🔢 mathematics

Classification and nonexistence for $t$ -structures on derived categories of schemes

This paper classifies tensor $t$ -structures on the bounded derived category of coherent sheaves over suitable Noetherian schemes and demonstrates that the existence of such structures restricting to perfect complexes serves as a criterion for detecting the regularity of the scheme.

Original authors: Alexander Clark, Pat Lank, Kabeer Manali-Rahul, Chris J. Parker

Published 2026-05-19

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Alexander Clark, Pat Lank, Kabeer Manali-Rahul, Chris J. Parker

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 the mathematical world of algebraic geometry as a vast, complex city called Scheme City. In this city, the buildings are "schemes," and the things living inside them (like data, functions, or shapes) are organized into different neighborhoods called derived categories.

The authors of this paper are like urban planners and detectives trying to figure out how to organize these neighborhoods. Specifically, they are looking for a special kind of organization system called a t-structure.

Here is a simple breakdown of what they did, using everyday analogies:

1. The Problem: Sorting the City's Data

Think of the "derived category" as a giant warehouse filled with boxes of data. Some boxes are well-organized (perfect complexes), some are messy but bounded (bounded coherent sheaves), and some are huge and unbounded (quasi-coherent sheaves).

A t-structure is like a sorting rule. It tells you which boxes go into the "Morning Shift" (the aisle) and which go into the "Night Shift" (the co-aisle).

The Goal: The authors wanted to classify all possible valid sorting rules for these warehouses.
The Twist: They weren't just looking for any sorting rule. They wanted rules that play nicely with the city's "tensor product" (a way of combining data, like mixing colors or multiplying numbers). They call these tensor t-structures.

2. The Map: Thomason Filtrations

To describe these sorting rules, the authors use a tool called a Thomason filtration.

The Analogy: Imagine the city has a map where every point (location) is assigned a "time stamp" or a "level." A filtration is just a list of these levels that gets stricter as you go down the list.
The Discovery: The authors found a perfect one-to-one match (a bijection) between these "time-stamped maps" and the valid sorting rules. If you have a map, you can build a sorting rule. If you have a sorting rule, you can draw the map.

3. The First Big Discovery: The "Weak Cousin" Rule

The authors focused on specific neighborhoods where the data is "coherent" (well-behaved). They asked: When does a sorting rule designed for the whole city work perfectly inside a specific, smaller neighborhood?

They found the answer lies in a condition they call "Weak Cousin."

The Metaphor: Imagine a family tree in the city. If a "cousin" (a point in the city) is assigned to a specific "level" in your sorting rule, then their "direct ancestor" (a point they generalize from) must be assigned to the previous level.
The Result: If your map follows this "Weak Cousin" rule, the sorting system works perfectly for the well-behaved data. If it doesn't, the system breaks down when you try to use it on the smaller neighborhood.

4. The Second Big Discovery: The "Regularity" Test

This is the most dramatic part of the paper. The authors investigated what happens when the sorting rule is applied to the most perfect, well-behaved boxes in the warehouse: the Perfect Complexes.

They discovered a "Regularity Test":

The Scenario: Imagine you have a specific neighborhood (a closed subset $Z$ ) in Scheme City.
The Test: Can you create a sorting rule that works perfectly for the "Perfect" boxes in this neighborhood?
The Verdict:
- YES: If and only if the neighborhood is Regular. In math terms, "Regular" means the neighborhood is smooth, without any sharp corners, singularities, or "cracks" in its geometry.
- NO: If the neighborhood has any "cracks" (singularities), you simply cannot create such a sorting rule.

Why is this cool?
It turns a very abstract algebraic problem (can we sort these boxes?) into a geometric one (is the neighborhood smooth?). It's like saying, "If you can't organize this library perfectly, it's because the building itself is crooked."

5. Local-to-Global Principles

The authors also showed that you don't need to check the whole city at once.

The Analogy: If you want to know if a sorting rule works for the whole city, you just need to check if it works in every single small block (local rings) and every open street (open subschemes).
The Takeaway: If the rule works everywhere locally, it works globally. This allows them to solve big, scary problems by breaking them down into tiny, manageable pieces.

Summary of Their Achievements

Classification: They mapped out exactly which sorting rules (t-structures) exist for well-behaved data in these geometric cities, using "Thomason filtrations" as the blueprint.
The "Weak Cousin" Condition: They identified the specific topological rule (the family tree logic) that determines if a sorting rule works for coherent data.
Regularity Detection: They proved that the existence of a sorting rule for "perfect" data is a litmus test for whether a geometric space is smooth (regular). If the space is "broken" (singular), the sorting rule cannot exist.

In short, the paper provides a new dictionary to translate between geometric shapes (smooth vs. broken) and algebraic sorting systems (t-structures), showing that the two are inextricably linked.

Technical Summary: Classification and Nonexistence for t-Structures on Derived Categories of Schemes

Problem Statement
The paper addresses the classification of tensor t-structures on derived categories of coherent sheaves ( $D^b_{\text{coh}}(X)$ ) and perfect complexes ( $\text{Perf}(X)$ ) over Noetherian schemes, particularly those with prescribed support in a closed subset $Z$ . While the classification of compactly generated tensor t-structures on the unbounded derived category of quasi-coherent sheaves ( $D_{\text{qc}}(X)$ ) is well-understood via Thomason filtrations (established in works like [ATJLS10], [DS23], [Lan25]), the problem of determining when these structures restrict to subcategories of coherent or perfect complexes remains subtle.

A central motivation is the Antieau–Gepner–Heller conjecture, which posits that $\text{Perf}(X)$ admits a bounded t-structure if and only if $X$ is regular. This was recently resolved by Neeman [Nee24] in the general case and Smith [Smi22] in the affine case. This paper seeks to extend these results to arbitrary closed subsets $Z$ and to provide a topological characterization of when t-structures on $D_{\text{qc}}(X)$ restrict to $D^b_{\text{coh}, Z}(X)$ and $\text{Perf}_Z(X)$ .

Methodology
The authors employ a "local-to-global" strategy, leveraging the fact that many properties of t-structures on schemes can be detected locally on affine open subschemes or at stalks. The core technical tools include:

Thomason Filtrations: The classification relies on the bijective correspondence between compactly generated tensor t-structures on $D_{\text{qc}}(X)$ and Thomason filtrations (sequences of specialization-closed subsets).
Relative Weak Cousin Condition: The authors introduce and utilize a "relative weak Cousin" condition for Thomason filtrations across a closed subset $Z$ . This generalizes the classical weak Cousin condition from [ATJLS10] to the relative setting.
Restriction and Localization: The paper establishes that the restriction of a t-structure to a closed subset $Z$ (or an open subscheme) corresponds to the restriction of the associated Thomason filtration. Lemmas 3.5, 3.6, and 3.17 demonstrate that truncation functors commute with restriction to open affines and local rings, allowing global questions to be reduced to local algebraic conditions.
Support Theory and Regularity: The nonexistence results rely on analyzing the behavior of Koszul complexes and local cohomology. Specifically, the authors use the dimension of supports and the properties of local rings (regular vs. non-regular) to derive contradictions when t-structures are assumed to restrict to perfect complexes in non-regular settings.

Key Contributions and Results

Classification on Coherent Sheaves with Support (Theorem 1.2 & Proposition 1.1):
For a quasi-compact CM-excellent scheme $X$ of finite Krull dimension and a closed subset $Z$ , the paper establishes a one-to-one correspondence between:
- Tensor t-structures (or $\otimes$ -aisles) on $D^b_{\text{coh}, Z}(X)$ .
- Thomason filtrations on $X$ that are "weak Cousin across $Z$ ."
A filtration $\phi$ is weak Cousin across $Z$ if, for any direct generalization $p \leadsto q$ in $Z$ , $q \in \phi(n)$ implies $p \in \phi(n-1)$ . This result generalizes affine classification results [ATJLS10, Tak23] to the global setting and provides a topological criterion for the existence of such restrictions.
Nonexistence and Regularity Detection (Theorem 1.3):
The paper proves that a nonempty closed subset $Z$ of a Noetherian scheme $X$ is contained in the regular locus of $X$ if and only if there exists a non-constant Thomason filtration on $X$ whose associated t-structure on $D_{\text{qc}}(X)$ restricts to $\text{Perf}_Z(X)$ .
- If $Z$ contains a singular point, no non-constant t-structure can restrict to $\text{Perf}_Z(X)$ .
- This recovers Neeman's theorem [Nee24] in the affine case ( $Z=X$ ) and extends Smith's result [Smi22] to arbitrary Krull dimensions.
- The proof utilizes local algebra techniques (specifically properties of local cohomology and Koszul complexes) rather than the approximable triangulated category methods used by Neeman.
Classification on Perfect Complexes with Support (Corollary 1.4):
Combining the above results, the authors classify t-structures on $\text{Perf}_Z(X)$ . A Thomason filtration $\phi$ yields a t-structure restricting to $\text{Perf}_Z(X)$ if and only if, for each connected component $Z_t$ of $Z$ :
- If $Z_t$ is in the regular locus, $\phi \cap Z_t$ satisfies the weak Cousin condition.
- If $Z_t$ is not in the regular locus, $\phi \cap Z_t$ must be constant.

Significance and Claims
The paper claims to establish robust local-to-global principles for tensor t-structures, bridging the gap between the well-understood affine case and the global scheme setting. By characterizing the restriction of t-structures to coherent and perfect complexes via topological conditions on Thomason filtrations, the work provides a complete classification for these subcategories under the CM-excellent hypothesis.

Furthermore, the nonexistence results offer a new perspective on the relationship between homological algebra and geometric regularity. The authors note that their methods provide an alternative proof for the affine case of the Antieau–Gepner–Heller conjecture and extend it to infinite Krull dimensions, applying to specific examples of non-regular Noetherian integral domains constructed by Hochster [Hoc73]. The results also have ramifications for the classification of t-structures arising from perversity functions, offering an effective conclusion to that classification problem on $D^b_{\text{coh}}(X)$ .