{"id":"f25530d0-a1f8-4515-9898-ba7b26d6b756","arxiv_id":"2412.03476","paper_version":2,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Toric sheaves on smooth projective toric varieties are encoded by polyhedral 'Weil decorations', which are used to construct universal extensions of nef line bundles and a spectral sequence computing sheaf cohomology from reduced cohomology of polyhedral subsets.","lead":"This math paper gives a way to translate 'equivariant reflexive sheaves' (certain symmetry-invariant geometric structures) on toric varieties into polyhedral shapes, and uses the shapes to build universal extensions of line bundles and to compute cohomology groups. It matters because it turns sheaf computations into polyhedral inclusion/exclusion problems that can be handled with pictures and finite combinatorial data.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 6.1 depends on Lemma 7.10's claim that the dual boundary subcomplex Δ' contracts to u; the boundary case is only justified by Figure 15 and the ad hoc relations (54)-(55), so exactness of K(F)^• is not fully established.","rationale":"The reader identified exactly the same load-bearing premise: the proof of Theorem 6.1 hinges on exactness of K(F)^•, whose critical input is Lemma 7.10's contractibility claim. My reading of Section 7 confirms this is the least secure point. The paper's own Remark 7.8 concedes the nonstandard behavior of the boundary relations, and the proof of Lemma 7.10 is a figure plus an appeal to Figure 15 rather than a derivation. Because R lim← F• = F is proven stalkwise via (52), any failure of exactness at a single boundary point would break the sheaf-theoretic bridge and with it the spectral sequence of Theorem 8.3. The second reader-flagged item, Lemma 6.11, is also a contractibility assertion but is explicitly deferred to a published source [AP20]; even if that source is accepted, it does not cover arbitrary strata, so Lemma 7.10 remains the essential gap. The indexing typo in Theorem 8.3 (S<T instead of S≤T) is real but does not threaten the method: it is an obvious correction and does not affect the spectral sequence once G^0 is defined as ⊕_S G_S. Thus the verdict CONDITIONAL is appropriate: the framework is coherent and the examples work, but the central theorem is not fully rigorous until Lemma 7.10 is either proved by a complete contractibility argument or checked computationally. I do not recommend ACCEPT, and nothing here suggests REJECT.","tokens_in":42609,"tokens_out":13821,"duration_ms":129578,"concrete_test":"Implement a computational check of Lemma 7.10 for a nontrivial family of smooth projective toric surfaces (e.g., P^2 and the Hirzebruch surfaces F_n). Fix an ample Δ and several amply decorated toric sheaves, including line bundles with different D+(S) and a rank-2 example. For a dense sample of u∈Δ—including vertices, relative interiors of edges, and interior points—and for every stratum S, compute Σ_S(1) = {ρ∈Σ(1) | min⟨D+(S),ρ⟩ <_ρ ⟨u,ρ⟩} using (54)-(55), form the complex C(Σ_S)^• of (53), and compute its cohomology (equivalently, the reduced homology of the dual subcomplex Δ' of ∂Δ). If any instance has nonzero H^•(C(Σ_S)^•), Lemma 7.10 is false and Theorem 6.1 loses its proof; if thousands of random cases all yield exact complexes, the contractibility claim is supported and the remaining gap is a missing rigorous proof, not a counterexample.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central isomorphism H^ℓ(X,E)_0 ≅ H^ℓ(Δ,F(E)) (Theorem 6.1) is proved by showing R lim← F• = F (Proposition 7.5). By Lemma 7.6 this reduces to exactness of the augmented stalk complex K(F)^• in (52). The key exactness input is Lemma 7.10: for every stratum S and u∈Δ, the Čech complex C(Σ_S)^• of the subfan Σ_S is exact because the dual boundary subcomplex Δ' 'can be contracted to u' (Subsection 7.2.2, Figure 15). This is the single point where a false geometric assertion would make the argument collapse: if C(Σ_S)^• has nonzero cohomology for some (S,u), then the strata double complex C■ has nontrivial cohomology, Lemma 7.11 is false, R lim← F• ≠ F, and Theorem 6.1 fails. The proof of Lemma 7.10 is not a complete argument: contractibility of Δ' is asserted from a figure, with the boundary cases (u∈∂Δ) regulated by relations (54)-(55) that the paper itself flags as nonstandard in Remark 7.8. For u on a facet with ⟨u,ρ⟩=min⟨Δ,ρ⟩, the inequalities reverse strictness, so Δ' may contain facets that would be excluded in the standard case; no proof shows the resulting subcomplex is still contractible. Lemma 6.11 also defers an analogous contractibility claim for line bundles to [AP20, Claim 3.3.2], but that does not cover arbitrary strata. The indexing error in Theorem 8.3 (sums over S<T instead of S≤T) is real but cosmetic; the contractibility of Δ' is load-bearing.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops a polyhedral framework for toric sheaves (equivariant reflexive sheaves) on smooth projective toric varieties. Every toric sheaf is encoded by a Weil decoration, a map from the nonzero vectors of the associated torus-invariant vector space to toric divisors, equivalently to virtual polytopes. The authors use this to construct the torus-invariant universal extension of two nef line bundles via inclusion/exclusion of polyhedra, and then to attach to each toric sheaf E a constructible sheaf F(E) on an ample polytope Δ. The central result, Theorem 6.1, asserts that H^ℓ(X,E)_0 is isomorphic to H^ℓ(Δ,F(E)); a spectral sequence in Theorem 8.3 computes these groups from reduced singular cohomology of polyhedral subsets of Δ. The paper closes with worked examples, including line bundles and the twisted tangent sheaves of the projective plane.","tokens_in":42848,"tokens_out":3769,"duration_ms":39729,"significance":"If the main theorem is correct, it gives a genuinely polyhedral-topological description of the T-invariant cohomology of every toric sheaf after a twist, which is a substantial generalization of the known line-bundle formula of [ABKW20] and [AP20]. The construction of the universal extension by Weil decorations and inclusion/exclusion sequences is elegant and appears to explain and extend [AFH23], including cases where the virtual intersection is not an honest lattice polytope. The spectral sequence of Theorem 8.3, with E1-terms involving only reduced cohomology of polyhedral subsets, is concrete and well suited to examples such as the tangent sheaf of P2. The paper is ambitious, mostly well organized, and the large supply of worked examples is a real strength. The main reservation is that two load-bearing geometric inputs in the proof of Theorem 6.1 are only sketched or deferred.","major_comments":[{"comment":"Lemma 7.10 is the key exactness input: the subfan Čech complex C(Σ_S)• is exact because the dual boundary subcomplex Δ′ 'can be contracted to u'. The proof of this contractibility is not a complete argument: for u in the relative interior of Δ the picture in Figure 15 is plausible, but for u∈∂Δ the relations (54) and (55) reverse strictness, so Δ′ may include facets with min⟨D+(S),ρ⟩ = ⟨u,ρ⟩ that the standard case would exclude. Remark 7.8 explicitly flags that in this boundary regime a>ρ b need not imply a≥ρ b. No proof is given that the resulting larger subcomplex is still contractible. This matters because Lemma 7.10 is used to show exactness of the augmented stalk complex K(F)• in (52), which is exactly what Proposition 7.5 and hence Theorem 6.1 require. The authors should supply a rigorous proof of the contractibility of Δ′ for every stratum S and every u∈Δ, with the boundary cases handled explicitly rather than by appeal to a figure.","section":"§7.2.2, Lemma 7.10"},{"comment":"Lemma 6.11 asserts that for a line bundle with ample twist and Δ nef, the local sheaves Fσ have vanishing higher cohomology; the proof is only sketched and refers to [AP20, Claim 3.3.2] for the key claim that the set S(σ) is empty or retractible to rσ. This lemma is load-bearing: Proposition 6.7, which follows from it, is used in Lemma 7.3 to conclude that RΓΔ(F•) = ΓΔ(F•), and that equality is needed in equation (49) to pass from the Klyachko–Čech complex to H(Δ, R lim← F•). The manuscript should either give a complete proof of Lemma 6.11 or state precisely which assertion is imported from [AP20] and verify that the cited claim covers all cones σ∈Σ, not only maximal cones. As written, this is a second gap in the proof of the central theorem.","section":"§6.3, Lemma 6.11 and Proposition 6.7"},{"comment":"The spectral sequence in Eq. (67) is stated with sums over S<T and chains chℓ(S,T) for S<T. For ℓ=0, ch0(S,T) is empty unless S=T, so the displayed E1-term makes E^{0,q}_1 = 0. This contradicts Theorem 8.2, where G^0 = ⊕_S G_S, and also contradicts the surrounding discussion and Figure 16, which visibly uses E^{0,q}_1 from the generic and minimal strata. The sums should be over S≤T, with the ℓ=0 case giving S=T. This is a straightforward indexing correction but it affects the statement of a main theorem and therefore should be fixed in the text.","section":"§8.2, Theorem 8.3, Eq. (67)"}],"minor_comments":[{"comment":"Remark 7.8 acknowledges the nonstandard behavior of the relations >ρ and ≥ρ on ∂Δ but gives no concrete example where a>ρ b occurs without a≥ρ b. Since this is exactly the regime that makes Lemma 7.10 delicate, a short explicit example would substantially help the reader.","section":"§7.2.2, Remark 7.8"},{"comment":"The last sentence of Section 8.5 ends with 'In particular, it follows that' followed immediately by Section 9; the intended display or conclusion appears to be missing. Please complete the sentence and give the promised Euler-characteristic conclusion.","section":"§8.5, final paragraph"},{"comment":"The derivation of H^ℓ(Δ,F) = H̃^{ℓ−1}(Z) uses the reduced cohomology convention in which H̃^{-1}(∅) = k; this is stated in the introduction but it would be helpful to repeat the convention at first use in the Gysin sequence.","section":"§6.2.1, Eq. (30)"},{"comment":"The Weil decoration of the tangent sheaf is written as D(ρ_i) = Δ_1, Δ_1−[1,0], Δ_1−[0,1]; the reader has to infer that the rays ρ_i are ordered consistently with the labels in Figure 9. A one-sentence explanation of the labeling would remove ambiguity.","section":"§6.2.3, Eq. (33)"}],"recommendation":"major_revision","confidential_remarks":"The paper is likely to be an important contribution if the two geometric inputs can be proved or properly imported. The authors lean heavily on their own published results ([ABKW20], [AP20], [AFH23]); this is not by itself a problem, but a short 'imported results' list in Section 7 would make the proof's dependencies transparent. The indexing error in Theorem 8.3 is real and should be corrected. No concerns about novelty or scope for this journal."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things you should know. First, this is the real thing: the paper gives a uniform polyhedral calculus for arbitrary rank toric sheaves, extending the rank-one Klyachko dictionary. The main theorem (6.1) identifies T-invariant cohomology with cohomology of a constructible sheaf F(E) on a polytope, and Theorem 8.3 gives a spectral sequence whose E1 terms are reduced cohomology of polyhedral subsets. The universal extension construction in Section 5 genuinely goes beyond [AFH23] by handling the case where the intersection is not a compatible lattice polytope. Second, the proof of the main theorem is not fully rigorous as written: two load-bearing geometric inputs are only sketched. Lemma 6.11 defers the line-bundle acyclicity to [AP20, Claim 3.3.2], and Lemma 7.10 asserts contractibility of the dual boundary subcomplex Δ' to u, justified by a figure and the ad hoc relations (54)-(55). If either fails for some boundary position, Proposition 7.5 and Theorem 6.1 collapse. That is a gap, not an observed falsehood; the standard case is convincing, but a complete argument is needed.\n\nThe indexing error in Theorem 8.3 (sums over S<T instead of S≤T) is real and should be fixed; also Example 8.9's first display misses a minus sign. These are cosmetic relative to the main argument.\n\nCredit where due: the stratification-height bound on complex length is nice; the treatment of the tangent sheaf examples is illuminating; the paper is honest about the necessity of ample decoration (Remark 6.4). The self-citation pattern is not a problem: they cite their own published line-bundle results as black boxes, which is legitimate. The rank>1 cohomology theorem is derived from scratch.\n\nWho is this for? Anyone working on toric vector bundles, Klyachko's classification, or polyhedral methods in sheaf cohomology. It deserves a serious referee; the gaps in Section 7 need to be filled before acceptance, but the framework is valuable and likely correct.","headline":"A substantial generalization of the rank-one Klyachko dictionary to arbitrary rank toric sheaves; the main theorem is plausible and useful, but two geometric contractibility lemmas in the proof are only sketched and need completion.","tokens_in":43652,"tokens_out":2226,"would_cite":true,"duration_ms":19966,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["14C20","14F06","14F08","14M25","18G10","52C07","55N05"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proves that the torus-invariant cohomology of a toric sheaf equals the cohomology of a constructible sheaf built from the polytopes of its Weil decoration.","keywords":["toric sheaves","reflexive sheaves","Weil decorations","constructible sheaves","polyhedra","cohomology","spectral sequence","toric varieties"],"falsifier":"Find a smooth projective toric variety, an amply decorated toric sheaf, and a boundary point $u \\in \\partial \\Delta$ for which the subcomplex $\\Delta'$ of facets with $\\min\\langle D^+(S), \\rho\\rangle \\ge_\\rho \\langle u, \\rho\\rangle$ has nonvanishing reduced homology, contradicting the contraction asserted in Lemma 7.10; then the complex $K(F)^\\bullet$ becomes non-exact and Theorem 6.1 must fail.","tokens_in":42118,"feed_emoji":"📐","tokens_out":5407,"duration_ms":44486,"temperature":0.7,"pith_summary":"This paper establishes that the torus-invariant cohomology of a toric sheaf on a smooth projective toric variety is a polyhedral-topology invariant. Concretely, after twisting to an ample 'materialisation', the invariant piece $H^\\ell(X,E)_0$ is isomorphic to the sheaf cohomology $H^\\ell(\\Delta, F(E))$ of a constructible sheaf built from the polytopes decorating the sheaf. It also constructs the torus-invariant universal extension of two nef line bundles through polyhedral inclusion/exclusion, and gives a spectral sequence that computes $H^\\ell(X,E)_0$ from reduced cohomology groups of polyhedral subsets. A sympathetic reader would care because it converts cohomological questions about equivariant reflexive sheaves into concrete convex geometry.","feed_headline":"Polyhedra capture toric sheaf cohomology","feed_subtitle":"A constructible sheaf on a polytope computes the torus-invariant cohomology exactly, via a new spectral sequence.","key_machinery":"The central objects are Weil decorations: maps $D$ from nonzero vectors of the torus-invariant vector space $E$ to toric divisors (or polytopes), satisfying the subadditivity condition $D(e+e') \\ge D(e) \\wedge D(e')$. They completely encode a toric sheaf, and after an ample twist they define the constructible sheaf $F(E)$ on $\\Delta$. The proof that $F(E)$ computes cohomology runs through the inverse system of local sheaves $F_\\sigma$ on the fan and hinges on exactness of the augmented stalk complex $K(F)^\\bullet$; the critical geometric input is Lemma 7.10, which asserts that the dual boundary subcomplex $\\Delta'$ contracts to the point $u$ for every stratum and boundary position.","core_discovery":"Theorem 6.1 states that if $X$ is a smooth projective toric variety and $E^+ = E(\\Delta)$ is amply decorated for $\\Delta \\in \\mathrm{Pol}^+(\\Sigma)$, then $H^\\ell(X,E)_0 \\cong H^\\ell(\\Delta, F(E))$, where $F(E)(U) = \\{e \\in E \\mid U \\subseteq D^+(e)\\}$ is a constructible sheaf on the polytope $\\Delta$. Together with Theorem 8.3, this says the $T$-invariant cohomology of every toric sheaf is the abutment of a spectral sequence whose $E_1$ terms are sums of reduced cohomology groups $\\tilde H^{q-1}(P(T))$ of polyhedral subsets $P(T) = \\Delta \\setminus \\mathrm{int}_\\Delta D^+(T)$.","pith_inferences":["The constructible-sheaf bridge suggests a practical algorithm: from a Weil decoration one can compute $H^\\ell(X,E)_0$ directly by polyhedral topology, bypassing resolutions by line bundles.","If the local acyclicity assumptions survive without smoothness, the same polyhedral formula might extend to singular or non-complete toric varieties.","The spectral sequence could be compared with the coherent-constructible correspondence to translate Morse-theoretic or tropical data on $\\Delta$ into sheaf cohomology.","Because the complex length is governed by the height of the stratification, the method is especially short for rank-two sheaves, as the paper demonstrates."],"forward_implications":["Amply decorated toric sheaves are acyclic: $H^\\ell(X,E) = 0$ for all $\\ell \\ge 1$ (Corollary 6.3).","The $T$-invariant cohomology of any toric sheaf is computable from the reduced cohomology of polyhedral subsets via the spectral sequence of Theorem 8.3.","The torus-invariant universal extension of two nef line bundles admits an explicit polyhedral inclusion/exclusion description (Theorem 5.2).","The Euler characteristic of a toric sheaf in degree zero is governed by the Möbius function of its stratum poset (Corollary 8.8).","Shifting $\\Delta$ by $m \\in M$ recovers the $m$-graded piece $H^\\ell(X,E)_m$ (Remark 6.2)."],"supporting_citations":[{"why":"Supplies the line-bundle cohomology formula $H^\\ell(X,O_X(D))_0 \\cong \\tilde H^{\\ell-1}(\\nabla_- \\setminus \\nabla_+)$ that the paper extends from line bundles to general toric sheaves.","marker":"[ABKW20]"},{"why":"Provides the acyclicity of local sheaves $F_\\sigma$ for line bundles (Claim 3.3.2), used in Lemma 6.11 to prove Proposition 6.7.","marker":"[AP20]"},{"why":"Determines the toric universal extension $\\mathrm{Ext}(\\nabla_-,\\nabla_+)_0$ and handles the case $\\nabla_+ \\subseteq \\nabla_-$, which the paper's Theorem 5.2 generalizes.","marker":"[AFH23]"},{"why":"Supplies the Klyachko filtration formalism and the Klyachko-Čech complex used to set up the cohomology computation.","marker":"[Kly90]"},{"why":"Provides the sheaf theory on finite poset spaces used in the inverse-limit argument that links the constructible sheaf $F(E)$ to the toric sheaf $E$.","marker":"[Lad08]"}],"fun_headline_variants":["Toric sheaf cohomology via constructible sheaves on polyhedra","Polytopes compute torus-invariant cohomology exactly","New spectral sequence links toric sheaves to polyhedral subsets","Cohomology of toric sheaves from reduced cohomology of polyhedra"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The theorem stands or falls on the geometric contraction claim in Lemma 7.10: for every stratum and every boundary position of $u$, the dual boundary subcomplex $\\Delta'$ must be contractible to $u$; together with the deferred acyclicity of local sheaves for line bundles in Lemma 6.11, this exactness drives the entire bridge.","fun_headline_variants_meta":{"raw":{"variants":["Toric sheaf cohomology via constructible sheaves on polyhedra","Polytopes compute torus-invariant cohomology exactly","New spectral sequence links toric sheaves to polyhedral subsets","Cohomology of toric sheaves from reduced cohomology of polyhedra"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000453,"raw_usage":{"total_tokens":2231,"prompt_tokens":850,"completion_tokens":1381,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":466,"completion_tokens_details":{"reasoning_tokens":1300}},"tokens_in":466,"tokens_out":1381,"duration_ms":9616,"temperature":1.0,"reasoning_tokens":1300,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T22:25:30.705810+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Find a smooth projective toric variety, an amply decorated toric sheaf, and a boundary point $u \\in \\partial \\Delta$ for which the subcomplex $\\Delta'$ of facets with $\\min\\langle D^+(S), \\rho\\rangle \\ge_\\rho \\langle u, \\rho\\rangle$ has nonvanishing reduced homology, contradicting the contraction asserted in Lemma 7.10; then the complex $K(F)^\\bullet$ becomes non-exact and Theorem 6.1 must fail.","supporting_citations":[],"review_version":1}