{"id":"57f06004-b53e-472d-a0a3-d448cb4038af","arxiv_id":"2508.19047","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A sharp dimension bound for Furstenberg sets built from transversal families of graphs, with an application to Fourier decay of fractal measures on convex curves.","lead":"This paper proves a Furstenberg set theorem for families of graphs obeying a transversality condition, extending the known result for lines. It then bounds Fourier transforms of fractal measures on general convex curves, not just the parabola.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Lemma 6.21 is stated without proof and is load-bearing for the t=s and semi-well-spaced cases; if its adaptation from [20] fails, Theorem 1.11 collapses.","rationale":"The central claim of the paper is a genuine generalization of the Ren-Wang theorem to transversal families, and the bulk of the proof is a careful adaptation of the known linear machinery. I found no definite internal contradiction: the rescaling lemmas, the curvilinear high-low estimate, and the induction in Proposition 4.13 are consistent with the stated hypotheses. The non-tangential transversality condition in Definition 1.2 is restrictive—it excludes cinematic and tangential families—but that is an acknowledged limitation rather than a flaw. The most load-bearing concern is the explicitly omitted proof of Lemma 6.21, which the paper itself flags. This lemma is needed for the t=s case and for the semi-well-spaced argument, so the written proof of Theorem 1.11 is incomplete at a central juncture. The reader's verdict is already CONDITIONAL and correctly identifies the omitted proof as a reason for conditionality, though the reader's formal 'weakest assumption' emphasizes the transversality restriction rather than this missing lemma. I therefore partially agree and see no reason to change the verdict.","tokens_in":84378,"tokens_out":11051,"duration_ms":111099,"concrete_test":"Write out a complete proof of Lemma 6.21 by translating every step of [20, Lemma 4.11] into the C^2 dyadic system of Definition 2.23, verifying at each stage that only the geometric bound in Remark 6.22 is used. In particular, check the Katz-Tao discretization step: if it requires a Euclidean identity (e.g., about intersections of tubes with grids) that has no C^2 analogue, then Lemma 6.21 is unproved and the t=s and semi-well-spaced estimates lack support.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper explicitly omits the proof of Lemma 6.21, saying only that it is identical to the original and that the only geometric property needed is the pairwise intersection bound in Remark 6.22. This lemma is not peripheral: it is used in Corollary 6.23 to obtain the t=s case of the Furstenberg estimate and in Proposition 6.13 for the semi-well-spaced case, both of which feed directly into the discretized Theorem 1.11 and hence into Theorem 1.9. The adaptation from [20, Lemma 4.11] is not automatic: the original proof is formulated for straight tubes and Euclidean dyadic grids, whereas here F is a subset of C^2(I) with its own dyadic system (Definition 2.23), and the 'Katz-Tao' small-scale structure must interact with C^2-balls. The authors assert that Remark 6.22 suffices, but do not demonstrate that every step of the original proof—especially the dyadic pigeonholing and the construction of the low-density subsets—survives verbatim. This is a missing proof of a load-bearing ingredient, not merely a cosmetic omission. If Lemma 6.21 is false or requires an additional transversality-type hypothesis, the t=s and semi-well-spaced cases are unsupported, and Theorem 1.11 would not follow from the written argument.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves a nonlinear analogue of the Ren–Wang planar Furstenberg set theorem. For a family F ⊂ C^2(I) satisfying the non-tangential transversality condition of Definition 1.2 and a plane set F whose intersection with each graph Γ_f has Hausdorff dimension at least s, Theorem 1.9 gives dim_H F ≥ min{s+t, (3s+t)/2, s+1}, where t is the dimension of F. The proof follows the Ren–Wang three-regime strategy: a δ-discretised theorem (Theorem 1.11) is built from a projection theorem (Theorem 4.1), a regular-case estimate (Theorem 5.3), a semi-well-spaced estimate (Proposition 6.13), and a branching-function interpolation (Theorem 7.1). The discretised theorem is then converted to the Hausdorff-dimension statement in §1.3. As an application, Theorem 1.16 derives L^p Fourier decay bounds for measures on convex curves from the special case of translates of a fixed convex graph.","tokens_in":84695,"tokens_out":11103,"duration_ms":96956,"significance":"If correct, Theorem 1.9 is a substantial and natural extension of the linear Ren–Wang estimate to nonlinear transversal curve families. The paper contains genuinely new ingredients, in particular the Fourier-free high-low incidence estimate (Proposition 3.5) and the curvilinear incidence lemma (Lemma 6.19, with a full proof in Appendix A). The Fourier application to measures on general convex curves is a meaningful advance beyond the parabola. The paper is technically demanding, carefully structured, and transparent about its black boxes: the linear Ren–Wang theorem is used explicitly as the linear base case, and Proposition 7.2 is imported from the literature. The proof of the discretised-to-dimension passage in §1.3 is clear and, as far as I can check, correct.","major_comments":[{"comment":"The proof of Proposition 4.13 (Step 5) closes by applying Theorem 4.90, the linear δ-discretised projection theorem. The text says the details are not repeated and are ‘scattered in the literature’, referring to [17, Cor. 6.1] and [20, Thm 4.1]. This is a load-bearing input: it is the only place where the linear Ren–Wang theorem enters the projection theorem, and Theorem 4.90 is a dual formulation not literally identical to [17, Cor. 6.1]. Please either state Theorem 4.90 as a known theorem with a precise reference, or supply the short deduction from [20, Thm 4.1] and [17, Cor. 6.1]. The current ‘reader should check’ is too terse for a result on which Theorem 1.9 depends.","section":"§4.2, Theorem 4.90"}],"minor_comments":[{"comment":"At the end of the proof, the vertical dilation (x,y) ↦ (x, Λ^{-1}y) is stated to increase the transversality constant by O_Λ(1), and the details are left to the reader. Since this corollary is used in the Fourier application, please write out the effect of the dilation on the family and on the dyadic squares.","section":"§1.4, Corollary 1.14"},{"comment":"The proof is said to follow from the definition and is left to the reader. A one-line proof would improve readability.","section":"§4, Lemma 4.5"},{"comment":"The proof says ‘We only treat explicitly the special case where n := 4/ϵ ∈ N.’ For arbitrary ϵ > 0 a short reduction should be mentioned, e.g. by replacing ϵ with a smaller value for which 4/ϵ is an integer.","section":"§6.1, Corollary 6.11"},{"comment":"The choice of the arbitrary centre f_F in each dyadic cube is used implicitly in later rescaling lemmas (e.g. Lemma 2.28). It would help to state explicitly that the results are independent of this choice or that a fixed selection is made once and for all.","section":"§2.3, Definition 2.26"}],"recommendation":"major_revision","confidential_remarks":"This is a strong paper with a serious but fixable gap: Lemma 6.21 is load-bearing and currently rests on an assertion that the proof is identical to one in the straight-tube setting. I would not recommend acceptance until the authors supply a full proof or a precisely referenced proof of this lemma. The rest of the manuscript appears solid, and the paper should be of interest to the journal's readership."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The main result is Theorem 1.9: the Ren-Wang threshold min{s+t, (3s+t)/2, s+1} holds when lines are replaced by any transversal family of C^2 graphs. That is genuinely new; prior nonlinear cases needed stronger separation assumptions. The proof follows the Ren-Wang architecture — regular case, semi-well-spaced case, interpolation — and the hard new input is the curvilinear high-low estimate (Prop 3.5) and the incidence lemma (6.19) proved in Appendix A. I looked at those arguments and they are substantive: the transversality condition is chosen so that it survives both rescaling operations, and the high-low lemma uses the Cohen–Pohoata–Zakharov Fourier-free argument, which is the right tool for curved tubes.\n\nThe paper is honest about what it imports: Ren–Wang is used as a black box in the regular case, and Proposition 7.2 comes from Orponen–Shmerkin. That is acceptable. The Fourier application to general convex curves is a nice payoff and follows from Corollary 1.14.\n\nThe soft spot is Lemma 6.21, stated without proof on the grounds that it is identical to Ren–Wang's Lemma 4.11. The stress-test note is right that this lemma is not peripheral: it drives the t=s case (Corollary 6.23) and the semi-well-spaced case. The stated justification — that only the pairwise intersection bound in Remark 6.22 is needed — is plausible; that bound follows from Lemma 2.14, and the rest of the proof is the usual dyadic pigeonholing plus the Katz–Tao condition. So the omission is probably repairable rather than fatal. But \"probably\" is not \"certainly,\" and an 80-page proof with a load-bearing lemma on trust is exactly what a referee should ask to be cleaned up. The authors should either write out the adaptation or give a very precise translation table from [20, Lemma 4.11] to their setting.\n\nMinor issues: Corollary 1.14 leaves a few details to the reader, and Lemma 4.5 is trivial but stated without proof. Nothing that bothers me.\n\nBottom line: this deserves a serious referee. It is a real theorem, not a repackaging, and the main proof is in the paper. The authors should be asked to supply the proof of Lemma 6.21 in revision, but I would not desk-reject over it.","headline":"Real extension of Ren-Wang to transversal graph families, with the main proof mostly present; the one load-bearing omitted proof (Lemma 6.21) is disclosed but should be supplied before publication.","tokens_in":85155,"tokens_out":2538,"would_cite":true,"duration_ms":26373,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["28A80","28A78","42B10"],"pacs":[],"model":"deepseek-v4-flash","headline":"For families of C^2 curves that are transversal—any two graphs separate in height or slope at every point—the paper proves the Furstenberg set threshold for lines still holds: dimension at least min{s+t, (3s+t)/2, s+1}.","keywords":["Furstenberg set","transversal family","Hausdorff dimension","Fourier transform","fractal measures","incidence estimates","plane curves","projection theorem"],"falsifier":"Take a transversal family at a small scale and test the key incidence estimate in Proposition 3.5: it predicts that incidences between δ-separated graph fragments and δ-squares are at most a constant times (S^3δ^{-1}|F||P|)^{1/2} plus S^{-1} times the large-scale incidence term. A configuration whose incidence count exceeds this bound would break the proof's central lemma. Alternatively, run the discretised Theorem 1.11 on an admissible configuration: if |P| falls below δ^η·min{δ^{-t}, δ^{-(s+t)/2}, δ^{-1}}·M for small enough δ, the theorem's key quantitative claim is false.","tokens_in":84282,"feed_emoji":"📐","tokens_out":7270,"duration_ms":76735,"temperature":0.7,"pith_summary":"The paper proves that the sharp Furstenberg set bound known for families of lines remains true when the lines are replaced by a family of C^2 curves satisfying a transversality condition: any two graphs must separate in value or slope uniformly at every point. The main theorem says that if the curve family has dimension at least t and a planar set intersects each graph in dimension at least s, then the set itself has dimension at least min{s+t, (3s+t)/2, s+1}. The proof proceeds through a δ-discretised estimate and a projection theorem for transversal families, matching the structure of the linear argument. As an application, the authors derive L^p Fourier-decay bounds for fractal measures on plane curves with non-vanishing second derivative, extending earlier parabola results to arbitrary strictly convex curves. The result matters because it indicates that the Furstenberg phenomenon is driven by non-tangential separation and rescaling invariance, not by algebraic straightness.","feed_headline":"Curved Furstenberg sets keep the sharp linear bound","feed_subtitle":"For families of curves that always separate in value or slope, any set meeting them in dimension s has the same sharp lower bound as the lin","key_machinery":"The load-bearing object is the transversal family (Definition 1.2): a family of C^2 functions in which any two graphs separate, at every point, in value or slope by an amount comparable to their full C^2 distance. This non-tangential separation is what makes the graphs behave like a nonlinear analogue of a line family; it is invariant under two natural rescaling operations (Lemmas 2.19 and 2.20), which enables the induction-on-scales scheme. The other central mechanism is the curvilinear high-low incidence estimate (Proposition 3.5), a Fourier-free replacement for the tube Fourier lemma that bounds incidences between δ-separated graph fragments and δ-squares by a small-scale term plus a larg","core_discovery":"The central claim is Theorem 1.9: if F⊂C^2(I) is a transversal family with dim_H F≥t, and F⊂R^2 satisfies dim_H(F∩Γ_f)≥s for every graph Γ_f of f∈F, then dim_H F≥min{s+t,(3s+t)/2,s+1}. This matches the sharp linear Furstenberg threshold and is proved through a discretised version (Theorem 1.11) that gives polynomial lower bounds on the number of δ-squares. The transversality condition (Definition 1.2) requires inf_{θ∈I}(|f(θ)−g(θ)|+|f′(θ)−g′(θ)|)≥T^{-1}‖f−g‖_{C^2(I)}, which rules out tangential intersections and is preserved under the rescaling operations used for induction on scales. The paper also proves Theorem 1.16: for g∈C^3 with g″ never vanishing, any measure on the graph segment sati","pith_inferences":["Beyond the paper: because the threshold matches the linear case, any genuine failure of Furstenberg-type bounds for curved leaves would likely have to come from tangential or higher-order tangency, not from curvature itself; the paper's Remark 1.4 leaves exactly this open for cinematic families.","The Fourier-free high-low lemma suggests the machinery is portable: one could test whether other rescaling-invariant families of curves—for example algebraic curve families with a uniform non-tangency condition—admit the same induction without Fourier tube structure.","The paper's Remark 1.18 notes that p=6 may always suffice in the Fourier application; a concrete testable strengthening would be to prove or disprove the p=6 bound for all g with g″≠0, rather than only the known s≥2/3 range.","If the proof's rescaling invariance is robust, a natural extension would be higher-dimensional analogues where leaves are hypersurfaces satisfying an analogous first-order separation condition, although the discretised machinery would need a new projection theorem."],"forward_implications":["If Theorem 1.9 is correct, the sharp linear Furstenberg threshold min{s+t,(3s+t)/2,s+1} holds unchanged when the leaf family is any transversal family of graphs, not just lines.","The discretised Theorem 1.11 gives quantitative bounds: for admissible configurations, |P| ≥ δ^η·min{δ^{-t}, δ^{-(s+t)/2}, δ^{-1}}·M, which underlies the Hausdorff-dimension statement.","Corollary 1.14 shows that translates of a fixed strictly convex C^3 curve form a transversal family with constants independent of location, so the theorem applies uniformly at all scales and positions.","Theorem 1.16 extends the Fourier-decay estimate for measures on the parabola to measures on any plane curve segment with non-vanishing second derivative, for t