The cotangent bundle of K3 surfaces of degree two

Basic Information

• Paper ID: 2207.09294
• Title: The cotangent bundle of K3 surfaces of degree two
• Authors: Fabrizio Anella, Andreas Höring
• Classification: math.AG (Algebraic Geometry)
• Published Journal: Épijournal de Géométrie Algébrique, Special volume in honour of C. Voisin, Article No. 3 (2023)
• Paper Link: https://arxiv.org/abs/2207.09294

Abstract

K3 surfaces have been extensively studied from multiple perspectives, yet the positivity properties of their cotangent bundles remain insufficiently understood. This paper explores the surprisingly rich geometric structure of the projectivization of the cotangent bundle $\mathbb{P}(\Omega_S)$ of a general polarized K3 surface $S$ of degree two. In particular, the authors describe the geometric properties of the surface $D_S \subset \mathbb{P}(\Omega_S)$, which plays a role analogous to that of bitangent planes to smooth quartic surfaces in $\mathbb{P}^3$.

Research Background and Motivation

1. Core Problem: The positivity properties of the cotangent bundle $\Omega_S$ of K3 surfaces are insufficiently understood. Although $\Omega_S$ is known to be stable for every polarization, these stability results do not fully characterize the positivity properties of $\Omega_S$.
2. Problem Significance:
   • The cotangent bundle of K3 surfaces is never pseudoeffective, necessitating new methods to measure its "negativity"
   • The description of the pseudoeffective cone of the projectivized cotangent bundle $\mathbb{P}(\Omega_S)$ is crucial for understanding K3 surface geometry
   • This research helps establish connections between pseudoeffectivity and families of elliptic curves
3. Limitations of Existing Methods:
   • The embedding $\mathbb{P}(\Omega_S) \subset S^{[2]}$ approach of Gounelas and Ottem can only determine that $\zeta_S + \pi^*2L$ is pseudoeffective, but cannot determine its extremality in $\text{Pseff}(\mathbb{P}(\Omega_S))$
   • For degree two K3 surfaces, $\alpha_S := 2L$ satisfies $\alpha_S^2 = 8$, precisely corresponding to the critical case of the optimal bound
4. Research Motivation: To find an analogue of the theory of bitangent planes to smooth quartic surfaces in $\mathbb{P}^3$, establishing a corresponding geometric framework for degree two K3 surfaces.

Core Contributions

1. Characterization of the Main Geometric Object: Construction and complete description of the surface $D_S \subset \mathbb{P}(\Omega_S)$, which is dominated by canonical lifts of singular elliptic curves in $|L|$, whose normalization is a smooth (non-minimal) elliptic surface.
2. Precise Numerical Equivalence: Proof that $D_S \equiv 30\zeta_S + 54\pi^*L \equiv 30(\zeta_S + 1.8\pi^*L)$.
3. Boundary Estimates of the Pseudoeffective Cone:
   • Proof of the existence of a prime divisor $Z_S \subset \mathbb{P}(\Omega_S)$ such that $Z_S \equiv a(\zeta_S + \lambda\pi^*L)$ with $\lambda \leq 1.7952024$
   • Establishment of the lower bound $\lambda \geq \frac{39}{22} = 1.772$
4. Complete Description of Birational Geometry: Through analysis of birational morphisms, transfer of information from the more understandable $\mathbb{P}(f^*\Omega_{\mathbb{P}^2})$ to the mysterious $\mathbb{P}(\Omega_S)$.

Detailed Methodology

Task Definition

Study the geometric properties of the projectivized cotangent bundle $\mathbb{P}(\Omega_S)$ of a general polarized K3 surface $(S,L)$ of degree two, where $S$ is obtained via a double cover $f: S \to \mathbb{P}^2$ with branch divisor a smooth degree 10 curve.

Model Architecture

1. Basic Setup

• $f: S \to \mathbb{P}^2$: double cover, defining the linear system $|L|$
• $B \subset \mathbb{P}^2$: branch locus (sextic curve), $g(B) = 10$
• $R \subset S$: ramification divisor, $R \in |3L|$

2. Birational Geometry Framework

Construction of the key birational morphism diagram:

Y ----µS----> P(ΩS)
 \           /
  µP        /
   \       /
    \     /
P(f*ΩP²) 

where $Y$ is obtained by blowing up along curves $R_P$ and $R_S$.

3. Elliptic Fibration Structure

• Normalization $\bar{D} \to q^{-1}(R^∨)$ yields a smooth elliptic fibration $\bar{q}: \bar{D} \to R$
• Contains 648 singular fibers of Kodaira type I₁ (nodal cubic curves)
• Further normalization $\tilde{D} \to D$ is the blowup of $\bar{D}$ at 720 points

Technical Innovations

1. Base Change Technique: Utilization of the base change $\mathbb{P}(f^*\Omega_{\mathbb{P}^2}) \dashrightarrow \mathbb{P}(\Omega_S)$ via the exact sequence: $0 \to f^*\Omega_{\mathbb{P}^2} \to \Omega_S \to \Omega_f \to 0$
2. Canonical Lifting: For an irreducible curve $C \subset S$, definition of the canonical lift $C_S \subset \mathbb{P}(\Omega_S)$ as the image of the morphism $\tilde{C} \to \mathbb{P}(\Omega_S)$ corresponding to the line bundle $Q_C$.
3. Double Cover Analysis: Through analysis of the double cover $\tilde{f}_{\bar{D}}: \bar{D} \to T$, where the branch locus $B_T \equiv 4\zeta_T + 288l_T$.

Experimental Setup

Geometric Objects

• Research objects: general K3 surfaces of degree two
• Main tools: Plücker formula, birational geometry, elliptic fibration theory
• Computational framework: intersection theory, Néron-Severi space

Computational Methods

• Intersection number formulas for blowups
• Application of Teissier simultaneous normalization theorem
• Zariski decomposition of modified nef divisors

Main Results

Theorem 1.3 (Main Geometric Result)

Let $(S,L)$ be a general polarized K3 surface of degree two, and $D_S \subset \mathbb{P}(\Omega_S)$ be the surface dominated by canonical lifts of singular elliptic curves in $|L|$. Then the normalization of $D_S$ is a smooth (non-minimal) elliptic surface, and: $D_S \equiv 30\zeta_S + 54\pi^*L \equiv 30(\zeta_S + 1.8\pi^*L)$

Theorem 1.4 (Existence Result)

There exists a prime divisor $Z_S \subset \mathbb{P}(\Omega_S)$ such that: $Z_S \equiv a(\zeta_S + \lambda\pi^*L)$ where $\lambda \leq 1.7952024$.

Theorem 1.5 (Lower Bound Estimate)

If there exists a prime divisor $Z_S \subset \mathbb{P}(\Omega_S)$ such that $Z_S \equiv a(\zeta_S + \lambda\pi^*L)$, then: $\lambda \geq \frac{39}{22} = 1.772$

Key Intersection Number Calculations

Important intersection numbers in $Y$:

• $E_S^3 = 18$, $E_P^3 = -72$
• $E_S \cdot \tilde{R} = -36$, $E_P \cdot \tilde{R} = 54$
• $D^3 = -10224$ ($D$ is a big but non-nef divisor)

Related Work

Historical Background

1. Classical Results: Work of Tikhomirov and Welters on bitangent planes to smooth quartic surfaces in $\mathbb{P}^3$
2. Modern Developments: Results of Bayer-Macrì on pseudoeffective cones of Hilbert schemes
3. K3 Surface Theory: Observations of Oguiso-Peternell on special properties of degree two K3 surfaces

Technical Tools

• Stability Theory: Stability of vector bundle restrictions on K3 surfaces by Hein, Feyzbakhsh, and others
• Pseudoeffectivity: Zariski decomposition theory of divisors by Boucksom and others
• Birational Geometry: Base change theory by Maruyama

Conclusions and Discussion

Main Conclusions

1. Geometric Characterization: Complete description of the key surface $D_S$ in the projectivized cotangent bundle of degree two K3 surfaces, which is analogous to bitangent planes to quartic surfaces but possesses more complex structure.
2. Numerical Bounds: Establishment of precise numerical bounds in the pseudoeffective cone, restricting possible extremal ray parameters to a small interval $[1.772, 1.7952024]$.
3. Elliptic Fibration: Revelation of the rich elliptic fibration structure of the normalization of $D_S$, including 648 nodal fibers and 72 special points.

Limitations

1. Generality Assumption: Results require K3 surfaces to be "general" to ensure Picard number equals one and the branch curve satisfies Plücker theorem conditions.
2. Computational Complexity: Although precise bounds are provided, determining the true extremal ray requires deeper analysis.
3. Generalization Questions: Whether the methods can be generalized to K3 surfaces of other degrees or other types of surfaces remains unclear.

Future Directions

1. Hilbert Scheme Connections: Further understanding of the pseudoeffective cone through the relative Hilbert scheme $\text{Hilb}^2(\mathcal{U}/|L|)$.
2. Families of Elliptic Curves: Exploration of deeper connections between pseudoeffectivity and families of elliptic curves on $S$.
3. Higher-Dimensional Generalizations: Extension of techniques to the study of cotangent bundles of higher-dimensional hyperkähler manifolds.

In-Depth Evaluation

Strengths

1. Technical Innovation: Skillful combination of birational geometry, elliptic fibration theory, and intersection theory, establishing a powerful computational framework.
2. Completeness: Provides a complete geometric picture from abstract birational mappings to concrete numerical calculations.
3. Precision: Provides very precise numerical bounds, demonstrating deep technical mastery.
4. Geometric Insight: Reveals the unexpectedly rich structure of the cotangent bundle of K3 surfaces, providing new perspectives for the field.

Weaknesses

1. Technical Threshold: The paper is highly technical, requiring readers to possess deep background in algebraic geometry.
2. Specificity: Results are primarily focused on the degree two case, with limited generalizability.
3. Computational Verification: Some complex intersection number calculations may require more detailed verification.

Impact

1. Theoretical Contribution: Opens new directions for research on positivity of cotangent bundles of K3 surfaces.
2. Technical Value: Developed birational geometry techniques may be applicable to other geometric problems.
3. Inspirational Significance: Provides important examples for understanding cotangent bundles of higher-dimensional algebraic varieties.

Applicable Scenarios

This research is primarily applicable to:

• K3 surface theory research in algebraic geometry
• Studies of positivity problems for vector bundles
• Applications of birational geometry and elliptic fibrations
• Concrete calculations of pseudoeffective and nef cones

References

The paper cites 30 important references, including:

• BM14 Bayer-Macrì on MMP theory for K3 surface moduli spaces
• GO20 Gounelas-Ottem on positivity of K3 surface cotangent bundles
• Bou04 Boucksom's Zariski decomposition theory of divisors
• Mar72 Maruyama's theory of families of algebraic vector bundles

These references form the important theoretical foundation of this research.