Sharp Thresholds for $ε$-Adjoint Singularities of Foliated Surfaces

This paper establishes that the stability of $\epsilon$-adjoint log canonical singularities on foliated surfaces undergoes sharp thresholds at $\epsilon=1/5$ and $\epsilon=1/4$, which are proven to be optimal through a complete classification of these singularities for $\epsilon \in (0,1/3)$ in terms of negative definite exceptional configurations.

The Gist

Imagine you are an architect designing a city, but instead of buildings, you are designing foliated surfaces. In math terms, a "foliation" is like a pattern of flowing water or wind lines drawn across a surface. Sometimes, these lines get messy, tangled, or crash into each other at specific points. These messy points are called singularities.

In the world of geometry, we love to classify these messy points. We ask: "Is this messiness 'acceptable' (log canonical) or is it 'too messy' (bad)?"

This paper by Shi Xu is about a very specific, tricky experiment: What happens if we gently nudge our rules for what counts as "acceptable" messiness?

The Core Idea: The "Adjoint" Nudge

Usually, mathematicians have a strict rule for what makes a singularity "good." But sometimes, that rule is too rigid. So, they introduce a knob (represented by the Greek letter $\epsilon$) that mixes two different rules together.

Think of it like adjusting the seasoning in a soup:

• Rule A (The Foliation Rule): "The flow lines must behave nicely."
• Rule B (The Surface Rule): "The ground beneath the flow must be smooth."

The "Adjoint" rule is a recipe: $1 \times$ Rule A + $\epsilon \times$ Rule B.

• When $\epsilon$ is tiny (close to 0), we mostly care about the flow lines.
• As we turn the knob and increase $\epsilon$, we start caring more about the ground beneath.

The big question the author asks is: "If I turn this knob just a tiny bit, does the soup still taste good? Or does it suddenly become inedible?"

The Discovery: Two Critical "Tipping Points"

The author discovered that the answer isn't a smooth slide. Instead, there are sharp cliffs (thresholds) where the rules suddenly change. It's like walking on a frozen lake; you can walk safely for a long time, but at a specific spot, the ice suddenly cracks.

1. The First Cliff: $\epsilon = 1/5$ (The "Boundary" Break)

• The Safe Zone: If you keep the knob below $1/5$ (0.2), everything is safe. Any singularity that passes the "mixed" test also passes the strict "flow line" test. The system is stable.
• The Break: The moment you hit exactly $1/5$, a new, weird configuration appears. Imagine a specific type of knot in the flow lines that was previously forbidden. At $1/5$, this knot suddenly becomes "legal" under the mixed rules, even though it's still "illegal" under the strict flow rules.
• The Metaphor: It's like a bouncer at a club. Below $1/5$, if you don't have a VIP pass (strict flow rules), you can't get in. At $1/5$, the bouncer suddenly lets in a specific group of people who don't have the VIP pass but have a special "mixed" pass. The rules of the club have fundamentally changed.

2. The Second Cliff: $\epsilon = 1/4$ (The "Canonical" Break)

• The Safe Zone: If we demand an even higher standard (called "canonical," which is stricter than just "log canonical"), the safe zone shrinks. You can only turn the knob up to $1/4$ (0.25).
• The Break: At $1/4$, another specific, very messy knot becomes legal under the mixed rules, even though it violates the strictest standards.
• The Metaphor: This is like a stricter bouncer. He lets you in up to $1/4$, but at $1/4$, he suddenly allows a very rowdy group in that he usually kicks out.

Why Does This Matter? (The "Why Should I Care?")

You might wonder, "Who cares about these specific numbers like 1/5 or 1/4?"

1. Stability in Math: In mathematics, we want our theories to be robust. If a tiny change in a parameter (like turning a knob slightly) causes the whole structure to collapse or change category, the theory is "unstable." This paper proves that for these specific geometric surfaces, the theory is rock solid up to $1/5$ (or $1/4$), but then it hits a wall. Knowing exactly where the wall is helps mathematicians build better theories without accidentally falling off the cliff.
2. Building New Worlds (The MMP): Mathematicians use a process called the "Minimal Model Program" (MMP) to simplify complex shapes into their most basic forms. This paper shows that if you use the "mixed" rules (the adjoint rules), you can successfully simplify these surfaces all the way up to the $1/4$ mark. Beyond that, the simplification process might break down. This gives us a clear "speed limit" for how far we can go with these new tools.

The "Detective Work"

How did the author find these numbers?
He didn't just guess. He acted like a detective looking at negative definite configurations.

• Imagine a pile of tangled wires (the "exceptional divisors").
• He classified every possible way these wires could tangle.
• He found that most tangles are "safe" (they follow the rules).
• But he found a few "boundary" tangles that are just barely safe.
• By doing complex algebra (solving systems of inequalities), he calculated exactly how much "mixing" ($\epsilon$) it takes for those barely-safe tangles to become legal.
• The math revealed that the first one becomes legal at exactly 0.2 ($1/5$) and the second at 0.25 ($1/4$).

Summary in One Sentence

This paper proves that when we mix two different geometric rules for surface singularities, the system remains stable and predictable up to a precise tipping point of 20% (or 25% for stricter rules), but beyond that, strange and previously forbidden geometric shapes suddenly become allowed, changing the fundamental nature of the geometry.

Technical Summary

1. Problem Statement

The paper investigates the stability of singularities in the context of the Minimal Model Program (MMP) for foliated surfaces. Specifically, it addresses the behavior of $\epsilon$-adjoint singularities, defined via the divisor $K_{\mathcal{F}} + \epsilon K_X$ (where $K_{\mathcal{F}}$ is the canonical divisor of the foliation $\mathcal{F}$, $K_X$ is the canonical divisor of the surface $X$, and $\epsilon > 0$ is a small parameter).

• Core Question: Does the property of being "log canonical" (lc) or "canonical" for a foliated singularity remain stable under small perturbations of $\epsilon$?
• Motivation: While the intrinsic birational geometry of foliations and the ambient geometry of the surface are linked via adjoint divisors, it is known that the canonical condition is not always stable under small variations of $\epsilon$. The paper seeks to identify the precise sharp thresholds (critical values of $\epsilon$) where the classification of admissible singularities changes, specifically where new, "boundary" configurations enter the admissible region.

2. Methodology

The author employs a combination of birational geometry, intersection theory, and combinatorial classification of negative definite configurations.

• Resolution and Discrepancy Analysis: The study focuses on the minimal resolution $\pi: (X, \mathcal{F}) \to (Y, \mathcal{G})$ of a foliated surface germ. The condition of being $\epsilon$-adjoint log canonical is translated into a system of linear inequalities involving the discrepancies $a_i$ of the exceptional divisors $E_i$.
• Linear Deformation: The variation of $\epsilon$ is interpreted as a linear deformation of these discrepancy inequalities. As $\epsilon$ increases, the polyhedral region of admissible configurations shrinks. The "critical values" occur when an inequality becomes sharp, allowing new configurations to enter the admissible set.
• Classification of Configurations: The core technical work involves a complete classification of the exceptional divisors $E = \sum E_i$ for $\epsilon \in (0, 1/3)$. This is achieved by:
  • Analyzing negative definite configurations of curves.
  • Defining special $K$-chains (linear configurations where the first curve has negative intersection with the adjoint divisor, and subsequent curves have non-negative intersection).
  • Using numerical reduction and determinant bounds (via Hirzebruch-Jung strings and intersection matrices) to limit the possible combinatorial types of the dual graphs.
  • Applying the Separatrix Theorem to rule out impossible geometric configurations.
• Case Distinction: The classification is split into:
  1. Foliated Canonical Configurations: Invariant curves forming chains, forks, or elliptic leaves.
  2. Foliated lc (Non-canonical) Configurations: Involving non-invariant curves (star graphs/chains).
  3. Boundary Configurations: Specific configurations that are $\epsilon$-adjoint lc but not foliated lc.

3. Key Contributions and Results

A. Complete Classification for $\epsilon \in (0, 1/3)$

The paper provides a full classification of $\epsilon$-adjoint log canonical singularities (Theorem 4.15). The exceptional divisors fall into three structural types:

1. Type I (Foliated Canonical): Standard foliated canonical configurations (F-chains, F-dihedral forks, elliptic Gorenstein leaves).
2. Type II (Foliated lc, Non-canonical): Configurations involving a smooth non-invariant curve (F-star chains/forks) attached to F-chains.
3. Type III (Boundary Configurations): Specific configurations involving a smooth rational F-invariant curve with $Z(\mathcal{F}, C)=3$ connecting two F-chains of type $(2,3)$.

B. Sharp Thresholds

The most significant contribution is the identification of two sharp stability thresholds:

1. The Log Canonical Threshold ($\epsilon = 1/5$):

   • Result: For $\epsilon \in (0, 1/5)$, every $\epsilon$-adjoint log canonical singularity is foliated log canonical.
   • Transition: At $\epsilon = 1/5$, the Type III (Boundary) configurations enter the admissible region. These configurations are $\epsilon$-adjoint lc but fail to be foliated lc.
   • Significance: This proves that $\epsilon = 1/5$ is the sharp first stability threshold. Below this value, the adjoint condition implies the intrinsic foliated condition.

2. The Canonical Threshold ($\epsilon = 1/4$):

   • Result: For $\epsilon \in (0, 1/4)$, every $\epsilon$-adjoint canonical singularity is foliated log canonical (and the surface singularity is klt).
   • Transition: At $\epsilon = 1/4$, the boundary between foliated canonical and non-canonical behavior shifts. Specifically, Type II configurations (specifically Type II-2-a, F-star forks) cease to be $\epsilon$-adjoint canonical at this threshold.
   • Significance: $\epsilon = 1/4$ is the maximal range where the adjoint canonical condition preserves the expected singularity behavior (foliated lc and surface klt).

C. Applications to the Adjoint MMP

The paper establishes an Adjoint Minimal Model Program for the range $\epsilon \in (0, 1/4)$ (Proposition 1.5).

• For a smooth projective surface $X$ with a rank-one foliation $\mathcal{F}$ having canonical singularities, there exists a birational morphism $\phi: X \to Y$ such that:
  • $Y$ has klt singularities and $\mathcal{G} = \phi^*\mathcal{F}$ has log canonical singularities.
  • Either $K_{\mathcal{G}} + \epsilon K_Y$ is nef, or there is a Mori fiber space structure where $-(K_{\mathcal{G}} + \epsilon K_Y)$ is ample.
• This extends previous results (which were limited to $\epsilon < 1/5$) to the optimal range $\epsilon < 1/4$.

D. Gap Phenomena

The results confirm a "1-gap" conjecture for interpolated lc thresholds in dimension 2. The interpolated threshold $t_0$ (where $t K_{\mathcal{F}} + (1-t)K_X$ is lc) satisfies either $t_0 = 1$ or $t_0 \leq 5/6$. The constant $\tau(2) = 1/6$ is shown to be sharp.

4. Significance

• Optimality: The thresholds $1/5$ and $1/4$ are proven to be optimal (sharp). The paper constructs explicit examples (Examples 6.1 and 6.3) demonstrating that for any $\epsilon$ slightly larger than these values, the stability of the singularity types breaks down.
• Structural Insight: The work clarifies the precise geometric configurations (specifically the "boundary configurations" involving curves with $Z(\mathcal{F}, C)=3$) that cause the failure of stability.
• Foundation for Higher Dimensions: By solving the problem completely in dimension 2 and identifying the combinatorial mechanisms (negative definite configurations and linear inequalities), the paper provides a blueprint for investigating similar threshold phenomena in higher-dimensional foliated pairs.
• Unification: It unifies the study of foliated singularities with the ambient surface geometry, showing how the parameter $\epsilon$ interpolates between the two and where the "pathologies" of foliation theory (non-finite generation of canonical rings) can be controlled.

In summary, Shi Xu's paper resolves the stability question for adjoint singularities on surfaces by providing a complete combinatorial classification and identifying the exact numerical thresholds ($1/5$ and $1/4$) that govern the transition between intrinsic foliated properties and adjoint properties.