{"id":"992c0e3c-63fa-40c1-99ab-efba96fceba6","arxiv_id":"2506.01397","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For two surfaces glued along a curve, the paper classifies the gluing as cylindrical, conical, cuspidal-edgy, or swallowtailed according to explicit conditions on the frame invariants of the two surfaces and the angle between their normals.","lead":"A differential geometry paper defines auxiliary developable surfaces along the curve where two surfaces are glued, and gives formulas for when these surfaces are cylinders, cones, or have cusp-edge or swallowtail singularities. A smart generalist might read it because it offers a local classification of surface gluings, with potential uses in discrete geometry and computer graphics.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No significant objection identified to the central classification; the l=0 frame ambiguity reduces to a global sign and leaves the invariants unchanged. The example errors are real but do not affect the theorem.","rationale":"The reader's conditional verdict is reasonable, but the specific weakest assumption named by the reader is not the main source of risk. The smooth unit vector field e is not arbitrary: away from l=0 it must be ± gamma'/|gamma'|, and continuity across a zero of l fixes it up to one global sign. A direct computation shows beta_nu, rho_nu, beta_b and rho_b are invariant under this sign flip, so the singularity classification at l=0 does not change. The central results are internal substitutions of Lemma 4.2 into Section 3; my re-derivation of Corollary 4.9 reproduces the paper's conditions exactly. The real soft spot is that the paper's own examples do not instantiate the hypotheses: Example 5.6 uses a vector nu_1 that is not normal to f_1, and the stated theta does not satisfy the rotation formula it claims to illustrate. This is a serious manuscript defect and explains the need for revision, but it does not falsify the classification theorem. Hence the verdict should remain conditional pending corrected examples and completed proof details, not be moved to accept or reject.","tokens_in":15998,"tokens_out":36367,"duration_ms":385600,"concrete_test":"Recompute Example 5.6 using the correct unit normal nu_1=(-1,0,0) for f_1=(0,4u^3+v,3u^4) and verify the frontal condition d f_1(X)·nu_1=0. Then recompute (kappa_11,kappa_12,kappa_13) and (kappa_21,kappa_22,kappa_23) from the frame and check whether any choice of theta satisfies (4.2); if no theta works, the example is invalid and must be replaced. Independently, symbolically substitute (4.2) into Corollary 3.3 to confirm the two displayed swallowtail branches of Corollary 4.9.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is that Theorems 4.5 and 4.8, together with Corollary 4.9, classify special gluings via the invariants (kappa_i1,kappa_i2,kappa_i3,theta). I re-derived Corollary 4.9 by substituting Lemma 4.2's rotation formulas into Corollary 3.3; the two displayed conditions match exactly, including the term (kappa_12 cosθ - kappa_11 sinθ)(kappa_13 + 4θ') + 3(kappa_11' cosθ + kappa_12' sinθ). The alleged non-uniqueness of e at l=0 does not undermine this: if a smooth unit vector field e exists with gamma'=l e, then for every t with l(t)≠0 the vector e is forced to be ± gamma'/|gamma'|, and continuity determines e up to one global sign. Under e→-e, l→-l, kappa_1→-kappa_1, kappa_3→-kappa_3, kappa_2→kappa_2, and theta→-theta, the quantities beta_nu i, rho_nu i and their S_b analogues are invariant, so the cuspidal-edge and swallowtail conditions in Corollaries 3.3, 3.6 and 4.9 are unchanged. Thus the classification is well-posed despite the pointwise ambiguity. The genuine problems are in the examples: in Example 5.6, nu_1=(0,0,1) is not a unit normal of f_1=(0,4u^3+v,3u^4), since d f_1(d/du)=(0,12u^2,12u^3) has nonzero dot product with nu_1; the stated theta also fails to satisfy the rotation law (4.2). Similar normalization errors appear elsewhere. These invalidate the illustrations but are not load-bearing for the theorem, whose proof is a direct substitution into the Section 3 structure.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the local geometry of two surfaces glued along a curve. For each surface, a moving frame {e, ν_i, b_i} is introduced along the gluing locus, and two developable surfaces S_{ν_i} and S_{b_i} are associated with the frame. The main results give explicit criteria, in terms of the frame invariants κ_{i1}, κ_{i2}, κ_{i3} and the angle θ between the two normal vector fields, for these developable surfaces to be cylinders, cones, cuspidal edges, or swallowtails (Theorems 3.1, 3.2, 3.4, 3.5 and the gluing versions Theorems 4.5 and 4.8, with Corollaries 3.3, 3.6, and 4.9). The paper then presents five examples intended to illustrate the classification.","tokens_in":16363,"tokens_out":16588,"duration_ms":140969,"significance":"If the main theorems are correct, the paper provides a complete and explicit local classification of 'special gluings' in terms of standard differential invariants; the formulas for β and ρ are explicit, and no parameters are fitted. The proof strategy is direct and largely computational, and the rotation formula (4.2) correctly links the second surface's invariants to the first. However, the conceptual novelty is modest: the results are a direct translation of the known developable-surface theory of Izumiya-Otani into the gluing setting, and several of the illustrative examples contain mathematical errors that must be repaired before the paper can be accepted.","major_comments":[{"comment":"The example is internally inconsistent. The vector ν_1=(0,0,1) is not a unit normal of f_1(u,v)=(0,4u^3+v,3u^4), because ∂_u f_1·ν_1 = 12u^3, which is not identically zero. Moreover, the stated angle θ satisfies sinθ(0)=1 and cosθ(0)=0, so θ(0)=π/2, while ν_1(0)=(0,0,1) and ν_2(0)=(0,0,-1) are antiparallel. Consequently equation (4.2) is violated at u=0: the listed κ_{21}(0) is -1, whereas κ_{11}(0)cosθ(0)+κ_{12}(0)sinθ(0) = 1·0+0·1 = 0. The claimed verification of the swallowtail condition (κ_{11}cosθ+κ_{12}sinθ)l''|_0 = -24 is also inconsistent with the displayed formulas, since the expression is 0. Thus Example 5.6 does not illustrate Corollary 4.9 and must be corrected or replaced.","section":"Section 5, Example 5.6"},{"comment":"The vector ν_1=(cos u, sin u, 1) is not a unit normal of f_1(u,v)=(cos u, sin u, v), because ∂_v f_1·ν_1 = 1 ≠ 0. The invariants (κ_{11},κ_{12},κ_{13})=(-1,0,0) given in the example are exactly those computed from the unit normal (cos u, sin u, 0), so the example as written is inconsistent. The normal vector should be corrected, and the subsequent verification of β_{ν1}=0 should be redone with the corrected normal.","section":"Section 5, Example 5.2"}],"minor_comments":[{"comment":"At a point where l(t)=0, the unit vector e is not uniquely determined by the condition γ'=le. The paper should state explicitly that a smooth unit vector field e is chosen on the whole interval I and that the classification is invariant under the global sign change (e,l,κ_1,κ_3,θ) → (-e,-l,-κ_1,-κ_3,-θ), under which β_ν, ρ_ν, β_b, ρ_b and the conditions in the corollaries are unchanged. Without this remark, the well-posedness of the invariants at isolated zeros of l is not manifest.","section":"Section 2.2 and Corollaries 3.3, 3.6, 4.9"},{"comment":"There are small typos in the examples: in Example 5.2, 'cosu.sinu' should be 'cosu, sinu', and in Example 5.4 the symbol 'a+' appears to be a typo for 'v+'.","section":"Section 5, Examples 5.2 and 5.4"},{"comment":"The notation overloads γ: it is first introduced as a curve in the parameter domain U and then used for its image f∘γ on the surface. This makes Definition 4.1 and the subsequent formulas harder to parse; using a separate symbol for the image curve would improve readability.","section":"Section 2.2"}],"recommendation":"major_revision","confidential_remarks":"The central classification theorems appear sound and the proof is essentially a direct computation, but the errors in Examples 5.2 and 5.6 are more than typographical: they involve incorrect unit normal vectors and a violation of equation (4.2). Since these examples are presented as demonstrations of the classification, the manuscript needs substantive correction before publication. The reference list and the absence of fitted parameters are strengths."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Core take: the classification of special gluings via the frame invariants is right. The reported problems in the examples are real but do not affect the theorems.\n\nWhat's actually new is Section 4: setting up the gluing locus with two frames, the rotation formula (4.2) relating the two normal frames, and the definitions of Sνi- and Sbi-cylindrical/conical/cuspidal-edgy/swallowtailed gluings. Theorems 4.5 and 4.8/4.9 are genuine extensions of Izumiya-Otani's flat approximation to the gluing context, and the criteria in terms of βνi and ρνi are computable from the local frame data. The attribution to [7] is honest, and the Section 3 machinery is reproduced carefully.\n\nThe soft spots are in the examples. Example 5.6 is internally inconsistent: ν1=(0,0,1) is not a unit normal to f1=(0,4u^3+v,3u^4) because ∂_u f1·ν1 = 12u^3, which is nonzero away from u=0. The proposed θ also fails to satisfy the rotation law (4.2). Similar normalization slips appear in other examples. These do not invalidate the central derivation; the theorems are direct substitutions from Section 3. But the illustrations should be treated as unreliable until recomputed. There is also a typo in the proof of Theorem 3.4 where s'(t) uses βν instead of βb, and Theorem 4.8/Corollary 4.9 are stated without showing the substitution; that missing detail should be added.\n\nThe l=0 frame ambiguity that might worry a reader turns out to be a non-issue: if a smooth unit e exists with γ'=le, then e is pointwise determined up to sign, and the quantities βνi and ρνi are invariant under the induced sign changes. So the classification is well-posed.\n\nWho this is for: people working on discrete surface gluing, curved foldings, or developable surfaces along curves. It does not solve a major open problem, but it gives a clean local taxonomy that could be useful.\n\nRecommendation: send to peer review. The central math holds up and the extension is legitimate. The referee report should demand corrected examples and the missing proof details. A serious referee will find this fixable.","headline":"The core classification is correct, but the example data are unreliable and must be fixed before the paper is usable as an illustration.","tokens_in":16912,"tokens_out":2389,"would_cite":true,"duration_ms":24666,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["53A05","53A55","58K30"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper claims that the geometry of a gluing of two surfaces along a curve is fully encoded by three frame invariants and the angle between the two surface normals, with cylinder, cone, cuspidal-edge, and swallowtail cases…","keywords":["gluing of surfaces","developable surfaces","moving frame","gluing locus","frontals","cuspidal edge","swallowtail","cylinder and cone"],"falsifier":"At a point where l(t0) = 0 (for example the cusp of gamma(t) = ($t^{2}$, $t^{3}$, 0)), take two different smooth unit vector fields e1 and e2 that both satisfy gamma' = l e, compute the resulting kappa_i1, kappa_i2, kappa_i3 and hence beta_nu and rho_nu for S_nu1, and check whether the predicted singularity type at t0 changes. If it does, the classification of special gluings at singular points is not intrinsic; if it never does, the choice of e is immaterial.","tokens_in":15754,"feed_emoji":"📐","tokens_out":6403,"duration_ms":52152,"temperature":0.7,"pith_summary":"The paper studies two surfaces glued along a common curve, the gluing locus. A moving frame is attached to the curve using the two surface normals, and from this frame two developable surfaces are constructed for each side. The central claim is that the local geometry of the gluing is completely controlled by three frame invariants and the angle between the normals: the auxiliary developable surfaces are cylinders, cones, cuspidal edges, or swallowtails exactly when explicit algebraic conditions on those invariants hold. If this is right, special gluings can be detected purely from frame data, which matters for gluing surfaces in discrete and singular geometry.","feed_headline":"Two surfaces glued along a curve: their type is read from three invariants","feed_subtitle":"A gluing's cylinder, cone, cuspidal, or swallowtail nature is encoded in frame curvatures and the angle between normals.","key_machinery":"The key machinery is the Frenet–Serret type frame {e, nu_i, b_i} attached to the gluing locus gamma, with gamma' = l e, and the two developable surfaces S_nu_i and S_b_i defined as envelopes of planes orthogonal to nu_i and b_i. Their cylinder and cone conditions are captured by the functions beta_nu_i and rho_nu_i, and their singularities by the same rho_nu_i together with its derivative. Lemma 4.2 shows how the second frame's invariants relate to the first via the angle theta, so the entire classification reduces to four scalar data on one side of the gluing.","core_discovery":"The central claim is Theorem 4.5 and Theorem 4.8: for each side i=1,2, the developable surface S_nu_i is a cylinder if and only if beta_nu_i is identically zero, a cone if and only if beta_nu_i is nonzero and rho_nu_i is identically zero, and at a striction point it has a cuspidal edge if and only if beta_nu_i is nonzero and rho_nu_i is nonzero, and a swallowtail if and only if beta_nu_i is nonzero, rho_nu_i is zero, and rho'_nu_i is nonzero. Here beta_nu_i and rho_nu_i are built from the frame invariants kappa_i1, kappa_i2, kappa_i3. Lemma 4.2 shows that the second frame's invariants are obtained from the first by rotating the normal plane by the angle theta, so all four developable surfaces S_nu1, S_b1, S_nu2, S_b2 are governed by three invariants plus theta.","pith_inferences":["Because the frame depends on the choice of the unit vector field e wherever l = 0, the classification at singular points of the gluing locus is well-posed only after a definite choice is fixed; an intrinsic reformulation would need to show the verdicts are independent of that choice or to quotient out this freedom.","The conditions are numerically testable: for any pair of fronts glued along a curve, one can compute kappa_i1, kappa_i2, kappa_i3 and theta, predict whether each S_nu_i or S_b_i is a cylinder, cone, cuspidal edge, or swallowtail, and compare directly with the computed surface.","The same frame-and-envelope construction may extend to gluings of higher-codimension submanifolds or to other ambient spaces, where the role of developable surfaces is played by envelopes of families of hyperplanes."],"forward_implications":["If theta = k pi/2, then S_nu2 is a cylinder (or cone) exactly when S_b1 is a cylinder (or cone), so the two sides of the gluing exchange roles under a quarter-turn of the normals.","If theta = k pi, then S_nu2 and S_nu1 have the same cylinder/cone status, so the gluing is symmetric under swapping the two normals.","At a point where the gluing locus is singular (l = 0), the cuspidal-edge and swallowtail conditions simplify to conditions involving l', l'', and the rotated invariants, as in Corollary 4.9.","The four developable surfaces are not independent: S_b1 is obtained from S_nu1 by rotating the rulings by pi/2, and S_nu2 is obtained from S_nu1 by rotating by theta, so any special gluing is classified by the same three invariants plus theta."],"supporting_citations":[{"why":"Supplies the construction of developable surfaces as envelopes of planes orthogonal to the frame vectors, and the beta/rho invariant method used throughout.","marker":"[7]"},{"why":"Provides the singularity criteria (via singularity identifier and null vector field) used to characterize cuspidal edges and swallowtails.","marker":"[11]"},{"why":"Provides the definitions and framework for frontals and their singularities used in the paper.","marker":"[12]"},{"why":"Background on ruled fronts and developable surfaces that underlies the classification of cylinders and cones.","marker":"[6]"},{"why":"Background on singularities of ruled surfaces, used for the striction-curve and singular-point arguments.","marker":"[10]"}],"fun_headline_variants":["Gluing surfaces: three invariants classify the developable type","Curve gluing classification: frame invariants decide cone, cusp, or more","Three frame invariants tell if a surface gluing is a cylinder or swallowtail","Surface gluing: beta and rho invariants encode the singularity type","Gluing two surfaces: type is decided by three invariants plus angle"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The classification requires a smooth unit vector field e along the gluing locus with gamma' = l e; at points where l = 0 the vector e is not uniquely determined, so the invariants (and the verdicts based on them) can change unless a specific choice is adopted, and the nondegeneracy conditions (kappa_i1, kappa_i3) not both zero and (kappa_i2, kappa_i3) not both zero must also hold.","fun_headline_variants_meta":{"raw":{"variants":["Gluing surfaces: three invariants classify the developable type","Curve gluing classification: frame invariants decide cone, cusp, or more","Three frame invariants tell if a surface gluing is a cylinder or swallowtail","Surface gluing: beta and rho invariants encode the singularity type","Gluing two surfaces: type is decided by three invariants plus angle"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000817,"raw_usage":{"total_tokens":3513,"prompt_tokens":817,"completion_tokens":2696,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":433,"completion_tokens_details":{"reasoning_tokens":2595}},"tokens_in":433,"tokens_out":2696,"duration_ms":16596,"temperature":1.0,"reasoning_tokens":2595,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T11:45:14.931106+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"At a point where l(t0) = 0 (for example the cusp of gamma(t) = ($t^{2}$, $t^{3}$, 0)), take two different smooth unit vector fields e1 and e2 that both satisfy gamma' = l e, compute the resulting kappa_i1, kappa_i2, kappa_i3 and hence beta_nu and rho_nu for S_nu1, and check whether the predicted singularity type at t0 changes. If it does, the classification of special gluings at singular points is not intrinsic; if it never does, the choice of e is immaterial.","supporting_citations":[{"cited_title":"Umehara, K","cited_arxiv_id":null,"evidence_quote":"Provides the definitions and framework for frontals and their singularities used in the paper."},{"cited_title":"Izumiya and S","cited_arxiv_id":null,"evidence_quote":"Supplies the construction of developable surfaces as envelopes of planes orthogonal to the frame vectors, and the beta/rho invariant method used throughout."},{"cited_title":"Kokubu, W","cited_arxiv_id":null,"evidence_quote":"Provides the singularity criteria (via singularity identifier and null vector field) used to characterize cuspidal edges and swallowtails."},{"cited_title":"Izumiya, Ruled fronts and developable surfaces.Publicationes Mathematicae Debrecen,61(2002), 139–144","cited_arxiv_id":null,"evidence_quote":"Background on ruled fronts and developable surfaces that underlies the classification of cylinders and cones."},{"cited_title":"Izumiya and N","cited_arxiv_id":null,"evidence_quote":"Background on singularities of ruled surfaces, used for the striction-curve and singular-point arguments."}],"review_version":1}