expansion claim is load-bearing, please provide or cite a convergence theorem for balanced multifold Mellin-Barnes integrals of the type (23), or state the convergence guarantee as an explicit assumption throughout the statements of Results 2–4.","section":"Sec. 2.3 and App. D"},{"comment":"The proof of the factorization theorem relies in Step 3 on 'the analysis of Sec. 3' to conclude that only the fully factorized term can be singular at ω_G → 0. This makes the theorem contingent on the same unproven Landau completeness argument identified in the first comment. If the singularity classification is repaired, the factorization argument is sound; as written, the theorem is not an independent proof. Please either prove the needed direction of the classification within the theorem's proof or cite a verification that the Landau criterion applies to the integral (29).","section":"Sec. 6, Factorization theorem (Eq. (70))"}],"minor_comments":[{"comment":"The notation in (50) omits the (2πi)^{-(N-M)} factors and contour specifications that are explicit in (23); please make the measure and contours unambiguous for reproducibility.","section":"Sec. 2.3, Eq. (50)"},{"comment":"The Pochhammer symbol is defined as (q)_n = Γ(q+n)/Γ(q) in Eq. (82), but footnote 6 writes (a)_n ≡ Γ(a+n)/Γ(n); this is a typo and should be corrected.","section":"Footnote 6, Sec. 5.1"},{"comment":"The phrase 'series with finite convergent domains' in the abstract is stronger than what is established in Sec. 2.3, where the exact boundary is left to future work; consider wording such as 'finite, not fully characterized, convergence domains'.","section":"Abstract and Sec. 2.3"},{"comment":"The text repeatedly refers to Figs. 4–6 for the pole-collecting algorithms, but the corresponding figure content is not visible in the body text provided; please ensure the figures are present, legible, and captioned so that the pole choices are self-explanatory.","section":"Figures 4–6"}],"recommendation":"major_revision","confidential_remarks":"The main risk is the unproven Landau step in Sec. 3, which underpins the completeness claims. This is not a disagreement with external consensus; it is an internal completeness gap that the authors can plausibly close with a direct pinch analysis of the Mellin-Barnes integral or a precise external theorem. The novelty relative to the authors' previous works [75,76] is clear, and the paper contains many valuable explicit computations. If the authors supply the missing proof or carefully restate the completeness result under the needed assumption, I would support acceptance."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a solid, technically serious paper. It proves the previously unproved total-energy series (34), derives the new large and small partial-energy series (36) and (68), the factorization theorem (70), and the twist-space meromorphicity result. The Mellin-Barnes machinery is systematic, the pole-collecting algorithm is laid out clearly, the conic-hull equivalence in App. D is a real check, and the worked examples in App. E are cross-validated. I found no circularity, no parameter fitting, and no invented physics.\n\nThe soft spot is the completeness claim. Result 1's \"iff\" depends on applying Landau analysis to the energy integral (29), whose integrand has branch points at epsilon_i = 0 and infinity. The paper says these are \"inert\" and just gives nontrivial monodromy, but does not prove that a branch point cannot participate in a pinch with a moving pole at a non-root-bearing partial-energy condition. On reading the integral, I think the assertion is correct: the branch points sit at the fixed integration endpoints, and the only moving poles are at root-bearing successive energy sums. Non-root-bearing partial energies never appear as poles of any chain integrand. Still, the proof is missing. A direct pinch analysis of the meromorphic MB representation (23) would settle it, and I would ask the authors to add that.\n\nThe convergence-boundary point the reader raised is minor. The abstract says \"finite convergent domains,\" which is exactly what is proven: each series converges in some finite subdomain, with the exact boundary left open. I do not read that as an overstatement.\n\nI also want to give credit for honesty: the paper flags the limitations itself, including the unknown convergence boundaries and the \"abuse of language\" in pole collecting. The proof of the factorization theorem is partly formal, but the theorem is also a direct corollary of the explicit series (68), so it stands on firmer ground than the completeness claim.\n\nWho this is for: anyone computing or bootstrapping massive tree-level inflation correlators. The local expansions and the factorization theorem should be directly useful, and the proof of the total-energy series fills a real gap. It deserves a serious referee. Even if the Landau step needs strengthening, the individual series and the factorization are valuable and publishable. My recommendation: send it to review, with a request to tighten the completeness argument.\n\nSend it.","headline":"A serious, technically rich paper on family-tree singularities; the completeness claim rests on one unproven Landau step that should be tightened but is likely correct.","tokens_in":48885,"tokens_out":3420,"would_cite":true,"duration_ms":32960,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proves that a family tree—a multivariate hypergeometric function arising from time-ordered integrals in cosmological correlators—is singular exactly when a root-bearing partial energy tends to zero or infinity, and derives…","keywords":["family trees","cosmological correlators","multivariate hypergeometric functions","Mellin-Barnes integrals","analytic continuation","factorization theorem","time-ordered integrals","singularity structure"],"falsifier":"Compute the three-site star family tree $[1(2)(3)]$ along a curve where the non-root partial energy $\\omega_2+\\omega_3$ goes to zero while $\\omega_1$, $\\omega_{12}$, $\\omega_{13}$, and $\\omega_{123}$ stay finite; the completeness claim predicts a regular function there, so any divergence or branch cut on that curve would refute it. A second check is to verify numerically that the small partial-energy series (68) converges in a nonzero neighborhood of a zero partial-energy point, since the paper leaves the exact convergence boundary open.","tokens_in":47897,"feed_emoji":"🌳","tokens_out":10945,"duration_ms":100833,"temperature":0.7,"pith_summary":"This paper sets out to give a complete local description of family trees—the multivariate hypergeometric functions produced by the recently introduced family-tree decomposition of time-ordered multilayer integrals in cosmological correlators. The central claim is that a family tree, viewed as a function of complex energies, is singular exactly when the sum of energies over a connected subgraph containing the root goes to zero or to infinity, and that no other singularities exist. Around every such point the paper constructs explicit hypergeometric series with finite convergence domains, including a large partial-energy series, a small partial-energy series, and a factorization theorem asserting that the singular part at a zero partial-energy limit factorizes into subgraph contributions to all orders. These expansions matter because the family trees are the building blocks of tree-level massive correlators; knowing their singularities and having local series at each singularity makes the functions analytically controllable rather than opaque special functions.","feed_headline":"One rule finds every singularity of a cosmological family tree","feed_subtitle":"Exact hypergeometric expansions now cover the neighborhood of every zero or infinite partial energy.","key_machinery":"The load-bearing mechanism is the Mellin-Barnes representation of family trees, used a second time inside the time integral. By integrating the nested time integral layer by layer and resolving the resulting special functions (exponential integrals or confluent hypergeometric functions) into gamma-function products, an arbitrary family tree becomes a multi-fold Mellin-Barnes integral in which energies appear only as ratio powers and twists as gamma-function arguments. The paper evaluates these integrals with a pole-collecting algorithm: after choosing a total order of the energy magnitudes, one selects the set of gamma-function pole families whose residue series is formally convergent, and the paper proves this algebraic selection is equivalent to the geometric conic-hull method. The flexibility to resolve the special functions completely or partially—and to collect or not collect energies along lines—is what enables reaching every singularity, not just the previously known large-single-energy and large-total-energy expansions.","core_discovery":"Result 1 of the paper states: a family tree has a singularity in the complex energy space if and only if the total energy of a root-bearing subgraph goes to zero or to infinity. The argument converts the time-ordered integral into an energy integral through a Fourier transform, applies a standard Landau singularity analysis to the poles of that integrand, and finds only endpoint singularities controlled by successive sums of energies starting at the root. At each infinite partial-energy singularity the whole family tree is a single multivariate hypergeometric series times a complex power (Eq. 36), with the nonanalyticity entirely in the power. At each zero partial-energy singularity the tree splits into a universal singular piece—a hypergeometric series times a singular power—plus regular pieces whose form depends on the ordering of the remaining energies (Eq. 68). A corollary is the factorization theorem (Eq. 70): at a zero partial-energy limit the singular part of the whole tree is the singular part of the root-bearing subgraph multiplied by the product of the untouched subgraphs, to all orders in the small partial energy. In twist space the singularities are only simple poles from gamma factors, and the regularized family trees are entire functions of all twists.","pith_inferences":["Editorial inference: the same Mellin-Barnes plus pole-collecting strategy should apply to massive family trees used for full massive correlators, so the singularity classification and the factorization theorem would likely survive that extension; the paper only suggests this as future work.","Editorial inference: because the new series provide boundary data at every singularity, computing their exact convergence radii via Stirling asymptotics is a concrete next test of how much of energy space these expansions actually cover.","Editorial inference: the factorization theorem at zero partial energies suggests that the nonanalytic part of a cosmological correlator is determined entirely by the singular subgraph, giving a direct physical picture of divergent limits as on-shell processes localized in that subgraph."],"forward_implications":["Every family tree now has a convergent hypergeometric expansion in the neighborhood of each of its singularities, so analytic continuation across the energy space can be assembled from local data.","The previously unproved large total-energy series (Eq. 34) is proved as a special case of the general infinite partial-energy expansion.","At zero partial energies the singular part factorizes to all orders into the singular subgraph times the disjoint subgraphs, so power-law divergences of the full tree are localized in a single subgraph.","Removing gamma factors leaves regularized family trees that are entire functions of all twists, so the twist-space singularity structure is exactly those simple poles.","The results supply the local singular data needed to reconstruct correlators by dispersion and to seed numerical evaluation of these hypergeometric functions."],"supporting_citations":[{"why":"Introduces the family-tree decomposition and the large-single-energy series that define the objects whose singularities this paper characterizes.","marker":"[75]"},{"why":"Defines family trees as multivariate hypergeometric functions and states the total-energy series that this paper proves as a corollary.","marker":"[76]"},{"why":"Supplies the Landau-analysis lemmas used to classify all candidate energy singularities of the integral representation.","marker":"[88]"},{"why":"Provides the conic-hull method for producing multiple series representations of multi-fold Mellin-Barnes integrals, which the paper shows equivalent to its pole-collecting procedure.","marker":"[86]"},{"why":"Extends the conic-hull and triangulation approach to general point configurations, supporting the formal-convergence check for each selected pole family.","marker":"[87]"}],"fun_headline_variants":["All singularities of cosmological family trees pinpointed","Exact series expansions at every family tree singularity","One criterion finds all singularities of family trees","Family trees factor at zero partial energy singularities","Mellin maps every singularity of a cosmological family tree"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The completeness of the singularity list rests on the assertion that the branch points of the energy-integral representation are inert—they sit at fixed positions and do not move with the external energies—so that only pole pinches can create singularities; if that assertion fails, singularities beyond the listed zero or infinite root-containing partial-energy sums would exist.","fun_headline_variants_meta":{"raw":{"variants":["All singularities of cosmological family trees pinpointed","Exact series expansions at every family tree singularity","One criterion finds all singularities of family trees","Family trees factor at zero partial energy singularities","Mellin maps every singularity of a cosmological family tree"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000189,"raw_usage":{"total_tokens":1330,"prompt_tokens":931,"completion_tokens":399,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":547,"completion_tokens_details":{"reasoning_tokens":326}},"tokens_in":547,"tokens_out":399,"duration_ms":3959,"temperature":1.0,"reasoning_tokens":326,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T16:35:38.942551+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the three-site star family tree $[1(2)(3)]$ along a curve where the non-root partial energy $\\omega_2+\\omega_3$ goes to zero while $\\omega_1$, $\\omega_{12}$, $\\omega_{13}$, and $\\omega_{123}$ stay finite; the completeness claim predicts a regular function there, so any divergence or branch cut on that curve would refute it. A second check is to verify numerically that the small partial-energy series (68) converges in a nonzero neighborhood of a zero partial-energy point, since the paper leaves the exact convergence boundary open.","supporting_citations":[],"review_version":2}