Tannakian Categories attached to abelian Varieties
Rainer Weissauer

TL;DR
This paper constructs a semisimple super-Tannakian category from specific perverse sheaves on abelian varieties, including intersection cohomology sheaves of curves and smooth ample divisors.
Contribution
It introduces a new categorical framework linking perverse sheaves on abelian varieties to Tannakian categories, expanding the understanding of their geometric and algebraic structures.
Findings
Construction of a semisimple super-Tannakian category from perverse sheaves
Inclusion of intersection cohomology sheaves of curves and ample divisors
Establishment of categorical properties related to abelian varieties
Abstract
Starting from certain perverse sheaves on an abelian variety, including the intersection cohomology sheaves of curves and smooth ample divisors, we construct a semisimple super-Tannakian category.
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsAlgebraic Geometry and Number Theory · Advanced Algebra and Geometry · Algebraic structures and combinatorial models
\rotdriver
dvips
Tannakian Categories attached to abelian Varieties
Rainer Weissauer
Let be an algebraically closed field , where is either the algebraic closure of a finite field or a field of characteristic zero. Let be a prime different from the characteristic of .
Notations. For a variety over let denote the triangulated category of complexes of etale -sheaves on in the sense of [5]. For a complex let denote its Verdier dual, and denote its etale cohomology -sheaves with respect to the standard -structure. The abelian subcategory of middle perverse sheaves is the full subcategory of all , for which and its Verdier dual are contained in the full subcategory of semi-perverse sheaves, where is semi-perverse if and only if holds for all integers , where denotes the support of the cohomology sheaf of .
If is the algebraic closure of a finite field , then a complex of etale -Weil sheaves is mixed of weight , if all its cohomology sheaves are mixed etale -sheaves with upper weights for all integers . It is called pure of weight , if and its Verdier dual are mixed of weight . Concerning base fields of characteristic zero, we assume mixed sheaves to be sheaves of geometric origin in the sense of the last chapter of [1], so we still dispose over the notion of the weight filtration and purity and Gabber’s decomposition theorem in this case. In this sense let denote the abelian category of mixed perverse sheaves on . The full subcategory of of pure perverse sheaves is a semisimple abelian category.
Abelian varieties. Let be an abelian variety of dimension over an algebraically closed field . The addition law of the abelian variety defines the convolution product of two complexes and in by the direct image
[TABLE]
For the skyscraper sheaf concentrated at the zero element [math] notice .
Translation-invariant sheaf complexes. More generally , where is a closed -valued point in , the skyscraper sheaf with support in and where denotes the translation by . In fact holds for all . For let be the abstract group of all closed -valued points of , for which holds. A complex is called translation-invariant, provided . If is a surjective homomorphism between abelian varieties, then the direct image of a translation-invariant complex is translation-invariant. As a consequence of the formulas above, the convolution of an arbitrary with a translation-invariant complex on is a translation-invariant complex. A translation-invariant perverse sheaf on is of the form , for an ordinary etale translation-invariant -sheaf . For a translation-invariant complex the irreducible constituents of the perverse cohomology sheaves are translation-invariant.
Multipliers. The subcategory of of all perverse sheaves, whose irreducible perverse constituents are translation-invariant, is a Serre subcategory of the abelian category . Let denote its abelian quotient category and the image of , which is a full subcategory of semisimple objects. The full subcategory of of all , for which , is a thick subcategory of the triangulated category . Let
[TABLE]
be the corresponding triangulated quotient category, which contains . Then the convolution product
[TABLE]
still is well defined, by reasons indicated above.
Definition. A perverse sheaf on is called a multiplier, if the convolution induced by
[TABLE]
preserves the abelian subcategory .
Obvious from this definition are the following properties of multipliers: If and are multipliers, so are the product and the direct sum . Direct summands of multipliers are multipliers. If is a multiplier, then the Verdier dual is a multiplier and also the dual
[TABLE]
Examples: 1) Skyscraper sheaves are multipliers 2) If is a projective curve, which generates the abelian variety , and is an etale -sheaf on with finite monodromy, then the intersection cohomology sheaf attached to is a multiplier. 3) If is a smooth ample divisor, then the intersection cohomology sheaf of is a multiplier.
The proofs. 1) is obvious. For 2) we gave in [7] a proof by reduction mod using the Cebotarev density theorem and counting of points. Concerning 3) the morphism is affine for ample divisors . Hence and are perverse sheaves, which coincide in . The morphism is affine. Indeed is affine for affine subsets of , being isomorphic under the isomorphism of to the affine product . By the affine vanishing theorem of Artin: For perverse sheaves we get and for all . The distinguished triangle \bigl{(}Ra_{*}(\lambda_{Y}\boxtimes L),R\pi_{!}(\lambda_{U}\boxtimes L),Ra_{*}(\delta_{X}\boxtimes L)\bigr{)} for and the corresponding long exact perverse cohomology sequence gives isomorphisms for the integers . Since is a direct sum of translates of constant perverse sheaves , we conclude for to be zero in . For smooth the intersection cohomology sheaf is , and it is self dual. Hence by Verdier duality has image in . Thus is a multiplier.
Let denote the full category of semisimple multipliers. Let denote its image in the quotient category of . Then, by the definition of multipliers, the convolution product preserves
[TABLE]
Theorem. With respect to this convolution product the category is a semisimple super-Tannakian -linear tensor category, hence as a tensor category is equivalent to the category of representations of a projective limit
[TABLE]
of supergroups.
Outline of proof. The convolution product obviously satisfies the usual commutativity and associativity constraints compatible with unit objects. See [7] 2.1. By [7], corollary 3 furthermore one has functorial isomorphisms
[TABLE]
where denotes the degree zero cohomology sheaf and sections with support in the neutral element. Let be simple and nonzero. Then the left side becomes . On the other hand is a direct sum of a perverse sheaf and translates of translation-invariant perverse sheaves. Hence is the direct sum of a skyscraper sheaf and translation-invariant etale sheaves. Therefore . By a comparison of both sides therefore . Notice is the unit element of the convolution product. Using the formula above we not only get
[TABLE]
but also find a nontrivial morphism
[TABLE]
By semisimplicity is a direct summand of the complex . In particular the Künneth formula implies, that the etale cohomology groups do not all vanish identically
[TABLE]
Therefore the arguments of [7] 2.6 show, that the simple perverse sheaf is dualizable. Hence is a rigid -linear tensor category. Let be a finitely -generated tensor subcategory with generator say . To show is super-Tannakian, by [4] it is enough to show for all
[TABLE]
where is a suitable constant. For any let , by abuse of notation, also denote the perverse semisimple representative in without translation invariant summand. Put . Then , since every summand of is a multiplier and therefore has nonvanishing cohomology. For the Künneth formula gives . Therefore the estimate above holds for . This completes the outline for the proof of the theorem.
Principally polarized abelian varieties. Suppose is a divisor in defining a principal polarization. Suppose the intersection cohomology sheaf of is a multiplier. Then a suitable translate of is symmetric, and again a multiplier. So we may assume is symmetric. Let denote the super-Tannakian subcategory of generated by . The corresponding super-group attached to acts on the super-space defined by the underlying super-fiber functor of . By assumption is self dual in the sense, that there exists an isomorphism . Obviously . This defines a nondegenerate pairing on , and the action of on respects this pairing.
Curves. If is the Jacobian of smooth projective curve of genus over , carries a natural principal polarization . If we replace this divisor by a symmetric translate, then is a multiplier. The corresponding group is the semisimple algebraic group or depending on whether the curve is hyperelliptic or not. The representation of defined by as above is the unique irreducible -representation of of highest weight, which occurs in the -th exterior power of the -dimensional standard representation of . See [7], section 7.6.
Conjecture. One could expect, that a principal polarized abelian variety of dimension is isomorphic to a Jacobian variety of a smooth projective curve (up to translates of the divisor in as explained above) if and only if is a multiplier with corresponding super-Tannakian group equal to one of the two groups
[TABLE]
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1[1] Beilinson A., Bernstein J., Deligne P., Faisceaux pervers, Asterisque 100 (1982)
- 2[2] Deligne P., Milne J.S., Tannakian categories, in Lecture Notes in Math 900, p.101 –228
- 3[3] Deligne P., Categories tannakiennes, The Grothendieck Festschrift, vol II, Progr. Math, vol. 87, Birkhäuser (1990), 111 – 195
- 4[4] Deligne P., Categories tensorielles, Moscow Math. Journal 2 (2002) no.2, 227 – 248
- 5[5] Kiehl R., Weissauer R., Weil conjectures, perverse sheaves and l-adic Fourier transform, Ergebnisse der Mathematik und ihrer Grenzgebiete 42, Springer (2001)
- 6[6] Weissauer R., Torelli’s theorem from the topological point of view, ar Xiv math.AG/0610460
- 7[7] Weissauer R., Brill-Noether Sheaves, ar Xiv math.AG/0610923
