Tannakian Fusion Category Overview

• Tannakian fusion categories are symmetric fusion categories equivalent to Rep(G), encoding finite-group representation theory within the framework of fusion categories.
• They serve as essential tools for classifying and analyzing braided, modular, and Morita-equivalent fusion categories by uncovering underlying group symmetries.
• Practical applications include de-equivariantization and Drinfeld center analysis, which are pivotal in studying topological phases and quantum group theories.

A Tannakian fusion category is a symmetric fusion category equivalent, as a braided tensor category, to the category of finite-dimensional representations of a finite group, denoted $\Rep(G)$, over an algebraically closed field of characteristic zero. This structure encapsulates the group-theoretical content within the framework of fusion categories and underpins the detection of classical symmetries within braided or modular fusion categories. Tannakian subcategories serve as building blocks in the classification, extension, and structure theory of fusion categories and their associated invariants, such as the Drinfeld center, Morita equivalence class, and mantle.

1. Formal Definitions and Characterization

Let $k$ be an algebraically closed field of characteristic zero. A fusion category $\mathcal{C}$ over $k$ is a $k$-linear, semisimple, rigid tensor category with finitely many isomorphism classes of simple objects, finite-dimensional $\operatorname{Hom}$-spaces, and a simple unit object. The braided structure is furnished by a family of isomorphisms $$c_{X, Y} : X \otimes Y \xrightarrow{\cong} Y \otimes X$$ satisfying the hexagon axioms.

A symmetric fusion category $\mathcal{E} \subseteq \mathcal{C}$ is called Tannakian if it is equivalent, as a symmetric tensor category, to $\Rep(G)$ for some finite group $G$, with the standard symmetry $x \otimes y \mapsto y \otimes x$. Equivalently, $\mathcal{E}$ admits a symmetric fiber functor $F: \mathcal{E} \to \mathrm{Vec}_k$, that is, an exact, faithful, $k$-linear tensor functor (Dong et al., 2014, Nikshych, 2012).

In characteristic $p>0$, \cite{(Coulembier, 2018)} provides a categorical internal characterization: a symmetric fusion category $\mathcal{C}$ is Tannakian if and only if, for every object $X$, there exists $n$ such that $\Lambda^n X = 0$, where $\Lambda^n$ is the $n$th exterior power. This recovers Ostrik’s criterion for positive characteristic.

2. Tannakian Subcategories in Braided Fusion Categories

In a braided fusion category, a Tannakian subcategory detects finite group symmetry. By Deligne's theorem, all symmetric fusion subcategories of $\mathcal{C}$ are of the form $\Rep(G, t)$, where $t$ is a central involution and the Tannakian ones correspond to $t=1$.

A central result is that, under substantial integrality conditions, many braided fusion categories necessarily contain nontrivial Tannakian subcategories. For instance, if in an integral braided fusion category every noninvertible simple object has Frobenius–Perron dimension divisible by a fixed prime $p$, then the largest pointed subcategory contains a nontrivial Tannakian subcategory. An analogous statement holds for categories of type $(1, m; p, n; q, s)$ under additional constraints on $p$ and $q$ (Dong et al., 2014).

These inclusions lead to classification results: for example, if all simple objects have $p$-power dimensions, the category is solvable; if the pointed part is of order $p$, the category is group-theoretical.

3. Morita Equivalence and Drinfeld Centers

In the Morita-equivalence approach, Tannakian subcategories play a critical role in the structure of the Drinfeld center $\mathcal{Z}(\mathcal{A})$ of a fusion category $\mathcal{A}$. $\mathcal{A}$ is Morita-equivalent to a $G$-extension (for some finite group $G$) of a category $\mathcal{B}$ if and only if $\mathcal{Z}(\mathcal{A})$ contains a Tannakian subcategory $\mathcal{E} \simeq \Rep(G)$. De-equivariantization and equivariantization methods recover the “base” category and establish equivalences: $$\mathcal{Z}(\mathcal{A})_G \simeq \mathcal{Z}(\mathcal{B}), \quad \mathcal{A} \sim_{\text{Morita}} \mathcal{B} \rtimes G$$ (Nikshych, 2012). This framework provides decisive classification results for fusion categories of small or prime-power dimension, ensuring that presence of a central Tannakian subcategory often forces strong algebraic structure.

4. Tannakian Radical, Mantle, and Reductive Categories

The Tannakian radical, denoted $\mathrm{Rad}_{\text{Tan}}(\mathcal{C})$, is defined as the intersection of all maximal Tannakian subcategories of a braided fusion category $\mathcal{C}$. It always has the form $\mathrm{Rep}(G_{\mathcal{C}})$ for a canonical group $G_{\mathcal{C}}$ (the “radical group”). The associated mantle, $\mathrm{Mantle}(\mathcal{C})$, is obtained by de-equivariantization with respect to $\mathrm{Rep}(G_{\mathcal{C}})$ and is a reductive braided fusion category—i.e., one with no nontrivial Tannakian subcategories (Green et al., 2024).

Structural classification then proceeds via a triple: $$\left( G_{\mathcal{C}},\, \mathrm{Mantle}(\mathcal{C}),\, F_{\mathcal{C}} \right)$$ where $F_{\mathcal{C}}$ is a canonical braided 2-functor encoding the central extension data. Classification of reductive categories demonstrates that, below dimension $p^5$, all reductive fusion categories are pointed; at $p^6$, non-pointed integral examples correspond to Drinfeld doubles of elementary Abelian groups.

5. Minimal Non-Degenerate Extensions and Cohomological Invariants

For a Tannakian fusion category $\mathcal{E} = \Rep(G)$, the group of minimal non-degenerate extensions, $\mathrm{Mext}(\mathcal{E})$, is isomorphic to $H^3(G; U(1))$, classifying the twisted Drinfeld doubles $Z(\operatorname{Vec}_G^\omega)$ (Nikshych, 2022). This connects higher group cohomology to categorical extensions.

A precise Künneth-type formula for tensor products of Tannakian categories holds: $\mathrm{Mext}(\Rep(G) \boxtimes \mathcal{E}) \cong \mathrm{Mext}(\mathcal{E}) \times 2\text{-Fun}(G, \operatorname{Pic}(\mathcal{E}))$ where $\operatorname{Pic}(\mathcal{E})$ is the 2-categorical Picard group of $\mathcal{E}$.

For pointed or super-Tannakian categories, there is a canonical filtration

$$M_0 \subset M_1 \subset M_2 \subset M_3 = \mathrm{Mext}(\mathcal{E})$$

each level corresponding to concrete cohomological or quadratic obstruction data (Nikshych, 2022). This structure directly describes the classification of topological phases with symmetry and the structure of braided extensions in terms of classical invariants.

6. Applications and Examples

Tannakian fusion categories, and their subcategories in larger braided settings, yield powerful classification results:

• Any integral, braided near-group category arises as an equivariantization by an abelian group (Dong et al., 2014).
• Group-theoretical fusion categories can be characterized by the presence of a Tannakian subcategory of the same Frobenius–Perron dimension in the center (Nikshych, 2012).
• The Drinfeld center of $\Rep(G)$ is itself the module category over the Drinfeld double of $G$, and contains $\Rep(G)$ as the “trivial” conjugacy class.
• In pointed categories, Tannakian subcategories correspond to self-dual subgroups with trivial quadratic form.

These results provide a framework in which Tannakian substructures are not only detected but algorithmically manipulated within the theory of modular tensor categories, topological quantum computation, and invariants of 3-manifolds.

7. Broader Context and Open Questions

Tannakian subcategories provide a mechanism for identifying and extracting group-theoretical "layers" from braided fusion categories. This isolation of group symmetry is central both to the Morita equivalence program and to the study of invariants such as the Witt group.

Open problems include:

• Extending existence and classification results to weakly integral (nonintegral) and metabolic reductive categories (Dong et al., 2014, Green et al., 2024).
• Characterizing minimal groups and the number of successive Tannakian layers needed in various categories.
• Cohomological description and computation of the canonical braided 2-functor $F_{\mathcal{C}}$ associated to a given mantle (Green et al., 2024).
• Application to modular data classification up to moderate dimension, as well as in settings arising from quantum groups at roots of unity and orbifold models in conformal field theory—where Tannakian subcategories correspond to hidden or enhanced gauge symmetries.

The theoretical developments surrounding Tannakian fusion categories have clarified the interplay between genuinely quantum phenomena and classical group symmetry, demonstrating that most naturally arising braided fusion categories are inherently not purely exotic but contain, at their core, an honest finite-group rep-theoretic structure (Dong et al., 2014, Green et al., 2024, Nikshych, 2012, Nikshych, 2022, Coulembier, 2018).