REVIEW 4 major objections 4 minor 17 references
Line Bundle Resolutions via the Coherent-Constructible Correspondence
T0 review · 4 major / 4 minor · reviewed 2026-08-12 · deepseek-v4-flash
Pith's one-line read A finite toric morphism's pushforward of the structure sheaf admits a minimal line bundle resolution of length equal to the codimension, with every Betti number given by compactly supported cohomology of a cube stratum intersected with…
desk verdict A genuinely useful toric geometry paper: minimal resolutions with explicit topological Betti numbers, but the proof rests on a chain of imported equivalences where Proposition 4.9 deserves close scrutiny. read the letter →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
What carries the argument
The load-bearing device is the commuting triangle of Theorem 3.14, which identifies the derived category of coherent sheaves on $X_\Sigma$, the category of constructible sheaves on the mirror torus with singular support in the cube skeleton $\Lambda^c$, and the derived category of modules over the entrance path algebra $A_{\mathrm{Ent}}$ of the cube stratification. Under this equivalence, line bundles $O(-a)$ correspond to indecomposable projectives $P_{[a]}$, the left and right wrapping functors move between the smaller singular-support category of the Coherent–Constructible Correspondence and the larger stratified category, and right wrapping preserves co-probe sheaves. For pushforwards, the mirror of $u_*O$ is $v_!\mathbb{C}_{\{0\}}$; Proposition 4.11 shows this object is represented, after a Serre twist and exodromy, by a pure module over $A_{\mathrm{Ent}}$, so its minimal projective resolution exists and transfers back to a minimal line bundle resolution. The Betti numbers are then Ext groups, computed as compactly supported cohomology of the strata intersections.
What would settle it
Compute the minimal resolution of the normalization module $N=\mathbb{C}[x_0,x_1,x_2,t]/(tx_1-x_0x_2,\,tx_0-x_1^2,\,t^2-x_1x_2)$ as a module over the Cox ring of $\mathbb{P}^2$, arising from the finite toric morphism $u:\mathbb{P}^1\to\mathbb{P}^2$ with lattice map $(a)\mapsto(2a,3a)$. Since $\dim\mathbb{P}^2-\dim\mathbb{P}^1=1$, the theorem predicts exactly the two-term complex $O(-2)^{\oplus 2}\to O(-1)\oplus O\to u_*O_{\mathbb{P}^1}$; any extra term, any missing summand, or a minimal resolution of length different from $1$ would refute Theorem 4.13.
Extended reading notes
Core claim
The paper's central discovery is a mirror-symmetry formula for minimal resolutions. Theorem 4.13 states that for a finite toric morphism $u:X_{\Sigma_1}\to X_{\Sigma_2}$ of smooth projective toric varieties, the pushforward $u_*O_{X_{\Sigma_1}}$ admits a minimal resolution by sums of line bundles of length $k=\dim X_{\Sigma_2}-\dim X_{\Sigma_1}$, and the multiplicity of $O(-a)$ in the $i$-th term is $\beta_{i,-a}=\dim H^i_c(S^c_{[a]}\cap V)$, where $V$ is the zero fiber of the dual map on tori and $S^c_{[a]}$ are the strata of the cube stratification of the mirror torus. The same mechanism gives, for an arbitrary coherent sheaf, a minimal line bundle resolution of length at most $n$ after a twist by a sufficiently high power of an ample line bundle. The proof transfers the sheaf across the non-equivariant Coherent–Constructible Correspondence, views it as a module over the entrance path algebra of the stratified torus, and resolves that module; minimality and the length bound come from the finite-dimensional algebra's homological dimension.
Load-bearing premise
The load-bearing premise is that the mirror-symmetry bridge is lossless: the non-equivariant Coherent–Constructible Correspondence and the exodromy identification with entrance-path-algebra modules are imported as black boxes, so if either equivalence fails for the toric varieties considered, the topological Betti formula would not transfer back to resolutions.
Editorial extensions
If this is right
- Any coherent sheaf on a smooth projective $n$-dimensional toric variety admits a minimal resolution by sums of line bundles from the finite collection, after a sufficiently high twist, with length at most $n$.
- For a finite toric morphism of smooth projective toric varieties, the pushforward of the structure sheaf admits a minimal line bundle resolution of length exactly the codimension $k$.
- The multiplicities in that resolution are topological invariants: $\beta_{i,-a}=\dim H^i_c(S^c_{[a]}\cap V)$, so Betti numbers can be computed from compactly supported cohomology on the mirror torus.
- Whenever the singular-support skeleton equals the cube skeleton, for example on projective space and products of projective spaces, the minimal resolution is unique and hence irreducible.
- The cellular resolutions of [HHL23] are recovered as the non-minimal CW version of this construction, and Frobenius pushforwards decompose as direct sums of line bundles.
Reading between the lines
- The Betti formula suggests a practical algorithm that avoids constructing differentials first: triangulate the stratified torus, intersect with $V$, compute $H^i_c$ of each stratum intersection, and assemble the terms of the minimal resolution; this could be benchmarked against direct computer algebra computations of syzygies.
- Because a generically finite toric morphism factors as a birational map followed by a finite one, the same mirror mechanism may yield minimal resolutions for birational pushforwards, with a length bound that adds a correction to the codimension; the paper does not work out this case.
- The uniqueness that holds when $\Lambda_\Sigma=\Lambda^c$ suggests that on projective space and products of projective spaces, minimal line bundle resolutions are canonical invariants of coherent sheaves; comparing them across toric embeddings may yield a toric analogue of multigraded Betti tables.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper studies minimal resolutions of coherent sheaves on smooth projective toric varieties by sums of line bundles from Bondal's finite collection Φ. The main results are Theorem 4.4, which gives a minimal resolution of length at most n for any coherent sheaf, and Theorem 4.13, which for a finite toric morphism u:XΣ1→XΣ2 gives a minimal resolution of u_*O_{XΣ1} of length k=dim XΣ2−dim XΣ1 with multiplicities β_{i,-a}=dim H^i_c(S^c_[a]∩V), where V is the zero fiber of the dual map on tori. The proof passes through Zhou's non-equivariant Coherent-Constructible Correspondence, the Favero–Huang exodromy identification with entrance path algebras, and wrapping functors between singular-support categories. A secondary result recovers the cellular resolutions of Hanlon–Hicks–Lazarev from a non-minimal injective resolution of the constant sheaf on V.
Significance. If correct, the paper gives a striking and useful bridge from topology to explicit minimal resolutions: the Betti numbers of the resolution of u_*O_{XΣ1} are computed by compactly supported cohomology of the cube stratification intersected with a subtorus. This generalizes and improves the Hanlon–Hicks–Lazarev resolutions, and Theorem 4.4 strengthens known bounds via the pdim bound of Favero–Huang. The main formulas are concrete and falsifiable, and the examples, including the Macaulay2 check for the normalization of the cubic, support the claimed terms. The chief weaknesses are the heavy reliance on imported equivalences and the terse, notationally inconsistent treatment of the functoriality of the mirror functor, which is load-bearing for the transfer of the topological Betti numbers.
major comments (4)
- [§4.3, Proposition 4.9 and equation (13)] The functoriality statement for finite toric morphisms is not written in a mathematically well-formed way. The map v is defined as v:M2,R/M2→M1,R/M1, so v_! is a pushforward from T^{n2} to T^{n1}. The expression v_!j_{0*}C_0 applies v_! to a sheaf on T^{n1}, which is not a valid composition. The proof also asserts that the right adjoint of v_! is ρR∘v_!, which is not a standard adjunction and is not justified. The intended statement is presumably that the mirror of u_*O_{XΣ1} is the pullback (up to the appropriate shift) of the skyscraper at 0 along v, i.e., the constant sheaf on V. Because this step transfers the topological computation into the algebraic resolution, the direction and the sign/shift bookkeeping must be corrected and proved explicitly.
- [§4.3, proof of Proposition 4.11] The purity argument for MΣ1 uses the equality t! = t* 'since t is proper by Proposition 2.21', but Proposition 2.21 only asserts that exit spaces are closed and contractible; it does not state compactness or properness. For the cube stratification the exit spaces are in fact compact polytopes under the strong convexity assumption, but this needs to be stated and proved. If an exit space were unbounded, the compactly supported cohomology computation that proves concentration in degree 0 would fail. This is a load-bearing step because it is how the paper concludes that MΣ1 is quasi-isomorphic to a pure module.
- [§4.3, proof of Theorem 4.13, equations (17)-(18)] The Betti-number computation conflates Hom in the derived category with the graded Hom complex. Equation (17) correctly asks for Ext^i(MΣ1,S[a]), but the following chain computes Hom_{D(mod-A)}(MΣ1,S[a]) and then concludes β_{i,-a}=dim H^i_c(S^c_[a]∩V). To make this valid, the chain should be an isomorphism of graded vector spaces, with the Serre-functor step accompanied by the appropriate shift. The final formula may be correct, but the homological-degree bookkeeping needs to be written out precisely; as it stands, the displayed equalities do not directly imply the claimed Ext computation.
- [§4.5, Proposition 4.25] The proof of acyclicity of the cellular injective complex relies on the assertion that every (k−1)-cell in the induced CW structure on V is contained in the closure of exactly two k-cells. This is true for a regular CW decomposition of a closed manifold, but the induced stratification of a subtorus is only introduced in §2.3; Proposition 2.8 establishes the CW properties for the torus T^n, not for the induced stratification of V. Since Example 2.11 shows that strata of a subtorus can have multiple connected components, the frontier axiom and regularity for S^V_cw should be proved before this assertion is used. This affects both the length bound in Theorem 4.13 and the recovery of the Hanlon–Hicks–Lazarev resolutions.
minor comments (4)
- [§1 and throughout] There are many typos and minor grammatical errors: 'Strumfels' should be 'Sturmfels', 'Hirzerbruch' should be 'Hirzebruch', 'verticies' should be 'vertices', 'morpshism' should be 'morphism', and 'startification' should be 'stratification'. A careful proofreading pass is needed.
- [§1, paragraph on Bayer–Peeva–Sturmfels] The sentence about 'the pioneering work of Bayer, Strumfels, and Peeva' does not cite the correct reference; the citation [BHS21] is the Bruce–Heller–Sayrafi paper, not the monomial-ideal resolution work. The appropriate reference should be added.
- [§4.3, Example 4.17] The phrase 'the zero-dimensional stratum has a trivial intersection' is ambiguous: if the intersection is a single point, the compactly supported cohomology is indeed one-dimensional in degree 0, but 'trivial' normally means empty. Please rephrase.
- [§4.3, Proposition 4.11] The notation MΣ1[a] is used before the module value at a vertex is defined; a short sentence explaining that this is the stalk at the vertex [a] would improve readability.
Circularity Check
No circularity: the topological Betti formula is derived from imported equivalences, not assumed.
full rationale
Theorem 4.13 does not define its Betti numbers to be the compactly supported cohomology groups; it computes them as Ext groups over the entrance-path algebra and then identifies those Ext groups topologically via the exodromy equivalence, Serre duality, and proper base change (Eq. 17–18 and the “same argument as in Proposition 4.11”). The existence of the line-bundle resolution is obtained by taking a minimal projective resolution of the module MΣ over the finite-dimensional algebra A_Ent and applying −⊗_A T, using the commuting diagram in Theorem 3.14 and Corollary 3.15. None of these steps presupposes the formula β_{i,−a} = dim H^i_c(S^c_[a] ∩ V); the formula is the output of the computation. The load-bearing inputs—Zhou’s non-equivariant CCC, Favero–Huang exodromy and homological-dimension bounds, Treumann’s functoriality, and Kuo’s wrapping adjunction—are external results with separate proofs and stated assumptions that do not include the present theorem. Although FH22/FH23 share an author, they are not consequences of this paper and are used as independent published theorems. The examples are checked against known Koszul and Macaulay2 resolutions, but the general claim is not fitted to those examples. The least-checked step, Proposition 4.9’s functoriality for finite toric morphisms, is a correctness risk rather than a circularity, since an error there would invalidate the transfer of the topological computation without making it an input. Accordingly the paper receives no circularity flags.
Assumptions & free parameters
assumptions (6)
- domain assumption Non-equivariant Coherent-Constructible Correspondence for smooth projective toric varieties (κ: Coh(XΣ) ≃ Sh(Tn,ΛΣ)).
- domain assumption Functoriality of Treumann's mirror functor under finite toric morphisms, giving u_* ↔ v_! commuting diagrams.
- domain assumption Exodromy equivalence Sh_S(Tn) ≃ mod-A_Ent and identification of probe/co-probe sheaves with projectives/injectives.
- domain assumption Projective dimension bound pdim(A_Ent) ≤ dimension of the stratified torus.
- domain assumption Generation of Sh(Tn,Λc) by pure probe sheaves.
- standard math Standard homological algebra: Serre vanishing, Serre duality, proper base change, Verdier duality.
Cite this review
Pith. "Pith review of Line Bundle Resolutions via the Coherent-Constructible Correspondence." pith.science (2026). https://pith.science/paper/NAGBFFA3
@misc{pith2026241117873,
author = {Pith},
title = {Pith review of: Line Bundle Resolutions via the Coherent-Constructible Correspondence},
year = {2026},
howpublished = {\url{https://pith.science/paper/NAGBFFA3}},
note = {Machine review of arXiv:2411.17873}
}
abstract
We consider a finite collection of line bundles $\Phi$ introduced by Bondal on a smooth, projective toric variety $X$. For any coherent sheaf $F$ on $X$, we construct minimal resolutions of $F$ by line bundles in $\Phi$, up to twist, with length bounded by the dimension of $X$ and provide explicit formulae for their Betti numbers. For a toric subvariety $Y \subset X$ of codimension $k$, we give a construction of the minimal resolution of $f_{*}\mathcal{O}_{Y}$ of length $k$ by line bundles in $\Phi$ and relate their Betti numbers to the topology of a stratified real torus. Additionally, we recover a (generally non-minimal) cellular resolution of $f_{*}\mathcal{O}_{Y}$ constructed in Hanlon-Hicks-Lazarev. Aspects of our proof run through the Coherent Constructible Correspondence, a form of homological mirror symmetry for toric varieties.
Reference graph
Works this paper leans on
-
[1]
Christine Berkesch, Daniel Erman and Gregory G. Smith. Virtual Resolutions for a Product of Projective Spaces, 2017, Algebraic Geometry, 7 (2020) no. 4, 460-481; arXiv:1703.07631. DOI: 10.14231/AG-2020-013
work page Pith review arXiv 2020
-
[2]
A categorification of Morelli's theorem
Bohan Fang, Chiu-Chu Melissa Liu, David Treumann, Eric Zaslow. A categorification of Morelli's theorem. Invent. Math. 186 (2011), no.1, 79-114
work page 2011
-
[3]
Alexey Bondal Derived categories of toric varieties Oberwolfach Rep. 3 (2006), 284-286
work page 2006
-
[4]
Juliette Bruce, Lauren Cranton Heller and Mahrud Sayrafi. Characterizing Multigraded Regularity and Virtual Resolutions on Products of Projective Spaces, 2021; arXiv:2110.10705
arXiv 2021
-
[5]
Sheaves, Cosheaves and Applications, 2013; arXiv:1303.3255
Justin Curry. Sheaves, Cosheaves and Applications, 2013; arXiv:1303.3255
arXiv 2013
-
[6]
Homotopy Path Algebras, 2022; arXiv:2205.03730
David Favero and Jesse Huang. Homotopy Path Algebras, 2022; arXiv:2205.03730
arXiv 2022
-
[7]
Rouquier dimension is Krull dimension for normal toric varieties
David Favero and Jesse Huang. Rouquier dimension is Krull dimension for normal toric varieties, 2023, Eur. J. Math. 9, 91 (2023); arXiv:2302.09158. DOI: 10.1007/s40879-023-00686-1
work page Pith review arXiv 2023
-
[8]
W. Fulton, Introduction to toric varieties, Annals of Mathematics Studies 131, Princeton University Press, 1993
work page 1993
Show all 17 references
-
[9]
Zelevinsky Discriminants, Resultants, and Multidimensional Determinants; 1994
Izrail Moiseevich Gelfand and Mikhail Kapranov and Andrey V. Zelevinsky Discriminants, Resultants, and Multidimensional Determinants; 1994
1994
-
[10]
Green and Øyvind Solberg
Edward L. Green and Øyvind Solberg. An algorithmic approach to resolutions, 2005, Journal of Symbolic Computation, vol. 42, 11-12 (2007) 1012-1033, Non-commutative Grobner bases and applications; arXiv:math/0509020. DOI: 10.1016/j.jsc.2007.05.002
2007 arXiv
-
[11]
Resolutions of toric subvarieties by line bundles and applications, 2023; arXiv:2303.03763
Andrew Hanlon, Jeff Hicks and Oleg Lazarev. Resolutions of toric subvarieties by line bundles and applications, 2023; arXiv:2303.03763
2023 arXiv
-
[12]
Variation of GIT and Variation of Lagrangian Skeletons II: Quasi-Symmetric Case, 2020, Adv
Jesse Huang and Peng Zhou. Variation of GIT and Variation of Lagrangian Skeletons II: Quasi-Symmetric Case, 2020, Adv. Math. (2022); arXiv:2011.06114
2022 arXiv
-
[13]
Wrapped sheaves, 2021, Advances in Mathematics 415 (2023): 108882; arXiv:2102.06791
Christopher Kuo. Wrapped sheaves, 2021, Advances in Mathematics 415 (2023): 108882; arXiv:2102.06791. DOI: 10.1016/j.aim.2023.108882
2023 arXiv
-
[14]
The nonequivariant coherent-constructible correspondence for toric stacks, 2016, Duke Math
Tatsuki Kuwagaki. The nonequivariant coherent-constructible correspondence for toric stacks, 2016, Duke Math. J. 169, no. 11 (2020), 2125-2197; arXiv:1610.03214. DOI: 10.1215/00127094-2020-0011
2020 arXiv
-
[15]
Frobenius direct images of line bundles on toric varieties Journal of Algebra 226.2 (2000), pp
Jesper Funch Thomsen. Frobenius direct images of line bundles on toric varieties Journal of Algebra 226.2 (2000), pp. 865–874
2000
-
[16]
David Treumann, Remarks on the nonequivariant coherent-constructible correspondence for toric varieties, arXiv preprint arXiv:1006.5756 (2010)
2010 arXiv
-
[17]
Twisted Polytope Sheaves and Coherent-Constructible Correspondence for Toric Varieties, 2017; arXiv:1701.00689
Peng Zhou. Twisted Polytope Sheaves and Coherent-Constructible Correspondence for Toric Varieties, 2017; arXiv:1701.00689
2017 arXiv
Reviewed August 12, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.