{"id":"afabdbdb-e43f-4a31-9138-a1d1a284cbe3","arxiv_id":"2505.10417","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The paper proves a structure theorem for the trivial Hodge module of toric varieties and uses it to compute Hodge-Du Bois diamonds for cones over simple polytopes.","lead":"This paper identifies the precise way local cohomology of a toric variety is assembled from intersection cohomology of smaller torus-invariant pieces, with weights controlled by the combinatorics of the defining cone. It then turns this structure into explicit formulas for the Hodge-Du Bois numbers of projective toric varieties attached to simple polytopes, a family where no such formula existed before.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The concrete polytope results depend on the unproved Theorem 5.4 from [KV25]; if that theorem is unsound or circularly relies on Theorem 1.1, Proposition 1.6 and Theorem 1.12 are unverified.","rationale":"The reader's weakest-assumption analysis correctly identifies Theorem 5.4 as the most fragile load-bearing premise. This theorem is not proved in the manuscript; it is stated as a result from the companion preprint [KV25], and Sections 6.2 and 7.3 rely on it directly for Proposition 1.6 and Theorem 1.12. The paper's abstract and introduction highlight the Hodge–Du Bois diamond and the two polytope classes as main applications, so the central claim of the paper, taken as the full package of results, is conditional on this external result. The concern is not an internal inconsistency but a verification gap: a conscientious reader cannot check the long exact sequence without going to [KV25], and if [KV25] depends on Theorem 1.1 of this paper, the dependence becomes a coupled pair that must be vetted together. I reviewed the proof of Theorem 1.1: Steps 1–4 form a coherent induction using the two spectral sequences, and the use of [PP24, Proposition 6.4] is a secondary black box but appears limited to the weight-range assertion. Thus I agree with the reader's conditional verdict and do not recommend changing it.","tokens_in":40276,"tokens_out":25672,"duration_ms":236825,"concrete_test":"Inspect [KV25] to verify that Theorem 1.4 (quoted here as Theorem 5.4) is proved without prior use of Theorem 1.1 of the present paper, and trace the dependency graph to rule out a cycle. Independently, recompute the Hodge–Du Bois diamond for the binomial hypersurface S={xyz=uvw} in P^5 from Example 7.3 by direct computation of the Ishida complex of the cone sigma^vee using the explicit lattice vectors given, and compare with the displayed diamond; a mismatch would falsify the application of Theorem 5.4.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The load-bearing concern is the unproved external Theorem 5.4, quoted without proof as [KV25, Theorem 1.4]. Proposition 1.6 (cones over simplicial polytopes) and Theorem 1.12 (Hodge–Du Bois diamond for simple polytopes) are derived from this theorem in Sections 6.2 and 7.3, respectively; the worked Example 7.3 and the advertised Betti-number computations therefore stand only if [KV25] is sound. Moreover, [KV25] may itself invoke Theorem 1.1 of the present paper, so verifying the chain requires reading both papers together and confirming there is no dependency cycle. The proof of Theorem 1.1 itself does not use Theorem 5.4; its main external inputs are [PP24, Proposition 6.4] and [KV24], which are also black boxes but are less likely to be circular. Without a proof or an independent source for Theorem 5.4, the concrete formulas for the two polytope classes remain conditional.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper studies the mixed Hodge module structure of the trivial Hodge module QH'_X = QH_X[n] on an n-dimensional toric variety X. The main structural theorem (Theorem 1.1) asserts that QH'_X has cohomology only in degrees [-(n-3),0], that its weight-graded pieces are described by intersection cohomology Hodge modules supported on torus-invariant subvarieties with explicit Tate twists, and that the multiplicities depend only on the corresponding cone. The authors use this to prove that the singular cohomology of proper toric varieties is mixed of Hodge-Tate type (Corollary 1.2), to obtain vanishing and depth results for Ext sheaves of reflexive differentials (Theorem 1.3), and to derive a combinatorial consequence on Ishida complexes. They then specialize to cones over simplicial and simple polytopes, giving complete descriptions of the local cohomology modules and computing the Hodge-Du Bois diamond for projective toric varieties associated to simple polytopes (Theorem 1.12). The paper also contains a self-contained account of the Ishida complex in Section 4 and an appendix computing the multiplicities in low dimensions.","tokens_in":40332,"tokens_out":7647,"duration_ms":66273,"significance":"If the main results are correct, the paper gives a near-complete description of the Hodge module structure of local cohomology for two important classes of toric varieties, and the Hodge-Tate corollary is a clean generalization of a known fact. The paper is clearly written, and the self-contained treatment of the Ishida complex in Section 4 is valuable. The explicit low-dimensional computations in Appendix A and the worked binomial hypersurface example in Example 7.3 make the abstract structures concrete. The main caveat is the dependence on unproved external results, especially Theorem 5.4, which is load-bearing for the polytope formulas.","major_comments":[{"comment":"The proofs of Proposition 1.6 and Theorem 1.12, including the Hodge-Du Bois diamond formula for simple polytopes and the Betti numbers in Example 7.3, rely on Theorem 5.4, a long exact sequence quoted from the companion preprint [KV25] without proof in this manuscript. Because [KV25] is an unpublished preprint that may itself invoke Theorem 1.1 of the present paper, the two papers form a coupled pair, and the concrete formulas for the two polytope classes are conditional on the soundness of [KV25]. To make the present paper self-contained for its advertised conclusions, the authors should either include a proof of Theorem 5.4 in this manuscript or explicitly state where it is proved and confirm that the proof does not use Theorem 1.1.","section":"Section 5.2, Theorem 5.4; used in Sections 6.2 and 7.3"},{"comment":"The proof invokes [PP24, Proposition 6.4] and [KV24] as black boxes to conclude that the local contribution K^{k,l} at the torus fixed point vanishes unless l+1 ≤ k ≤ n-2 and is Hodge-Tate. The contents of these propositions are not stated, so the reader cannot verify the key vanishing and purity assertions from the present text. Please reproduce the statements (or provide self-contained proofs) and explain exactly how they imply the required properties of K^{k,l}.","section":"Section 5.1, proof of Theorem 1.1"}],"minor_comments":[{"comment":"In deriving the Hodge-Du Bois diamond, the paper uses the fact that the Hodge structure of H^i(X) is mixed of Hodge-Tate type to identify the Hodge-Deligne polynomial with the weight filtration. It would help to spell out this identification explicitly, since the Hodge-Deligne polynomial is typically defined via the Hodge filtration.","section":"Section 7.3, proof of Theorem 1.12"},{"comment":"The notation Ish^l_σ is introduced for the degree-zero part of the Ishida complex, but Lemma 4.6 and elsewhere reuse similar notation for the full complex associated to a cone τ. Please distinguish the degree-zero part from the full complex, for example by a subscript or a parenthetical.","section":"Section 4, Definition 4.5"},{"comment":"The definition of the sets eP_m inside the proof depends on the shelling order ≺, but this dependence is not made explicit in the notation. The reader must infer that eP_m is relative to a fixed shelling of σ; please state this.","section":"Section 4, Lemma 4.9"},{"comment":"The face numbers f_l of the simple polytope are quoted from Grünbaum with a reindexing footnote. It would be helpful to display the reindexing formula explicitly in the example so that the subsequent use of Theorem 1.12 can be checked without consulting the reference.","section":"Example 7.3"}],"recommendation":"major_revision","confidential_remarks":"The main concern is the unproved Theorem 5.4 from the companion preprint [KV25]. This is not a routine citation: Proposition 1.6 and Theorem 1.12 are central advertised results, and the proof of Theorem 1.12 is a direct application of that theorem. I would ask the editor to verify the status of [KV25] and, if possible, to have the companion paper refereed jointly or to request the authors to prove the needed statement in this manuscript. The other external inputs, [PP24] and [KV24], are also preprints but are less likely to be circular."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear colleague,\n\nThe headline: this is a serious paper, and the main structure theorem is probably right, but the two big payoff formulas rest on a theorem you can't check in this text. Treat Proposition 1.6 and Theorem 1.12 as conditional on [KV25, Thm 1.4] until that companion is available.\n\nWhat's genuinely new: Theorem 1.1 gives a complete description of the mixed Hodge module QH'_X for any toric variety in terms of intersection cohomology Hodge modules on torus-invariant subvarieties, with a tight weight range and Tate twists. That is a strong result, and the proof is a coherent induction over orbits with spectral sequences. The depth inequality for Du Bois complexes (Theorem 1.3) is an elegant consequence, and the Hodge–Tate corollary is honestly flagged as likely known. The explicit formulas for cones over simplicial and simple polytopes are new and useful. Appendix A is candid: it computes the multiplicities up to dimension 5 and says plainly why dimension 6 doesn't close.\n\nThe soft spot is exactly where the reader's stress-test lands. Theorem 5.4 is quoted without proof from [KV25]; Proposition 1.6 and Theorem 1.12 depend on it. The companion may in turn use Theorem 1.1, so you need both papers to verify either fully. That's a normal scientific dependency, not fatal, but it's load-bearing for the concrete formulas. The proof of Theorem 1.1 itself doesn't use Theorem 5.4; its external inputs [PP24, Prop 6.4] and [KV24] are also black boxes, but those are published or separately available and less likely to be circular. The spectral sequence steps in the proof of Theorem 1.1 are dense but hang together; I didn't see an internal contradiction.\n\nSo: the central argument is likely sound, the dependence is the issue. The paper deserves a serious referee, but the referee should be instructed to verify the [KV25] dependency, and I'd want the authors to either include a proof of Theorem 5.4 or a precise statement with a no-cycle guarantee before acceptance.\n\nRecommendation: send it out; the main theorem is important enough to referee even in conditional form, but the final decision should wait on the companion.","headline":"A substantial and likely correct structure theorem for toric local cohomology whose explicit polytope formulas are conditional on an unproved companion theorem.","tokens_in":41069,"tokens_out":1828,"would_cite":false,"duration_ms":17508,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["14B05","14B15","14F10","14M25","32S35","52B20"],"pacs":[],"model":"deepseek-v4-flash","headline":"Every proper toric variety has Hodge–Tate cohomology","keywords":["toric varieties","mixed Hodge modules","local cohomology","intersection cohomology","Hodge–Tate type","Ishida complex","Hodge–Du Bois numbers","local cohomological defect"],"falsifier":"For a 4-dimensional non-simplicial cone $\\sigma$ over a simplicial polytope, compute the Ishida complex $\\mathrm{Ish}^3_\\sigma$ directly from the face lattice: Theorem 1.3 predicts $H^3(\\mathrm{Ish}^3_\\sigma)=0$ while Theorem 1.4 predicts $H^1(\\mathrm{Ish}^3_\\sigma)\\neq 0$. A single cone where either prediction fails would refute the corresponding theorem.","tokens_in":39860,"feed_emoji":"📐","tokens_out":11504,"duration_ms":102467,"temperature":0.7,"pith_summary":"This paper sets out to compute the mixed Hodge module structure of the trivial Hodge module of a toric variety $X$ — the object whose cohomology computes the singular cohomology of $X$. The central claim is that this object is supported in only a narrow band of cohomological degrees, and that every weight-graded piece is a direct sum of intersection cohomology Hodge modules of torus-invariant subvarieties, with Tate twists and multiplicities determined by the combinatorics of the fan. If the claim is right, it settles the local cohomological defect of toric varieties, gives the depth of the Du Bois complexes of a toric variety, and yields explicit Betti numbers and Hodge–Du Bois diamonds for cones over simplicial and simple polytopes. A direct corollary is that the singular cohomology of every proper toric variety is mixed of Hodge–Tate type.","feed_headline":"Every proper toric variety has Hodge–Tate cohomology","feed_subtitle":"A mixed Hodge module decomposition also fixes local cohomological defect and Betti numbers for polytope cones.","key_machinery":"The central object is the trivial Hodge module $\\mathrm{QH}'_X := \\mathrm{QH}_X[n]$ in the derived category of mixed Hodge modules, together with its Verdier dual, whose cohomology sheaves are the local cohomology sheaves of an affine embedding. The Ishida complex $\\mathrm{Ish}^l_X$, defined as the Grothendieck dual $\\mathrm{RHom}_{O_X}(\\Omega^l_X,\\omega_X)$, is a combinatorial complex of structure sheaves of torus-invariant subvarieties; it is the bridge between the Hodge module structure and the depth statements. The proof machinery combines the strict-support decomposition of pure Hodge modules, two spectral sequences for stalk cohomology at the torus fixed point, induction along open charts $U_\\tau \\simeq V_\\tau \\times O_\\tau$, and exactness results proved by lexicographic shellings. The simplicity conditions on the cone control which faces contribute to the decomposition.","core_discovery":"Let $X$ be an $n$-dimensional toric variety. Theorem 1.1 asserts that $\\mathrm{QH}'_X$ (the trivial Hodge module, shifted so that its underlying perverse sheaf is the constant sheaf) has cohomology only in degrees $[-(n-3),0]$. Its zero-th cohomology has weights in $[2,n]$, and the top weight piece is the intersection cohomology Hodge module $\\mathrm{IC}^H_X$. For $k\\ge l+1$, the weight $n-k$ graded piece of $H^{-l}\\mathrm{QH}'_X$ decomposes as $\\bigoplus_{j\\ge1}\\bigoplus_{\\lambda\\in P_{k+2j}} \\mathrm{IC}^H_{S_\\lambda}(-j)^{a^{l,j}_\\lambda}$, where $S_\\lambda$ is the torus-invariant closed subvariety corresponding to the cone $\\lambda$ and the multiplicities $a^{l,j}_\\lambda$ depend only on $\\lambda$. The paper proves this by analyzing the stalk of the complex at the torus fixed point and by induction over torus orbits, then dualizes to control local cohomology sheaves and the Ishida complex.","pith_inferences":["If the decomposition of Theorem 1.1 is as canonical as stated, the same intersection-cohomology summands should determine the Hodge filtration on all local cohomology sheaves, not just their support and weights; this could be tested by computing the graded de Rham complexes of the summands.","The formulas for simple polytopes turn a Hodge-theoretic computation into polytope face arithmetic; one can generate random simple lattice polytopes and compare the predicted $h^{p,q}$ with an independent computation of Du Bois cohomology as a check.","The dependence on the companion paper [KV25] suggests the two results should be read as one package; if the companion's long exact sequence can be proved without Theorem 1.1, the explicit polytope formulas become unconditional.","The coefficients $a^{l,j}_\\lambda$ are claimed to depend only on the cone $\\lambda$; an algorithm built from the Ishida complex that computes them in dimension 6 and beyond would give an independent test of Theorem 1.1 and a purely combinatorial route to the depth statements."],"forward_implications":["Every proper toric variety has singular cohomology groups whose Deligne weight pieces are all of Hodge–Tate type, so nonzero cohomology classes appear only in bidegrees $(p,p)$ after passing to associated graded.","For any toric variety, $\\mathrm{lcdef}(X)\\le \\max\\{0,n-3\\}$, and the paper recovers this bound from the mixed Hodge module decomposition rather than from the earlier Cohen–Macaulay argument.","The Du Bois complexes of an $n$-dimensional toric variety satisfy $\\mathrm{Ext}^l_{O_X}(\\Omega^k_X,\\omega_X)=0$ for $l+k>n$ and $\\mathrm{Ext}^{n-k}_{O_X}(\\Omega^k_X,\\omega_X)=0$ for $k\\le n/2$; hence $\\mathrm{depth}(\\Omega^k_X)\\ge k$, with strict inequality for $0