Existence of ACM Bundles on Polarized Abelian Varieties

arXiv:2606.04741 · math.AG · Submitted 2026-06-03 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Mathematics. The day in mathematics: new theorems, new proofs, and what they open.

Ada: We're looking at the paper: "Existence of ACM Bundles on Polarized Abelian Varieties".

Paper summary: Ada: So we've covered that the paper, "Existence of ACM Bundles on Polarized Abelian Varieties," tackles what it means for arithmetically Cohen-Macaulay bundles when you have a polarized abelian variety with dimension two or higher. We know the authors aim to settle this existence problem and they achieve that by making a few key claims about line bundles and vector bundles.

Emmy: They start with an acyclicity statement concerning non-trivial, algebraically trivial line bundles, proving that these are ACM for any ample polarization. Then they move on to explicitly building indecomposable ACM vector bundles in every rank. Finally, the big structural result is showing that the category of ACM bundles becomes wild representation type when the dimension is two or higher.

Kolya: So what’s the practical takeaway for someone who just wants to know what this means for their field? It sounds like they are proving existence and then showing that there's no simple structure to be found in these categories.

Ada: Exactly, Kolya. The fact that they find a family of bundles in every rank suggests that ACM bundles are abundant, not scarce. And the wild representation type part means you can't just classify them nicely with some simple algebraic tools; the theory is inherently more complicated than it first appeared.

Emmy: It really shows how delicate these objects are on abelian varieties. The structure sheaf itself isn't ACM, which makes this testing ground very interesting for the underlying theory of vector bundles on these spaces.

Conclusion: Ada: So to wrap up, the paper "Existence of ACM Bundles on Polarized Abelian Varieties" by Sondal and Barik has shown that there are abundant ACM bundles on polarized abelian varieties in dimensions two or higher. They proved that non-trivial algebraically trivial line bundles are always ACM, they constructed infinite families of indecomposable bundles in every rank, and they established the category is of wild representation type.

Emmy: What this means simply is that even though the structure sheaf fails to be ACM on these varieties, you still have a rich world of these well-behaved vector bundles that exist everywhere. The wildness part just tells us the classification problem is going to be quite hard.

Kolya: I think what stands out for me is how they tie this into unipotent bundles and power series modules. That connection, where the category of unipotent bundles on an abelian variety relates to modules over a specific regular local ring, that’s a deep structural insight.

Ada: It is deep because it provides the machinery to classify these ACM objects completely, even though they are wild. The authors also point out that M-regularity isn't enough on its own; you need the extra vanishing conditions for all twists, including the untwisted case.

Emmy: So in short, they’ve provided a solid foundation for finding these bundles and showing how complex the underlying theory is when you move past dimension one. It shows that abelian varieties are fertile ground for these kinds of objects despite their inherent structural quirks.

SOHAM MONDAL, PABITRA BARIK

math.AG

Submitted: 2026-06-03

Updated: 2026-11-15

Comments: 34 Pages, New results Included. All comments are welcome

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Importance score: 90/100

The gist: Existence of ACM bundles on polarized abelian varieties settles the existence problem for arithmetically Cohen-Macaulay (ACM) bundles on polarized abelian varieties of dimension 2 or higher by…

Key concepts

Arithmetically Cohen-Macaulay (ACM) Bundle
An ACM bundle is a vector bundle on an abelian variety that satisfies specific vanishing conditions related to its cohomology. The paper shows that many important bundles, including non-trivial line bundles and constructed indecomposable bundles, possess this special property with respect to a given polarization.
Wild Representation Type
This refers to the complexity of the category of ACM bundles. Wild representation type means the category is so complicated it contains structures similar to modules over a formal power series ring, indicating an extremely rich and difficult classification problem.
Fourier–Mukai Transform
This is a powerful mathematical tool used to relate different categories of sheaves on an abelian variety. It allows mathematicians to transform problems about vector bundles into equivalent problems about modules over specific rings, which is key to understanding the ACM property.
M-Regularity
M-regularity is a condition related to the regularity of a bundle and its dual. The paper uses this framework, along with vanishing conditions, to characterize when an ACM bundle exists, though it shows M-regularity alone is not enough.

Terminology

Summary

Existence of ACM bundles on polarized abelian varieties settles the existence problem for arithmetically Cohen-Macaulay (ACM) bundles on polarized abelian varieties of dimension 2 or higher by proving that every non-trivial, algebraically trivial line bundle is ACM, constructing explicit infinite families of indecomposable ACM vector bundles in every higher rank, and establishing that for dimension two or greater, the category of ACM bundles is of wild representation type

Acyclicity of Trivial Line Bundles

The starting point involves an acyclicity statement with strong consequences, proving that every nontrivial element of Pic0(A) is ACM with respect to every ample polarization This is achieved by using the Fourier–Mukai transform to prove that a nontrivial element P in Pic0(A) is acyclic, meaning Hi(A, P) = 0 for all 0 ≤ i ≤ dim A The full cohomology table of P ⊗ L ⊗ m is determined by the vanishing theorems and Riemann–Roch theorem Consequently, OA is shown to be the unique non-ACM element of Pic0(A) when g ≥ 2, and Pic0(A) excluding OA gives a natural g-dimensional family of rank-one ACM line bundles

Construction of Indecomposable Bundles

The paper constructs indecomposable ACM vector bundles Er of every rank r ≥ 1 by iterated non-split extensions starting from a nontrivial line bundle P The main difficulty in this construction is proving that suitable nontrivial extension classes exist at every step and that the resulting bundles remain indecomposable This process leads to the conclusion that for abelian varieties of dimension at least two, the category of ACM bundles is of wild representation type

Representation Type and Wildness

The category of unipotent bundles U on A is shown to be equivalent to flmod(k[[x1,..., xg]]) when g ≥ 2 Since the canonical surjection k[[x1,. p. 8.., xg]] ↠ k[[x1, x2]] induces a full exact restriction of scalars embedding flmod(k[[x1, x2]]) into flmod(k[[x1,. p. 1.., xg]]), the category U is wild Furthermore, the functor Φ: U −→ P ⊗ U is a full exact embedding of U into the category ACM(A, L) of ACM vector bundles on (A, L), implying that ACM(A, L) is of wild representation type

Characterization via M-Regularity

The Pareschi–Popa framework connects M-regularity to the ACM property, showing that if F and F∨⊗L are M-regular and satisfy certain intermediate cohomology vanishing conditions, then F is ACM with respect to L The paper demonstrates that being an ACM bundle does not automatically guarantee M-regularity because it fails the continuous global generation (CGG) requirement under severe negative twists

Classification of Line Bundles

For abelian varieties with Picard number ρ(A) = 1, every line bundle M can be uniquely written as M ∼= H⊗m ⊗ α, and M is ACM with respect to L if and only if one of three conditions holds The failure condition is characterized by the negation of these three good cases, specifically when Nn = OA or N2n < 0 or N2n = 0, N∈/Pic0(A) and PF ∼= OF This classification applies to every twist Nn = M ⊗ L ⊗ n

Existence of Higher Rank Bundles

The construction of indecomposable bundles Er(P) for a fixed rank r shows that as P varies in Pic0(A) excluding OA, the construction yields a g-dimensional family of pairwise non-isomorphic indecomposable ACM vector bundles This relies on the Jordan–H¨older filtration of Er(P) being determined by P, leading to non-isomorphism for distinct P and Q

Ulrich Bundle Behavior

No nontrivial line bundle P in Pic0(A) is Ulrich with respect to L Specifically, for g ≥ 2, the bundle Ma = P ⊗ L ⊗ a is never Ulrich If g = 1, Ma is Ulrich if and only if a = 1

Cohomological Structure

The complete cohomology table of P ⊗ L ⊗ m is given by Theorem 4.1 The only possible nonzero cohomology groups of P ⊗ L ⊗ m are H0(A, P ⊗ L ⊗ m) for m > 0 and Hg(A, P ⊗ L ⊗ m) for m < 0 For the untwisted case (m = 0), all cohomology groups vanish if g ≥ 2

M-Regularity vs. ACM

The bundle F is ACM with respect to L if and only if both F and its twisted dual bundle F∨⊗L are M-regular, provided the intermediate cohomology groups vanish for all twists However, M-regularity alone is not sufficient to guarantee ACM because an ACM bundle must satisfy vanishing for all twists, including the untwisted case where OA cannot lie in the cohomological support locus V i(F) This necessitates the untwisted assumption for Theorem 8.4

Fourier-Mukai Equivalence

The category of unipotent vector bundles U on A is equivalent to flmod(k[[x1,..., xg]]) via a chain of equivalences involving the Fourier–Mukai transform and the Cohen Structure Theorem This equivalence establishes the categorical link between unipotent bundles on an abelian variety A and the module category of a specific regular local ring The stalk functor F 7→ F0 yields an exact equivalence Coh(0)(Ab) ∼−−→ flmod OA, b 0

Conclusion on ACM Property

The paper concludes that the existence of abundant ACM bundles on polarized abelian varieties is supported by explicit constructions and the classification of line bundles, placing these varieties firmly on the wild side of the representation spectrum The results provide explicit families of indecomposable ACM bundles in every rank, clarify the behavior of algebraically trivial line bundles with respect to a polarization, and place abelian varieties firmly on the wild side of the ACM representation-type spectrum The results show that polarized abelian varieties carry abundant ACM bundles despite the fact that the structure sheaf itself fails to be ACM in dimension at least two

Appendix Details

The appendix establishes the chain of equivalences Unip(A) ≃ Coh(0)(Ab) ≃ flmod OA, b 0 ≃ flmod(k[[x1,..., xg]]) via the Fourier–Mukai transform and the Cohen Structure Theorem The equivalence between the category of unipotent vector bundles U on A and the category of coherent sheaves supported scheme-theoretically at the origin of Ab, denoted Coh(0)(Ab), is a classical consequence of the Fourier-Mukai equivalence The stalk functor F 7→ F0 yields an exact equivalence Coh(0)(Ab) ∼−−→ flmod OA, b 0 Since the dual abelian variety Ab is smooth of dimension g over an algebraically closed field, the local ring OA, b 0 is a regular local ring of dimension g By the Cohen Structure Theorem, its completion at the maximal ideal m0 is isomorphic to the formal power series ring This yields flmod OA, b 0 ≃ flmod(k[[x1,. p. 1.., xg]])

Representation Type of ACM Bundles

The category ACM(A, L) is of wild representation type because it contains the wild category U, and the functor Φ embeds U fully faithfully and exactly into ACM(A, L) via Lemma 7.2 This establishes that polarized abelian varieties support not only explicit ACM bundles in every rank but also a maximally complicated ACM theory from the point of view of classification The results show that polarized abelian varieties support not only explicit ACM bundles in every rank, clarify the behavior of algebraically trivial line bundles with respect to a polarization, and place abelian varieties firmly on the wild side of the ACM representation-type spectrum

Pareschi–Popa Framework

The Pareschi and Popa framework provides a bridge between M-regularity and ACM bundles, allowing one to prove that if F is M-regular and F∨⊗L is M-regular, then F is ACM with respect to L The necessity of the untwisted assumption for the ACM property stems from the fact that M-regularity does not mathematically prevent the bad locus V i(F) from containing OA This shows that F must have vanishing intermediate cohomology for all twists, including the untwisted case where OA cannot lie in the cohomological support locus V i(F) The results show that polarized abelian varieties carry abundant ACM bundles despite the fact that the structure sheaf itself fails to be ACM in dimension at least two

Classifications of Line Bundles

The classification of line bundles M on A with H1(A, M) = 0 is given by three conditions: N2 > 0, N ∈ Pic0(A) excluding OA, or N2 = 0 and PF not isomorphic to OF Applying this criterion to every twist Nn = M ⊗ L ⊗ n yields the asserted characterization of those M which are arithmetically Cohen–Macaulay with respect to L The equivalent description of the non-ACM locus is just the negation of the above three good cases

The paper determines that no nontrivial line bundle P in Pic0(A) is Ulrich with respect to L This is because the Ulrich condition would require Hi(A, P ⊗ L−1) = 0 for every i, which contradicts Theorem 4.

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. Improved ability to classify ACM bundles on polarized abelian varieties of dimension two or higher by determining their representation type, which is shown to be wild for these varieties. The AI can now predict whether an ACM bundle category exhibits tame or wild behavior based on the variety's dimension and polarization structure.

  2. Enhanced capability in constructing explicit infinite families of indecomposable ACM vector bundles in every higher rank on polarized abelian varieties. The system can generate these bundles by iterating non-split extensions starting from a nontrivial algebraically trivial line bundle P.

  3. Refined understanding of the cohomology table for twisted algebraically trivial line bundles on polarized abelian varieties. The AI can now calculate Hi(A, P tens L tens m) for all integers m based on the established formula derived from Riemann-Roch and vanishing theorems.

  4. Precise determination of when a vector bundle is Ulrich with respect to a polarization L on an abelian variety A of dimension g greater than or equal to two. The system can definitively state that "if g >= 2, the bundle Ma is never Ulrich" unless specific conditions related to the twist a = j are met (which they are not).

  5. Advanced classification of line bundles M on abelian varieties based on Picard rank. The AI can uniquely decompose any line bundle M into the form M tens H tens alpha, where alpha is an element of Pic0(A), and precisely determine the conditions under which this decomposition leads to a non-ACM bundle with respect to L.

  6. Sophisticated analysis of the Pareschi-Popa framework for M-regularity versus ACM bundles. The AI can distinguish between being M-regular and being ACM by identifying that M-regularity alone is not enough to guarantee that a bundle is Arithmetically Cohen–Macaulay.

  7. Categorical understanding of unipotent bundles on abelian varieties. The system can recognize the chain of equivalences: Unip(A) is equivalent to flmod OA, b 0, which in turn is equivalent to flmod k[[x1,..., xg]].

  8. Application of representation theory to determine the wild representation type of ACM bundles. For dimension g >= 2, the AI can conclude that the category ACM(A, L) is of wild representation type by embedding the category of unipotent bundles into it.

Related papers