{"id":"0818a5fb-a344-4aab-aa08-6519b516a36f","arxiv_id":"2501.09578","paper_version":2,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For hyper-Kähler fourfolds of K3[2] type with Picard rank two and two divisorial contractions, the pair of conic-bundle types must be one of seven combinations; for Fano varieties of cubic fourfolds, only four combinations occur.","lead":"This paper classifies the pairs of exceptional-divisor contractions that can occur on four-dimensional hyper-Kähler manifolds of K3[2] type with Picard rank two: exactly seven combinations are possible. For Fano varieties of cubic fourfolds only four combinations occur, and the paper provides explicit examples.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The central classification in Theorem 3.1 depends on prior embedding classifications in [vGK], and one cited divisibility computation in the B1 case appears to need care; a direct re-derivation of the B1 embedding normal form is the most load-bearing check.","rationale":"The paper's main theorem is a clean, well-structured reduction to Pell equation data and the five-type contraction classification. Proposition 2.4 is internally sound: the argument that (a,b) must be positive and minimal is correct, and the resulting formulas H' = aH - bd*tau and tau' = bH - a*tau are consistent. The possible pairs listed in Theorem 3.1 follow from Table 1's divisibility/congruence data together with the explicit embedding forms. I checked several of the divisibility computations and they work: for example in case (e) with the M3 form H = 2(1,(d+1)/4)_1 + delta, the condition that tau' = bH - a*tau has divisibility 2 is that a is even, and the proof's parity argument is correct. The reader identified exactly the same weakness: the completeness/correctness of the five-type list and the embedded normal forms quoted from [vGK]. That is a genuine load-bearing assumption for Theorem 3.1. However, it is an external classification and, as far as can be seen from this text, it is not contradicted by any internal evidence. Table 1 itself is consistent with the examples and the later sections. Section 4's specialization to Fano fourfolds is also consistent with the known cases in the literature (e=14, e=16 via the BOSS bundle, e=22 Debarre-Voisin), and in fact those concrete coincidences provide independent support for the five-type classification. The e=30 example depends on a conjecture from [H], but the authors clearly flag that; it is not load-bearing for the central theorem. For these reasons the reader's ACCEPT at moderate confidence is reasonable. The one concrete place where an error would be most damaging is the B1 embedding in case (c), because the type H+B1 is one of the two mixed cases that would not otherwise be visible; hence the proposed check focuses there. No change of verdict is required; the paper should be accepted, with the caveat that a full independent verification of the [vGK] classification would further strengthen confidence.","tokens_in":19470,"tokens_out":2230,"duration_ms":19172,"concrete_test":"Independently recompute the B1 embedding from [vGK, Prop. 3.5] given in the proof of Theorem 3.1(c): write L = U^3 + E8(-1)^2 + <delta>, q(delta) = -2; set H = (1,d)_1, tau = (1,-d)_1 + 2(1,d/4)_2 + delta, assume d ≡ 0 mod 8 with the embedding read as a divisibility-compatible representative of the transcendental lattice T(S,alpha). Verify directly: q(tau) = -2, div(tau) = 1, div(H) = 1, and the perpendicular transcendental lattice has the discriminant form of T(S,alpha) from [vGK, Prop. 3.5]. Then verify that tau' = bH - a*tau has divisibility 1 if and only if b is odd, by computing the pairing of tau' with the dual basis / test elements of L; if this fails for some a,b, recalculate the H+B1 condition. This single check would settle whether the B1 case of Theorem 3.1 rests on a correct embedding.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Theorem 3.1 is the paper's central claim, and its proof in cases (c) and (e) reduces the type of the second extremal ray to the divisibility of tau' = bH - a*tau, computed from explicit embedding representatives quoted from [vGK, Props. 3.4, 3.5, 3.8]. The weakest point is not internal to the Pell argument (Prop. 2.4), which is solid, but the reliance on the completeness and correctness of the five-type classification of embeddings of <2d> + <-2> into Lambda_K3[2], together with those explicit representatives. In particular, case (c) uses the B1 representative H = (1,d)_1, tau = (1,-d)_1 + 2(1,d/4)_2 + delta; one needs to verify that the given vector really has q = -2 and divisibility 1, and that the stated condition 'b odd' exactly captures when tau' has divisibility 1. The paper quotes these facts rather than reproving them, and [vGK] is co-authored by the second author. The second extremal ray is assumed to be a divisorial contraction by the hypothesis that |det(Pic(X))| is not a square, which is also external to this paper. Since the theorem's conclusion (the seven-case list) is exactly as strong as the five-type input, an error in that input would change the list.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper classifies the possible types of the two extremal rays of the movable cone of a projective hyper-Kähler fourfold of K3[2] type with Picard rank two, under the assumptions that the fourfold admits a divisorial contraction and that the determinant of the Picard lattice is not a square. Theorem 3.1 shows that the pair of types is one of M1+M1, B0+B0, H+H, B1+B1, H+B1, M3+M3, or H+M3, with parity conditions on the minimal Pell solution governing the mixed cases. The proof reduces each case to divisibility computations in explicit lattice embeddings quoted from Debarre–Macrì and van Geemen–Kapustka. The paper then specializes to Hilbert squares of K3 surfaces and to Fano varieties of cubic fourfolds, providing tables of examples.","tokens_in":19749,"tokens_out":29887,"duration_ms":265629,"significance":"The result is a useful, explicit classification with a clear computational structure. The reduction of the geometry to the minimal solution of a Pell equation is elegant, and the tables for S[2] and Fano fourfolds will be a convenient reference. I checked the load-bearing computations in the proof of Theorem 3.1, including the B1 embedding τ = (1,-d)_1 + 2(1,d/4)_2 + δ: since the two copies of U are orthogonal, q(τ) = -2d + 2d - 2 = -2, and the parity criterion for the divisibility of τ' = bH - aτ is correct. The main caveat is that the classification is conditional on the completeness and correctness of the five-type classification and the embedding normal forms in [vGK] and [DM]; these are cited but not reproved here. That dependency is explicit and does not, in my assessment, affect the internal soundness of the paper.","major_comments":[],"minor_comments":[{"comment":"In the row e = 74, the entry (6,1) in the (a,b) column solves x^2 - 37 y^2 = -1, not the positive Pell equation x^2 - 37 y^2 = 1; the minimal positive solution is (73,12). Since no contractions occur for e = 74, the entry is not used in the analysis, but it should be corrected or clearly labeled as a negative-Pell solution.","section":"Table 3"},{"comment":"The phrase \"There are exactly seven cases\" could be read as asserting that each of the seven pairs occurs. Theorem 3.1 establishes that these are the only possible pairs; if realizability of every case is not intended, I suggest rephrasing to \"at most seven cases\" or adding a sentence clarifying which cases are known to occur.","section":"Abstract and Introduction"},{"comment":"The proof uses the symmetry (x,y) ↦ (x,-y) of the Picard lattice and concludes that the other extremal ray has the same type. It should be stated explicitly that this conclusion concerns the situation where both extremal rays are divisorial contractions; for e ≡ 0 mod 6 this is automatic when a (-2)-class exists (as shown in the proof of part (d)), but the point is not stated.","section":"Theorem 4.4(e)"},{"comment":"There is a typo in \"equvalently\", which should be \"equivalently\".","section":"Section 4.4(d)"},{"comment":"The expression \"H62(X,Z)\" is a typo for \"H^2(X,Z)\".","section":"Section 1.5"},{"comment":"The notation \"Pic( X) ⁄= ZH ⊕ Zτ\" should use the standard symbol \"≠\".","section":"Section 1.6"},{"comment":"The heading \"Divisorial contractions on a Hilbert square of S[2]\" is redundant; it should be \"Divisorial contractions on S[2]\".","section":"Table 2"},{"comment":"In the proof, the display \"τ 3H′\" should be typeset as τ^3 H′ for clarity.","section":"Proposition 2.4"}],"recommendation":"minor_revision","confidential_remarks":"The paper is sound and clearly written; the only significant dependency is on prior classification work of Debarre–Macrì and van Geemen–Kapustka, which is cited openly. I verified the potentially delicate B1 embedding computation in the proof of Theorem 3.1 and found it correct. The remaining issues are local presentation matters, including a small error in Table 3 for e = 74."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Bert — the short version: this paper does what it says. It classifies all possible pairs of extremal rays that define divisorial contractions on a K3[2] fourfold with Picard rank two and non-square discriminant. The list is short — seven cases — and the proof is explicit. I went through the delicate parity step in case (c) of Theorem 3.1 and it checks out; the B1 embedding normal form quoted from [vGK] gives the claimed divisibility condition.\n\nWhat is genuinely new here is the pair classification, not the individual contraction types. The five types were already in [vGK] and Debarre–Macrì. The new content is the seven-case list, the parity conditions that govern the mixed cases H+M3 and H+B1, and the restriction to four pairs for Fano varieties of cubic fourfolds. Tables 2 and 3 are concrete and consistent with the Pell data — I spot-checked several rows. For the Hilbert square table, the conditions reduce to parity of b, and the e=16 example (H+B1, BOSS bundle) is a nice touch.\n\nThe main soft spot is the same one the reader flagged: Theorem 3.1 takes as input the completeness of the five-type classification of embeddings of <2d> + <-2> into Λ_K3[2], quoted from [vGK], a paper co-authored by the second author. That is a real dependency, but it is not a hidden one — the authors state it plainly and [vGK] is published and peer-reviewed. The present paper does not reprove the embedding classification, which is fine; it would be a lot of extra length for little gain. The e=30 example depends on an external conjecture, but the authors flag it and it is not needed for the main theorem. Minor complaint: the wording around Theorem 4.4(d) is a bit tangled, but the argument is clear enough.\n\nI did not find a load-bearing error. The Pell equation argument in Proposition 2.4 is clean, and the use of [BM] for the movable cone is standard. The paper is aimed at people working on hyperkähler birational geometry or cubic fourfolds; they will want these tables.\n\nRecommendation: send to a serious referee. This deserves a proper peer review, and it should be accepted after minor revision — the only real request I would make is that the dependence on [vGK] be stated even more explicitly in the introduction, and that the e=30 conjecture dependence be kept visibly separate, which it already is.","headline":"A complete, explicit classification of pairs of divisorial contractions on K3[2] fourfolds; the main theorem is sound and the tables are useful.","tokens_in":20273,"tokens_out":4591,"would_cite":true,"duration_ms":43279,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["14J42","14J28","14E30","11D09"],"pacs":[],"model":"deepseek-v4-flash","headline":"A Hyper-Kähler fourfold of K3[2] type with two divisorial contractions has one of only seven possible pairs of contraction types.","keywords":["Hyper-Kähler fourfolds","K3[2] type","divisorial contractions","conic bundles","movable cone","Pell equation","Fano variety of cubic fourfold","lattice embeddings"],"falsifier":"Exhibit, or find by lattice search, a projective K3[2]-type fourfold of Picard rank two with a divisorial contraction, $|\\det(\\mathrm{Pic}(X))|$ not a square, and a pair of extremal-ray types not among M1+M1, B0+B0, H+H, B1+B1, H+B1, M3+M3, H+M3 — for instance H+M1 — or an H+B1 pair with $d \\equiv 0 \\pmod 8$ and $b$ even; either would refute Theorem 3.1.","tokens_in":19261,"feed_emoji":"🧮","tokens_out":7692,"duration_ms":68423,"temperature":0.7,"pith_summary":"This paper answers, for Hyper-Kähler fourfolds of K3[2] type with Picard rank two, which pairs of divisorial contractions can coexist. The contractions contract conic bundles over K3 surfaces, and prior work lists five types, conventionally H, M1, M3, B0, and B1. The paper proves that when a fourfold has two contractions and $|\\det(\\mathrm{Pic}(X))|$ is not a square, the pair must be one of exactly seven combinations, not the a priori fifteen. Five of these are the double types T+T, and the only mixed pairs are H+M3 and H+B1, with parity conditions dictating when the mixed pairs occur. The result pins down the second extremal ray by a single Pell equation and gives the concrete lists for Hilbert squares and Fano varieties of cubic fourfolds.","feed_headline":"Only seven type-pairs for two contractions on a K3[2] fourfold","feed_subtitle":"A Pell equation and two divisibility numbers decide which of the five conic-bundle types can pair up.","key_machinery":"The argument runs on the rank-two Picard lattice $\\mathrm{Pic}(X) = \\mathbb{Z}H \\oplus \\mathbb{Z}\\tau$ with $q(H) = 2d$, $q(\\tau) = -2$, whose primitive embedding into the cohomology lattice $\\Lambda_{\\mathrm{K3}^{[2]}}$ has five isometry classes. Each class is a type of exceptional divisor, a conic bundle over a K3 surface: H, M1, M3, B0, B1. The second extremal ray is forced by the minimal positive solution $(a,b)$ of the Pell equation $x^2 - d y^2 = 1$: the other contraction is defined by $H' = aH - bd\\tau$ and $\\tau' = bH - a\\tau$. The divisibilities of $\\tau$ and $\\tau'$ in $H^2(X,\\mathbb{Z})$, together with the residue of $d$ modulo 4 and the parity of $a$ and $b$, select which of the five types the two rays have. A proposition drawn from [Dr] certifies that any conic bundle embedded in a fourfold appears as the exceptional fiber of a divisorial contraction, which is how the tables for Hilbert squares and Fano fourfolds are read.","core_discovery":"Let $X$ be a projective Hyper-Kähler fourfold of K3[2] type with Picard rank two, admitting a divisorial contraction, and with $|\\det(\\mathrm{Pic}(X))|$ not a square. Then the movable cone has two extremal rays that both define divisorial contractions, and the pair of types is exactly: M1+M1 when $|\\det(\\mathrm{Pic}(X))| \\equiv 1 \\pmod 4$; B0+B0 when the discriminant group of the transcendental lattice is not cyclic; and, when that group is cyclic with $|\\det(\\mathrm{Pic}(X))| = 4d$, H+H, B1+B1, M3+M3, H+B1, or H+M3 according to $d$ modulo 4, the divisibilities of the two $(-2)$-classes, and the parity of the minimal Pell solution $(a,b)$. The mixed pair H+B1 occurs exactly when $d \\equiv 0 \\pmod 8$ and $b$ is odd, and H+M3 occurs exactly when $a$ is even. The theorem turns the apparent fifteen possibilities into seven.","pith_inferences":["Editorial extension: the same Pell-equation mechanism likely extends to contractions on higher-dimensional K3$^{[n]}$-type manifolds with Picard rank two, though the conic-bundle classification used here is specific to fourfolds.","Editorial extension: the parity criteria suggest a closed congruence rule for Hilbert squares, where the second type should change with $e \\bmod 8$ and the parity of the minimal Pell solution; checking all admissible $e$ would be a quick computational confirmation.","Editorial extension: the seven-case list can be stress-tested by an exhaustive lattice search over primitive rank-two sublattices of $\\Lambda_{\\mathrm{K3}^{[2]}}$ using the five embedding normal forms, recomputing each second ray from the Pell solution; any pair outside the list would expose a missing case."],"forward_implications":["For a Hilbert square $S^{[2]}$ with $S$ of degree $e$, one contraction is always of type H, and the second ray's type is H, M3, or B1; the paper's Table 2 lists these types for $e = 2,\\dots,32$.","For Fano varieties of cubic fourfolds with rank-two Picard lattice, only four pairs occur: H+H, H+M3, M3+M3, and B1+B1.","In the mixed cases H+M3 and H+B1, the fourfold is birational to a Hilbert square $S^{[2]}$ for a K3 surface $S$.","When the determinant condition forces an isotropic second ray instead, the fourfold has a Lagrangian fibration rather than a second contraction; the non-square assumption in the theorem is exactly what rules this out.","In the degree-14 cubic fourfold case, the second exceptional divisor of type M3 is identified with the known scroll over the K3 surface, and the paper presents this identification as new."],"supporting_citations":[{"why":"Supplies the classification of primitive embeddings of $\\langle 2d\\rangle \\oplus \\langle -2\\rangle$ into $\\Lambda_{\\mathrm{K3}^{[2]}}$, which yields the five contraction types.","marker":"[DM]"},{"why":"Gives the five embedding classes and the description of movable and nef cones used to identify extremal rays.","marker":"[De]"},{"why":"Supplies the classification of conic bundles and the explicit embedding normal forms used for the parity checks in Theorem 3.1.","marker":"[vGK]"},{"why":"Establishes that a contraction of a moduli space of twisted sheaves induces a conic bundle structure on the exceptional divisor.","marker":"[BM]"},{"why":"Provides the result that an isotropic extremal ray gives a Lagrangian fibration, so the non-square determinant assumption forces the second ray to be a divisorial contraction.","marker":"[Ma2]"},{"why":"Gives the criterion used in Proposition 1.11 that an embedded conic bundle with negative fiber class is the exceptional divisor of a contraction.","marker":"[Dr]"},{"why":"Classifies the rank-two lattices of special cubic fourfolds and the admissibility conditions used in Section 4 to restrict the four Fano pairs.","marker":"[Ha00]"}],"fun_headline_variants":["Two rays on K3[2] fourfold: exactly seven type-pairs","Seven conic-bundle type-pairs for two contractions","K3[2] fourfold contractions: from fifteen pairs to seven","Two extremal rays: only seven possible type combinations","Pell equation decides seven type-pairs on Hyper-Kähler"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole list rests on the earlier five-type classification of embeddings of $\\langle 2d\\rangle \\oplus \\langle -2\\rangle$ into $\\Lambda_{\\mathrm{K3}^{[2]}}$ being complete and on the explicit embedding representatives quoted from [vGK] being correct; if a sixth embedding type exists or one representative is wrong, a pair could be missing from the seven cases.","fun_headline_variants_meta":{"raw":{"variants":["Two rays on K3[2] fourfold: exactly seven type-pairs","Seven conic-bundle type-pairs for two contractions","K3[2] fourfold contractions: from fifteen pairs to seven","Two extremal rays: only seven possible type combinations","Pell equation decides seven type-pairs on Hyper-Kähler"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000282,"raw_usage":{"total_tokens":1640,"prompt_tokens":889,"completion_tokens":751,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":505,"completion_tokens_details":{"reasoning_tokens":661}},"tokens_in":505,"tokens_out":751,"duration_ms":7775,"temperature":1.0,"reasoning_tokens":661,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T19:53:00.172052+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Exhibit, or find by lattice search, a projective K3[2]-type fourfold of Picard rank two with a divisorial contraction, $|\\det(\\mathrm{Pic}(X))|$ not a square, and a pair of extremal-ray types not among M1+M1, B0+B0, H+H, B1+B1, H+B1, M3+M3, H+M3 — for instance H+M1 — or an H+B1 pair with $d \\equiv 0 \\pmod 8$ and $b$ even; either would refute Theorem 3.1.","supporting_citations":[],"review_version":1}