{"id":"6a4f9546-6712-488b-88d9-fdbc33258b23","arxiv_id":"1908.02696","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Every Finsler metric on a surface with a 3-dimensional projective symmetry algebra is locally projectively equivalent to a Randers metric, with an explicit list of normal forms.","lead":"This paper classifies all two-dimensional Finsler metrics whose geodesics are preserved by at least three independent vector fields. It shows each such metric is locally projectively equivalent to a Randers metric or a Riemannian metric, and lists the possible normal forms.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 1 only classifies metrics near transitive points; the abstract's unqualified claim is not established for metrics whose projective algebra is nowhere transitive.","rationale":"After reviewing the proof, the transitive-point restriction is the only genuinely load-bearing gap. The internal logic from Lemma 3 to Lemma 1 to Lemma 2 to the construction of Finsler metrics is coherent for transitive algebras: the ODE list, the elimination via smoothness of g±, and the matching with explicit sprays are all plausible, and the explicit Randers and Riemannian metrics are correctly associated to the sprays. The unproved computations in Lemma 1 are a presentation issue, not a suspected error; they are routine given Lemma 4 and the stated method. The more serious issue is that the classification of Lie algebras of vector fields in Lemma 3 is explicitly restricted to the transitive case and to algebras with no element vanishing on an open neighborhood. If the projective algebra has no transitive point, the proof cannot start. The abstract states the result without this qualifier, so the advertised 'every Finsler metric' is not justified. This is not a disagreement with known results but a genuine scope gap in the argument. A fix would require either proving that any 3-dimensional projective algebra of a Finsler spray is transitive somewhere or classifying the intransitive case separately. The concrete test is to classify intransitive 3-dimensional Lie algebras of vector fields and see whether their invariant ODEs can be induced by a spray; if none survive, the gap is harmless, if some survive, the theorem is incomplete. The reader's verdict of CONDITIONAL is appropriate.","tokens_in":13203,"tokens_out":33708,"duration_ms":334779,"concrete_test":"Classify all 3-dimensional Lie algebras of vector fields on R^2 without the transitivity assumption, including rank-1 cases. For each representative, solve the symmetry condition (Eq. 1) for the induced ODEs f± and test the smoothness/compatibility conditions of §2.3 (existence of smooth g± and fiber-global definition). In particular, take the intransitive algebra span{∂x, x∂x, x^2∂x} acting on R^2 and compute whether a spray invariant under its prolonged action has exactly these three projective vector fields and is not projectively flat. If such a spray arises from a Finsler metric and is not in the list of Lemma 2, the classification in Theorem 1 is incomplete.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The most load-bearing assumption is the transitivity of the projective algebra at the base point, introduced in the Setup and used essentially in Lemma 3: the classification of 3-dimensional Lie algebras of vector fields on the plane assumes the algebra is transitive at the origin (isotropy 1-dimensional and no nonzero element vanishing on an open neighborhood). Lemma 1 inherits this assumption and Lemma 2 uses Lemma 1 to list sprays, so Theorem 1 is proved only 'near any transitive point'. The abstract, however, claims that every Finsler metric on a surface with at least a 3-dimensional projective algebra is locally projectively equivalent to a Randers metric, with no transitivity qualifier. If a metric has a 3-dimensional projective algebra that is nowhere transitive (for example, a rank-1 action such as span{∂x, x∂x, x^2∂x} extended trivially in y), the proof gives no information. No argument is supplied that such metrics are projectively flat, have larger projective algebra, or are projectively equivalent to one of the listed normal forms. Thus the completeness claim in the abstract is unsupported in this case, and even the theorem, as a classification of all such surfaces, is incomplete unless the intransitive case is handled separately.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies Finsler metrics on surfaces whose Lie algebra of projective vector fields has dimension at least three. The main result (Theorem 1) states that near any transitive point such a metric is projectively equivalent either to a Randers metric F = α + β with α of constant sectional curvature and p(F) equal to the Killing algebra of α, or to a Riemannian metric, and gives an explicit list of normal forms (a), (b±_k), (c±). The proof reduces the problem to the classification of 3-dimensional Lie algebras of vector fields in the plane (Lemma 3), the classification of second-order ODEs admitting three infinitesimal point symmetries (Lemma 1), and the conversion of these ODEs into spray normal forms (Lemma 2); Section 3 constructs the Finsler metrics whose geodesic sprays realize these normal forms.","tokens_in":13388,"tokens_out":11408,"duration_ms":110318,"significance":"If correct, the paper would settle the submaximal case dim p = 3 in the surface version of Lie's problem, complementing the classical projectively flat case (dim p = 8) and the pseudo-Riemannian classification of [5]. The explicit normal forms and the construction via Randers metrics are valuable, and the paper draws a useful connection between projective symmetry and geodesic curvature in magnetic geodesic terms (Lemma 7). However, the completeness of the classification is not yet established because of the restrictions and unproved steps listed in the major comments; the paper's main claim as stated in the abstract is not fully supported.","major_comments":[{"comment":"The abstract claims that every Finsler metric on a surface with at least three independent projective vector fields is locally projectively equivalent to a Randers metric, but the proof and Theorem 1 only apply 'near any transitive point' (Setup, p. 2). The paper gives no argument for metrics whose projective algebra is nowhere transitive, such as a rank-1 action extended trivially in the transverse direction; it is not shown that such metrics are projectively flat, have larger projective algebra, or are projectively equivalent to one of the listed normal forms. This gap affects the central completeness claim and should be addressed either by proving the intransitive case or by amending the abstract and theorem.","section":"Abstract; Setup; Theorem 1"},{"comment":"The proof of Lemma 3(2) only shows that the isotropy element X0 is not central: it assumes X0 commutes with X1 and X2 and derives a contradiction. The statement being proved is stronger, namely that the isotropy subalgebra g0 is not an ideal. The subsequent classification of pairs (g,h) uses the condition that h is not an ideal, so the list of Lie algebras of vector fields in Section 2.1 is incomplete if there exists a transitive 3-dimensional action whose 1-dimensional isotropy is a non-central ideal. The proof must rule out [X0,X1] ∈ span{X0} and [X0,X2] ∈ span{X0} with at least one nonzero bracket, or the classification must be extended.","section":"§2.1, Lemma 3"},{"comment":"Lemma 1 is load-bearing: it provides the list of second-order ODEs with three independent point symmetries from which all subsequent normal forms are derived. However, the paper only states that the system (1) 'can be solved by elementary methods' and lists the results without showing the integration or providing a reference that contains this exact classification. As written, the completeness of the list cannot be checked. A derivation for each of the cases D1, D2, J1, J2, C1, C2, or a precise citation, is needed.","section":"§2.2, Lemma 1"},{"comment":"The non-equivalence assertion 'None of these sprays can be transformed into one projectively equivalent to one of the others' is not proved: the text refers to 'direct calculations or using invariants for the induced ODEs, see [7]' but gives no calculation, invariant, or reference to a specific result. This statement is used in Theorem 1 to claim that the listed metrics are distinct up to projective equivalence. The authors should supply the missing argument or a concrete reference.","section":"§2.3, Lemma 2"}],"minor_comments":[{"comment":"The abstract's 'locally projectively equivalent to a Randers metric' is stronger than the theorem's disjunction 'a Randers metric ... or a Riemannian metric', and the theorem's qualifier 'near any transitive point' is absent from the abstract; the abstract should be aligned with the theorem.","section":"Abstract vs. Theorem 1"},{"comment":"The formula for D2, f = C z^{(λ-2)/(λ-1)}, is undefined for λ=1. The special role of λ=1 is mentioned in Lemma 4(2) but should be stated explicitly in Lemma 1.","section":"§2.2, Lemma 1"},{"comment":"The formulas for g± involve 1/z and the text writes conditions 'if z ≥ 0' and 'if z ≤ 0'; at z = 0 these expressions must be interpreted by limits, which is not explained.","section":"§2.3, equation (3)"},{"comment":"There are several typos: 'prooving' in Section 2, 'stritcly' in Definition 1, and 'Affilation' in the author footnote. Reference [7] is an arXiv preprint; if a published version exists, it should be cited.","section":"Throughout"},{"comment":"In the classification of Lie algebras, the case J3 is presented with parameters γ0,γ1; the text says one can assume γ0,γ1 ∈ {0,1} but then gives a general formula. The normalization deserves a sentence of explanation.","section":"§2.1, J3"}],"recommendation":"major_revision","confidential_remarks":"The paper contains a promising approach and the main classification is likely correct, but the proof is too compressed in several load-bearing places (Lemma 1 and Lemma 2) and the transitive-point restriction is not reconciled with the abstract. In its current form I would not accept. After the authors supply the missing computations and correct the statements, the paper could be a solid contribution. There is no suspicion of misconduct; the issues are mathematical rigor and presentation."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"First, the substance: this is one of the first real classifications of Finsler metrics with a 3-dimensional projective algebra on a surface, and the normal forms are explicit and believable. The reduction to Lie's ODE classification is a sensible route, and the argument that most invariant ODEs cannot come from a fiber-globally defined spray is a nice piece of filtering. The construction of Randers metrics with geodesics of constant curvature from a constant-curvature base metric (Lemma 7) is clean and gives a clear geometric picture for the examples. If the paper is correct, it settles the submaximal case of Lie's problem for Finsler metrics in the transitive regime.\n\nThe soft spots are real but not fatal. The abstract states that every Finsler metric with at least three projective vector fields is locally projectively equivalent to a Randers metric, without any qualifier. The theorem itself only claims this near a transitive point, i.e. a point where the projective vector fields span the tangent plane. The proof depends on that assumption essentially: Lemma 3 classifies 3-dimensional Lie algebras of vector fields under transitivity, and the ODE classification inherits it. If a metric has a 3-dimensional projective algebra that is nowhere transitive, the paper gives no argument that it must be projectively flat or that it is projectively equivalent to one of the listed forms. That is a genuine gap between abstract and theorem, and it should be fixed by either handling the intransitive case or making the abstract match the theorem.\n\nA secondary issue is that Lemma 1, which is load-bearing, lists the invariant ODEs with the note that the systems are 'solved by elementary methods' but does not show the derivations. A referee will want to see enough of the computation to trust the list, especially because the subsequent filtering by global definedness depends on that list being complete. This is likely fillable, but it is exactly the kind of thing that should not be left as an exercise.\n\nOverall, the paper is a solid contribution and deserves a serious referee. With a corrected abstract and either a proof of the intransitive case or an explicit restriction, it should be publishable. For anyone working on projective geometry of Finsler metrics, this is a useful reference.","headline":"Substantive local classification with an overreaching abstract; the intransitive case and unshown ODE derivations are the main gaps.","tokens_in":13940,"tokens_out":4478,"would_cite":true,"duration_ms":49928,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["53B40","53A20","58B20"],"pacs":[],"model":"deepseek-v4-flash","headline":"A complete local classification of Finsler surfaces with three projective vector fields.","keywords":["Finsler metrics","projective vector fields","Randers metrics","projective equivalence","sprays","second-order ODEs","geodesics","constant curvature"],"falsifier":"Exhibit a Finsler metric on a surface with a three-dimensional projective algebra that has no transitive point and whose geodesic spray is not projectively equivalent to any spray in Lemma 2; that example would refute the abstract's unqualified claim while leaving the transitive-point theorem intact.","tokens_in":12966,"feed_emoji":"📐","tokens_out":12505,"duration_ms":112984,"temperature":0.7,"pith_summary":"This paper settles the submaximal case of a classical problem: which Finsler metrics on a surface admit many projective vector fields? The answer, argued in Theorem 1, is that any Finsler metric with at least three independent projective vector fields is locally projectively equivalent, near any transitive point, either to a Randers metric built from a Riemannian metric of constant sectional curvature, or to a Riemannian metric. Up to coordinate change and projective equivalence, the possibilities reduce to a short list: the Euclidean metric, two Randers families associated with constant-geodesic-curvature dynamics on the Euclidean, spherical, and hyperbolic planes, and two Riemannian metrics. The paper also proves that no two metrics on this list are locally isometric to projectively equivalent metrics, so the list is both complete and minimal.","feed_headline":"Three projective fields put Finsler surfaces on a short list","feed_subtitle":"Local classification into Randers and Riemannian normal forms, up to projective equivalence.","key_machinery":"The proof works with sprays rather than metrics. The central object is the pair of second-order ODEs induced by a spray when its geodesics are parametrized by $x$, one for $\\dot x > 0$ and one for $\\dot x < 0$; this pair determines the spray up to projective equivalence. The paper classifies all three-dimensional Lie algebras of vector fields on the plane that are transitive at a point, solves the infinitesimal point-symmetry equations for each, and obtains the normal-form list of second-order ODEs in Lemma 1. Filtering out equations that cannot come from a fiber-globally defined spray yields the spray normal forms (a), (b$_k^\\pm$), (c$^\\pm$) of Lemma 2. The Randers metrics are produced by Lemma 7: for a constant-curvature metric $\\alpha$, adding a 1-form $\\beta$ whose exterior derivative is a constant multiple of the volume form makes the geodesics of $F = \\alpha + \\beta$ the curves of constant geodesic curvature $k$, so the whole Killing algebra of $\\alpha$ becomes projective for $F$.","core_discovery":"On the paper's own terms, the central discovery is Theorem 1: in dimension two, a Finsler metric whose projective vector fields form a Lie algebra of dimension at least three is, near any transitive point, projectively equivalent to a Randers metric $F = \\alpha + \\beta$ with $\\alpha$ of constant sectional curvature and with the projective algebra equal to the Killing algebra of $\\alpha$, or to a Riemannian metric. Written out, the local normal forms are the Euclidean metric; the Randers metric (a), $\\sqrt{dx^2+dy^2} + \\tfrac{1}{2}(y\\,dx - x\\,dy)$; the two Randers families (b$_k^\\pm$) for $k>0$, obtained by adding a 1-form to the round sphere metric and to the hyperbolic plane metric; and the Riemannian metrics (c$^\\pm$), $\\sqrt{\\frac{e^{3x}}{(2e^x-1)^2}\\,dx^2 + \\frac{e^x}{2e^x-1}\\,dy^2}$ and $\\sqrt{e^{3x}\\,dx^2 + e^x\\,dy^2}$. None of these metrics is locally isometric to a Finsler metric projectively equivalent to a different one, so the classification is complete and irredundant.","pith_inferences":["The same second-order ODE machinery would classify any spray with a three-dimensional projective algebra, not only geodesic sprays of Finsler metrics, so the normal-form list may serve as a general model for projectively symmetric second-order ODEs on the plane.","The measure-theoretic construction mentioned in the paper shows the list classifies projective classes, not all metrics in a class: there exist non-trivial Finsler metrics projectively equivalent to (a), so a classification of metrics rather than sprays would need an additional description of these deformations.","Because the theorem is proved only at transitive points, metrics whose three-dimensional projective algebra is nowhere transitive remain unexplored; checking whether such metrics exist would either close the gap or force a refinement of the statement."],"forward_implications":["Any Finsler surface with exactly three independent projective vector fields is locally projectively equivalent to one of the listed normal forms, so the search for submaximal examples ends with a finite list.","A Finsler metric with more than three projective fields is projectively flat, so the dimension of the projective algebra jumps from 3 to 8 in dimension two.","The Randers metrics (a) and (b$_k^\\pm$) are geodesically irreversible, while the Riemannian metrics (c$^\\pm$) are geodesically reversible; this distinction is invariant under projective equivalence within the list.","For each Randers metric on the list the projective algebra equals the Killing algebra of the underlying constant-curvature metric, so the three-dimensional symmetry algebra is exactly the isometry algebra of a classical geometry."],"supporting_citations":[{"why":"Supplies the classical classification of second-order ODEs by infinitesimal point symmetries, the technique behind Lemma 1.","marker":"[9]"},{"why":"Gives the result that more than three point symmetries force the ODE into the flat form used in Lemma 5.","marker":"[12]"},{"why":"Provides the normal forms for two-dimensional metrics with two projective fields and the PDE system used to construct the Riemannian metrics (c$^\\pm$).","marker":"[5]"},{"why":"Proves that two Randers metrics are projectively equivalent only when trivially related, supporting the rigidity discussion.","marker":"[10]"},{"why":"Shows that measures on the plane satisfying a ball condition produce Finsler metrics projectively equivalent to (a), connecting to the Pompeiu problem.","marker":"[16]"},{"why":"Supplies invariants of second-order ODEs used to distinguish the spray normal forms (c$^+$) from (c$^-$) and (b$_k^+$) from (b$_k^-$).","marker":"[7]"}],"fun_headline_variants":["Three projective fields pin down Finsler surfaces to Randers or Riemannian","Finsler surfaces with 3 projective vector fields fully classified","Complete local list: Finsler surfaces with 3 projective fields","Randers or Riemannian: Finsler surfaces with 3 projective fields","Short list completes projective classification of Finsler surfaces"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof assumes the three projective fields span the tangent plane at the point around which the classification is made; a metric whose projective algebra is three-dimensional but has no such transitive point is not covered by the argument.","fun_headline_variants_meta":{"raw":{"variants":["Three projective fields pin down Finsler surfaces to Randers or Riemannian","Finsler surfaces with 3 projective vector fields fully classified","Complete local list: Finsler surfaces with 3 projective fields","Randers or Riemannian: Finsler surfaces with 3 projective fields","Short list completes projective classification of Finsler surfaces"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000684,"raw_usage":{"total_tokens":3049,"prompt_tokens":835,"completion_tokens":2214,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":451,"completion_tokens_details":{"reasoning_tokens":2123}},"tokens_in":451,"tokens_out":2214,"duration_ms":14873,"temperature":1.0,"reasoning_tokens":2123,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T14:38:17.988993+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Exhibit a Finsler metric on a surface with a three-dimensional projective algebra that has no transitive point and whose geodesic spray is not projectively equivalent to any spray in Lemma 2; that example would refute the abstract's unqualified claim while leaving the transitive-point theorem intact.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the classical classification of second-order ODEs by infinitesimal point symmetries, the technique behind Lemma 1."},{"cited_title":"Romanovski ˘ ı","cited_arxiv_id":null,"evidence_quote":"Gives the result that more than three point symmetries force the ODE into the flat form used in Lemma 5."},{"cited_title":"Bryant, G","cited_arxiv_id":null,"evidence_quote":"Provides the normal forms for two-dimensional metrics with two projective fields and the PDE system used to construct the Riemannian metrics (c$^\\pm$)."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Proves that two Randers metrics are projectively equivalent only when trivially related, supporting the rigidity discussion."},{"cited_title":"Tabachnikov","cited_arxiv_id":null,"evidence_quote":"Shows that measures on the plane satisfying a ball condition produce Finsler metrics projectively equivalent to (a), connecting to the Pompeiu problem."},{"cited_title":"The geometry of second-order ordinary differential equations","cited_arxiv_id":"1602.00913","evidence_quote":"Supplies invariants of second-order ODEs used to distinguish the spray normal forms (c$^+$) from (c$^-$) and (b$_k^+$) from (b$_k^-$)."}],"review_version":1}