Constructing prime $\mathbb{Q}$-Fano threefolds of codimension four via key varieties related with $\mathbb{P}^2\times \mathbb{P}^2$-fibrations

Basic Information

• Paper ID: 2407.06200
• Title: Constructing prime $\mathbb{Q}$-Fano threefolds of codimension four via key varieties related with $\mathbb{P}^2\times \mathbb{P}^2$-fibrations
• Author: Hiromichi Takagi (Department of Mathematics, Gakushuin University)
• Classification: math.AG (Algebraic Geometry)
• Publication Date: July 2024 (arXiv v2: October 31, 2025)
• Paper Link: https://arxiv.org/abs/2407.06200

Abstract

This paper constructs prime quasi-smooth $\mathbb{Q}$-Fano threefolds with anticanonical embeddings of codimension four in weighted projective spaces, based on affine varieties $\Sigma_{\mathbb{A}}^{13}$ and $\Pi_{\mathbb{A}}^{14}$ constructed by the author previously. Their partial projectivizations admit $\mathbb{P}^2\times\mathbb{P}^2$-fibrations with relative Picard number one. These threefolds belong to 23 classes (from $\Sigma_{\mathbb{A}}^{13}$) and 8 classes (from $\Pi_{\mathbb{A}}^{14}$ or its cone) in the Graded Ring Database. Furthermore, the author proves that for a general prime $\mathbb{Q}$-Fano threefold, a general member of the anticanonical linear system is a quasi-smooth K3 surface with at most Du Val singularities.

Research Background and Motivation

Core Problem

This paper addresses a fundamental problem in algebraic geometry: the classification of $\mathbb{Q}$-Fano threefolds. Specifically, it focuses on:

1. Existence and construction of prime $\mathbb{Q}$-Fano threefolds (whose anticanonical divisor generates the numerical equivalence class group of Weil divisors)
2. The case of anticanonical embeddings of codimension 4 (which is the most challenging in the classification)

Importance of the Problem

• Classification of $\mathbb{Q}$-Fano varieties is a central problem in Mori theory for projective threefolds
• The Graded Ring Database (GRDB) provides a massive table of candidate objects requiring verification of actual existence
• Cases with codimension ≤2 have been completely classified; codimension 3 has systematic constructions (Altınok using 5×5 skew-symmetric matrices), but codimension 4 still has many unresolved cases

Limitations of Existing Methods

1. Type I unprojection method (Brown-Kerber-Reid): Constructs examples for 116 out of 143 classes, but does not verify whether Picard number equals one
2. Cluster variety method (Coughlan-Ducat): Uses $C_2$ or $G_2^{(4)}$-cluster varieties to construct some examples, but with limited coverage
3. Type II unprojection (Taylor): Constructs some examples, but several classes remain uncovered

Research Motivation

In previous work, the author constructed more general key varieties $H_{\mathbb{A}}^{13}$ and $U_{\mathbb{A}}^{14}$, which include cluster varieties as special cases. This paper further develops this approach by introducing new key varieties $\Sigma_{\mathbb{A}}^{13}$ and $\Pi_{\mathbb{A}}^{14}$ (whose partial projectivizations admit special fibration structures) to systematically construct new examples and fill gaps in the classification.

Core Contributions

1. Systematic construction method: Through weighted complete intersections in weighted projectivizations of key varieties $\Sigma_{\mathbb{A}}^{13}$ and $\Pi_{\mathbb{A}}^{14}$, constructs 31 classes of prime $\mathbb{Q}$-Fano threefolds (23 classes from $\Sigma_{P}^{12}$, 8 classes from $\Pi_{P}^{13}$ or $\Pi_{P}^{14}$)
2. Existence verification: Proves that these constructions indeed yield quasi-smooth prime $\mathbb{Q}$-Fano threefolds, verifying consistency with numerical data in GRDB
3. K3 surface properties: Proves that the general member of the anticanonical linear system is a quasi-smooth K3 surface with only A-type Du Val singularities
4. Classification progress: Combined with existing work, existence is now confirmed for 141 out of 143 classes (only No.166 and No.12960 remain unresolved; the latter has been resolved in subsequent work)
5. New examples: The constructed examples include several new classes previously inaccessible to other methods (such as No.308, 393, 642, 644, etc.)

Detailed Methods

Task Definition

Objective: For given numerical data (Hilbert numerator, singularity basket, coordinate weights) from GRDB, construct a threefold $X$ satisfying:

• Input: Numerical data triple (a,b,c)
• Output: Prime $\mathbb{Q}$-Fano threefold $X$ with anticanonical embedding of codimension 4
• Constraints:
  1. $X$ is quasi-smooth (the affine cone is smooth outside the vertex)
  2. $X$ has only terminal singularities
  3. Picard number of $X$ equals one
  4. The anticanonical divisor of $X$ is primitive

Construction of Key Varieties

Definition of $\Sigma_{\mathbb{A}}^{13}$

In an 18-dimensional affine space $A_{\Sigma}$ with coordinates: $p = \begin{pmatrix} p_1 \\ p_2 \\ p_3 \end{pmatrix}, \quad p_4, \quad q = \begin{pmatrix} q_1 \\ q_2 \\ q_3 \end{pmatrix}, \quad r, \quad u, \quad S = \begin{pmatrix} s_{11} & s_{12} & s_{13} \\ & s_{22} & s_{23} \\ & & s_{33} \end{pmatrix}, \quad t = \begin{pmatrix} t_1 \\ t_2 \\ t_3 \end{pmatrix}$

$\Sigma_{\mathbb{A}}^{13}$ is defined by 9 polynomial equations $F_1,\ldots,F_9$ (on the hyperplane $s_{33}=1$):

• $F_1 = {}^t p q$
• ${}^t(F_2, F_3, F_4) = (rI + A_q S)p + p_4 A_q t$
• $F_5 = {}^t p S p + p_4 {}^t p t$
• ${}^t(F_6, F_7, F_8) = up - (rI - A_q S)A_q t$
• $F_9 = up_4 + (r^2 + {}^t q S^\dagger q)$

where $A_q$ is the skew-symmetric matrix defined by $q$, and $S^\dagger$ is the adjugate matrix of $S$.

Definition of $\Pi_{\mathbb{A}}^{14}$

In a 19-dimensional affine space $A_{\Pi}$ with coordinates: $p_1, p_2, p_3, p_4, u_1, u_2, s_1, s_2, s_3, t_1, t_2, t_{123}, t_{124}, t_{125}, t_{126}, t_{135}, t_{136}, t_{245}, t_{246}$

$\Pi_{\mathbb{A}}^{14}$ is defined by 9 polynomial equations $G_1,\ldots,G_9$ (on the hyperplane $t_{246}=1$), involving quadratic and cubic terms in these coordinates.

Construction Strategy

Weighted Complete Intersection Construction

For each class, the threefold $X$ is constructed as: $X = K \cap (a_1)^{m_1} \cap \cdots \cap (a_k)^{m_k}$ where:

• $K = \Sigma_P^{12}$ or $\Pi_P^{13}$ (weighted projectivization of the key variety)
• $(a_i)$ denotes a general hypersurface section of weight $a_i$
• $m_1 + \cdots + m_k = \dim K - 3$

Key observation: For each weight $a_i$, the number of coordinates of weight $a_i$ is $\geq m_i$, so section equations can be chosen in the form: $(a_i\text{-weight coordinate}) = (a_i\text{-weight polynomial in other coordinates})$

K3 Surface Construction

The anticanonical divisor section $T$ is defined as: $T = K \cap (b_1)^{n_1} \cap \cdots \cap (b_l)^{n_l}$ where weights $b_i$ are appropriately adjusted to include weight-1 sections.

Verification Strategy

The paper employs a three-step verification strategy, proving the following three key claims:

Claim (A): $X$ is a quasi-smooth threefold, $T$ is a quasi-smooth surface

• That is, $X_A^o$ is a smooth 4-dimensional variety, $T_A^o$ is a smooth 3-dimensional variety

Claim (B): Singularity analysis

• $\text{Sing } X = \text{Sing } T$
• $X$ and $T$ have only cyclic quotient singularities of specified types

Claim (C): Boundary is a prime divisor

• $X \cap \{b=0\}$ is a prime divisor (where $b=p_1$ for $\Sigma_P^{12}$, $b=p_3^2+t_1p_4^2+p_4u_2$ for $\Pi_P^{13}$)

Technical Innovations

1. Linear Part Computation (LPC)

This is the core technique for verifying singularity types:

• At a singularity $t$, select a nonzero coordinate $x$ and localize the equations
• Compute the linear part of the localized equations
• Verify that the linear part spans a 4-dimensional subspace in the cotangent space
• Determine the weights of local coordinates relative to the stabilizer group action
• Directly determine the singularity type (e.g., $1/\alpha(\beta, \alpha-\beta)$)

Advantage: Compared to the Jacobi criterion, LPC not only verifies smoothness but directly determines singularity types.

2. Reduction Strategy

Simplify verification through the following reductions:

• Claim (A) for $T$ implies Claim (A) for $X$: Since $\text{Sing } X_A^o \subset \text{Sing } T_A^o$
• Claim (B) reduces to $T$: Since the non-free locus of $X$ is contained in $T$
• Claim (C) reduces to dimension estimates: Only need to prove $\dim \text{Sing}(T \cap \{b=0\}) \leq 0$

3. Chart Partition Analysis

Analysis of $T$ is divided into:

• $p_1$-chart: Usually a complete intersection or affine space
• $p_2$-chart: Assume $p_1=0$ to simplify calculations
• Locus $\{p_1=p_2=0\}|_T$: Usually a finite set of points, verified using LPC

4. Computational Strategy

Different strategies for different complexity levels:

• $h^0(\mathcal{O}_{P_X}(1))=1$: Direct analysis of $T$'s equations
• $h^0(\mathcal{O}_{P_X}(1))=2$: Analyze the curve $C=T\cap T'$
• $h^0(\mathcal{O}_{P_X}(1))\geq 3$: Only base locus information needed

Experimental Setup

Data Sources

• Graded Ring Database (GRDB): Provides 143 classes of candidate prime $\mathbb{Q}$-Fano threefolds of codimension 4
• Excludes two classes: No.29374 (smooth Fano threefold of genus 6, classified by Gushel) and No.26989 (genus 5 with one $1/2(1,1,1)$ singularity, already classified)
• Remaining 141 classes, with 31 classes handled in this paper

Computational Tools

1. Magma: For verifying ideal generators, computing elimination ideals of Jacobi ideals
2. Mathematica: For symbolic computation, linear part computation, local coordinate changes
3. Complete code: Author provides Mathematica code (CodeQFano.nb) on GitHub

Verification Workflow

For each class (e.g., No.360, 1185, 577):

1. Determine section weights and parameter counts
2. Write out the embedding of $T$ or $C$
3. Determine singularity locations (through weight analysis)
4. Verify smoothness on various charts (Jacobi criterion or LPC)
5. Verify singularity types at special points (LPC)
6. Verify irreducibility of the boundary

Experimental Results

Main Results (Theorem 1)

Part (1): For each numerical datum in Table 1 (23 classes) and Table 3 (8 classes), there exists a quasi-smooth prime $\mathbb{Q}$-Fano threefold $X$ satisfying:

• Anticanonical embedding of codimension 4
• Constructed as a weighted complete intersection in $\Sigma_P^{12}$ (or $\Pi_P^{13}$/$\Pi_P^{14}$)
• Numerical data consistent with GRDB

Part (2): For a general $X$, a general member $T$ of $|{-K_X}|$ is a quasi-smooth K3 surface satisfying:

• Has only A-type Du Val singularities
• $\text{Sing } T = \text{Sing } X$
• If $X$ has a $1/\alpha(\beta,-\beta,1)$ singularity at some point, then $T$ has a $1/\alpha(\beta,-\beta)$ singularity at that point

Concrete Examples

Case No.360

• Embedding: $T \subset \mathbb{P}(4,5,6,7^2,8,9)$
• Parameters: 6 complex parameters $a_3,\ldots,b_0$
• Singularities:
  • 2 singularities of type $1/4(1,3)$ (on $T\cap\mathbb{P}(t_2,p_2)$)
  • 1 singularity of type $1/6(1,5)$ ($p_4$-point)
  • 1 singularity of type $1/7(2,5)$ ($p_1$-point)
• Verification method:
  • $p_1$-chart: Complete intersection, verified using Jacobi criterion
  • $p_2$-chart: After removing weight-8 section, becomes a hypersurface, easy to verify smoothness
  • Finite points: Determine singularity types using LPC

Case No.1185

• Embedding: $T \subset \mathbb{P}(2,3,4,5^2,6,8)$
• Parameters: 17 complex parameters
• Singularities:
  • 3 singularities of type $1/2(1,1)$
  • 1 singularity of type $1/5(1,4)$
  • 1 singularity of type $1/8(3,5)$ ($u$-point)
• Special handling: Analysis of $\{p_1=p_2=0\}|_T$ requires verifying the freeness of $Z_8$ action

Case No.577 (Most Complex)

• Embedding: $T \subset \mathbb{P}(3,4,5^2,6^2,7)$
• Parameters: 8 complex parameters
• Singularities:
  • 1 singularity of type $1/2(1,1)$
  • 3 singularities of type $1/3(1,2)$
  • 2 singularities of type $1/5(1,4)$
• Technical difficulty: Requires proving that three open subsets are isomorphic to open subsets of affine 9-space, using refined local analysis

Statistical Results

Classification Progress

• 141 out of 143 classes now have constructed examples
• Only 2 classes remain:
  • No.166: Existence still unknown (if it exists, it is birationally superrigid)
  • No.12960: Resolved in subsequent work Tak10

Related Work

Historical Background

1. Fano (1942): Classification of smooth prime Fano threefolds
2. Iskovskikh-Mori-Mukai: Development of Mori theory framework
3. Graded Ring Database: Provides systematic list of candidates

Cases with Codimension ≤3

• Codimension ≤2: Fano-Iskovskikh completed classification (quasi-smooth case)
• Codimension 3: Altınok systematically constructs using 5×5 skew-symmetric matrices

Existing Work on Codimension 4

Brown-Kerber-Reid (2012)

• Method: Type I unprojection (Tom and Jerry type)
• Coverage: 116 out of 143 classes
• Limitation: Does not verify Picard number equals one

Coughlan-Ducat (2020)

• Method: Rank 2 cluster varieties ($C_2$ and $G_2^{(4)}$)
• Contribution: First use of key variety method
• Limitation: Limited coverage

Author's Previous Work

• $H_{\mathbb{A}}^{13}$: Contains $C_2$-cluster varieties, constructs No.20652, etc.
• $U_{\mathbb{A}}^{14}$: Contains $G_2^{(4)}$-cluster varieties, constructs No.20544, etc.
• Key insight: Fibration structure of partial projectivization guarantees Picard number equals one

Taylor (2020)

• Method: Type II unprojection
• Contribution: Constructs 14-dimensional affine variety $\Upsilon_{\mathbb{A}}^{14}$
• Relation: Inspired the construction of $\Pi_{\mathbb{A}}^{14}$ in this paper

Advantages of This Paper

1. Systematicity: Handles multiple classes through unified key variety framework
2. New examples: Obtains classes inaccessible to previous methods
3. Theoretical guarantee: Fibration structure automatically ensures Picard number equals one
4. Verifiability: Provides complete computational code

Conclusions and Discussion

Main Conclusions

1. Completeness of construction: Through two key varieties $\Sigma_{\mathbb{A}}^{13}$ and $\Pi_{\mathbb{A}}^{14}$, systematically constructs 31 classes of prime $\mathbb{Q}$-Fano threefolds
2. Classification progress: Combined with existing work, 141 out of 143 classes now have constructions, with only No.166 remaining unresolved
3. K3 surface properties: The general member of the anticanonical linear system is a quasi-smooth K3 surface with only A-type Du Val singularities
4. Method effectiveness: "Construction via key varieties" is proven to be a powerful tool for systematically constructing $\mathbb{Q}$-Fano threefolds

Limitations

1. Computational complexity:
   • Verification of some cases (e.g., No.577) requires very refined local analysis
   • Depends on computer algebra systems (Magma and Mathematica)
   • Manual verification is nearly infeasible
2. Coverage range:
   • Handles only 31 classes, though combined with other work covers 141 classes
   • Existence of No.166 remains an open problem
3. Generality issues:
   • Construction depends on choice of "general" parameters
   • Does not discuss degenerate cases with special parameters
4. Uniqueness:
   • Does not discuss classification of different topological types within the same class
   • Though mentions that different weighted projectivizations produce different topological types

Future Directions

Directions explicitly proposed by the author:

1. Further applications of $U_{\mathbb{A}}^{14}$:
   • Expected to yield more codimension 4 examples
   • Planned for future work
2. Existence of No.166:
   • The only remaining unresolved class
   • If it exists, it is known to be birationally superrigid (Okada)
3. Alternative constructions for No.12960:
   • Already constructed in Tak10 via other key varieties
   • Partial projectivization admits 5-dimensional cone over $\mathbb{P}^2\times\mathbb{P}^2$ fibration
4. Higher-dimensional generalizations:
   • Key variety method may generalize to higher-dimensional Fano varieties
   • Relationship between fibration structure and Picard number deserves deeper study

Theoretical Significance

1. Mori theory: Provides systematic tools for classification of $\mathbb{Q}$-Fano threefolds
2. Moduli space theory:
   • Different topological types within the same class correspond to different branches of moduli space
   • Different weighted projectivizations of key varieties provide systematic parametrization
3. Fibration method:
   • Fibration structure of partial projectivization is an effective tool for controlling Picard number
   • Relative Picard number 1 + general fiber Picard number ≥2 → total space Picard number 1

In-Depth Evaluation

Strengths

1. Methodological Innovation (★★★★★)

• Key variety framework: Systematizes the key variety method, generalizing from cluster varieties to more general structures
• Fibration insight: Deeply exploits geometric properties of $\mathbb{P}^2\times\mathbb{P}^2$-fibrations to control Picard number
• Technical tools: LPC method is more efficient than traditional Jacobi criterion for verifying singularity types

2. Theoretical Rigor (★★★★★)

• Complete proofs: Though computation-dependent, the logical framework is clear and complete
• Reduction strategy: Systematically simplifies verification through $X \to T \to C$ reduction
• Proposition support: Key properties of key varieties (irreducibility, normality, UFD property) have complete proofs

3. Computational Verifiability (★★★★☆)

• Open code: Provides complete Mathematica code on GitHub
• Detailed data: Tables 1-4 give complete data for all 31 classes
• Representative examples: Complete computational process for No.360, 1185, 577
• Limitation: Some computations remain complex, reproduction requires specialized knowledge

4. Academic Value (★★★★★)

• Classification progress: Advances known examples from 116 to 141 classes (combined with other work)
• New examples: First construction of 8 entirely new classes
• Methodological contribution: Provides systematic framework for subsequent research

5. Writing Quality (★★★★☆)

• Clear structure: Strategy section (Section 2) explains verification logic in detail
• Technical details: Appendix reviews key variety definitions
• Readability: Easy to follow for algebraic geometry experts
• Room for improvement: Still quite technical for non-specialists

Weaknesses

1. Computational Dependency (★★★☆☆)

• Black-box computation: Many verifications depend on built-in commands of Magma/Mathematica
• Reproducibility issues: Though code is provided, complete reproduction of all 31 classes still requires substantial computation time
• Theory vs. computation: Rigor of certain claims (e.g., "general parameters") depends on computational verification

2. Coverage Range (★★★☆☆)

• Partial coverage: Handles only 31/143 classes (though combined with other work is sufficient)
• Method limitations: $\Sigma_{\mathbb{A}}^{13}$ and $\Pi_{\mathbb{A}}^{14}$ cannot cover all cases
• No.166: Still unresolved, may require entirely new methods

3. Generality Discussion (★★☆☆☆)

• Parameter space: Does not discuss moduli space structure of parameters in detail
• Degenerate cases: Behavior under special parameters not analyzed
• Topological types: Different topological types within the same class not completely classified

4. Application Limitations (★★★☆☆)

• Specialization: Method is highly specialized, primarily for algebraic geometry experts
• Generalizability: Generalization to higher dimensions or other types of Fano varieties not obvious

Impact Assessment

Contribution to the Field (★★★★★)

• Milestone progress: Classification of $\mathbb{Q}$-Fano threefolds nears completion
• Methodology: Key variety method becomes standard tool
• Database completion: Provides existence verification for candidates in GRDB

Practical Value (★★★☆☆)

• Theoretical tool: Provides concrete examples for studying geometric properties of Fano varieties
• Birational geometry: Some examples (e.g., No.308) proven to be birationally superrigid
• Application scope: Primarily limited to pure mathematics research

Reproducibility (★★★★☆)

• Code availability: Mathematica code available on GitHub
• Data completeness: Tables provide all necessary numerical data
• Documentation: Paper explains verification strategy in detail
• Challenge: Complete reproduction still requires specialized background and computational resources

Applicable Scenarios

1. Algebraic geometry research:
   • Fano variety classification
   • Birational geometry
   • Applications of Mori theory
2. Moduli space theory:
   • Study of moduli spaces of $\mathbb{Q}$-Fano threefolds
   • Understanding parametrization of different topological types
3. Computational algebraic geometry:
   • Development of new symbolic computation methods
   • Verification of theoretical predictions of existence
4. Teaching purposes:
   • Case study for advanced algebraic geometry courses
   • Illustration of computational methods in modern algebraic geometry

Selected References

1. GRDB Altınok et al., The Graded Ring Database (database of $\mathbb{Q}$-Fano threefold candidates)
2. BKR Brown-Kerber-Reid (2012), Fano 3-folds in codimension 4, Tom and Jerry (Type I unprojection method)
3. CD Coughlan-Ducat (2020), Constructing Fano 3-folds from cluster varieties (cluster variety method)
4. Tak9 Takagi, Key varieties related with $\mathbb{P}^2\times\mathbb{P}^2$-fibration (construction of $\Sigma_{\mathbb{A}}^{13}$)
5. Tak6 Takagi, Key varieties defined by Jordan algebras, Part II (construction of $\Pi_{\mathbb{A}}^{14}$)
6. Tay Taylor (2020), Type II unprojections (inspired $\Pi_{\mathbb{A}}^{14}$)
7. O Okada (2020), Birationally superrigid Fano 3-folds (birational superrigidity of No.308 and No.166)

Overall Assessment: This is a high-quality algebraic geometry research paper that makes significant progress on the core problem of classifying $\mathbb{Q}$-Fano threefolds. The methodological innovation (key varieties + fibrations), theoretical rigor, and computational verifiability provide a systematic construction framework for the field. Though highly technical and computation-dependent, it is an important reference for algebraic geometry experts. The paper's main value lies in methodological contributions and advancement of the classification problem, rather than construction of individual examples.