Inductive systems of the symmetric group, polynomial functors and tensor categories

Imagine you are a master chef trying to understand the fundamental rules of cooking. You have a giant kitchen (a Tensor Category) filled with ingredients (objects) and recipes (operations). You want to know: "If I mix these ingredients together in every possible way, what new flavors (representations) can I create?"

This paper, written by Kevin Coulembier, is like a massive cookbook that tries to solve a very specific, high-level puzzle about how these flavors behave, especially when the kitchen is running on "spicy" rules (mathematical fields with positive characteristic, which behave differently than our usual "mild" zero-characteristic world).

Here is the breakdown of the paper's big ideas using everyday analogies:

1. The Main Problem: The "Symmetric Group" Puzzle

In math, the Symmetric Group ( $S_d$ ) is like a group of dancers who can swap places in $d$ different positions. When you mix ingredients in your kitchen, the order matters. Swapping two ingredients might change the dish, or it might not.

The paper asks: Which specific "dance moves" (representations of the symmetric group) can actually happen inside these magical kitchens?

In a normal kitchen (characteristic zero), the answer is simple: you can do almost any dance. But in a "spicy" kitchen (characteristic $p$ ), some moves are forbidden, and the rules are much more complex. The author wants to map out exactly which dances are allowed in which kitchens.

2. The Three Ways to Look at the Same Coin

The author discovers that this problem can be solved in three different ways, which are actually just different lenses looking at the same reality. Think of it like describing a mountain: you can describe it by its height, its shape, or the rocks on its surface. They are all the same mountain.

Lens A: The "Inductive System" (The DNA of the Kitchen)

Imagine every kitchen has a "DNA sequence" that dictates which dance moves are possible.

The author defines a system called an Inductive System. This is like a rulebook that says, "If you can do a dance with 3 steps, you must also be able to do a specific dance with 2 steps."
They found that for certain famous kitchens (like the Verlinde categories), this DNA is very specific. For example, one kitchen only allows "completely splittable" dances (dances that break down into simple, non-overlapping moves).
The Discovery: They proved that the "DNA" of these kitchens is the key to understanding the whole structure of the kitchen.

Lens B: The "Universal Polynomial Functors" (The Magic Recipe Book)

Imagine you have a magical recipe book that works in every kitchen in the universe.

A Polynomial Functor is a recipe that takes an ingredient (like a vector space) and turns it into a new dish (like a tensor power) in a consistent way.
The author creates a "Universal Recipe Book" ( $P^d_k$ ). This book contains the master recipes that work everywhere.
The Discovery: The author proves that the list of all possible recipes in this Universal Book is exactly the same as the list of all possible dance moves (Inductive Systems) found in Lens A. If you know the recipes, you know the dances, and vice versa.

Lens C: "Strict Polynomial Functors" (The Local Chef's Special)

This is like looking at a specific chef in a specific kitchen and asking, "What can you make with your ingredients?"

The author generalizes a classic math concept (Schur functors) to work in any magical kitchen.
The Discovery: They show that the "Local Chef's Specials" are just a specific packaging of the "Universal Recipes." If a chef has enough ingredients (a "discerning object"), their local menu is a perfect reflection of the universal one.

3. The "Annihilator" (The Forbidden Moves)

In the spicy kitchens, some dance moves are so "bad" that they destroy the dish. In math, these are called Annihilator Ideals.

The author figured out that for the most famous spicy kitchens, these "forbidden moves" are generated by just one or two simple rules (like "don't swap these two specific dancers").
This is a huge simplification! Instead of a complex list of forbidden moves, they found that a single "symmetrizer" (a rule to make things symmetric) or "skew-symmetrizer" (a rule to make things anti-symmetric) is enough to ban everything else.

4. The Big Picture: Why Does This Matter?

The paper connects three huge areas of mathematics that were previously thought to be separate:

Representation Theory: How groups (like dancers) act on things.
Tensor Categories: The abstract structure of these magical kitchens.
Polynomial Functors: The algebraic recipes used to build new things from old ones.

The Analogy:
Imagine you are trying to understand a new language.

Lens A studies the grammar rules (Inductive Systems).
Lens B studies the dictionary of universal words (Universal Functors).
Lens C studies how a specific person speaks (Strict Functors).

The author's breakthrough is proving that Grammar = Dictionary = Speech. If you understand one, you automatically understand the others.

Summary of the "Takeaways"

The "Verlinde" Connection: The author confirmed a major conjecture that the most complex "moderate growth" kitchens are built from a specific chain of simpler kitchens ( $Vec \subset sVec \subset Ver_p \dots$ ).
The "Frobenius-Exact" Test: They gave a new, simple test to see if a kitchen is "well-behaved" (Frobenius-exact). If the kitchen passes this test, its "DNA" (Inductive System) is closed and stable.
New Tools: They provided a new way to classify "completely splittable" representations, showing they are "algebraic" (they follow a polynomial pattern).

In a nutshell: This paper is a unifying theory. It says, "Stop looking at these problems as three separate puzzles. They are the same puzzle seen from three different angles. Once you solve one, you've solved them all." It provides a new, powerful toolkit for mathematicians to explore the structure of these abstract, high-dimensional worlds.

Here is a detailed technical summary of the paper "Inductive Systems of the Symmetric Group, Polynomial Functors and Tensor Categories" by Kevin Coulembier.

1. Problem Statement and Motivation

The paper addresses a fundamental gap in the structure theory of tensor categories over fields of positive characteristic ( $p > 0$ ).

Context: In characteristic zero, Deligne's theorems establish that tensor categories of "moderate growth" are essentially representation categories of groups or supergroups. In positive characteristic, the landscape is richer and more complex, involving a chain of "incompressible" categories (e.g., $\text{Vec} \subset \text{sVec} \subset \text{Ver}_p \subset \text{Ver}_{p^2} \dots$ ).
The Core Question: When a tensor category $\mathcal{C}$ $C$ acts on an object $X$ $X$ via the braiding (symmetric group action on $X^{\otimes d}$ $X^{\otimes d}$ ), which representations of the symmetric group $S_d$ $S_{d}$ appear?
- Specifically, the paper seeks to classify the subcategories of $\text{Rep}(S_d)$ generated by the images of the braid morphism $\beta^d_X: kS_d \to \text{End}_{\mathcal{C}}(X^{\otimes d})$ .
Related Questions:
1. Inductive Systems: How do these representations behave as $d$ varies?
2. Universal Polynomial Functors: Can we define functors that act coherently across all tensor categories (generalizing Schur functors)?
3. Strict Polynomial Functors: How do classical strict polynomial functors (defined on vector spaces) generalize to arbitrary tensor categories?

The author posits that these three questions are different formulations of the same underlying structural problem.

2. Methodology

The paper employs a synthesis of modular representation theory, category theory, and invariant theory.

Inductive Systems: The author introduces the concept of an inductive system $\mathcal{A}$ , which assigns a pseudo-abelian subcategory $\mathcal{A}_n \subset \text{Rep}(S_n)$ for each $n$ , compatible with restriction to $S_{n-1}$ . This framework allows for the systematic study of representations appearing in tensor categories.
Braid Action and Annihilators: For an object $X \in \mathcal{C}$ , the paper analyzes the representations $\omega(X^{\otimes d})$ (where $\omega$ is a fiber functor) to define an inductive system $\mathcal{B}_X[\mathcal{C}]$ . It connects these systems to annihilator ideals in the finitary symmetric group algebra $kS_\infty$ .
Tensor Product of Categories: The paper utilizes the Deligne tensor product ( $\mathcal{C} \boxtimes \mathcal{D}$ ) to relate representations in a specific category to universal functors.
Strict Polynomial Functors: The author generalizes the category of strict polynomial functors (originally defined on $\text{Vec}$ by Friedlander and Suslin) to arbitrary tensor categories $\mathcal{C}$ by defining them as enriched functors on a specific category $\Gamma_d\mathcal{C}$ .
Frobenius-Exactness: A crucial technical tool is the notion of Frobenius-exact tensor categories (where the Frobenius functor is exact), which allows for the control of inductive systems via fiber functors to Verlinde categories ( $\text{Ver}_p$ ).

3. Key Contributions and Results

A. Classification of Inductive Systems from Tensor Categories

The Inductive System $\mathcal{B}[\mathcal{C}]$ : For any tensor category $\mathcal{C}$ , the author constructs a closed inductive system $\mathcal{B}[\mathcal{C}] = \sum_{X \in \mathcal{C}} \mathcal{B}_X[\mathcal{C}]$ .
Connection to Verlinde Categories:
- For $\mathcal{C} = \text{Vec}$ , $\mathcal{B}[\text{Vec}]$ is the system of Young modules.
- For $\mathcal{C} = \text{sVec}$ , it is the system of signed Young modules.
- For $\mathcal{C} = \text{Ver}_p$ (the Verlinde category), $\mathcal{B}[\text{Ver}_p]$ corresponds to the system of completely splittable (CS) representations, classified by Kleshchev.
New Systems: The paper identifies new closed inductive systems, such as those arising from $\text{Ver}_{p^2}$ and $\text{Ver}_4$ (for $p=2$ ), leading to strict inclusions like $\mathcal{B}[\text{Vec}] \subsetneq \mathcal{B}[\text{Ver}_4^+] \subsetneq \mathcal{B}[\text{Ver}_4]$ .
Algebraicity: A significant result is that every completely splittable representation is algebraic (its tensor powers involve only finitely many indecomposable types).

B. Annihilator Ideals in $kS_\infty$

The paper reformulates the classification of maximal ideals in $kS_\infty$ (for $p>2$ ) using tensor categories.
Result: The maximal ideals in $kS_\infty$ are precisely the annihilators $\text{Ann}(L)$ of simple objects $L$ in $\text{Ver}_p$ .
Generators: These ideals are shown to be generated by a single symmetrizer and/or a single skew-symmetrizer (e.g., $a_{i+1}$ and $s_{p+1-i}$ ). This provides a new, simple description of these ideals.

C. Equivalence of Three Perspectives

The paper's central theoretical achievement is proving that three seemingly distinct concepts are equivalent:

Inductive Systems: The classification of representations appearing in tensor categories.
Universal Polynomial Functors: Functors defined on the 2-category of tensor categories that commute with tensor functors.
Strict Polynomial Functors: Functors defined on a specific tensor category $\mathcal{C}$ via the category $\Gamma_d\mathcal{C}$ .

Theorem 9.1.1 (Main Equivalence):
Let $\mathcal{C}$ be a tensor category and $X \in \mathcal{C}$ .

The category of strict polynomial functors on $\mathcal{C}$ , denoted $\text{SPol}^\circ_d(\mathcal{C})$ , is equivalent to the category of polynomial representations $\text{Rep}^d_{\mathcal{C}}(\text{GL}_X)$ if and only if the inductive system of $X$ equals the system of the whole category ( $\mathcal{B}^d_X = \mathcal{B}^d[\mathcal{C}]$ ).
The category of universal polynomial functors $\text{Pol}^d_k$ is equivalent to $\text{mod}(\mathcal{TC}_d)$ , where $\mathcal{TC}_d$ is the sum of $\mathcal{B}^d[\mathcal{C}]$ over all tensor categories.
Specifically, $\text{Pol}^d_k \simeq \text{mod}(\mathcal{B}^d[\text{Ver}_{p^\infty}])$ (assuming the conjecture that all moderate growth categories map to $\text{Ver}_{p^\infty}$ ).

D. Discerning Objects

The author defines $d$ -discerning objects (objects $X$ where $\mathcal{B}^d_X = \mathcal{TC}_d$ ).

It is proven that in categories of moderate growth, no object is $d$ -discerning for all $d$ .
However, for specific categories (like $\text{Ver}_p$ ), specific objects (like the generator) can be discerning for specific degrees, allowing for a complete classification of polynomial representations in those contexts.

4. Significance and Impact

Unification of Theory: The paper provides a unified framework linking modular representation theory of symmetric groups, the structure of tensor categories, and the theory of polynomial functors. It demonstrates that classifying polynomial functors is equivalent to classifying which symmetric group representations appear in tensor categories.
Advancement in Structure Theory: By characterizing the inductive systems of $\text{Ver}_p$ and related categories, the paper offers tools to test the BEO Conjecture (that all moderate growth tensor categories admit a functor to $\text{Ver}_{p^\infty}$ ). If a category has an inductive system not contained in $\mathcal{B}[\text{Ver}_{p^\infty}]$ , the conjecture would be false.
New Tools for Cohomology: The generalization of strict polynomial functors to arbitrary tensor categories opens new avenues for proving cohomological finite generation for finite tensor categories (a major open problem in the field).
Explicit Descriptions: The paper provides explicit generators for maximal ideals in $kS_\infty$ and clarifies the structure of "completely splittable" modules, showing they are algebraic.
Counter-examples and Nuance: The work highlights the complexity of positive characteristic by showing that the "universal" system $\mathcal{TC}_d$ is strictly larger than the systems generated by specific categories like $\text{Vec}$ or $\text{sVec}$ , and that the chain of incompressible categories leads to strictly ascending chains of inductive systems.

In summary, Coulembier's work establishes that the "shape" of a tensor category is encoded in the inductive systems of symmetric group representations it generates, and that polynomial functors provide the correct language to describe and classify these shapes across all tensor categories.