{"id":"37a518b1-aa61-4d41-9c5b-7b32f185fac6","arxiv_id":"2504.15488","paper_version":1,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For R-ball convex bodies, the volume lost by the R-ball floating body at scale δ behaves like δ^{2/(n+1)} times an integral of (κ_i - 1/R)^{1/(n+1)} over the boundary, defining a rigid-motion invariant valuation.","lead":"This paper defines a floating body for ball-convex shapes, where translated balls, rather than half-spaces, cut off small volume pieces. It proves that the volume loss has a sharp limit that defines a new boundary quantity, the relative affine surface area, with natural geometric properties.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Lemma 8's lower-bound inequality (32) is underived and dimensionally suspicious; if it fails, Theorem 1 lacks a lower bound.","rationale":"The reader located the weak point in the local ellipsoidal approximation imported from [46], which is a fair concern. My stress-test goes one step further into the proof of Lemma 8: the cap-volume comparison cannot be completed without the distance comparison (32), and that inequality is the least documented part of the argument. It is derived through an opaque minimization with unstated regime assumptions, and as typeset the formula does not have the correct length scaling. Because Lemma 8 is the pointwise engine of Theorem 1, this is load-bearing: if (32) is wrong, the lower bound on δ in terms of the integral over S^{n−2} fails, and the limit theorem is not established. I do not claim the theorem is false; the construction is standard and the authors are expert, so the most likely resolution is a correctable typo or omitted computation. But the proof as written is not checkable at this step, and the claim is strong enough that a complete derivation should be required. Hence I move from ACCEPT to CONDITIONAL rather than REJECT: the mathematics is probably right, but the paper should supply a rigorous proof of (32) or replace it with a cited lemma before acceptance.","tokens_in":2,"tokens_out":29920,"duration_ms":833751,"concrete_test":"In the n=2 case, set E(ε−) to the ellipse x1^2/((1−ε)a1)^2 + (x2−a2)^2/a2^2 = 1, let xδ = t(sinθ, cosθ) with t small, and compute the exact Euclidean distance d0(t) to the ellipse. Compare the first two terms of its Taylor expansion with the claimed bound (32). If the expansion is d0 = t cosθ − a2 t^2 sin^2θ / (2(1−ε)^2 a1^2) + O(t^3), then the correct dimensionless correction in (32) after factoring out t cosθ must be a2 t sin^2θ / ((1−ε)^2 a1^2), not the printed expression; this would confirm a typographical error and the intended estimate. If the expansion differs by a power of t, Lemma 8's lower bound is false in dimension 2 and Theorem 1 is unproved.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"Theorem 1 rests on the pointwise limit in Lemma 8, and the lower half of that lemma requires the two-sided distance estimate (32), which compares d0 = inf_{y in ∂E(ε−)} ||xδ − y|| with the normal height t cosθ = ⟨x/||x||, N_K(x)⟩ ||xδ − x||. The proof of (32) is the least secure step in the paper: it is a long 2D distance minimization containing several unstated truncations (e.g., 'we may assume ξ(1)^2 ≤ 2((1−ε)a1)^2' and 'we may also assume a_n ≥ xδ(2)'), and the displayed formula in (32) is not homogeneous in length as typeset, so it cannot be checked as written. If the correction term has the wrong power of t or a missing factor, the subsequent lower bound for vol(E(ε−) \\ B^n(a,R)) has the wrong leading term, Lemma 8 collapses, and the proof of Theorem 1 gives only an upper bound for the limit. This is an internal gap rather than merely an imported premise: no external reference is cited for (32), and the surrounding assertion that the ball centered at (R+d0)e_n 'cuts off strictly less than δ' is also not fully justified from xδ being interior to that ball. The reader's concern about the imported ellipsoidal approximation (30) is real, but even granting (30), the proof still needs a clean derivation of (32).","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper introduces a floating-body construction for ball-convex bodies. For an R-ball convex body K, the R-ball floating body K_R^δ is defined as the intersection of all R-balls that cut off from K a set of volume at most δ. The main result (Theorem 1) asserts that, if all principal curvatures of K are strictly larger than 1/R, then the right derivative of the volume difference at δ=0 has the explicit limit c_n ∫_{∂K} ∏_{i=1}^{n-1} (κ_i(K,x) − 1/R)^{1/(n+1)} dμ_K(x). The integral, called the relative affine surface area as_R(K), recovers the classical affine surface area as R→∞. The paper proves that as_R is invariant under rigid motions, homogeneous, a valuation, and upper semicontinuous, and it establishes an affine isoperimetric-type inequality. The proofs combine a cap-volume estimate for ellipsoidal caps cut by R-balls (Lemma 7), a pointwise limit for the radial displacement x−xδ (Lemma 8), and an integration argument using a bound from the rolling function (Lemma 6).","tokens_in":17944,"tokens_out":18058,"duration_ms":154794,"significance":"If Theorem 1 is correct, the paper provides a natural extension of affine surface area to ball-convex bodies, with a geometric definition via floating bodies and the expected R→∞ limit. The proof structure is largely self-contained: Proposition 2 is proved from first principles, Lemma 7 supplies two-sided cap estimates with explicit constants, and Lemma 6 gives the integrable control needed for dominated convergence. The additional valuation and semicontinuity properties make the new functional a promising tool for approximation theory of ball-convex bodies, where only planar versions were previously available. The main weakness is that the pointwise lower bound in Lemma 8, and the uniformity of the ellipsoidal approximation used there, are not fully established; these gaps are localized and appear fixable, but they are load-bearing for the central limit formula.","major_comments":[{"comment":"The lower-bound part of Lemma 8, which is the only source of the lower bound in Theorem 1, is not rigorously established. The two-sided estimate (32) as displayed is not dimensionally homogeneous, so it cannot be checked as written. The derivation of the left-hand side contains an invalid step: after obtaining an inequality with the extra term 2(xδ(2)−a_n)a_n ξ(1)^3/((1−ε)a1)^4, the proof drops this term and concludes the inequality without it, even though the assumption a_n ≥ xδ(2) makes the dropped term non-positive, which weakens the right-hand side rather than preserving the inequality. In addition, the paragraph after (32) asserts that the R-ball centered at (R+d0)e_n 'cuts off strictly less than δ from K as xδ is in the interior of this R-ball' without a proof; this is not a consequence of interiority alone and needs a quantitative argument relating vol(K\\B) to δ. Please replace (32) and the subsequent paragraph by a complete, dimensionally consistent derivation.","section":"§3.2, Lemma 8, inequality (32)"},{"comment":"The ellipsoidal approximation (30) is quoted from [46] for a fixed boundary point x, but Lemma 8 requires a uniform version along the sequence xδ→x: the proof asserts that for all sufficiently small δ and all support R-balls at xδ, the set E(ε−)\\(z+RB^n) is contained in the fixed neighborhood H^-(x−Δε e_n,e_n)∩E(ε−). This uniformity is not proved and does not follow from the pointwise statement (30), because the support ball may vary with δ. Since the cap-volume comparison of Lemma 7 is transferred from K to E(ε±) through this inclusion, the lower bound in Lemma 8 depends on this missing uniformity. Please provide a proof or a precise citation for the uniform version.","section":"§3.2, equations (30)–(31)"}],"minor_comments":[{"comment":"The sentence 'Such an R-ball exists by Theorem 5 (i)' should refer to Lemma 5; the paper contains no Theorem 5.","section":"§3.2, proof of Lemma 6"},{"comment":"Equation (20) contains a corrupted expression: 'B^n_2(x−r_K(x)N_K(x), r_K(x)−K(x))' should presumably be 'B^n_2(x−r_K(x)N_K(x), r_K(x))'.","section":"§3.2, proof of Lemma 6"},{"comment":"The statement of Proposition 2 is for integrals over S^{n−1}, while the proof in §3.1 treats the S^{n−2} case used later; please make the reduction explicit.","section":"§3.1, Proposition 2"},{"comment":"The application of [32] is terse; please spell out the hypotheses of the cited semicontinuity theorem and verify them for the integrand f(∏(κ_i−1/R)) on the class of R-ball convex bodies.","section":"§3.3, Proposition 3(iv)"},{"comment":"In the definition of as_L(K), the term κ_i(L, N_L^{-1}(N_K(x))) is undefined when N_K(x) is not a regular value of the Gauss map of L; please add a convention for such boundary points.","section":"§2, Definition 2"},{"comment":"There are several typos: 'surface surface area' in the Introduction, the duplicated email address in the author affiliation, and the inconsistent spelling 'covarigram' for 'covariogram' in §2.","section":"Introduction and affiliations"}],"recommendation":"major_revision","confidential_remarks":"The paper is from a group with a strong track record in this area, and the main theorem is plausible. The blocking issue is isolated to the lower-bound argument in Lemma 8; if the authors can supply a complete derivation of (32) and justify the uniformity of the ellipsoidal approximation, the paper should be acceptable. The rest of the manuscript (Proposition 2, Lemma 7, the valuation and semicontinuity arguments) is in reasonable shape."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Main takeaway: this is a competent, honest extension of the classical floating body template, and the main limit theorem is probably correct. The one genuinely shaky spot is the lower-bound estimate in Lemma 8, which should be rewritten before I'd trust it without checking every line.\n\nWhat is actually new: the R-ball floating body and the limit formula for the volume difference in all dimensions. In dimension 2, related quantities appeared in [22,24], but the general n-dimensional statement, the definition of relative affine surface area, and the valuation/semicontinuity properties are new. The proof follows the Schuett-Werner template faithfully: cap volume estimates for ellipsoids cut by R-balls (Lemma 7), a pointwise limit (Lemma 8), and integration using the rolling function bound. Proposition 2, the spherical integral, is a useful standalone fact. The paper is honest about the 2D precedent.\n\nThe soft spots are in proportion. The biggest is (32), the lower-distance inequality in Lemma 8. As typeset, the displayed formula is dimensionally suspect, and the derivation relies on two 'we may assume' truncations without full justification. This is load-bearing: without a lower bound on the distance from x_delta to the inner approximating ellipsoid, the lower bound in Theorem 1 fails. I don't think the argument is wrong--the geometry of a tangent paraboloid makes the needed (1 - o(1)) t cos(theta) bound very plausible--but the paper should present a clean, checkable proof of (32). A referee should ask for that. The imported ellipsoidal approximation (30) from [46] is standard, though the uniformity over the boundary is not spelled out. The semicontinuity proof via [32] is terse; I'd like a sentence confirming the hypotheses. Minor stuff.\n\nWho gets value: convex geometers working on floating bodies, affine surface area, or ball-convex bodies, especially anyone interested in approximation theory in this class. The result does not change practice outside the subfield, but it fills a natural gap.\n\nRecommendation: send it to peer review. It deserves a serious referee; with a cleaned-up Lemma 8 it would be a solid paper.","headline":"Solid extension of classical floating body theory to ball-convex bodies; the main theorem is likely correct, but the lower-bound proof in Lemma 8 needs a clean rewrite before I'd fully trust it.","tokens_in":18472,"tokens_out":16985,"would_cite":true,"duration_ms":144317,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["52A20","52A38","53A15"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves that the volume lost by cutting an R-ball-convex body with R-balls has a universal δ^{2/(n+1)} limit given by an integral of (curvature − 1/R)^{1/(n+1)}, defining a relative affine surface area.","keywords":["ball-convex bodies","floating body","affine surface area","relative affine surface area","valuation","upper semicontinuity","convex geometry","curvature"],"falsifier":"For a planar ellipse that is R-ball convex with both principal curvatures strictly greater than $1/R$, compute the exact area of $K_R^\\delta$ by direct integration for small $\\delta$ and compare the coefficient of $\\delta^{2/3}$ with $c_2\\int_{\\partial K}(\\kappa-1/R)^{1/3}\\,ds$, where $c_2=\\frac12(3/2)^{2/3}$; any disagreement in the leading coefficient would falsify Theorem 1.","tokens_in":17459,"feed_emoji":"📐","tokens_out":5279,"duration_ms":46422,"temperature":0.7,"pith_summary":"This paper extends the classical floating-body construction from convex bodies to ball-convex bodies: bodies that can be written as intersections of congruent R-balls. It defines the R-ball floating body by cutting off caps with R-balls, and proves a sharp asymptotic formula for the volume decrease as the cap volume goes to zero. The coefficient is an integral over the boundary of the product of $(\\kappa_i - 1/R)^{1/(n+1)}$, which the authors name the relative affine surface area. A sympathetic reader should care because this gives ball-convex geometry its own affine surface area with the same formal properties as the classical one: rigid-motion invariance, valuation, homogeneity, and upper semicontinuity, with the classical affine surface area recovered as $R\\to\\infty$.","feed_headline":"Cutting with R-balls defines a relative affine surface area","feed_subtitle":"Volume loss under ball cuts has a sharp δ^{2/3} limit; the classical affine surface area returns as R grows.","key_machinery":"The central object is the R-ball floating body $K_R^\\delta = \\bigcap_{\\operatorname{vol}_n(K\\setminus (z+RB_2^n))\\le \\delta} (z+RB_2^n)$, the intersection of all R-balls that cut off at most $\\delta$ of K's volume. The proof machinery combines three ingredients: Lemma 7 compares the volume of the cap removed from K with caps of the approximating ellipsoids $E(\\varepsilon^-)$ and $E(\\varepsilon^+)$ that squeeze $\\partial K$ near a boundary point (equation (30), imported from [46]); Lemma 8 converts this cap-volume comparison into the pointwise limit of $\\|x-x_\\delta\\|/\\delta^{2/(n+1)}$; and Proposition 2 evaluates the resulting spherical integral as a product of square roots, yielding the explicit constant $c_n$. Lemma 6, based on McMullen's rolling function $r_K(x)$, controls the convergence uniformly over the boundary so that integration and limit can be interchanged.","core_discovery":"The central discovery is Theorem 1: for every R-ball convex body K whose principal curvatures are all strictly larger than $1/R$, the volume difference between K and its R-ball floating body satisfies $$\\lim_{\\delta\\to 0} \\frac{\\operatorname{vol}_n(K)-\\operatorname{vol}_n(K_R^\\delta)}{\\$delta^{{2/(n+1)}}$} = c_n \\int_{\\partial K} \\prod_{i=1}^{n-1} \\left(\\kappa_i(K,x)-\\frac{1}{R}\\right)^{1/(n+1)} d\\mu_K(x),$$ with the explicit constant $c_n = \\frac{1}{2}\\left(\\frac{n+1}{\\operatorname{vol}_{n-1}(B_2^{n-1})}\\right)^{2/(n+1)}$. This justifies calling the integral the relative affine surface area $\\operatorname{as}_R(K)$, and it recovers Blaschke's affine surface area when $R\\to\\infty$.","pith_inferences":["A natural extension the authors flag is an $L_p$-version of $\\operatorname{as}_R$; if the derivative formula holds with a $p$-power weight, it would yield a family of relative affine invariants on ball-convex bodies.","The constant $c_n$ being exactly the classical floating-body constant suggests the relative formula is the leading term of an expansion in $1/R$, possibly connecting to spherical or hyperbolic floating bodies when the ball radius is allowed to become imaginary.","One could probe stability by computing the second-order term in $\\delta$ for small $n$; the theorem fixes only the leading order, and the next coefficient would distinguish genuinely different relative affine structures.","The strict curvature assumption $\\kappa_i>1/R$ suggests that the boundary case $\\kappa_i=1/R$, where the integrand vanishes, may produce a different power of $\\delta$; testing this on bodies with flat arcs made of R-ball pieces would clarify the boundary behavior of the relative affine surface area."],"forward_implications":["The formula defines a rigid-motion invariant and upper-semicontinuous valuation on R-ball-convex bodies, giving ball-convex geometry a natural affine-analytic invariant alongside its metric ones.","Taking $R\\to\\infty$ recovers the classical affine surface area, so the new notion is a genuine one-parameter relative version rather than a separate construction.","The inequality $\\operatorname{as}_R(K) \\le n\\operatorname{vol}_n(B_2^n)^{2/(n+1)} \\operatorname{vol}_n(K)^{(n-1)/(n+1)}$, with equality only for $R=\\infty$ and ellipsoids, provides a relative affine isoperimetric bound.","R-ball polyhedra, intersections of finitely many R-balls, have $\\operatorname{as}_R=0$, which is consistent with the upper semicontinuity and with the fact that every smooth body can be approximated by such polyhedra.","In dimension 2 the construction recovers the r-spindle floating body, linking the result to existing approximation questions for random disc polygons."],"supporting_citations":[{"why":"Supplies the classical convex floating body construction and the cap-volume derivative method that Theorem 1 adapts to R-balls.","marker":"[45]"},{"why":"Provides the local ellipsoidal approximation of a smooth convex body near a boundary point, equation (30), on which Lemma 8 relies.","marker":"[46]"},{"why":"Introduces the class of C-ball convex bodies and the intersection representation used throughout the paper.","marker":"[31]"},{"why":"Gives the classical floating body definition and the economic cap-covering viewpoint that motivates the R-ball floating body.","marker":"[7]"},{"why":"Introduces McMullen's rolling function, used in Lemma 6 to bound the displacement $\\|x-x_\\delta\\|$ uniformly over the boundary.","marker":"[36]"},{"why":"Establishes the semicontinuity of curvature integrals that the proof of upper semicontinuity in Proposition 3(iv) invokes.","marker":"[32]"},{"why":"Provides the valuation proof technique used to show that $\\operatorname{as}_R$ is a valuation on R-ball-convex bodies.","marker":"[44]"}],"fun_headline_variants":["Relative affine surface area emerges from ball-convex floating bodies","Floating bodies for ball-convex sets define new surface area","Ball-cut volume loss gives relative affine surface area","Classical affine surface area recovered as ball radius grows","Ball-convex floating bodies yield a rigid-motion invariant surface area"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The formula rests on the assumption that near each boundary point the body can be squeezed between two nearly identical ellipsoids in a neighborhood large enough to contain the caps that the cutting R-balls remove; if that ellipsoidal approximation is not uniform along the boundary, the cap-volume comparison that produces the limit breaks down.","fun_headline_variants_meta":{"raw":{"variants":["Relative affine surface area emerges from ball-convex floating bodies","Floating bodies for ball-convex sets define new surface area","Ball-cut volume loss gives relative affine surface area","Classical affine surface area recovered as ball radius grows","Ball-convex floating bodies yield a rigid-motion invariant surface area"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00069,"raw_usage":{"total_tokens":3042,"prompt_tokens":780,"completion_tokens":2262,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":396,"completion_tokens_details":{"reasoning_tokens":2180}},"tokens_in":396,"tokens_out":2262,"duration_ms":13614,"temperature":1.0,"reasoning_tokens":2180,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T11:25:41.122890+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"For a planar ellipse that is R-ball convex with both principal curvatures strictly greater than $1/R$, compute the exact area of $K_R^\\delta$ by direct integration for small $\\delta$ and compare the coefficient of $\\delta^{2/3}$ with $c_2\\int_{\\partial K}(\\kappa-1/R)^{1/3}\\,ds$, where $c_2=\\frac12(3/2)^{2/3}$; any disagreement in the leading coefficient would falsify Theorem 1.","supporting_citations":[{"cited_title":"Sch¨ utt and E","cited_arxiv_id":null,"evidence_quote":"Supplies the classical convex floating body construction and the cap-volume derivative method that Theorem 1 adapts to R-balls."},{"cited_title":"Sch¨ utt and E","cited_arxiv_id":null,"evidence_quote":"Provides the local ellipsoidal approximation of a smooth convex body near a boundary point, equation (30), on which Lemma 8 relies."},{"cited_title":"L´ angi, M","cited_arxiv_id":null,"evidence_quote":"Introduces the class of C-ball convex bodies and the intersection representation used throughout the paper."},{"cited_title":"B´ ar´ any and D.G","cited_arxiv_id":null,"evidence_quote":"Gives the classical floating body definition and the economic cap-covering viewpoint that motivates the R-ball floating body."},{"cited_title":"McMullen, On the inner parallel body of a convex body , Israel J","cited_arxiv_id":null,"evidence_quote":"Introduces McMullen's rolling function, used in Lemma 6 to bound the displacement $\\|x-x_\\delta\\|$ uniformly over the boundary."},{"cited_title":"Ludwig, On the semicontinuity of curvature integrals , Mathematische Nachrichten 227 (2001), 99–108","cited_arxiv_id":null,"evidence_quote":"Establishes the semicontinuity of curvature integrals that the proof of upper semicontinuity in Proposition 3(iv) invokes."},{"cited_title":"Sch¨ utt,On the affine surface area , Proceedings of the American Mathematical So- ciety 118(4) (1993), 1213–1218","cited_arxiv_id":null,"evidence_quote":"Provides the valuation proof technique used to show that $\\operatorname{as}_R$ is a valuation on R-ball-convex bodies."}],"review_version":1}