Local cohomology and singular cohomology of toric varieties via mixed Hodge modules

Hyunsuk Kim; Sridhar Venkatesh

REVIEW 2 major objections 4 minor 1 cited by

Local cohomology and singular cohomology of toric varieties via mixed Hodge modules

T0 review · 2 major / 4 minor · reviewed 2026-08-15 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Every proper toric variety has Hodge–Tate cohomology

desk verdict A substantial and likely correct structure theorem for toric local cohomology whose explicit polytope formulas are conditional on an unproved companion theorem. read the letter →

arxiv 2505.10417 v2 pith:K75L3WC5 submitted 2025-05-15 math.AG math.CO

Hyunsuk Kim , Sridhar Venkatesh This is my paper · ORCID

classification math.AGmath.CO MSC 14B0514B1514F1014M2532S3552B20

keywords toricvarietiesmixedHodgemoduleslocalcohomologyintersectionHodge–TatetypeIshidacomplexHodge–DuBoisnumberscohomologicaldefect

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

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.

What carries the argument

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.

What would settle it

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.

Watch

Extended reading notes

Core claim

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.

Load-bearing premise

The load-bearing premise is the correctness of the long exact sequence stated as Theorem 5.4 and proved in the companion preprint [KV25]: if inserting an interior ray does not produce that exact sequence, the explicit lcdef and Hodge–Du Bois formulas for polytope cones collapse.

Editorial extensions

If this is right

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 $0<k\le n/2$.
For cones over simplicial polytopes, the local cohomological defect is exactly $n-3$ when the cone is not simplicial, and all relevant Ext sheaves are concentrated in one degree with dimensions given by $g$-polynomial coefficients.
For cones over simple polytopes, $\mathrm{lcdef}(X)=0$, and the Hodge–Du Bois diamond of the projective toric variety of a simple lattice polytope is given by an explicit face-number formula, which is highly asymmetric.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

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.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Dear colleague,

The 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.

What'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.

The 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.

So: 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.

Recommendation: send it out; the main theorem is important enough to referee even in conditional form, but the final decision should wait on the companion.

Referee Report

2 major / 4 minor

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.

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 (2)

[Section 5.2, Theorem 5.4; used in Sections 6.2 and 7.3] 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 5.1, proof of Theorem 1.1] 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}.

minor comments (4)

[Section 7.3, proof of Theorem 1.12] 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 4, Definition 4.5] 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 4, Lemma 4.9] 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.
[Example 7.3] 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.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity; the central Hodge-module derivation is self-contained, with companion-preprint dependencies that are correctness risks rather than circular reductions.

full rationale

The paper's main theorem (Theorem 1.1) is proved directly from Saito's mixed Hodge module formalism, the strict-support decomposition of pure Hodge modules, an induction on dimension, and the external bound [PP24, Proposition 6.4]; none of the displayed spectral-sequence steps (Steps 1–4 of §5.1) assumes the conclusion. The multiplicities a^{l,j}_λ are not fitted: Theorem 1.5 and Theorem 1.10 compute them from the face numbers via the stalk cohomology of the intersection cohomology Hodge module [KV24] and via the Euler characteristic of the Ishida complex, respectively, and Theorem 1.12 derives the Hodge–Du Bois diamond from Grothendieck duality, the long exact sequence of Theorem 5.4, and the standard Hodge–Deligne polynomial of a toric variety. The only self-citations are the companion preprints [KV24] and [KV25]; Theorem 5.4 is quoted from [KV25] without proof and is load-bearing for Proposition 1.6 and Theorem 1.12, but the paper gives no equation showing that Theorem 5.4 is equivalent to Theorem 1.1, and there is no fitted input renamed as a prediction. Whether [KV25] itself depends on Theorem 1.1 is a possible dependency-cycle risk, not an exhibited circular reduction, so under the stated rules it belongs to correctness risk rather than to the circularity score.

Assumptions & free parameters 0 free parameters · 8 assumptions · 0 invented entities

The central claim rests on standard Hodge module theory (Saito), standard toric geometry, and three external results from preprints: [KV24] and [PP24] used in the proof of Theorem 1.1, and [KV25] used for the explicit polytope computations. No parameters are fitted to data, and no new entities are introduced; the coefficients a^{l,j}_λ are outputs of the structure theorem, not inputs.

assumptions (8)

domain assumption Saito's six-functor formalism, the category of polarizable Hodge modules, and the Decomposition theorem.
Invoked throughout Section 2.3 and used for Lemma 3.2, Theorem 1.1, and Lemma 5.3; these are accepted black boxes of the theory.
domain assumption Structure theorem of pure Hodge modules: decomposition by strict support and unique extension of variations of Hodge structures.
Used in Step 1 of the proof of Theorem 1.1 to decompose gr^W of cohomologies of QH'_X into summands supported on torus-invariant subvarieties.
domain assumption [KV24]: the generating function dR_σ(K,L) for graded de Rham of IC^H_X and the stalk cohomology H_σ(q) are determined combinatorially by the cone σ.
Quoted in Section 3.3 and used to compute the coefficients in Sections 6.1, 7.2, and Appendix A. It is the authors' own prior result, independent of this paper.
domain assumption [PP24, Proposition 6.4]: the point-supported summand K^{k,l} in the decomposition of gr^W_{n-k}H^{-l}QH'_X vanishes unless k≥l+1.
Used in the proof of Theorem 1.1, Step 1, to establish the weight range; cited without proof from a preprint.
ad hoc to paper Theorem 5.4, quoted as [KV25, Theorem 1.4]: inserting an interior ray into the cone gives a long exact sequence relating cohomologies of Ishida complexes of the original cone and the exceptional divisor.
Load-bearing for Proposition 1.6 and Theorem 1.12; it is stated here without proof and the proof lives in the authors' companion preprint [KV25].
domain assumption [DT16]: for a Cohen-Macaulay variety of dimension n, lcdef(X) ≤ max{0,n-3}.
Used in the Introduction and in the proof of Theorem 1.4 to show cones over simplicial polytopes reach the upper bound.
standard math Bruggesser-Mani: every polytope and every cone admits a shelling.
Used in Section 3.4 and in Lemma 4.9 to filter the Ishida complex and prove its exactness.
standard math Toric varieties are normal, have rational singularities, and are Cohen-Macaulay, so the dualizing complex ω^•_X is a shifted dualizing sheaf.
Invoked in Sections 2, 4, and 5 to identify ω^•_X[dim X] with ω_X and to apply Theorem 2.1; standard references [Ful93], [CLS11].

how reviews work

0 comments

Cite this review

Pith. "Pith review of Local cohomology and singular cohomology of toric varieties via mixed Hodge modules." pith.science (2026). https://pith.science/paper/K75L3WC5

@misc{pith2026250510417,
  author       = {Pith},
  title        = {Pith review of: Local cohomology and singular cohomology of toric varieties via mixed Hodge modules},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/K75L3WC5}},
  note         = {Machine review of arXiv:2505.10417}
}

abstract

Given an affine toric variety $X$ embedded in a smooth variety, we prove a general result about the mixed Hodge module structure on the local cohomology sheaves of $X$. As a consequence, we prove that the singular cohomology of a proper toric variety is mixed of Hodge-Tate type. Additionally, using these Hodge module techniques, we derive a purely combinatorial result on rational polyhedral cones that has consequences regarding the depth of reflexive differentials on a toric variety. We then study in detail two important subclasses of toric varieties: those corresponding to cones over simplicial polytopes and those corresponding to cones over simple polytopes. Here, we give a comprehensive description of the local cohomology in terms of the combinatorics of the associated cones, and calculate the Betti numbers (or more precisely, the Hodge-Du Bois diamond) of a projective toric variety associated to a simple polytope.

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

Lefschetz morphisms on singular cohomology and local cohomological dimension of toric varieties
math.AG 2025-06 conditional novelty 7.0 of 10

The local cohomological defect of an affine toric variety is characterized by Lefschetz cup-product maps on a projective toric variety of one dimension lower, which shows it is not a combinatorial invariant and allows...

Reference graph

Works this paper leans on

4 extracted references · 4 canonical work pages · cited by 1 Pith paper

[1]

Bruggesser and P

[BM71] H. Bruggesser and P. Mani, Shellable decompositions of cells and spheres , Math. Scand. 29 (1971), 197–205. MR328944 ↑16 [Bra06] Tom Braden, Remarks on the combinatorial intersection cohomology of fans , Pure Appl. Math. Q. 2 (2006), no. 4, 1149–1186. MR2282417 ↑6, 7 [CLS11] David A. Cox, John B. Little, and Henry K. Schenck, Toric varieties, Gradu...

work page 1971
[58]

MR4491455 ↑1, 3, 6, 7, 11, 12, 21 [Ogu73] Arthur Ogus, Local cohomological dimension of algebraic varieties, Ann. of Math. (2) 98 (1973), 327–365. MR506248 ↑1 44 HYUNSUK KIM AND SRIDHAR VENKATESH [OR24] Sebastian Olano and Debaditya Raychaudhury, On the local cohomology of secant varieties , preprint arXiv:2407.16688 (2024). ↑3 [Per24] Michael Perlman, Lo...

work page Pith review arXiv 1973
[2008]

MR2357361 ↑9 [Ish80] Masa-Nori Ishida, Torus embeddings and dualizing complexes , Tohoku Math. J. (2) 32 (1980), no. 1, 111–146. MR567836 ↑4, 8, 16 [Ish87] , Torus embeddings and de Rham complexes , Commutative algebra and combinatorics (Kyoto, 1985), 1987, pp. 111–145. MR951199 ↑4, 8, 16 [KV24] Hyunsuk Kim and Sridhar Venkatesh, The intersection cohomolo...

work page arXiv 1980
[2011]

MR2810322 ↑13, 14 [DB81] Philippe Du Bois, Complexe de de Rham filtr´ e d’une vari´ et´ e singuli` ere, Bull. Soc. Math. France 109 (1981), no. 1, 41–81. MR613848 ↑9 [dCM09] Mark Andrea A. de Cataldo and Luca Migliorini, The decomposition theorem, perverse sheaves and the topology of algebraic maps , Bull. Amer. Math. Soc. (N.S.) 46 (2009), no. 4, 535–633...

work page 1981

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.

[1] [1]

Bruggesser and P

[BM71] H. Bruggesser and P. Mani, Shellable decompositions of cells and spheres , Math. Scand. 29 (1971), 197–205. MR328944 ↑16 [Bra06] Tom Braden, Remarks on the combinatorial intersection cohomology of fans , Pure Appl. Math. Q. 2 (2006), no. 4, 1149–1186. MR2282417 ↑6, 7 [CLS11] David A. Cox, John B. Little, and Henry K. Schenck, Toric varieties, Gradu...

work page 1971

[2] [58]

MR4491455 ↑1, 3, 6, 7, 11, 12, 21 [Ogu73] Arthur Ogus, Local cohomological dimension of algebraic varieties, Ann. of Math. (2) 98 (1973), 327–365. MR506248 ↑1 44 HYUNSUK KIM AND SRIDHAR VENKATESH [OR24] Sebastian Olano and Debaditya Raychaudhury, On the local cohomology of secant varieties , preprint arXiv:2407.16688 (2024). ↑3 [Per24] Michael Perlman, Lo...

work page Pith review arXiv 1973

[3] [2008]

MR2357361 ↑9 [Ish80] Masa-Nori Ishida, Torus embeddings and dualizing complexes , Tohoku Math. J. (2) 32 (1980), no. 1, 111–146. MR567836 ↑4, 8, 16 [Ish87] , Torus embeddings and de Rham complexes , Commutative algebra and combinatorics (Kyoto, 1985), 1987, pp. 111–145. MR951199 ↑4, 8, 16 [KV24] Hyunsuk Kim and Sridhar Venkatesh, The intersection cohomolo...

work page arXiv 1980

[4] [2011]

MR2810322 ↑13, 14 [DB81] Philippe Du Bois, Complexe de de Rham filtr´ e d’une vari´ et´ e singuli` ere, Bull. Soc. Math. France 109 (1981), no. 1, 41–81. MR613848 ↑9 [dCM09] Mark Andrea A. de Cataldo and Luca Migliorini, The decomposition theorem, perverse sheaves and the topology of algebraic maps , Bull. Amer. Math. Soc. (N.S.) 46 (2009), no. 4, 535–633...

work page 1981