{"id":"0f369611-b532-4713-9c2a-b982522c624f","arxiv_id":"2411.16584","paper_version":3,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Area-weighted vertex quadrature on a triangulation of a polygon yields Marcinkiewicz-Zygmund inequalities for polynomials of sufficiently low degree relative to the mesh size.","lead":"This paper builds quadrature rules that use arbitrary scattered points on a polygon as integration points, and proves two-sided bounds known as Marcinkiewicz-Zygmund inequalities for them. These bounds let numerical methods such as hyperinterpolation work without requiring the quadrature rule to be exact for the polynomials they use.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 3.6 rests on Lemma 3.1's degree-independent Markov constant, but the proof uses the degree-dependent constant K0(d) from Lemma 2.3 without proving it is bounded uniformly in d.","rationale":"The reader's verdict is CONDITIONAL, and I agree: the central issue is the unproved degree-uniformity of the Markov constant in Lemma 3.1, which Theorem 3.6 uses to make the MZ constants independent of N. The proof as written absorbs K0(d) from Lemma 2.3 into C5 and declares C5 boundary-only; this is not justified because K0(d) is allowed to depend on the degree d. The same pattern appears in Lemma 3.2, where C7 is claimed to depend only on the boundary but the proof introduces γ_△ = |△|/ρ_Tmin, which depends on the triangulation and can be unbounded for badly shaped triangles; this reinforces the concern that boundary-only constants are not established for arbitrary scattered points. I did not find a definitive counterexample to Theorem 3.6 itself, and the construction is plausible enough that a revised proof (using classical degree-independent Markov inequalities on the fixed triangulation, or adding a shape-regularity assumption) could repair the argument. Therefore the CONDITIONAL verdict remains appropriate; no change is needed. The p = ∞ branch also contains an unjustified step (the bound by C4|△||χ|_∞,2,T2 after Taylor expansion), but that is a more localized flaw and is secondary to the Markov-constant issue.","tokens_in":56,"tokens_out":25469,"duration_ms":346408,"concrete_test":"Compute K0(d) from Lemma 2.1 for a reference triangle by forming the Bernstein interpolation matrix B from §3.1 (rows/columns indexed by domain points ξ_{ijk} and Bernstein polynomials B^T_{lmn}, d = 1,...,30) and evaluating ||B^{-1}||∞. If ||B^{-1}||∞ grows with d, the proof of Lemma 3.1 cannot yield a boundary-only constant independent of d, confirming the gap. A complementary analytic check: reprove Lemma 3.1 directly from the standard Markov inequality on the fixed coarsest triangulation △0 (without Lemma 2.3) and verify the resulting constant is independent of d; if this succeeds, the theorem is salvageable but the manuscript must be revised.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim of Theorem 3.6 is that the MZ inequalities hold with constants depending only on p and the boundary of Omega, under the condition N^2|△|^{2/p} < 1/c1 with c1 boundary-only. The critical bridge is Lemma 3.1, which is invoked to bound |χ|_{p,2,Omega} by C5 N^2 ||χ||_{p,Omega} with C5 independent of N. The proof of Lemma 3.1 applies Lemma 2.3 on the coarsest triangulation △0. However, Lemma 2.3's constant is K2 = K0(2d)^{α+β}, where K0 = K0(d) is the Bernstein norm-equivalence constant from Lemma 2.1. The proof absorbs K0(d) into C5 and asserts C5 depends only on the boundary of Omega, but no argument shows K0(d) is bounded uniformly in d. For the Bernstein basis, the conditioning of the interpolation matrix B in §3.1 typically grows with d, so K0(d) is not known to be uniformly bounded; indeed it generally is not. Without a degree-uniform bound, the constant c1 in Theorem 3.6 may depend on N, and the stated condition 'if N^2|△|^{2/p} < 1/c1' with c1 independent of N is unsupported. This directly affects the p < ∞ branch and also the p = ∞ branch, which uses Lemma 3.1 to bound Sobolev seminorms. The proof of Theorem 3.6 therefore has a load-bearing gap at the level of its main hypothesis.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes quadrature rules for triangles and polygons built from Bernstein–Bézier polynomials and scattered points. On a triangle, the rule uses domain points with weights defined through column sums of the inverse Bernstein–Bézier interpolation matrix; the paper proves exactness for polynomials of degree d and derives Marcinkiewicz–Zygmund estimates for d = 1, 3, 5, plus error estimates for d ≥ 1. On a polygon, the scattered points are used as vertices of a triangulation, the quadrature weights are one third of the sum of adjacent triangle areas, and the paper claims exactness for linear polynomials, an O(|△|^2) error bound, and Marcinkiewicz–Zygmund inequalities for 1 ≤ p ≤ ∞. Numerical experiments on a triangle and a Georgia-shaped polygon illustrate quadrature accuracy.","tokens_in":16648,"tokens_out":38517,"duration_ms":372872,"significance":"If the main theorem were established, the paper would provide an explicit, positive-weight quadrature rule for scattered data on polygons with Marcinkiewicz–Zygmund inequalities, a practically useful extension of the sphere and manifold results. The triangle part is genuinely explicit: the weights are given in closed form, the quadrature rule is exact for P_d, and the constants are not fitted. The numerical experiments support the quadrature error claims. However, the central polygon result is not established: the proof relies on a Markov inequality with a degree-uniform constant that is false as stated, and on geometric bounds whose constants are not controlled by the boundary alone. The paper therefore needs substantial revision before its main claim can be accepted.","major_comments":[{"comment":"Lemma 3.1 is false as stated, and the proof does not provide the degree-uniform constant it claims. Let [a,b] be the projection of Ω on the x-axis and set P_d(x,y) = T_d(2(x-a)/(b-a)-1). Then P_d ∈ P_d and ||P_d||_{∞,Ω} ≤ 1, while ||∂_x P_d||_{∞,Ω} = 2 d^2/(b-a). Lemma 3.1 with α+β=1 would require a boundary-only C5 satisfying C5 d ≥ c d^2 for all d, which is impossible. The proof of Lemma 3.1 absorbs the degree-dependent constant K0(d) from Lemma 2.1 into C5, but K0(d) is not shown to be uniformly bounded and in the Bernstein basis it grows with d. This is load-bearing: equations (3.18)–(3.19) and the p=∞ branch of Theorem 3.6 use Lemma 3.1 to replace |χ|_{p,2,Ω} by C5 N^2 ||χ||_{p,Ω}, so the stated condition N^2|△|^{2/p} < 1/c1 with c1 boundary-only is unsupported. A correct Markov estimate changes the admissible N-|△| relation in Theorem 3.6.","section":"§3.2, Lemma 3.1 (eq. (3.12))"},{"comment":"The constant C7 in Lemma 3.2 is not determined by the boundary of Ω. The proof defines γ△ = |△|/ρ_{Tmin}; this ratio depends on the triangulation and can be arbitrarily large for a fixed polygon, for example in triangulations containing very thin triangles with longest edge of order one and arbitrarily small inradius. The sentence in the proof asserting that γ△ depends on the boundary of Ω is therefore incorrect, and the inequality |#△|^{1-1/p}|△|^2 ≤ C7|△|^{2/p} is not valid uniformly over all triangulations of a fixed polygon. This affects the error bound in Theorem 3.5 and the constants c1,c2 in Theorem 3.6. Relatedly, Lemma 2.2 states that for 1 ≤ p < ∞ the constant K1 depends on κ_T, but in the proof of Theorem 3.6 K1 is declared to depend only on p 'since d=1'; without a shape-regularity assumption on the triangulation this is not justified.","section":"§3.2, Lemma 3.2 and Theorem 3.6 constants"},{"comment":"The step bounding the vertex value near an extremum is not justified. The proof writes |χ(x*,y*) - χ(x_J,y_J)| ≤ 2|T2||χ|_{∞,1,T2} and then passes from ‖χ‖_{∞,Ω} - K1|T2|^2|χ|_{∞,2,T2} - 2|T2||χ|_{∞,1,T2} to ‖χ‖_{∞,Ω} - C4|△||χ|_{∞,2,T2}. This implicitly requires a bound of the first-order seminorm |χ|_{∞,1,T2} by the second-order seminorm |χ|_{∞,2,T2}, which is false for polynomials: for example, a nonconstant linear polynomial has zero second seminorm but nonzero first seminorm. The argument should instead use a first-order Markov estimate on the polygon; as written, the p=∞ branch contains a genuine gap.","section":"§3.2, proof of Theorem 3.6(b), p=∞ branch"}],"minor_comments":[{"comment":"The proof states '|#△||△|^2 ≤ C7', but the estimate available from Lemma 3.2 is |#△|^{1-1/p}|△|^2 ≤ C7|△|^{2/p} (equivalently |#△|^{1/q}|△|^2 ≤ C7|△|^{2/p}); the displayed line should be corrected.","section":"§3.2, Theorem 3.5 proof"},{"comment":"The citation '[16, Theorems 2.32]' should be '[16, Theorem 2.32]'.","section":"§2, Lemma 2.3"},{"comment":"There are several typos: 'Combing both inequalities' should be 'Combining both inequalities'; 'inscribed circle in T' should be 'inscribed in T'; 'out arguments' in the final remark should be 'our arguments'; and 'BT(ξ010))' has an extra parenthesis.","section":"Throughout"},{"comment":"The numerical experiments test quadrature exactness and integration error only; they do not directly test the Marcinkiewicz–Zygmund inequalities asserted in Theorem 3.6, so the claim that they 'validate our construction' is stronger than what is demonstrated.","section":"§4"},{"comment":"The p=∞ proof contains a mismatched delimiter, '|χ‖_{∞,T}', which should be '‖χ‖_{∞,T}'.","section":"§3.3, Theorem 3.3 proof"}],"recommendation":"major_revision","confidential_remarks":"The construction is explicit and the triangle-based results appear salvageable, but the polygon theorem rests on two unsupported geometric/approximation claims: the degree-uniform Markov inequality in Lemma 3.1 and the boundary-only constant in Lemma 3.2. Both are load-bearing for the main Marcinkiewicz–Zygmund claim. A revision should either add explicit shape-regularity assumptions and correct Markov constants, or substantially modify the statements and proofs."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: the construction is clean and the triangle part is solid, but the main theorem on polygons is not proved as written. The stress-test's central criticism is correct: Lemma 3.1's proof absorbs a degree-dependent constant into a boundary-only constant, and the p=∞ branch has an additional invalid estimate.\n\nWhat's new: the paper takes the standard vertex-area quadrature rule (3.10)-(3.11) and claims it satisfies Marcinkiewicz–Zygmund inequalities on polygons for scattered points. The idea of using the coarsest triangulation to get a Markov inequality on the polygon is good; it sidesteps the shape-parameter blow-up that blocks the single-triangle case. The exactness for P1, the error bound in Theorem 3.5, and the triangular MZ estimates for d=1,3,5 are all sound. The numerical experiments match the analysis.\n\nWhere it falls apart: Lemma 3.1 states that there is a constant C5 depending only on the boundary with |χ|_{p,2,Ω} ≤ C5 N^2 ||χ||_{p,Ω}. The proof applies Lemma 2.3 on the coarse triangulation. That lemma's constant is K0(2d)^{α+β}, with K0 depending on d. For the Bernstein basis, K0 is not uniform in d: coefficients c_k = (-1)^k C(d,k) give the polynomial (1-2t)^d, sup norm 1, but max |c_k| ~ 2^d/√d. So C5 cannot be boundary-only in general. Without a uniform K0, c1 in Theorem 3.6 depends on N, and the hypothesis N^2|Δ|^{2/p}<1/c1 is not the stated boundary-only condition. This affects both p<∞ and p=∞ branches.\n\nThe p=∞ branch has a separate gap: it bounds the first-order seminorm |χ|_{∞,1} by |Δ||χ|_{∞,2}, which is false (a linear polynomial is a counterexample). A first-order Markov inequality on the polygon would repair this, but that's not what is written.\n\nEven after repair, the theorem is weaker than the classical MZ form: the constants grow with the number of triangles, and on quasi-uniform meshes the degree is restricted to N ~ m^{1/(2p)} for p<∞ and N ~ m^{1/4} for p=∞. For p>1 that is well below the usual N ~ m^{1/2}, so the MZ constants are not uniform in the number of points. The paper should state this limitation explicitly.\n\nBottom line: the construction is worth knowing and the triangle part is a solid contribution, but the polygon theorem needs a real rewrite. The paper deserves refereeing, with instructions to focus on Lemma 3.1 and the p=∞ case.","headline":"A clean quadrature construction with a solid triangle part, but the polygon MZ theorem rests on a Markov inequality whose proof absorbs a degree-dependent constant, and the p=∞ branch has an invalid step.","tokens_in":100,"tokens_out":16921,"would_cite":true,"duration_ms":210698,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["41A17","41A05","65D32","42C15"],"pacs":[],"model":"deepseek-v4-flash","headline":"A quadrature rule built from scattered points on a polygon satisfies Marcinkiewicz–Zygmund inequalities for all 1≤p≤∞.","keywords":["Marcinkiewicz–Zygmund inequalities","Bernstein–Bézier polynomials","quadrature","interpolation","scattered points","triangles","polygons","Markov inequality"],"falsifier":"Evaluate the left and right sides of (3.13) on a long thin quadrilateral whose triangulation uses only boundary and a few interior points, for $\\chi(x,y)=x^N$ with $N$ just below $c^{-1}|\\triangle|^{-2/p}$ and a range of $p$; if the relative discrepancy exceeds the claimed $\\eta$ with constants as stated, the theorem cannot hold. Independently, maximizing the ratio in Lemma 3.1 over $P_N$ on the same polygon would expose whether its constant stays bounded as $N$ grows.","tokens_in":16101,"feed_emoji":"📐","tokens_out":11248,"duration_ms":97334,"temperature":0.7,"pith_summary":"Marcinkiewicz–Zygmund inequalities compare the $L^p$ norm of a polynomial with a weighted sum of its values at finitely many points, without requiring the quadrature rule to be exact for that polynomial. This paper claims that on any 2D polygon, any scattered point set that includes the vertices can be used as quadrature points: triangulate the polygon with those points as vertices and give each point one third of the sum of the areas of the triangles meeting there. For polynomials whose degree $N$ is small compared with the triangulation size, the resulting positive-weight quadrature satisfies these inequalities for every $1\\le p\\le\\infty$, with constants depending only on $p$ and on the polygon's boundary. The payoff is that methods which only need norm control from discrete samples, such as discrete least-squares projection, can work on general polygonal domains without enforcing quadrature exactness. The paper also proves analogous estimates for 3-, 10-, and 21-point rules on a single triangle and supplies quadrature error bounds.","feed_headline":"Scattered polygon points can serve directly as quadrature nodes","feed_subtitle":"Weights equal one-third of adjacent triangle areas; the rule gives norm inequalities for all p.","key_machinery":"Bernstein–Bézier polynomials, the basis of degree $d$ built from barycentric coordinates on a triangle, supply the norm equivalence, closed-form integrals, and interpolation error estimates that tie values at domain points to $L^p$ norms. The polygon argument is carried by the $d=1$ triangle rule, where the quadrature matrix is the identity and each vertex of a triangle receives weight $|T|/3$; summing these over the triangulation gives the weights in (3.11). The second load-bearing mechanism is Lemma 3.1, a Markov inequality on the polygon obtained by triangulating $\\Omega$ with only its vertices: because those coarse triangles are fixed by the boundary, their shape parameters and the constant $C_5$ do not depend on $N$ or on the fine triangulation $\\triangle$. This Markov inequality converts the second-derivative semi-norm $|\\chi|_{p,2,\\Omega}$ into $C_5 N^2 \\|\\chi\\|_{p,\\Omega}$, which is what turns the triangle-level estimates into the polygon inequalities.","core_discovery":"The paper establishes that scattered data on a polygon can directly define a quadrature rule with norm control. The construction triangulates the polygon with the scattered points as vertices, assigns each point weight $w_j = \\frac13 \\sum_{T\\ni (x_j,y_j)} |T|$, and proves in Theorem 3.6 that for every polynomial $\\chi$ of degree $N$ satisfying $N^2|\\triangle|^{2/p}<1/c_1$ ($1\\le p<\\infty$), the discrete $L^p$ sum $\\sum_j w_j |\\chi(x_j,y_j)|^p$ differs from $\\int_\\Omega |\\chi|^p$ by at most $\\eta$ times the integral, with $\\eta$ given explicitly; for $p=\\infty$ the maximum over the scattered points is within relative error $\\eta$ of the continuous maximum whenever $N$ is at most a constant times $\\min\\{1/|\\triangle|, |\\triangle|^{-1/2}\\}\\sqrt{\\eta}$. The constants involve a polygon Markov constant that depends only on the boundary of $\\Omega$, not on the fine triangulation, which is why the polygon setting succeeds where a single triangle with a fixed quadrature point set cannot. For triangles, the paper proves positive-weight Marcinkiewicz–Zygmund estimates for the 3-, 10-, and 21-point domain-point quadrature rules ($d=1,3,5$), exactness for all polynomials of degree at most $d$, and error bounds in $W^{d+1,p}$.","pith_inferences":["A reader could take the boundary-only Markov constant and build polygon hyperinterpolation schemes whose discrete inner products use these scattered-point weights; the paper does not carry out that application.","If the positivity pattern seen for $d=1,3,5$ continues for all odd $d$, the same coefficient-free construction would yield positive-weight triangle rules at every odd degree; this is a testable extension, not a claim in the paper.","One could replace the 3-point triangle rule inside each coarse triangle by the 10- or 21-point rule to raise the exactness degree of the polygon quadrature, at the cost of requiring interior domain points in every triangle."],"forward_implications":["In any polygonal domain, a positive-weight quadrature that is exact for linear polynomials and satisfies norm-control inequalities for higher-degree polynomials now exists for arbitrary scattered point sets containing the vertices.","The polygon Marcinkiewicz–Zygmund inequalities make it possible to run least-squares projection and related discretizations on scattered polygonal data without assuming quadrature exactness, since the discrete norm is controlled by the continuous norm whenever $N$ is below the stated threshold.","For a single triangle, scattered point sets containing the vertices inherit the polygon inequalities by applying the construction to the triangle itself (Corollary 3.1).","The triangle rules at $d=1,3,5$ provide positive-weight quadrature exact through degree $d$, and the associated error analysis bounds the quadrature error for $f\\in W^{2,p}$ on polygons by $K_1 C_7 \\max_T |A_T|^{1/q} |f|_{2,p,\\Omega}$.","If the Markov inequality on polygons is sharp enough, the same quadrature weights give uniform bounds that do not degrade as the scattered points are refined, provided the degree $N$ is kept within the stated size condition."],"supporting_citations":[{"why":"Supplies the Bernstein–Bézier basis, its norm equivalence, and the triangle Markov inequality on which the quadrature analysis rests.","marker":"[16]"},{"why":"Supplies the interpolation error bound used to pass from polynomial values at nodes to functions in $W^{d+1,p}$.","marker":"[15]"},{"why":"Provides the scattered-data Marcinkiewicz–Zygmund analogue on spheres that motivates the polygon construction.","marker":"[10]"},{"why":"Situates the multivariate-domain Marcinkiewicz–Zygmund results that this paper extends to arbitrary polygons.","marker":"[8]"},{"why":"Establishes the connection between positive quadrature rules and spherical Marcinkiewicz–Zygmund inequalities that the polygon rule mirrors.","marker":"[22]"}],"fun_headline_variants":["Scattered points become quadrature nodes on polygons, with all-p norm bounds","Polygon quadrature from scattered data: Marcinkiewicz–Zygmund for every p","Triangle-area weights turn scattered polygon points into quadrature nodes","Scattered polygon points yield quadrature rules with full p-norm control","From scattered points to polygon quadrature: explicit norm inequalities for all p"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that on a fixed polygon, every derivative of a degree-$N$ polynomial is bounded by a constant times $N^2$ times the polynomial's norm, with the constant depending only on the polygon's boundary, not on $N$ or on how finely the polygon is triangulated; the proof gets this from the coarsest vertex-only triangulation and hinges on absorbing a degree-dependent factor from the triangle Markov inequality.","fun_headline_variants_meta":{"raw":{"variants":["Scattered points become quadrature nodes on polygons, with all-p norm bounds","Polygon quadrature from scattered data: Marcinkiewicz–Zygmund for every p","Triangle-area weights turn scattered polygon points into quadrature nodes","Scattered polygon points yield quadrature rules with full p-norm control","From scattered points to polygon quadrature: explicit norm inequalities for all p"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000342,"raw_usage":{"total_tokens":1915,"prompt_tokens":1012,"completion_tokens":903,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":628,"completion_tokens_details":{"reasoning_tokens":805}},"tokens_in":628,"tokens_out":903,"duration_ms":8523,"temperature":1.0,"reasoning_tokens":805,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T12:59:14.704939+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Evaluate the left and right sides of (3.13) on a long thin quadrilateral whose triangulation uses only boundary and a few interior points, for $\\chi(x,y)=x^N$ with $N$ just below $c^{-1}|\\triangle|^{-2/p}$ and a range of $p$; if the relative discrepancy exceeds the claimed $\\eta$ with constants as stated, the theorem cannot hold. Independently, maximizing the ratio in Lemma 3.1 over $P_N$ on the same polygon would expose whether its constant stays bounded as $N$ grows.","supporting_citations":[{"cited_title":"Lai and L","cited_arxiv_id":null,"evidence_quote":"Supplies the Bernstein–Bézier basis, its norm equivalence, and the triangle Markov inequality on which the quadrature analysis rests."},{"cited_title":"Lai and L","cited_arxiv_id":null,"evidence_quote":"Supplies the interpolation error bound used to pass from polynomial values at nodes to functions in $W^{d+1,p}$."},{"cited_title":"De Marchi and A","cited_arxiv_id":null,"evidence_quote":"Situates the multivariate-domain Marcinkiewicz–Zygmund results that this paper extends to arbitrary polygons."}],"review_version":1}