{"id":"d9a5dd05-7ee2-400f-84fd-02d42085779e","arxiv_id":"2608.10253","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"A Federer-style projection framework for nonlinear maps yields structural bounds on exceptional pins, radial projection vantage points, and circle unions of 1-rectifiable sets.","lead":"This paper develops a general technique that turns a classical theorem about linear projections into new results for nonlinear measurements such as distances, angles, and circle intersections. It shows that one-dimensional curved sets always have points from which distances spread over a positive-length range, and that unions of variable-radius circles over such sets have positive area.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The WLOG reduction to Ω_ε in Theorems 2.3 and 4.2 is asserted without proof; the determinant lower bound alone does not imply H^1(H(E)) > 0 for unbounded E, so the central framework rests on an unstated bounded-subset step.","rationale":"The central claim is that the nonlinear-projection framework is sound: if the canonical embedding H_α(E) is 1-rectifiable with positive H^1, then Federer's projection theorem yields a nonlinear projection of positive length. Every application tries to establish the positive-measure condition by proving a Jacobian lower bound on a domain Ω_ε. The load-bearing weakness is that the determinant lower bound alone does not make the map Lipschitz on an unbounded domain, so the positivity of H^1(H(E)) is not automatic; one must first pass to a bounded positive-measure subset. This is exactly the reduction that the proofs of Theorems 2.3 and 4.2 assert without proving. The gap is real but clearly fixable, since H^1 is sigma-finite and some bounded part of the positive-measure piece must carry positive measure. No counterexample to the results themselves is apparent; the framework and the three applications remain plausible. Theorem 3.1's reduction is less problematic because its Ω_ε is bounded and the text first establishes H^1(E∩Ω_ε)>0 before replacing E. For this reason I partially agree with the reader's weakest_assumption. The concrete test above — rewriting the reduction explicitly in Theorem 2.3 — would settle whether the concern lands. Since the reader already issued a CONDITIONAL verdict and this stress-test does not move it, the verdict remains unchanged.","tokens_in":14644,"tokens_out":25510,"duration_ms":222056,"concrete_test":"Re-prove Theorem 2.3 replacing the unstated WLOG with the explicit step: choose R so that H^1(E∩Ω_ε∩B_R)>0, apply Federer's theorem to H_P(E∩Ω_ε∩B_R) instead of H_P(E), and verify that the extracted good pin for the subset is also good for E. If this substitution cannot be completed from the text of Section 2, the proof gap is confirmed. As a sanity check in R^2, take E={(t,t): t≥1}, P={(−1,0),(1,0)}; verify that H_P is not Lipschitz on E∩Ω_ε but is bilipschitz on every bounded truncation, and that restricting to a bounded truncation still yields a good pin.","verdict_should_be":"UNCHANGED","load_bearing_attack":"In the proof of Theorem 2.3, after obtaining H^1(E\\V_P(ε)) > 0, the text states 'we may therefore assume without loss of generality that V_P = R^{d-1}×{0} and that E⊆Ω_ε', where Ω_ε = R^{d-1}×(ε,1/ε). This is logically inverted: E\\V_P(ε) is the set of points at distance >ε from V_P, not a subset of Ω_ε. Moreover, Ω_ε is unbounded in the first d-1 coordinates. Lemma 2.4 only bounds det DH_P on Ω_ε, and a determinant bound does not control the operator norm of DH_P (the columns 2(z−p_j) have unbounded first d−1 entries). The next assertion, 'Therefore H_P(E) is a 1-rectifiable set in R^d with positive length', consequently lacks its needed justification: the Jacobian lower bound yields positivity of H^1(H_P(A)) only on bounded subsets A on which H_P is genuinely Lipschitz. The gap is fixable by first choosing a positive-measure bounded subset E′⊆E∩Ω_ε∩B_R, applying the area formula to H_P(E′), and then observing that a good pin for E′ is good for E. The same unstated reduction appears in Theorem 4.2, where the proof says 'we may also assume E⊆(a/10,11a/100)×R'; this requires passing to a positive-measure subset of E lying in some translate of that vertical strip and then translating coordinates. If the bounded-subset reduction were unavailable, the determinant lower bounds of Lemmas 2.4 and 4.3 would not by themselves imply positivity of the canonical image for unbounded E, so the central Step (2) of the framework would fail.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The manuscript develops a transfer principle for Federer's projection theorem: given a family of nonlinear maps φ_α: Ω→R whose canonical embedding H_α=(φ_{α1},...,φ_{αd}) is locally bilipschitz or has Jacobian bounded below, one applies Federer's theorem to H_α(E) to conclude that at least one nonlinear image φ_{α_j}(E) has positive length. This framework is applied to three problems: pinned distance sets for 1-rectifiable sets not essentially (d−1)-flat (Theorem 2.3) with structural consequences for the set of bad pins (Theorem 2.6); radial projections in the plane (Theorem 3.1); and unions of variable-radius circles centered on a 1-rectifiable set (Theorem 4.2). The paper also proves a short d-lines corollary to illustrate the method.","tokens_in":15049,"tokens_out":11361,"duration_ms":106478,"significance":"The framework is attractive and, if correct, would provide a genuinely common route to several results at the rectifiable endpoint, including sharp statements on exceptional sets. The Jacobian computations in Lemmas 2.4 and 3.2 are explicit and check out, and the use of Federer's theorem as a black box makes the combinatorial step in Sections 2 and 3 conceptually clean. However, the proof of Theorem 4.2 rests on a determinant estimate in Lemma 4.3 that appears to be false as stated, and the proofs of Theorems 2.3, 3.1, and 4.2 contain an unstated bounded-subset reduction. These issues are local and fixable in principle, but they are load-bearing for the central claims.","major_comments":[{"comment":"The inference from H^1(E \\ V_P(ε))>0 to \"we may therefore assume without loss of generality that ... E⊆Ω_ε\" is logically inverted and skips a necessary boundedness reduction. Having H^1(E \\ V_P(ε))>0 only means that a positive-measure part of E lies at distance at least ε from the affine hull, not that all of E lies in Ω_ε=R^{d-1}×(ε,1/ε); moreover Ω_ε is unbounded in the first d−1 coordinates. Since Lemma 2.4 only bounds det DH_P on Ω_ε and the columns of DH_P are 2(z−p_j), a determinant bound does not control the operator norm of DH_P on unbounded sets, and the Jacobian lower bound alone does not imply H^1(H_P(E))>0 for unbounded E. The proof should first choose a bounded positive-measure Borel subset E'⊆E∩V_P(ε)^c, reflect or rotate so that E'⊆Ω_ε∩B_R, apply the area formula to H_P(E'), and then note that a good pin for E' is a good pin for E because ∆_p(E')⊆∆_p(E). The same missing reduction appears in the proof of Theorem 3.1 (assumption E⊆Ω_ε) and in the proof of Theorem 4.2 (assumption E⊆(a/10,11a/100)×R).","section":"Section 2, Proof of Theorem 2.3; also Section 3 and Section 4"},{"comment":"In the decomposition det DH_(α,β) = (A)+(B)+(C), term (C) has the wrong sign: since α<β and r_y≥0, the expression (C) = (r_y/r)·(α−β)/(√(1−((x−α)/r)^2)√(1−((x−β)/r)^2)) is nonpositive, not nonnegative as claimed. The lower bound for det DH therefore depends on whether |C| can dominate the positive contribution (A) ≥ (α−β)/∥r∥∞. It can: with r(x,y)=a+(b−a)/(1+e^{−Ky}) (so r_x=0, r_y≥0, bounded range and bounded gradient), term (C) is approximately −K(b−a)(α−β)/(4r), which for large K exceeds the positive lower bound (α−β)/b. Thus Lemma 4.3's assertion that the Jacobian determinant is bounded away from zero is false without an additional smallness condition on ∥∇r∥∞, and the proof of Theorem 4.2 collapses at this point. A corrected estimate or a modified hypothesis on r is required before the circle-union theorem can be accepted.","section":"Lemma 4.3, determinant decomposition"}],"minor_comments":[{"comment":"The word \"bilipshictz\" should be \"bilipschitz\", and \"1-unrectifable\" should be \"1-unrectifiable\".","section":"Section 1.2"},{"comment":"The notation π_j(t)∩H(E) should be π_j^{-1}(t)∩H(E) in two places, since π_j maps R^d to R and H(E) is a subset of R^d.","section":"Proof of Corollary 1.3"},{"comment":"The sentence claiming that H_P is \"locally Lipschitz due to the upper bound on detDH_P\" is imprecise: local Lipschitzness follows from boundedness of the full derivative matrix on bounded subsets, not from a determinant upper bound alone.","section":"Proof of Theorem 2.3"},{"comment":"The definition of essentially k-flat in Remark 2.2 differs from the Orponen–Sahlsten definition, and Section 3 later refers to \"k-flat\" without repeating the variant; this distinction should be stated where the term is used in Theorem 3.1.","section":"Remark 2.2 and Section 3"},{"comment":"The final integration uses |φ_γ(E)|>0 for all γ in an interval, but the argument before it only shows this for γ in a set of positive measure or for a dichotomy of intervals; the Fubini step should be written more carefully to make clear how pointwise positivity on a full positive-length interval is obtained.","section":"Proof of Theorem 4.2"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is generally carefully written and the central transfer idea is appealing. The main concerns are the fixable bounded-subset gap in Sections 2–4 and the determinant sign error in Lemma 4.3; the latter is load-bearing for Theorem 4.2 and needs a genuine repair. I would be willing to review a revision. Given the heavy citation of 2026 preprints, the authors should ensure these references are publicly available or clearly marked as forthcoming."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThis paper gives a clean transfer principle: if you can show the canonical embedding H_α = (φ_{α_1},...,φ_{α_d}) is bilipschitz with Jacobian bounded away from zero on the relevant domain, then Federer's projection theorem does the rest. The three applications — pinned distances, planar radial projections, and variable-radius circle unions — are honest illustrations, and the Jacobian calculations in Lemmas 2.4, 3.2, and 4.3 are explicit and correct. I particularly like the bad-pin subspace bound: the exceptional set of pins is contained in a (d−2)-plane, and the circle example shows the dimension is sharp.\n\nThe paper is also candid about what is new. The introduction discloses that Li–Taylor already give alternative proofs of the pinned-distance and radial-projection results, and the radial projection theorem recovers Orponen–Sahlsten. So the real novelty is the framework itself, the d-dimensional bad-pin bound, and the circle-union theorem. That's a fair and honest package.\n\nWhere I have reservations: the 'without loss of generality' reductions in Theorems 2.3 and 4.2. In Theorem 2.3 the proof goes from H^1(E \\ V_P(ε)) > 0 to 'we may assume E ⊆ Ω_ε'. That is not a valid WLOG: the first set is outside the ε-neighborhood of the hyperplane, not inside a slab. To get into the slab you need to pass to a positive-measure bounded subset, and the proof doesn't say so. The determinant lower bound in Lemma 2.4 does not by itself control the operator norm of DH_P on the unbounded domain, so the claimed positivity of H^1(H_P(E)) needs the bounded-subset step. The same issue appears in Theorem 4.2's reduction to a vertical strip; you need to translate and restrict to a subset, and the monotonicity of the area conclusion makes that harmless, but it has to be stated. In Theorem 3.1 the WLOG 'E ⊆ Ω_ε' is actually valid because E∩Ω_ε has positive measure, but the text compresses the argument.\n\nThese are real but fixable gaps. I don't think they sink the framework; they just need a short paragraph explaining the standard restriction to a bounded positive-measure subset, after a translation/reflection if necessary. The area formula then does exactly what the authors want.\n\nThe citation pattern is fine. Federer is used as an external black box, the self-citations are disclosed, and none of the maps are defined in terms of the conclusions.\n\nWho is this for? Geometric measure theorists working on projection theory, distance sets, or circle unions. It deserves a serious referee: the framework is reusable, the applications are at the right endpoint, and the gaps are fixable. I'd encourage you to send it out rather than desk-reject, and ask the authors to tighten the WLOG arguments.","headline":"A clean, reusable framework for nonlinear projections at the rectifiable endpoint, with two sloppy but fixable WLOG reductions.","tokens_in":15591,"tokens_out":8228,"would_cite":true,"duration_ms":75868,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["28A75","28A78"],"pacs":[],"model":"deepseek-v4-flash","headline":"A classical projection theorem for rectifiable sets is extended to nonlinear maps: distance, radial, and circle families with a nondegenerate canonical embedding each force at least one positive-length image, with bad indices in a…","keywords":["nonlinear projections","Federer projection theorem","1-rectifiable sets","pinned distance sets","radial projections","unions of circles","canonical embedding","exceptional sets"],"falsifier":"A direct disproof would be a 1-rectifiable planar set $E$, not essentially a line, with two distinct vantage points $p\\neq q$ for which $H^1(\\pi_p(E))=H^1(\\pi_q(E))=0$, since Theorem 3.1 permits at most one such point. A natural test case is a union of three non-collinear segments: the radial image from the common endpoint is a finite set, and Theorem 3.1 predicts every other point is good, so checking the radial image from a second segment endpoint would already put the mechanism at risk.","tokens_in":14440,"feed_emoji":"📐","tokens_out":13181,"duration_ms":113464,"temperature":0.7,"pith_summary":"Federer's projection theorem says that among any $d$ linearly independent orthogonal projections of a 1-rectifiable set in $\\mathbb{R}^d$, at least one image has positive length. This paper claims the same conclusion survives when the linear projections are replaced by nonlinear maps—distance to a pin, angle from a vantage point, or the height of a circle intersection—provided the maps can be assembled into a single embedding whose Jacobian determinant is bounded away from zero. The payoff is a unified proof of three results at the critical dimension 1: every 1-rectifiable set contains a pin whose pinned distance set has positive measure, planar radial projections have at most one bad vantage point unless the set is essentially linear, and variable-radius circles centered on a 1-rectifiable set cover positive area. A curious reader should care because these statements are usually attacked with Fourier or dimension-theoretic tools, while here they reduce to a determinant calculation plus a classical projection theorem.","feed_headline":"Nonlinear projections keep positive length on rectifiable sets","feed_subtitle":"Federer's projection theorem is extended to distance maps, radial views, and circles, with small exceptional sets.","key_machinery":"The load-bearing object is the canonical embedding $H_{\\vec{\\alpha}}(z)=(\\varphi_{\\alpha_1}(z),\\dots,\\varphi_{\\alpha_d}(z))$ built from a family of nonlinear scalar maps. The proofs work by computing $\\det DH_{\\vec{\\alpha}}$ and showing it is bounded away from zero and infinity on a conveniently chosen slab or strip: for distance-squared pins the determinant is $2^d\\,\\pi_d(z)\\det[p_1-p_2\\,\\cdots\\,p_1-p_d]$, for radial angle maps it is $y/((x_+^2+y^2)(x_-^2+y^2))$, and for circle intersections it is a sum of three separately controlled terms. Boundedness away from zero makes $H_{\\vec{\\alpha}}$ locally bilipschitz, hence positivity-preserving for length, and converts the nonlinear problem into the linear projection problem solved by Federer's theorem. The black-box use of that theorem is deliberate: all technical difficulty is relocated into the Jacobian estimate.","core_discovery":"The central claim is a transfer principle. Given indices $\\alpha_1,\\dots,\\alpha_d$ and maps $\\varphi_{\\alpha_j}:\\Omega\\to\\mathbb{R}$, form the canonical embedding $H_{\\vec{\\alpha}}(z)=(\\varphi_{\\alpha_1}(z),\\dots,\\varphi_{\\alpha_d}(z))$. If $H_{\\vec{\\alpha}}$ is bilipschitz—or more generally if $\\det DH_{\\vec{\\alpha}}$ is bounded away from zero and infinity on the relevant domain—then for every 1-rectifiable set $E\\subseteq\\Omega$ with positive length, $H_{\\vec{\\alpha}}(E)$ is again a 1-rectifiable set of positive length. Applying the classical projection theorem to $H_{\\vec{\\alpha}}(E)$ forces at least one coordinate projection $\\pi_j(H_{\\vec{\\alpha}}(E))$ to have positive length, and that coordinate projection is exactly the nonlinear image $\\varphi_{\\alpha_j}(E)$. The paper realizes this scheme for distance-squared maps $\\varphi_p(z)=|z-p|^2$, for planar arctangent angle maps, and for circle-intersection maps $\\varphi_\\alpha(x,y)=y+\\sqrt{r(z)^2-(\\alpha-x)^2}$, then converts 'at least one good index' into sharp geometric restrictions on the exceptional set of bad pins or vantage points.","pith_inferences":["The exceptional-set bounds are qualitative; extending the determinant estimates to families of maps with parameters would likely give quantitative dimension bounds for bad pins or vantage points, with the $(d-2)$-flat bound for pins a natural sharpness target.","The same canonical-embedding recipe should apply to other curve families with computable Jacobians, such as parabolas or hyperbolas, producing new 'good curve' theorems for 1-rectifiable sets.","The localization gap in the proofs suggests a standalone lemma—every 1-rectifiable set of positive length contains a bounded positive-measure subset on which a nondegenerate $C^1$ embedding is bilipschitz—that would complete the 'without loss of generality' steps and might generalize the framework beyond the three examples."],"forward_implications":["Every 1-rectifiable set $E\\subseteq\\mathbb{R}^d$ with $H^1(E)>0$ contains a pin $p$ with $H^1(\\Delta_p(E))>0$, and the set of bad pins lies in an affine subspace of dimension at most $d-2$.","In the plane, a 1-rectifiable set that is not essentially 1-flat has at most one bad radial vantage point; an essentially 1-flat set has bad vantage points forming exactly a line.","For any admissible radius function $r$—bounded above and below, differentiable with bounded gradient, and with monotone partial derivatives—the union of circles centered on a 1-rectifiable set $E\\subseteq\\mathbb{R}^2$ has positive two-dimensional Lebesgue measure.","The framework itself yields a nonlinear d-lines theorem: any $d$ nonlinear maps whose canonical embedding satisfies the Jacobian condition contain at least one good index, and this combinatorial fact is what produces the low-dimensional exceptional sets."],"supporting_citations":[{"why":"Supplies the classical projection theorem, the black-box inequality that the framework reduces every nonlinear problem to.","marker":"[7]"},{"why":"Provides a prior quantified two-projection theorem for nonlinear projections, noted by the authors as yielding alternative proofs of the pinned-distance and radial-projection results.","marker":"[12]"},{"why":"Records the antecedent radial-projection result for rectifiable sets that Theorem 3.1 recovers.","marker":"[24]"},{"why":"Establishes the fixed-radius circle-union theorem identifying rectifiability as the critical condition for positive measure.","marker":"[27]"},{"why":"Gives the zero-measure circle-union construction that motivates the admissibility condition on the radius function.","marker":"[28]"},{"why":"Presents the variable-hypersurface result compared with the circle theorem to highlight the different geometric mechanism.","marker":"[9]"},{"why":"Shows the unrectifiable counterpart, framing the contrast the paper emphasizes at the critical dimension.","marker":"[11]"}],"fun_headline_variants":["Federer's theorem goes nonlinear for rectifiable sets","Curved projections keep rectifiable sets visible","Nonlinear projection theorem: pins, radii, and circles","Nonlinear maps keep rectifiable sets positive in measure","Without straight lines: measure still holds for nonlinear projections"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proofs for pinned distances and radial projections assume, without proof, that a 1-rectifiable set of positive length can be cut down to a bounded piece, still of positive length, on which the encoding map is one-to-one with stretch factors bounded above and below; if this cutting-down step fails, the determinant estimate alone does not imply that the image has positive length.","fun_headline_variants_meta":{"raw":{"variants":["Federer's theorem goes nonlinear for rectifiable sets","Curved projections keep rectifiable sets visible","Nonlinear projection theorem: pins, radii, and circles","Nonlinear maps keep rectifiable sets positive in measure","Without straight lines: measure still holds for nonlinear projections"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001224,"raw_usage":{"total_tokens":5075,"prompt_tokens":1032,"completion_tokens":4043,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":648,"completion_tokens_details":{"reasoning_tokens":3967}},"tokens_in":648,"tokens_out":4043,"duration_ms":33097,"temperature":1.0,"reasoning_tokens":3967,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T04:13:49.712400+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A direct disproof would be a 1-rectifiable planar set $E$, not essentially a line, with two distinct vantage points $p\\neq q$ for which $H^1(\\pi_p(E))=H^1(\\pi_q(E))=0$, since Theorem 3.1 permits at most one such point. A natural test case is a union of three non-collinear segments: the radial image from the common endpoint is a finite set, and Theorem 3.1 predicts every other point is good, so checking the radial image from a second segment endpoint would already put the mechanism at risk.","supporting_citations":[{"cited_title":"Federer.Geometric Measure Theory","cited_arxiv_id":null,"evidence_quote":"Supplies the classical projection theorem, the black-box inequality that the framework reduces every nonlinear problem to."},{"cited_title":"A Quantified Two-projection Theorem for Nonlinear Projections","cited_arxiv_id":"2606.00381","evidence_quote":"Provides a prior quantified two-projection theorem for nonlinear projections, noted by the authors as yielding alternative proofs of the pinned-distance and radial-projection results."},{"cited_title":"Orponen and T","cited_arxiv_id":null,"evidence_quote":"Records the antecedent radial-projection result for rectifiable sets that Theorem 3.1 recovers."},{"cited_title":"Simon and K","cited_arxiv_id":null,"evidence_quote":"Establishes the fixed-radius circle-union theorem identifying rectifiability as the critical condition for positive measure."},{"cited_title":"Talagrand","cited_arxiv_id":null,"evidence_quote":"Gives the zero-measure circle-union construction that motivates the admissibility condition on the radius function."},{"cited_title":"Favard length and generalized projections","cited_arxiv_id":"2607.28793","evidence_quote":"Shows the unrectifiable counterpart, framing the contrast the paper emphasizes at the critical dimension."}],"review_version":1}