{"id":"ac25d33b-2dde-433f-98b0-03fb72ba849d","arxiv_id":"2501.18541","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"By extending Kuga-Satake period morphisms to singular models, the authors prove the Tate conjecture for many geometric-genus-one surfaces and BSD for height-one elliptic curves over genus-one function fields.","lead":"This paper proves new cases of the Tate conjecture for surfaces with geometric genus one, including the Birch-Swinnerton-Dyer conjecture for every height-one elliptic curve over a genus-one function field in characteristic at least 11. The key innovation is a period-morphism technique that handles singular natural models of such surfaces instead of avoiding them.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The D^- case of Theorem 4.2 hinges on the boundary extension of the crystalline period matching; the compatibility proved in Lemma 3.10 via p-adic RH functors is the least secure step.","rationale":"The reader identified the D^- locus and the observation 'any pi-exceptional curve is also tau-exceptional' as the structural load-bearing premise, and I agree that this is the most critical point in the proof of Theorem 4.2. My stress-test narrows the concern to the boundary extension of the crystalline period matching, because the expected-Hodge-number branch of Proposition 4.27(ii) ultimately depends on the constancy of L_crys|_C via the matching isomorphism alpha_crys. The flop-decomposition and constant-family-resolution inputs (Theorem 2.17 and Corollary 2.18) are used to prove Y|_C is constant, but the jump from 'Y|_C is constant' to 'L_crys|_C is constant' requires the F-crystal matching over the boundary. That matching is established through a novel compatibility between p-adic and classical Riemann-Hilbert functors (Lemma 3.10), which is only sketched and is not covered by Blasius' theorem for non-abelian motives. This is not a described error in the text, but it is the least externally verified link in the chain leading to the headline BSD result. The reader's verdict is CONDITIONAL, and my analysis does not change that verdict; it sharpens the condition to a specific technical step that should be independently checked before the paper is accepted as fully verified.","tokens_in":57044,"tokens_out":29069,"duration_ms":307867,"concrete_test":"Re-derive Proposition 3.11 for the universal height-one Weierstrass family over a complete DVR at a boundary point of D^+, using only the classical crystalline-de Rham comparison and the rigidity theorem (Theorem 2.2), without invoking the p-adic Riemann-Hilbert functor of [GY24]. Check that the resulting isomorphism alpha_crys sends the Frobenius on L_crys to the correct Frobenius on the primitive crystalline cohomology of the minimal resolution of the special fiber. If the two Frobenius structures differ on the boundary, Lemma 3.10 fails and the D^- case collapses; if they agree, the concern is resolved.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The proof of Theorem 4.2 for points s in D^- is the only place where singular natural models are truly essential, and it rests on the observation that any pi-exceptional curve is also tau-exceptional. In the expected-Hodge-number case, which covers Theorems 1.1 and 1.2, this is proved via Proposition 4.27(ii) by showing that L_crys|_C is a constant F-crystal. That constancy is obtained by extending the matching isomorphism alpha_crys over the boundary of the smooth locus (Construction 4.4). The construction of alpha_crys in Proposition 3.11 depends on the p-adic Riemann-Hilbert functor of [GY24] and on Lemma 3.10, which proves a compatibility between the p-adic and classical Riemann-Hilbert outputs by analytic density of Noether-Lefschetz loci. This compatibility is exactly the point where the paper departs from Blasius' theorem for abelian motives (Remark 1.7). If Lemma 3.10 fails for the non-abelian motives arising from elliptic surfaces of height 1, then the isomorphism alpha_crys cannot be shown to extend to the singular boundary, the F-crystal L_crys|_C need not be constant, Proposition 4.27(ii) collapses, and the D^- case of Theorem 4.2 is unproved. This would remove the unconditional BSD result of Theorem 1.1 and the general-type Tate result of Theorem 1.2.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a general framework (Theorem 4.2) for proving the Tate conjecture for divisors on minimal resolutions of fibers of a family of surfaces of geometric genus one over a Z_(p)-scheme, subject to conditions on the discriminant locus, relative boundary, Kodaira–Spencer map, and rational double point singularities. The proof constructs extended period morphisms to orthogonal Shimura varieties and establishes motivic alignment of special endomorphisms with divisor classes, using separate arguments for the smooth/mildly singular locus D^+ and the worse singular locus D^-. Applications include the Birch–Swinnerton-Dyer conjecture for every height-one elliptic curve over a genus-one global function field in characteristic p ≥ 11 (Theorem 1.1) and the Tate conjecture for minimal surfaces with p_g = K^2 = 1 and q = 0 (Theorem 1.2).","tokens_in":57311,"tokens_out":60382,"duration_ms":680842,"significance":"If the proof is correct, Theorem 1.1 is a substantial new unconditional BSD result, and Theorem 1.2 significantly extends the earlier results of [HYZ25] by allowing singular natural models. The paper is also valuable for introducing and combining several technical tools: Artin–Brieskorn simultaneous resolutions, the minimal model program for threefolds in positive characteristic, a Beauville–Laszlo gluing criterion, rigidity of F-crystals, and a p-adic Riemann–Hilbert matching for non-abelian motives. The authors are transparent about the conditional status of the remaining p = 5, 7 cases (Remark 4.35) and about the insensitivity of their method to the type of RDP with respect to p (Remark 1.8). However, several load-bearing arguments are only sketched, and at least one proof contains an incorrect Riemann–Hurwitz computation; the main theorems as stated also overclaim characteristic zero cases that the proof does not treat.","major_comments":[{"comment":"The proof of Proposition 2.26 is not valid for wildly ramified covers, which are explicitly allowed by Remark 1.8. The equality (10) and the subsequent comparison (12) use the contribution (m_Q − 1)·e_Q to the Euler characteristic difference, but this is the tame-ramification formula; for wild ramification the different contributes a larger term. The appeal to relative Abhyankar’s lemma (Proposition 2.25) only gives tameness if the cover of the geometric generic fibre is already tame, and that hypothesis is not present in Proposition 2.26. Since Proposition 2.26 is used in Step 2 of Theorem 4.18 to ensure that the reduced subschemes of the pullback of the boundary are étale over W, this gap directly affects the D^+ case and the verification of motivic alignment in Corollary 4.29. The authors should either add an explicit tameness hypothesis and verify it in the applications, or replace the Riemann–Hurwitz argument with a correct one that handles wild ramification.","section":"§2.8, Proposition 2.26, equations (10)–(16)"},{"comment":"Theorem 4.2 is stated for an arbitrary point s ∈ M with residue field k := k(s), which includes points of the generic fibre over Q_p. The proof, however, only treats characteristic p: Corollary 4.19 and Corollary 4.29 assume k is a perfect field of characteristic p, and the entire D^- construction in §4.7 uses crystalline cohomology and Frobenius in an essential way. As written, the theorem would claim the Tate conjecture for all such surfaces over number fields, which is not established by the arguments in the paper. The statement should be restricted to points lying over the special fibre M_{F_p} (or an equivalent characteristic-p hypothesis), with the characteristic-zero case either removed or treated separately.","section":"Theorem 4.2 and §4.7"},{"comment":"Proposition 4.20 is load-bearing for Theorem 4.2(ii), the non-supersingular case, but its proof is only a sketch: it says that the Hodge bundle λ is positive on orthogonal Shimura varieties and that the height stratification has the same properties as in [GK00, Thm. 15.1] by “suitably adapting” the K3-surface arguments. No details are given for the general orthogonal Shimura variety, and the proportionality λ = Q_+·λ_A is asserted without proof. Since the headline applications (Theorems 1.1 and 1.2) proceed through condition (iii) rather than (ii), this is not fatal to those theorems, but it is a gap in Theorem 4.2 as stated. The proof should be completed or the proposition should be replaced by a precise citation of a general result.","section":"§4.6, Proposition 4.20"},{"comment":"Lemma 3.10 is the key new step that replaces Blasius’ theorem for the matching of de Rham and p-adic realizations, and its proof is only sketched. In particular, the application of Lemma 3.9 to an arbitrary basis of rational classes requires that the vectors lie in the open ball produced by Lemma 3.9 and that there exists a point with a Hodge class whose Kodaira–Spencer map is surjective; neither point is justified in the text. The argument can likely be repaired using density of rational vectors and the fact that h^{2,0}=1, but as written the compatibility between α_dR and α^†_{dR,C} is not fully established. This matters because Proposition 3.11 and the boundary extension of α_crys in the D^- case depend on it.","section":"§3.3, Lemma 3.10"}],"minor_comments":[{"comment":"The proof of Proposition 4.25(a) uses the geometric connectedness of |τ^{-1}(Q)|, but this is not guaranteed for an arbitrary proper morphism from a surface to a curve; one should pass to the Stein factorization (or state an additional connectedness assumption) before invoking [Sta25, Tag 0AY8].","section":"§4.6, Proposition 4.25 and Theorem 4.28"},{"comment":"The notation “V_{π(t)} = V ×_S {π(t)}” in case (c) is confusing: π at that point is not the map to the Shimura variety, and the fiber should be over the relevant point of S or of U. Also, “an étale neighborhood of the point t” should be “an étale neighborhood of τ(t)”. These are presentation issues, but they make the argument hard to check.","section":"§4.7, Theorem 4.28(c)"},{"comment":"In the proof of Lemma 3.10, the phrase “the isomorphism α_B,s sends c1,p(ξ) to cl_p(ζ)” should read “α_B,s sends cl_p(ζ) to c1,p(ξ)”, since α_B maps the Shimura side to the cohomology side. The intended meaning is clear from the surrounding diagram, but the typo is confusing.","section":"§3.3, Lemma 3.10"},{"comment":"The notation M is overloaded: M denotes the smooth moduli scheme over Z_(p), its complement H, and also the projective compactification M. Consider using a different symbol for the projective bundle P(V_4 ⊕ V_6 ⊕ O) and its open subsets.","section":"§4.8, Construction 4.31"},{"comment":"The footnote about the cases p = 5, 7 is terse. It would be helpful to state explicitly that the three inseparable cases of Proposition 2.23 are excluded unless the purity hypothesis of Remark 4.35 holds, and to indicate whether Theorems 1.1 and 4.33 are meant to include those cases.","section":"Theorem 1.1 and Remark 4.35"}],"recommendation":"major_revision","confidential_remarks":"The paper is ambitious and, if correct, contains results of significant interest. My main concern is that the proof of Proposition 2.26 is incorrect for wild ramification, and this step is used in the D^+ argument of Theorem 4.18 and in Corollary 4.29. The authors should also restrict Theorem 4.2 to characteristic p or add a separate treatment of characteristic-zero points; as written the theorem overclaims. Proposition 4.20 and Lemma 3.10 need to be filled in before the general theorem can be considered established. These issues appear fixable, but they require real work, so I recommend major revision rather than acceptance."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"You should read this before the next seminar. Guo and Yang prove unconditional Tate conjecture for a large class of surfaces of geometric genus one with singular natural models, and as a corollary BSD for height-one elliptic curves over genus-one function fields in characteristic p >= 11. Both are new, and the architecture is coherent. The key new idea is the extended period morphism over singular fibers and the principle that any pi-exceptional curve is tau-exceptional. That is a genuine contribution, not a repackaging.\n\nThe paper is also honest about where it departs from prior work. It avoids Blasius by matching crystalline and de Rham realizations through the p-adic Riemann-Hilbert functor; Lemma 3.10 is the hinge. The examples in Section 4.8 verify the hypotheses of Theorem 4.2 in enough detail to be convincing.\n\nSoft spots: the D^- case is the only place singular models truly matter, and the proof there depends on the crystalline matching extending over the singular boundary. The compatibility in Lemma 3.10 is proved by analytic density of Noether-Lefschetz loci; it is plausible, but the argument is compressed and the paper itself notes it is addressing something close to [DLLZ23, Conj. 1.4]. Proposition 4.20 also adapts Maulik and van der Geer-Katsura without full details. These are not fatal, but they are the spots to check first. The p=5,7 supersingular cases remain conditional on a Newton purity conjecture, which is stated clearly.\n\nOverall this is a serious paper with a load-bearing technical step that deserves careful refereeing. The main theorems are significant enough that refereeing should happen, and a good referee will either tighten Lemma 3.10 or find a real gap. I would send it out.","headline":"A serious, detailed proof of new Tate/BSD cases that hinges on a delicate crystalline matching over singular boundaries; deserves a careful referee.","tokens_in":57879,"tokens_out":2050,"would_cite":true,"duration_ms":25550,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["14C25","14F30","14G17","14J27","11G40"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves the Tate conjecture for the minimal resolutions of surfaces of geometric genus one whose natural models are singular, and derives the Birch–Swinnerton-Dyer conjecture for every height-one elliptic curve over a genus-one…","keywords":["Tate conjecture","Birch–Swinnerton-Dyer conjecture","geometric genus one","Kuga–Satake","period morphism","rational double points","F-crystals","positive characteristic"],"falsifier":"Construct a constant family $X_0 \\times C$ where $X_0$ has a rational double point and $C$ is a smooth curve, and check whether every simultaneous resolution is isomorphic to the constant resolution $\\breve{X}_0 \\times C$; a nonconstant simultaneous resolution in characteristic $p \\ge 5$ would refute the paper's Corollary 2.18 and, with it, the argument at the worst singular locus.","tokens_in":56805,"feed_emoji":"📐","tokens_out":13041,"duration_ms":114563,"temperature":0.7,"pith_summary":"This paper aims to prove the Tate conjecture for divisor classes on smooth projective surfaces of geometric genus one, in the cases where the natural or canonical model of the surface is singular. Its central claim is that the period-morphism strategy based on the Kuga–Satake construction, previously confined to smooth families such as K3 surfaces, can be extended across the singular locus of the moduli space as long as the singularities are only rational double points and the moduli base satisfies two mild genericity conditions. A sympathetic reader would care because the payoff is concrete and unconditional: every elliptic curve of height one over a global function field of genus one and characteristic $p \\ge 11$ satisfies the Birch–Swinnerton-Dyer conjecture, and every minimal surface with $p_g=1$, $K^2=1$, and $q=0$ over a finitely generated field of characteristic $p \\ge 5$ satisfies the Tate conjecture.","feed_headline":"Height-one elliptic curves obey BSD over genus-one function fields","feed_subtitle":"The new period machinery tolerates singular natural models, yielding unconditional BSD for height-one curves.","key_machinery":"The central object is the extended period morphism (Definition 4.3): a map from a normal scheme $T$ over the moduli space $M$ to the Kuga–Satake Shimura variety $S$ that lifts the ordinary period morphism on the smooth locus and is defined only over subschemes on which the family of surfaces admits a simultaneous resolution. The argument is carried by four mechanisms: rigidity of F-crystals, so that crystalline matching isomorphisms extend over the boundary; simultaneous resolution for rational double point singularities; a gluing lemma for projective schemes; and the minimal model program for threefolds in characteristic $p \\ge 5$, whose flop decomposition yields the key fact that any curve contracted by the projection to the moduli space is also contracted by the extended period morphism.","core_discovery":"On the paper's own terms, the discovery is that the singularities of the natural model are not merely tolerated but are structurally necessary: the exceptional divisors on the resolution are themselves sources of algebraic classes, and the singular fibers are what create the geometric monodromy that makes the period map nontrivial. The proof constructs an extended period morphism (Definition 4.3) that agrees with the ordinary Kuga–Satake period morphism on the smooth locus and extends over the discriminant locus after a simultaneous resolution of the family. It then shows, through rigidity of F-crystals and a flop decomposition of birational maps from the minimal model program, that any curve contracted by the projection to the moduli space is also contracted by the extended period morphism; this uniqueness of periods is what allows special endomorphisms of the associated abelian variety to be compared with divisor classes on the resolution, completing the reduction of the Tate conjecture to a known Tate theorem for special endomorphisms.","pith_inferences":["A natural next test is to run the same extended-period programme on other families listed as amenable in the paper, such as canonical models of surfaces with $p_g=1$ and $q=1$; the paper indicates only routine modifications are needed.","The bottleneck for the remaining $p=5,7$ cases is the purity of the Newton stratification for resolutions of this family, a question the paper raises explicitly; a purity theorem there would close the last BSD gap.","The p-adic Riemann–Hilbert matching used here may work more broadly as a tool to compare crystalline and de Rham realizations in Kuga–Satake-type arguments without assuming the underlying motives are abelian, which could be useful for other Shimura-variety constructions."],"forward_implications":["Over a global function field of genus one and characteristic $p \\ge 11$, every elliptic curve of height one satisfies the Birch–Swinnerton-Dyer conjecture (Theorem 4.33).","A minimal smooth projective surface over a finitely generated field of characteristic $p \\ge 5$ with $p_g=1$, $K^2=1$, and $h^{1,0}=h^{0,1}=0$ satisfies the Tate conjecture (Theorem 1.2).","Under the general hypotheses of Theorem 4.2, the minimal resolution of any fiber of a family of surfaces of geometric genus one satisfies the Tate conjecture whenever its Hodge diamond matches the generic fiber or it is non-supersingular.","The method covers singular fibers of type $\\mathrm{IV}^*$ with $E_6$ singularities even when $p=5$ divides the order of the Weyl group, a case where local monodromy results fail (Example 4.34).","If Newton-stratification purity holds on the moduli space, the remaining $p=5,7$ exceptional inseparable $j$-invariant cases for BSD would also follow (Remark 4.35)."],"supporting_citations":[{"why":"It supplies the Kuga–Satake period-morphism method and the reduction of the Tate conjecture to special endomorphisms of the associated abelian variety.","marker":"[Mad15]"},{"why":"It establishes the period morphism and matching isomorphisms on the smooth locus in positive characteristic and proves the Tate conjecture for smooth fibers; the present paper extends this framework to singular fibers.","marker":"[HYZ25]"},{"why":"It provides simultaneous resolutions for families of surfaces whose fibers have at worst rational double point singularities, the foundation for the extended period morphism.","marker":"[Art74a]"},{"why":"It provides integral canonical models of Shimura varieties and their extension property, which is used to extend period morphisms from mixed-characteristic bases.","marker":"[Kis10]"},{"why":"It is the source of the rigidity theorem for F-isocrystals that lets crystalline matching isomorphisms extend from the smooth locus to the boundary.","marker":"[DK17]"},{"why":"It shows that birational maps between minimal models decompose into flops in characteristic zero, the model for the flop decomposition used in positive characteristic.","marker":"[Kaw08]"},{"why":"It provides the minimal model program for threefolds in characteristic $p \\ge 5$ that makes the flop decomposition and the contraction property valid in this paper's setting.","marker":"[HW22]"},{"why":"It supplies integral p-adic Hodge theory, used to control the p-adic lattices and prove the matching of crystalline and de Rham realizations.","marker":"[BMS18]"},{"why":"It provides the p-adic Riemann–Hilbert functor used to compare the de Rham and crystalline realizations of the period morphism.","marker":"[DLLZ23]"},{"why":"It supplies the equivalence between the Tate conjecture for an elliptic surface and the Birch–Swinnerton-Dyer conjecture for the corresponding elliptic curve in characteristic $p>0$, which converts the main geometric result into a BSD statement.","marker":"[KT03]"}],"fun_headline_variants":["Tate conjecture proven for singular genus-one surfaces","Singularity-tolerant proof yields BSD for height-one curves","Extended period map proves Tate for singular models","Singular natural models conquered for Tate, then BSD","Singularities necessary: Tate proof embraces them"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof for the most singular fibers depends on the claim that any curve the moduli projection collapses is also collapsed by the period map, a claim proved via flop decompositions and constancy of simultaneous resolutions; if that claim fails for a surface in the stated class, the strategy of giving degenerate fibers unique periods would collapse.","fun_headline_variants_meta":{"raw":{"variants":["Tate conjecture proven for singular genus-one surfaces","Singularity-tolerant proof yields BSD for height-one curves","Extended period map proves Tate for singular models","Singular natural models conquered for Tate, then BSD","Singularities necessary: Tate proof embraces them"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000833,"raw_usage":{"total_tokens":3557,"prompt_tokens":785,"completion_tokens":2772,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":401,"completion_tokens_details":{"reasoning_tokens":2699}},"tokens_in":401,"tokens_out":2772,"duration_ms":17653,"temperature":1.0,"reasoning_tokens":2699,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-09T23:03:28.300833+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Construct a constant family $X_0 \\times C$ where $X_0$ has a rational double point and $C$ is a smooth curve, and check whether every simultaneous resolution is isomorphic to the constant resolution $\\breve{X}_0 \\times C$; a nonconstant simultaneous resolution in characteristic $p \\ge 5$ would refute the paper's Corollary 2.18 and, with it, the argument at the worst singular locus.","supporting_citations":[],"review_version":1}