{"id":"2ee6b509-2dfa-4cd7-ae91-8cc010934c9f","arxiv_id":"2412.14857","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Existence of standard models for degree 4 del Pezzo fibrations over curves is proved in characteristic not 2, using semistable pencils of quadrics and terminal singularity analysis.","lead":"This paper proves that every smooth del Pezzo surface of degree 4 over the function field of a curve, in any characteristic other than 2, has a standard model fibration with only terminal singularities. It extends a complex-number result to positive characteristic using Kollár stability on pencils of quadrics.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The terminality proof rests on sketched case analyses in Lemma 5.3 and Proposition 5.4; an unverified coefficient configuration could admit a semistable model with a non-du Val elephant.","rationale":"The reader's weakest-assumption analysis correctly identifies the completeness of the singularity configurations and the correctness of the coordinate normalizations as the load-bearing part of the proof. My stress-test pass found no independent internal contradiction or obvious counterexample, but it did locate the precise places where the argument is only sketched: Lemma 4.4 Cases (5) and (6), Lemma 5.3's normal-form reductions, and Proposition 5.4's final cD classification. These are exactly the points on which Theorem 1.5 depends: a missed configuration could allow a semistable model whose special fibre has a non-normal double curve or a non-isolated elephant, which would destroy the terminality conclusion. Because the paper is conditional on these verifications, and because the omissions are concrete and checkable rather than fatal, the appropriate verdict remains CONDITIONAL. The proposed computational test over k[[t]]/(t^N) is finite and would settle whether any unlisted configuration survives the semistability inequalities, or whether all branches are ruled out as claimed. The sign typo in Proposition 3.7 was noted but does not affect the logic of the ideal-chain termination. Overall, the central claim is plausible and the strategy is sound; the remaining work is to supply the missing coefficient-level checks or a machine-assisted verification of them.","tokens_in":23838,"tokens_out":43790,"duration_ms":357011,"concrete_test":"Run a symbolic coefficient verification over R = k[[t]]/(t^N), N ≥ 4, for each LPS normal form in Lemma 4.4 and for each Jacobian-rank case in Lemma 5.3. Impose the singularity-along-the-specified-locus equations, solve for the t-adic divisibilities of the λ_{ij} and μ_{ij}, and then test every effective weight system with entries in {0,...,4} against the semistability inequality (3.1). In particular, check Lemma 4.4 Cases (5) and (6), Lemma 5.3 Case 1-B-a and Case 2, and the final degree-≤3 computation of Proposition 5.4. If any branch yields a solution satisfying all semistability inequalities, that solution is a counterexample to Proposition 5.4 and to Theorem 1.5; if none does, the omitted calculations are confirmed.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim of Theorem 1.5 requires every semistable degree-4 del Pezzo threefold to have only terminal singularities. The proof reduces this to a finite singularity analysis, but the analysis is not fully written out at several load-bearing points. In Lemma 4.4, Cases (5) and (6) are dismissed with \"one can show\" that certain coefficients satisfy t^2-divisibility as in Case (4); a failure or missed subcase would allow a non-normal central fibre and break regularity in codimension 2. In Lemma 5.3, the reduction of Case 1-B-a to the two normal forms {fE=fE^(2)(x1,x2), gE=gE^(2)(x1,x2)+x1x3} and {fE=fE^(2)(x1,x2), gE=gE^(2)(x1,x2)+x1x5}, and the reduction of Case 2 to (5.6)/(5.7), are asserted without proof; a missed Jacobian-rank configuration or an invalid coordinate normalization could leave an elephant singular along a curve through P, contradicting Lemma 5.3. In Proposition 5.4, the final cD classification relies on the degree-3 part being \"not a cube\" and on coefficient divisibilities such as t^2|λ_{4,5}; these are verified only by displayed partial computations. The paper also contains a sign typo in Proposition 3.7 (\"divisible by tD\" is reversed), although the ideal-chain direction is consistent. These are not internal contradictions, but the omitted coefficient-level checks are the load-bearing assumption: if any of them fails, a semistable model could have a non-terminal singularity.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves that, over an algebraically closed field of characteristic different from 2, every smooth del Pezzo surface of degree 4 over the function field of a smooth curve admits a standard model: a terminal threefold fibration with integral fibres and relatively ample anticanonical bundle. The proof follows the Kollár stability framework of Abban–Fedorchuk–Krylov. After establishing existence of semistable models via a discriminant-valuation argument on the Grassmannian of pencils of quadrics, the paper analyzes the possible singularities of the resulting complete intersections. It shows the central fibre is integral, the threefold is regular in codimension 2, and every singular point is a hypersurface singularity. The main technical work is a case analysis, using the classification of non-normal intersections of two quadrics and an elephant argument, to show that every singularity is of type cA or cD, hence terminal.","tokens_in":24189,"tokens_out":6146,"duration_ms":50449,"significance":"If the proof is completed as sketched, the result is a genuine advance: it extends Corti's existence theorem for standard models of degree-4 del Pezzo fibrations from characteristic zero to all characteristics different from 2, using a method different from Corti's MMP-based approach. The paper gives explicit semistable reduction for pencils of quadrics via the discriminant, and it makes systematic use of Reid's elephant criterion and the LPS classification of non-normal quadric intersections. The proof is structurally self-contained relative to cited theorems, and it does not assume the target statement. The main risk is not circularity but incompleteness: several coefficient-level case analyses that are load-bearing for the terminality claim are only sketched.","major_comments":[{"comment":"The proof that semistable models are regular in codimension 2 depends on ruling out the six LPS normal forms. Cases (1)–(4) contain explicit coefficient calculations, but Cases (5) and (6) are dismissed with “one can show that t^2 | λ_{i,j} for i,j = 4,5,...,n as in case (4)”. The normal forms in Cases (5) and (6) contain additional monomials (x1x5, x3x4, x2^2) that change the shape of the linear parts on the affine chart x4=1, so the reduction to Case (4) is not immediate. Please supply the missing calculation or a uniform argument that covers these cases.","section":"§4.2, Lemma 4.4, Cases (5) and (6)"},{"comment":"Two coordinate normalizations are asserted without proof. In Case 1-B-a, the pair {f_E, g_E} is said to be transformable into either {f_E = f_E^(2)(x1,x2), g_E = g_E^(2)(x1,x2) + x1x3} or {f_E = f_E^(2)(x1,x2), g_E = g_E^(2)(x1,x2) + x1x5}, while preserving the elephant equation x4 = u t x5 and the normalization (5.1). In Case 2, after assuming the conic lies in the plane x1=0 and f_E = x1^2, the pair is reduced to (5.6) or (5.7). These reductions involve changes of coordinates and linear combinations of the two equations; the constraints imposed by the fixed elephant form and by the semistable normalization are not checked. A missed Jacobian-rank configuration or an invalid normalization could leave an elephant singular along a curve through P. Please provide the linear-algebra details.","section":"§5.1, Lemma 5.3, Case 1-B-a and Case 2"},{"comment":"The cA/cD classification in the rank f^(2)=3, g^(1)=x1 case rests on several coefficient-level assertions that are not fully demonstrated. Specifically: (i) the claim that when (λ_{4,5}/t) ≠ 0 the degree-3 part of the substituted equation is “nonzero and not a cube” for general u; (ii) the claim that if either µ_{2,4} or µ_{4,4} is nonzero then P is a cD-type singularity; and (iii) the final step where all terms of the degree-3 part are said to be divisible by x2, hence not a cube. For (iii), the displayed expression contains many terms involving λ^{(1)}_{i,5}, µ^{(1)}_{5,5}, and λ_{3,4}/t, and after the given coordinate change the coefficient of x2 x4^2 must be shown to be nonzero; also the cD criterion requires verifying that the quadratic part has rank exactly 1. Please expand these computations or give a systematic argument that the stated coefficient conditions are the only ones possible under semistability.","section":"§5.2, Proposition 5.4"}],"minor_comments":[{"comment":"The sentence “D(det(λA′+µB′)) is divisible by tD(det(λA+µB))” has the divisibility direction reversed: from the formula in Lemma 3.6 and (3.4) one obtains that D(det(λA+µB)) is divisible by t D(det(λA′+µB′)), which is what makes the displayed chain of ideals ascending. Please correct the wording.","section":"§3.2, Proposition 3.7"},{"comment":"In the paragraph “If λ_{4,4} is divisible by t^2, for a weight system ρ=(1,1,1,0,0) we have multρ(P) ≤ 2+1 = 3”, the inequality should be ≥, since a lower bound of 3 is needed to exceed the semistability threshold 4/5 · 3 = 12/5.","section":"§5.2, Proposition 5.4"},{"comment":"There are two typographical errors in the notation for weight systems: “ρ = (w1, . . . , mn)” and “diag(w1, . . . , mn)” should both read wn in the last coordinate.","section":"§3.1"},{"comment":"The proof of Lemma 5.2 appeals to Bertini’s theorem via [Kle, 12 Corollary], but the exact statement used (smoothness of a general hyperplane section of the smooth generic fibre) should be spelled out, since the elephant equation involves coefficients a_i + u_i t^{n_i} that are not simply a hyperplane in the ambient projective space but a section of the anticanonical linear system.","section":"§5.1, Lemma 5.2"},{"comment":"In the proof of Lemma 3.6, the intermediate expression D(det((det F)^{2/n}(λA+µB))) uses an n-th root of det F, which need not exist in K; the subsequent equality is correct if one instead uses the homogeneity D(c Φ)=c^{2n-2}D(Φ) with c=(det F)^2. Please rewrite this step to avoid the fractional exponent.","section":"§3.2, Lemma 3.6"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is clearly a serious piece of work and the overall strategy is convincing, but the referee report reflects the fact that the terminality proof—the central new content—is not yet written out at several load-bearing points. The sketched computations in Lemma 4.4 and Lemma 5.3 and Proposition 5.4 are exactly the places where a subtle error could invalidate the main theorem, so they should be filled in before publication. The many small typos suggest the paper would also benefit from a careful proofreading pass."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Here's my read of the Kitagawa paper.\n\nIt proves something real: for any smooth del Pezzo surface of degree 4 over the function field of a curve in char ≠ 2, there is a standard model with terminal singularities and integral fibres. Corti did this in characteristic 0, and AFK did degree 1; degree 4 in positive characteristic was open. The strategy via Kollár stability is well chosen, and the paper does the heavy lifting of showing that semistable (2,2)-complete intersections are integral, regular in codimension 2, and eventually terminal.\n\nThe singularity analysis is the core. Lemma 4.4 rules out non-normal central fibres; Lemma 5.3 shows that a general elephant has an isolated singularity; Proposition 5.4 classifies the resulting hypersurface singularities as cA or cD. The computations in Prop 5.4 are actually written out in considerable detail—this is not a paper that hides behind 'routine'. The structure is sound.\n\nThe soft spots are real but manageable. Several steps in Lemma 4.4 (cases (5), (6)) and Lemma 5.3 (the reduction to the two normal forms in Case 1-B-a, and the reduction to (5.6)/(5.7) in Case 2) are asserted with 'one can show' or 'by some coordinate changes'. These are exactly the steps that ensure no configuration of the singular locus is missed. A referee should ask for these to be expanded. I don't think there is a known counterexample—the claims are plausible and likely true—but the completeness of the enumeration is load-bearing.\n\nThere are also two typos that should be fixed: the divisibility direction in the proof of Prop 3.7 is backwards as written, and in Prop 5.4 the inequality for the destabilizing weight system is written '≤' where it should be '≥'. Neither affects the argument's logic once corrected.\n\nBottom line: if you work on del Pezzo fibrations or positive-characteristic birational geometry, this is worth a careful read. The main result is new, the proof is credible but requires additional detail in a few key places. I'd send it to expert referees and get those calculations written out. I would accept this for peer review and would cite it once the gaps are closed.","headline":"New and plausible result for degree 4 del Pezzo fibrations in char ≠ 2; referee should request the skipped case checks.","tokens_in":24646,"tokens_out":4069,"would_cite":true,"duration_ms":31592,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["14E30","14J45","14J17"],"pacs":[],"model":"deepseek-v4-flash","headline":"Every smooth degree-4 del Pezzo surface over a curve in characteristic >2 extends to a standard model with terminal singularities.","keywords":["del Pezzo fibrations","standard models","Kollár stability","semistable models","terminal singularities","complete intersection of two quadrics","pencils of quadrics","positive characteristic"],"falsifier":"For n=5 over a characteristic-not-2 field, take each of the six normal forms from the LPS classification, lift it to a pencil over R = k[[t]] with a smooth generic fibre, and compute mult_ρ(P) for the weight systems named in Lemma 4.4; respectively, construct the elephants of Lemma 5.3 and test each type of singular locus. If any lift satisfies the semistability inequality mult_ρ(P) ≤ (4/5)∑w_i — that is, if the claimed destabilizing weight system does not actually destabilize — then the proof's enumeration has a gap. A positive result would be to exhibit a semistable degree-4 del Pezzo fibration whose singular point is not cA or cD, which would directly contradict Proposition 5.4.","tokens_in":23654,"feed_emoji":"🔷","tokens_out":8476,"duration_ms":67595,"temperature":0.7,"pith_summary":"The paper establishes that every smooth del Pezzo surface of degree 4 over the function field of a curve, over an algebraically closed field of characteristic not 2, can be extended to a fibration over the curve whose total space is terminal, whose fibres are integral, and whose anticanonical class is ample relative to the base curve — a 'standard model' in the sense of Corti. Corti had proved this over the complex numbers using the minimal model program and Kawamata–Vieweg vanishing; both tools are unavailable in positive or low characteristic, so the paper instead uses Kollár stability. The payoff is that degree 4 del Pezzo fibrations, which are (2,2)-complete intersections of two quadrics in $P^{4}$, now have good birational models in arbitrary characteristic not 2, matching what was already known for degrees 3 and 1. The proof works through semistable models of pencils of quadrics and a classification of the singularities that can appear on their central fibres.","feed_headline":"Standard models exist for degree-4 del Pezzo fibrations","feed_subtitle":"Beyond characteristic 2, degree-4 del Pezzo fibrations admit terminal, integral-fibre models.","key_machinery":"The machinery has three parts. First, Kollár stability for pencils of quadrics: an R-point of the Grassmannian Gr(2,V) of pencils is semistable when, for every weight system ρ=(w_1,...,w_5), the multiplicity mult_ρ(P) computed from Plücker coordinates satisfies mult_ρ(P) ≤ (4/5)∑ w_i. The engine is Lemma 3.6, an identity relating the discriminant D(det(λA+μB)) of a pencil to that of the ρ-transformed pencil: up to units, the discriminant valuation shifts by n(n−1)(−mult_ρ(P)+(4/n)∑ w_i), so any violation of semistability visibly divides the discriminant by t, forcing a noetherian descent to a semistable model. Second, the classification of non-normal (2,2)-complete intersections supplies the six normal forms that a singular central fibre could reduce to, and direct coefficient calculations show each is ruled out by an explicit destabilizing weight system. Third, the elephant method: a general member of |-K_X| through a singular point is an isolated du Val singularity, and the corresponding criterion converts that into terminality of the threefold singularity.","core_discovery":"The central claim is Theorem 1.5: over an algebraically closed field k with char k ≠ 2, for a smooth curve C with function field K, any smooth del Pezzo surface X_K of degree 4 over K admits a standard model π: X → C — that is, X has only terminal singularities, π has integral fibres, and -K_X is π-ample. The key intermediate result is Theorem 1.9: the semistability condition on the pencil of quadrics defining X_K yields a model over the local ring that has integral fibres, is anticanonically relatively ample, is regular in codimension 2, and has only hypersurface singularities. For the threefold case, the paper then shows that every such semistable model has only cA or cD (hence terminal) singularities, using the existence of a general elephant with an isolated du Val singularity.","pith_inferences":["The discriminant-valuation identity of Lemma 3.6 gives a concrete measure of how far a model is from semistability; one could use the most destabilizing weight system to define an explicit algorithmic procedure that terminates at a standard model, though the paper only proves existence via the noetherian chain.","The template — semistability inequality on a parameter space, noetherian descent via a discriminant, then an elephant-based singularity check — should extend to degree 2 and 1 del Pezzo fibrations in positive characteristic, where the required parameter spaces are double covers or weighted hypersurfaces and the stability slope changes with the anticanonical index.","A computer algebra check of the coefficient congruences in Lemmas 4.4 and 5.3 over finite fields of characteristic ≠2 could mechanically certify the case analysis that the paper leaves as 'one can show'."],"forward_implications":["Theorem 1.5 settles Question 1.4 for degree 4 del Pezzo fibrations over algebraically closed fields of characteristic not 2.","Semistable (2,2)-complete intersections of dimension at least two over a DVR, with smooth generic fibre, are regular in codimension 2 and have only hypersurface singularities; in the threefold case the singularities are cA or cD, hence terminal.","Standard models of degree 4 del Pezzo fibrations can be constructed without the MMP or Kawamata–Vieweg vanishing, using only a noetherian descent on the discriminant ideal of the defining pencil.","As the paper notes, these terminal threefolds in positive characteristic give examples relevant to the study of moduli spaces of rational curves on terminal del Pezzo fibrations, previously studied in characteristic zero."],"supporting_citations":[{"why":"Introduces Kollár stability, the notion of D-semistability used throughout, and proves the degree-1 existence theorem that this paper adapts to degree 4.","marker":"[AFK]"},{"why":"Defines standard models of del Pezzo fibrations and proves existence over C for degrees ≥2, establishing the framework and the positive-characteristic question.","marker":"[Cor]"},{"why":"Defines semistability for hypersurfaces over one-dimensional regular schemes and proves existence of semistable models of degree 3 del Pezzo fibrations; the proof of Proposition 3.7 is explicitly modeled on its argument.","marker":"[Ko97]"},{"why":"Classifies irreducible non-normal (2,2)-complete intersections; the six normal forms are the starting point for the destabilization calculations in Lemma 4.4.","marker":"[LPS]"},{"why":"Provides the elephant criterion used to upgrade an isolated du Val elephant singularity to terminality of the threefold.","marker":"[Ko21]"},{"why":"Gives the smoothness criterion for (2,2)-complete intersections via the discriminant having distinct roots, used to identify smooth generic fibres.","marker":"[Re72]"},{"why":"Supplies the normal forms of du Val singularities used to recognize cA/cD singularities on elephants.","marker":"[Re87]"},{"why":"Used to guarantee that general elephants have smooth generic fibre in arbitrary characteristic via a Bertini-type theorem.","marker":"[Kle]"}],"fun_headline_variants":["Degree-4 del Pezzo fibrations get standard models in char >2","Char >2: degree-4 del Pezzo fibrations admit standard models","Standard models exist for degree-4 del Pezzo fibrations in char >2"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the enumerated singular-locus configurations in Lemmas 4.4 and 5.3 are complete, and that each coordinate normalization and coefficient-level calculation summarized as 'one can show' is correct; a missed configuration or a faulty normalization could allow a non-terminal singularity.","fun_headline_variants_meta":{"raw":{"variants":["Degree-4 del Pezzo fibrations get standard models in char >2","Char >2: degree-4 del Pezzo fibrations admit standard models","Standard models exist for degree-4 del Pezzo fibrations in char >2"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000905,"raw_usage":{"total_tokens":3816,"prompt_tokens":789,"completion_tokens":3027,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":405,"completion_tokens_details":{"reasoning_tokens":2961}},"tokens_in":405,"tokens_out":3027,"duration_ms":18813,"temperature":1.0,"reasoning_tokens":2961,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T11:52:01.025007+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"For n=5 over a characteristic-not-2 field, take each of the six normal forms from the LPS classification, lift it to a pencil over R = k[[t]] with a smooth generic fibre, and compute mult_ρ(P) for the weight systems named in Lemma 4.4; respectively, construct the elephants of Lemma 5.3 and test each type of singular locus. If any lift satisfies the semistability inequality mult_ρ(P) ≤ (4/5)∑w_i — that is, if the claimed destabilizing weight system does not actually destabilize — then the proof's enumeration has a gap. A positive result would be to exhibit a semistable degree-4 del Pezzo fibration whose singular point is not cA or cD, which would directly contradict Proposition 5.4.","supporting_citations":[],"review_version":1}