{"id":"410634b7-2e21-4706-874a-9bc7616a293e","arxiv_id":"2412.18830","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A Calabi-Yau pair of complexity two is cluster type exactly when its standard model over a toric base is nodal, component-compatible, and volume-large; this classifies all rank-one Gorenstein del Pezzo surfaces.","lead":"This paper develops criteria to decide when certain Calabi-Yau pairs of complexity two are of cluster type, reducing the question to del Pezzo fibrations over toric varieties. It then proves a complete classification of which Gorenstein del Pezzo surfaces of Picard rank one are of cluster type.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 1.4, the reduction to surface fibrations over toric bases, is imported wholesale from the first author's unpublished [20, Thm 5.6]; if that tower theorem is false or incomplete, the paper's main criterion does not govern general complexity-two pairs.","rationale":"I read the paper as primarily claiming an effective reduction of complexity-two cluster-type detection to standard models over toric bases, with Theorem 1.3 as concrete payoff. The internal lemmas (Theorem 1.9, Proposition 5.1, Theorems 1.7/1.8) are detailed and the local surface computations are plausible; I do not see an internal contradiction. The load-bearing weak point is exactly the imported tower theorem [20, Thm 5.6]. The paper explicitly says 'By [20, Theorem 5.6]' and gives no proof. Since the same group's preprints [8] and [5] are also cited for essential base-case and cluster-type-over-base statements, a referee cannot certify the main theorem without access to those proofs. This does not mean the paper is wrong; it means the central claim is conditional on an unverified external result, so the reader's CONDITIONAL verdict is appropriate.","tokens_in":31634,"tokens_out":40197,"duration_ms":369609,"concrete_test":"Obtain the full statement and proof of [20, Theorem 5.6] and check that its hypotheses match Theorem 1.4 exactly. Concretely, re-prove the n=3 case (the first case where the 'at least n-2' condition is nontrivial): for a Fano-type threefold Calabi-Yau pair of complexity two, explicitly construct the crepant birational model with at least one strict conic fibration, or exhibit a pair for which no such tower exists. If the proof uses an additional assumption not stated in Theorem 1.4, the reduction collapses; if the n=3 case goes through, the main structural concern is answered.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central structural step is Theorem 1.4: every Fano-type, index-one, complexity-two Calabi-Yau pair is either finitely crepant-covered by a cluster-type pair, or admits a crepant birational model with a surface fibration over a toric pair. Its proof does not contain the main construction; it invokes [20, Theorem 5.6] verbatim ('By [20, Theorem 5.6], there exists... a tower of Mori fiber spaces ... of which at least n-2 are strict conic fibrations'). No proof, sketch, or independent verification of this theorem is given here, and it is a 2024 preprint by the first author. If [20, Theorem 5.6] requires extra hypotheses (e.g., toroidal models, Q-factoriality, or a bound on the number of boundary components) that are not available in the setting of Theorem 1.4, then the dichotomy (i)/(ii) can fail, and Theorems 1.7-1.8 would only characterize standard models rather than arbitrary complexity-two pairs. Theorem 1.9 and Proposition 5.1 also feed into the converses, but they are proved in the paper; the main unverified input is the tower theorem. The classification theorem 1.3 additionally uses the external complexity-one theorem [8, Thm 1.1] and [5, Thm 3.4] for the 'cluster type over the base' step, but these are less central to the title claim than the reduction itself.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper develops criteria for deciding when a Calabi-Yau pair of index one and complexity two is of cluster type. The main structural theorem (Theorem 1.4) asserts that, up to a crepant birational map extracting only log canonical places, such a pair is either finitely crepant-covered of degree at most two by a cluster-type pair, or admits a surface fibration over a toric Calabi-Yau pair after a composition of strict conic fibrations. The paper then restricts to standard models over toric bases and proves two precise criteria (Theorems 1.7 and 1.8) for cluster type in terms of the general fiber: existence of a node, component-wise restriction, and either a positive-self-intersection component (relative Picard rank two) or volume at least five (relative Picard rank one). These criteria are applied to Gorenstein del Pezzo surfaces of Picard rank one, yielding Theorem 1.3: cluster type holds if and only if the surface has only A-type singularities and either vol(X)>1 or |X^sing|≤3, with X(4A2) and X(2A1+2A3) as the only non-cluster A-type exceptions. The paper closes with worked examples and two open questions.","tokens_in":31920,"tokens_out":8745,"duration_ms":77453,"significance":"If the claimed results are correct, this is a substantial advance: it gives an actionable, numerical characterization of cluster type for a natural class of log Calabi-Yau pairs and settles the rank-one Gorenstein del Pezzo case. The proof strategy is transparent and largely verifiable: the reduction to standard models, the coregularity invariance theorem, and the del Pezzo analysis are written out in detail, with no fitted parameters. The main caveat is that the proof of Theorem 1.4 imports the central tower-of-Mori-fiber-spaces statement [20, Theorem 5.6] from an unpublished preprint, and the converse direction of Theorem 1.7 imports [5, Theorem 3.4]; these dependencies must be resolved before the characterization can be regarded as self-contained.","major_comments":[{"comment":"The dichotomy (i)/(ii) is the foundation of the paper, but its proof is not self-contained: it invokes [20, Theorem 5.6] (an unpublished 2024 preprint by the first author) which supplies the tower of Mori fiber spaces with at least n-2 strict conic fibrations. No statement, proof, or verification of the hypotheses of that theorem is given. Since Theorems 1.7 and 1.8 apply only to standard models obtained after this reduction, any unstated hypothesis in [20, Theorem 5.6] would invalidate the claim that the criteria govern all Fano-type index-one complexity-two pairs. Please either include a proof of the tower theorem in an appendix or state it explicitly and confirm that it applies to the pairs considered here.","section":"§3, proof of Theorem 1.4"},{"comment":"The only-if direction of condition (2) (the 'monodromy reasons' argument) is delegated to [5, Theorem 3.4], another preprint. This is load-bearing for the equivalence: if that statement is not available, Theorem 1.7 gives only a sufficient condition. The paper should state the imported theorem and either prove it or give a peer-reviewed reference.","section":"§5, proof of Theorem 1.7 (converse)"},{"comment":"The proof passes from the Miyanishi-Zhang classification [18] to the table of eleven singularity types with the sentence 'we are left with checking the following classes,' but the completeness of this reduction is not demonstrated. Since Theorem 1.3's 'if and only if' and the count of fourteen cluster-type classes depend on this completeness, the table should be accompanied by the precise correspondence to [18] (for instance, the relevant rows of Figures 1 and 1' or the enumeration of the 31 families).","section":"§6, proof of Theorem 1.3"}],"minor_comments":[{"comment":"In the final step, 'By Lemma 2.13, we conclude that (Y0,BY0) is of cluster type' should cite Lemma 3.2; Lemma 2.13 only gives coregularity, index, and complexity inequalities.","section":"§3, proof of Theorem 1.4, Case 2"},{"comment":"There is a typo: 'the infimum among of c(X,B;Σ)' should read 'the infimum of c(X,B;Σ)'.","section":"§2.4, Definition 2.9"},{"comment":"There are small typos: 'F urthermore' in the abstract and 'Asume' in Theorem 6.1 should be corrected.","section":"Abstract and §6, Theorem 6.1"},{"comment":"The proofs of Theorems 1.7 and 1.8 depend on Figures 1-8; the figures should be checked for legibility and label consistency, and color-coded components should be annotated so that the arguments are reproducible in grayscale.","section":"§5, Figures 1-8"}],"recommendation":"major_revision","confidential_remarks":"The main concern is the paper's heavy reliance on the first author's own preprint [20] for the central structural theorem; I recommend asking for either a proof of the tower theorem or a clear statement of its status. The dependence on [5] and [8] should also be clarified. These are correctness-risk issues rather than evidence of intent. The paper's scope and novelty are well aligned with the journal."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"This is a solid, useful paper with a real result, and the main caveat is exactly the one you'd suspect: its backbone is an unproved theorem imported from a preprint by the first author.\n\nThe genuinely new content is Theorems 1.7 and 1.8, clean criteria for cluster type for standard models of relative dimension two over toric bases, in terms of the general fiber boundary having a node, componentwise restriction, and a volume/self-intersection inequality. The payoff, Theorem 1.3, settles the cluster type classification for Gorenstein del Pezzo surfaces of Picard rank one: cluster type iff only A-type singularities and either vol>1 or at most three singular points, with exactly two exceptions X(4A2) and X(2A1+2A3). That is a concrete, checkable advance; the paper provides explicit 1-complements witnessing cluster type in the positive cases. The writing is careful and the arguments in Sections 3 and 5 are detailed.\n\nThe soft spot is load-bearing, not cosmetic. The reduction of an arbitrary Fano-type, index-one, complexity-two Calabi-Yau pair to a surface fibration over a toric base is Theorem 1.4, and its proof is essentially a citation to [20, Theorem 5.6], a 2024 preprint by the first author. If that tower theorem is false or needs extra hypotheses, Theorems 1.7-1.8 only characterize standard models, not general complexity-two pairs. The paper gives neither a proof nor a statement of the missing hypotheses. That is a real gap, and the stress-test note is on target. The classification also leans on [8, Theorem 1.1] and [5, Theorem 3.4], both preprints from the same group; those are less central but still unverified inputs.\n\nThis is not a circular or fitted argument: the criteria are not definitionally equivalent to the inputs, and there are no free parameters. The issue is simply that the scope of the main theorem is conditional on an external result that hasn't been independently checked.\n\nWho should read it: anyone working on cluster type, mirror symmetry for log Calabi-Yau pairs, or del Pezzo surface classification. A serious referee should engage, and should ask for a proof or precise statement of [20, Theorem 5.6] or for an independent verification. The classification section also deserves scrutiny of the local intersection computations behind Figures 9-13; the text relies on 'as shown in the figure' a bit too often.\n\nNet: accept-shaped with one large dependency. I'd send it to review, with a clear request to de-risk the imported tower theorem.","headline":"A clean criterion for cluster type in complexity two, with a classification payoff for Gorenstein del Pezzo surfaces, but its reduction to surface fibrations rests on an unproved preprint theorem by the first author.","tokens_in":32475,"tokens_out":3264,"would_cite":false,"duration_ms":27444,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["14E30","14D10","14M25","14J17"],"pacs":[],"model":"deepseek-v4-flash","headline":"For Calabi-Yau pairs of complexity two, cluster type is decided by three fiber conditions: a node, irreducible restrictions, and volume at least 5.","keywords":["Calabi-Yau pairs","complexity","cluster type","toric geometry","Gorenstein del Pezzo surfaces","coregularity","conic fibrations","del Pezzo fibrations"],"falsifier":"Take a standard model of relative Picard rank one over a toric base whose general fiber has a nodal boundary, irreducible component restrictions, and $\\operatorname{vol}(F)=4$; Theorem 1.8 predicts it is not cluster type, so exhibiting a crepant birational map from $(\\mathbb{P}^3,\\Sigma_3)$ would disprove the theorem. Concretely, one can search for an anti-canonical boundary on $X(4A_2)$ or $X(2A_1+2A_3)$ whose complement is covered by algebraic tori; the paper says no such boundary exists.","tokens_in":31417,"feed_emoji":"📐","tokens_out":13484,"duration_ms":109669,"temperature":0.7,"pith_summary":"This paper tries to turn the question 'is this Calabi-Yau pair of complexity two of cluster type?' into a finite geometric check. The authors show that, after birational modifications that preserve the Calabi-Yau structure, the question reduces to a surface fibration over a toric base, and that the answer is read off from a general fiber: the boundary curve must have a node, the boundary components must restrict to irreducible curves without splitting, and the fiber volume must be at least 5 in the relative Picard rank one case. This yields a complete answer for Gorenstein del Pezzo surfaces of Picard rank one: exactly fourteen isomorphism classes are cluster type, and among A-type surfaces only $X(4A_2)$ and $X(2A_1+2A_3)$ fail. The payoff is that a global birational search becomes a check of curve data on a single fiber.","feed_headline":"Volume 5 threshold decides cluster type for complexity-two pairs","feed_subtitle":"Gorenstein del Pezzo surfaces of rank one: 14 cluster-type classes, only two A-type exceptions.","key_machinery":"The machinery is built from the complexity invariant $c(X,B)=\\dim X+\\dim_{\\mathbb{Q}}\\mathrm{Cl}_{\\mathbb{Q}}(X)-|B|$, which measures how far a log Calabi-Yau pair is from being toric: complexity zero pairs are toric and complexity one pairs are already cluster type. Theorem 1.4 decomposes a Fano type pair of index one and complexity two into either a finite cover of degree at most two that is cluster type or a crepant tower of strict conic fibrations (conic fibrations whose two boundary components are horizontal and disjoint) ending in a surface fibration over a toric Calabi-Yau pair. Theorem 1.6 normalizes that fibration into a standard model, and Theorems 1.7 and 1.8 translate cluster type into fiber data. The coregularity invariance of Theorem 1.9 supplies the missing obstruction: over a toric base the coregularity of general fibers equals that of the total space, so the absence of a node on the general fiber is a genuine obstruction rather than an artifact of a bad model. The volume threshold 5 enters because blowing up a node subtracts 4 from the self-intersection of the fiber boundary, so the strict transform is positive exactly when $\\operatorname{vol}(F)\\geq 5$.","core_discovery":"The central claim is Theorem 1.8: if $f:(X,B)\\to(T,B_T)$ is a standard model of relative dimension two and relative Picard rank one over a toric Calabi-Yau pair, then $(X,B)$ is of cluster type if and only if the general fiber boundary $B_F$ has a nodal point, each irreducible component of $B$ restricts to an irreducible component of $B_F$, and $\\operatorname{vol}(F)\\geq 5$. A companion theorem, Theorem 1.7, handles relative Picard rank two with the volume bound replaced by positivity of self-intersection of some boundary component. Applied to surfaces, Theorem 1.3 states that a Gorenstein del Pezzo surface of Picard rank one is cluster type exactly when it has only A-type singularities and either $\\operatorname{vol}(X)>1$ or $|X^{\\mathrm{sing}}|\\leq 3$; the only A-type surfaces with volume one and four singular points, $X(4A_2)$ and $X(2A_1+2A_3)$, are not cluster type. Here 'cluster type' means the open variety $X\\setminus B$ is covered by algebraic tori up to codimension two, so the theorem says precisely which singular surfaces admit such a torus-rich logarithmic structure.","pith_inferences":["The three conditions are decidable in practice: once a standard model is exhibited, checking a node, component irreducibility, and volume is finite computation; the paper supplies the reduction but does not advertise it as an algorithm.","The threshold $\\operatorname{vol}(F)\\geq 5$ coincides with the classical boundary where del Pezzo fibrations over $\\mathbb{P}^1$ tend to be rational, suggesting cluster type may be the logarithmic version of that rationality threshold; this connection is not made in the paper.","The two exceptional surfaces sit at the simultaneous extreme of volume one and four singular points; a natural test is whether every anti-canonical boundary on them fails the node condition, or whether some non-general boundary choice could still be cluster type.","For relative Picard rank one, the node condition and volume condition appear independent; one could try replacing volume $\\geq 5$ by self-intersection $\\geq 1$ in higher relative dimension, a reformulation left implicit."],"forward_implications":["Theorem 1.3 identifies the fourteen cluster-type Gorenstein del Pezzo surfaces of Picard rank one; the only A-type non-cluster cases are $X(4A_2)$ and $X(2A_1+2A_3)$.","For higher-dimensional complexity-two pairs, cluster type is decided at the level of surface fibrations over toric bases, so the three fiber tests can be applied fibration by fibration.","Strict conic fibrations preserve cluster type in both directions (Theorem 1.5), so inserting or contracting such fibrations does not change the answer.","Complexity-two pairs can fail to be rational while having index one, coregularity zero, and complexity two (Example 7.1), so the criterion does not secretly measure rationality.","Coregularity is constant on general fibers of crepant fibrations to toric Calabi-Yau pairs (Theorem 1.9), giving a ready-made invariant for families."],"supporting_citations":[{"why":"It establishes that a Calabi-Yau pair of complexity below one has toric underlying variety, which is the base case of the reduction.","marker":"[3]"},{"why":"It supplies the theorem used to show that a cluster type pair would have no monodromy splitting of boundary components over the base.","marker":"[5]"},{"why":"It gives the torus-covering characterization of cluster type and the fact that cluster type surfaces have only A-type singularities.","marker":"[7]"},{"why":"It proves that complexity-one pairs are cluster type, the base case used throughout the inductions.","marker":"[8]"},{"why":"It provides the dual-complex and connectedness results used to compare coregularity across fibrations and adjunctions.","marker":"[11]"},{"why":"It supplies the classification and minimal resolutions of Gorenstein del Pezzo surfaces of rank one on which Theorem 1.3 is checked.","marker":"[18]"},{"why":"It supplies the structural tower of Mori fiber spaces with strict conic fibrations from which the reduction to surface fibrations begins.","marker":"[20]"}],"fun_headline_variants":["Cluster type iff A-singularities, volume>1 or ≤3 singular points","Exactly two A-type del Pezzos fail cluster type: vol=1 with 4 points","Complexity-two Calabi-Yau cluster type decided by del Pezzo fibrations","Rank-one Gorenstein del Pezzo: cluster type iff A-type and (vol>1 or ≤3 points)"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof leans on an external structural claim, not proved here, that every such pair can be birationally decomposed without changing its Calabi-Yau structure into conic fibrations ending in a surface fibration over a toric variety; if that claim is false or incomplete, the whole reduction collapses.","fun_headline_variants_meta":{"raw":{"variants":["Cluster type iff A-singularities, volume>1 or ≤3 singular points","Exactly two A-type del Pezzos fail cluster type: vol=1 with 4 points","Complexity-two Calabi-Yau cluster type decided by del Pezzo fibrations","Rank-one Gorenstein del Pezzo: cluster type iff A-type and (vol>1 or ≤3 points)"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001575,"raw_usage":{"total_tokens":6287,"prompt_tokens":945,"completion_tokens":5342,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":561,"completion_tokens_details":{"reasoning_tokens":5244}},"tokens_in":561,"tokens_out":5342,"duration_ms":39476,"temperature":1.0,"reasoning_tokens":5244,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T04:25:33.566524+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a standard model of relative Picard rank one over a toric base whose general fiber has a nodal boundary, irreducible component restrictions, and $\\operatorname{vol}(F)=4$; Theorem 1.8 predicts it is not cluster type, so exhibiting a crepant birational map from $(\\mathbb{P}^3,\\Sigma_3)$ would disprove the theorem. Concretely, one can search for an anti-canonical boundary on $X(4A_2)$ or $X(2A_1+2A_3)$ whose complement is covered by algebraic tori; the paper says no such boundary exists.","supporting_citations":[{"cited_title":"Algebraic tori in the complement of quartic surfaces","cited_arxiv_id":"2411.03506","evidence_quote":"It supplies the theorem used to show that a cluster type pair would have no monodromy splitting of boundary components over the base."},{"cited_title":"Log Calabi-Yau pairs of birational complexity zero","cited_arxiv_id":"2404.05878","evidence_quote":"It gives the torus-covering characterization of cluster type and the fact that cluster type surfaces have only A-type singularities."},{"cited_title":"Complexity one varieties are cluster type","cited_arxiv_id":"2504.17369","evidence_quote":"It proves that complexity-one pairs are cluster type, the base case used throughout the inductions."},{"cited_title":"Filipazzi and R","cited_arxiv_id":null,"evidence_quote":"It provides the dual-complex and connectedness results used to compare coregularity across fibrations and adjunctions."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"It supplies the structural tower of Mori fiber spaces with strict conic fibrations from which the reduction to surface fibrations begins."}],"review_version":1}