🔢 mathematics

A construction of tame sheaves and tame de Rham--Witt cohomology

This paper introduces an algebraic definition of the tame site for a pair $(X, \widetilde{X})$ to establish a general construction of tame sheaves from étale sheaves and local tame sections, which is then applied to big de Rham–Witt sheaves and reciprocity sheaves to compare tame syntomic cohomology with the Nygaard filtration on the tame de Rham–Witt complex.

Original authors: Alberto Merici, Kay Rülling, Shuji Saito

Published 2026-05-21

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Alberto Merici, Kay Rülling, Shuji Saito

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

The Big Picture: Fixing a Broken Map

Imagine you are trying to map a country (a mathematical object called a "scheme") to understand its hidden structures. In the world of mathematics, there are two main ways to look at a country:

The "Wild" Way (Étale Topology): This looks at the land with a very high-resolution microscope. It sees every tiny crack, every wild twist, and every chaotic edge. It's great for many things, but in certain conditions (specifically when working with numbers in "positive characteristic," like counting in a circle), this microscope gets confused by "wild ramification." It's like trying to map a stormy sea; the waves are too chaotic to get a clear, stable picture.
The "Tame" Way: This is a gentler approach. It ignores the chaotic, wild storms and focuses only on the "tame" parts of the landscape—places where the terrain changes smoothly and predictably.

The Problem: For a long time, mathematicians had a powerful tool called de Rham–Witt cohomology (a way to measure the "shape" and "holes" of these mathematical spaces) that worked beautifully in characteristic zero (like real numbers). But when they tried to use it in positive characteristic (like counting in a circle), the "Wild" microscope broke the tool. The tool didn't work well because it couldn't handle the wild chaos.

The Solution: The authors of this paper, Merici, Rülling, and Saito, built a new, custom-made lens called the "Tame Site." They also invented a machine (the $\beta$ -construction) that can take any standard mathematical object and "tame" it, forcing it to behave nicely even in chaotic environments.

Key Concepts Explained

1. The Tame Site: A "Filtered" View of Reality

Imagine you are looking at a forest through a window.

The Standard View (Étale): You see every leaf, every bug, and every branch. If a branch snaps violently (wild ramification), your view gets blurry.
The Tame Site (This Paper): The authors built a special window with a filter. This filter blocks out the violent snapping branches. It only lets you see the parts of the forest that grow smoothly.
The Pair $(X, \tilde{X})$ : To make this work, they don't just look at the forest ( $X$ ); they look at the forest inside a larger, well-behaved garden ( $\tilde{X}$ ). Think of $X$ as a messy room and $\tilde{X}$ as the whole house. By looking at the room relative to the house, they can define what "tame" means more precisely.

2. The $\beta$ -Construction: The "Taming Machine"

This is the paper's main invention. Imagine you have a raw, chaotic dataset (an "étale sheaf"). You want to turn it into a "tame sheaf" that works with your new Tame Site.

The Input: You feed the machine a standard mathematical object and a set of rules called $\beta$ . These rules say, "If you encounter a specific type of chaotic edge (a valuation ring), here is how you should behave to stay calm."
The Output: The machine spits out a new, "tamed" version of the object. This new object ignores the wild chaos and only cares about the smooth, predictable parts.
Why it matters: This allows mathematicians to take old, familiar tools and adapt them to work in difficult, positive-characteristic environments where they previously failed.

3. The Result: A New Way to Measure "Holes"

Using this new Tame Site and the Taming Machine, the authors successfully reconstructed the de Rham–Witt complex.

The Analogy: Think of de Rham–Witt cohomology as a way to count the number of holes in a donut, a coffee mug, or a pretzel.
The Achievement: In the past, trying to count these holes in "positive characteristic" (the chaotic environment) gave wrong answers because the wild ramification messed up the count.
The Fix: By using the Tame Site, they proved that you can count these holes correctly. They showed that their new "Tame de Rham–Witt cohomology" matches up perfectly with other known, reliable ways of measuring these shapes (like "log-crystalline cohomology").

4. The Comparison: Tame vs. Wild

The paper also compares their new "Tame Site" with an older, similar concept called the "Tame Site of [HS20]."

The Metaphor: Imagine two different maps of the same city.
- Map A (Old Tame Site): Shows the city but loses some details when you zoom in on specific neighborhoods.
- Map B (New Tame Site): Shows the city with more detail, specifically designed to keep the "tame" information intact even when you zoom in.
The Finding: The authors show that while both maps are related, the New Tame Site (Map B) contains more information. If you try to translate a map from the New Site to the Old Site, you lose data. It's like compressing a high-definition video into a low-resolution one; the smooth parts look okay, but you lose the nuance that the new method was designed to preserve.

Summary of What They Claim

They built a new mathematical framework (the Tame Site) that filters out "wild" chaos in algebraic geometry.
They created a general method (the $\beta$ -construction) to turn any standard mathematical object into a "tame" one that works in this new framework.
They applied this to de Rham–Witt cohomology, proving that it works correctly in positive characteristic when viewed through this new lens.
They showed that this new method recovers known results (like cycle maps) and connects to other theories (like syntomic cohomology) in a way that was previously difficult or impossible.

In short: They found a way to smooth out the rough edges of a chaotic mathematical world, allowing them to use powerful tools to measure shapes and holes that were previously too messy to measure accurately.

Technical Summary: A Construction of Tame Sheaves and Tame de Rham–Witt Cohomology

Problem and Motivation
The paper addresses a fundamental limitation in the construction of integral $p$ -adic cohomology theories for schemes over a field $k$ of positive characteristic $p$ . While de Rham cohomology with logarithmic poles provides a well-behaved theory in characteristic zero satisfying étale descent, the analogous construction using the de Rham–Witt complex with log poles in characteristic $p$ fails to satisfy étale descent. This failure is attributed to wild ramification. Previous work by Hübner and Saito ([HS20]) established that these theories satisfy "tame descent" relative to a specific tame site. However, the existing framework relies on adic spaces, making the construction of sheaves technically difficult.

The authors aim to introduce an algebraic variant of the tame topology for pairs of schemes $(X, \tilde{X})$ , where $X \hookrightarrow \tilde{X}$ is a quasi-compact open immersion. The goal is to establish foundational results for this site that enable the systematic construction of tame sheaves and cohomology theories satisfying tame descent, specifically for the de Rham–Witt complex and reciprocity sheaves.

Methodology
The core methodology involves defining a new site, the tame site $(X, \tilde{X})_t$ , and developing a general machinery to construct sheaves on it.

The Tame Site: The underlying category consists of pairs $(U, \tilde{U})$ where $U \hookrightarrow \tilde{U}$ is a quasi-compact open immersion over a base pair $(S, \tilde{S})$ , with specific finiteness conditions (ift morphisms). The topology is generated by tame coverings, defined via the behavior of valuations on the adic space $Spa(U, \tilde{U})$ . A morphism is a tame covering if, for every valuation in the target, there exists a lift in the source where the induced extension of valuation fields is tame.
The $\beta$ -Construction: The authors introduce a general mechanism (Proposition 8.2) to construct a tame sheaf $F_\beta$ on $(X, \tilde{X})_t$ from an étale sheaf $F$ on $X_{\text{ét}}$ equipped with a "local datum" $\beta$ . This datum specifies, for every valuation ring centered on $\tilde{X}$ , a subset of "tame sections" of $F$ . The resulting sheaf $F_\beta$ consists of sections of $F$ that, when pulled back to the henselization of the valuation ring, lie within the specified tame subsets.
Cohomological Computations: The paper establishes a comparison theorem (Theorem 7.2) relating the cohomology of a tame sheaf on $(X, \tilde{X})_t$ to the direct limit of étale cohomology groups over admissible blow-ups of $\tilde{X}$ . This relies on the existence of acyclic objects (pro-objects that are local for the topology) and a comparison between Čech and sheaf cohomology (Theorem 6.5).

Key Contributions and Results

Foundational Theory of the Tame Site: The authors define the tame topos and prove that the cohomology of a tame sheaf is independent of the choice of compactification $\tilde{X}$ (Lemma 2.7). They establish that the cohomology can be computed as a colimit over modifications (Theorem 7.2), provided the sheaf satisfies a specific condition regarding its stalks being $\mathbb{Z}(p_x)$ -modules.
Tame de Rham–Witt Complex: Applying the $\beta$ $β$ -construction to the de Rham–Witt complex with log poles, the authors define the tame big de Rham–Witt complex $W_S\Omega^\bullet_t$ $W_{S} Ω_{t}^{∙}$ (Definition 9.5). They prove that for a smooth scheme $\tilde{X}$ $\tilde{X}$ with a simple normal crossing divisor $D = \tilde{X} \setminus X$ $D = \tilde{X} ∖ X$ , the cohomology of this tame complex on $(X, \tilde{X})_t$ $(X, \tilde{X})_{t}$ is isomorphic to the logarithmic crystalline cohomology of $(\tilde{X}, D)$ $(\tilde{X}, D)$ (Theorem 10.3).
- Specifically, $R\Gamma((X, \tilde{X})_t, W\Omega^\bullet_t) \cong R\Gamma_{\text{log-crys}}((\tilde{X}, D)/W(k))$ .
Tame Syntomic Cohomology: The authors define tame syntomic complexes $Z_p(r)_t$ using the Nygaard filtration on the tame de Rham–Witt complex. They show that these complexes recover the tame motivic cohomology considered in previous literature (Theorem 10.8), establishing an isomorphism $Z/p^\bullet(r)_t \cong W^\bullet\Omega^r_{t, \log}[-r]$ .
Reciprocity Sheaves: The construction is extended to reciprocity sheaves (Theorem 11.3). For a reciprocity sheaf $F$ , the associated tame sheaf $F^{\text{rec}}_{\log}$ has cohomology isomorphic to the Nisnevich cohomology of the maximal cube-invariant modulus sheaf on the compactification.
Comparison with Previous Tame Sites: The paper compares the new site $(X, \tilde{X})_t$ with the tame site $(X/S)_t$ defined in [HS20]. While there is an adjoint pair of functors between the categories of sheaves, the cohomology theories are generally distinct. The authors demonstrate that passing from $(X, \tilde{X})_t$ to $(X/S)_t$ results in a loss of information; for instance, the cohomology of the tame structure sheaf $O_t$ on $(X, \tilde{X})_t$ computes the cohomology of the structure sheaf on the compactification, whereas on $(X/S)_t$ it computes the cohomology of a constant sheaf related to étale cohomology (Remark 12.6).

Significance and Claims
The paper claims to provide a robust, algebraic framework for constructing integral models of $p$ -adic cohomology theories that satisfy tame descent. By shifting from adic spaces to schemes, the authors simplify the construction of sheaves, making it accessible to standard algebraic geometry techniques.

The primary significance lies in the ability to:

Systematically generate tame sheaves from étale data via the $\beta$ -construction.
Prove that the de Rham–Witt complex with log poles satisfies tame descent, thereby recovering logarithmic crystalline cohomology in a purely algebraic manner.
Establish a bridge between tame syntomic cohomology and motivic cohomology.
Clarify the relationship between different tame topologies, showing that the pair-based site $(X, \tilde{X})_t$ retains more geometric information (specifically regarding the compactification) than the relative site $(X/S)_t$ .

The authors note that these results rely on resolution of singularities or specific dimension bounds (e.g., $d \leq 3$ in positive characteristic) for certain isomorphisms, such as the identification of cohomology with logarithmic crystalline cohomology. The work sets the stage for future constructions of integral models for coherent and de Rham cohomology over non-archimedean fields.