🔢 mathematics

Recent progress on the Minimal Model Program for foliations

This paper surveys recent advancements in the birational geometry of foliations on complex varieties, with a specific focus on singularities, adjunction, and their applications to the Minimal Model Program for surfaces and the existence of flips on threefolds.

Original authors: Paolo Cascini, Calum Spicer

Published 2026-04-13

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Paolo Cascini, Calum Spicer

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 a landscape architect trying to redesign a vast, overgrown garden. Your goal is to tidy it up, remove the weeds, and create a clean, efficient layout that reveals the garden's true structure. In the world of mathematics, this garden is a complex variety (a shape defined by equations), and the "weeds" are the messy, tangled parts of the design.

This paper, written by Paolo Cascini and Calum Spicer, is a guidebook for a new kind of gardening project: The Minimal Model Program (MMP) for Foliations.

Here is the breakdown of what they are doing, using simple analogies.

1. What is a "Foliation"?

In a normal garden, you might look at the whole plot. But a foliation is like looking at the garden through a specific lens: the direction the plants are growing.

Imagine a field of wheat. The wheat stalks are all leaning in a specific direction. If you trace a path following the wind blowing through the wheat, you are following a "leaf" of the foliation.
In math, a foliation is a way of slicing a shape into layers (like pages in a book) or tracing paths (like leaves on a tree).
The authors want to know: If we start with a messy, tangled set of paths, can we "clean up" the shape so the paths become simple and organized?

2. The Problem: Blowing Up Doesn't Always Help

In standard geometry, if you have a messy point, you can often "blow it up" (zoom in and replace the point with a smooth surface) to fix it.

The Twist: With foliations, blowing up a messy point often makes the paths more tangled. It's like trying to untangle a knot by pulling on the ends; sometimes you just make a bigger knot.
The Goal: Instead of making everything perfectly smooth (which is impossible), the goal is to reduce the mess to a "distinguished class" of manageable problems. Think of it as sorting the weeds into specific, known categories rather than trying to remove every single one.

3. The Toolkit: The "Adjunction" Formula

The paper's secret weapon is a tool called Adjunction.

The Analogy: Imagine you have a complex 3D sculpture (the variety) and you slice a piece off it (a divisor). The "Adjunction" formula is a rule that tells you how the "messiness" (singularities) of the whole sculpture relates to the messiness of the slice you just cut.
Why it matters: It allows the mathematicians to take a big, hard problem and break it down into smaller, 2D problems. If they understand the slice, they can understand the whole. This formula acts as a bridge, translating the rules of the big shape to the rules of the smaller piece.

4. The Journey: The Minimal Model Program (MMP)

The MMP is a step-by-step process to simplify a shape.

Find a Problem Spot: Look for a path (curve) where the "energy" (mathematically, the canonical class $K_F$ ) is negative. This is like finding a patch of garden that is growing in the wrong direction.
Contract or Flip:
- Contraction: If the bad patch is small and isolated, you squash it down to a point.
- The Flip (The Big Challenge): In 3D gardens, sometimes you can't just squash the bad patch. You have to perform a "flip." Imagine a bridge that is collapsing. Instead of removing it, you magically swap the supports so the bridge is now stable but looks different.
- The Breakthrough: The authors explain that for 3D foliations, they can now prove these "flips" always exist. They do this by finding a special "invariant" path (a separatrix) that acts like a safety rail, allowing them to perform the swap without breaking the math.

5. The Surprises: When Things Go Wrong

In standard geometry, if you keep simplifying, you eventually reach a "Minimal Model"—a perfect, finished garden.

The Foliations Twist: With foliations, sometimes you don't reach a perfect finish.
The "Elliptic Gorenstein Leaf": The authors found that sometimes, the garden settles into a loop of rational curves (a ring of paths) that behaves like an "elliptic leaf."
The Consequence: In these specific cases, the "recipe book" (the ring of sections) for the garden is infinite. You can never write down a finite list of instructions to describe the whole thing. This is a major difference from standard geometry, where the recipe is always finite.

6. The Big Picture: Why Does This Matter?

This paper is a survey of recent progress. It tells us:

We can simplify 2D foliations: We have a complete map for surfaces (like a flat sheet of paper).
We can flip 3D foliations: We have proven that the "bridge swap" (flip) works in 3D, which was a huge missing piece of the puzzle.
We have new mysteries: We found that sometimes the "recipe" is infinite, and we are still figuring out the rules for higher dimensions (4D and beyond).

Summary Metaphor

Think of the authors as explorers mapping a new continent.

The Terrain: Complex shapes with flowing paths (foliations).
The Map: The Minimal Model Program.
The Compass: The Adjunction formula (telling them how the terrain changes when they zoom in).
The Discovery: They found that while they can smooth out most of the terrain, there are some "swirling vortexes" (elliptic leaves) that can't be flattened, and the rules for navigating them are different from the rest of the world.

They have successfully built a bridge (the flip) to cross the 3D chasm, but they warn us that the landscape is stranger and more complex than we originally thought.

1. Problem Statement and Context

The paper addresses the extension of the Minimal Model Program (MMP) to the setting of foliations on complex varieties. In classical algebraic geometry, the MMP aims to simplify a variety $X$ via birational modifications (contractions and flips) to reach a "minimal model" where the canonical divisor $K_X$ is nef, or a "Mori fiber space."

For foliations $\mathcal{F}$ of rank $r$ on a normal variety $X$ , the guiding principle is to replace the canonical divisor $K_X$ with the canonical divisor of the foliation, $K_{\mathcal{F}}$ , defined via the dual of the top exterior power of the tangent sheaf of the foliation ( $\omega_{\mathcal{F}} = (\wedge^r T_{\mathcal{F}})^*$ ).

Core Challenges:

Singularities: Unlike varieties, foliations cannot generally be resolved to smooth ones via blow-ups; blowing up often creates new singularities. The goal is instead to reduce singularities to a "distinguished class" (e.g., canonical or log canonical).
Failure of Standard Theorems: The Base Point Free theorem, crucial for constructing flips in the classical MMP, does not hold in the same generality for foliations.
New Phenomena: The behavior of $K_{\mathcal{F}}$ differs from $K_X$ in that rational curves with $K_{\mathcal{F}} \cdot C < 0$ may not exist in the same way, and the "Bend and Break" lemma requires adaptation.

The paper aims to survey recent progress in establishing an MMP for foliations, focusing on singularities, adjunction, surface foliations, and the existence of flips in dimension three.

2. Methodology and Technical Framework

The authors develop a framework that adapts classical MMP tools to foliations, relying heavily on the concept of Adjunction for Foliations.

A. Definitions and Preliminaries

Foliations: Defined as saturated, Lie-bracket-closed subsheaves $T_{\mathcal{F}} \subset T_X$ .
Singularities: Defined via discrepancy $a(E, \mathcal{F}, \Delta)$ . A pair $(\mathcal{F}, \Delta)$ is log canonical if $a(E, \mathcal{F}, \Delta) \geq -1$ for all divisors $E$ .
Pfaff Fields: A morphism $\phi: \Omega^{[r]}_X \to \mathcal{O}_X(K_{\mathcal{F}})$ used to define the singular locus $\text{Sing } \mathcal{F}$ even on singular varieties.
Invariant Subvarieties: A subvariety $W$ is invariant if the foliation is tangent to it. The indicator function $\epsilon(W)$ is 0 if invariant, 1 otherwise.

B. Adjunction for Foliations (The Central Tool)

The paper establishes a Foliated Adjunction Formula (Theorem 4.1). For a codimension-one subvariety $D$ with normalization $\nu: W \to D$ :
$(K_{\mathcal{F}} + \epsilon(D)D)|_W \sim K_{\mathcal{F}_W} + \text{Diff}_D(\mathcal{F})$
where:

$\mathcal{F}_W$ is the restricted foliation on $W$ .
$\text{Diff}_D(\mathcal{F})$ is the different, an effective $\mathbb{Q}$ -divisor correcting the discrepancy.
Significance: This formula allows translating numerical negativity on $X$ into restrictions on subvarieties and comparing singularities between $X$ and $D$ . It is the primary engine for proving the existence of contractions and flips.

C. Singularities Analysis

Rank 1 Foliations: The authors utilize the Jordan decomposition of vector fields ( $\partial = \partial_S + \partial_N$ ). They prove that a rank 1 foliation is log canonical if and only if the linear part of its generating vector field is non-nilpotent (Proposition 3.9).
Co-rank 1 Foliations: The distinguished class of singularities is simple singularities (Definition 3.20), characterized by specific formal normal forms.
Separatrices: A key result is that canonical singularities on threefolds admit separatrices (invariant divisors). This allows "gluing" formal invariant divisors along curves, a crucial step for constructing flips.

3. Key Contributions and Results

A. MMP for Surface Foliations (Section 5)

The authors provide a streamlined proof of McQuillan's Minimal Model Program for surfaces using adjunction.

Contraction Theorem: If $K_{\mathcal{F}} \cdot C < 0$ for a curve $C$ , then $C$ is $\mathcal{F}$ -invariant. This allows the construction of a contraction $\phi: X \to Z$ to a point (Proposition 5.2).
Termination: The process terminates in a model where $K_{\mathcal{F}}$ is nef or a fibration in rational curves.
General Type & Finite Generation: For surfaces of general type ( $K_{\mathcal{F}}$ $K_{F}$ big), the authors analyze the structure of the minimal model. They identify elliptic Gorenstein leaves (cycles of rational curves with specific intersection properties).
- Result: The ring of sections $R(X, K_{\mathcal{F}})$ is not always finitely generated (Theorem 5.11), contrasting with the classical MMP for varieties. This failure is linked to the existence of elliptic Gorenstein leaves.

B. Existence of Flips in Dimension Three (Section 6)

The most significant recent progress highlighted is the proof of the existence of flips for foliations on threefolds.

The Obstacle: The Base Point Free theorem fails, so standard flip construction methods do not apply.
The Solution: The authors construct flips by finding an F-invariant divisor $G \geq 0$ $G \geq 0$ such that the flipping curve $C$ $C$ is a log canonical center of the pair $(X, G)$ $(X, G)$ .
- Rank 1 Case: Using the Reeb Stability Theorem and adjunction, they show that near the flipping curve, the foliation admits a meromorphic first integral (Proposition 6.3). This allows the construction of $G$ and the subsequent $(K_X + G)$ -flip, which coincides with the $K_{\mathcal{F}}$ -flip (Theorem 6.4).
- Co-rank 1 Case: The construction relies on the existence of separatrices for canonical singularities. By taking the union of separatrices meeting the flipping curve and extending them (using Artin/Elkik approximation or formal MMP techniques), they construct the required invariant divisor $G$ .
Main Theorem: The $K_{\mathcal{F}}$ -flip exists for rank 1 foliations with simple singularities and co-rank 1 foliations with canonical singularities on $\mathbb{Q}$ -factorial klt threefolds.

4. Significance and Open Problems

Significance:

Foundational Step: This work establishes the existence of flips, a critical missing piece for running the full MMP in dimension 3 and higher for foliations.
New Geometry: It highlights that foliations introduce genuinely new geometric phenomena (e.g., non-finite generation of canonical rings, the necessity of separatrices) that do not appear in the classical variety setting.
Unification: The adjunction formula unifies the treatment of singularities and provides a robust mechanism for transferring MMP techniques from varieties to foliations.

Open Questions Raised:

Non-dicritical Singularities: What is the correct definition of "non-dicritical" for foliations of arbitrary rank? (Currently well-understood for rank 1 and co-rank 1).
Finite Generation: Is there a structural characterization in higher dimensions for when $R(X, K_{\mathcal{F}})$ fails to be finitely generated?
Independence of Flips: Can the auxiliary divisor $G$ used to construct flips be chosen independently of the specific flipping ray?
Resolution of Singularities: While reduction to simple singularities is known in dimension $\leq 3$ , a general reduction theory for higher dimensions remains elusive.

Conclusion

Cascini and Spicer demonstrate that while the MMP for foliations shares deep structural similarities with the classical MMP (via adjunction and discrepancy), it requires novel tools to handle the specific nature of foliation singularities. The proof of flip existence in dimension three marks a major milestone, opening the door to a comprehensive classification theory for foliated threefolds.