{"id":"694c1580-eab5-404f-b896-13dca208451e","arxiv_id":"2607.04514","paper_version":1,"verdict":"ACCEPT","confidence":"HIGH","novelty_score":7.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"Near a boundary or corner of a manifold support, the heat-regularized density admits a two-term conical-layer expansion whose leading term is the Gaussian mass of the inward tangent cone, yielding uniform singular expansions of the score, log-Hessian, and scale derivative of the score.","lead":"Gaussian smoothing of measures living on manifolds with corners has a precise two-term small-noise expansion near edges and corners, governed by the Gaussian mass of the inward tangent cone. The formulas identify how lower-dimensional support, boundaries, corners, and curvature appear as singular powers of the noise in the score and its derivatives—objects used by diffusion generative models.","discovery_kind":"extension","skeptic_critique":{"model":"grok-4.5","headline":"No significant objection identified","rationale":"The reader correctly identifies the local corner-chart/density/positivity package as the weakest (and only) load-bearing premise, and correctly notes that the paper is a complete pure-math asymptotic analysis with no data, circularity, or structural red flags. After reading the full manuscript and appendices, I find no stronger or more concrete concern that would move the verdict. The proofs close every estimate needed for the local expansions; the sketched globalization is standard and does not affect the theorems as stated. The recommended concrete test is a low-cost analytic sanity check that would still be worth performing, but is expected to pass. Hence the reader’s ACCEPT / HIGH confidence verdict stands unchanged.","tokens_in":59410,"tokens_out":463,"duration_ms":4695,"concrete_test":"Independently re-derive the leading coefficient C0 (eq. 9) and the first correction C1 (eq. 12) for the model case of uniform measure on the half-space H^d_1 (or the unit ball near a boundary point) by direct Gaussian integration; verify that they match the general formulas specialized in Section 4 and that the resulting score/log-Hessian expansions recover the classical half-space factors Φ_N and λ=ϕ_N/Φ_N.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central two-term conical-layer expansions (Theorems 3–6) rest on a standard local package: a single Cr+1 corner chart over a contractible base, a Cr density representative, and positivity of that density on the compact stratum piece KS (Assumption 1). The appendix supplies complete, self-contained proofs of the geometric frame/tubular package, admissible-class calculus, scaled Taylor expansions, Gaussian domination, tail absorption, far-field exponential smallness, and conical-layer chain rules. The only genuine limitation is that globalization beyond one chart is sketched (Remark 2) via intrinsicness of C0/C1 (Remark 25) and finite covers; this is routine and does not undermine the local claims that constitute the paper’s strongest result. No internal inconsistency, hidden unboundedness, or missing estimate appears in the argument.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.5","summary":"The paper establishes small-noise (small-σ) asymptotics for the Euclidean heat regularization p_σ of a probability measure q supported on an embedded m-dimensional manifold with corners M ⊂ R^d. In the conical boundary layer where the observation point approaches a codimension-c stratum at scale σ, the support is replaced to leading order by its inward tangent cone. Under a local corner-chart hypothesis with C^{r+1} regularity, a C^r density representative, and positivity on a compact stratum piece (Assumption 1), Theorems 3–4 give a two-term expansion of p_σ and log p_σ, with leading coefficient C_0 the Gaussian mass of the linearized cone weighted by density and adapted Jacobian, and first correction C_1 recording density/Jacobian variation and quadratic embedding geometry. Theorems 5–6 then yield uniform expansions for the score (∼σ^{-1}), log-Hessian (∼σ^{-2}), and scale derivative of the score. Localization shows the far-field contribution is exponentially small. Particular cases recover classical interior, smooth-manifold, and half-space boundary formulas.","tokens_in":59569,"tokens_out":855,"duration_ms":17453,"significance":"The work supplies a uniform, geometrically intrinsic two-term description of how ambient codimension, boundary faces, corners, and curvature enter the singular differential structure of Gaussian-smoothed singular measures. The expansions for score, log-Hessian, and ∂_σ-score are directly relevant to denoising, diffusion generative models, and Hessian-based analyses under the manifold hypothesis, going beyond leading normal attraction to next-order and corner corrections. Strengths include a complete, self-contained appendix pipeline (adapted frames and tubular coordinates; admissible classes with closure; scaled Taylor expansions and Gaussian domination; tail absorption; far-field control; conical-layer chain rules), explicit intrinsicness of C_0 and C_1 (Remark 25), and clean specialization to classical regimes. The results are rigorous local asymptotics under stated assumptions rather than heuristic scaling arguments.","major_comments":[],"minor_comments":[{"comment":"Remark 2 sketches globalization by finite charts and intrinsicness of coefficients. A short explicit sentence that the local expansions glue on overlaps because C_0, C_1 (and derived logarithmic coefficients) are chart-independent would make the routine step fully self-contained for readers who skip the appendix.","section":null},{"comment":"Section 4 (smooth manifold without boundary): the mean-curvature correction L_1 = (1/2)⟨N(x)a, h_M(x)⟩ is stated after specializing to normal geodesic coordinates. A one-line reminder that the general C_1 formula reduces to this after odd terms integrate to zero would help readers connecting Section 3 to Section 4.","section":null},{"comment":"Notation table (Table 1) and appendix Table 2 are helpful; a few symbols (e.g., the reconstructed fields J_ν(a,x,σ), Q_ν) appear in Theorems 5–6 before their full reconstruction is recalled. Cross-referencing (15)–(16) at first use in the theorem statements would reduce page-flipping.","section":null},{"comment":"Figure 2 caption and the heat-regularization illustration are useful; ensuring axis labels and the σ-sequence are legible in the final production version would improve readability.","section":null},{"comment":"Section 6 lists natural open directions (higher-order jets, multi-stratum transition regimes, weaker stratified supports). These are appropriately scoped; no change needed beyond optional brief pointers to related heat-content literature already cited.","section":null}],"recommendation":"accept","confidential_remarks":"The manuscript is unusually complete for a first arXiv version: the appendix is a full proof package rather than a sketch. Fit for a probability/analysis journal is strong; the generative-modeling motivation is well calibrated and does not overclaim empirical results. No novelty or citation concerns stood out."},"author_rebuttal":null,"desk_editor":{"model":"grok-4.5","letter":"This is a careful pure-math paper that delivers exactly what the abstract promises: uniform two-term expansions for the heat regularization of a measure on a manifold with corners, in the O(σ) conical layer around a stratum, plus the induced singular expansions of the score, log-Hessian, and scale derivative of the score.\n\nWhat is new is the package. Classical short-time heat kernels, half-space KDE bias, and leading normal attraction under the manifold hypothesis are recovered as special cases (Section 4). The contribution is the simultaneous treatment of ambient codimension k, stratum/corner codimension c, two-term remainders with uniform derivative control, and the differentiated objects that actually appear in denoising and diffusion models. C0 is the Gaussian mass of the linearized inward cone times density and adapted Jacobian; C1 folds in density/Jacobian variation and quadratic embedding geometry. The localization argument makes the far-field contribution exponentially small, so the local expansion is the full expansion at the stated order.\n\nThe appendix is the real work: adapted frames and tubular coordinates, admissible classes with closure, scaled Taylor expansions, Gaussian domination, tail absorption, far-field control, and conical-layer chain rules. The proofs look complete and self-contained. Coefficients are shown intrinsic (Remark 25), so the single-chart localization is not a conceptual hole; globalization by finite covers is routine (Remark 2).\n\nSoft spots are minor and proportional. The load-bearing package is the usual local corner chart + Cr density + positivity on the compact stratum piece; without positivity the log and differentiated expansions fail, which the authors flag. There is no data, no circularity, and no hidden unboundedness. The generative-modeling discussion is motivational framing, not a claim that the paper solves sampling rates.\n\nThis is for people who need precise small-noise geometry of scores near singular supports—math.PR, geometric statistics, and anyone analyzing stiffness of reverse SDEs or DDPM/DDIM discretizations near manifolds with boundary. It deserves a serious referee. I would engage with it and expect to cite the expansions when the geometry of the support matters.","headline":"Solid, complete two-term conical-layer asymptotics for Gaussian-smoothed measures on manifolds with corners, with clean score/Hessian expansions; the local package is standard and the proofs are thorough.","tokens_in":60176,"tokens_out":530,"would_cite":true,"duration_ms":8615,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["60H30","58J35","62G07"],"pacs":[],"model":"grok-4.5","headline":"Near edges and corners, Gaussian smoothing of a singular measure is governed by the Gaussian mass of the inward tangent cone, which sets the leading score and Hessian singularities.","keywords":["Gaussian smoothing","singular measures","manifolds with corners","tangent cones","score functions","heat regularization","boundary-layer asymptotics"],"falsifier":"On a concrete example with a known corner (e.g., uniform measure on a square or simplex), compute the exact Gaussian convolution numerically for small sigma and check whether the leading coefficient matches the predicted Gaussian mass of the inward cone and whether the score scales as sigma^{-1} times the predicted normal attraction.","tokens_in":60312,"feed_emoji":"△","tokens_out":688,"duration_ms":6740,"temperature":0.7,"pith_summary":"When a probability measure lives on a lower-dimensional set that may have boundary faces or corners, its Gaussian smoothing at small noise level sigma develops sharp layers of thickness sigma. This paper zooms into those layers and proves a two-term expansion for the smoothed density: after rescaling, the support looks like its inward tangent cone, and the leading coefficient is simply the Gaussian mass of that cone, weighted by the density and the local volume Jacobian. Differentiating the expansion then gives uniform formulas for the score, the log-Hessian, and the scale derivative of the score, with explicit negative powers of sigma whose coefficients are intrinsic geometric quantities. A localization argument shows that mass far from the observation point contributes only an exponentially small remainder. The resulting formulas make precise how ambient codimension, boundary constraints, corners, and curvature are encoded in the singular differential structure of the regularized density—objects that appear as population targets in denoising and generative modeling.","feed_headline":"Gaussian smoothing near corners is set by the tangent cone","feed_subtitle":"Two-term expansions give the singular score and Hessian that encode edges, corners, and curvature","key_machinery":"The linearized-cone coefficient C_0(a,x): after rescaling the observation point to a=sigma^{-1} times transverse displacement, C_0 is the integral of the Gaussian e^{-Psi} over the local quadrant H_m^c, multiplied by density times |det L(x)|. It carries the leading geometry; its logarithmic derivatives produce the singular score and Hessian coefficients.","core_discovery":"In the O(sigma) conical layer around a codimension-c stratum of a manifold with corners, the heat-regularized density admits the two-term expansion p_sigma(y)=sigma^{-k}(2pi)^{-d/2}[C_0(a,x)+sigma C_1(a,x)+O(sigma^2)], where C_0 is the Gaussian mass of the linearized inward tangent cone weighted by density and adapted Jacobian; the same expansion yields logarithmic asymptotics and uniform expansions for the score (order sigma^{-1}), log-Hessian (order sigma^{-2}), and scale derivative of the score.","pith_inferences":[],"forward_implications":[],"fun_headline_variants":["Conical layer sets two-term heat density near corners","Tangent-cone mass drives singular score and Hessian","Boundary-layer Gaussians encode edges corners curvature","O(sigma) cone expansion yields score log-Hessian asymptotics","Heat densities near strata track linearized inward cones"],"cache_read_input_tokens":49280,"weakest_assumption_plain":"The support must look locally like a manifold with corners covered by a single smooth corner chart, the density must be smooth enough and strictly positive on the compact piece of the stratum being studied; without that positivity the passage from density to log and score expansions fails.","fun_headline_variants_meta":{"raw":{"variants":["Conical layer sets two-term heat density near corners","Tangent-cone mass drives singular score and Hessian","Boundary-layer Gaussians encode edges corners curvature","O(sigma) cone expansion yields score log-Hessian asymptotics","Heat densities near strata track linearized inward cones"]},"model":"grok-4.5","effort":"low","cost_usd":0.003334,"raw_usage":{"total_tokens":1117,"prompt_tokens":794,"num_sources_used":0,"completion_tokens":79,"cost_in_usd_ticks":33340000,"prompt_tokens_details":{"text_tokens":794,"audio_tokens":0,"image_tokens":0,"cached_tokens":128},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":244,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":794,"tokens_out":79,"duration_ms":3286,"temperature":1.0,"reasoning_tokens":244,"cache_read_input_tokens":128,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-07-11T18:13:14.638173+00:00","model_set":{"reader":"grok-4.5"},"falsifier":"On a concrete example with a known corner (e.g., uniform measure on a square or simplex), compute the exact Gaussian convolution numerically for small sigma and check whether the leading coefficient matches the predicted Gaussian mass of the inward cone and whether the score scales as sigma^{-1} times the predicted normal attraction.","supporting_citations":[],"review_version":1}