{"id":"c2b4feb5-a8f1-4203-b1b5-7baf3f9543da","arxiv_id":"2602.13821","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A single Kawamata-Viehweg-Kollár-Nadel vanishing theorem for higher direct images holds on compact Kähler manifolds, with the vanishing range determined by the numerical dimension of a closed positive current T on the base.","lead":"This paper proves a vanishing theorem for higher direct images: on compact Kähler manifolds, if a line bundle's curvature dominates the pullback of a positive current T, then cohomology vanishes in degrees governed by the numerical dimension of T. The result folds together several classical vanishing theorems—Kodaira, Kawamata-Viehweg, Nadel, and Kollár—into a single statement.","discovery_kind":"unification","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Lemma 4.1's equality I(ψ′+π^*φ_k)=I(ψ′+π^*φ) is asserted via an unproved adaptation of Skoda-type estimates to ψ′+π^*φ_k; the J_π^{-ε} integrability is also stated without proof.","rationale":"The reader identifies the J_π integrability as the most exposed premise; I agree that Lemma 4.1 is the hinge of Theorem 4.2, but I would weight the unproved Skoda adaptation more heavily than the J_π claim. The latter is a standard one-line local fact for holomorphic maps (Jπ=|det Jac|², finite vanishing order), so it is fixable. The former is a substantive transfer of Cao14's mass-concentration estimates to a pulled-back weight with an arbitrary quasi-psh ψ′; the paper's one-sentence 'we can also apply Skoda' is not a proof and Remark 2.9 explicitly postpones the technicalities. Neither issue is visible as a counterexample, and the overall strategy is credible. The main theorem would be accepted after a completed proof of Lemma 4.1 and a reference for Lemma 2.7. Hence I do not move the reader's CONDITIONAL verdict.","tokens_in":19783,"tokens_out":19130,"duration_ms":165915,"concrete_test":"Independently re-derive the first inclusion of Lemma 4.1 from Cao14's Lemma 5.10: prove the estimate ∫_U |f|²e^{-(ψ′+π^*φ_k)} ≤ C (∫_U |f|²e^{-(1+s1)(ψ′+π^*φ)})^{1/(1+s1)} for an arbitrary quasi-psh ψ′ on X, using only the properties listed in Lemma 2.7. If the derivation goes through without extra hypotheses, the concern is resolved; if it requires an additional bound on the Lelong numbers of ψ′ along π-fibers (or on the pullback of the Monge-Ampère data), then Lemma 4.1 is underproved as stated. As a quick falsification probe, test the product case X=Y×C with ψ′(y,z)=2log|z| and f=1: the inequality factors and the k-dependence of the constant can be computed explicitly.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central mechanism of Theorem 4.2 is the equality of multiplier ideals I(ψ′+π^*φ_k)=I(ψ′+π^*φ) in Lemma 4.1, which lets Step 2 replace h by h_k without changing I(h). The proof of Lemma 4.1 rests on two unsupported analytic claims. (a) The inclusion I(ψ′+π^*φ)⊂I(ψ′+π^*φ_k) is justified by 'we can replace φ_k in the proof of Lemma 2.7(4) ... by ψ′+π^*φ_k, since we can also apply Skoda's uniform integrability theorem.' This is not a proof: Cao14's Lemma 5.10 is a statement about an approximation of a single weight φ on Y, and transferring it to the pulled-back weight ψ′+π^*φ_k requires checking that the Lelong numbers of ψ′+π^*φ_k are uniformly controlled and that the mass-concentration data survive pullback. No such check is given, and the paper explicitly defers the relevant Skoda statement (Remark 2.9). (b) The assertion that ∫_X J_π^{-ε} dV<∞ for some ε>0 (Jπ being the squared Jacobian determinant) is stated without proof or reference; it is true (locally Jπ=|h|² for a holomorphic function h), but it is load-bearing for the coarea-formula step. If either fails, I(h_k)=I(h) collapses and Step 3's ε_k C_k→0 cannot be converted into vanishing. Lemma 2.7 itself is imported from Cao14 with only a sketch.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves two vanishing theorems for higher direct images of adjoint bundles with multiplier ideal sheaves. The central result, Theorem 1.8 (= Theorem 4.2), asserts that if π:X→Y is a holomorphic surjection between compact Kähler manifolds, L is a holomorphic line bundle with a singular Hermitian metric h, and T is a closed positive (1,1)-current on Y such that iΘ_{L,h}≥π^*T, then H^p(Y,R^qπ_*(K_X⊗L⊗I(h)))=0 for every p≥m−nd(T)+1 and every q≥0. The proof combines the cohomological representative construction of [QZ25] with a regularization of T via Monge–Ampère equations, Skoda-type uniform integrability, the strong openness theorem of Guan–Zhou, and Bochner-type L² estimates. A Girbau-type version (Theorem 1.9), valid on weakly pseudoconvex Kähler manifolds when a continuous semipositive form has rank at least l, and a Fujiki-class version (Theorem 1.10) are also established.","tokens_in":1986,"tokens_out":2888,"duration_ms":109685,"significance":"If the proof is completed, Theorem 1.8 is a substantial unification: it contains Kollár's vanishing theorem (when nd(T)=m) and the Cao–Guan–Zhou Kawamata–Viehweg–Nadel theorem for pseudo-effective line bundles (when Y=X and q=0), and it gives a new statement even for projective manifolds. The numerical-dimension formulation for higher direct images is natural and likely to be useful. The paper also gives a clean mechanism—cohomology representatives from [QZ25] plus multiplier-ideal equality under approximation—that is conceptually attractive. However, the main proof depends on a small number of analytic facts that are either only sketched or asserted without proof, so the result is credible but not yet fully demonstrated in the submitted form.","major_comments":[{"comment":"The crucial equality I(ψ′+π^*φ_k)=I(ψ′+π^*φ) is the mechanism that allows replacing h by h_k without changing I(h). The first inclusion is justified by the sentence: “we can replace φ_k in the proof of Lemma 2.7(4) ... by ψ′+π^*φ_k, since we can also apply Skoda’s uniform integrability theorem.” This is an assertion rather than a proof. Cao14's Lemma 5.10 concerns an approximation of a single weight on Y; transferring it to the pulled-back weight requires controlling the Lelong numbers of ψ′+π^*φ_k and checking that the mass-concentration data survive pullback. Remark 2.9 explicitly defers the precise Skoda statement, so the reader cannot verify the inclusion. Since Lemma 4.1 is load-bearing for Step 2 of Theorem 4.2, this gap must be filled with a detailed argument or a precise reference that covers the pulled-back family.","section":"§4, Lemma 4.1"},{"comment":"The proof uses the assertion that for a proper surjective holomorphic map π:X→Y between compact Kähler manifolds there exists ε>0 with ∫_X J_π^{-ε} dV<∞, where J_π is the square Jacobian determinant. This is stated in two lines: “Note that Jπ=eΨ ... There exists a constant ε>0...” and is then used to prove the integrability of the coarea expression. The finiteness is true, but it is not demonstrated; one should give the local normal form J_π=∑|h_i|² with h_i holomorphic (or a direct reference), show that the inverse power is locally integrable for sufficiently small ε, and then choose a uniform ε on the compact manifold. This is a small but load-bearing step in the proof of Lemma 4.1.","section":"§4, Lemma 4.1, coarea estimate"},{"comment":"Lemma 2.7 is imported from Cao14 with only a sketch, yet Theorem 4.2 uses its quantitative properties in an essential way: Vol(U_k)≤ε_k^β, ε_k≫τ_k+1/k, and the uniform integrability bound (4). The paper should either state the full lemma with a proof or give a precise reference to the exact statement in Cao14, indicating which parts are verbatim and which are adapted. Remark 2.8 is a heuristic sketch and does not provide enough detail to check that all constants β, γ, s1 can be chosen compatibly. This matters in Step 3 of Theorem 4.2 where the convergence ε_kC_k→0 relies on the rate of decay of Vol(U_k).","section":"§2, Lemma 2.7 and Remark 2.8"},{"comment":"The operator B_{1,k}=[π^*(α+i∂∂φ_k+4ε_kω_Y),Λ_{ω_X}] is defined using the current i∂∂φ_k, which is not smooth; Lemma 3.1 is stated for continuous semipositive forms. The pointwise estimate (and hence the L² estimate) is valid only away from the analytic singularities of φ_k. The proof should explicitly justify the passage from X\\Z_k to all of X, for example by removing the combined singular locus of ψ′_i and π^*φ_k and using Lemma 2.2–2.3, as is done elsewhere in the paper. This is a technical but necessary step; without it the ∂-equation argument is not fully rigorous.","section":"§4, Step 2 of Theorem 4.2"}],"minor_comments":[{"comment":"The symbol φ is used both for a quasi-plurisubharmonic function and for an L-valued (n,q)-form in Lemma 3.1. Please use a different letter for the form to avoid confusion.","section":"§3, Lemma 3.1"},{"comment":"The inequality “(a−b)^q≤a^q−b^q” is stated for “any t>0” via g_k^t. This inequality requires the exponent qt≥1. Taking t=1 is sufficient because q>1, but the current wording is inaccurate and should be corrected.","section":"§4, Lemma 4.1 proof"},{"comment":"The proof uses a fixed p>1 with I(pψ′)=I(ψ′), obtained from the strong openness theorem. On a compact manifold this can be justified by coherence and compactness, but the argument is not given; please add a sentence explaining why such a uniform p exists.","section":"§4, Lemma 4.1 proof"},{"comment":"The notation H(α_{i_0...i_p}) appearing in the proof of Theorems 3.2 and 4.2 is used before it is defined; it would help to recall that H is the inverse of the quotient map ι from Theorem 2.10(1).","section":"§2, Theorem 2.10"},{"comment":"The list of references uses nonstandard labels such as [GZ15-a], [Kol86-a], and [QZ25]; this is acceptable but should be normalized to the journal style. Also, the arXiv reference [XZ25] is cited as “to appear”; please update if possible.","section":"General"}],"recommendation":"major_revision","confidential_remarks":"The manuscript is a serious piece of work with a genuinely unifying statement. The main risk is concentrated in Lemma 4.1 and in the use of the non-smooth operators in Theorem 4.2; these gaps appear fillable rather than fatal. I recommend sending the revision back to the same referee or to an expert in L² estimates for singular metrics. The paper also relies heavily on [QZ25] by two of the authors; this is not improper, but the editor may want to ensure that the relevant statements from [QZ25] and Cao14 are quoted accurately and are not themselves under dispute."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nThe main theorem is real and worth your attention. Theorem 1.8 subsumes Kollár, Cao's q=0 numerical-dimension vanishing, and the QZ25 Kähler-form higher-direct-image result in a single statement: for a holomorphic surjection π:X→Y between compact Kähler manifolds and a closed positive current T on Y, H^p(Y,R^qπ_*(K_X⊗L⊗I(h)))=0 for p≥m−nd(T)+1 and all q≥0, assuming iΘ_{L,h}≥π^*T. That is a genuine unification and it is new even for projective manifolds when T is only pseudo-effective and q>0. The proof mechanism — use the QZ25 injective map φ^{p,q}, regularize T via Cao's mass-concentration technique, show [b] is represented by arbitrarily small forms, then use Hausdorff property — is coherent, and the pointwise operator estimate in Lemma 3.1 is cleanly derived.\n\nThe soft spots are concentrated in Lemma 4.1, which is the load-bearing step that equates I(ψ′+π^*φ_k) with I(ψ′+π^*φ). The inclusion I(ψ′+π^*φ)⊂I(ψ′+π^*φ_k) is justified by saying the proof of Lemma 2.7(4) can be repeated with ψ′+π^*φ_k in place of φ_k, using Skoda's uniform integrability. That is plausible, but it is not demonstrated. You need to check uniform control of Lelong numbers of ψ′+π^*φ_k so that Skoda gives a uniform exponent and that Cao's mass-concentration estimates survive pullback. The paper explicitly defers the Skoda statement to Remark 2.9. Separately, the coarea step asserts ∫_X J_π^{-ε} dV<∞ for some ε>0; this is true (locally Jπ=|h|²), but it is stated without proof or reference. Both are fill-in gaps rather than fatal flaws — the central argument appears sound — but a referee should ask for the details.\n\nThe paper leans heavily on [QZ25] (same two authors) for the injective map and on [Cao14] for the regularization, but those are independent published results; there is no circularity. The prose is honest about what is imported.\n\nI would send this to a serious referee. The statement is important, the proof strategy is credible, and the gaps are localized. Someone working in transcendental algebraic geometry or vanishing theorems will want to know this result; if you are in that area, cite it.\n\nBest.","headline":"Genuinely unifying higher-direct-image vanishing theorem for compact Kähler manifolds; proof is credible, with localized analytic gaps in Lemma 4.1.","tokens_in":20727,"tokens_out":10825,"would_cite":true,"duration_ms":83714,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["32C35","14F17","14F18","32L10","32L20","32U05","14C30"],"pacs":[],"model":"deepseek-v4-flash","headline":"A unified vanishing theorem: cohomology of higher direct images of adjoint bundles vanishes once the degree exceeds a threshold set by the numerical dimension of a positive current.","keywords":["vanishing theorem","higher direct images","pseudo-effective line bundles","numerical dimension","multiplier ideal sheaf","Kähler manifolds","L2 estimates","strong openness"],"falsifier":"The most direct falsifier would be a counterexample to the unproved integrability assertion: a proper surjective holomorphic map π:X→Y between compact Kähler manifolds for which ∫_X J_π^{-ε} dV diverges for every ε>0; such a map would break Lemma 4.1 and therefore Theorem 4.2. Alternatively, one could look for a specific compact Kähler fibration and current T with nd(T)<m where the predicted L2 solutions or the limit ε_k C_k → 0 fail, producing a nonzero class in H^p(Y,R^qπ_*(K_X⊗L⊗I(h))) for some p ≥ m−nd(T)+1.","tokens_in":19674,"feed_emoji":"📐","tokens_out":12230,"duration_ms":92769,"temperature":0.7,"pith_summary":"This paper proves a single vanishing theorem that covers and unifies the classical higher-direct-image vanishing theorems and the multiplier-ideal vanishing theorems on compact Kähler manifolds. The main result says: for a holomorphic surjection π:X→Y between compact Kähler manifolds, if a line bundle L with singular Hermitian metric h satisfies iΘ_{L,h} ≥ π^*T for a closed positive (1,1)-current T on Y, then H^p(Y,R^qπ_*(K_X⊗L⊗I(h))) = 0 for every p ≥ m − nd(T) + 1 and every q ≥ 0, where nd(T) is the numerical dimension of T and m = dim Y. In particular, choosing T with full numerical dimension recovers the classical higher-direct-image vanishing theorem, while the case Y=X and q=0 recovers the multiplier-ideal vanishing theorem. The proof constructs approximating metrics that preserve the multiplier ideal sheaf, solves ∂-equations with L2 estimates, and lets the approximation error tend to zero; the result is new even for projective manifolds.","feed_headline":"Numerical dimension sets the vanishing range for direct images","feed_subtitle":"Vanishing in all degrees above a threshold set by the numerical dimension of a positive current on the base","key_machinery":"The proof rests on three pieces. (1) An injective map φ^{p,q}: H^p(Y,R^qπ_*(K_X⊗L⊗I(h))) → H^{p+q}(X,K_X⊗L⊗I(h)) is constructed via a Stein cover and a smooth partition of unity; proving it is zero gives the vanishing. (2) The Bochner-type curvature operator B=[π^*σ,Λ_{ω_X}] and the pointwise estimate of Lemma 3.1 bound ⟨B^{-1}b,b⟩ in terms of 1/(λ_1+⋯+λ_p) times |φ|^2, enabling L2 solutions of ∂u=b with explicit error terms. (3) Lemma 4.1 shows that for the approximating potentials φ_k of Lemma 2.7, I(ψ' + π^*φ_k)=I(ψ' + π^*φ) for large k; this equality, proved using strong openness and the coarea formula, allows the replacement of h by h_k without changing the multiplier ideal sheaf, which","core_discovery":"The central claim, Theorem 1.8, is that the vanishing of H^p(Y,R^qπ_*(K_X⊗L⊗I(h))) depends only on the numerical dimension nd(T) of the positive current T on the base, not on the fiber geometry or the singularities of h, as long as the curvature of h dominates π^*T. The proof shows that the natural map φ^{p,q} from H^p(Y,R^qπ_*(K_X⊗L⊗I(h))) into H^{p+q}(X,K_X⊗L⊗I(h)) is the zero map. To do this, the paper constructs a sequence of quasi-plurisubharmonic approximants φ_k of the potential of T, with analytic singularities and controlled lower eigenvalue bounds, and proves (Lemma 4.1) that after pulling back to X, the multiplier ideal sheaf I(ψ' + π^*φ_k) equals I(ψ' + π^*φ) for all large k. Thi","pith_inferences":["The equality I(ψ' + π^*φ_k)=I(ψ' + π^*φ) in Lemma 4.1 is the technical heart; if it could be proved by a different method, the mass-concentration rates and the integrability assumption on J_π^{-ε} might be bypassed, potentially widening the theorem to singular bases or non-Kähler settings.","The theorem suggests that the numerical dimension of the dominating current, rather than positivity of the line bundle itself, is the effective invariant for vanishing of adjoint cohomology; this points toward analogous statements for Higgs bundles or twisted de Rham cohomology where a similar dominant current exists.","One testable extension would be to drop compactness of Y: combining the coarea-formula argument with a Girbau-type exhaustion might yield a relative vanishing theorem for families over Stein or weakly pseudoconvex bases, a case the paper only partially covers in Theorem 3.2."],"forward_implications":["When nd(T)=m, the theorem gives H^p(Y,R^qπ_*(K_X⊗L⊗I(h)))=0 for all p≥1 and q≥0, recovering the classical higher-direct-image vanishing theorem for pseudo-effective line bundles over Kähler bases.","When Y=X and π is the identity, q=0 and the theorem reduces to H^p(X,K_X⊗L⊗I(h))=0 for p≥n−nd(T)+1, the standard multiplier-ideal vanishing theorem for pseudo-effective line bundles on compact Kähler manifolds.","The same proof yields Theorem 3.2: on a weakly pseudoconvex Kähler manifold X, if σ is a continuous semi-positive (1,1)-form with σ^l ≠ 0 everywhere and iΘ_{L,h} ≥ π^*σ, then H^p(Y,R^qπ_*(K_X⊗L⊗I(h)))=0 for p≥m−l+1 and all q≥0.","Theorem 5.1 extends the main result to compact manifolds admitting a Kähler modification, showing the vanishing statement does not depend on the Kähler condition of X itself."],"fun_headline_variants":["Numerical dimension alone sets vanishing range for direct images","Vanishing theorem: numerical dimension of base current decides","Base current's numerical dimension controls vanishing of direct images","Higher direct images vanish above numerical dimension threshold"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"Lemma 4.1 asserts, without proof, that for a proper surjective holomorphic map between compact Kähler manifolds there exists ε>0 with ∫_X J_π^{-ε} dV_{ω_X} < ∞, where J_π is the squared Jacobian determinant; the equality of multiplier ideals under pullback, and hence the replacement of h by h_k, rests on this integrability.","fun_headline_variants_meta":{"raw":{"variants":["Numerical dimension alone sets vanishing range for direct images","Vanishing theorem: numerical dimension of base current decides","Base current's numerical dimension controls vanishing of direct images","Higher direct images vanish above numerical dimension threshold"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00057,"raw_usage":{"total_tokens":2467,"prompt_tokens":613,"completion_tokens":1854,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":357,"completion_tokens_details":{"reasoning_tokens":1802}},"tokens_in":357,"tokens_out":1854,"duration_ms":13671,"temperature":1.0,"reasoning_tokens":1802,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-02T23:24:50.077432+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"The most direct falsifier would be a counterexample to the unproved integrability assertion: a proper surjective holomorphic map π:X→Y between compact Kähler manifolds for which ∫_X J_π^{-ε} dV diverges for every ε>0; such a map would break Lemma 4.1 and therefore Theorem 4.2. Alternatively, one could look for a specific compact Kähler fibration and current T with nd(T)<m where the predicted L2 solutions or the limit ε_k C_k → 0 fail, producing a nonzero class in H^p(Y,R^qπ_*(K_X⊗L⊗I(h))) for some p ≥ m−nd(T)+1.","supporting_citations":[],"review_version":1}