A common home for finite symmetric tensor categories

← All digests

Family 208 · Algebra
Importance 4/5 · Provisional editorial rating
Sources consulted 2026-10-21

1. The question

Many mathematical objects can be added, tensored and dualized. Vector spaces are the basic example; group representations carry the same operations together with symmetry information. A tensor category records these operations abstractly, without first specifying what the objects are made of.

A natural question is whether such an abstract category can always be interpreted inside a familiar concrete setting while preserving its linear and tensor structure. In characteristic zero, supervector spaces provide a remarkably broad answer. Positive characteristic introduces new behavior that ordinary or supervector spaces cannot capture.

Family 208 contains a manuscript dated 24 September 2026 claiming that every finite symmetric tensor category in positive characteristic has a faithful exact interpretation in one finite higher Verlinde category. This is the finite case of the Benson–Etingof–Ostrik conjecture, including characteristic two. The finiteness qualification is a substantive part of the result.

2. Background and essential definitions

Categories that behave like representations

A category has objects and maps between them. In a $k$-linear category, each space of maps is a vector space over the field $k$, and composition is bilinear.

An abelian category supports kernels, quotients and exact sequences, so it provides a setting for linear algebra and extensions. A tensor product combines objects, with a unit object playing the role of the one-dimensional trivial representation. Rigidity means every object has a dual. Symmetry means there are compatible swap maps $X\otimes Y\to Y\otimes X$, whose successive swaps give the identity.

The prototype is the category of finite-dimensional representations of a group. Its tensor product uses the diagonal group action, its dual is the dual representation, and its maps are equivariant linear maps.

Here a finite symmetric tensor category has finite-dimensional map spaces, finite-length objects, finitely many simple isomorphism classes and enough projective objects. Tensoring is exact in each variable, and the endomorphism ring of the unit is $k$. Finite length means that each object has a finite composition series with simple factors. “Finite category” does not mean finitely many objects in total.

What a fiber functor preserves

A fiber functor in this paper is a $k$-linear, exact, faithful, strong symmetric monoidal functor. In more familiar terms, it preserves linear combinations of maps, short exact sequences, tensor products, the unit and the symmetry; it does not identify two distinct maps.

Faithful does not mean full. The target may have maps between the images that did not come from the source. Thus a fiber functor is an interpretation preserving the relevant structure, not necessarily an identification with a full subcategory.

The higher Verlinde tower

Over an algebraically closed field $k$ of characteristic $p>0$, the higher Verlinde categories form a tower

$$\operatorname{Ver}_p(k)\subset \operatorname{Ver}_{p^2}(k)\subset \operatorname{Ver}_{p^3}(k)\subset\cdots.$$

The first level comes from representations of the cyclic group of order $p$, after discarding morphisms invisible to all trace pairings. Its simple objects correspond to Jordan blocks of lengths $1,\ldots,p-1$. In particular, $\operatorname{Ver}_2$ is the category of vector spaces, while $\operatorname{Ver}_3$ is the category of supervector spaces with the usual sign rule.

A supervector space is a vector space split into even and odd parts; swapping two odd vectors introduces a minus sign. Higher levels of the tower need not be semisimple: objects may be genuine extensions rather than direct sums of simples. They arise from quotients and abelian envelopes of tilting representations of $\operatorname{SL}_2$. For the main theorem, their role is a fixed hierarchy of increasingly expressive symmetric tensor targets.

3. Development history and previous partial results

Tannakian theory reconstructs groups from categories of representations together with suitable fiber functors. Deligne’s characteristic-zero theorem extended the common target from vector spaces to supervector spaces for symmetric tensor categories of moderate growth. Moderate growth means that, for each object, the lengths of its tensor powers grow at most exponentially.

Positive characteristic changes the picture. Ostrik proved that symmetric fusion categories admit fiber functors to $\operatorname{Ver}_p$. Fusion categories are semisimple finite tensor categories, so this covers an important but restricted case.

Etingof and Ostrik developed the Frobenius functor for nonsemisimple categories. Coulembier independently constructed its underlying additive functor. Exactness is the decisive issue: the ordinary Frobenius functor need not preserve short exact sequences. Etingof–Ostrik characterized finite categories where it is exact by the existence of a fiber functor to the first Verlinde level. Coulembier, Etingof and Ostrik extended this characterization to moderate-growth categories.

The higher tower is needed to accommodate categories beyond that first-level criterion. Benson and Etingof constructed characteristic-two examples, followed by the all-prime tower of Benson, Etingof and Ostrik. Coulembier independently developed the relevant abelian-envelope constructions.

Coulembier and Flake then introduced higher Frobenius constructions using representations of elementary abelian $p$-groups. Their construction becomes symmetric monoidal under a specific restricted-tilting conjecture. Earlier work established the conjecture in low rank and characteristic, and for substantial module classes, including modules of Loewy length at most two and Carlson modules, together with categories generated from them by standard operations.

The new manuscript claims the full restricted-tilting statement for arbitrary finite-dimensional modules. It then uses that result, together with existing higher-Frobenius machinery, to select an exact functor and obtain the finite-category theorem.

4. What the new manuscripts claim

For every prime $p$, every algebraically closed field $k$ of characteristic $p$, and every finite symmetric tensor category $\mathcal C$ over $k$, the manuscript asserts the existence of an integer $n\geq1$ and a fiber functor

$$F:\mathcal C\longrightarrow\operatorname{Ver}_{p^n}(k).$$

The height $n$ may depend on $\mathcal C$. The theorem does not say that every category maps to the first level or to every prescribed level.

An independent representation-theoretic statement supplies the main new input. Embed

$$E_r=(\mathbb Z/p\mathbb Z)^r$$

into the upper unipotent subgroup of $\operatorname{SL}_2(k)$, and let $R$ denote restriction of representations. If $\operatorname{St}_{r-1}$ is the simple Steinberg module of highest weight $p^{r-1}-1$, then for every finite-dimensional $E_r$-module $D$, the manuscript claims

$$R(\operatorname{St}_{r-1})\otimes D\cong R(T)$$

for some tilting $\operatorname{SL}_2$-module $T$. Tilting means that the module has both a filtration by Weyl modules and one by their dual counterparts. The assertion applies to every embedding of the specified kind, not only a specially chosen module class.

The fiber theorem also yields consequences through established results. It classifies finite incompressible symmetric tensor categories using the known subcategories of the Verlinde tower. Incompressible means that every exact symmetric tensor functor out of the category is fully faithful.

In characteristic two, a previous uniqueness theorem combines with the new existence claim to give exactly one fiber functor to the union of the tower, up to monoidal natural isomorphism. It has a representative landing in a finite level. The manuscript does not assert this uniqueness for every odd characteristic.

Finally, generalized Tannaka reconstruction realizes $\mathcal C$ as a suitable representation category of an affine group scheme internal to the chosen Verlinde level, with the required canonical fundamental-group compatibility.

5. Why the advance matters

Importance: 4/5, provisional. This would establish a broad common-target theorem for finite symmetric tensor categories in every positive characteristic, extending far beyond the semisimple and ordinary Frobenius-exact cases.

A common target makes abstract categories more comparable. It provides a setting for reconstruction and separates the universal background structure from the additional internal group symmetry defining a particular category.

The restricted-tilting theorem would also be useful independently: it says that tensoring by one specified restricted Steinberg module brings every elementary-abelian-group representation into the restricted tilting world.

The rating reflects a major finite-case advance while retaining the distinction from the full moderate-growth conjecture. That broader conjecture permits categories without a finite list of simple objects or a projective generator, precisely the features used in the proposed proof.

6. The main idea

Make arbitrary modules accessible to tilting theory

An elementary abelian $p$-group can sit inside the additive upper-unipotent matrices

$$\begin{pmatrix}1&t\\0&1\end{pmatrix}.$$

The paper first lifts an individual module for this finite subgroup to a representation of the additive algebraic group, with a bound on its polynomial degree. It does not require this preliminary lifting operation to preserve tensor products.

The lifted representation defines an equivariant vector bundle on the punctured plane. The bundle is then extended across the origin while controlling the highest weights in its fiber there. Tensoring by the specified Steinberg module makes that origin fiber tilting through an established bounded-weight result.

A formal comparison of fibers links the origin construction back to the original module away from the origin. This is the geometric step that the manuscript uses to handle arbitrary $D$, beyond the previously known special classes.

Choose a higher Frobenius functor that is exact

The restricted-tilting result activates the Coulembier–Flake construction. It produces an additive strong symmetric monoidal functor of the form

$$\Phi:\mathcal C\longrightarrow \mathcal C\boxtimes\operatorname{Ver}_{p^r}(k).$$

The symbol $\boxtimes$ is the Deligne product, the categorical counterpart of taking modules over a tensor product of finite-dimensional algebras. Initially, $\Phi$ is semilinear: multiplying a map by $a\in k$ multiplies its image by $a^{p^r}$.

Exactness depends on the embedding of $E_r$. Choose a projective generator $P$, an object from which every object is obtained as a quotient of a finite direct sum. Permuting the factors of $P^{\otimes p^r}$ gives an $E_r$-module on

$$\operatorname{Hom}_{\mathcal C}(P,P^{\otimes p^r}).$$

The manuscript finds the first rank where this module is nonprojective. Rank-support methods then select an embedding that makes $\Phi$ exact. The finiteness hypotheses are doing concrete work here by supplying $P$.

Remove the original category from the target

A functor landing in $\mathcal C\boxtimes\operatorname{Ver}_{p^r}$ has not yet solved the problem: the original unknown category remains in the target.

The paper compares Frobenius–Perron dimensions, positive real sizes determined by tensor multiplication. For a simple object $S$, any first-factor simple constituent $A$ arising from $\Phi(S)$ has dimension at most that of $S$. Equality forces the object into the subcategory where the ordinary Frobenius functor is exact.

Draw a finite directed graph with an arrow $S\to A$ for each such possibility. Every cycle lies in that controlled subcategory and cannot escape it. Because there are only finitely many simple objects, repeated application eventually places all first factors in its extension closure.

Earlier theorems give that controlled category a fiber functor to $\operatorname{Ver}_p$ in odd characteristic, or to a specified subcategory of $\operatorname{Ver}_4$ in characteristic two. Combining the finitely many target factors inside one tower level and correcting the scalar semilinearity produces the claimed $k$-linear fiber functor.

7. Scope, limitations, and open questions

The theorem assumes an algebraically closed field of positive characteristic and a finite symmetric tensor category with the stated exactness, duality and projectivity properties. It includes characteristic two, where the ordinary sign intuition alone is inadequate.

The full Benson–Etingof–Ostrik conjecture for arbitrary moderate-growth categories is not claimed. The proof’s finite directed graph and projective generator do not automatically extend to that setting.

Existence at some finite height is different from a uniform bound on height or an explicit minimum level for every category. The theorem supplies neither a universal prescribed height nor a general classification of all finite symmetric tensor categories by a short list of individual objects.

The internal reconstructed group scheme is asserted to be affine. The manuscript expressly makes no finiteness or finite-type claim for it. Reconstruction includes compatibility with the target category’s canonical fundamental-group action; it is not simply an ordinary group representation category over vector spaces.

8. Sources and further reading

Deligne, Catégories tensorielles, Moscow Mathematical Journal 2 (2002), 227–248, provides the characteristic-zero comparison. Ostrik, On symmetric fusion categories in positive characteristic, Selecta Mathematica 26 (2020), Article 36, gives the semisimple positive-characteristic theorem.

Benson, Etingof and Ostrik, New incompressible symmetric tensor categories in positive characteristic, Duke Mathematical Journal 172 (2023), 105–200, constructs the tower and formulates the conjecture. Coulembier, Etingof and Ostrik, Incompressible tensor categories, Advances in Mathematics 457 (2024), Article 109935, supplies the uniqueness and reconstruction framework.

Coulembier and Flake, Towards higher Frobenius functors for symmetric tensor categories, and A conjecture on the tensor ideal for an elementary p-group generated by the restriction of a Steinberg module, give the higher-functor construction and preceding restricted-tilting results.

Source manuscripts

• M1. OpenAI, Fiber functors for finite symmetric tensor categories in positive characteristic (September 24, 2026).

Sources consulted on 21 October 2026 at repository revision 1581d4c. Download the manuscript citations.