{"id":"29d6071f-7f04-4405-ba33-2a10e53715c8","arxiv_id":"1908.09498","paper_version":1,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"For Poisson hyperplane K-cells, the expected mean width excess is between n^{-1} log^{d-1} n and n^{-2/(d+1)}, with exact limits for smooth bodies and simplicial polytopes.","lead":"This math paper determines how quickly a random polyhedron formed by random planes that avoid a convex shape converges to the shape's mean width. It proves the convergence rate ranges and exact limit constants, extending earlier isotropic results to non-isotropic plane direction distributions.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No significant objection identified: Theorem 1's proof is internally sound; remaining risk is standard reliance on prior theorems.","rationale":"The reader's ACCEPT verdict is well supported. I examined the proof of Theorem 1 in detail, including the dual map eta, the measure comparison (17), the use of Bárány–Larman (13), and the Bronshtein net argument for the upper bound. The central argument is internally consistent; the only issues are a harmless missing factor 2 in the μ/W identity and a proof gap in Theorem 2 for degenerate directional distributions, neither of which affects Theorem 1 because its assumptions (3) and (4) exclude the degenerate case. The exact asymptotic relations rely on prior results by overlapping authors, but this is a standard citation dependency and not a reason to downgrade the verdict. I therefore recommend leaving the verdict unchanged.","tokens_in":15088,"tokens_out":47707,"duration_ms":449456,"concrete_test":"Re-derive the normalization in (15)–(16) after replacing W(Z)-W(K) with 2∫(h_Z-h_K)dσ; confirm that the factor 2 is absorbed into the constants a2,a3 and that the lower bound remains of order n^{-1} log^{d-1}n for every phi satisfying phi ≤ a0 sigma. If the order in n changes, the lower-bound proof of Theorem 1 would need revision.","verdict_should_be":"UNCHANGED","load_bearing_attack":"After a careful check of the proof of Theorem 1, I find no load-bearing gap. The lower bound is obtained through the dual wet-part argument: inequality (15) and the inclusion eta(K^o(t)) subset H^psi(const t) correctly transfer the Bárány–Larman estimate (13) to the mean-width difference. There is a missing factor 2 in the identity W(Z)-W(K)=integral 1_{H∩Z neq empty} mu(dH), since W=2∫ h dσ while mu uses dσ; however, this factor is absorbed into the unspecified constants a2,a3 in (16) and does not affect the exponent n^{-1} log^{d-1}n. The upper bound via the Bronshtein net in Theorem 3 is sound, and the conversion from the hitting functional to the mean width under phi ≥ a1 sigma in Theorem 1 is correct. The proof of Theorem 2 as written uses Lemma 1, which may fail for degenerate phi where Z is unbounded; but Theorem 1 invokes Theorem 2 only under phi ≥ a1 sigma, where the needed tail estimate is valid. The exact asymptotic Theorems 5, 6, 8, and 9 depend on the published results [4, Thm. 5.2, 5.3] and [5, Thm. 1.3], which is a normal citation dependency rather than an internal inconsistency. Overall, no load-bearing concern about the central claim was identified.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the K-cell Z_K^(n) of a stationary Poisson hyperplane process in R^d, defined as the intersection of all closed halfspaces containing a convex body K whose boundary hyperplanes do not meet K. The central object is the expected deviation of a functional evaluated at Z_K^(n) from its value at K, as the intensity n tends to infinity. For the mean width W, Theorem 1 establishes the two-sided rate n^{-1} log^{d-1} n ≪ E W(Z_K^(n)) - W(K) ≪ n^{-2/(d+1)} under the respective assumptions φ ≤ a0 σ and φ ≥ a1 σ on the directional distribution φ. For the hitting functional Φ, Theorem 2 gives a general upper bound of order n^{-2/(d+1)}, with moment versions in Theorem 4 and facet-number moment bounds in Theorem 7. Under a positive continuous density of φ, Theorems 5 and 8 provide exact asymptotic constants for the mean width difference and the expected facet number; in the isotropic simplicial-polytope case, Theorems 6 and 9 give exact logarithmic asymptotics. The proofs combine a dual wet-part argument based on Bárány-Larman, Bronshtein's entropy bound for convex bodies, and Poissonization lemmas that transfer finite-k asymptotics to Poisson intensities.","tokens_in":15390,"tokens_out":23855,"duration_ms":262703,"significance":"If correct, the paper significantly extends the isotropic results of [21] to general directional distributions, resolves the sharp order of the mean-width approximation for the K-cell model, and provides exact constants in smooth and simplicial cases. The hitting-functional bounds and the facet-number estimates are of independent interest and give a Poisson-dual counterpart to classical random-polytope results. The proofs are detailed and largely cross-checkable, and the main external dependencies are published theorems (Bárány-Larman, [4], [5]) rather than new claims. The paper does not provide machine-checked proofs or code, but the analytic arguments are transparent and the key inequalities are explicitly derived.","major_comments":[{"comment":"Lemma 1 as stated is false without a non-degeneracy assumption on φ. For example, in d=2 take φ=(δ_{e1}+δ_{-e1})/2; then every hyperplane of the process is vertical and the K-cell is almost surely an unbounded vertical strip, so Ro(Z_K^(n))=∞ and the claimed bound P(Ro(Z_K^(n))>b(Ro+x)) ≤ 2d e^{-a n x} fails for large x. The condition that φ is not concentrated on a great subsphere is used later in Section 3 but is not assumed in Section 2, where inequalities (6)-(7) and the proof of Theorem 4 rely on Lemma 1. Consequently Theorem 2 and Theorem 7 are proved only under an implicit extra assumption as they stand. Since the hypotheses (3) and (4) of Theorem 1 imply the needed non-degeneracy, the main theorem is not affected, but the statements of Theorems 2-4 and 7 should either explicitly assume φ is not concentrated on a great subsphere (or a comparable condition) or provide a separate argument that handles degenerate φ.","section":"Section 2, Lemma 1"}],"minor_comments":[{"comment":"If ν is defined as the image measure of Lebesgue measure under the map η(ru)=H(u,1/r), then the displayed formula should contain the factor d in front of ω_d, i.e., dω_d ∫ τ^{-(d+1)} dτ σ(du), because the normalized spherical measure σ is related to the surface measure by dS = dω_d σ. The missing factor is harmless for the constant comparisons c5,c6 that follow, but the displayed equation should be corrected.","section":"Section 3, Eq. (17)"},{"comment":"The identity W(Z)-W(K)=∫_{H\\H_K} 1_{H∩Z≠∅} μ(dH) deserves a one-sentence justification: the integration over the two families τ>h(K,u) and τ<-h(K,-u) produces the factor 2 that appears in W=2∫ h dσ. As written, the reader must infer this from the definition of μ in (14).","section":"Section 3, proof of Theorem 1"}],"recommendation":"major_revision","confidential_remarks":"The central Theorem 1 appears sound, and the exact asymptotic results are a substantial contribution. The main issue is the overgeneralized statement of Lemma 1 in Section 2, which is a formal error in a supporting argument and should be corrected either by adding the non-degeneracy assumption or by supplying a degenerate-case proof. The paper's reliance on [4], [5], and [21] by overlapping authors is a normal citation dependency in this research program, but it would strengthen the paper to state explicitly in the introduction which prior results are assumed. I see no evidence of circularity: the cited results are independent published theorems, not the target results."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper does what it says: it settles the mean-width approximation order for K-cells of Poisson hyperplane processes when the directional distribution is not isotropic. Theorem 1 gives the right two-sided bounds, n^{-1} log^{d-1}n below and n^{-2/(d+1)} above, and the exact asymptotics for smooth bodies and simplicial polytopes (Theorems 5, 6, 8, 9) are genuinely new. The moment bounds for facet numbers are also a useful addition. The proofs are careful and largely self-contained, and the dualization of the Bárány–Larman argument is well executed. The missing factor 2 in the mean-width identity is real but harmless; it disappears into the unspecified constants a2, a3, exactly as the stress-test note says.\n\nThe soft spot worth flagging is Theorem 2. It is stated for every directional distribution φ, but the proof leans on Lemma 1, quoted from [21], which gives an exponential tail bound for the circumradius of the K-cell. That bound cannot hold when φ is degenerate, because then the K-cell is unbounded with positive probability and the radius is infinite. So either Theorem 2 needs a non-degeneracy hypothesis, or Lemma 1 needs to be stated with the right assumptions and proved. Since Theorem 1 uses Theorem 2 only under φ ≥ a1 σ, the main results are not endangered. This is a blemish in a side theorem, not a crack in the foundation.\n\nThe citation pattern is healthy. The exact asymptotics lean on [4, Thm. 5.2/5.3] and [5, Thm. 1.3] by overlapping authors, but those are published, independent theorems; the paper is not restating its own results. The Bárány–Larman estimate is deep but standard. I see no circularity.\n\nThe intended reader is anyone working in stochastic geometry or random polytopes. The paper is worth a serious referee: it extends a known isotropic result to the non-isotropic setting, adds exact constants, and opens a clean path for the remaining intrinsic volumes. I would send it to peer review and ask for a minor revision to tighten Theorem 2's hypotheses or proof.","headline":"Solid extension of Schneider's isotropic K-cell result to non-isotropic processes; exact constants and facet bounds are new, and the only real blemish is a possibly overbroad Theorem 2.","tokens_in":15897,"tokens_out":4047,"would_cite":true,"duration_ms":42628,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["60D05","52A27"],"pacs":[],"model":"deepseek-v4-flash","headline":"Random halfspace-cut polytopes approximate convex bodies at sharp, dimension-dependent rates.","keywords":["Poisson hyperplane process","convex body approximation","K-cell","mean width","hitting functional","random polytopes","facet number","stochastic geometry"],"falsifier":"For a numerical check, take $K$ as a disk in $\\mathbb{R}^2$ and an isotropic Poisson line process, simulate the $K$-cell for increasing $n$, and verify that $n^{2/3}(E[W(Z_K^{(n)})]-W(K))$ approaches the constant from Theorem 5 computed for the circle; if it does not settle on that value, the sharp upper-order claim fails. The logarithmic lower order can be tested by repeating the simulation for a square: $(n/\\log n)(E[W(Z_K^{(n)})]-W(K))$ should approach the stated constant from Theorem 6.","tokens_in":14904,"feed_emoji":"📐","tokens_out":10896,"duration_ms":102967,"temperature":0.7,"pith_summary":"The paper studies a natural dual model of random approximation in which a fixed convex body $K$ is enclosed by the intersection of all halfspaces of a Poisson hyperplane process that contain $K$, called the $K$-cell. It establishes that the expected value of $W(Z_K^{(n)})-W(K)$ has order at least $n^{-1}\\log^{d-1}n$ when the directional distribution is bounded above by a constant times spherical measure, and order at most $n^{-2/(d+1)}$ when it is bounded below by a positive multiple of spherical measure. For smooth bodies with a positive continuous directional density the exact limit of $n^{2/(d+1)}$ times the error is computed, and for isotropic processes and simplicial polytopes the logarithmic lower order is shown to be sharp. The paper also proves moment bounds and exact asymptotic relations for the expected facet number of the $K$-cell. These rates matter because they make precise how well a natural halfspace-generated random polytope approximates a convex body, extending known volume results to mean width and facet counts.","feed_headline":"Random hyperplane cuts approximate bodies at exact rates","feed_subtitle":"New limiting results fix the mean-width error and facet counts as cut intensity grows.","key_machinery":"The central object is the $K$-cell $Z_K^{(n)}$, defined as the intersection of all closed halfspaces bounded by hyperplanes of the process that do not meet $K$. The proofs run through three mechanisms. A metric entropy estimate for the space of convex bodies gives a deviation inequality for the hitting functional $\\Phi$, from which all moment bounds of order $n^{-2k/(d+1)}$ follow. A dualization of the floating-body construction maps the wet part of the polar body $K^\\circ$ to the family of hyperplanes separating $K$ from a point, so cap-volume lower bounds for the polar body become lower bounds for the mean-width error. A Poissonization lemma transfers asymptotic relations known for $k$ independent random hyperplanes to the Poisson process, and the Slivnyak–Mecke identity $E f_{d-1}(Z_K^{(n)})=2n\\,E[\\Phi(Z_K^{(n)})-\\Phi(K)]$ turns the hitting-functional estimates into facet-number estimates.","core_discovery":"At the center is a pair of order-sharp bounds for the mean width. If the directional distribution $\\varphi$ satisfies $\\varphi\\le a_0\\sigma$, then $E[W(Z_K^{(n)})-W(K)]\\gg n^{-1}\\log^{d-1}n$; if $\\varphi\\ge a_1\\sigma$, then $E[W(Z_K^{(n)})-W(K)]\\ll n^{-2/(d+1)}$. Here $\\sigma$ is normalized spherical Lebesgue measure and $\\gg,\\ll$ mean up to constants depending only on $d,K,\\varphi$. Under stronger smoothness assumptions the paper evaluates the limits: for positive continuous density $q$, $n^{2/(d+1)}E[W(Z_K^{(n)})-W(K)]$ converges to $2^{-2/(d+1)}F(K,q)$, where $F(K,q)$ is an explicit boundary integral of the density and the Gauss–Kronecker curvature; and for an isotropic process with simplicial polytopes, $n/\\log^{d-1}n$ times the same error converges to $rd(\\log 2/(d+1))^{d-1}$. It also proves that the $k$-th moment of the facet number of the $K$-cell is at most $c(k)n^{k(d-1)/(d+1)}$ and that, for smooth $K$, $n^{-(d-1)/(d+1)}E f_{d-1}(Z_K^{(n)})$ converges to $2^{(d-1)/(d+1)}G(K,q)$.","pith_inferences":["Because the asymptotic constant $F(K,q)$ weights the boundary by $q^{-2/(d+1)}\\kappa^{d/(d+1)}$, the paper's result implies that for a fixed process the body's local curvature, not just its volume, controls the leading approximation error; this dependence could be tested by comparing bodies with the same width but different curvature.","The same entropy-net plus Poissonization route is the natural template for the open problem the paper names: exact rates for intrinsic volumes $V_i$ with $i\\ne 1,d$.","The exact identity linking expected facet count to the hitting-functional error means that in applications one can estimate approximation quality by counting facets of the $K$-cell, a quantity directly observable from the hyperplane arrangement.","For anisotropic processes, Theorem 5 predicts that the mean-width error's leading constant changes by the factor $q^{-2/(d+1)}$ integrated over $\\partial K$; tilting an otherwise isotropic process and measuring the error would provide a clean test of that prediction."],"forward_implications":["The mean-width error of the $K$-cell is bounded below by $n^{-1}\\log^{d-1}n$ whenever the directional distribution is not too concentrated, so no isotropy assumption is needed for the lower order.","For any body and any directional distribution, the expected hitting-functional error is $O(n^{-2/(d+1)})$, even in cases where the $K$-cell does not converge almost surely to $K$.","For smooth bodies with a positive continuous directional density, the limiting constant is given explicitly as a curvature- and density-weighted boundary integral, showing that the exponent $2/(d+1)$ is sharp.","For isotropic processes and simplicial polytopes, the logarithmic lower order is attained, so both bounds in the main theorem are optimal.","The expected facet number of the $K$-cell grows like $n^{(d-1)/(d+1)}$ up to constants, with a matching asymptotic constant for smooth $K$."],"supporting_citations":[{"why":"Supplies the cap-volume lower bound for the polar body that the dual mean-width lower bound relies on.","marker":"[3]"},{"why":"Gives the limit of $k^{2/(d+1)}$ times the circumscribed-polytope mean-width error used in Theorem 5.","marker":"[4]"},{"why":"Gives the isotropic simplicial-polytope limit used for the logarithmic-order mean-width relation in Theorem 6.","marker":"[5]"},{"why":"Provides the metric entropy bound for $\\varepsilon$-nets of convex bodies used in the deviation estimate.","marker":"[6]"},{"why":"Supplies the random-polytope deviation approach that the paper dualizes into a Poisson version.","marker":"[7]"},{"why":"Establishes the volume approximation order for the same model, the result being extended here to mean width.","marker":"[12]"},{"why":"Supplies the Poissonization lemma that transfers finite-$k$ asymptotic relations to Poisson expectations.","marker":"[15]"},{"why":"Proves the isotropic case of the mean-width bounds and contains the deviation lemma reused here.","marker":"[21]"},{"why":"Provides the intensity-measure representation and Slivnyak–Mecke formula behind the facet-number identity.","marker":"[22]"}],"fun_headline_variants":["Hyperplane cuts yield sharp convergence rates","Mean-width error decays at exact orders","Poisson hyperplanes shrink error predictably","Explicit limits fix random polytope error","Facet counts follow tight growth bounds"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The exact constants in the asymptotic limits rest on prior limiting results for the mean width of polytopes circumscribed around $K$ by $k$ independent random tangent hyperplanes; if either prior result is incorrect, the corresponding constants in this paper would not follow, and the lower bound also relies on a cap-volume lower bound for the polar body.","fun_headline_variants_meta":{"raw":{"variants":["Hyperplane cuts yield sharp convergence rates","Mean-width error decays at exact orders","Poisson hyperplanes shrink error predictably","Explicit limits fix random polytope error","Facet counts follow tight growth bounds"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000251,"raw_usage":{"total_tokens":1579,"prompt_tokens":992,"completion_tokens":587,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":608,"completion_tokens_details":{"reasoning_tokens":523}},"tokens_in":608,"tokens_out":587,"duration_ms":6226,"temperature":1.0,"reasoning_tokens":523,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:09:16.160647+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"For a numerical check, take $K$ as a disk in $\\mathbb{R}^2$ and an isotropic Poisson line process, simulate the $K$-cell for increasing $n$, and verify that $n^{2/3}(E[W(Z_K^{(n)})]-W(K))$ approaches the constant from Theorem 5 computed for the circle; if it does not settle on that value, the sharp upper-order claim fails. The logarithmic lower order can be tested by repeating the simulation for a square: $(n/\\log n)(E[W(Z_K^{(n)})]-W(K))$ should approach the stated constant from Theorem 6.","supporting_citations":[{"cited_title":"Mathematika 35 (1988), 274–291","cited_arxiv_id":null,"evidence_quote":"Supplies the cap-volume lower bound for the polar body that the dual mean-width lower bound relies on."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the limit of $k^{2/(d+1)}$ times the circumscribed-polytope mean-width error used in Theorem 5."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Gives the isotropic simplicial-polytope limit used for the logarithmic-order mean-width relation in Theorem 6."},{"cited_title":"Siberian Math","cited_arxiv_id":null,"evidence_quote":"Provides the metric entropy bound for $\\varepsilon$-nets of convex bodies used in the deviation estimate."},{"cited_title":"arXiv:1704","cited_arxiv_id":null,"evidence_quote":"Supplies the random-polytope deviation approach that the paper dualizes into a Poisson version."},{"cited_title":"Doctoral The- sis, Albert-Ludwigs-Universit¨ at, Freiburg i","cited_arxiv_id":null,"evidence_quote":"Establishes the volume approximation order for the same model, the result being extended here to mean width."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the Poissonization lemma that transfers finite-$k$ asymptotic relations to Poisson expectations."},{"cited_title":"Interaction of Poisson hyperplane processes and convex bodies","cited_arxiv_id":"1812.08443","evidence_quote":"Proves the isotropic case of the mean-width bounds and contains the deviation lemma reused here."},{"cited_title":"Springer, Berlin, 2008","cited_arxiv_id":null,"evidence_quote":"Provides the intensity-measure representation and Slivnyak–Mecke formula behind the facet-number identity."}],"review_version":1}