On the localization theorem for F-pure rings

This paper resolves Grothendieck's localization problem for a specific class of rings from tight closure theory by utilizing a detailed analysis of the relative Frobenius map.

The Gist

The Big Picture: The "Smoothness" Traveling Problem

Imagine you are an architect designing a building. You have a blueprint (a mathematical ring) that represents a perfect, smooth structure. You then decide to build a new wing attached to it (a new ring) using a specific method called a "flat map."

In mathematics, a flat map is like a very careful, non-distorting construction. It ensures that the new wing doesn't accidentally warp the old building. However, there's a catch: even if the connection is perfect, the new parts of the building might develop cracks or weird shapes (singularities) that weren't there before.

The Core Question (Grothendieck's Localization Problem):
If the "base" of your new wing (the closed fiber) is perfect and smooth, and the "foundation" you are building on (the original ring) is also perfect, does that guarantee that every single part of the new wing, all the way up to the roof, is also perfect?

Usually, the answer is "No." Bad shapes in the foundation can sometimes "infect" the upper floors. But this paper asks: Are there special types of buildings where the answer is YES?

The Special Building: "F-Pure" Rings

The authors focus on a specific class of mathematical structures called F-Pure rings.

• The Analogy: Think of an F-Pure ring as a building made of a special, self-healing material. If you poke a hole in it, the material naturally snaps back into place. It resists "badness."
• The Context: These rings come from a field of math called "Tight Closure Theory," which is used to solve deep problems in algebra and geometry.

The Magic Tool: The Radu-Andrè Morphism

To prove their point, the authors use a special tool called the Radu-Andrè morphism.

• The Analogy: Imagine you want to check if a new building is stable. You can't just look at it; you need to take a "snapshot" of it under a special microscope that reveals its internal structure.
• In this paper, the Radu-Andrè morphism is that microscope. It takes the original ring and the new ring and creates a "hybrid" structure. By studying this hybrid, the authors can see if the "self-healing" property (F-purity) spreads from the bottom of the building to the top.

The Main Discovery

The authors prove a powerful theorem (Theorem 3.10):

If you start with a perfect foundation, and you build a new wing where the base is "self-healing" (F-pure) and "strong" (Gorenstein), then the entire new wing, no matter where you look, will also be "self-healing."

In other words, the "goodness" of the base doesn't just stay at the bottom; it travels up and protects the whole structure.

Why This Matters (The Geometric Consequences)

The paper doesn't just stop at the math; it shows how this helps us understand shapes in geometry (Theorem 4.4).

• The "Open Set" Concept: Imagine you have a map of a country. You want to know where the soil is fertile (good for farming).
• The Result: The authors show that if the soil is fertile in one specific spot (the closed fiber), then the fertile soil isn't just a tiny isolated island. Instead, it forms a large, continuous open region.
• The Metaphor: If you find a patch of perfect, self-healing soil, you don't have to check every single grain of sand around it. You can be sure that a whole neighborhood around it is also perfect. This makes it much easier for mathematicians to predict where "good" shapes exist and where "bad" shapes might appear.

Summary of the Journey

1. The Problem: Does a perfect base guarantee a perfect whole in a flat construction? Usually, no.
2. The Special Case: But if the materials are "F-Pure" (self-healing), maybe yes?
3. The Method: The authors used a "microscope" (Radu-Andrè morphism) to look at the connection between the base and the new structure.
4. The Proof: They showed that the self-healing property does travel up the structure.
5. The Result: This means that in these special mathematical worlds, if you have a good start, you are guaranteed a good finish, and this "goodness" covers large, continuous areas, not just tiny spots.

In a nutshell: This paper solves a long-standing puzzle about how "goodness" spreads in mathematical structures, proving that for a specific, important class of rings, a strong foundation guarantees a strong building everywhere.

Technical Summary

1. Problem Statement: Grothendieck's Localization Problem

The paper addresses a specific instance of Grothendieck's Localization Problem, which concerns the behavior of fiber rings under flat local homomorphisms.

• The Setup: Let $\phi: R \to S$ be a flat local homomorphism of Noetherian rings.
• The Question: If the closed fiber ($S/\mathfrak{m}S$) and all formal fibers of $R$ satisfy a specific ring-theoretic property $P$, does every fiber of $\phi$ (i.e., $S \otimes_R k(\mathfrak{p})$ for all $\mathfrak{p} \in \text{Spec } R$) also satisfy $P$?
• Context: Previous work by Avramov and Foxby solved this for properties like Cohen-Macaulay, Gorenstein, and Complete Intersection using numerical invariants (defects).
• The Gap: This paper focuses on properties arising from tight closure theory in characteristic $p > 0$, specifically:
  1. Maximal Cohen-Macaulay (MCM) modules.
  2. F-purity (and geometrically F-pure rings).
• Specific Constraint: The authors restrict their study to F-finite rings. This is crucial because F-finite rings are excellent, meaning their formal fibers are geometrically regular, simplifying the problem by removing the need to check formal fiber conditions explicitly.

2. Methodology and Key Tools

The authors employ a combination of tight closure theory, homological algebra, and a specific morphism introduced by Radu and Andrè.

A. The Radu-Andrè Morphism

The central technical tool is the Radu-Andrè morphism ($w_{S/R}^e$).

• Definition: For a homomorphism $\phi: R \to S$ in characteristic $p$, the $e$-th Radu-Andrè ring is $W_{S/R}^e = S \otimes_R {}^eR$, where ${}^eR$ is $R$ viewed as an $R$-module via the $e$-th iterate of the Frobenius map.
• The Map: $w_{S/R}^e: S \otimes_R {}^eR \to {}^eS$ is defined by $s \otimes r \mapsto s^p \phi(r)$.
• Significance: Theorem 3.2 states that $\phi$ is a regular homomorphism if and only if $w_{S/R}^1$ is flat. This allows the authors to study fiber properties using the structure of these tensor products, bypassing limitations of standard cohomological arguments.

B. Trivial Extensions and Depth Analysis

To handle the Maximal Cohen-Macaulay (MCM) property, the authors utilize trivial extensions (idealization).

• They construct a ring $S * N$ (where $N$ is an $S$-module) to relate the depth of a module to the depth of a ring.
• Key Lemma (Proposition 3.7): They prove that if the closed fiber is Cohen-Macaulay and the formal fibers of $R$ are Cohen-Macaulay, then the property of being MCM propagates to all fibers for flat maps. This serves as a stepping stone for the F-purity results.

C. Splitting of Frobenius

The characterization of F-purity relies on the splitting of the Frobenius morphism.

• A ring is F-pure if the Frobenius map splits.
• The authors use the fact that for F-finite rings, F-purity is equivalent to the Frobenius map being a pure extension.
• They utilize Lemma 3.9 to handle field extensions, showing that F-purity is preserved under finite separable extensions and reduced to purely inseparable extensions via the Radu-Andrè morphism.

3. Key Contributions and Results

A. Localization for MCM Modules (Proposition 3.7)

The authors establish that for a flat local map $\phi: R \to S$ of Noetherian rings:

• If formal fibers of $R$ are Cohen-Macaulay and the closed fiber $S/\mathfrak{m}S$ is Cohen-Macaulay, then for any $R$-flat finite $S$-module $N$, the fiber $N \otimes_R k(\mathfrak{p})$ is MCM over $S \otimes_R k(\mathfrak{p})$ for all $\mathfrak{p}$.
• Technique: Uses Nagata's trivialization to reduce the module problem to a ring problem, then applies Avramov-Foxby's results (Theorem 3.5).

B. The Main Theorem: Localization for Geometrically F-Pure Rings (Theorem 3.10)

This is the paper's primary contribution.

• Hypothesis: Let $\phi: R \to S$ be a flat local map of F-finite rings. Assume the closed fiber $S/\mathfrak{m}S$ is Gorenstein and geometrically F-pure over $k_R$.
• Conclusion: For every $\mathfrak{p} \in \text{Spec } R$, the fiber $S \otimes_R k(\mathfrak{p})$ is geometrically F-pure over $k(\mathfrak{p})$.
• Proof Strategy:
  1. Establish that fibers are Gorenstein and geometrically reduced (using Theorem 3.4 and 3.5).
  2. Reduce the problem to purely inseparable field extensions using the structure of field extensions in characteristic $p$.
  3. Construct the Radu-Andrè ring $W_q^e$ associated with the fiber.
  4. Show that the cokernel of the Radu-Andrè map is MCM over the Radu-Andrè ring (using the MCM localization result from Prop 3.7).
  5. Use local duality and the splitting of the Frobenius on the closed fiber to prove that the Frobenius splits on the generic fiber, thereby proving F-purity.

C. Geometric Consequences (Section 4)

The authors translate these algebraic localization results into geometric statements regarding the openness of certain loci in the spectrum.

• Generic Principle I (Proposition 4.2): For a flat map of finite type between excellent rings, the set of points $\mathfrak{p} \in \text{Spec } R$ where the fiber is MCM is a Zariski open subset.
• Generic Principle II (Theorem 4.4): For a flat map of finite type between F-finite rings, if all fibers are Gorenstein, the set of points $\mathfrak{p}$ where the fiber is geometrically F-pure is a Zariski open subset of $\text{Spec } R$.
  • This confirms that geometric F-purity is a "generic" property in families of F-finite rings.

4. Significance and Impact

1. Solving a Specific Case of Grothendieck's Problem: The paper successfully solves the localization problem for the class of F-pure rings, a property central to tight closure theory and singularity theory in characteristic $p$.
2. Methodological Innovation: The paper demonstrates the power of the Radu-Andrè morphism in investigating fibers of flat maps. It provides a concrete alternative to cohomological methods, which often fail or are difficult to apply in the context of tight closure and F-singularities.
3. Geometric Applications: By proving that geometric F-purity forms an open set in the base scheme, the authors provide a foundational tool for studying families of varieties in positive characteristic. This is analogous to results known for smoothness or Gorenstein properties but was previously unresolved for F-purity in this generality.
4. Bridging Tight Closure and Commutative Algebra: The work integrates deep concepts from tight closure (F-purity, F-injectivity) with classical commutative algebra (flatness, Cohen-Macaulayness, dualizing complexes), showing that these "tight closure" properties behave well under flat base change, similar to classical geometric properties.

In summary, Shimomoto and Zhang provide a rigorous proof that geometric F-purity is preserved and open in flat families of F-finite rings, utilizing the Radu-Andrè morphism to overcome technical obstacles in positive characteristic commutative algebra.