🔢 mathematics

Global smoothing of singular Fano and Calabi-Yau varieties

This paper establishes global smoothing criteria for Fano and Calabi-Yau varieties with isolated Du Bois lci singularities by demonstrating that such varieties deform to those with milder singularities or become smooth under specific Hodge-Du Bois number conditions, thereby generalizing and extending prior results on threefolds and rational hypersurface singularities to higher dimensions.

Original authors: Anda Tenie

Published 2026-05-07

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Anda Tenie

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 have a beautiful, complex sculpture made of clay. In the world of mathematics, these sculptures are called varieties. Sometimes, these sculptures are perfect and smooth, but other times, they have rough spots, sharp corners, or even holes where the clay didn't quite form right. These rough spots are called singularities.

This paper is about a specific type of mathematical sculpture:

Fano varieties: Think of these as sculptures that naturally want to curve inward, like a bowl or a sphere.
Calabi-Yau varieties: Think of these as sculptures that are perfectly balanced, like a flat sheet or a donut, with a very specific kind of "weight" (mathematically, their curvature is zero).

The author, Anda Tenie, is asking a big question: If our sculpture has rough spots, can we gently reshape the clay to make it perfectly smooth without breaking it?

In math terms, this is called "smoothing." It's not just about fixing one spot; it's about finding a way to fix all the spots at once using a single, continuous motion.

The Problem: Local vs. Global

Imagine you are a sculptor. You look at one rough corner and think, "I can fix this!" You look at another corner and think, "I can fix that one too!"

The problem is that just because you can fix each corner individually doesn't mean you can fix them all together. Sometimes, the instructions for fixing corner A conflict with the instructions for fixing corner B. If you try to fix both, the whole sculpture might collapse or warp in a weird way.

This paper provides a set of rules (criteria) to tell us when we can successfully fix the whole sculpture at once.

The New Tools: "1-Rational" and "1-Du Bois"

To understand the rules, the author uses a new way of measuring how "rough" a spot is. Think of it like a grading system for the quality of the clay:

The "1-Du Bois" Grade: This is a measure of how "mathematically nice" a rough spot is. Some spots are very messy; others are just slightly bumpy.
The "1-Rational" Grade: This is an even stricter standard. If a spot is "1-Rational," it's so well-behaved that it's almost as good as being smooth.
The "1-Liminal" Grade: These are the spots that are "nice enough" (1-Du Bois) but not "perfect enough" (1-Rational). They are the tricky middle ground.

The Main Discoveries

1. The "Fano" Sculptures (The Bowls)

For the bowl-shaped sculptures (Fano varieties), the author found a surprisingly simple rule:

The Rule: If the rough spots are not "1-Rational" (meaning they are messy enough), you can almost always smooth the whole sculpture out.
The Magic: You don't need any special global conditions. If the spots are messy enough, the universe of math forces a way to fix them all together. It's like saying, "If the clay is sticky enough, it will naturally mold into a smooth shape."

2. The "Calabi-Yau" Sculptures (The Balanced Shapes)

For the balanced sculptures (Calabi-Yau varieties), it's a bit more complicated, but the author found two paths:

Path A: The Messy Spots
If none of the rough spots are "1-Du Bois" (meaning they are all very messy), then no special rules are needed. Just like the Fano case, the messiness guarantees you can smooth the whole thing out.
Path B: The Tricky Spots (1-Liminal)
If you have those tricky "middle-ground" spots (1-Liminal), you can't just rely on the spots themselves. You need a Global Check.
- The Check: The author looks at the "Hodge-Du Bois numbers." Imagine these as a scorecard that counts the different types of holes and curves in the sculpture.
- The Condition: If two specific numbers on this scorecard are equal (specifically, $h_{n-1,2} = h_{1,n-2}$ ), then the sculpture can be smoothed.
- The Analogy: Think of this like balancing a scale. If the weight on the left side of the sculpture matches the weight on the right side (in a very specific mathematical way), the sculpture is stable enough to be smoothed out. If they don't match, the sculpture might break when you try to fix it.

Why This Matters

Before this paper, mathematicians knew how to smooth these sculptures in 3 dimensions (like our everyday world). But when you go to 4, 5, or 10 dimensions, the rules get very confusing.

This paper takes the old rules for 3D and upgrades them for any number of dimensions. It also removes some of the "technical assumptions" that previous mathematicians thought were necessary.

Previous thought: "You need a very specific type of rough spot to fix it."
This paper says: "Actually, as long as the spots aren't too perfect (1-Rational), or if the global scorecard balances out, you can fix it."

Summary

In simple terms, this paper is a repair manual for high-dimensional mathematical sculptures. It tells us:

If the damage is "messy enough," we can fix it automatically.
If the damage is "tricky," we need to check a global balance sheet (the Hodge numbers) to see if a fix is possible.
These rules work for all dimensions, not just the 3D world we are used to.

The author proves that with the right conditions, even the most broken, singular mathematical shapes can be gently coaxed into becoming perfectly smooth.

Technical Summary: Global Smoothing of Singular Fano and Calabi-Yau Varieties

Problem Statement
This paper addresses the problem of smoothing singular Fano and Calabi-Yau varieties $Y$ of dimension $n \geq 3$ with isolated Du Bois local complete intersection (lci) singularities. While local deformations of isolated lci singularities are unobstructed and admit smoothing directions, a global smoothing requires that these local directions can be lifted to a single global deformation. The central challenge is that even when global deformations are unobstructed, there is no guarantee that local smoothing directions "glue" to form a global deformation that smooths all singular points simultaneously. The paper investigates conditions under which such global smoothings exist, extending previous work restricted to threefolds or rational singularities to higher dimensions and the broader class of Du Bois lci singularities.

Methodology
The approach relies on comparing global and local deformation spaces.

Deformation Spaces: Global first-order deformations are parametrized by $T^1_Y = \text{Ext}^1(\Omega^1_Y, \mathcal{O}_Y)$ , while local deformations at a singular point $x$ are parametrized by $T^1_{Y,x}$ . The natural restriction map $\eta: T^1_Y \to \bigoplus_{x \in \text{Sing}(Y)} T^1_{Y,x}$ determines whether local smoothings can be globalized.
Hodge-Theoretic Tools: The paper utilizes the Du Bois complex $\Omega^\bullet_Y$ and its associated Hodge-Du Bois numbers $h^{p,q}(Y) = \dim H^q(Y, \Omega^p_Y)$ . It employs the theory of $k$ -rational and $k$ -Du Bois singularities (specifically $k=1$ ) to classify singularities.
Local-to-Global Lifting: The core technical mechanism involves analyzing exact sequences derived from local cohomology. For a strong log resolution $\pi: \tilde{Y} \to Y$ with exceptional divisor $E$ , the authors relate $T^1_{Y,x}$ to local cohomology groups such as $H^2_E(\Omega^{n-1}_{\tilde{X}_x}(\log E))$ and $H^2_E(\Omega^{n-1}_{\tilde{X}_x}(\log E)(-E))$ .
Vanishing Theorems: The proofs leverage Nakano-type vanishing theorems for Du Bois complexes and Serre duality to show that certain global cohomology groups vanish (e.g., $H^2(\Omega^{n-1}_{\tilde{Y}}(\log E)) = 0$ for Fano varieties). This vanishing ensures the surjectivity of specific maps in the deformation diagrams, allowing local smoothing directions to be lifted globally.

Key Contributions and Results

Unobstructedness for Fano Varieties: The paper proves that for a Fano variety $Y$ with isolated lci singularities, the obstruction space $T^2_Y$ vanishes. Consequently, deformations of $Y$ are unobstructed. This result holds even when singularities are Du Bois but not rational (0-liminal), extending previous results known for rational singularities.
Smoothing Fano Varieties (Theorem 1.5):
- Any singular Fano variety with isolated Du Bois lci singularities can be deformed to a Fano variety with only 1-rational singularities.
- If none of the singularities of $Y$ are 1-rational, then $Y$ is smoothable, and every small smoothing is Fano.
- This generalizes results for Fano threefolds by Namikawa and Friedman-Laza to higher dimensions and the Du Bois lci setting.
Smoothing Calabi-Yau Varieties (Theorems 1.3 and 1.4):
- Case 1 (No 1-Du Bois singularities): If $Y$ is a singular Calabi-Yau variety where no singularities are 1-Du Bois, $Y$ is smoothable without any additional global conditions. Every small smoothing is Calabi-Yau.
- Case 2 (Allowing 1-liminal singularities): If $Y$ contains 1-liminal singularities (1-Du Bois but not 1-rational), a global condition is required. The paper establishes that if the Hodge-Du Bois numbers satisfy the equality $h^{n-1,2}(Y) = h^{1,n-2}(Y)$ , then $Y$ can be deformed to a Calabi-Yau with 1-rational singularities. If, furthermore, no singularities are 1-rational, $Y$ is smoothable.
Generalization of Friedman-Laza Results: The paper provides an alternative smoothing criterion for Calabi-Yau varieties with 1-liminal singularities. While Friedman-Laza previously established a condition involving the vanishing of a specific element in a cohomology group (Theorem 1.10), this paper offers a condition based on the vanishing of a morphism related to Hodge-Du Bois numbers (Theorem 1.4 and Theorem 1.11). This new condition is presented as a natural higher-dimensional generalization of the Q-factoriality condition used for threefolds.

Significance and Relation to Prior Work
The paper claims to recover and generalize seminal results for threefolds by Friedman, Namikawa, Namikawa-Steenbrink, and Gross.

Threefold Recovery: In dimension three, 1-rational singularities correspond to smooth points, and 1-liminal singularities correspond to Ordinary Double Points (ODPs). The authors' results recover the Namikawa-Steenbrink theorem that Q-factoriality (equivalent to $h^{2,2}=h^{1,1}$ for rational singularities) guarantees smoothability for Calabi-Yau threefolds.
Higher Dimensions: In dimensions $n \geq 4$ , prior results on smoothing Calabi-Yau varieties were largely restricted to rational hypersurface singularities. This work extends the theory to the lci setting and to Du Bois singularities that are not rational.
Technical Advancement: The paper removes technical assumptions found in previous work (such as the splitting of specific short exact sequences required in Friedman-Laza's earlier theorems) and extends the scope from hypersurface singularities to general lci singularities.
New Unobstructedness: The proof of unobstructedness for Fano varieties with isolated lci singularities (including 0-liminal cases) is presented as a new result, distinct from previous approaches that relied on $T^1$ -lifting or log Calabi-Yau contexts.

The authors emphasize that their global criterion for Calabi-Yau varieties ( $h^{n-1,2} = h^{1,n-2}$ ) serves as a natural replacement for Q-factoriality in higher dimensions, linking the smoothing problem directly to the Poincaré duality properties of the singular variety's Hodge-Du Bois numbers.