{"id":"7e8e662a-8bad-4a69-8d1a-95cf80ccadcd","arxiv_id":"2607.04536","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Compact Kähler fourfolds with pseudo-effective K_X and no codimension-1 or -2 subvarieties have torsion K_X, hence are torus quotients or IHS manifolds.","lead":"A compact Kähler fourfold with pseudo-effective canonical bundle and no divisors or surfaces must have torsion canonical bundle, so it is a torus quotient or irreducible holomorphic symplectic. This advances the Kähler minimal-model program by classifying the extreme case with almost no subvarieties.","discovery_kind":"extension","skeptic_critique":{"model":"grok-4.5","headline":"No significant objection identified beyond the reader's already-flagged recursion; the support-dimension claims for multiplier ideals and reflexive quotients appear to hold under the stated hypotheses.","rationale":"The central claim (Theorem 4) rests on three reductions: pseudo-effective \to nef (Lemma 11), irregularity dichotomy (Lemma 9), and the recursive Ext construction that forces K_X torsion when χ ≤ 0 or when nefness supplies the numerical hypotheses of Proposition 45. The first two reductions use standard tools (Cao–Höring rational curves, Albanese geometry, Miyaoka–Yau) and appear solid. The third is delicate but, under the paper's explicit geometric hypotheses, the support-dimension statements required by Lemmas 20–22 hold for every coherent ideal that arises—multiplier ideals included—because any proper analytic set is forced to dimension ≤1. Consequently the isomorphisms and the infinite supply of sections survive. The reader's CONDITIONAL verdict already correctly flags that this recursion has not been independently re-checked; that remains the appropriate posture. No stronger or different load-bearing concern emerges from a second reading, so the verdict is left unchanged.","tokens_in":29033,"tokens_out":672,"duration_ms":6889,"concrete_test":"Re-derive the initial non-split extension (7) and the first recursive step (14)/(17) from the two holomorphic 3-forms and the rank-2 reflexive sheaf E without invoking external results beyond Demailly's Frobenius theorem and Lemma 24; if both extension classes remain nonzero and the support of each successive Z stays of dimension ≤1, the recursion is sound.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The reader's weakest_assumption correctly isolates the technical core of Lemma 8: the chain H^{3}(K^{m+1} \times I(h^m)) \to H^{1}(K^{-m}) \to Ext^{1}(I_Z, K^{-m}) and the recursive production of nonzero classes. Under the paper's hypotheses those steps are internally consistent. Lemma 20 uses only that any proper analytic support of a coherent quotient has codimension ≥3 (hence dim ≤1), which follows immediately from the absence of codim-1 and codim-2 subvarieties; the same bound applies to the cosupport of any multiplier ideal I(h^m) because that cosupport is analytic. Lemma 22 then converts the resulting H^{1} into Ext^{1} by local depth vanishing (depth_J R ≥3 on a regular local ring of dimension 4). The recursion itself (Cases A/B after the initial non-split extension (7)) never re-introduces higher-dimensional supports: each new ideal sheaf again arises as a rank-1 torsion-free quotient of a reflexive sheaf, so its support is again forced to dimension ≤1. Thus the infinite supply of sections of Ω^{1} \times K^m, and the appeal to [1, Prop. 2.6], go through. No independent load-bearing gap appears.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.5","summary":"The paper proves that a compact Kähler fourfold X with pseudo-effective canonical bundle and no analytic subvarieties of codimension 1 or 2 has torsion canonical bundle (Theorem 4). By Beauville–Bogomolov this implies X is a torus quotient or an irreducible holomorphic symplectic manifold. The argument proceeds by reducing to the nef case (Lemma 11, via Cao–Höring rational curves, Horikawa deformations, and Demailly regularisation), establishing the irregularity dichotomy q(X) ∈ {0,4} (Lemma 9), and then treating the nef case by a recursive construction of nonzero classes in Ext^{1}(I_Z, K^{-m}) that, via Hard Lefschetz with multiplier ideals, produces infinitely many sections of Ω^{1}_X ⊗ K^m and forces torsion by Anella–Huybrechts (Lemma 8 and Lemma 10). Supporting examples of elliptic fibrations without divisors or surfaces are given in §3.3.","tokens_in":29302,"tokens_out":1259,"duration_ms":13190,"significance":"If correct, the result settles the four-dimensional pseudo-effective case of the Campana–Demailly–Verbitsky conjectures on simple Kähler manifolds under the natural minimal-model hypothesis that excludes only low-codimension subvarieties (allowing curves). The method—Hard Lefschetz with multiplier ideals, reflexive extension classes, and foliation positivity—is elementary relative to the surrounding literature and is presented as the first complete instance of a technique intended for higher dimensions. The elliptic-fibration examples of §3.3 usefully separate the “no divisors or surfaces” hypothesis from the stricter “no positive-dimensional subvarieties” condition, showing that the statement is not vacuous.","major_comments":[{"comment":"The load-bearing step is the recursive production of nonzero classes in Ext^{1}(I_Z, K^{-m}) inside the proof of Lemma 8 (pp. 20–25). The chain H^{3}(K^{m+1} ⊗ I(h^m)) ≅ H^{1}(K^{-m}) ≅ Ext^{1}(I_Z, K^{-m}) rests on Lemmas 20–22. Lemma 20 correctly uses that any proper analytic support has codimension ≥3 (hence dimension ≤1) under the standing hypotheses; the same bound applies to cosupports of multiplier ideals. Lemma 22’s local depth vanishing (depth_J R ≥3) is standard on a regular local ring of dimension 4. The recursion (Cases A/B after the initial non-split extension (7)) never re-introduces higher-dimensional supports, because each new ideal again arises as a rank-1 torsion-free quotient of a reflexive sheaf. Thus the infinite supply of sections of Ω^{1} ⊗ K^m and the appeal to [1, Prop. 2.6] go through. No independent gap appears, but the argument is delicate and would benefit fr","section":null},{"comment":"Lemma 11 (nef reduction) invokes Cao–Höring [20, Cor. 1.4] for a rational curve with K_X · C < 0, Horikawa’s deformation estimate, and Demailly regularisation of the class c_{1}(K_X) + ε[ω]. The claim that the singular locus S of the resulting Kähler current has dim S ≤ 1 follows immediately from the absence of divisors and surfaces, and the subsequent contradiction with a positive-dimensional family of cycles supported on S is clean. The step is therefore sound, but the paper should record that the family of cycles is obtained after quotienting by Aut(P^{1}) and that the germ is chosen so that the cycles are not all equal; a one-sentence clarification would remove any residual ambiguity.","section":null},{"comment":"In the positive-Euler-characteristic case of Lemma 10 / Lemma 46 the appeal to the Miyaoka–Yau inequality of Liu [37, Thm. 1.1] for nef canonical bundles on compact Kähler fourfolds is essential for the sign of c_{1}(K)^{2} · c_{2}. The inequality is cited correctly, yet the paper does not spell out the elementary rearrangement that yields c_{1}(K)^{2} · c_{2} ≥ (2/5)c_{1}(K)^{4} ≥ 0. Inserting this short calculation would make the numerical hypothesis of Proposition 45 completely self-contained.","section":null}],"minor_comments":[{"comment":"Several typographical slips: “setp” for “step” (p. 4), “albanese” uncapitalised (p. 12), “cosupport” sometimes written “co-support”, and occasional missing spaces after punctuation.","section":null},{"comment":"Notation for the multiplier ideals I(h^m) versus I_m is not entirely uniform; a single convention would improve readability.","section":null},{"comment":"The reference list contains a few incomplete or future-dated entries (e.g., [13], [39], [43]); standard arXiv identifiers or DOIs should be supplied where available.","section":null},{"comment":"In Example 39 the Baire-category argument that a generic A yields NS(X_A)=0 is correct but slightly terse; a sentence recalling that the exceptional sets S_Q are proper closed algebraic subsets would help non-specialists.","section":null}],"recommendation":"minor_revision","confidential_remarks":"The manuscript is independent of Campana’s concurrent preprint [13] and treats a strictly weaker hypothesis (allowing curves). The self-citation [43] is used only for the torsion case of Beauville–Bogomolov and does not affect novelty. The paper is a natural fit for a geometry journal that publishes Kähler classification results; the technical core is sound and the requested revisions are local clarifications rather than conceptual repairs."},"author_rebuttal":null,"desk_editor":{"model":"grok-4.5","letter":"This paper proves that a compact Kähler fourfold with pseudo-effective K_X and no codimension-1 or -2 subvarieties has torsion canonical bundle, hence is a torus quotient or IHS by Beauville–Bogomolov. That is the four-dimensional pseudo-effective case of Campana’s simple-manifold picture, under the weaker (and MMP-natural) hypothesis that only curves are allowed.\n\nWhat is new is the method and the examples. The author reduces to the nef case via Cao–Höring + Horikawa + Demailly regularisation, gets the irregularity dichotomy q=0 or 4 by Albanese arguments that kill intermediate images, then runs a recursive construction of nonzero Ext^{1} classes from a holomorphic 2-form and two 3-forms, feeding Hard Lefschetz with multiplier ideals to produce infinitely many sections of Ω^{1} ⊗ K^m and invoking Anella–Huybrechts. The elliptic-fibration examples over a simple 3-torus with NS=0 show that “no divisors or surfaces” is strictly weaker than “no positive-dimensional subvarieties,” so the statement is not just a special case of Campana’s concurrent preprint.\n\nThe soft spot is exactly the one the reader flagged: the chain H^{3}(K^{m+1} ⊗ I) ≅ H^{1}(K^{-m}) ≅ Ext^{1}(I_Z, K^{-m}) and the recursion that keeps producing nonzero classes. Under the paper’s hypotheses the support-dimension claims are forced (any proper analytic cosupport of a coherent ideal or of a rank-1 torsion-free quotient of a reflexive sheaf has codim ≥3, hence dim ≤1), so the isomorphisms and the infinite supply of sections go through. It is still a long, case-by-case argument that a specialist should re-check line by line; several external inputs (Ou’s uniruledness preprint, Liu’s Miyaoka–Yau for Kähler, PRT) are deep. No circularity, no free parameters.\n\nThis is for people working on Kähler MMP or simple manifolds. It deserves a serious referee. I would send it out.","headline":"Solid fourfold classification under a natural minimal-model hypothesis; the recursion is delicate but holds under the stated support bounds.","tokens_in":29932,"tokens_out":547,"would_cite":true,"duration_ms":6457,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["32J27","32Q15","32Q57","14E30"],"pacs":[],"model":"grok-4.5","headline":"Compact Kähler fourfolds with no divisors or surfaces and pseudo-effective canonical bundle must have torsion canonical bundle, hence are torus quotients or irreducible holomorphic symplectic manifolds.","keywords":["compact Kähler fourfolds","pseudo-effective canonical bundle","torsion line bundle","no codimension-one subvarieties","no codimension-two subvarieties","Beauville–Bogomolov decomposition","Hard Lefschetz with multiplier ideals","simple manifolds"],"falsifier":"A compact Kähler fourfold with pseudo-effective but non-torsion canonical bundle that contains no analytic divisors and no analytic surfaces would falsify the main theorem; equivalently, an explicit fourfold whose multiplier ideals for powers of a singular metric on K_X have two-dimensional support would break the key isomorphism step.","tokens_in":29872,"feed_emoji":"🔷","tokens_out":610,"duration_ms":5152,"temperature":0.7,"pith_summary":"The paper classifies smooth compact Kähler fourfolds that contain no analytic divisors and no analytic surfaces, under the hypothesis that the canonical bundle is pseudo-effective. It proves that the canonical bundle must then be torsion. Once the canonical bundle is torsion, the Beauville–Bogomolov decomposition theorem reduces the manifold to a finite quotient of a complex torus or to an irreducible holomorphic symplectic manifold. The result therefore settles a four-dimensional case of the expectation that simple Kähler manifolds are built from tori and hyperkähler pieces, while still allowing curves. The argument proceeds by first proving that the canonical bundle is automatically nef, then by constructing infinitely many sections of twisted cotangent bundles via Hard Lefschetz with multiplier ideals and recursive non-split extension classes; those sections force the canonical bundle to be torsion.","feed_headline":"Kähler fourfolds without divisors or surfaces have torsion K","feed_subtitle":"They must be torus quotients or irreducible holomorphic symplectic manifolds","key_machinery":"Recursive non-split extension classes in Ext^{1}(I_Z,K_X^{-m}) together with the Hard Lefschetz theorem for pseudo-effective line bundles with multiplier ideals; absence of codimension-2 subvarieties collapses the relevant cohomology groups to these Ext groups, producing infinitely many sections of Ω^{1}_X⊗K_X^m that force torsion.","core_discovery":"If X is a compact Kähler fourfold whose canonical bundle K_X is pseudo-effective and which contains no irreducible analytic subvarieties of codimension 1 or 2, then K_X is a torsion line bundle. Consequently X is, up to finite étale cover, a complex torus or an irreducible holomorphic symplectic manifold.","pith_inferences":[],"forward_implications":[],"fun_headline_variants":["Kähler fourfolds without codim-1/2 subvars have torsion K","No divisors or surfaces: compact Kähler 4-folds yield torsion K_X","Pseudo-effective K plus no low-codim subvars forces torsion K","Kähler fourfolds free of surfaces and divisors carry torsion K","Absence of codim ≤2 subvars implies torsion K on Kähler fourfolds"],"cache_read_input_tokens":16512,"weakest_assumption_plain":"The whole recursion collapses if the multiplier ideals or the ideal sheaves arising from rank-one quotients of reflexive sheaves can have two-dimensional support even when the manifold has no surfaces.","fun_headline_variants_meta":{"raw":{"variants":["Kähler fourfolds without codim-1/2 subvars have torsion K","No divisors or surfaces: compact Kähler 4-folds yield torsion K_X","Pseudo-effective K plus no low-codim subvars forces torsion K","Kähler fourfolds free of surfaces and divisors carry torsion K","Absence of codim ≤2 subvars implies torsion K on Kähler fourfolds"]},"model":"grok-4.5","effort":"low","cost_usd":0.007874,"raw_usage":{"total_tokens":1769,"prompt_tokens":589,"num_sources_used":0,"completion_tokens":91,"cost_in_usd_ticks":78740000,"prompt_tokens_details":{"text_tokens":589,"audio_tokens":0,"image_tokens":0,"cached_tokens":256},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":1089,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":589,"tokens_out":91,"duration_ms":9734,"temperature":1.0,"reasoning_tokens":1089,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-07-11T17:45:35.202203+00:00","model_set":{"reader":"grok-4.5"},"falsifier":"A compact Kähler fourfold with pseudo-effective but non-torsion canonical bundle that contains no analytic divisors and no analytic surfaces would falsify the main theorem; equivalently, an explicit fourfold whose multiplier ideals for powers of a singular metric on K_X have two-dimensional support would break the key isomorphism step.","supporting_citations":[],"review_version":1}