{"id":"db519ea8-d552-4d72-a2fd-a1c81ac8475a","arxiv_id":"1908.02208","paper_version":4,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"On elliptic surfaces, all but finitely many multiples of a non-torsion section meet the zero section transversely, yielding new transversality results and surfaces with fixed geometric genus and unbounded K^2.","lead":"An elliptic surface with a non-torsion section P has only finitely many points where a multiple of P is tangent to the zero section. The paper uses this to build surfaces of arbitrary fixed geometric genus with ample canonical class and unbounded self-intersection.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Proof of main finiteness theorem (Thm 2.5) omits a required case: tangencies near additive potentially good fibers II, II*, III, III*, IV, IV* are left as an exercise.","rationale":"The reader's weakest assumption is the same as the one I would single out: the proof of the central finiteness theorem explicitly defers the additive potentially good reduction types. I see no other assumption that is comparably load-bearing. The uniqueness/positivity of the height pairing, the divisor-sequence reformulations, and the geography construction are checked in enough detail; the I0/I1/Ib/I*b local arguments are explicit and internally consistent. A useful independent check is that [CDMZ19] is claimed to prove a closely related finiteness statement, but that result is not used in the proof here and does not remove the need for a complete proof of Theorem 2.5 in this paper. The authors' own sentence 'We leave the details as an exercise for the reader' is an explicit admission of the missing case, and hard rules require flagging it. Verdict remains CONDITIONAL: the gap is real but plausibly fixable, and the rest of the paper, including the very-general transversality and the geography application, appears coherent. I therefore do not adjust the reader's conditional acceptance, and I agree with the reader's identification of the weakest point.","tokens_in":28031,"tokens_out":12377,"duration_ms":143189,"concrete_test":"Work out the omitted local case for Kodaira type IV as a test: let X→Δ have type IV reduction, take the degree-3 cover Δ~→Δ, z=u^3, pull back to obtain a good-reduction surface, and derive the pulled-back tangency equation from the analogue of (3.3). Verify that a sequence z_i→0 of tangencies of P to local Betti leaves would produce a sequence u_i→0 where the pulled-back section is tangent to a Betti leaf, contradicting the good-reduction estimate of Claim 3.1, and that the pulled-back section is not itself contained in a Betti leaf. Repeat with the degree-4 covers for III/III* and degree-6 covers for II/II*; if the calculation cannot be carried out, Theorem 2.5 is incomplete for that reduction type.","verdict_should_be":"UNCHANGED","load_bearing_attack":"At the end of Section 3, the proof of Theorem 2.5 handles I0, I1, Ib, and I*b by explicit local models and Betti-leaf calculations, but for the additive potentially good reduction types II, II*, III, III*, IV, IV* it merely says a ramified base change of degree 2, 3, 4, or 6 reduces to good reduction and 'We leave the details as an exercise.' This is load-bearing because Claim 3.1 needs to exclude sequences of tangencies accumulating at every point of C, including bad fibers. For the cases written out, the paper compares the holomorphic section P with the local Betti-leaf family and derives a contradiction (e.g., equation (3.3) and the growth estimates (3.5)). For the additive types, no tangency equation is derived, and no check is supplied that the pulled-back section stays outside the Betti foliation, that the Betti-leaf condition transforms correctly under the cyclic cover, and that the good-reduction argument on the cover really rules out accumulation in the original punctured disk. If one of these checks fails for e.g. type IV or IV*, the discreteness proof would not cover tangencies accumulating at that fiber, and Theorem 2.5 would be unproved. The omission is explicit in the text and not supplied elsewhere in the paper.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies a relatively minimal Jacobian elliptic surface E over a curve C in characteristic zero, equipped with a zero section O and a non-torsion section P. The central result (Theorem 1.1, reformulated as Theorem 2.5) asserts that the set of base points t at which O is tangent to nP for some nonzero integer n is finite, equivalently that the elliptic divisibility divisors D_n = O^*(nP) are reduced for all n outside a finite exceptional set. Section 3 proves this via local analysis of the Betti foliation: for good reduction and the multiplicative types I1, Ib, and I*_b, the proof derives explicit tangency equations and uses growth estimates to rule out accumulation of tangencies at the corresponding fibers. For the additive potentially good types II, II*, III, III*, IV, and IV*, the proof asserts a reduction to the good-reduction case by a ramified base change and leaves the details as an exercise. The paper then proves genericity and explicit-construction results (Theorems 1.7 and 1.8) in which no tangencies occur, and applies them to construct normal surfaces with fixed geometric genus, one log-terminal singularity, ample Q-Cartier canonical class, and arbitrarily large K^2 (Theorem 1.9).","tokens_in":28263,"tokens_out":8063,"duration_ms":84623,"significance":"If the main theorem is fully proved, this is a substantial contribution: it gives a sharp function-field analogue of square-freeness of elliptic divisibility sequences, replacing a conjectural o(n^2) bound by an actual finiteness statement; it establishes a non-obvious global transversality theorem through the Betti foliation; and it supplies new geography examples with unbounded K^2 for fixed geometric genus. The generic transversality theorem for very general data and the explicit height-2 examples over small fields are valuable in their own right. The paper is clearly written and uses standard tools (Kodaira models, Deligne-Rapoport, canonical height pairings, Artin contraction) in an appropriate way. However, the proof of the central finiteness theorem has a load-bearing gap in the additive reduction cases, so the manuscript is not yet suitable for publication as is.","major_comments":[{"comment":"The proof of Theorem 2.5 is incomplete because the additive potentially good reduction types are dismissed with 'We leave the details as an exercise for the reader.' This is load-bearing: Claim 3.1 must exclude tangencies accumulating at every point of C, including at bad fibers. For the cases that are written out, the argument derives a concrete tangency equation (for example equation (3.3) for type I1) and uses the growth estimates in (3.5). For the additive types no such equation is derived, and the asserted reduction to good reduction after a ramified base change of order 2, 3, 4, or 6 requires several non-formal checks: that the pulled-back section stays outside the Betti foliation, that the tangency condition transforms correctly under the ramified cyclic cover, and that the absence of accumulation points on the cover implies the absence of accumulation in the original punctured disk. Since none of these checks appears in the manuscript, the discreteness argument does not currently cover points of types II, II*, III, III*, IV, or IV*, and Theorem 2.5, and therefore Theorems 1.1 and 1.4, are unproved as written.","section":"Section 3, final case ('The cases of II, II*, III, III*, IV, and IV* reduction')"}],"minor_comments":[{"comment":"The word 'compex manifolds' should be 'complex manifolds'.","section":"Section 3, first paragraph"},{"comment":"'Thereom 1.4' should be 'Theorem 1.4'.","section":"Remark 1.5"},{"comment":"'let Cj be the strict transform the section jP' should read 'let Cj be the strict transform of the section jP'.","section":"Section 8, paragraph after blowing up"},{"comment":"In the sentence 'Composing φ with a linear fractional transformation', φ should be f, since f is the morphism C→P^1 just chosen.","section":"Proof of Theorem 1.8"},{"comment":"The phrase 'there exist infinitely many pairs (E,P ) where E is a Jacobian elliptic surface E→C equipped with a section P such that' is grammatically awkward and should be rephrased.","section":"Statement of Theorem 1.8"}],"recommendation":"major_revision","confidential_remarks":"The omitted additive-reduction cases are the only serious obstacle I see to the main theorem. The rest of the paper is strong and the gap appears fixable, but it is not a presentation issue: the authors must either supply the missing argument or cite a precise reference that covers the additive potentially good cases. I would be willing to review a revised version."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nRead this one if you care about unlikely intersections on elliptic surfaces or geography of surfaces. The headline is that the paper has a real main theorem, a smart method, and one honest gap in the proof that should be fixed before the result is taken as established.\n\nWhat is new: the tangency finiteness theorem (equivalent to a finiteness statement for non-reduced elliptic divisibility divisors) was independently obtained by Corvaja-Demeio-Masser-Zannier; the authors say so. What they add is the Betti-foliation proof, which is a genuinely different and more geometric route. The new material is the very general no-tangency results (Theorems 1.7/1.8) and the geography application (Theorem 1.9): log-terminal surfaces of fixed geometric genus with K ample and K^2 unbounded. That is a striking answer to a natural boundedness question, and the construction is explicit and checks out. The explicit examples over number fields (Section 7) are also valuable and self-contained. Credit where due: the paper is honest about the overlap and mostly writes down real arguments rather than deferring to references.\n\nThe soft spot is in the proof of the central Theorem 2.5. The cases I0, I1, Ib, I*b are written in detail, but the additive potentially good types II, II*, III, III*, IV, IV* are disposed of with 'we leave the details as an exercise for the reader.' That line is load-bearing: Claim 3.1 needs discreteness of tangencies near every fiber, including those. A ramified base change of degree 2, 3, 4, or 6 ought to reduce to the good case, but you still have to check that the Betti leaf tangency condition transforms correctly and that accumulation in the punctured disk is really ruled out. The paper does not do that check. I don't see an obvious reason it fails—the reduction is plausible and the authors know this material—but as written the main theorem has a hole. This is not a manufactured nitpick; it is an explicit gap flagged in the text.\n\nThe geography section engages with the theorem as a black box, and the theorem is plausible; if the gap is filled, the paper is strong. The citation pattern is fine. No circularity: the proof uses standard results, and the overlap with CDMZ is properly credited.\n\nWho this is for: anyone working on elliptic surfaces, divisibility sequences, or surface geography. It deserves a serious referee; the right call is to send it out with a request that the additive case be written up. I would not accept it as is.\n\nBest.","headline":"Genuinely new Betti-foliation results and a striking geography construction, but the proof of the central finiteness theorem has a load-bearing gap in the additive reduction cases.","tokens_in":28847,"tokens_out":1843,"would_cite":true,"duration_ms":18993,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["14J27","11B39","14J29"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proves that, over a field of characteristic zero, the zero section of a Jacobian elliptic surface is tangent to nP for only finitely many base points; in very general families no tangencies occur at all.","keywords":["elliptic surfaces","sections","transversality","elliptic divisibility sequences","Betti coordinates","geography of algebraic surfaces","log-terminal singularities","unlikely intersections"],"falsifier":"Take a concrete elliptic surface with one additive potentially good fiber and a non-torsion section P, pull back by the asserted ramified cover of degree 2, 3, 4, or 6, and compute the tangency points of P with the torsion multisections E[n] near the ramification point. If those tangencies accumulate at the fiber, the omitted case in Section 3 collapses the discreteness argument; if the pulled-back equation lands in the already-proved I0, I1, Ib, or I*b cases with no accumulation, the load-bearing step is confirmed.","tokens_in":27793,"feed_emoji":"📐","tokens_out":9726,"duration_ms":108569,"temperature":0.7,"pith_summary":"The paper studies a Jacobian elliptic surface π:E→C with a zero section O and another section P of infinite order. It proves that the set of points of C where O is tangent to some multiple nP is finite, and equivalently that the elliptic divisibility divisors Dn=O∗(nP) are reduced for every n not divisible by one of a finite set of integers. For a very general choice of coefficients in a sufficiently positive linear system, it proves the stronger statement that O meets nP transversally in exactly d(n2−1) points for every n prime to the characteristic. This transversality drives an explicit geography result: there exist normal surfaces of any fixed geometric genus with one mild singular point, ample canonical class, and arbitrarily large K2.","feed_headline":"Only finitely many tangencies between a section and its multiples","feed_subtitle":"Very generally there are none, yielding square-free divisibility sequences and unbounded canonical degree.","key_machinery":"The load-bearing object is the local Betti foliation: over a disk in the good-reduction locus, the elliptic surface is (Δ×C)/(Zτ+Z), and the leaves Lr,s are the images of z↦(z,rτ(z)+s). In the real coordinates (r,s) on the universal cover, every torsion section becomes a constant rational leaf, and the section P becomes a real-analytic map φ:Δ→R2 whose derivative vanishes exactly at tangencies with some leaf. The Lojasiewicz gradient inequality then rules out accumulating zeros of the derivative away from the point where φ equals its base value, giving discreteness. Near bad fibers, Kodaira's models convert the same idea into an inequality comparing zf′(z)/f(z) with log|f(z)|/log|z|, proving no tangencies can accumulate at a multiplicative fiber. The other main mechanism is the moduli space of triples (E,ω,P), represented by Spec k[a2,a3,a4], where the condition P∈E[n] is a hypersurface; the very-general transversality theorem follows by showing the incidence correspondence over V=H0(L2⊕L3⊕L4) is smooth and étale over a nonempty open parameter set.","core_discovery":"The paper's central claim is that for a relatively minimal Jacobian elliptic surface over the complex numbers, with zero section O and a non-torsion section P, the set T=⋃n≠0{t∈C:nP is tangent to O at t} is finite. The proof works with the n-torsion multisection E[n]: P is tangent to E[n] exactly when nP is tangent to O, so proving P is transverse to all E[n] away from finitely many points proves the theorem. The authors show that the larger set of tangencies of P with all local Betti leaves Lr,s is discrete, hence finite, using Kodaira's local models and the Lojasiewicz gradient inequality; the torsion tangencies are contained in this larger set. The same finiteness is equivalent to a square-freeness statement for function-field elliptic divisibility sequences: a finite set M exists such that Dn is reduced if and only if n is not divisible by any element of M. For very general data in H0(C,L2⊕L3⊕L4), the paper proves there are no tangencies at all, and applies this to construct, for every g≥0 and every N, a normal projective surface with geometric genus g, one log-terminal singular point, K ample and Q-Cartier, and K2>N.","pith_inferences":["If the omitted base-change verification for the six additive potentially-good fiber types is completed, the same Betti-coordinate method should yield effective upper bounds on the number of tangencies, since the Lojasiewicz inequality produces quantitative separation near each fiber.","The very-general transversality statement suggests a concrete algebraic test in positive characteristic: for p>3 one expects the same transversality to hold for all n prime to p, and the moduli-space argument in the paper could likely be adapted without passing through complex analysis.","The geography construction implicitly answers a question about accumulation of volumes in moduli of surfaces: fixing the geometric genus does not bound the canonical self-intersection even for surfaces with only one log-terminal singularity, so any lower-bound results of that type cannot extend to upper bounds.","The Betti-leaf viewpoint may also yield a constructive algorithm for the finite set M: instead of checking each n separately, one could track the leaves met by P and compute the finitely many rational leaf parameters where the derivative vanishes."],"forward_implications":["For every non-torsion section in characteristic zero, the function-field elliptic divisibility sequence Dn has bounded 'non-square-free part': the degree of gcd(fn,dfn/dt) is bounded, not merely subquadratic.","For a very general parameter in H0(C,L2⊕L3⊕L4), every multiple nP meets O transversally in exactly d(n2−1) points, so the full sequence of sections is uniformly transverse.","There exist explicit pairs (E,P) over number fields and global function fields, with singular fibers of type I0*, in which nP meets O transversally in the exact counts (n2−1)/2 for odd n and (n2−4)/2 for even n.","Given any integers g≥0 and N, there is a normal projective surface with pg=g, one log-terminal singular point, ample K, and K2>N, so K2 has no upper bound in terms of pg among mildly singular surfaces of general type."],"supporting_citations":[{"why":"Supplies the local analytic models of elliptic fibrations, including the multiplicative and additive cases, on which the discreteness proof is built.","marker":"[Kod63]"},{"why":"The Lojasiewicz gradient inequality is the tool that rules out accumulating vanishing derivatives in the good-reduction case.","marker":"[Łoj64]"},{"why":"Establishes the group-scheme structure of the smooth locus and the torsion multisections E[n] used to reformulate tangency.","marker":"[DR73]"},{"why":"Provides the formal-group computation that [n] acts as multiplication by n on fiber tangent spaces, plus the division polynomials and height basics.","marker":"[Sil09]"},{"why":"Gives the intersection and height formulas that convert (nP).O into d(n2−1) and identify the correction terms used in explicit examples.","marker":"[CZ79]"},{"why":"Defines the function-field divisibility sequence framework, primitive divisors, and the Dn/Dn′ notation that the finite-set theorem sharpens.","marker":"[IMS+12]"},{"why":"Provides the moduli description of elliptic curves with differential, represented by Spec R[a2,a3,a4], used in the very-general transversality proof.","marker":"[Del75]"},{"why":"The contractibility criterion that contracts the two rational curves C0∪Cn to the single log-terminal singularity in the geography construction.","marker":"[Art62]"}],"fun_headline_variants":["Finitely many tangencies for elliptic sections","Very general elliptic surfaces avoid tangencies","Tangency finiteness gives square-free sequences","Elliptic surfaces with unbounded K^2 from tangencies","No tangencies for very general elliptic data"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof relies on an unverified local step: near the six exceptional types of singular fiber with additive potentially good reduction, the authors assert that a ramified base change of degree 2, 3, 4, or 6 reduces the problem to cases already proved, and they leave the details as an exercise for the reader.","fun_headline_variants_meta":{"raw":{"variants":["Finitely many tangencies for elliptic sections","Very general elliptic surfaces avoid tangencies","Tangency finiteness gives square-free sequences","Elliptic surfaces with unbounded K^2 from tangencies","No tangencies for very general elliptic data"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000387,"raw_usage":{"total_tokens":2063,"prompt_tokens":984,"completion_tokens":1079,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":600,"completion_tokens_details":{"reasoning_tokens":1010}},"tokens_in":600,"tokens_out":1079,"duration_ms":11778,"temperature":1.0,"reasoning_tokens":1010,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T14:51:51.399492+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a concrete elliptic surface with one additive potentially good fiber and a non-torsion section P, pull back by the asserted ramified cover of degree 2, 3, 4, or 6, and compute the tangency points of P with the torsion multisections E[n] near the ramification point. If those tangencies accumulate at the fiber, the omitted case in Section 3 collapses the discreteness argument; if the pulled-back equation lands in the already-proved I0, I1, Ib, or I*b cases with no accumulation, the load-bearing step is confirmed.","supporting_citations":[],"review_version":1}