K-stability for varieties with a big anticanonical class

Basic Information

• Paper ID: 2210.16631
• Title: K-stability for varieties with a big anticanonical class
• Author: Chenyang Xu (Princeton University)
• Classification: math.AG (Algebraic Geometry), math.DG (Differential Geometry)
• Published Journal: Épijournal de Géométrie Algébrique, Special volume in honour of C. Voisin, Article No. 7 (2023)
• Paper Link: https://arxiv.org/abs/2210.16631

Abstract

This paper extends the theory of algebraic K-stability to projective klt pairs with a big anticanonical class. Although such pairs may exhibit pathological behavior in general, the paper observes that the K-semistability condition forces them to possess a klt anticanonical model whose stability properties coincide with those of the original pair.

Research Background and Motivation

Problem Background

1. Success of K-stability theory: Significant progress has been made in the algebraic K-stability theory for log Fano pairs, which provides an algebraic approach to studying the existence of Kähler-Einstein metrics.
2. Need for extension: Recent work DZ22 and DR22 has considered the Kähler-Einstein problem for Kähler manifolds $(X,ω)$ where $-K_X$ is big, particularly DZ22 proved a transcendental Yau-Tian-Donaldson theorem for twisted big Kähler-Einstein metrics.
3. Pathological phenomena: For projective varieties $X$ with big anticanonical class $-K_X$, pathological examples may exist, such as the anticanonical ring $R(X,-K_X) = \bigoplus_{m \in \mathbb{N}} H^0(X,-mK_X)$ not necessarily being finitely generated.

Research Motivation

This paper aims to demonstrate that for projective klt pairs with a big anticanonical class, K-stability theory essentially reduces to the original (log) Fano case, meaning that K-stability conditions force these pairs to possess log Fano-type properties.

Core Contributions

1. Established the connection between K-semistability and log Fano type: Proved that if $(X,\Delta)$ is a klt projective pair with big $-K_X-\Delta$ and $\delta(X,\Delta) \geq 1$, then there exists an effective Q-divisor $\Gamma$ such that $(X,\Delta+\Gamma)$ is a log Fano pair.
2. Proved finite generation: Under K-semistability conditions, the anticanonical ring $R(X,-r(K_X+\Delta))$ is finitely generated.
3. Established equivalence of stability: Proved that $(X,\Delta)$ and its anticanonical model $(Z,\Delta_Z)$ possess the same K-stability properties.
4. Unified stability concepts: Proved that for such pairs, uniform K-stability is equivalent to K-stability.

Methodology Details

Problem Formulation

Study K-stability of projective klt pairs $(X,\Delta)$ with a big anticanonical class, where $-K_X-\Delta$ is big but not necessarily ample.

Core Technical Framework

Definition of S-invariant

For an $n$-dimensional projective normal pair $(X,\Delta)$ such that $-K_X-\Delta$ is big, for any prime divisor $E$ appearing on a birational model $\mu: Y \to X$, the S-invariant is defined as:

$S_{X,\Delta}(E) := \frac{1}{\text{vol}(-K_X-\Delta)} \int_0^{\infty} \text{vol}(-\mu^*(K_X+\Delta)-tE) dt$

δ-invariant

If $(X,\Delta)$ is klt, define: $\delta(X,\Delta) := \inf_E \frac{A_{X,\Delta}(E)}{S_{X,\Delta}(E)}$

where $E$ ranges over all valuations on $(X,\Delta)$, and $A_{X,\Delta}(E)$ is the log discrepancy.

Key parameter $a(X,\Delta)$

Define the constant $a(X,\Delta)$ as: $a(X,\Delta) = \sup\{t \in \mathbb{R} : \exists \text{ ample divisor } A \text{ such that } A-t(K_X+\Delta) \text{ is ample and } -K_X-\Delta-A \text{ is pseudoeffective}\}$

Proof Strategy for Main Theorems

Proof Strategy for Theorem 1.1

1. Direct case: When $\delta(X,\Delta) > 1$, directly apply Lemma 3.1 via the log canonical threshold of the base divisor.
2. Perturbation argument: When $\delta(X,\Delta) \leq 1$ but condition (3.2) is satisfied, use an ample divisor $A$ for perturbation and construct appropriate Q-complements to obtain log Fano properties.

Core of Theorem 1.2's Proof

Using a common resolution $Y$, establish the relationship: $\pi^*(K_Z+\Delta_Z) - \mu^*(K_X+\Delta) = B \geq 0$

Prove that for any prime divisor $E$:

• $A_{X,\Delta}(E) = A_{Z,\Delta_Z}(E) + \text{ord}_E(B)$
• $S_{X,\Delta}(E) = S_{Z,\Delta_Z}(E) + \text{ord}_E(B)$

Experimental Setup

Theoretical Verification

This is primarily theoretical work, with results verified through rigorous mathematical proofs.

Construction of Counterexamples

Example 3.8: Constructs a concrete counterexample illustrating pathological phenomena:

• Let $S$ be the blowup of $\mathbb{P}^2$ at nine very general points
• $-K_S$ is nef but not semiample
• Construct $X = \mathbb{P}_S(\mathcal{E})$, where $\mathcal{E} := \mathcal{O}_S + \mathcal{O}_S(H)$
• Prove that $-K_X$ is big but the corresponding algebra is not finitely generated
• Direct computation verifies $\delta(X) < \frac{3}{5} < 1$

Experimental Results

Main Results

Theorem 3.4 (Finite Generation)

Let $(X,\Delta)$ satisfy Assumption 3.3, namely: $\delta(X,\Delta) > \frac{n+1}{n+1+a_0}$ where $a_0 = a(X,\Delta)$. Then $(X,\Delta)$ is of log Fano type, and in particular, any Cartier divisor $E$ satisfies that $R(X,E) := \bigoplus_{m \in \mathbb{N}} H^0(X,mE)$ is finitely generated.

Corollary 3.5 (Properties of Anticanonical Model)

Under the conditions of Theorem 3.4, the anticanonical model $(Z,\Delta_Z)$ is a log Fano pair.

Stability Equivalence Results

Theorem 1.2: $(X,\Delta)$ is K-semistable (resp. K-stable, uniformly K-stable) if and only if $(Z,\Delta_Z)$ is K-semistable (resp. K-stable, uniformly K-stable).

Technical Results

Lemma 2.6: If $A$ is an effective ample Q-divisor on $X$ such that $-K_X-\Delta-A$ is pseudoeffective, then $S_{X,\Delta}(A) \geq \frac{1}{n+1}$.

Related Work

Development of K-stability Theory

1. Classical theory: The Fujita-Li criterion Fuj19, Li17, BX19 established the equivalence between K-stability defined using valuations and the original definition via test configurations.
2. Big anticanonical class case: DZ22 formulated the current definition for the case of big anticanonical class.
3. Ding stability: DR22 developed the concept of Ding stability for projective klt pairs with big $-K_X-\Delta$.

Birational Geometry Techniques

Utilizes finite generation results from BCH+10 and standard techniques from the minimal model program.

Conclusions and Discussion

Main Conclusions

1. K-semistability conditions force klt pairs with big anticanonical class to possess log Fano-type properties
2. The K-stability problem for such pairs reduces to the corresponding problem for their anticanonical models
3. For such pairs, uniform K-stability is equivalent to K-stability

Theoretical Significance

The results in this paper demonstrate that, despite the potential pathological behavior of pairs with big anticanonical class, K-stability conditions provide sufficient constraints to ensure good geometric properties.

Application Prospects

These results provide a theoretical foundation for applying existing birational geometry techniques to study K-stability problems in more general settings.

In-Depth Evaluation

Strengths

1. Theoretical depth: Successfully extends K-stability theory to more general geometric settings
2. Technical innovation: Cleverly reduces pathological cases to the classical log Fano case
3. Completeness: Provides a complete theoretical framework, including finite generation and stability equivalence
4. Rigor: Mathematical arguments are rigorous and logically clear

Technical Contributions

1. Introduction of parameter $a(X,\Delta)$: Provides an effective tool for quantifying the degree of "bigness"
2. Perturbation techniques: The perturbation arguments used in the proofs have general value
3. Unified framework: Unifies different stability concepts within a single framework

Limitations

1. Restrictive conditions: Main results require relatively strong hypotheses
2. Concrete computations: Verification of conditions may be difficult for specific examples
3. Scope of applications: Primarily theoretical results; practical applications require further development

Impact

This paper makes important contributions to the development of K-stability theory, particularly by providing theoretical foundations for handling more general geometric objects. It is expected to have sustained impact on the fields of algebraic geometry and complex geometry.

References

• BJ20 H. Blum and M. Jonsson, Thresholds, valuations, and K-stability
• DZ22 T. Darvas and K. Zhang, Twisted Kähler-Einstein metrics in big classes
• DR22 R. Dervan and R. Reboulet, Ding stability and Kähler-Einstein metrics on manifolds with big anticanonical class
• LXZ22 Y. Liu, C. Xu, and Z. Zhuang, Finite generation for valuations computing stability thresholds and applications to K-stability