{"id":"b034e808-6d8b-45be-80a8-e1ab354b0ad6","arxiv_id":"1908.07867","paper_version":3,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For generic parabolic surfaces in R^3 modulo the special affine group, the algebra of differential invariants is generated through invariant differentiation by the fourth-order invariant W and one new fifth-order invariant M.","lead":"This paper studies surfaces in 3D space whose curvature-like Hessian has rank one (parabolic surfaces) and classifies them up to volume-preserving affine transformations. It proves that in the main generic case, all differential invariants can be generated from two basic ones, the known fourth-order invariant W and a new fifth-order invariant M.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The pulled-back Fels–Olver recurrence is used on PJ^n without proving freeness or a genuine cross-section at the jet order of the phantom invariants; the Cramer solvability in Section 20.1 is asserted, not established.","rationale":"The reader identified the validity of the pulled-back Fels–Olver recurrence on the non-free parabolic jet bundle as the weakest assumption. I agree, and I refined the concern: the paper does not prove that the action is locally free on the higher-order parabolic jet space where the phantom invariants live, nor that the phantom equations form a genuine transversal. The Cramer systems in Section 20.1 are asserted to be uniquely solvable, but the required determinant is not analyzed. My own symbolic evaluation of the displayed 6x6 matrix suggests a determinant proportional to W(M+3), which would vanish on a subbranch not excluded by the hypotheses; this makes the concern concrete rather than a general worry about missing formalism. The proposed test, a direct rank and determinant computation at a generic normalized jet, would settle whether the recurrence is justified. Since the concern does not amount to a demonstrated contradiction, but rather an unproven load-bearing assumption, the appropriate verdict remains CONDITIONAL, matching the reader's assessment, so no change is needed.","tokens_in":81121,"tokens_out":11122,"duration_ms":106925,"concrete_test":"At a generic normalized jet in the branch S≠0, W≠0 (e.g., with the phantom values I2,0=1, I1,1=0, I3,0=0, I2,1=1, I4,0=0, I4,1=0, and generic W,M), compute: (1) the rank of the eleven pushed-forward prolonged generators pj_*(v_1^(5)), ..., pj_*(v_6^(5)), pj_*(w_1^(5)), pj_*(w_2^(5)), pj_*(w_3^(5)) on PJ^5; and (2) the determinant of the 6x6 phantom coefficient matrix used in the D1 and D2 recurrence systems of Section 20.1, as a function of W and M. If the rank equals 11 and the determinant is nonzero for all W≠0 (including M=-3), the pulled-back Fels–Olver recurrence is valid on the branch and the concern is resolved. If the rank is less than 11 or the determinant vanishes on an open subbranch, the recurrence proof and hence the generation theorem require a new argument.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central generation claims (Theorems 2.12 and 2.13) rest on the recurrence formulas of Section 20, which are pulled back from Fels–Olver theory to the parabolic jet bundles PJ^n. The paper itself flags the obstruction in Question 4.14: on PJ^4, the SA3(R)-orbits have dimension 10 instead of 11, so no moving frame exists (Section 17.15, Assertion 17.18). The response in Section 20 is to introduce six phantom invariants, including I4,1 of order 5, and to solve a 6x6 Cramer system for the Maurer–Cartan terms K^σ_j. This only works if the action on the relevant higher-order parabolic jet space is locally free, i.e., if the eleven pushed-forward prolonged generators have rank 11 on PJ^5 (or some PJ^n), and if the phantom equations define a genuine transversal. Neither condition is proved. The displayed 6x6 matrix in Section 20.1 is stated to have a unique solution, but its determinant is not computed; a direct evaluation at the normalized phantom values gives a determinant proportional to W(M+3), which vanishes on the subbranch M=-3, W≠0. That subbranch is not excluded by the assumptions. If the determinant can vanish or if the rank of the prolonged action on PJ^5 is below 11, then the recurrence formulas and the induction in Proposition 20.2 lose their justification, and the generation theorem does not follow from the arguments given. The paper's own Section 21 verifies a few low-order recurrences with explicit D1,D2, but this does not settle the infinite induction or the Cramer solvability at all orders.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper studies the algebra of differential invariants of parabolic surfaces S^2 \\subset R^3 under the special affine group SA3(R). After a branching analysis based on relative invariants, it claims that in the main branch S \\neq 0, W \\neq 0, the full algebra of differential invariants is generated, through invariant differentiations, by the fourth-order invariant W and a single fifth-order invariant M whose explicit expression has 57 differential monomials; analogous generation statements are given in the branches S=0 and S \\neq 0, W=0. The proof combines power-series normalization, rank computations on parabolic jet bundles, and Fels-Olver recurrence formulas pulled back to the parabolic jet spaces PJ^n, with explicit invariant differentiation operators D1,D2 in Section 21. The paper also gives invariant-theoretic characterizations of cylinders, cones, and tangential surfaces, and classifies special affinely homogeneous parabolic surfaces.","tokens_in":81444,"tokens_out":24242,"duration_ms":204896,"significance":"If the generation theorems are correct, this is a substantial contribution to the computational theory of differential invariants: it provides explicit generators, normal forms, and invariant differential operators for a non-free group action on a constrained jet submanifold, going beyond the standard moving-frame framework. The explicit expressions for W, M, X, Y and the verification of D1,D2 are valuable and machine-checkable data. The paper also cleanly characterizes cylinders, cones, and tangential surfaces in terms of differential invariants. The main novelty, the use of recurrence formulas pulled back to the parabolic jet bundles, is also the main source of risk, and the proof as written does not fully justify that pullback.","major_comments":[{"comment":"The generation theorems (Theorems 2.12 and 2.13) rest on the recurrence formulas of Section 20, which are pulled back from Fels-Olver theory (Theorem 14.6) to the parabolic jet bundles PJ^n. The paper itself flags in Question 4.14 that on PJ^4 the SA3(R)-orbits have dimension 10 < 11, so no moving frame exists there. Section 17.15 and Assertion 17.18 only establish that the relevant matrix has rank 5 on PJ^4, and Section 20 replaces the missing moving frame by six phantom invariants and Cramer systems. A general proof that the pulled-back recurrence remains valid on PJ^n is not supplied; in particular, local freeness on PJ^5 and the transversality of the phantom equations are asserted rather than established. Section 21 verifies only finitely many low-order recurrences. This gap is load-bearing for Theorem 2.13 and for the homogeneous-model conclusions in Section 23. The authors should either prove that the action is locally free on PJ^5 and that the normalization defines a genuine cross-section (the power-series normal form in Section 18 already shows the stabilizer is reduced to the identity at order 5, so this is plausible), or give a direct derivation of the recurrence on the parabolic submanifold.","section":"20, 17.15, 4.14"},{"comment":"Propositions 20.2 and 20.4 state that the generation result follows by 'an elementary induction' without displaying the general recurrence matrices for arbitrary order. Since the entire generation theorem depends on this induction, the authors should spell out the induction step: for each order n >= 6, the recurrences for D1 I_{n,0} and D2 I_{n,0} express I_{n+1,0} and I_{n,1} in terms of known lower-order invariants and the Maurer-Cartan terms. The compressed statement is acceptable only after the recurrence validity from the previous comment is established.","section":"20.2, 20.4"}],"minor_comments":[{"comment":"The displayed 6x6 Cramer matrix in Section 20.1 is stated to have a unique solution without presenting its determinant. Direct expansion gives det = 108 W, so the system is nonsingular precisely on the branch W \\neq 0; the matrix is not singular at M = -3. The authors should report the determinant or an equivalent nondegeneracy argument, because the displayed solution divides by W and the branch assumption is essential.","section":"20.1"},{"comment":"In the proof of Proposition 20.4, the text says D2X = 5X, but the displayed computation immediately before gives D2X = 3X. The earlier value is the one used in the homogeneous-model argument in Section 23, so the proof should be corrected.","section":"20.3"},{"comment":"The branching diagrams before Theorems 2.7, 2.11, and 2.12 are hard to parse when printed; a table or explicitly labeled tree with node names S, P, C, W, X, Y, M would improve readability.","section":"2"},{"comment":"The notation S is used both for the relative invariant (Fxx Fxxy - Fxy Fxxx)/Fxx^2 and for the transformation parameter s in Section 7; this is occasionally confusing and should be flagged or renamed.","section":"17.10"}],"recommendation":"major_revision","confidential_remarks":"The paper is computationally substantial and the main claims are likely correct, but the missing proof of the pulled-back recurrence on the parabolic jet bundles is a genuine gap in the proof of the generation theorems. The stress-test claim that the 6x6 matrix in Section 20.1 has determinant proportional to W(M+3) is not supported by the manuscript as printed; direct expansion gives 108W, so the Cramer systems are nonsingular on the assumed branch. I recommend major revision rather than rejection because the missing proof appears repairable within the manuscript's scope: the power-series normal form already suggests local freeness at order 5, and the explicit D1,D2 checks in Section 21 provide strong evidence for the recurrence structure."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Chen and Merker finish a real gap: on the S≠0, W≠0 parabolic branch under SA3(R), the full differential invariant algebra is generated by W and the new fifth-order M, and every surface has a normal form with Taylor coefficients determined by W, M, and their invariant derivatives. The explicit M with 57 monomials is a serious computation, and Section 21's explicit D1, D2 gives an independent check of low-order recurrences. The branch analysis also cleanly separates cylinders (S=0), cones (W=0), and tangential surfaces (W≠0). That is the core value.\n\nThe soft spots are real but manageable. The generation theorems (2.12, 2.13) rely on Fels-Olver recurrence pulled back to parabolic jet bundles. The paper itself flags that PJ4 has orbit dimension 10 not 11, so no moving frame exists there; Section 20.1 substitutes six phantom invariants and solves a 6x6 Cramer system. That system is nonsingular on the W≠0 branch — I get det = -108W, not something vanishing on M=-3 — so the specific worry in the stress-test note about M=-3 is unfounded. But the general concern remains: the rank condition that makes the pulled-back recurrence valid at higher orders is asserted rather than proved. A referee should ask for a lemma showing the prolonged action has rank 11 on PJ5 or PJn, or a direct derivation of the recurrence on parabolic jets. Propositions 20.2 and 20.4 also compress the induction into 'an elementary induction' without showing it. That is probably fixable, but it is the heart of generation, so it should be displayed. The computer algebra has no scripts, but the paper's explicit expressions are checkable; minor.\n\nThe citation to Pocchiola for W is slightly loose in the body, and the paper's own earlier affine rigidity [26] is used without restating, but neither affects the main claim. Overall the central argument is coherent and the explicit invariants are reproducible in principle. This deserves a serious referee and likely acceptance after the recurrence justification is tightened.","headline":"Genuine branch completion with a new fifth-order invariant, but the generation proof leans on a pulled-back recurrence whose validity is asserted more than proved.","tokens_in":81975,"tokens_out":4200,"would_cite":true,"duration_ms":41651,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["53A15","53A55","58A20"],"pacs":[],"model":"deepseek-v4-flash","headline":"For generic parabolic surfaces in 3-space, the full algebra of differential invariants under the special affine group is generated by one fourth-order invariant W and one fifth-order invariant M together with their invariant derivatives.","keywords":["differential invariants","parabolic surfaces","special affine group","jet spaces","recurrence formulas","moving frames","normal forms","developable surfaces"],"falsifier":"On the explicit normalized surface of the main branch, compute the sixth-order invariants $I_{6,0}$ and $I_{5,1}$ independently by power-series normalization and check whether the recurrence-derived identities $D_1 I_{6,0} = I_{7,0} - \\frac{3}{2}W(7M-2I_{5,1})(4M-I_{5,1}) + \\frac{4}{3}W I_{6,0}$ and $D_2 I_{6,0} = I_{6,1} - I_{6,0} + 21WM - 8WI_{5,1}$ hold identically; if any term is missing, the pulled-back recurrence formula is invalid. Equally decisive, any sixth- or higher-order differential invariant not expressible as a rational function of $W$, $M$ and their invariant derivatives would disprove the generation theorem.","tokens_in":80845,"feed_emoji":"📐","tokens_out":12406,"duration_ms":600153,"temperature":0.7,"pith_summary":"This paper determines the differential invariants of generic parabolic surfaces—surfaces whose Hessian matrix has rank one everywhere—under the volume-preserving affine group $\\mathrm{SA}_3(\\mathbb{R})$. It proves that in the principal branch, every such invariant is a rational function of just two basic invariants, $W$ of order 4 and $M$ of order 5, together with their invariant derivatives. Since these invariants are the data that decide when two surfaces are equivalent, the result provides a complete local normal form: the independent Taylor coefficients of a normalized surface are exactly the values of $W$, $M$, and their invariant derivatives. The achievement is that this resolves a branch of the classical equivalence problem where the standard recurrence formulas had not previously been pushed through, because the group action is not free on the relevant parabolic jet bundle.","feed_headline":"Two invariants generate all symmetry classes of parabolic surfaces","feed_subtitle":"An order-4 invariant W and an order-5 invariant M give a complete local normal form under the special affine group.","key_machinery":"The load-bearing mechanism is the pulled-back recurrence formulas of moving-frame theory, which express the invariant derivative of any invariantized jet monomial as the next-order monomial plus correction terms involving Maurer–Cartan invariants; these formulas determine all higher invariants from low-order ones without computing a moving frame explicitly. On the parabolic jet bundle $\\mathrm{PJ}^n$ (the space of all Taylor coefficients of parabolic graphs up to order $n$), the $\\mathrm{SA}_3(\\mathbb{R})$-orbits on $\\mathrm{PJ}^4$ have dimension 10 rather than 11, so no moving frame exists; the paper substitutes rank computations and Cramér systems, solving for the Maurer–Cartan terms from the six phantom invariants $I_{2,0}=1$, $I_{1,1}=0$, $I_{3,0}=0$, $I_{2,1}=1$, $I_{4,0}=0$, $I_{4,1}=0$. The resulting invariant differentiation operators $D_1,D_2$ are made explicit and checked against independently computed invariants, closing the logical loop.","core_discovery":"The central discovery is that the algebra of differential invariants for $\\mathrm{SA}_3(\\mathbb{R})$-equivalence of parabolic surfaces is finitely generated in each branch. In the main branch $S\\neq 0$, $W\\neq 0$, the full algebra is generated by $W$ and $M$ together with their invariant derivatives; $W$ is the unique fourth-order invariant, $M$ is the unique fifth-order invariant, with an explicit expression containing 57 differential monomials, and $M$ cannot be obtained from $W$ by invariant differentiation. Equivalently, every surface in this branch has a unique normal form $u = \\frac{x^2}{2} + \\frac{x^2 y}{2} + \\frac{F_{3,1} x^3 y}{6} + \\frac{x^2 y^2}{2} + \\frac{F_{5,0} x^5}{120} + \\cdots$, with $F_{3,1}=W$, $F_{5,0}=M$, and two surfaces are equivalent exactly when the remaining normalized Taylor coefficients, obtained from $W$ and $M$ by invariant differentiation, match. The neighbouring branches are governed by the same structure: $C$ alone generates when $S=0$ and $P\\neq 0$, while $X$ and $Y$ generate when $S\\neq 0$ and $W=0$.","pith_inferences":["A general proof that moving-frame recurrence formulas remain valid on jet bundles with differential relations, in the absence of a free action, would turn the rank-and-Cramér checks in this paper into a systematic method for other branchings; the paper itself raises this as an open question.","The same explicit strategy should apply to Levi-degenerate CR hypersurfaces or other geometric structures with non-free orbit behavior, where invariant algebras are currently described only abstractly.","From the commutator $[D_1,D_2] = -D_1 + \\frac{1}{3}W D_2$, one could try to eliminate $W$ when $D_2M\\neq 0$; if successful, a single invariant and a single differentiation operator might generate the algebra in a Zariski-open subbranch.","One testable extension is to compute the full syzygy ideal among the invariant derivatives of $W$ and $M$; the paper gives the commutator formula but not the complete syzygy basis."],"forward_implications":["Every local analytic parabolic surface in the branch $S\\neq 0$, $W\\neq 0$ is uniquely determined up to $\\mathrm{SA}_3(\\mathbb{R})$-equivalence by the values of $W$, $M$, and their invariant derivatives; the normal form lists these values as Taylor coefficients.","The full infinite-dimensional algebra of differential invariants is finitely generated: all higher-order invariants are rational functions of $W$, $M$, and their invariant derivatives.","The vanishings of the relative invariants give a complete invariant classification of developable surfaces: $S=0$ means a cylinder, $S\\neq 0, W=0$ means a cone, and $S\\neq 0, W\\neq 0$ means a tangential surface.","Apart from the straight cone $u = x^2/(2(1-y))$, there are no non-cylindrical special affinely homogeneous parabolic surfaces; all other homogeneous models in this setting are products of a plane curve with a line.","The explicit invariant differentiation operators $D_1,D_2$ obtained from the recurrences permit algorithmic computation of arbitrarily high-order invariants without re-solving the cross-section equations."],"supporting_citations":[{"why":"Supplies the recurrence formulas that express invariant derivatives in terms of higher-order monomial invariants and correction terms; the paper pulls these back to parabolic jet bundles.","marker":"[13]"},{"why":"Established the analogous one-generator theorem for elliptic and hyperbolic surfaces; the normal-form and generation argument here is modeled on that result.","marker":"[33]"},{"why":"Provides the power-series normal-form method used to compute the explicit invariants from normalized Taylor coefficients.","marker":"[36]"},{"why":"Develops the invariantization and relative-invariant machinery behind the phantom invariants and cross-section normalizations used in the paper.","marker":"[12]"},{"why":"Gives the affine-invariance formula for the Hessian rank and the transfer formula for the relative invariant S, which produce the invariant branching structure.","marker":"[26]"},{"why":"Provides the classical classification of developable surfaces as cylinders, cones, and tangential surfaces, re-expressed here in terms of S and W.","marker":"[14]"}],"fun_headline_variants":["Two invariants generate all parabolic surface classes","Parabolic surfaces need just two invariants","W and M suffice for parabolic surface invariants","Special affine invariants: two generators close the algebra","Order-4 W and order-5 M generate all parabolic classes"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof assumes that the standard recurrence formulas for differential invariants, proved for actions that admit a moving frame, remain valid when pulled back to the parabolic jet bundle even though the action is not free there; if that pulled-back validity failed, the derivation of the Maurer–Cartan terms and hence the generation theorem would collapse.","fun_headline_variants_meta":{"raw":{"variants":["Two invariants generate all parabolic surface classes","Parabolic surfaces need just two invariants","W and M suffice for parabolic surface invariants","Special affine invariants: two generators close the algebra","Order-4 W and order-5 M generate all parabolic classes"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000245,"raw_usage":{"total_tokens":1518,"prompt_tokens":911,"completion_tokens":607,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":527,"completion_tokens_details":{"reasoning_tokens":533}},"tokens_in":527,"tokens_out":607,"duration_ms":152028,"temperature":1.0,"reasoning_tokens":533,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:55:14.555277+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"On the explicit normalized surface of the main branch, compute the sixth-order invariants $I_{6,0}$ and $I_{5,1}$ independently by power-series normalization and check whether the recurrence-derived identities $D_1 I_{6,0} = I_{7,0} - \\frac{3}{2}W(7M-2I_{5,1})(4M-I_{5,1}) + \\frac{4}{3}W I_{6,0}$ and $D_2 I_{6,0} = I_{6,1} - I_{6,0} + 21WM - 8WI_{5,1}$ hold identically; if any term is missing, the pulled-back recurrence formula is invalid. Equally decisive, any sixth- or higher-order differential invariant not expressible as a rational function of $W$, $M$ and their invariant derivatives would disprove the generation theorem.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the recurrence formulas that express invariant derivatives in terms of higher-order monomial invariants and correction terms; the paper pulls these back to parabolic jet bundles."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Established the analogous one-generator theorem for elliptic and hyperbolic surfaces; the normal-form and generation argument here is modeled on that result."},{"cited_title":"Springer Proc","cited_arxiv_id":null,"evidence_quote":"Provides the power-series normal-form method used to compute the explicit invariants from normalized Taylor coefficients."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Develops the invariantization and relative-invariant machinery behind the phantom invariants and cross-section normalizations used in the paper."},{"cited_title":"Affine Rigidity Without Integration","cited_arxiv_id":"1903.00889","evidence_quote":"Gives the affine-invariance formula for the Hessian rank and the transfer formula for the relative invariant S, which produce the invariant branching structure."},{"cited_title":"Dover Books on Advanced Mathematics","cited_arxiv_id":null,"evidence_quote":"Provides the classical classification of developable surfaces as cylinders, cones, and tangential surfaces, re-expressed here in terms of S and W."}],"review_version":1}