Classification of nondegenerate $G$-categories (with an appendix written jointly with Germán Stefanich)

This paper classifies nondegenerate categories with a reductive group action entirely in terms of the group's root datum and applies these methods to establish a monoidal equivalence for bi-Whittaker $\mathcal{D}$-modules and provide evidence for a conjecture regarding the parabolic restriction of very central sheaves.

The Gist

The Big Picture: Organizing a Chaotic Universe

Imagine you are an archivist trying to organize a massive, chaotic library. This library doesn't contain books; it contains universes of mathematical objects (called "categories"). These universes are being shaken and stirred by a giant, invisible force called a Group (specifically, a "reductive group" $G$).

In this paper, the author, Tom Gannon (with help from Germán Stefanich), tries to answer a huge question: "If we have a universe being shaken by this group, can we describe exactly what that universe looks like?"

Usually, this is incredibly hard. The shaking creates complex patterns that are impossible to map. However, Gannon discovers a special "safe zone" within these universes. He calls these Nondegenerate Categories.

Think of "Nondegenerate" as the "healthy" or "generic" parts of the library. In these parts, the chaos isn't broken or stuck; it's moving in a very specific, predictable way. Gannon's main achievement is proving that all these healthy universes can be perfectly described using a single, elegant blueprint based on the group's "DNA" (called the root datum).

---

Key Concepts & Analogies

1. The Group Action: The Great Shaker

Imagine a group of dancers (the Group $G$) moving around a stage.

• Vector Spaces (Old Math): Usually, we study how dancers move a single object (like a ball). We can easily see the ball spin or slide.
• Categories (New Math): Here, the dancers are moving an entire room full of objects. The room itself is the "category."
• The Problem: When the dancers move the room, the objects inside get mixed up. It's hard to tell what the room looks like after the dance.

2. The "Nondegenerate" Condition: The Filter

Gannon introduces a filter. He says, "Let's ignore the rooms where the dancers get stuck or where the movement is broken."

• Analogy: Imagine a sieve. If you shake a bucket of sand and rocks, the rocks (the "degenerate" parts) might get stuck in the holes. The sand that falls through smoothly is the "nondegenerate" part.
• The Result: Once you filter out the stuck parts, the remaining sand (the nondegenerate category) has a beautiful, uniform structure.

3. The Blueprint: The Root Datum

Every group has a "skeleton" or "DNA" called the Root Datum. It's like the architectural plan for the building.

• The Discovery: Gannon proves that for any healthy (nondegenerate) room being shaken by the dancers, you don't need to look at the room itself. You just need to look at the Architectural Plan (Root Datum).
• The Map: He builds a map (a mathematical object called $\Gamma_{\tilde{W}_{aff}}$) based only on that plan. He then shows that every healthy room is just a "module" (a specific arrangement of furniture) sitting on top of this map.
• In Simple Terms: "If you know the group's DNA, you can predict exactly what any healthy, shaken universe looks like."

---

The Two Big Applications (The "So What?")

The paper isn't just about abstract theory; it solves two specific, long-standing puzzles in the field.

Application 1: The Symmetric Monoidal Whittaker-Hecke Category

• The Puzzle: Mathematicians had a specific type of room called the "Whittaker-Hecke category." They knew how to move things around in it (multiplication), but they weren't sure if the rules were "symmetric" (i.e., does $A \times B$ look the same as $B \times A$?). A famous mathematician, Drinfeld, asked if it was symmetric.
• The Solution: Gannon proves that because this room fits into his "Nondegenerate" blueprint, it must be symmetric.
• Analogy: Imagine a dance floor where everyone was worried that if you swapped partners, the dance would break. Gannon proves that because the floor is built on this specific blueprint, swapping partners is perfectly safe and the dance remains beautiful. This upgrades the rules of the game from "okay" to "perfectly symmetric."

Application 2: The Parabolic Restriction of "Very Central" Sheaves

• The Puzzle: There is a process called "Parabolic Restriction." Imagine taking a complex sculpture (a "sheaf") and projecting its shadow onto a simpler wall.
• The Conjecture: Ben-Zvi and Gunningham guessed that if the sculpture is "Very Central" (a special, highly symmetric type of sculpture), its shadow would land on a specific, clean part of the wall (the "coarse quotient").
• The Solution: Gannon proves this is true. He shows that these special sculptures, when projected, not only land on the right spot but also gain a new "superpower": they become Weyl Group Equivariant.
• Analogy: Imagine you have a complex 3D sculpture. When you shine a light on it to cast a shadow, you expect a messy blob. But Gannon proves that if the sculpture is "Very Central," the shadow isn't just a blob; it's a perfectly organized 2D pattern that fits a specific grid. Furthermore, this pattern has a hidden symmetry (the Weyl group) that makes it even more structured.

---

The "Secret Sauce": How Did They Do It?

The paper uses a technique called Categorical Representation Theory.

• The Old Way: Trying to solve these problems by looking at the individual pieces (like counting atoms).
• The New Way: Looking at the relationships between the pieces.
• The Magic Trick: They use a tool called the Mellin Transform.
  • Analogy: Think of the Mellin Transform as a magical prism. You shine a complex, colorful light (a D-module) through it, and it comes out the other side as a simple, clear spectrum (a sheaf on a dual space).
  • Gannon and Stefanich upgraded this prism to work in the "higher" world of categories, proving it preserves the "symmetric" nature of the objects. This allowed them to translate the messy, shaken rooms into clean, organized blueprints.

Summary

This paper is a masterclass in finding order in chaos.

1. It identifies a "healthy" subset of mathematical universes shaken by groups.
2. It proves that these universes are completely determined by the group's DNA.
3. It uses this insight to solve two major puzzles: proving a dance floor is symmetric and showing that special shadows land perfectly on a grid.

It's like taking a tangled ball of yarn, realizing that if you pull on the right thread (the "nondegenerate" part), the whole ball unravels into a perfect, straight line that you can measure and understand completely.

Technical Summary

1. Problem Statement and Motivation

The paper addresses a fundamental problem in geometric representation theory: the classification of categories equipped with an action of a split reductive group $G$. While the study of groups acting on vector spaces is classical (via eigenspace decomposition), the categorical analogue (groups acting on DG categories) is significantly more complex due to the existence of non-trivial morphisms between invariants of different subgroups.

The authors focus on a specific, "dense open" subclass of these categories called nondegenerate G-categories.

• Definition: A G-category $\mathcal{C}$ is nondegenerate if, for every rank-one parabolic subgroup $P_\alpha$, the invariants $\mathcal{C}^{[P_\alpha, P_\alpha]}$ vanish.
• Motivation: Many central objects in representation theory, such as the category of Whittaker D-modules or the BGG category $\mathcal{O}$, exhibit nondegenerate behavior. The authors aim to provide a "coherent" description of these categories entirely in terms of the root datum of $G$, specifically relating them to sheaves on the dual Cartan subalgebra $\mathfrak{t}^*$.

A key motivation is to upgrade existing equivalences (specifically those by Ginzburg and Lonergan) from the level of abelian categories to monoidal equivalences of DG categories, thereby answering a question posed by Drinfeld regarding the symmetric monoidal structure of the Whittaker-Hecke category.

2. Methodology

The paper employs a sophisticated blend of categorical representation theory, ind-coherent sheaves, and descent theory.

• Categorical Representation Theory: The authors utilize the framework where a group $G$ acts on a category $\mathcal{C}$. They leverage the relationship between invariants ($\mathcal{C}^H$) and coinvariants, and the interplay between different subgroups (e.g., $N$, $B$, $T$).
• The Coarse Quotient: A central geometric object is the coarse quotient $\mathfrak{t}^* // \widetilde{W}_{aff}$, where $\widetilde{W}_{aff}$ is the extended affine Weyl group acting on the dual Cartan $\mathfrak{t}^*$. This is a prestack defined in the companion paper [Gan26].
• Ind-Coherent Sheaves ($\text{IndCoh}$): Instead of working with quasi-coherent sheaves, the authors use $\text{IndCoh}$, which is better suited for singular stacks and ind-schemes. They study the category $\text{IndCoh}(\Gamma_{\widetilde{W}_{aff}})$, where $\Gamma_{\widetilde{W}_{aff}}$ is an ind-scheme constructed from the graphs of the affine Weyl group action.
• Comonadicity and Barr-Beck-Lurie: A major technical tool is the Barr-Beck-Lurie theorem. The authors prove that certain functors (like averaging functors and pushforwards) are comonadic. This allows them to reconstruct the entire category of nondegenerate G-categories from the category of comodules over a specific comonad acting on $\text{IndCoh}(\mathfrak{t}^*)$.
• Mellin Transform: The paper utilizes a higher-categorical version of the Mellin transform, which provides a symmetric monoidal equivalence between D-modules on a torus $T$ and ind-coherent sheaves on the dual torus (or its quotient).

3. Key Contributions and Results

A. Classification of Nondegenerate G-Categories (The Main Theorem)

The primary result (Theorem 1.2 and Theorem 1.14) establishes an equivalence of 2-categories:
$$\text{G-mod}_{\text{nondeg}} \simeq \text{IndCoh}(\Gamma_{\widetilde{W}_{aff}})\text{-mod}$$
This implies that the 2-category of nondegenerate G-categories is equivalent to the 2-category of module categories over the monoidal category $\text{IndCoh}(\Gamma_{\widetilde{W}_{aff}})$.

• Geometric Interpretation: $\Gamma_{\widetilde{W}_{aff}}$ is identified with the fiber product $\mathfrak{t}^* \times_{\mathfrak{t}^* // \widetilde{W}_{aff}} \mathfrak{t}^*$.
• Universal Object: The category $D(N \backslash G / N)_{\text{nondeg}}$ serves as a universal nondegenerate G-category, equivalent to $\text{IndCoh}(\Gamma_{\widetilde{W}_{aff}})$.

B. Monoidal Upgrade of Whittaker-Hecke Equivalences

The authors upgrade the classical equivalence between bi-Whittaker D-modules and equivariant sheaves on $\mathfrak{t}^*$ to a monoidal equivalence (Theorem 1.4).

• Result: There is a monoidal, $t$-exact, fully faithful functor:
  $$\mathcal{H}_\psi \to D(T)^W$$
  where $\mathcal{H}_\psi$ is the category of bi-Whittaker D-modules.
• Significance: This proves that the Whittaker-Hecke category is symmetric monoidal, resolving a question by Drinfeld. The proof is more direct than previous approaches relying on the geometric Satake equivalence.

C. W-Equivariance of Parabolic Restriction

The paper proves a modified conjecture of Ben-Zvi and Gunningham regarding very central D-modules (Theorem 1.22).

• Definition: A sheaf is "very central" if its horocycle transform is supported on the closed embedding of the torus.
• Result: The parabolic restriction of a very central D-module acquires a canonical Weyl group equivariant structure such that the resulting sheaf descends to the coarse quotient $\mathfrak{t}^* // \widetilde{W}_{aff}$.
• Method: This is achieved by constructing a "nondegenerate horocycle functor" that lifts the standard horocycle functor to the nondegenerate setting, where the Weyl group action is well-defined.

D. The Mellin Transform (Appendix A)

In collaboration with Germán Stefanich, the authors rigorously construct the Mellin transform as a symmetric monoidal equivalence of DG categories (Theorem A.2).

• They upgrade the classical abelian equivalence to the derived setting using operadic methods and the theory of Grothendieck categories with enough projectives.
• This provides the necessary foundation for the monoidal structures used in the main theorems.

4. Technical Highlights

• Nondegeneracy as Localization: The paper treats nondegeneracy as a localization of the 2-category of all G-categories. This perspective allows the authors to "bootstrap" results from the universal case to general nondegenerate categories.
• Comonadicity on the Heart: A significant technical hurdle is that the pushforward functor to the coarse quotient is not conservative on the full $\text{IndCoh}$ category. The authors overcome this by proving comonadicity on the eventually coconnective subcategory (the "heart" in a derived sense) and then extending the result to the full category via ind-completion.
• W-Action on Invariants: The authors show that for any nondegenerate G-category $\mathcal{C}$, the $N$-invariants $\mathcal{C}^N$ naturally acquire a $W$-action. This is a crucial step in relating $N$-invariants to Whittaker invariants.

5. Significance and Impact

1. Unification of Geometric Representation Theory: The paper provides a unified framework describing a vast class of representation-theoretic categories (including Whittaker models and parts of category $\mathcal{O}$) purely in terms of the root datum and the geometry of the dual Cartan.
2. Resolution of Open Problems: It definitively answers Drinfeld's question about the symmetric monoidal structure of the Whittaker-Hecke category and proves the Ben-Zvi–Gunningham conjecture on the descent of very central sheaves.
3. Methodological Advancement: The work demonstrates the power of categorical representation theory and ind-coherent sheaves in solving classical problems. It moves beyond the geometric Satake equivalence, offering a more direct path to monoidal structures via averaging functors and descent.
4. Foundations for Local Geometric Langlands: The classification of nondegenerate categories is interpreted as an "upgraded" version of the Whittaker functor, which is central to the local geometric Langlands program. The results suggest that generic G-categories "diagonalize" over the coarse quotient $\mathfrak{t}^* // \widetilde{W}_{aff}$.

In summary, Gannon's paper establishes a deep structural link between the representation theory of reductive groups and the geometry of the dual Cartan subalgebra, utilizing advanced categorical tools to upgrade classical equivalences to the monoidal level and resolve long-standing conjectures.