Ranks of matrix factorizations and sheaf cohomology

Mark E. Walker; Michael K. Brown

REVIEW 1 major objections 4 minor 19 references

Ranks of matrix factorizations and sheaf cohomology

T0 review · 1 major / 4 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read A graded rank conjecture for matrix factorizations and a new sheaf-cohomology conjecture are equivalent on Calabi-Yau hypersurfaces.

desk verdict Solid bridge theorem between graded BGS and a new sheaf cohomology conjecture; the main argument holds, but Example 4.7 has a genuine a=1/0 error that must be fixed. read the letter →

arxiv 2412.01060 v2 pith:6KDKVELI submitted 2024-12-02 math.AG math.AC

Michael K. Brown , Mark E. Walker This is my paper · ORCID

classification math.AGmath.AC MSC 13D0214F08

keywords matrixfactorizationsrankconjecturehypersurfacesingularitiessheafcohomologynon-Fanohypersurfacesderivedcategoriessingularityspectralsequences

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

The paper studies the graded version of a 1987 rank conjecture for matrix factorizations, which predicts that every nontrivial matrix factorization of a hypersurface singularity has rank at least $2^{\lfloor n/2\rfloor}$. It aims to show that, for smooth non-Fano projective hypersurfaces, this algebraic bound is really a geometric statement: the graded rank conjecture implies that every nonzero sheaf complex $C$ on $X$ satisfies $\rho(C)\ge 2^{\lfloor n/2\rfloor+1}$, where $\rho$ counts cohomology against twisted differential forms on the ambient projective space. For Calabi-Yau hypersurfaces ($a=0$) the two statements are proved equivalent. A reader should care because this converts an open local algebra conjecture into a concrete lower bound on the cohomology of every sheaf, and the paper exhibits an example where the bound is attained.

What carries the argument

The load-bearing mechanism is the Betti-number/cohomology dictionary of Theorem 3.3. For a reduced graded matrix factorization $F$ with $\Psi_a(C)\cong\operatorname{coker}(F)$, writing $a-j=qd-r$ with $0\le r<d$, it asserts $b^i_j=h^{r+a-2q-i+1}(\mathbb{P}^n, i_*C\otimes\Omega^{r+a}_{\mathbb{P}^n}(r+a))$, and hence $\operatorname{rank}(F)=\rho(C)$. The dictionary is built from an equivalence between the graded singularity category and $D^b(X)$, a duality isomorphism for the singularity category that turns the relevant Hom groups into Ext groups, and Proposition 3.1, which computes $\Phi_0(k(\ell))$ for $\ell=qd-r$ as $i_*(\wedge^{r+a}T_{\mathbb{P}^n})(-r-a)[2q+n-r-a-1]$. That residue-field computation is the step that uses $a\le 0$ through the inequality $\ell-a+d>0$.

What would settle it

Compute $\Phi_0(k(\ell))$ directly for a non-Fano hypersurface, say a quartic surface in $\mathbb{P}^3$ or a quintic threefold in $\mathbb{P}^4$, and compare the cohomological shift with $[2q+n-r-a-1]$; any discrepancy of one, or any computed value of $\rho(C)$ below $2^{\lfloor n/2\rfloor+1}$ for a nonzero $C$ on a Calabi-Yau hypersurface, would disprove the bridge.

Watch

Extended reading notes

Core claim

For a smooth hypersurface $X=V(f)\subseteq \mathbb{P}^n_k$ of degree $d$ with $a=n+1-d\le 0$, the central claim is: if every nontrivial graded matrix factorization of $f$ has rank at least $2^{\lfloor n/2\rfloor}$, then every nonzero object $C\in D^b(X)$ satisfies $\rho(C)\ge 2^{\lfloor n/2\rfloor+1}$; and when $a=0$ the converse also holds. The bridge is Theorem 3.3, which identifies the Betti numbers of a graded matrix factorization with cohomology dimensions of $C$ twisted by $\Omega^{r+a}_{\mathbb{P}^n}(r+a)$, so that twice the rank of the factorization equals $\rho(C)$. The paper does not settle either conjecture, but it reduces one to the other exactly, and it constructs a vector bundle on even-dimensional hypersurfaces given by a sum of $d$th powers that attains the bound.

Load-bearing premise

The whole argument rests on one exact shift in a long calculation, and on the still-open graded 1987 rank conjecture for the first direction.

Editorial extensions

If this is right

If the graded 1987 rank conjecture holds for a smooth non-Fano hypersurface, then every nonzero object in $D^b(X)$, including every vector bundle, satisfies $\rho(C)\ge 2^{\lfloor n/2\rfloor+1}$.
On Calabi-Yau hypersurfaces, Conjecture 1.3 and the graded rank conjecture are the same statement: proving either settles the other.
The bound is sharp: on even-dimensional hypersurfaces given by a sum of $d$th powers there is a reduced graded matrix factorization whose associated vector bundle has $\rho=2^{\lfloor n/2\rfloor+1}$.
The invariant $\rho(C)$ is the total rank of the $E_1$ page of the spectral sequence from Remark 3.4, so the conjecture says every nonzero sheaf complex on a non-Fano hypersurface needs at least $2^{\lfloor n/2\rfloor+1}$ total entries there.
The non-Fano condition $a\le 0$ is necessary: on linear hypersurfaces, line bundles achieve $\rho=2$, below the conjectured bound.

Reading between the lines

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

One could test Conjecture 1.3 numerically on well-known vector bundles over Calabi-Yau threefolds; an indecomposable bundle with $\rho<8$ would disprove it and, by Theorem 1.4, the graded rank conjecture in that case.
The dictionary may extend to Fano hypersurfaces if $\rho$ is replaced by a truncated invariant, since the paper traces the failure on Fano examples precisely to line bundles.
The pattern suggests a general translation principle: rank statements about graded matrix factorizations correspond to total-cohomology statements about sheaves whenever the images of residue fields in the derived category are computable.

Share X Bluesky LinkedIn Reddit HN

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 short version: this paper proves a clean bridge theorem between a graded form of the BGS rank conjecture and a new sheaf-cohomology rank conjecture, and the main argument looks sound. The one real defect I found is Example 4.7, which as printed does not achieve what it claims. That's fixable, but it needs fixing.

What's actually new: Theorem 3.3 gives an exact formula expressing the graded Betti numbers of a reduced matrix factorization of f in terms of cohomology h^*(P^n, i_*C ⊗ Ω^{r+a}(r+a)), with rank(F) = ρ(C). This generalizes Pavlov's Koszul computation beyond the Calabi-Yau case to all non-Fano hypersurfaces. The bridge theorem, Theorem 1.4, shows that the graded BGS bound is equivalent to their Conjecture 1.3 in the Calabi-Yau case, and implies it in general. That is a genuinely useful reframing: it turns a local algebra question into a geometric statement about all sheaves on a hypersurface, and the Beilinson spectral interpretation is a nice touch.

The proof machinery is standard but used competently: Orlov, Auslander duality, Shamash/Koszul resolution. I cannot see a load-bearing error in the main computation. The shift concern raised by the stress-test is a red herring—the displayed [n] in the proof of Proposition 3.1 is a typo for [−n], and the final shift in the statement is consistent.

Where it's soft: Example 4.7 is internally inconsistent. For f = x_0^n + ... + x_n^n, deg f = n, so a = n+1−n = 1, not 0. The sharpness claim relies on a = 0 for both Conjecture 1.3 and the inverse equivalence, so as written the example lies outside the hypotheses. The likely fix is to take n odd and deg d = n+1, giving a = 0, and build rank 2^{(n-1)/2} factorizations; that recovers the claimed bound. The authors cite [BD20] for tensor products, which is appropriate and not a substitute for fixing the example.

Verdict: the main theorem stands, but the example must be corrected before publication. A serious referee will catch this, and the paper deserves referee time. I'd bring it to our reading group—it's a good illustration of how to connect singularity categories with sheaf cohomology.

Referee Report

1 major / 4 minor

Summary. The paper studies a graded version of the Buchweitz-Greuel-Schreyer rank conjecture for matrix factorizations and relates it to a new conjecture (Conjecture 1.3) about the total cohomology invariant ρ(C) of sheaves on non-Fano projective hypersurfaces. The main result, Theorem 1.4, shows that the graded BGS conjecture implies Conjecture 1.3 for all non-Fano hypersurfaces, and conversely in the Calabi-Yau case a=0. The technical core is Theorem 3.3, which uses Orlov's equivalence and graded Auslander duality to express the Betti numbers of a graded matrix factorization in terms of cohomology of P^n with coefficients in Ω^{r+a}(r+a), yielding a rank–cohomology equality. The paper also computes ρ for points, structure sheaves, and Fano counterexamples, and Example 4.7 purports to show sharpness of Conjecture 1.3.

Significance. If the main arguments are correct, the paper provides a clean bridge between two open conjectures: it reduces the graded BGS conjecture in the Calabi-Yau case to a concrete statement about sheaf cohomology, and it gives a new prediction about the total rank of the Beilinson spectral sequence. The explicit formula in Theorem 3.3 is checkable and is the main technical contribution. The paper is also transparent about the conditional nature of its main theorem, which depends on the still-open graded BGS conjecture. The flaw in Example 4.7 identified below does not invalidate the proof of Theorem 1.4 or Theorem 3.3, but it does affect the advertised sharpness of Conjecture 1.3.

major comments (1)

[Section 4, Example 4.7] Example 4.7 is internally inconsistent as written. The polynomial f=x_0^n+...+x_n^n has degree n, so a=n+1-n=1, not 0; therefore the line "Since a=0" is false, and Theorem 2.3(3) cannot be invoked. Since Conjecture 1.3 is only formulated for a≤0, the example falls outside the hypotheses of the conjecture and does not demonstrate sharpness. In addition, the tensor-product construction as written is problematic: one copy of F, which factorizes x_0^n, together with e copies of F', each of which factorizes x_0^n+x_1^n, would produce a factorization of (e+1)x_0^n+e x_1^n, not of f, unless the copies of F' are understood to be relabeled so as to act on disjoint pairs of variables. The example should be repaired, for instance by taking n odd and f=x_0^{n+1}+...+x_n^{n+1}, and then taking the tensor product of (n+1)/2 pair factorizations; this gives a=0 and recovers the claimed rank bound. Because the Introduction states that the lower bound in Conjecture 1.3 is sharp by citing this example, the correction is necessary before publication.

minor comments (4)

[Theorem 3.3(2)] The statement "rank(F)=ρ(C)" is ambiguous. Under the convention in Section 2.1, rank(F) means the common rank rank(F^0)=rank(F^1), but the proof computes rank(F)=rank(F^0)+rank(F^1)=Σ_j(b^0_j+b^1_j), and Theorem 1.4 uses the equality in the form 2·rank(F^0)=ρ(C). Please define rank(F) in Theorem 3.3(2) explicitly as the sum of the two ranks, or state the equality as 2·rank(F^0)=ρ(C).
[Section 2.2] The definition of the truncation G_{≺i} appears garbled: "free summands of the form R(i) with i > 0" cannot be right as written, since the index i is already the truncation parameter. Please clarify whether G_{≺i} consists of summands R(j) with j < i or j > i.
[Proof of Theorem 3.3] In the chain of isomorphisms in the proof of Theorem 3.3, the reference "(Theorem 3.1)" should be "(Proposition 3.1)".
[Example 4.3] The citation "[OSS11] Chapter 1, Section 1.1]" has a missing opening bracket or comma; it should read "[OSS11, Chapter 1, Section 1.1]".

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: Theorem 1.4 is a conditional equivalence between independent conjectures, and the technical bridge rests on external results; the only self-citation is non-load-bearing.

full rationale

The central derivation is not circular. Theorem 1.4 does not assume Conjecture 1.3 to prove Conjecture 1.3; it assumes the graded BGS conjecture as an explicitly unproved hypothesis and proves the implication to Conjecture 1.3, while the converse direction is proved only in the Calabi-Yau case. The two conjectures concern different quantities, matrix factorization ranks versus the cohomological invariant rho, so the implication is substantive. The load-bearing bridge, Theorem 3.3, is a genuine computation rather than a renaming: the Betti numbers b^i_j are related to Ext groups by Serre duality from Proposition 2.7, the adjunction from Proposition 2.6, and the explicit identification of Phi_0 on residue fields in Proposition 3.1; the equality rank(F)=rho(C) then follows by summing those Betti numbers. Proposition 3.1 itself is derived from the Shamash resolution and standard exterior-power identifications, not from either conjecture under discussion. The external citations, including Orlov, Buchweitz, Eisenbud, Knörrer, Erman, Pavlov, KMVdB11, Burke-Stevenson, Tu, and OSS11, are independent of the present authors' claims. The only self-citation is [BD20] in Example 4.7, where Brown and Dyckerhoff's tensor product construction is used to build an illustrative matrix factorization; this does not support the central theorems. I also flag, for the record, a concrete correctness defect that is not circularity: Example 4.7 takes f=x_0^n+...+x_n^n with n even, which has degree d=n and therefore a=n+1-d=1, not 0, so the example's invocation of Theorem 2.3(3) and its statement "Since a=0" are outside the hypotheses of Conjecture 1.3. This is a peripheral hypothesis error and does not undermine the conditional theorems. Overall, no step in the claimed derivation chain reduces by construction to its own input, and no load-bearing argument depends on an unverified self-citation.

Assumptions & free parameters 0 free parameters · 5 assumptions · 0 invented entities

No free parameters or invented entities appear in the paper: the conjectures are open statements, and the only new object is the invariant rho(C) defined in (1.2), which is a definition. The main theorems rest on external standard results (Orlov's theorem, Eisenbud's theorem, Auslander duality, Shamash construction) and on the structural assumption that X is smooth and non-Fano. The conditional direction additionally assumes the open graded BGS conjecture.

assumptions (5)

standard math Orlov's theorem (Theorem 2.3): semiorthogonal decompositions relating Dsing_gr(R) and D^b(X), with functors Phi_i and Psi_i satisfying Phi_i left adjoint to Psi_{i+a}.
Invoked in Section 2.2 and used continuously in Theorems 3.3 and 1.4; quoted from [Orl09, Theorem 2.5].
standard math Graded Eisenbud theorem: equivalence between the homotopy category of graded matrix factorizations and the stable category of graded MCM modules.
Used in Section 2.1 to establish the equivalence in (2.1); cited from [Eis80, Theorem 6.3].
standard math Graded Auslander duality (Proposition 2.7): Hom(M,N)* is isomorphic to Hom(N,M(-b)[t-1]) for Gorenstein isolated singularity rings.
Used in the proof of Theorem 3.3 to convert Hom groups in the singularity category; cited from [KMVdB11, Section 4.3].
domain assumption Smoothness of X and a <= 0 are assumed throughout Theorem 3.3 and Conjecture 1.3.
The paper states the formula only for smooth non-Fano hypersurfaces and notes in Remark 4.6 that the Fano case fails; this is a domain restriction, not proven.
standard math Shamash construction gives the minimal free resolution of k over the hypersurface ring R.
Used in the proof of Proposition 3.1 to compute Phi_0(k(l)); cited from [EP16, Construction 4.1.3].

how reviews work

0 comments

Cite this review

Pith. "Pith review of Ranks of matrix factorizations and sheaf cohomology." pith.science (2026). https://pith.science/paper/6KDKVELI

@misc{pith2026241201060,
  author       = {Pith},
  title        = {Pith review of: Ranks of matrix factorizations and sheaf cohomology},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/6KDKVELI}},
  note         = {Machine review of arXiv:2412.01060}
}

read the original abstract

Buchweitz-Greuel-Schreyer conjectured in 1987 a lower bound on the ranks of matrix factorizations over certain local hypersurface rings. We study a graded version of this conjecture, and we show that it implies a novel conjecture concerning the cohomology of sheaves over non-Fano projective hypersurfaces.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

19 extracted references · 17 canonical work pages

[1]

C onf., T emple U niv., P hiladelphia, P a., 1976), Lect

Maurice Auslander, Functors and morphisms determined by objects, Representation theory of algebras ( P roc. C onf., T emple U niv., P hiladelphia, P a., 1976), Lect. Notes Pure Appl. Math., Vol. 37, Dekker, New York-Basel, 1978, pp. 1--244. 480688

work page 1976
[2]

Brown and Tobias Dyckerhoff, Topological K -theory of equivariant singularity categories , Homology Homotopy Appl

Michael K. Brown and Tobias Dyckerhoff, Topological K -theory of equivariant singularity categories , Homology Homotopy Appl. 22 (2020), no. 2, 1--29. 4073075

work page 2020
[3]

Be linson, Coherent sheaves on P ^n and problems of linear algebra , Functional Analysis and its Applications 12 (1978), no

Alexander A. Be linson, Coherent sheaves on P ^n and problems of linear algebra , Functional Analysis and its Applications 12 (1978), no. 3, 214--216

work page 1978
[4]

Buchweitz, G.-M

R.-O. Buchweitz, G.-M. Greuel, and F.-O. Schreyer, Cohen- M acaulay modules on hypersurface singularities. II , Invent. Math. 88 (1987), no. 1, 165--182. 877011

work page 1987
[5]

Jesse Burke and Greg Stevenson, The derived category of a graded G orenstein ring , Commutative algebra and noncommutative algebraic geometry. V ol. II , Math. Sci. Res. Inst. Publ., vol. 68, Cambridge Univ. Press, New York, 2015, pp. 93--123. 3496862

work page 2015
[6]

262, American Mathematical Society, Providence, RI, [2021] 2021, With appendices and an introduction by Luchezar L

Ragnar-Olaf Buchweitz, Maximal C ohen- M acaulay modules and T ate cohomology , Mathematical Surveys and Monographs, vol. 262, American Mathematical Society, Providence, RI, [2021] 2021, With appendices and an introduction by Luchezar L. Avramov, Benjamin Briggs, Srikanth B. Iyengar and Janina C. Letz. 4390795

work page 2021
[7]

David Eisenbud, Homological algebra on a complete intersection, with an application to group representations, Trans. Amer. Math. Soc. 260 (1980), no. 1, 35--64. 570778

work page 1980
[8]

150, Springer-Verlag, New York, 1995, With a view toward algebraic geometry

, Commutative algebra, Graduate Texts in Mathematics, vol. 150, Springer-Verlag, New York, 1995, With a view toward algebraic geometry. 1322960

work page 1995

Show all 19 references

[9]

2152, Springer, Cham, 2016

David Eisenbud and Irena Peeva, Minimal free resolutions over complete intersections, Lecture Notes in Mathematics, vol. 2152, Springer, Cham, 2016. 3445368

2016
[10]

Daniel Erman, Matrix factorizations of generic polynomials, arXiv:2112.08864 (2021)

2021 arXiv
[11]

Bernhard Keller, Daniel Murfet, and Michel Van den Bergh, On two examples by I yama and Y oshino , Compos. Math. 147 (2011), no. 2, 591--612. 2776613

2011
[12]

I , Invent

Horst Kn \"o rrer, Cohen- M acaulay modules on hypersurface singularities. I , Invent. Math. 88 (1987), no. 1, 153--164. 877010

1987
[13]

Daniel Murfet, Residues and duality for singularity categories of isolated G orenstein singularities , Compos. Math. 149 (2013), no. 12, 2071--2100. 3143706

2013
[14]

Dmitri Orlov, Derived categories of coherent sheaves and triangulated categories of singularities, Algebra, arithmetic, and geometry: in honor of Y u. I . M anin. V ol. II , Progr. Math., vol. 270, Birkh\" a user Boston, Boston, MA, 2009, pp. 503--531. 2641200

2009
[15]

a user Classics, Birkh\

Christian Okonek, Michael Schneider, and Heinz Spindler, Vector bundles on complex projective spaces, Modern Birkh\" a user Classics, Birkh\" a user/Springer Basel AG, Basel, 2011, Corrected reprint of the 1988 edition, With an appendix by S. I. Gelfand. 2815674

2011
[16]

Alexander Pavlov, Koszul duality between B etti and cohomology numbers in the C alabi- Y au case , Proc. Amer. Math. Soc. 148 (2020), no. 4, 1373--1381. 4069177

2020
[17]

, Betti tables of MCM modules over the cone of a plane cubic , Math. Z. 297 (2021), no. 1-2, 223--254. 4204691

2021
[18]

Roberts, Multiplicities and C hern classes in local algebra , Cambridge Tracts in Mathematics, vol

Paul C. Roberts, Multiplicities and C hern classes in local algebra , Cambridge Tracts in Mathematics, vol. 133, Cambridge University Press, Cambridge, 1998. 1686450

1998
[19]

Junwu Tu, Matrix factorizations via K oszul duality , Compos. Math. 150 (2014), no. 9, 1549--1578. 3260141

2014

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.

[1] [1]

C onf., T emple U niv., P hiladelphia, P a., 1976), Lect

Maurice Auslander, Functors and morphisms determined by objects, Representation theory of algebras ( P roc. C onf., T emple U niv., P hiladelphia, P a., 1976), Lect. Notes Pure Appl. Math., Vol. 37, Dekker, New York-Basel, 1978, pp. 1--244. 480688

work page 1976

[2] [2]

Brown and Tobias Dyckerhoff, Topological K -theory of equivariant singularity categories , Homology Homotopy Appl

Michael K. Brown and Tobias Dyckerhoff, Topological K -theory of equivariant singularity categories , Homology Homotopy Appl. 22 (2020), no. 2, 1--29. 4073075

work page 2020

[3] [3]

Be linson, Coherent sheaves on P ^n and problems of linear algebra , Functional Analysis and its Applications 12 (1978), no

Alexander A. Be linson, Coherent sheaves on P ^n and problems of linear algebra , Functional Analysis and its Applications 12 (1978), no. 3, 214--216

work page 1978

[4] [4]

Buchweitz, G.-M

R.-O. Buchweitz, G.-M. Greuel, and F.-O. Schreyer, Cohen- M acaulay modules on hypersurface singularities. II , Invent. Math. 88 (1987), no. 1, 165--182. 877011

work page 1987

[5] [5]

Jesse Burke and Greg Stevenson, The derived category of a graded G orenstein ring , Commutative algebra and noncommutative algebraic geometry. V ol. II , Math. Sci. Res. Inst. Publ., vol. 68, Cambridge Univ. Press, New York, 2015, pp. 93--123. 3496862

work page 2015

[6] [6]

262, American Mathematical Society, Providence, RI, [2021] 2021, With appendices and an introduction by Luchezar L

Ragnar-Olaf Buchweitz, Maximal C ohen- M acaulay modules and T ate cohomology , Mathematical Surveys and Monographs, vol. 262, American Mathematical Society, Providence, RI, [2021] 2021, With appendices and an introduction by Luchezar L. Avramov, Benjamin Briggs, Srikanth B. Iyengar and Janina C. Letz. 4390795

work page 2021

[7] [7]

David Eisenbud, Homological algebra on a complete intersection, with an application to group representations, Trans. Amer. Math. Soc. 260 (1980), no. 1, 35--64. 570778

work page 1980

[8] [8]

150, Springer-Verlag, New York, 1995, With a view toward algebraic geometry

, Commutative algebra, Graduate Texts in Mathematics, vol. 150, Springer-Verlag, New York, 1995, With a view toward algebraic geometry. 1322960

work page 1995