🔢 mathematics

Remarks on diagonal dimension for algebraic stacks

This paper investigates the Rouquier dimension of the bounded derived category of coherent complexes on Noetherian algebraic stacks by introducing diagonal dimension to establish upper bounds for smooth morphisms, recover strong generators for fiber products, and prove that the diagonal dimension of varieties with mild singularities is at most twice their Krull dimension.

Original authors: Pat Lank, Fei Peng

Published 2026-05-14

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Pat Lank, Fei Peng

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 complexity of a massive, intricate city. In the world of mathematics, this "city" is a shape called an algebraic stack. These aren't just simple lines or circles; they are complex geometric structures that can have "twists," "folds," and even "singularities" (rough spots or sharp corners).

Mathematicians want to know: How complicated is this city? To measure this, they use a tool called the Rouquier dimension. Think of this dimension as a "complexity score." A low score means the city is simple and easy to navigate; a high score means it's a labyrinth.

The paper you provided is a guidebook written by Pat Lank and Fei Peng. They are trying to figure out how to calculate this complexity score for these tricky geometric cities, especially when the cities have rough edges or exist in different mathematical "universes" (like different number systems).

Here is a breakdown of their main ideas using simple analogies:

1. The Problem: Measuring the Maze

The authors are looking at a specific type of mathematical object called a derived category. Imagine this as a giant library containing every possible "map" or "blueprint" of the city.

The Goal: They want to find a "Master Key" (called a strong generator). If you have this one key, you can unlock every single door in the library by combining it with itself a certain number of times.
The Metric: The Rouquier dimension is the minimum number of times you need to combine that Master Key to unlock the whole library. The fewer times you need to combine it, the simpler the city is.

2. The New Tool: "Diagonal Dimension"

Calculating the complexity score directly is like trying to count every brick in a castle while it's being built. It's hard.
The authors introduce a new, easier-to-calculate tool called Diagonal Dimension.

The Analogy: Imagine the city has a "diagonal" path that connects every point to itself. The Diagonal Dimension measures how "thick" or "complex" this diagonal path is.
The Benefit: They prove that if you know the thickness of this diagonal path, you can set a ceiling (an upper limit) on the complexity score. You don't need to count every brick; you just need to measure the diagonal to know the building can't be more complex than a certain amount.

3. Their Three Main Discoveries

A. The Smooth City Bound (Theorem 1.2)

If the city is "smooth" (no sharp corners) and built over a regular foundation (like a standard field of numbers), the authors found a specific formula to calculate the maximum complexity.

The Metaphor: Imagine you are building a smooth, glass skyscraper. The authors say, "If you know the height of the foundation and the width of the building, you can calculate exactly how many floors the building could have at most."
Why it matters: Before this, we only had these formulas for simple shapes. Now, they work for much more complex, "stacky" shapes (cities with extra layers of structure).

B. The Lego Block Rule (Proposition 1.3)

What happens if you take two cities and smash them together to make a new, bigger city (a "fiber product")?

The Analogy: Imagine you have two sets of Lego blocks. Set A can build City 1, and Set B can build City 2. The authors prove that if you mix Set A and Set B together, you can build the combined city.
The Result: They showed that the "Master Keys" for the two original cities can be combined to create the "Master Key" for the new, merged city. This is a powerful shortcut because it lets mathematicians build complex solutions from simple, known pieces.

C. The "Rough Spot" Rule (Theorem 1.4 & Corollary 1.5)

Real cities often have potholes, craters, or sharp corners (singularities). The authors looked at cities with "mild" rough spots.

The Discovery: They found that even if the city has these rough spots, the complexity score doesn't explode.
The Rule of Thumb: For a city with mild rough spots, the complexity score is at most twice the size of the city itself.
- Example: If a city is 3-dimensional, its complexity score won't exceed 6.
The Twist: This works even in "positive characteristic" (a weird mathematical universe where numbers wrap around like on a clock). This is a big deal because previous rules often broke down in these weird universes.

Summary

In short, Lank and Peng have given mathematicians a new ruler.

They created a way to measure the "thickness" of a geometric shape's diagonal to estimate its overall complexity.
They proved that for smooth shapes, this estimate is very precise.
They showed how to combine the complexity of two shapes to understand a merged shape.
They proved that even shapes with "rough edges" aren't infinitely complex; their complexity is strictly limited to twice their size.

This helps mathematicians navigate the "labyrinth" of algebraic geometry with a better map, ensuring they know exactly how complex their shapes can get without getting lost in the details.

Technical Summary: Remarks on Diagonal Dimension for Algebraic Stacks

Problem Statement
The paper addresses the complexity of the bounded derived category of coherent complexes, denoted $D^b_{\text{coh}}(X)$ , for a Noetherian algebraic stack $X$ . A central challenge in this field is determining the existence of "strong generators" (objects $G$ such that $D^b_{\text{coh}}(X) = \langle G \rangle_{n+1}$ for some finite $n$ ) and establishing explicit, computable upper bounds for the Rouquier dimension (the smallest such $n$ ). While these questions are well-understood for schemes, they remain subtle for algebraic stacks, where effective bounds are scarce and explicit constructions are rare. The authors aim to refine the understanding of this complexity by utilizing the notion of diagonal dimension.

Methodology
The authors employ the framework of triangulated categories and derived algebraic geometry, specifically focusing on the diagonal dimension of a morphism $f: Y \to X$ . Defined as the smallest integer $n$ such that the pushforward of the structure sheaf of the diagonal, $R\Delta_* \mathcal{O}_Y$ , lies in the subcategory generated by external tensor products of objects from $D^b_{\text{coh}}(Y)$ , this invariant serves as a tractable proxy for the Rouquier dimension.

The methodology involves:

Reduction to Single Generators: Proving that the diagonal dimension can be computed using a single object $G$ rather than a pair, simplifying the generation condition.
Spectral Sequence Analysis: Utilizing spectral sequences relating local cohomology and Ext groups to bound the cohomological dimension of regular algebraic stacks.
Base Change and Projection Formulas: Leveraging properties of concentrated, smooth, and proper morphisms to transfer generation properties between fiber products and their factors.
Resolution of Singularities: Applying results on resolutions of singularities and good moduli spaces to relate the diagonal dimension of singular varieties to that of smooth covers.

Key Contributions and Results

Explicit Upper Bounds for Smooth Morphisms (Theorem 1.2):
The authors establish an explicit, computable upper bound for the diagonal dimension of a concentrated, separated, smooth, finitely presented, quasi-DM morphism $f: Y \to \text{Spec}(R)$ , where $R$ is a regular affine scheme. If $U \to Y$ is a smooth surjective morphism from a scheme, the bound is given by:
$\dim_\Delta(f) \leq (\dim R + 1)(2 \max\{\dim U, \text{cd}(Y)\} + 1) - 1$
This result generalizes known bounds for smooth varieties, algebraic spaces, and Deligne–Mumford stacks to a broader context, including stacks over DVRs and those with affine stabilizers.
Generators for Fiber Products (Proposition 1.3):
The paper identifies strong generators for the derived category of a fiber product $Y_1 \times_S Y_2$ in terms of generators for the factors $Y_1$ and $Y_2$ . Specifically, if $G_i$ are classical generators for $D^b_{\text{coh}}(Y_i)$ , then the external tensor product of their pullbacks generates $D^b_{\text{coh}}(Y_1 \times_S Y_2)$ . This result recovers a theorem by Elagin–Lunts–Schnürer for varieties over perfect fields but extends it to algebraic stacks without requiring separatedness or finite Krull dimension on the base, and crucially, provides a proof that avoids dg-methods.
Comparison of Diagonal Dimensions (Theorem 1.4):
A comparison theorem is proven for morphisms $f_i: Y_i \to S$ where there exists an $S$ -morphism $f: Y_2 \to Y_1$ such that the unit $\mathcal{O}_{Y_1} \to Rf_* \mathcal{O}_{Y_2}$ is an isomorphism. Under these conditions, $\dim_\Delta(f_1) \leq \dim_\Delta(f_2)$ . This allows the transfer of dimension bounds from a "better" space to a "worse" one.
Bounds for Varieties with Mild Singularities (Corollary 1.5):
Applying the comparison theorem, the authors show that for a separated integral scheme $X$ over a field $k$ :
- If $X$ has rational singularities (characteristic zero), $\dim_\Delta(f) \leq 2 \dim X$ .
- If $X$ has linearly reductive singularities (arbitrary characteristic), $\dim_\Delta(f) \leq 2 \dim X$ .
  These results provide a new general bound for varieties with mild singularities, extending beyond the toric case where the diagonal dimension is often known to equal the Krull dimension.

Significance and Claims
The paper claims to provide a refined tool (diagonal dimension) for estimating the Rouquier dimension of algebraic stacks, a problem where explicit bounds are typically elusive. By establishing computable upper bounds for smooth morphisms and demonstrating how these bounds behave under fiber products and morphisms with isomorphic structure sheaf pushforwards, the work bridges the gap between the well-understood case of schemes and the more complex case of stacks.

The authors highlight that their results apply in greater generality than previous literature, covering stacks over mixed characteristic bases and those with specific stabilizer conditions. Furthermore, the identification of generators for fiber products offers a structural insight into the derived categories of stacky fiber products without relying on differential graded (dg) techniques. Finally, the bound of $2 \dim X$ for varieties with rational or linearly reductive singularities represents a significant step in understanding the complexity of derived categories in the presence of singularities, particularly in positive characteristic.