{"id":"386df665-6474-41e5-bb75-4f2d1eeb71ce","arxiv_id":"2608.10476","paper_version":1,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"For an admissible family of d smooth nonlinear measurements, every positive-length curve-like set has at least one measurement whose image has positive length.","lead":"For one-dimensional curve-like sets in space, the paper shows that from any suitable collection of d nonlinear measurements, at least one measurement must have positive length. It extends a classical projection theorem to curved and non-linear projections, and gives conditions under which the conclusion survives small perturbations.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No significant objection identified: the proof of Theorem 1.4 is internally sound under the stated admissibility hypotheses.","rationale":"The paper's central claim, Theorem 1.4, is a conditional statement: if a family satisfies the critical-hypersurface and tangential-immersion conditions, then it has the good witness property. The proof splits into a regular case (handled by a variant of Federer's theorem after a local bilipschitz change of variables) and a critical case (handled by an intrinsic argument on the hypersurface M). In the critical case, tangential immersion guarantees that, at each point of M, some (d-1)-tuple of the encoding's components has full rank on the tangent space. The finite collection of such index sets therefore covers M, so a positive-measure piece of E cap Sigma lies in a region where one H_I|M is a C^1 diffeomorphism. The image of that piece under H_I is a 1-rectifiable subset of R^{d-1} with positive H^1 measure, and Federer's theorem supplies a coordinate projection of it with positive length. That projection corresponds to one of the original phi_j, proving the witness. I checked the local compactness, measurability, and bilipschitz claims, and they are all justified by the cited standard theorems (Federer, Regular Level Set Theorem, Manifold Inverse Function Theorem). The radial projection example does not contradict the theorem; it merely shows the tangential-immersion condition is necessary for the critical-region argument. The structural exceptional-set results and stability theorems follow from the main theorem with straightforward product-set arguments. I therefore find no load-bearing gap in the proof, and the reader's ACCEPT verdict is consistent with my reading.","tokens_in":33099,"tokens_out":35693,"duration_ms":305166,"concrete_test":"Run a computational sanity check of the central theorem in d=2 with the admissible family H(x,y)=(x, y^3), whose critical set is M={y=0} and which satisfies tangential immersion on M. Let E be the union of the vertical segment {1} x [0,1] and the horizontal segment [0,1] x {0}. Compute H^1(phi_1(E)) and H^1(phi_2(E)) explicitly; verify that at least one is positive, exercising both the regular regime (vertical segment) and the critical regime (horizontal segment on M). Additionally, symbolically confirm that the sets U_{1} and U_{2} from the proof of Theorem 1.4 cover M, and that the chosen index does not depend on any hidden measurability assumption.","verdict_should_be":"UNCHANGED","load_bearing_attack":"I reviewed the proof of Theorem 1.4 in detail, focusing on the critical-set argument. The regular case (Corollary 4.3) correctly localizes to compact sets disjoint from Sigma and applies Federer's theorem via a locally bilipschitz encoding. The critical case is also sound: tangential immersion ensures that for each p in M, some (d-1)-index subset I has full rank on T_pM, so the relatively open sets U_I finitely cover M; hence a positive-measure part of E cap Sigma lies in some U_I. On a neighbourhood V_p where H_I|M is a local C^1 diffeomorphism (hence bilipschitz after shrinking), the image A = H_I(F) is a 1-rectifiable subset of R^{d-1} of positive H^1 measure. Federer's theorem in R^{d-1} then yields a coordinate projection with positive length, which is exactly phi_{i_l}(F) subset phi_{i_l}(E). No hidden assumption or gap appears in this chain. The supporting results (Theorem 4.2, Lemma 4.9, Theorem 4.10) were also checked: the block-determinant and kernel-transversality arguments are consistent, and the compact-localization steps are valid. The reader's identified weakest assumption, the tangential-immersion condition, is indeed the key hypothesis, but it is an explicit sufficient condition rather than an unsupported or circular step.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops a deterministic finite-witness principle for generalized curve projections at the 1-rectifiable endpoint. Given d C^2 scalar maps φ_1,...,φ_d, it forms the canonical encoding map H=(φ_1,...,φ_d) and studies its critical set Σ. If a positive-length part of E lies away from Σ, a local bilipschitz change of variables plus Federer's projection theorem gives a good witness. If a positive-length part of E lies in Σ, the paper assumes that Σ is contained in a C^1 hypersurface M on which H retains full rank d-1 when restricted to tangent spaces; the proof then restricts a suitable (d-1)-component submap H_I to M, obtains a local bilipschitz diffeomorphism into R^{d-1}, and applies Federer's theorem there. The paper introduces a fold non-degeneracy condition as a parameter-stability mechanism and applies the framework to pinned squared distances, planar radial projections (where the hypotheses fail on the critical line), nonlinear anisotropic distances, Bregman functionals generated by log-sum-exp regularizations of polyhedral norms, and families with arbitrarily prescribed C^2 critical graphs.","tokens_in":33376,"tokens_out":17106,"duration_ms":157327,"significance":"If the main theorem is correct, it is a clean deterministic complement to Peres-Schlag exceptional-parameter theory at the endpoint s=1, where dimension bounds degenerate. The proof of Theorem 1.4 is coherent and the supporting determinant computations in Sections 2, 5.1, 5.2, and 5.3 are explicit and checkable; there are no fitted parameters and no circularity. The radial-projection example in Section 3 is especially valuable because it shows that the tangential-immersion condition is not vacuous and identifies the precise obstruction. The fold non-degeneracy framework provides a verifiable stability mechanism, and the examples give a useful library of admissible and non-admissible families.","major_comments":[],"minor_comments":[{"comment":"The critical-hypersurface condition is stated as Σ_Φ⊆M, while several examples, such as Proposition 2.1, establish equality; please add a sentence clarifying that M may be strictly larger than Σ and that the proof of Theorem 1.4 only uses the inclusion E∩Σ⊆M.","section":"Definition 1.3"},{"comment":"In the proof of Theorem 4.10, the phrase 'component neighbourhoods G_x' near the finite-subcover step should read 'open neighbourhoods G_x'; the notation L=overline{W} is introduced after being used implicitly, so a short preamble would improve readability.","section":"Theorem 4.10"},{"comment":"In the final paragraph of the proof of Theorem 4.11, the statement that Lemma 4.9 gives a C^1 critical hypersurface is abbreviated; since Lemma 4.9 is formulated on an open set, the argument should explicitly invoke the neighbourhood W supplied by Theorem 4.10(2) and localize to it.","section":"Theorem 4.11"},{"comment":"The equivalence between positive length of the distance set Δ_p(E) and of the squared-distance set is stated in one sentence; a two-line countable-localization argument would remove any doubt, since t↦t^2 is only locally bilipschitz on (0,∞).","section":"Corollary 2.2"},{"comment":"In Proposition 5.3, the line 'recall that dim L_α=d−1, by the assumption of affine independence' should explicitly mention that affine independence gives linear independence of the differences ξ_1−ξ_2,...,ξ_1−ξ_d.","section":"Section 5.2.1"}],"recommendation":"minor_revision","confidential_remarks":"The manuscript is a good fit for a geometric measure theory or analysis journal. The main theorem is sound under its explicit admissibility hypotheses, and the examples are well chosen. The only reservation is editorial: the companion paper [3] is cited as an unpublished manuscript and [14], [16] as preprints; the editor may wish to confirm their availability. I see no issue with novelty or attribution."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Federer's theorem gives a deterministic finite-witness statement at the rectifiable endpoint: among d linearly independent orthogonal projections of a 1-rectifiable set of positive length, one has positive length. Marshall proves the nonlinear analogue for d scalar C^2 maps: if the critical set of the encoding map lies in a C^1 hypersurface and the encoding retains full rank tangentially, then one map in the family witnesses positive length for every positive-length rectifiable set. This is a real extension, not a repackaging. The proof is a clean combination of local bilipschitz changes of variables with Federer applied intrinsically on the critical hypersurface. I went through the critical-set argument in Theorem 1.4 and the supporting lemmas; it is sound. The fold non-degeneracy condition is well chosen: it implies admissibility and gives parameter-space stability, and the examples show it is sharp in the radial projection case where it fails. The pinned-distance, Bregman, and prescribed-C^2-graph examples are explicit and the determinant calculations check out.\n\nHonest caveats: the companion paper [3] is unpublished and overlapping in motivation, though the differential-geometric machinery appears to be new here. A few standard facts are used without proof, like the equivalence between positive length of a distance set and its squared version after compact localization; that is minor and fixable. Proposition 4.13's Area Formula argument is sketched rather than detailed, but it is routine. None of this touches the main theorem. I see no circularity and no hidden parameters.\n\nWho is it for? Geometric measure theorists working on projections and rectifiability, and anyone interested in deterministic finite-witness problems. It is a good, solid paper. I would accept it for peer review. Send it to a specialist.","headline":"A genuine nonlinear finite-witness generalization of Federer's projection theorem with careful proofs and useful examples; worth a serious referee.","tokens_in":33864,"tokens_out":2749,"would_cite":true,"duration_ms":26879,"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":"Every admissible family of d nonlinear scalar maps on R^d has a finite witness for positive-length rectifiable curves.","keywords":["good witness property","1-rectifiable sets","generalized curve projections","critical hypersurface","tangential immersion","fold non-degeneracy","exceptional parameter sets","pinned distance sets"],"falsifier":"To settle the claim, one could construct an admissible family—critical set a $C^1$ hypersurface with tangential rank $d-1$ throughout—together with a positive-length Borel 1-rectifiable set $E$ contained in that critical set, and compute $\\mathcal H^1(\\varphi_j(E))$ for each $j$. The theorem predicts at least one of these numbers is positive; if all were zero, the central claim would collapse. A concrete starting point is the prescribed-graph family $\\varphi_\\alpha^g(x',t)=\\langle\\alpha,x'\\rangle+(t-g(x'))^2$: taking $E$ to be a curve inside the graph $\\{t=g(x')\\}$ and checking whether the linear forms $\\langle\\alpha_j,x'\\rangle$ give positive length for some $j$ is a direct, finite-dimensional calculation.","tokens_in":32911,"feed_emoji":"📐","tokens_out":9855,"duration_ms":83073,"temperature":0.7,"pith_summary":"The paper proves a finite-witness theorem for curve projections at the borderline dimension where classical methods had only partial information. For any admissible family of $d$ twice-differentiable scalar maps $\\varphi_1,\\dots,\\varphi_d:\\mathbb R^d\\to\\mathbb R$, every Borel 1-rectifiable set $E\\subset\\mathbb R^d$ with positive length has at least one map $\\varphi_{j_E}$ whose image $\\varphi_{j_E}(E)$ still has positive length. The two admissibility conditions—the critical set of the encoding map lying inside a $C^1$ hypersurface, and the encoding map having full rank on that hypersurface—are exactly what lets the proof continue through the critical region where the inverse function theorem would break. The paper also introduces a fold non-degeneracy condition that makes the witness conclusion stable under small perturbations of the maps, and it verifies the hypotheses for pinned squared distances, anisotropic distances, Bregman functionals, and prescribed critical geometries.","feed_headline":"Every admissible curve-projection family has a finite witness","feed_subtitle":"New hypotheses let Federer's projection theorem pass through critical sets, settling the rectifiable endpoint.","key_machinery":"The canonical encoding map $H(x)=(\\varphi_1(x),\\dots,\\varphi_d(x))$ and its critical set $\\Sigma=\\{x:\\operatorname{rank} DH(x)\\le d-1\\}$ carry the argument. When $\\Sigma$ lies in a $C^1$ hypersurface $M$ and the restricted derivative $D_xH|_{T_xM}$ has rank $d-1$ everywhere on $M$, the restricted submap $H_I|_M$ for a suitable index set $I$ is a local bilipschitz diffeomorphism onto $\\mathbb R^{d-1}$, which reduces the critical part of $E$ to Federer's theorem one dimension down. The fold non-degeneracy condition $D_xJ(x)[v]\\ne0$ for every nonzero $v\\in\\ker D_xH(x)$ is the stability mechanism: it forces $\\Sigma$ to be a genuine hypersurface and its kernel direction to be transverse to $\\Sigma$, so the whole geometric configuration survives small perturbations of parameters.","core_discovery":"Theorem 1.4 is the central claim: if $\\Phi=\\{\\varphi_1,\\dots,\\varphi_d\\}$ is an admissible family of $C^2$ maps, then every Borel 1-rectifiable set $E$ with $\\mathcal H^1(E)>0$ admits a good witness. The proof splits into the regular region, where the encoding map $H=(\\varphi_1,\\dots,\\varphi_d)$ is locally bilipschitz and Federer's Projection Theorem applies directly, and the critical region, where the argument runs intrinsically along the $C^1$ hypersurface $M$ containing the critical set. There, the tangential-immersion hypothesis selects a subfamily of $d-1$ coordinates whose restriction to $M$ is locally bilipschitz, so Federer's theorem in $\\mathbb R^{d-1}$ produces the witness. The accompanying fold condition, which demands that the Jacobian determinant $J=\\det DH$ vanish transversely to the kernel direction, guarantees that this configuration persists under small parameter perturbations on compact sets.","pith_inferences":["A natural extrapolation, which the author signals as forthcoming work, is to extend the encoding-plus-hypersurface strategy to $k$-rectifiable sets for $k\\ge2$, replacing the $C^1$ hypersurface by a $(d-k)$-dimensional submanifold and requiring tangential rank $d-k$.","The affine-subspace bounds for bad parameters are stronger than dimension bounds; for parameter families where admissibility is open, one may expect similar thinness of exceptional sets in other projection problems.","Since fold non-degeneracy is often certified by a positive-definite Hessian, a testable extension is to verify the same two identities for other strictly convex potentials, beyond the log-sum-exp regularizations treated here.","The prescribed-$C^2$-graph example suggests the critical geometry itself is flexible; a plausible extension is to ask whether the conclusion persists for critical hypersurfaces that are only $C^1$ with Hölder tangent fields, though the proof as written needs only the bilipschitz parametrization."],"forward_implications":["For affinely independent pinned squared-distance maps, every positive-length rectifiable set has a pin whose distance set has positive length, and the set of bad pins is contained in an affine subspace of dimension at most $d-2$.","The planar radial angle pair fails the hypotheses precisely along the line through its two vantage points; positive-length sets lying on that line have no radial witness, while sets with positive length away from the line always do.","Under fold non-degeneracy, the good-witness property is open in parameter space on compact spatial regions: a successful index exists uniformly for all rectifiable sets in the region, with only the index depending on the set.","The same framework covers nonlinear anisotropic distances, Bregman functionals built from log-sum-exp regularizations of polyhedral norms, and families whose critical hypersurface is an arbitrary prescribed $C^2$ graph.","At the rectifiable endpoint the method gives deterministic finite witnesses, complementing exceptional-parameter dimension estimates that apply to fractal sets of Hausdorff dimension greater than one."],"supporting_citations":[{"why":"Supplies Federer's Projection Theorem and the bilipschitz reduction that carries the regular-region argument.","marker":"[8]"},{"why":"Provides the transversality framework for exceptional-parameter estimates above the rectifiable endpoint and the comparison baseline.","marker":"[20]"},{"why":"Documents the radial-projection failure for rectifiable sets and motivates the tangential-immersion condition.","marker":"[19]"},{"why":"Companion manuscript sharing the finite encoding principle, developed in tandem with this paper.","marker":"[3]"},{"why":"Supplies the standard fold-singularity condition used in Definition 1.5.","marker":"[10]"},{"why":"Gives a complementary quantified two-projection theorem for nonlinear projection families.","marker":"[16]"},{"why":"Context for nonlinear projection families and Favard-length estimates that the examples build on.","marker":"[4]"},{"why":"Provides a comparison principle relating generalized projections to orthogonal projections.","marker":"[14]"}],"fun_headline_variants":["Finite witnesses for curve projections at rectifiable endpoint","Generalized projections admit finite witnesses even at critical sets","Fold condition stabilizes witnesses for curve projections","Federer's theorem extended to nonlinear critical hypersurfaces"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is the tangential-immersion condition: on the $C^1$ hypersurface containing the critical set, the restricted derivative $D_xH|_{T_xM}$ must have full rank $d-1$ at every point. If that rank drops, the intrinsic argument on $M$ fails, and the planar radial-projection example shows that the obstruction is real, not a technicality.","fun_headline_variants_meta":{"raw":{"variants":["Finite witnesses for curve projections at rectifiable endpoint","Generalized projections admit finite witnesses even at critical sets","Fold condition stabilizes witnesses for curve projections","Federer's theorem extended to nonlinear critical hypersurfaces"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000307,"raw_usage":{"total_tokens":1827,"prompt_tokens":1085,"completion_tokens":742,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":701,"completion_tokens_details":{"reasoning_tokens":679}},"tokens_in":701,"tokens_out":742,"duration_ms":6434,"temperature":1.0,"reasoning_tokens":679,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T14:21:30.737913+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"To settle the claim, one could construct an admissible family—critical set a $C^1$ hypersurface with tangential rank $d-1$ throughout—together with a positive-length Borel 1-rectifiable set $E$ contained in that critical set, and compute $\\mathcal H^1(\\varphi_j(E))$ for each $j$. The theorem predicts at least one of these numbers is positive; if all were zero, the central claim would collapse. A concrete starting point is the prescribed-graph family $\\varphi_\\alpha^g(x',t)=\\langle\\alpha,x'\\rangle+(t-g(x'))^2$: taking $E$ to be a curve inside the graph $\\{t=g(x')\\}$ and checking whether the linear forms $\\langle\\alpha_j,x'\\rangle$ give positive length for some $j$ is a direct, finite-dimensional calculation.","supporting_citations":[{"cited_title":"Federer.Geometric Measure Theory","cited_arxiv_id":null,"evidence_quote":"Supplies Federer's Projection Theorem and the bilipschitz reduction that carries the regular-region argument."},{"cited_title":"Peres and W","cited_arxiv_id":null,"evidence_quote":"Provides the transversality framework for exceptional-parameter estimates above the rectifiable endpoint and the comparison baseline."},{"cited_title":"Bongers, P","cited_arxiv_id":null,"evidence_quote":"Companion manuscript sharing the finite encoding principle, developed in tandem with this paper."},{"cited_title":"Golubitsky and V","cited_arxiv_id":null,"evidence_quote":"Supplies the standard fold-singularity condition used in Definition 1.5."},{"cited_title":"Bongers and K","cited_arxiv_id":null,"evidence_quote":"Context for nonlinear projection families and Favard-length estimates that the examples build on."}],"review_version":1}