🔢 mathematics

Hochschild Cohomology, Modular Tensor Categories, and Mapping Class Groups II

This paper explicitly computes the action of mapping class groups on derived block spaces for Drinfel'd doubles of finite groups over fields of positive characteristic by connecting Lyubashenko's approach with representation varieties, thereby demonstrating that these representations generally differ from those on ordinary block spaces.

Original authors: Simon Lentner, Svea Nora Mierach, Christoph Schweigert, Yorck Sommerhaeuser

Published 2026-08-12

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Simon Lentner, Svea Nora Mierach, Christoph Schweigert, Yorck Sommerhaeuser

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are trying to understand the hidden rules that govern how shapes can be twisted, stretched, and woven together without tearing. This is the world of topology, a branch of mathematics where a coffee mug and a donut are considered "the same" because they both have exactly one hole. Now, take that idea and mix it with quantum physics, where particles can be in two places at once and the order in which you swap them changes the outcome. When mathematicians combine these fields, they create "modular categories." Think of these as rulebooks for a cosmic dance. In this dance, the "steps" are performed on surfaces like spheres or donuts (tori), and the "dancers" are abstract objects that follow very strict, magical laws.

For a long time, mathematicians only knew how to describe the dance when the rules were simple and the dancers were well-behaved (a state called "semisimple"). But in the real quantum world, things can get messy and tangled. The big question was: What happens to the dance when the rules get complicated? Specifically, if we look deeper into the math, do we find entirely new, hidden steps that were invisible before? This paper dives into that question, exploring whether these "messy" versions of the dance reveal new patterns that the simple versions completely miss.

The Paper's Story: Uncovering Hidden Steps in the Cosmic Dance

In this paper, the authors act like detectives investigating a mysterious dance troupe. They are studying a specific type of mathematical object called the Drinfel'd double of a finite group. To visualize this, imagine a finite group as a small, finite set of dance moves (like "spin left," "spin right," "jump"). The Drinfel'd double is a way of combining these moves into a massive, complex choreography.

The authors are particularly interested in what happens when the "floor" they are dancing on has a specific mathematical property: the characteristic of the field (think of this as the type of grid or coordinate system the dance is happening on). They focus on cases where this grid has a "characteristic" of 2 or 3. In these specific grids, the usual rules of arithmetic break down in a way that makes the dance "non-semisimple"—meaning the dancers can get stuck in loops or tangles that don't happen on a normal grid.

The Main Discovery: New Moves Exist
The central finding of the paper is a resounding "yes" to the question of whether new steps exist. The authors prove that when you look at the "derived block spaces" (a fancy term for the deeper, more complex layers of the dance), you find representations of the mapping class group that are completely different from the ones you see in the simple, "ordinary" layers.

To use an analogy: Imagine you are watching a magic show. In the front row (degree zero), you see the magician pull a rabbit out of a hat. It's a classic trick. But if you could see the "derived" version of the show (the hidden machinery behind the curtain), you would see the magician pulling out a dragon, a unicorn, and a phoenix. These are not just bigger versions of the rabbit; they are entirely different creatures that were never visible from the front row. The paper proves that for certain groups (specifically the symmetric group on three letters, $S_3$ ), these "new creatures" (representations) appear in higher degrees of the math, and they cannot be created just by combining the old tricks.

The Specific Cases: When the Grid Matters
The authors tested this theory on the smallest non-abelian group, the symmetric group on three letters ( $S_3$ ), which is like a group of three friends who can swap places in six different ways. They looked at a surface shaped like a torus (a donut), which has a mapping class group known as $SL(2, \mathbb{Z})$ (the group of all ways to twist the donut).

In Characteristic 2: When they ran the math on a grid with characteristic 2, they found that while the "derived" layers existed, they didn't actually contain any new types of moves. The new layers were just copies of the old ones, just stacked on top of each other. It was like finding a second floor in a house, but it was built exactly like the first floor.
In Characteristic 3: However, when they switched to a grid with characteristic 3, the magic happened. They discovered that in degrees where the number is 1 or 2 modulo 4 (like 1, 2, 5, 6, 9, 10...), entirely new representations appeared. These were moves that simply did not exist in the front row (degree zero).

The "Yoneda" Trap: You Can't Just Build Them
One of the most exciting parts of the paper is what they ruled out. In mathematics, there is a way to build complex structures by combining simple ones, called the Yoneda product. It's like building a tower by stacking blocks. A natural guess would be: "Maybe these new, complex moves are just the old moves stacked up in a clever way?"

The authors proved this is false. They showed that the new representations found in characteristic 3 cannot be generated by taking the simple, degree-zero moves and stacking them together using the Yoneda product. The new moves are fundamentally alien; they are not just a recombination of the old ones. They are genuinely new discoveries that require their own unique mathematical description.

How They Did It
To find these hidden steps, the authors used a clever trick. They connected the abstract dance of the Drinfel'd double to something called the representation variety. Imagine the representation variety as a map of all possible ways the group can act on a surface. By translating the complex quantum dance into this map, they could use tools from group cohomology (a branch of math that studies holes and tangles in groups) to calculate the exact shape of the dance.

They broke the problem down into smaller pieces, analyzing how the group $S_3$ acts on different parts of the map. They found that the "new" representations in characteristic 3 were linked to a specific "sign" representation (a move that flips the sign of the result) that only becomes visible when the grid has characteristic 3.

The Conclusion
The paper concludes that the "derived block spaces" are not just a fancy rewording of the old spaces. They are a richer, more complex world. In the case of the symmetric group $S_3$ on a torus with a characteristic 3 grid, the mapping class group (the group of twists) has representations in higher degrees that are:

Different from the ones in degree zero.
Not constructible from the degree zero ones using standard stacking methods.

This means that if you only look at the "ordinary" block spaces, you are missing a whole layer of the mathematical universe. The "derived" layers reveal a deeper, more intricate structure that is essential for understanding the full picture of these non-semisimple modular categories. The authors have successfully mapped out these new territories, showing us that the quantum dance has steps we never knew existed.

Technical Summary: Hochschild Cohomology, Modular Tensor Categories, and Mapping Class Groups II. Examples

Problem Statement
This work addresses the structure of mapping class group representations on "derived block spaces" within the context of non-semisimple modular categories. In the preceding part of this study (referenced as [LMSS1]), the authors generalized the action of mapping class groups on spaces of chiral conformal blocks to derived block spaces, defined as the cohomology groups of a cochain complex constructed from projective resolutions. A central open question remained: Are these representations on derived block spaces genuinely new, or are they isomorphic to the representations on the ordinary (degree zero) block spaces? Furthermore, it was unclear whether the higher-degree representations could be generated from the degree zero component via Yoneda products (the algebraic structure corresponding to the cup product in cohomology).

Methodology
To answer these questions, the authors focus on a specific, non-semisimple modular category: the category of representations of the Drinfel'd double $D(G)$ of a finite group $G$ over a field $K$ of positive characteristic (where the characteristic divides the order of $G$ ). The methodology proceeds through several key steps:

Alternative R-matrix and Representation Varieties: The authors utilize an "alternative" R-matrix $\tilde{R}$ for the Drinfel'd double, which simplifies formulas compared to the standard R-matrix. They establish an isomorphism (Theorem 4.28) between the mapping class group action on the derived block spaces (specifically the homogeneous components of the coend $L^{\otimes g}$ ) and the action on the "linearized representation variety" $M^g = K[\text{Hom}(\pi_1(\Sigma_g, x), G)]$ . This connects the categorical construction to the set of group homomorphisms from the surface fundamental group to $G$ .
Reduction to Group Cohomology: Using the fact that the cohomology of the Drinfel'd double can be computed via the group cohomology of the underlying finite group (Proposition 5.3), the problem is reduced to computing $H^m(G, M^g)$ .
Decomposition of the Representation Variety: The authors decompose the representation variety based on the conjugacy classes of subgroups of $G$ . For a subgroup $U$ , they analyze the orbit space under the conjugation action of $G$ and the mapping class group. This leads to a decomposition of the cohomology groups involving the centralizers and normalizers of these subgroups (Theorem 6.2, Corollary 6.3).
Explicit Computation for $S_3$ : The authors perform explicit calculations for the smallest non-abelian group, the symmetric group $S_3$ , on a torus ( $g=1$ ). They analyze the cases where the base field $K$ has characteristic 2 and characteristic 3, utilizing periodic resolutions for cyclic subgroups and analyzing the Yoneda algebra structure (cup products) of the cohomology ring $H^*(S_3, K)$ .

Key Contributions and Results

Isomorphism of R-matrix Representations: The authors prove that the mapping class group representations arising from the standard R-matrix and the alternative R-matrix are isomorphic (Theorem 4.36), justifying the use of the simpler alternative R-matrix for computations.
Non-Isomorphism of Representations: In the case of $S_3$ with characteristic 3, the authors demonstrate that the mapping class group representations appearing in higher cohomological degrees are not isomorphic to those in degree zero. Specifically, for cohomological degrees $m \equiv 1, 2 \pmod 4$ , new representations appear (Theorem 7.7). These new representations are distinct from the degree zero representations (which correspond to $K[P^1_3]$ ) and involve a different action of the modular group (specifically, $s^2$ acts as multiplication by $-1$ in the new representations, whereas it acts as the identity in the degree zero case).
Failure of Yoneda Generation: The paper shows that these new higher-degree representations cannot be generated from the degree zero component by taking Yoneda products with the cohomology ring of the unit object (Theorem 7.13). In characteristic 3, the submodule generated by the degree zero component misses the summands isomorphic to $K\chi_2[P^1_3]$ . This contrasts with the abelian case, where the derived block spaces form a free module over the cohomology ring of the unit object (Proposition 6.4).
Module Structure in Characteristic 2: In characteristic 2, while no new isomorphism classes of representations appear in higher degrees (all are present in degree zero), the module structure is not free over the cohomology ring of the unit object (Theorem 7.5). The derived block spaces are not generated by the degree zero component in the same way as in the abelian case.
Gluing Interpretation: The authors interpret the generation of missing summands (specifically those isomorphic to $H^*(S_3, K_\epsilon)$ ) via Yoneda products involving a different module $X = I(e, K_\epsilon)$ rather than the trivial module $K$ . This is interpreted geometrically as gluing a disk labeled by $X^*$ into a boundary component (Proposition 7.14).

Significance
The paper provides explicit examples demonstrating that derived block spaces in non-semisimple modular categories carry mapping class group representations that are fundamentally distinct from the ordinary block spaces. It establishes that these new representations are not merely "higher order" versions of the original ones generated by the Yoneda product structure. This confirms that the derived block spaces contain genuinely new topological and algebraic information that is lost if one considers only the degree zero (ordinary) conformal blocks. The work connects the abstract framework of Lyubashenko's modular functors with concrete group cohomology and representation varieties, offering a computational pathway to explore derived topological field theories over fields of positive characteristic.

The Paper's Story: Uncovering Hidden Steps in the Cosmic Dance

More like this