Optimal upper bound for degrees of canonical Fano threefolds of Picard number one

Basic Information

• Paper ID: 2501.16632
• Title: Optimal upper bound for degrees of canonical Fano threefolds of Picard number one
• Authors: Chen Jiang (Fudan University), Haidong Liu (Sun Yat-sen University), Jie Liu (Chinese Academy of Sciences)
• Classification: math.AG (Algebraic Geometry)
• Publication Date: November 20, 2025 (version 0.02)
• Paper Link: https://arxiv.org/abs/2501.16632

Abstract

This paper proves that for Q-factorial canonical Fano threefolds X with Picard number one, the anti-canonical degree satisfies $(-K_X)^3 \leq 72$, with equality if and only if $X \cong \mathbb{P}(1,1,1,3)$ or $\mathbb{P}(1,1,4,6)$. The main tool is establishing a Kawamata-Miyaoka type inequality that relates $(-K_X)^3$ to the generalized second Chern class $\hat{c}_2(X) \cdot c_1(X)$.

Research Background and Motivation

Core Problem

This paper investigates the upper bound problem for the anti-canonical degree $(-K_X)^3$ of Fano threefolds, which is one of the central problems in the classification theory of three-dimensional algebraic varieties.

Problem Significance

1. Classification Theory Foundation: The anti-canonical degree is an important invariant of Fano threefolds and plays a key role in the classification of smooth Fano threefolds
2. Minimal Model Program: According to the minimal model program, Fano varieties with mild singularities form a fundamental class of objects in birational geometry research
3. Theoretical Completeness: While results exist for the smooth and Gorenstein cases, understanding of the general canonical singularities case remains incomplete

Limitations of Existing Methods

1. Prokhorov (2005): Proved that for Gorenstein canonical Fano threefolds, $(-K_X)^3 \leq 72$ and the bound is optimal
2. Jiang-Zou (2023): Proved that for general canonical Fano threefolds, $(-K_X)^3 \leq 324$, but conjectured the optimal upper bound should be 72
3. Technical Obstacles: Generalizing from the Gorenstein case to the Q-factorial case requires handling more complex singularity structures and Chern class computations

Research Motivation

This paper aims to prove that the conjectured optimal bound of 72 holds for Q-factorial canonical Fano threefolds, thereby perfecting the theoretical framework in this field.

Core Contributions

1. Main Theorem (Theorem 1.1 = Theorem 5.4): Proves that for Q-factorial canonical Fano threefolds X with Picard number one, $(-K_X)^3 \leq 72$, and completely characterizes when equality holds
2. Kawamata-Miyaoka Type Inequality (Theorem 1.2, Theorem 3.8): Establishes a more general Kawamata-Miyaoka type inequality for ε-lc Fano varieties, generalizing previous results Liu19, LL25, LL24
3. Technical Innovations:
   • Develops Harder-Narasimhan filtration methods for handling non-semistable tangent bundles
   • Establishes precise formulas for the difference between Chern classes and generalized Chern classes (Theorem 4.6)
   • Provides integrality relations between degree and Fano index (Theorem 4.2)
4. Applications: Resolves remaining cases from Lai21 concerning Q-factorial terminal weak Fano threefolds with Picard number 2 (Corollary 5.6)

Methodology Details

Task Definition

Input: A Q-factorial canonical Fano threefold X with Picard number $\rho(X) = 1$
Objective: Prove $(-K_X)^3 \leq 72$ and characterize when equality holds
Constraints: X has canonical singularities (i.e., discrepancy $a(E,X) \geq 0$ for all divisors)

Overall Strategy Architecture

The proof strategy in the paper consists of three main steps:

Step 1: Establishing Kawamata-Miyaoka Type Inequality

By studying the stability of the tangent bundle $T_X$ and foliation theory, establish the relationship between degree and Chern classes:

Theorem 3.8: For Q-factorial canonical Fano threefolds X, let $q = q_Q(X)$ be the Q-Fano index, then $c_1(X)^3 \leq \begin{cases} \frac{16}{5}\hat{c}_2(X) \cdot c_1(X) & \text{if } q \leq 5\\ \frac{4q^2}{q^2 + 2q - 4}\hat{c}_2(X) \cdot c_1(X) & \text{if } q \geq 6 \end{cases}$

Step 2: Connecting Chern Classes to Singularity Data

Using Reid's basket and Riemann-Roch formula, establish the relationship between Chern classes and singularities:

Theorem 4.6: For canonical projective threefolds X and any Q-Cartier divisor H, $c_2(X) \cdot H - \hat{c}_2(X) \cdot H = \sum_{C \subset \text{Sing}(X)} \left(e_C - \frac{1}{g_C}\right)(H \cdot C)$

where $e_C, g_C$ are determined by singularity types.

Step 3: Combinatorial Analysis and Case Classification

Complete the proof through refined number-theoretic analysis and case classification.

Key Technical Modules

Module 1: Foliation Theory (Section 3.2)

For rank-1 subsheaves $\mathcal{L}$ of the tangent bundle $T_X$, prove:

Proposition 3.6: For ε-lc Fano varieties X with Picard number one, the divisor class $c_1(X) - (1+\varepsilon)c_1(\mathcal{L})$ is nef.

Technical Points:

• Use Bogomolov-McQuillan criterion to prove algebraic integrability of foliations
• Control multiplicities of exceptional divisors through geometric analysis of leaf families (Proposition 3.5)
• Combine discrepancy estimates for ε-lc singularities to obtain slope bounds

Module 2: Q-Variant of Langer Inequality (Section 3.1)

Theorem 3.2: For rank r reflexive sheaves $\mathcal{E}$, $(\alpha_1 \cdots \alpha_{n-2} \cdot \alpha_{n-1}^2) \cdot (\hat{\Delta}(\mathcal{E}) \cdot \alpha_1 \cdots \alpha_{n-2}) + r^2(\mu_{\max} - \mu)(\mu - \mu_{\min}) \geq 0$

where $\hat{\Delta}(\mathcal{E}) = 2r\hat{c}_2(\mathcal{E}) - (r-1)c_1(\mathcal{E})^2$ is the Q-Bogomolov discriminant.

Proof Strategy:

1. Decompose reflexive sheaves using Harder-Narasimhan filtrations
2. Apply Q-Bogomolov-Gieseker inequality to semistable quotients
3. Derive the inequality through Hodge index theorem and algebraic relations on slopes

Module 3: Integrality Relations between Degree and Fano Index (Section 4.1)

Theorem 4.2: Let $-K_X \equiv qA$, and $J_A$ be the minimal positive integer such that $J_A A$ is Cartier in codimension 2, then:

1. $J_A r^{n-2}(-K_X)^n/q^2$ is a positive integer
2. If $-K_X \sim qA$, then $J_A | q$ and $q | r^{n-2}(-K_X)^n$

This provides strong divisibility constraints on the degree.

Technical Innovations

1. Systematic Treatment of Non-Semistable Cases: Through refined analysis of Harder-Narasimhan filtrations, handle cases where the tangent bundle is non-semistable, which is the key obstacle in generalizing from Gorenstein to Q-factorial cases
2. Geometric Applications of Foliation Theory: Innovatively use geometric properties of algebraically integrable foliations (particularly precise estimates of exceptional divisor multiplicities) to control Chern class inequalities
3. Precise Computation of Singularity Data: Establish explicit formulas connecting Chern class differences to Du Val singularity types, enabling precise numerical analysis
4. Exploitation of Integrality Constraints: Systematically use integrality and torsion properties of Weil divisor class groups, performing reduction through index covering techniques

Experimental Setup

This is a pure mathematics theory paper without numerical experiments. The proof employs the following verification strategy:

Case Classification Verification

Through enumeration of Reid basket possibilities:

• Singularity Types: Consider all possible $R_X$ combinations (11 cases)
• Degree Ranges: Use Reid-Riemann-Roch formula to determine possible values of $r_X c_1(X)^3$
• Index Constraints: Combine divisibility of Fano index for screening

Numerical Verification Tables

The paper constructs detailed verification tables (Tables 1-5):

• Table 1: Enumerates all $(R_X, r_X c_1^3)$ combinations satisfying basic inequalities
• Tables 2-4: Progressively eliminate impossible cases through Kawamata-Miyaoka inequality
• Table 5: Complete analysis of Gorenstein case

Experimental Results

Main Results

Theorem 5.4 (Main Theorem): Let X be a Q-factorial canonical Fano threefold with Picard number one, then: $(-K_X)^3 \leq 72$ Equality holds if and only if $X \cong \mathbb{P}(1,1,1,3)$ or $\mathbb{P}(1,1,4,6)$.

Key Intermediate Results

1. Lemma 5.3: For varieties with $q_Q(X) \leq 6$, $(-K_X)^3 \leq 72$, with equality only for $\mathbb{P}(1,1,1,3)$
2. Corollary 5.5: For canonical weak Fano threefolds with $q_Q(X) \geq 7$, $(-K_X)^3 \leq 72$
3. Corollary 5.6: Resolves the open problem from Lai21: Q-factorial terminal weak Fano threefolds with Picard number 2 satisfy $(-K_X)^3 \leq 72$

Verification Process Analysis

Case 1: $q_Q(X) \leq 6$

• Use $c_2(X) \cdot c_1(X) = 24 - \sum(r_i - 1/r_i)$ and Kawamata-Miyaoka inequality
• Obtain $c_1(X)^3 \leq 72$ or derive contradiction

Case 2: $q_Q(X) \geq 7$

• Reduce to torsion-free case through index covering
• Enumerate 11 possible types of $R_X$
• Use divisibility of $J_A$ and prime factorization to eliminate all cases with $(-K_X)^3 > 72$

Elimination Mechanism

The paper eliminates impossible cases through three-layer screening:

1. First Layer: Integrality constraints from Reid formula
2. Second Layer: Basic Kawamata-Miyaoka inequality (Theorem 3.8)
3. Third Layer: Refined Kawamata-Miyaoka inequality (Inequality 4.8)

Related Work

Historical Development

1. Smooth Case: Iskovskikh-Prokhorov IP99 classification theory, where anti-canonical degree plays a central role
2. Gorenstein Case: Prokhorov Pro05 proved the optimal bound $(-K_X)^3 \leq 72$
3. General Canonical Case: Jiang-Zou JZ23 proved upper bound 324, but conjectured the optimal bound to be 72
4. Kawamata-Miyaoka Type Inequalities:
   • Liu Liu19: Fano manifold case
   • Liu-Liu LL25, LL24: Initial results for canonical singularities
   • This paper: Most general ε-lc case

Positioning of This Paper

This paper completes the generalization from Gorenstein to Q-factorial, resolving a core conjecture in the field. Advantages over related work:

1. Optimality: Achieves the conjectured optimal bound of 72
2. Completeness: Completely characterizes when equality holds
3. Generality: Methods apply to more general ε-lc Fano varieties

Technical Comparison

• vs Prokhorov Pro05: Must handle non-Gorenstein singularities with more complex Chern class computations
• vs Jiang-Zou JZ23: Obtains more precise bounds through Kawamata-Miyaoka inequality
• vs Liu-Liu LL25: Develops systematic treatment of non-semistable cases

Conclusions and Discussion

Main Conclusions

1. Completeness of Theorem: Proves the optimal upper bound of 72 for anti-canonical degrees of Q-factorial canonical Fano threefolds with Picard number one
2. Precision of Characterization: Completely determines varieties achieving the bound: weighted projective spaces $\mathbb{P}(1,1,1,3)$ and $\mathbb{P}(1,1,4,6)$
3. Generality of Methods: The established Kawamata-Miyaoka type inequality applies to broader classes of ε-lc Fano varieties

Limitations

1. Picard Number Restriction: Main theorem only applies to Picard number one case; higher Picard number cases remain open
2. Q-Factorial Assumption: Requires Q-factorial assumption; completely general canonical singularities case unresolved
3. Dimension Restriction: Methods highly depend on special properties of dimension three (Reid basket, Du Val singularity classification, etc.); generalization to higher dimensions requires new ideas
4. Combinatorial Nature of Proof: Final proof involves extensive case classification and numerical verification, lacking unified conceptual explanation

Future Directions

1. Higher Picard Number Cases: Study degree bounds for Q-factorial canonical Fano threefolds with Picard number $\geq 2$
2. Weak Fano Varieties: Generalize to weak Fano varieties (nef and big anti-canonical divisor)
3. Higher Dimensional Generalization: Explore degree bound problems for Fano varieties in dimension four and higher
4. Classification Theory: Use degree bounds to advance classification of canonical Fano threefolds
5. Moduli Spaces: Study moduli space structures of Fano varieties with bounded degree

In-Depth Evaluation

Strengths

1. Methodological Innovation

• Novel Application of Foliation Theory: First systematic application of algebraically integrable foliation theory to degree bound problems for Fano varieties, particularly precise control of exceptional divisor multiplicities (Proposition 3.5)
• Refined Harder-Narasimhan Filtration Analysis: Complete and optimized treatment of three cases $(l, r_1) = (2,1), (2,2), (3,1)$ for rank-3 vector bundles
• Systematic Use of Integrality Constraints: Creative combination of integrality and torsion properties of Weil divisor class groups with divisibility relations of Fano index

2. Technical Completeness

• Refinement of Theorem 3.8: Compared to the general form in Theorem 1.2, provides more precise coefficients $\frac{4q^2}{q^2+2q-4}$ for the case $q \geq 6$
• Chern Class Difference Formula: Theorem 4.6 provides explicit expression for $c_2(X) - \hat{c}_2(X)$, connecting geometry with combinatorial data
• Reduction Techniques: Index covering reduction in Lemma 5.1 ensures reduction to torsion-free case

3. Optimality of Results

• Achieves the conjectured optimal bound, completely characterizing extremal cases
• Proves that the Gorenstein bound remains optimal in the Q-factorial setting

4. Clarity of Exposition

• Clear structure with layered progression from general theory to specific applications
• Detailed tables (Tables 1-5) make verification process traceable
• Remark 5.7 provides alternative path avoiding Pro05

Weaknesses

1. Combinatorial Complexity of Proof

• Final proof relies on extensive case enumeration and numerical verification (Tables 1-5 involve dozens of cases)
• Lacks unified conceptual explanation for why the bound is exactly 72
• For $q \geq 7$ cases, requires table-by-table elimination without structural understanding

2. Technical Dependency

• Highly dependent on Reid basket theory and fine classification of Du Val singularities
• Foliation theory proofs (Lemma 3.4, Proposition 3.5) are quite technical with insufficient geometric intuition
• Theorem 4.6 proof requires delicate topological Euler number calculations

3. Generalization Obstacles

• Methods are inherently three-dimensional, difficult to generalize to higher dimensions (Reid basket and Du Val singularity classification are three-dimensional specific)
• Methods do not directly apply to Picard number $\geq 2$ cases
• ε-lc assumption is essential in some places (Example 3.7 shows cannot simply weaken to klt)

4. Unresolved Issues

• Completely general canonical singularities (non-Q-factorial) case unresolved
• No complete classification of varieties achieving bound 72 (only two examples known)
• Unclear which degrees close to the bound (e.g., 70, 68) can be realized

Impact Assessment

Contribution to Field

1. Theory Perfection: Resolves core conjecture in the field, completing generalization from Gorenstein to Q-factorial
2. Methodology: Established Kawamata-Miyaoka type inequality framework will influence subsequent research
3. Technical Arsenal: Developed foliation and Harder-Narasimhan filtration techniques applicable to other problems

Practical Value

• Classification Theory: Provides fundamental bounds for classification of canonical Fano threefolds
• Moduli Spaces: Degree boundedness is key property for constructing moduli spaces
• Birational Geometry: Deeper understanding of Fano varieties in minimal model program

Reproducibility

• Paper provides complete proof details and verification tables
• Primarily relies on established theories (Reid formula, foliation theory, etc.)
• Numerical verification portions can be checked with computer assistance

Applicable Scenarios

1. Algebraic Geometry Research:
   • Classification and moduli space theory of Fano varieties
   • Boundedness problems in minimal model program
   • Singularity theory and Chern class computations
2. Related Problems:
   • Degree bounds for Fano varieties with higher Picard number
   • Degree studies for weak Fano and log Fano varieties
   • Topological invariant estimates for Calabi-Yau varieties
3. Technical Applications:
   • Birational geometry problems requiring refined Chern class estimates
   • Geometric problems involving singularities and foliations
   • Applications of vector bundle stability in algebraic variety geometry

Overall Assessment

This is a high-level algebraic geometry paper that resolves an important conjecture in the field. The paper's main strengths are:

1. Establishes powerful general theoretical framework (Kawamata-Miyaoka type inequality)
2. Develops innovative technical methods (foliation applications, Harder-Narasimhan filtration analysis)
3. Achieves optimal results with complete characterization of extremal cases

Main limitations are:

1. Proof involves extensive case analysis; conceptual understanding needs deepening
2. Method generalizability limited by special properties of three-dimensional geometry
3. Some technical details quite complex with insufficient geometric intuition

Nevertheless, the paper has landmark significance in the field, providing solid foundation for subsequent research. Expected to have sustained impact in algebraic geometry, particularly in Fano variety theory and minimal model program.

References

The paper cites 18 key references, primarily including:

1. Pro05 Prokhorov: Degree bounds for Gorenstein case, direct predecessor of this work
2. JZ23 Jiang-Zou: Upper bound 324 for general canonical case
3. LL25, LL24 Liu-Liu: Prior work on Kawamata-Miyaoka type inequalities
4. Rei87 Reid: Canonical singularity theory and Reid basket
5. KM98 Kollár-Mori: Classical textbook on birational geometry of algebraic varieties
6. GKPT19 Greb-Kebekus-Peternell-Taji: Generalized Chern class theory
7. Dru21 Druel: Foliation theory
8. Lai21 Lai: Picard number 2 case; open problems resolved in this paper

These references form the theoretical foundation and research background of this paper.