K3 Surfaces with Two Involutions and Low Picard Number

Basic Information

• Paper ID: 2210.14623
• Title: K3 surfaces with two involutions and low Picard number
• Authors: Dino Festi, Wim Nijgh, Daniel Platt
• Classification: math.AG (Algebraic Geometry), math.DG (Differential Geometry), math.NT (Number Theory)
• Publication Date: February 13, 2024 (v2)
• Paper Link: https://arxiv.org/abs/2210.14623

Abstract

This paper studies complex algebraic K3 surfaces with two involutions, where one is holomorphic and the other is antiholomorphic. For K3 surfaces X of degree 2d with Picard number ρ, the authors establish that ρ ≥ 1 when d = 1 and ρ ≥ 2 when d ≥ 2. They construct new explicit K3 surfaces defined over the rationals that realize the minimal Picard number ρ = 2 for d = 2, 3, 4. Furthermore, they prove that nodal quartic surfaces can realize the minimal Picard number ρ = 2 for infinitely many distinct values of d. Finally, they strengthen Morrison's result by proving that for any even lattice N of rank 1 ≤ r ≤ 10 with signature (1, r−1), there exists a K3 surface Y defined over the reals such that Pic Y_C = Pic Y ≅ N.

Research Background and Motivation

Problem Background

1. Geometric Motivation: K3 surfaces play an important role in differential geometry, particularly in the construction of G₂ manifolds. The construction methods of Joyce-Karigiannis and Kovalev-Lee require K3 surfaces with two involutions—one non-symplectic holomorphic involution and one antiholomorphic involution.
2. Physical Applications: Studying the moduli space of G₂ metrics via the Donaldson-Thomas program requires counting G₂ instantons, which can be constructed from stable bundles on K3 surfaces. Verifying bundle stability becomes computationally more difficult as the number of line bundles (i.e., Picard number) increases.
3. Theoretical Problem: For K3 surfaces of given degree, what is the minimal Picard number when requiring both holomorphic and antiholomorphic involutions? This is the core theoretical question of the paper.

Research Significance

• Geometric Construction: Examples of K3 surfaces with low Picard number are crucial for constructing G₂ manifolds and instantons on them
• Computational Efficiency: Low Picard number makes stability verification computationally more feasible
• Theoretical Completeness: Provides a complete characterization of lower bounds on Picard numbers for K3 surfaces with involutions

Limitations of Existing Methods

• Elsenhans-Jahnel (2008) only provided examples with ρ = 1 for d = 1
• Kondō's examples, while realizing ρ = 2 for all d ≥ 2, consist of only a single example
• Lack of explicit construction methods for different degrees
• Morrison's result applies only to K3 surfaces over the complex numbers

Core Contributions

1. Theoretical Completeness (Theorem 1.1): Completely answers the question of minimal Picard number for K3 surfaces with two involutions:
   • When d = 1: ρ ≥ 1, with examples of ρ = 1 defined over ℚ
   • When d > 1: ρ ≥ 2, with examples of ρ = 2 defined over ℚ
2. New Explicit Constructions:
   • K3 surfaces of degree 4 and 8 with Picard lattice 4 5 2 (§6)
   • K3 surface of degree 6 with Picard lattice 6 6 2 (§7)
   • K3 surfaces of degree 2d (d > 3) with Picard lattice 2 d+1 2d (§8)
3. Infinite Family of Examples: Proves that nodal quartic surfaces can realize ρ = 2 for infinitely many distinct values of d (§4)
4. Strengthened Result over the Reals (Corollary 8.3): For any even lattice N of rank 1 ≤ r ≤ 10 with signature (1, r−1), there exists a K3 surface Y defined over ℝ satisfying Pic Y_C = Pic Y ≅ N
5. Computational Resources: Provides Magma code for constructing and verifying these examples

Detailed Methods

Task Definition

Given a positive integer d, construct a K3 surface X of degree 2d with Picard number 2 such that:

1. X is defined over the rationals ℚ (or reals ℝ)
2. X admits a holomorphic involution ι
3. X admits an antiholomorphic involution σ (automatically provided by definition over ℝ)
4. The two involutions commute

Core Observation (Remark 1.3)

Key Equivalence: A complex K3 surface admits commuting holomorphic and antiholomorphic involutions if and only if the underlying algebraic K3 surface can be defined over ℝ and admits a second-order automorphism.

Construction Strategy: Seek K3 surfaces defined over ℝ with an ample divisor D of self-intersection D² = 2. Such a divisor provides the required second-order automorphism (Lemma 3.1).

Method I: Double Covers of the Plane (§3)

Key Role of Lemma 3.1: If a K3 surface X has an ample divisor H with self-intersection H² = 2, then X is isomorphic to a double cover of ℙ² branched over a smooth sextic curve.

Involution from Double Cover: For a K3 surface defined by w² = f(x,y,z) in weighted projective space ℙ(1,1,1,3), the involution is given by: $ι: (x:y:z:w) \mapsto (x:y:z:-w)$

Example 3.4: Constructs an example X₂ with Picard lattice ⟨2⟩ and verifies ρ = 1 through modular reduction:

• Computes Picard lattice rank and discriminant modulo 5 and 13
• Uses the fact that detPicX₂,₅ ≢ detPicX₂,₁₃ mod ℚ² to conclude ρ(X₂) = 1

Method II: Nodal Quartic Surfaces (§4)

Construction Idea:

1. Start with a quartic surface X in ℙ³ with a single node
2. Obtain a smooth model S by blowing up the node
3. The Picard lattice contains ⟨H,E⟩ ≅ 4 0 -2, where H is a hyperplane section and E is the exceptional divisor

Key Technique (Proposition 4.4): For d > 2 where 2 is a quadratic residue modulo d, there exists a primitive ample class D ∈ PicS with D² = 2d.

Pell Equation Method: The problem reduces to solving the Pell equation $y² - 2x² = -d$ By Lagrange's result, this equation has solutions if and only if y² - 2x² = ±1 has solutions (which always holds).

Weyl Group Argument: Uses the transitive action of the Weyl group to map solutions in the positive cone into the ample cone.

Example 4.7: Provides a concrete nodal quartic surface defined over ℚ and verifies that its smooth model has Picard number 2.

Method III: Smooth Quartic Surfaces Containing Specific Curves (§5-6)

Theorem 5.1 (Mori, Knutsen): There exists a smooth quartic surface X containing a smooth curve C of degree e and genus g if and only if:

• g = e²/8 + 1, or
• g < e²/8 and (e,g) ≠ (5,3)

Proposition 5.2: If e > 4, a smooth quartic surface X contains a smooth genus 2, degree e curve C with PicX = ⟨H,C⟩, then C is ample and X admits a holomorphic involution induced by the linear system |C|.

Construction Method (Remark 6.5):

1. Construct a bidegree (2,3) genus 2 curve C' and a bidegree (2,1) genus 0 curve D' in ℙ¹ × ℙ¹
2. Map their union to a quadric surface Q ⊂ ℙ³ via the Segre embedding
3. The image C ∪ D is defined by a quartic polynomial
4. Verify smoothness and Picard number

Example 6.1 (X₄): Explicitly constructs a quartic surface of degree 4 containing a degree 5, genus 2 curve C with Picard lattice 4 5 2.

Explicit Involution Formula (Remark 6.6): The involution ι is given by a degree 9 polynomial expression through the linear system |9H − 5D|.

Other Models:

• |C| gives a double cover model X'₂ of ℙ²
• |3H − C| gives a model X₈ as the intersection of three quadrics in ℙ⁵

Method IV: Intersection of Quadric and Cubic Surfaces (§7)

Construction Strategy: Seek K3 surfaces in ℙ⁴ defined by a quadric x₄² = f and a cubic l₁g₁ + l₂g₂ + l₃g₃ = 0.

Example 7.1 (X₆): Explicitly provides a K3 surface of degree 6 containing a degree 6, genus 2 curve C₆ with Picard lattice 6 6 2.

Lattice-Theoretic Argument (Lemma 7.4): The lattice 6 6 2 does not represent any d ≡ 4 mod 6, in particular does not represent −2, hence contains no smooth rational curves, and the ample cone equals the positive cone.

Method V: General Construction over the Reals (§8)

Strengthening of Morrison's Result (Proposition 8.2): For a primitive sublattice N of U⊕² ⊕ E₈(−1)⊕² (rank r, signature (1, r−1)), there exists a K3 surface X defined over ℝ satisfying PicX = PicX_C ≅ N.

Construction Steps:

1. Find a primitive embedding of N in Λ := U⊕² ⊕ E₈(−1)⊕²
2. Choose w₋ in N⊥ ⊗ ℝ such that w₋² > 0
3. Choose w₊ in U ⊗ ℝ such that w₊² = w₋²
4. Construct w := w₊ + iw₋ satisfying the Riemann conditions
5. Obtain a K3 surface Y by the surjectivity of the period map
6. Define an involution τ = id ⊕ (−id) on ΛK₃ = U ⊕ Λ
7. Verify that τ satisfies the conditions of Theorem 8.1 to obtain a model X defined over ℝ

Existence of Real Points: Proves X(ℝ) ≠ ∅ by computing the Euler characteristic χ(X(ℝ)) = −16 ≠ 0.

Application (Proposition 8.5): For d > 3, there exists a K3 surface defined over ℝ with Picard lattice 2 d+1 2d generated by two ample classes.

Experimental Setup

Computational Tools

The paper uses the Magma computer algebra system for numerical verification and construction. All code is publicly available at: https://github.com/danielplatt/quartic-k3-with-involution

Verification Methods

Picard Number Upper Bound Computation (§2.3):

1. Select good reduction primes p
2. Compute the Picard number of the reduced surface X_p
3. Use the embedding PicX_C → PicX_{p,F̄_p} to obtain an upper bound
4. Apply the characteristic polynomial of the Frobenius automorphism and the Tate conjecture (proved by Kim-Pera)

Concrete Implementation:

• For degree 2 K3 surfaces: use the Magma built-in function of Elsenhans-Jahnel
• Verification strategy: compute at different primes and use different residues of discriminants to exclude high Picard numbers

Data Sets

The paper does not involve traditional datasets but rather constructs specific algebraic geometry objects. Each example is verified through:

1. Defining equations
2. Verifying smoothness
3. Computing Picard number
4. Verifying existence of involutions

Experimental Results

Main Results

Complete Proof of Theorem 1.1:

• d = 1: Example 3.4 provides an example with ρ = 1 (Picard lattice ⟨2⟩)
• d ≥ 2: Kondō's elliptic K3 surface X₆₆ realizes ρ = 2 for all d > 1 (Picard lattice U = 0 1 0)

Verification of New Examples:

1. Example 3.4 (X₂, degree 2):
   • Picard lattice: ⟨2⟩ (ρ = 1)
   • Verification: rank(PicX₂,₅) = rank(PicX₂,₁₃) = 2, but detPicX₂,₅ ≢ detPicX₂,₁₃ mod ℚ²
   • Conclusion: ρ(X₂) = 1
2. Example 4.7 (Nodal quartic, degree 2d):
   • Picard lattice: 4 0 -2 (ρ = 2)
   • Verification: reduction modulo 5 shows ρ = 2
   • Realizable degrees: d ∈ {7, 14, 17, 23, 31, 34, 41, 46, 47, 49, 62, 71, 73, 79, 82, 89, 94, 97, 98, ...} (2 is a quadratic residue modulo d)
3. Example 6.1 (X₄, degree 4):
   • Picard lattice: 4 5 2 (ρ = 2)
   • Contains a degree 5, genus 2 curve C
   • Verification: reduction modulo 2 shows ρ ≤ 2
   • Involution: ι*(H) = −H + 5C, ι*(C) = C
4. Example 7.1 (X₆, degree 6):
   • Picard lattice: 6 6 2 (ρ = 2)
   • Contains a degree 6, genus 2 curve C₆
   • Verification: reduction modulo 7 shows ρ ≤ 2
   • Non-representable: d ≡ 4 mod 6 (in particular, cannot be embedded as the intersection of a quadric and cubic in ℙ⁴)
5. Proposition 8.5 (degree 2d, d > 3):
   • Picard lattice: 2 d+1 2d (ρ = 2)
   • Definition field: reals ℝ
   • Does not contain −2 class (when d > 3), hence ample cone = positive cone

Technical Verification

Negative Result of Proposition 5.5: If a quartic surface X ⊂ ℙ³ admits a linear involution ι: ℙ³ → ℙ³ such that ι(X) = X, then ρ(X) ≥ 8. This shows that the involution in Example 6.1 cannot be induced by a linear involution.

Lemma 5.6: For the situation of Proposition 5.2, the action of the involution on the Picard lattice is:

• ι*(C) = C
• ι*(H) = −H + eC

Effectiveness of Construction Methods

Remarks 6.5 and 7.7: Provide systematic construction methods that can generate more examples through random search. The Magma files ConstructionQuartic and ConstructionK3Degree6 implement these constructions.

Related Work

Foundational K3 Surface Theory

1. Huybrechts 12: Standard textbook on K3 surfaces, providing foundational theory on Picard lattices, period domains, and Torelli theorems
2. Kondō 18: Monograph on K3 surfaces, particularly on classification of automorphisms and involutions

Classification of Involutions

3. Nikulin 22, 24:
   • Theorem 2.6: Symplectic involutions have 8 isolated fixed points, ρ ≥ 9
   • Classification of non-symplectic involutions by fixed point sets: empty, two elliptic curves, or mixed curves, leading to ρ ≥ 10 or ρ ≥ 11 − p_a(C) + k
4. Artebani-Sarti-Taki 1: Prove that the lattice L₊ related to the fixed point set is contained in the Picard lattice

Picard Number Computation

5. Elsenhans-Jahnel 7, 8, 9:
   • First provided examples with d = 1, ρ = 1
   • Developed methods for computing Weil polynomials of degree 2 K3 surfaces
   • Magma implementation for computing Picard numbers of reductions
6. van Luijk 34: Practical method for computing characteristic polynomials through point counting
7. Charles 4, Kim-Pera 15: Proof and refinement of the Tate conjecture

Curves and Quartic Surfaces

8. Mori 20, Knutsen 16: Theorem 5.1 on existence of curves on quartic surfaces
9. Bini 3: Computation of automorphism groups of K3 surfaces with Picard lattice 4 0 -2 (Proposition 4.1)

Real K3 Surfaces

10. Morrison 21: Proved that even lattices of rank ≤ 10 can be realized as Picard lattices of complex K3 surfaces
11. Silhol 33: Theory of real algebraic surfaces, particularly Theorem 8.1's discriminant criterion
12. Nikulin-Saito 25, 26: Moduli spaces and connected components of real K3 surfaces

Geometric Applications

13. Joyce-Karigiannis 14: Blowup construction of G₂ manifolds using K3 surfaces
14. Kovalev-Lee 19: Alternative construction of G₂ manifolds
15. Walpuski et al. 29, 35: Construction of G₂ instantons from bundles on K3 surfaces

Position of This Paper

Advantages of this paper over existing work:

1. Systematicity: Provides explicit construction methods for different degrees, rather than isolated examples
2. Computability: Provides Magma code making constructions reproducible
3. Theoretical Strengthening: Extends Morrison's result from ℂ to ℝ while ensuring Pic Y_C = Pic Y
4. Application-Oriented: Explicitly addresses the needs of G₂ geometry

Conclusions and Discussion

Main Conclusions

1. Complete Characterization: Theorem 1.1 completely answers the question of minimal Picard number for K3 surfaces with two involutions and provides realizations over the rationals.
2. Rich Library of Examples:
   • Degree 2: example with ρ = 1 (Example 3.4)
   • Degree 4: example with ρ = 2, Picard lattice 4 5 2 (Example 6.1)
   • Degree 6: example with ρ = 2, Picard lattice 6 6 2 (Example 7.1)
   • Degree 2d (d > 3): example with ρ = 2, Picard lattice 2 d+1 2d (Proposition 8.5)
   • Infinite family: nodal quartic surfaces realizing ρ = 2 for infinitely many d (Proposition 4.4)
3. Theoretical Strengthening: Corollary 8.3 extends Morrison's result from the complex numbers to the reals while ensuring Pic Y_C = Pic Y.
4. Computational Resources: Provides publicly available Magma code supporting further research and applications.

Limitations

1. Degree Restrictions:
   • The Picard lattice 4 5 2 of Example 6.1 does not represent 6, so this surface has no degree 6 polarization
   • One needs d = 9 to find a quartic surface simultaneously admitting a degree 6 divisor
2. Definition Field:
   • The examples in Proposition 8.5 are only defined over the reals ℝ, not necessarily over the rationals ℚ
   • Examples 3.4, 4.7, 6.1, 7.1 are defined over ℚ but do not cover all degrees
3. Construction Methods:
   • The construction methods in Remarks 6.5 and 7.7 require random search and verification
   • No deterministic algorithm is provided guaranteeing finding surfaces with desired Picard number
4. Explicit Involutions:
   • While existence of involutions is proved, explicit formulas are given only for the double cover case
   • Example 6.1 provides a degree 9 polynomial expression, but it is extremely complex
5. High Picard Number Cases:
   • Proposition 5.5 shows that quartic surfaces with linear involutions have Picard number at least 8
   • The paper does not systematically study cases with ρ > 2

Future Directions

1. Examples for More Degrees:
   • Systematically construct examples of ρ = 2 for degrees 10, 12, 14, etc.
   • Determine which degrees can realize ρ = 2 over ℚ
2. Algorithm Improvements:
   • Develop deterministic algorithms for constructing K3 surfaces of given degree and Picard lattice
   • Improve efficiency of Picard number computation
3. G₂ Geometry Applications:
   • Use these examples to construct concrete G₂ manifolds
   • Study corresponding G₂ instantons
4. Moduli Space Research:
   • Study the moduli space structure of K3 surfaces with two involutions
   • Classify connected components corresponding to different Picard lattices
5. Metric Degeneration:
   • Study degeneration behavior of Calabi-Yau metrics on these examples
   • Similar to work of Chen-Viaclovsky-Zhang
6. Higher Rank Cases:
   • Extend Corollary 8.3 to rank r > 10
   • Study uniqueness of embeddings over ℝ

In-Depth Evaluation

Strengths

1. Theoretical Completeness:
   • Theorem 1.1 provides a complete characterization of minimal Picard numbers
   • Corollary 8.3 strengthens Morrison's classical result
   • Theory results are combined with explicit examples
2. Systematic Construction Methods:
   • Multiple construction techniques provided (double covers, nodal quartics, quartics with curves, intersections of quadric and cubic)
   • Remarks 6.5 and 7.7 give reproducible construction steps
   • Different methods apply to different degree ranges
3. Rigorous Computational Verification:
   • Uses modular reduction and Tate conjecture to verify Picard numbers
   • Verification at multiple primes to exclude high Picard numbers
   • Publicly available Magma code ensures reproducibility
4. Technical Innovation:
   • Lemma 3.1 and Proposition 3.3 establish equivalences between involutions, degree 2, and double covers
   • Proposition 4.4 cleverly combines Pell equations with ampleness conditions
   • Proposition 8.2's construction uses Riemann conditions and surjectivity of the period map
5. Application-Oriented:
   • Explicitly addresses G₂ geometry needs (low Picard number for stability checking)
   • Provides rich explicit examples for subsequent applications
   • Code is publicly available for use by geometric physicists
6. Writing Quality:
   • Clear structure progressing from background through theory to examples
   • Detailed remarks, such as Remarks 6.5, 7.7 providing construction details
   • Precise statements of lemmas, propositions, and theorems

Weaknesses

1. Incomplete Coverage:
   • No explicit examples over ℚ for degrees 8, 10, 12, etc.
   • The nodal quartic method depends on 2 being a quadratic residue modulo d, excluding many d values
   • No systematic method to determine which (d, ρ) pairs can be realized over ℚ
2. Randomness in Construction:
   • Methods in Remarks 6.5 and 7.7 require random generation and verification
   • No estimates of success probability provided
   • No guarantee of finding examples for specific degree and Picard lattice
3. Lack of Explicit Involutions:
   • Except for double covers, most involutions lack explicit formulas
   • The degree 9 polynomial in Remark 6.6 is too complex for practical computation
   • Geometric description of involutions on fixed point sets is insufficient
4. Insufficient Connection to G₂ Geometry:
   • While motivation comes from G₂ geometry, no concrete G₂ manifold constructions are given
   • No discussion of G₂ instantons on these examples
   • Lack of connection to metric degeneration
5. Limited Experimental Scope:
   • Only finitely many examples are verified
   • No statistics on success rates of different construction methods
   • No analysis of efficiency of Picard number computation
6. Limited Theoretical Depth:
   • Proof of Corollary 8.3 is mainly technical, lacking deep geometric insight
   • No discussion of moduli space structure of Picard lattices
   • Lacks explanation for why certain Picard lattices are easier to realize

Impact

1. Contribution to Algebraic Geometry:
   • Enriches the library of K3 surface examples
   • Corollary 8.3 strengthens Morrison's classical result
   • Provides new construction techniques (e.g., Pell equation method in Proposition 4.4)
2. Contribution to Differential Geometry:
   • Provides raw material for G₂ manifold construction
   • Low Picard number examples facilitate study of metric degeneration
   • May inspire new G₂ instanton constructions
3. Contribution to Computational Number Theory:
   • Magma code can be used for further computational experiments
   • Demonstrates power of modular reduction techniques
   • Provides benchmark examples for algorithm development
4. Practical Value:
   • For geometric physicists: provides concrete K3 surfaces usable in G₂ geometry
   • For algebraic geometers: provides examples for studying moduli spaces and degenerations
   • For computational mathematicians: provides benchmarks for algorithm testing
5. Reproducibility:
   • Code publicly available on GitHub
   • Construction steps detailed
   • Verification methods explicit
   • Facilitates use and extension by other researchers

Applicable Scenarios

1. G₂ Manifold Construction:
   • Joyce-Karigiannis blowup method
   • Kovalev-Lee twisted connected sum method
   • Requires low Picard number for computation
2. G₂ Instanton Research:
   • Construction from stable bundles on K3 surfaces
   • Low Picard number simplifies stability checking
   • Applicable to moduli space compactification research
3. K3 Surface Moduli Spaces:
   • Study moduli spaces of K3 surfaces with involutions
   • Study branches corresponding to different Picard lattices
   • Study degeneration families
4. Arithmetic Geometry:
   • Study of rational points
   • Study of Galois actions
   • Study of reduction properties
5. Computational Algebraic Geometry:
   • Algorithm testing
   • Benchmarks for symbolic computation
   • Verification of Picard number computation methods
6. Teaching and Learning:
   • Concrete examples for K3 surface theory
   • Illustration of relationships between involutions and Picard lattices
   • Demonstration of computational verification techniques

Selected References

12 D. Huybrechts, Lectures on K3 surfaces, Cambridge Studies in Advanced Mathematics, vol. 158, 2016.

17 S. Kondō, Automorphisms of algebraic K3 surfaces which act trivially on Picard groups, J. Math. Soc. Japan 44 (1992), no. 1, 75–98.

18 S. Kondō, K3 surfaces, EMS Tracts in Mathematics, vol. 32, 2020.

21 D. R. Morrison, On K3 surfaces with large Picard number, Invent. Math. 75 (1984), no. 1, 105–121.

22 V. V. Nikulin, Finite groups of automorphisms of Kählerian K3 surfaces, Trudy Moskov. Mat. Obshch. 38 (1979), 75–137.

23 V. V. Nikulin, Integer symmetric bilinear forms and some of their geometric applications, Izv. Akad. Nauk SSSR Ser. Mat. 43 (1979), no. 1, 111–177.

33 R. Silhol, Real algebraic surfaces, Lecture Notes in Mathematics, vol. 1392, Springer-Verlag, Berlin, 1989.

Summary: This is a high-quality algebraic geometry paper that systematically studies the minimal Picard number problem for K3 surfaces with two involutions. The paper's main strengths lie in theoretical completeness, diversity of construction methods, rigor of computational verification, and attention to applications. Main weaknesses include incomplete coverage of degree ranges, randomness in constructions, and limited depth of connection to applications (G₂ geometry). The paper makes important contributions to algebraic geometry, differential geometry, and computational number theory, with provided code and examples having lasting impact on subsequent research.