Pure subrings of Du Bois singularities are Du Bois singularities

This paper establishes that Du Bois singularities descend under cyclically pure maps of Noetherian $\mathbb{Q}$-algebras, a result that extends to log canonical singularities in the complex setting and yields new insights in prime and mixed characteristics.

The Gist

Imagine you are an architect inspecting a building. In the world of mathematics, specifically algebraic geometry, "buildings" are shapes defined by equations, and "flaws" in these shapes are called singularities. Some flaws are minor (like a small scratch), while others are catastrophic (like a collapsed foundation).

Mathematicians have a special category of "acceptable flaws" called Du Bois singularities. Think of these as "safe" cracks. A building with Du Bois singularities is still structurally sound enough to be studied and understood, even if it isn't perfectly smooth.

This paper, written by Charles Godfrey and Takumi Murayama, tackles a fundamental question: If you have a "safe" building, and you find a smaller, hidden room inside it that is built using the same blueprints, is that hidden room also safe?

Here is the breakdown of their discovery using everyday analogies.

1. The Setup: The "Pure" Connection

The authors are looking at a relationship between two rings (mathematical structures that act like blueprints for buildings), let's call them Ring R (the small room) and Ring S (the big building).

They are connected by a specific type of map called a cyclically pure map.

• The Analogy: Imagine Ring S is a massive, high-quality warehouse. Ring R is a small, locked vault inside that warehouse.
• The "Pure" Rule: The connection is "pure" if the vault (R) is so tightly integrated into the warehouse (S) that you can't cheat. If you take a pile of items (an ideal) from the vault, put them in the warehouse, and then try to take them back out, you get exactly the same pile you started with. Nothing is lost, nothing is added, and the vault's contents are perfectly preserved within the warehouse.

2. The Big Question

The authors ask: If the big warehouse (S) has only "safe" cracks (Du Bois singularities), does the small vault (R) inside it also have "safe" cracks?

In the past, mathematicians knew this was true for some very specific types of connections (like if the vault was just a perfect copy of a section of the warehouse). But they didn't know if it held true for all "pure" connections, even if the vault was just a weird, twisted subset of the warehouse.

3. The Discovery: The "Inheritance" Theorem

The authors prove that Yes, the safety is inherited.

If the big structure (S) is Du Bois, then the smaller structure (R) inside it is also Du Bois.

• Why is this surprising? Usually, when you take a piece out of a whole, you might break it. If you take a slice of a cake, the slice is fine. But if you take a slice of a cake that has a specific type of structural integrity, you might think the slice loses that property. The authors proved that in this specific mathematical world, the "safety" property is robust enough to survive being a sub-part of a larger whole.

4. The Secret Weapon: The "H-Topology"

To prove this, the authors had to invent a new way of looking at the buildings.

• The Old Way: Traditionally, to check if a building is "Du Bois," you had to try to smooth it out completely (resolve the singularities). This is like trying to fix a broken vase by melting it down and recasting it. It works, but it's hard to do for complex, abstract shapes.
• The New Way (H-Topology): The authors used a concept called the h-topology.
  • The Analogy: Imagine you want to check the structural integrity of a building. Instead of just looking at the building itself, you imagine covering it with a giant, flexible, transparent net made of many different shapes (other buildings) that fit over it perfectly.
  • This "net" (the h-topology) allows you to see the building's flaws from every possible angle simultaneously. The authors showed that if the big building passes the test under this "net," the small vault inside it must also pass.

5. Why This Matters (The Real-World Impact)

The paper isn't just about abstract math; it solves a puzzle that has been bothering mathematicians for decades.

• The "Boutot" Connection: There was a famous theorem by Boutot that said if a big building has "rational singularities" (a very strict type of safety), a smaller pure part of it also has them. This paper proves the same thing for "Du Bois singularities," which are a broader, more flexible category.
• Log Canonical Singularities: The authors also applied their result to "Log Canonical" singularities. These are like "Code Red" safety warnings. They showed that if a big building is "Log Canonical" and the smaller part is "pure," the smaller part is also "Log Canonical." This helps engineers (mathematicians) classify which shapes are safe to work with in complex calculations.

Summary

Think of the universe of shapes as a giant library of books.

• Du Bois is a "Seal of Approval" for books that are slightly damaged but still readable.
• Cyclically Pure is a special binding that ensures a chapter (Ring R) is perfectly integrated into the whole book (Ring S).
• The Paper's Conclusion: If the whole book has the "Seal of Approval," then any chapter bound by this special method automatically gets the seal too. You don't need to re-read the whole book to know the chapter is safe; the safety of the whole guarantees the safety of the part.

This discovery gives mathematicians a powerful new tool to prove that complex, messy shapes are actually well-behaved, simply by looking at the bigger, well-behaved shapes they are built from.

Technical Summary

1. Problem Statement and Context

The paper addresses a fundamental question in commutative algebra and algebraic geometry regarding the descent of singularity properties under cyclically pure ring maps.

• The Core Question: If $R \to S$ is a cyclically pure map of Noetherian rings (meaning $IS \cap R = I$ for every ideal $I \subseteq R$), and $S$ possesses a specific type of singularity, does $R$ necessarily possess the same type of singularity?
• Historical Context:
  • Boutot's Theorem (1987): Established that if $R \to S$ is a map of finite type over a field of characteristic zero and $S$ has rational singularities, then $R$ has rational singularities.
  • Characteristic $p$ Counterexamples: In prime characteristic, the analogue of rational singularities is F-injectivity. It was known that F-injectivity does not generally descend under cyclically pure maps, even if the map splits.
  • Du Bois Singularities: These are a characteristic-zero analogue of F-injectivity (and rational singularities). While it was known that Du Bois singularities descend under split maps or faithfully flat maps in specific settings, the general case for cyclically pure maps remained open, particularly for rings that are not necessarily of finite type over a field.

The Main Problem: Does the property of having Du Bois singularities descend from $S$ to $R$ when $R \to S$ is a cyclically pure map of Noetherian $\mathbb{Q}$-algebras?

2. Methodology

The authors employ a sophisticated blend of modern algebraic geometry techniques, moving beyond classical resolution of singularities to utilize Grothendieck topologies and limit theorems.

A. Characteristic-Free Definition via Grothendieck Topologies

Instead of relying on the classical definition of Du Bois singularities (which requires resolutions of singularities and the Deligne–Du Bois complex), the authors adopt a definition due to Huber and Kelly based on sheafification with respect to the $h$-topology (and related topologies like $cdh$, $eh$, $sdh$).

• They define a pair $(X, \Sigma)$ to have Du Bois singularities if the natural map from the ideal sheaf $I_\Sigma$ to the derived pushforward of the structure sheaf under the $h$-topology, $\Omega^0_{X, \Sigma, h}$, is a quasi-isomorphism.
• This approach allows them to work with arbitrary Noetherian $\mathbb{Q}$-algebras, not just varieties of finite type.

B. The Key Injectivity Theorem (Generalized)

A central technical tool is a generalization of the Kovács–Schwede Key Injectivity Theorem.

• Classical Result: For complex varieties, if a point $x$ is such that the punctured neighborhood is Du Bois, the map on local cohomology $H^i_x(X, \mathcal{O}_X) \to H^i_x(X, \Omega^0_X)$ is surjective.
• Authors' Contribution: They prove a version of this theorem for separated Noetherian schemes of equal characteristic zero with isolated non-Du Bois points.
• Technique: To prove this without assuming the scheme is of finite type over a field, they construct an absolute canonical compactification $\langle X \rangle_{cpt}$ using Zariski–Riemann spaces (inverse limits of blowups).
  • They utilize Grothendieck's limit theorem for inverse limits of topoi to relate the cohomology of the limit space to the cohomology of the finite-type approximations.
  • They use the degeneration of the Hodge-to-de Rham spectral sequence on the compactified space to establish surjectivity, which is then pulled back to the local setting.

C. Descent via Injectivity

The authors establish that for a map $f: Y \to X$, if $Y$ is Du Bois and the map induces injective maps on local cohomology (a condition satisfied by cyclically pure maps), then $X$ is Du Bois. This shifts the problem from splitting conditions (which are strong) to injectivity conditions (which are weaker and satisfied by pure maps).

3. Key Contributions and Results

Theorem A (Main Result)

Let $R \to S$ be a cyclically pure map of Noetherian $\mathbb{Q}$-algebras. If $S$ has Du Bois singularities with respect to the $h$-topology, then $R$ has Du Bois singularities with respect to the $h$-topology.

• Significance: This is new even when $R \to S$ is faithfully flat. It extends Boutot's theorem from rational singularities to Du Bois singularities in a much broader context (Noetherian $\mathbb{Q}$-algebras).

Theorem 4.3 (Generalization to Pairs and Various Conditions)

The authors prove that Du Bois singularities descend under a wide variety of conditions on the morphism $f: Y \to X$, provided $Y$ is Du Bois:

1. Faithfully flat maps.
2. Pure maps where the map on local rings is pure at maximal points.
3. Partially pure maps (existence of a pure point in the fiber).
4. Affine maps where the module map on sections is pure.
   This generalizes previous results by Kovács and Schwede, which required the map to split in the derived category.

Corollary B (Log Canonical Singularities)

As a consequence, they address a question by Zhuang regarding log canonical (lc) type singularities:

• If $R \to S$ is a cyclically pure map of essentially finite type $\mathbb{C}$-algebras, $S$ has log canonical type singularities, and $K_R$ is Cartier, then $R$ has log canonical singularities.
• This provides a positive answer to the descent question for log canonical singularities under cyclically pure maps, a result previously unknown in this generality.

Theorem C (Key Injectivity Theorem for Schemes)

For a separated Noetherian scheme $X$ of equal characteristic zero, if $x \in X$ is a point such that $\text{Spec}(\mathcal{O}_{X,x}) \setminus \{x\}$ is Du Bois, then the natural map:
$$H^i_x(\text{Spec}(\mathcal{O}_{X,x}), \mathcal{O}_{X,x}) \longrightarrow H^i_x(\text{Spec}(\mathcal{O}_{X,x}), \Omega^0_{X,h,x})$$
is surjective.

• This theorem is a crucial intermediate step that links the geometric property (Du Bois) to the cohomological property (injectivity/surjectivity), enabling the descent proofs.

4. Significance and Impact

1. Unification of Characteristic Zero and Mixed Characteristic: The paper provides a framework (using Grothendieck topologies) that works uniformly across different characteristics, recovering known results in prime characteristic (descent of F-injectivity under specific pure maps) while solving the characteristic zero case for Du Bois singularities.
2. Relaxation of Splitting Conditions: Previous descent theorems often required the map to split (i.e., $S$ is a direct summand of $R$). This paper shows that cyclically pure maps (a weaker condition) are sufficient for Du Bois singularities, significantly broadening the scope of applicable morphisms.
3. New Tools for Non-Finite Type Schemes: By utilizing Zariski–Riemann spaces and limit theorems, the authors successfully extend results that were previously restricted to varieties of finite type over a field to general Noetherian $\mathbb{Q}$-algebras.
4. Resolution of Open Questions: The paper positively answers Question 1.1 for Du Bois singularities and provides a partial positive answer for log canonical singularities (under the Cartier condition), advancing the understanding of how singularities behave under pure subring inclusions.

In summary, this paper establishes that the class of Du Bois singularities is robust under cyclically pure maps, utilizing a novel combination of sheaf-theoretic definitions on Grothendieck topologies and advanced limit techniques to bypass the need for global resolutions of singularities.