{"id":"d9c0028a-7c08-4e91-b70d-c37df55b106d","arxiv_id":"2411.13857","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A coordinate-space averaging regularization is shown to be compatible with gluing partition functions on Riemannian manifolds with boundaries.","lead":"The paper defines a cutoff regularization for quantum field theories on curved spaces with boundaries, based on averaging fields inside small balls instead of cutting momenta. It proves that two regularized halves of a manifold can be glued by integrating over boundary data to recover the whole regularized theory.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The admissible averaging kernel in Definition 18 is never constructed; Theorem 1 and Lemma 10 both rest on it, so the gluing result is conditional on an existence assumption.","rationale":"I read the full text in good faith. The formal-power-series proof of Theorem 1 is internally coherent: the stage-by-stage replacement of pairings using formulas (41) and (42) is a plausible diagrammatic identity, and the free-field limit, where S reduces to 1, checks out. The main reason I cannot regard the central claim as established is the status of the admissible averaging kernel. Definition 18 is an existence assumption, and the text explicitly defers the verification of the only proposed example ('it can be argued'). The proof of Lemma 10 also requires more than Definition 18 states: it needs the averaged Green's function to be regular enough that products appearing in arbitrary Feynman diagrams are finite, and it needs the double average to be continuous. These are not automatic from a single-average continuity condition unless additional mapping properties of H_omega are proved. Lemma 9's restriction construction is likewise only shown on M_i,Lambda, not on the full submanifold, so the extension of admissibility to the submanifolds is incomplete. This is precisely the reader's weakest_assumption, and my reading does not change the verdict: if the kernel existence is supplied and Theorem 1 is restated as conditional on it, the gluing result appears defensible. I therefore recommend keeping the CONDITIONAL verdict.","tokens_in":30758,"tokens_out":21832,"duration_ms":228702,"concrete_test":"Take a compact Riemannian manifold with boundary, e.g., a hemisphere or a flat cylinder with two boundary components, and explicitly write down a candidate kernel: first the geodesic-sphere average of Definition 19, then a mollified kernel obtained by pulling back a smooth radial Euclidean kernel through the exponential map and normalizing. Verify numerically or analytically that (28) holds, that both delta-limits in Definition 18 are satisfied, and that Lemma 9's renormalized restriction omega_i belongs to Omega_0^Lambda(M_i) including near the cut. If the sphere kernel fails, Theorem 1 must be restated with an explicit admissible kernel; if it passes, the conditional verdict is resolved.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central gluing theorem depends on the class Omega_0^Lambda(M) of admissible averaging kernels (Definition 18), yet no such kernel is constructed on a general compact Riemannian manifold with boundary. After Definition 19 the paper only says that geodesic-sphere averaging 'can be argued' to be acceptable by Euclidean analogy. This matters at two load-bearing points. First, Lemma 10 (regularized amplitudes finite) requires the doubly averaged Green's function G^Lambda(p_i,p_j) in (35) to be continuous and bounded on M_Lambda x M_Lambda; Definition 18 only guarantees a single average int omega(p,p_1;Lambda)G(p_1,p_2) is continuous for fixed p_2, and it does not imply that H_omega maps the singular Green's function to a function to which H_omega can be applied again with the needed regularity. Second, Lemma 9 defines omega_i by restricting omega to M_i and renormalizing; near the cut surface Sigma the normalization factor int_{M_i} omega(p,.) differs from 1, and the resulting omega_i may fail the support or delta-limit conditions of Definition 18 on M_i. Equality (30) is only asserted on M_i,Lambda, so the replacement of G_i by G in the proof of Lemma 11 is not fully justified. The sphere-average example does not address boundary truncation or the C^j requirement needed for Corollary 10.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a coordinate-space cutoff regularization for scalar field theory on a compact Riemannian manifold with boundary, defined by deforming the interaction term of the classical action via an averaging operator with a kernel in a class Ω_0^Λ(M). The central claim is that this regularization is consistent with gluing of manifolds and partition functions: Theorem 1 (Eqs. (43)–(44)) states that the functional integral over boundary data on the gluing hypersurface reproduces the regularized partition function of the glued manifold in the limit where the deformed submanifolds fill the deformed manifold. The proof is carried out at the level of formal power series in √ℏ, using Wick contractions and external gluing identities for Green's functions from Refs. [62] and [24]. Corollaries extend the statement to position-dependent coupling coefficients and multiplicative renormalization.","tokens_in":31035,"tokens_out":3504,"duration_ms":35967,"significance":"If the theorem holds, it would provide a regularization scheme compatible with Atiyah-Segal-style gluing in all dimensions for scalar theories on curved manifolds, which would be a genuinely useful contribution to the constructive and perturbative QFT literature. The paper is careful to set up the formal power series framework, and the Wick-contraction proof in Section 4.2 is detailed and largely transparent. The reliance on previously established gluing identities for Green's functions is appropriate and does not make the argument circular. The main value is conditional: the theorem is proved for any kernel in the class Ω_0^Λ(M), but the paper does not establish that this class is nonempty in the stated geometric generality.","major_comments":[{"comment":"No admissible kernel is actually constructed on a general compact Riemannian manifold with boundary. Definition 18 imposes continuity of the single average, support in B_{1/Λ}(p,M), normalization, and two delta-function limits, but the only example given, geodesic-sphere averaging, is justified by a Euclidean asymptotic analogy and does not address boundary truncation or the regularity needed for Corollary 10. Since Lemma 10, Lemma 11, and Theorem 1 all quantify over ω ∈ Ω_0^Λ(M), the theorem is conditional on an existence assumption that is never proved. Either a construction of an admissible kernel should be supplied, or the main theorem should be stated explicitly as a conditional statement with the existence of Ω_0^Λ(M) as a hypothesis.","section":"Section 3.2, Definitions 18–19"},{"comment":"The proof that regularized amplitudes are finite requires that the doubly averaged Green's function G^Λ(p_i,p_j) in (35) be continuous and bounded on M_Λ × M_Λ. Definition 18 only guarantees that for fixed p_2 the single average H_ω G(·, p_2) is continuous; it does not imply that H_ω can be applied a second time with the needed regularity. A separate continuity/boundedness statement for the iterated average, or an additional condition in Definition 18, is needed to justify Lemma 10.","section":"Lemma 10 and Eq. (35)"},{"comment":"The proof of Lemma 9 defines ω_i by restricting ω and renormalizing by ∫_{M_i} ω(p, p_2). For p ∈ M_i,Λ this prefactor equals 1 because the support of ω(p, ·) lies away from Σ, so equality (30) holds there; however, the proof does not verify that the renormalized ω_i satisfies the support, normalization, and delta-limit conditions of Definition 18 for all p ∈ M_i, especially for p in the collar where B_{1/Λ}(p, M) crosses Σ. Since Lemma 11 uses ω_i only for arguments in M_i,Λ, the present gap may be repairable, but the stated existence claim for ω_i ∈ Ω_0^Λ(M_i) is not established as written.","section":"Lemma 9 and Eq. (30)"}],"minor_comments":[{"comment":"There are several typos that should be corrected, e.g., 'knowm' (Section 2.4), 'beetween' (Section 3.3), 'cinsists' (Lemma 10), 'correcp onding' (Section 3.2), and 'partitial functions' (Introduction).","section":"Section 2.4 and Section 3.1"},{"comment":"The definition of Λ_1 via the minimum of d(p,q) for p ∈ Σ and q ∈ Y assumes both Σ and Y are nonempty; the case ∂M = ∅, mentioned in Section 2.1, is not covered by this definition.","section":"Definition 3 and Remark 2"},{"comment":"The two limit transitions in (28) are stated informally as 'for Λ → +∞ on the class C∞(M,R)'. It would improve rigor to specify the precise topology or test-function space in which the delta-function limits are taken.","section":"Definition 18"},{"comment":"The proof of Corollary 10 asserts that the internal structure of the coefficients was not used, but for local differential operator couplings the regularization requires a kernel of suitable smoothness (as the corollary itself notes). This further highlights the need for a concrete construction of kernels in Ω_j^Λ(M) with j > 0.","section":"Section 4.3, Corollary 10"},{"comment":"The displayed chain of equalities for the free determinant ratio would benefit from an explicit statement of the normalization convention used for the functional measure, since the quotient of two formal Gaussian integrals is otherwise ambiguous.","section":"Section 5.1, 'On the regularization of the determinant'"}],"recommendation":"major_revision","confidential_remarks":"The paper's central theorem is a formal power series statement whose proof is internally consistent given the assumed class of admissible kernels. The main weakness is the unproved existence of such kernels on general manifolds with boundary; this is not a circularity or a falsity but a missing load-bearing construction. I would recommend major revision rather than rejection, because the gap appears fixable: a local construction of averaging kernels away from the boundary, combined with a careful treatment of the boundary collar, should suffice to fill the hole. I also note that the manuscript contains a substantial number of self-citations to the author's prior work; this is understandable given the topic but the novel content should be clearly delineated. The historical dedication on the title page is unusual for a math-ph journal but not inappropriate."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things to know. The paper proves a new gluing statement: under this coordinate-space averaging cutoff, the formal partition functions on two halves of a compact Riemannian manifold with boundary glue, after integration over boundary data, to the partition function on the whole manifold, in any dimension for scalar fields. That is genuinely new—the earlier Euclidean constructions did not address gluing on curved manifolds. Second, and more important, the proof is conditional on an existence assumption that is never discharged. The admissible averaging kernel of Definition 18 is not constructed on a general curved manifold with boundary; the paper says sphere averaging 'can be argued' to work by Euclidean analogy, and that is not a proof.\n\nCredit where it is due. The paper is explicit about the formal-power-series setting rather than pretending to a measure-theoretic functional integral. The Wick-contraction proof of Theorem 1 is detailed, the pairing bookkeeping is credible, and the use of the exterior gluing identities for Green's functions from Carron and from Kandel–Mnev–Wernli is legitimate. The heavy self-citation is for background regularization technology, not for the target theorem, so I do not read it as circular.\n\nThe soft spots, in order of seriousness. (1) No kernel in Ω_0^Λ(M) is exhibited. Near the boundary the geodesic ball is truncated, the normalization changes, and the two delta-function limits in Definition 18 are unchecked. (2) Lemma 10 needs the doubly averaged Green's function in (35) to be continuous, but Definition 18 controls only a single average; the second averaging step is not justified. (3) Lemma 9 defines the submanifold kernels by restriction and renormalization, but the resulting objects may fail the support and normalization conditions of Definition 18 on M_i; equality (30) is only asserted on M_{i,Λ}, and Lemma 11 relies on it. The stress-test note is accurate on all three points. These gaps make the theorem conditional rather than false; if a suitable kernel is constructed, or explicitly assumed as a hypothesis, the gluing argument looks sound. The corollaries on gauge and spinor fields and on multiplicative renormalization are asserted, and Section 5.1 itself presents the wider applicability as a recipe, not a theorem.\n\nWho gets value: mathematicians and mathematical physicists working on perturbative QFT on manifolds with boundary, especially those wanting a gluing-compatible regularization. It deserves a serious referee. The right outcome is major revision: construct the kernel or state Theorem 1 as conditional, and close the regularity gaps in Lemmas 9 and 10.","headline":"A genuinely new gluing statement for coordinate-space cutoff regularization, but Theorem 1 is conditional on an unconstructed averaging kernel on curved manifolds with boundary.","tokens_in":31509,"tokens_out":3548,"would_cite":false,"duration_ms":32902,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["81T15","81T20","81S40","58J32"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves that replacing the interaction fields by averages over a geodesic ball of radius $1/\\Lambda$ gives a cutoff regularization of scalar field theory on curved manifolds that is compatible with gluing manifolds and partition…","keywords":["effective action","cutoff regularization","coordinate-space cutoff","averaging operator","gluing of partition functions","quasi-locality","Riemannian manifold with boundary","scalar field theory"],"falsifier":"Take a compact hyperbolic manifold with boundary and attempt to write the sphere-averaging kernel of Definition 19 explicitly; if the averaged Green's function fails the continuity or delta-limit conditions of Definition 18 for some point near a boundary component, Lemma 9 and Theorem 1 have no instance on that manifold. A direct analytic or numerical check of the two-loop contribution to both sides of (44) for $\\varphi^4$ on a three-dimensional ball cut into two hemispheres, at finite $\\Lambda$, would also settle the claim order by order in $\\hbar$.","tokens_in":30541,"feed_emoji":"🧩","tokens_out":9955,"duration_ms":86170,"temperature":0.7,"pith_summary":"This paper proposes a coordinate-space cutoff regularization for the effective action of a scalar field theory on a compact Riemannian manifold with boundary, and proves that the regularization is compatible with gluing manifolds along a common boundary. The cutoff replaces the fields entering the interaction term by averages over geodesic balls of radius $1/\\Lambda$ and restricts the interaction to the deformed manifold $M_\\Lambda$ obtained by deleting a boundary collar. The central result, Theorem 1, says that the functional integral over boundary data on the cutting surface $\\Sigma$ of the product of the two regularized partition functions equals the regularized partition function of the glued manifold, with a limit transition that reconstructs $M_\\Lambda$. If the claim is correct, one can compute the regularized theory on pieces and assemble the pieces without introducing new ultraviolet divergences at the gluing step.","feed_headline":"Cutoff-regularized partition functions glue cleanly along boundaries","feed_subtitle":"Averaging interaction fields in a geodesic ball removes divergences while preserving the gluing identity.","key_machinery":"The load-bearing object is the deforming operator $H^\\Lambda_\\omega$, an integral operator with kernel $\\omega(\\cdot,\\cdot;\\Lambda)\\in\\Omega^\\Lambda_0(M)$ that averages a field over a geodesic ball of radius $1/\\Lambda$. Applied only to the interaction $S_{\\mathrm{int}}$ and accompanied by the replacement $M\\to M_\\Lambda$, it makes the interaction quasi-local and turns every Green's function appearing in Wick contractions into a bounded, regularized version $G^\\Lambda(p_1,p_2)$. The proof of Theorem 1 uses Lemma 11, which says that the deformed Green's function on the whole manifold decomposes into the deformed Green's functions on the two pieces plus a boundary-to-boundary term built from the Dirichlet-to-Neumann operators and the Green's function on the cutting surface $\\Sigma$; this decomposition is exactly what makes the functional integral over $\\eta_\\Sigma$ reproduce the glued partition function.","core_discovery":"On the paper's own terms, the central discovery is that the cutoff defined by a deforming operator $H^\\Lambda_\\omega$—averaging each interaction field with a kernel $\\omega(p,p_1;\\Lambda)$ supported in a geodesic ball of radius $1/\\Lambda$ and normalized to unit integral—respects the gluing of partition functions. For a manifold $M=M_l\\cup_\\Sigma M_r$, with boundary data split as $\\eta_l+\\eta_\\Sigma$ and $\\eta_r+\\eta_\\Sigma$ on the two pieces, the glued object $\\int \\mathcal{D}\\eta_\\Sigma\\, Z(\\eta_l+\\eta_\\Sigma;M_{l,\\Lambda})Z(\\eta_r+\\eta_\\Sigma;M_{r,\\Lambda})$ is shown in formula (43) to equal $Z(\\eta_l+\\eta_r;M_{l,\\Lambda}\\cup M_{r,\\Lambda})$, and formula (44) passes to the limit $M_{l,\\Lambda}\\cup M_{r,\\Lambda}\\to M_\\Lambda$ to give the partition function of the whole manifold. The proof works order by order in the perturbative expansion by rewriting the Wick pairings so that the deformed Green's functions on the pieces combine, through the standard gluing relations for Green's functions, into the deformed Green's function on the whole manifold.","pith_inferences":["If an admissible kernel can actually be constructed on every compact Riemannian manifold with boundary, the construction would give a dimension-independent, geometrically local regularization for perturbative QFT that is compatible with cutting and pasting the spacetime itself.","A testable extension is to compute both sides of the gluing identity (44) at finite $\\Lambda$ for a concrete model, such as $\\varphi^4$ on a ball in $\\mathbb{R}^3$ split into two hemispheres, and verify equality order by order in $\\hbar$.","The boundary pairing in (44) suggests a Hilbert-space interpretation of regularized partition functions as vectors in a space of boundary data, which may connect to state-sum and topological-QFT constructions without changing the local field content.","Because the kernel is supported in a geodesic ball, the regularization is quasi-local, so the counterterms of the renormalized theory should remain local; checking that the $\\Lambda$-dependent counterterms in a two-loop calculation are local densities on $M_\\Lambda$ would test this directly."],"forward_implications":["In every dimension, the same cutoff recipe removes the ultraviolet divergences of perturbative scalar theory on manifolds with boundary, and gluing the pieces introduces no new divergences beyond those already present on each piece.","The gluing identity (44) turns the regularized partition function into a pairing of boundary functionals, so one can compute on submanifolds and assemble the result by a functional integral over the common boundary.","If the model admits multiplicative renormalization, the renormalized partition functions inherit the gluing property, because the counterterms can be absorbed into position-dependent couplings while the theorem remains true.","The averaging construction extends to vector, gauge, and spinor fields, as well as to couplings that are local differential operators, provided an admissible kernel of sufficient smoothness exists.","The equality (43) holds whether or not the model is renormalizable, since it is a statement about the regularized formal series before the limit $\\Lambda\\to\\infty$."],"supporting_citations":[{"why":"It supplies the gluing relations for Green's functions and the boundary quadratic forms used in Lemma 2 and in formulas (41)-(42), which carry the proof of Theorem 1.","marker":"[24]"},{"why":"It establishes the identity for the inverse of the sum of Dirichlet-to-Neumann operators and the gluing formula for Green's functions used in Lemma 3 and Lemma 11.","marker":"[62]"},{"why":"It provides the boundary-value problem for harmonic extensions and the Dirichlet-to-Neumann operator behind Lemma 1 and Definition 8.","marker":"[61]"},{"why":"It defines the acceptability condition for averaging kernels and the non-negativity of the spectrum that underpin Definition 14 and Lemma 6.","marker":"[34]"},{"why":"It introduces the explicit coordinate-space cutoff regularization and the deforming operator that this paper generalizes to curved manifolds.","marker":"[33]"},{"why":"It introduces the cutoff deformation in a concrete scalar model, the template for applying the averaging operator to the interaction term.","marker":"[25]"},{"why":"It supplies the functional-derivative and Wick-expansion formalism used to define the formal functional integral and to reorganize pairings in the proof of Theorem 1.","marker":"[87]"}],"fun_headline_variants":["Averaging fields in geodesic balls preserves gluing","Geodesic-ball averaging yields consistent gluing for partition functions","Quasi-local cutoff keeps partition function gluing intact","Cutoff via averaging enables clean gluing of partition functions"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that, on every compact Riemannian manifold with boundary, there exists an admissible averaging kernel $\\omega(\\cdot,\\cdot;\\Lambda)\\in\\Omega^\\Lambda_0(M)$ satisfying Definition 18: the averaged Green's function is continuous, the kernel is supported in a geodesic ball of radius $1/\\Lambda$, it is normalized to unit integral, and it reproduces the delta function as $\\Lambda\\to\\infty$; the paper argues this by Euclidean analogy for sphere averaging but does not construct such a kernel in general.","fun_headline_variants_meta":{"raw":{"variants":["Averaging fields in geodesic balls preserves gluing","Geodesic-ball averaging yields consistent gluing for partition functions","Quasi-local cutoff keeps partition function gluing intact","Cutoff via averaging enables clean gluing of partition functions"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000543,"raw_usage":{"total_tokens":2601,"prompt_tokens":947,"completion_tokens":1654,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":563,"completion_tokens_details":{"reasoning_tokens":1587}},"tokens_in":563,"tokens_out":1654,"duration_ms":11362,"temperature":1.0,"reasoning_tokens":1587,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T15:48:15.154801+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a compact hyperbolic manifold with boundary and attempt to write the sphere-averaging kernel of Definition 19 explicitly; if the averaged Green's function fails the continuity or delta-limit conditions of Definition 18 for some point near a boundary component, Lemma 9 and Theorem 1 have no instance on that manifold. A direct analytic or numerical check of the two-loop contribution to both sides of (44) for $\\varphi^4$ on a three-dimensional ball cut into two hemispheres, at finite $\\Lambda$, would also settle the claim order by order in $\\hbar$.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"It supplies the gluing relations for Green's functions and the boundary quadratic forms used in Lemma 2 and in formulas (41)-(42), which carry the proof of Theorem 1."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"It establishes the identity for the inverse of the sum of Dirichlet-to-Neumann operators and the gluing formula for Green's functions used in Lemma 3 and Lemma 11."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"It provides the boundary-value problem for harmonic extensions and the Dirichlet-to-Neumann operator behind Lemma 1 and Definition 8."}],"review_version":1}