On quasi-Albanese morphisms for log canonical Calabi-Yau pairs

Basic Information

• Paper ID: 2511.14580
• Title: On quasi-Albanese morphisms for log canonical Calabi-Yau pairs
• Author: Yiming Zhu (University of Science and Technology of China)
• Classification: math.AG (Algebraic Geometry)
• Publication Date: November 18, 2025 (arXiv preprint)
• Paper Link: https://arxiv.org/abs/2511.14580
• Mathematics Subject Classification: 14E30

Abstract

This paper studies quasi-Albanese morphisms for log canonical Calabi-Yau pairs. This represents an important research direction in birational geometry and minimal model theory within algebraic geometry.

Research Background and Motivation

1. Core Problem

The central question addressed in this paper is: For a log canonical Calabi-Yau pair $(X, D_X)$ (satisfying $K_X + D_X \sim_{\mathbb{Q}} 0$), what are the properties of the quasi-Albanese morphism of its open part $X \setminus D_X$?

2. Significance of the Problem

The importance of this problem manifests in several aspects:

• Generalization of Classical Results: Kawamata proved in 1985 that for complex projective varieties with canonical singularities where $K_X$ is torsion, the Albanese morphism is an étale fibration. This paper extends this result to the log canonical setting.
• Fundamental Questions in Logarithmic Geometry: Understanding the structure of log Calabi-Yau pairs is one of the central problems in modern birational geometry, with deep connections to mirror symmetry and Gromov-Witten theory.
• Fibration Structure: Quasi-Albanese morphisms provide a way to understand complex algebraic varieties through fibration, allowing dimension reduction.

3. Limitations of Existing Methods

• The classical theorems of Kawamata and Fujino (Theorem 1) establish that when the log Kodaira dimension $\kappa(X_0) = 0$, the quasi-Albanese morphism is dominant with irreducible general fibers. However, other geometric properties of the morphism (such as codimension-one properties, flatness, and semistability) lack systematic study.
• For the log canonical singularity case, deeper techniques from minimal model theory are required.

4. Research Motivation

Inspired by work of Iitaka, Kawamata, Zhang, Fujino, BFPT and others, the author systematically studies codimension-one properties of quasi-Albanese morphisms for log canonical Calabi-Yau pairs.

Core Contributions

The main contributions of this paper include:

1. Main Theorem (Theorem 2): For a log smooth pair $(X, D_X)$ with $D_X$ reduced and $K_X + D_X \sim_{\mathbb{Q}} 0$, the quasi-Albanese morphism $\alpha: X_0 \to G$ satisfies:
   • Surjectivity in codimension one
   • Flatness in codimension one
   • Semistability in codimension one
2. Structure Theorem (Theorem 3): For Q-factorial dlt pairs, a crepant birational contraction $\mu: (W, D_W) \dashrightarrow (X, D_X)$ and fibration $g: W \to \mathbb{P}_A$ are constructed such that:
   • The canonical bundle formula has standard form
   • Restriction to $g^{-1}G$ is locally stable
   • The vertical part equals exactly $g^{-1}H$
3. Concrete Applications:
   • When $q(X) = 0$ and $q(X, D_X) = \dim X$, $(X, D_X)$ is crepant birationally equivalent to $(\mathbb{P}^n, (\prod_i z_i = 0))$
   • When $q(X, D_X) - q(X) \geq 2$, $D_X$ is connected
4. Reduction of Orbifold Fundamental Group (Proposition 22): The virtual nilpotency conjecture for the orbifold fundamental group of log canonical Calabi-Yau pairs is reduced to the case $q(X, D_X) = 0$.

Detailed Methodology

Task Definition

Input: A log canonical Calabi-Yau pair $(X, D_X)$, namely:

• $X$ is a complex projective variety
• $D_X$ is a reduced divisor
• $(X, D_X)$ is log canonical (lc)
• $K_X + D_X \sim_{\mathbb{Q}} 0$ (numerical Calabi-Yau condition)

Output: Description of the geometric properties of the quasi-Albanese morphism $\alpha: X \setminus D_X \to G$, particularly its codimension-one properties.

Core Technical Framework

The proof relies on several profound techniques:

1. Construction of Quasi-Albanese Morphism (Section 8)

For a log smooth pair $(X, D_X = \sum_{i=1}^m D_i)$, let $q = q(X) = \dim H^1(\mathcal{O}_X)$, $\bar{q} = q(X, D_X) = \dim H^0(\Omega^1_X(\log D_X))$, $d = \bar{q} - q$.

Construction Steps:

• Select $\eta_i \in H_1(X^0, \mathbb{Z})$ ($1 \leq i \leq d$) as a basis for the free part of $\ker(H_1(X^0, \mathbb{Z}) \to H_1(X, \mathbb{Z}))$
• Select $\omega_i \in H^0(\Omega^1_X)$ and $\varphi_j \in H^0(\Omega^1_X(\log D_X))$ satisfying period conditions
• Define the quasi-Abelian variety as: $G := \mathbb{C}^{\bar{q}} / \left(\sum_i \mathbb{Z}A_i + \sum_j \mathbb{Z}B_j\right)$
• Possessing Chevalley decomposition: $0 \to (\mathbb{C}^*)^d \to G \to A \to 0$

Compactification Construction: For representations corresponding to $L_j \in \text{Pic}^0(A)$, $G$ has natural compactification: $\mathbb{P}_A := \mathbb{P}_A(\mathcal{O}_A \oplus L_1^{-1}) \times_A \cdots \times_A \mathbb{P}_A(\mathcal{O}_A \oplus L_d^{-1})$

2. Minimal Model Theory Techniques (Theorem 15)

This is the core technical theorem of the paper, with proof divided into 5 parts:

(1) Control of Exceptional Divisors: Proves that any divisor exceptional on $\mathbb{P}_A$ must be contained in $E_Z + f^{-1}H$. Key uses:

• Fujino's subadditivity theorem (Theorem 13)
• Relationship between pseudoeffectivity and exceptionality
• Lazarsfeld lemma

(2) Existence of Non-Vertical Divisors: For any prime divisor $P \subset \mathbb{P}_A \setminus H$, there exists a prime divisor $Q \subset Z$ not contained in $E_Z$ that dominates $P$. Proof uses proof by contradiction and estimates of log Kodaira dimension.

(3) Existence of Good Minimal Models: Proves through establishing:

• The vertical part $E_Z^v$ is very exceptional on $\mathbb{P}_A$
• For general fiber $F$, $\kappa(K_F + D_Z|_F) = 0$
• Application of Gongyo's abundance theorem and Birkar's termination theorem

Key formula: $(f_*\mathcal{O}_Z(mE_Z))^\vee = \mathcal{O}_{\mathbb{P}_A}$ holds for sufficiently divisible $m > 0$.

(4) Standard Form of Canonical Bundle Formula: Proves the canonical bundle formula must have the form: $K_W + D_W \sim_{\mathbb{Q}} g^*(K_{\mathbb{P}_A} + H)$

Uses inversion of adjunction to prove local stability.

(5) Inductive Argument:

• Case 1 ($\dim \mathbb{P}_A - \dim A = 1$): Uses BFPT's result; Albanese morphism is a fibration
• Case 2 ($\dim \mathbb{P}_A - \dim A > 1$): Induction on dimension difference

3. Differential Forms Theory (Section 2.1)

Uses GKKP theory (Theorem 4): $\pi_* \Omega^k_Y(\log D_Y) \text{ is reflexive, and } H^0(Y, \Omega^k_Y(\log D_Y)) \cong H^0(X_0, \Omega^k_{X_0}(\log \lfloor D_X|_{X_0} \rfloor))$

This ensures the invariant $q(X, D_X)$ is invariant under birational transformations (Lemma 6).

Technical Innovations

1. Systematic Study of Codimension-One Properties: First systematic study of codimension-one properties of quasi-Albanese morphisms (surjectivity, flatness, semistability).
2. Construction of Crepant Birational Contractions: Construction of special birational contractions via minimal model theory that preserve log canonical properties.
3. Inductive Framework: Establishes framework for induction via $q(X, D_X)$, allowing dimension reduction.
4. Dimension Difference Induction: In the proof of Theorem 15, induction on $\dim \mathbb{P}_A - \dim A$ is a key innovation.

Experimental Setup

As a pure mathematics theory paper, this work contains no experimental section but rather establishes theoretical results through rigorous mathematical proof.

Example Analysis (Example 10)

The paper provides three important classes of examples:

Example 1: $X = \mathbb{P}^2$

• If $D_X$ consists of 3 lines, then $q(X, D_X) = 3$
• If $D_X$ is an elliptic curve, then $q(X, D_X) = 0$
• If $D_X = Q + L$ (conic + line), then $q(X, D_X) = 1$, quasi-Albanese morphism is $x \mapsto \frac{l^2}{q}$

Example 2: $X = \mathbb{P}^1 \times \mathbb{P}^1$, $D_X = F_1 + F_2 + \Delta$

• Quasi-Albanese morphism is $(z_1, z_2) \mapsto \frac{1}{z_1} - \frac{1}{z_2}$

Example 3: $X = \mathbb{P}_A(\mathcal{O}_A \oplus L)$

• If $L \in \text{Pic}^0(A)$, then $q(X, D_X) = \dim A + 1$
• If $L \notin \text{Pic}^0(A)$, then $q(X, D_X) = \dim A$

Main Results Analysis

Significance of Theorem 2

For log smooth Calabi-Yau pairs, the quasi-Albanese morphism $\alpha: X_0 \to G$ satisfies:

1. Codimension-One Surjectivity: For any prime divisor $P \subset G$, there exists a prime divisor $Q \subset X_0$ dominating $P$
   • This shows the morphism is "as surjective as possible" in codimension one
2. Codimension-One Flatness: For any prime divisor $Q \subset X_0$, if $Q$ does not dominate $G$, then $\alpha(Q)$ has codimension 1
   • This controls dimension jumps in fibers
3. Codimension-One Semistability: For any prime divisor $P \subset G$, the pair $(X, \alpha^*P)$ is log canonical near a general point of $P$
   • This is a mild singularity condition

Structure of Theorem 3

This theorem provides a standardized geometric model:

W \ D_W ----h----> G = P_A \ G
    |                    |
    g                    p
    |                    |
    v                    v
   P_A -------p-------> A

Where:

• $\mu: (W, D_W) \dashrightarrow (X, D_X)$ is a crepant birational contraction
• The vertical part of $g$ equals exactly $g^{-1}H$
• Canonical bundle formula is standardized: $K_W + D_W \sim_{\mathbb{Q}} g^*(K_{\mathbb{P}_A} + H)$

Corollary Applications

Corollary 1: If $q(X) = 0$ and $q(X, D_X) = \dim X$, then $(X, D_X)$ is crepant birationally equivalent to $(\mathbb{P}^n, (\prod_i z_i = 0))$.

Corollary 2: If $q(X, D_X) - q(X) \geq 2$, then $D_X$ is connected.

These results are important for understanding the classification of log Calabi-Yau pairs.

Related Work

1. Classical Foundations

• Iitaka (1976, 1977): Introduced log Kodaira dimension and quasi-Albanese morphism concepts
• Kawamata (1981, 1985): Proved étale fibration property of Albanese morphisms
• Fujino (2024): Recent systematic study of quasi-Albanese morphisms

2. Minimal Model Theory

• Birkar (2012): Existence of log canonical flips and special LMMP
• Hacon-Xu (2013): Existence of log canonical closures
• Gongyo (2011): Abundance theorem for dlt pairs with numerical log Kodaira dimension zero

3. Differential Forms Theory

• GKKP (2011): Theory of differential forms on log canonical spaces
• Maehara (1986), Fujino (2017): Subadditivity theorems

4. Recent Developments

• BFPT (2024): Strong counterexamples to log canonical Beauville-Bogomolov decomposition
• GLM (2023): Fundamental groups of log Calabi-Yau surfaces

Positioning of This Paper

This paper extends Kawamata's classical results to the log canonical setting and systematically studies codimension-one properties. Compared to related work, the advantages are:

• More general singularity conditions (log canonical vs. canonical)
• More refined geometric properties (three codimension-one properties)
• Provides inductive framework for further research

Conclusions and Discussion

Main Conclusions

1. Geometric Structure Theorem: Quasi-Albanese morphisms of log canonical Calabi-Yau pairs possess good codimension-one properties and can be standardized through crepant birational contractions.
2. Fibration Structure: There exists a fibration to a $(\mathbb{P}^1)^d$-bundle whose canonical bundle formula has standard form.
3. Inductive Principle: Many problems can be reduced via induction on $q(X, D_X)$ to the case $q(X, D_X) = 0$.

Limitations

1. Technical Restrictions:
   • Main results require Q-factoriality and dlt conditions; further work needed for general log canonical pairs
   • Proofs rely on deep results from minimal model theory, with high technical threshold
2. Dimension Restrictions: While theoretically applicable to arbitrary dimension, concrete applications may be limited by computational complexity.
3. Characteristic Restrictions: Works only over the complex numbers; positive characteristic cases require different techniques.

Future Directions

Research directions explicitly indicated in the paper:

1. Orbifold Fundamental Group Conjecture (Conjecture 20): Prove that the orbifold fundamental group of log canonical Calabi-Yau pairs is virtually nilpotent. Proposition 22 reduces the problem to the case $q(X, D_X) = 0$.
2. Classification Problems: Use this framework to classify low-dimensional log Calabi-Yau pairs.
3. Generalization to More General Singularities: Study non-Q-factorial or non-dlt cases.
4. Connections to Mirror Symmetry: Explore applications of these results in mirror symmetry.

In-Depth Evaluation

Strengths

1. Theoretical Depth:
   • Extends Kawamata's classical results to log canonical setting with significantly increased technical difficulty
   • Systematically studies codimension-one properties, filling a gap in this direction
   • Proof techniques synthesize minimal model theory, differential forms theory, Hodge theory and other fields
2. Clear Structure:
   • Well-organized paper with rigorous logical flow from preliminaries to main theorems
   • Rich examples aid understanding
   • Standardization of canonical bundle formula provides unified treatment framework
3. Technical Innovation:
   • Construction of crepant birational contractions is key innovation
   • Dimension difference induction argument is clever
   • Reduction of orbifold fundamental group problem to simpler case demonstrates method's power
4. Application Value:
   • Provides inductive framework for further research
   • Important for understanding geometric structure of log Calabi-Yau pairs
   • Potential connections to fundamental groups, classification theory, etc.

Weaknesses

1. Technical Threshold:
   • Proofs depend on extensive advanced techniques, difficult for non-specialists to fully understand
   • Some key steps (such as proof of Theorem 15) are highly technical
2. Concrete Calculations:
   • Lacks more examples in dimension ≥ 3
   • Unclear how to effectively compute quasi-Albanese morphisms for specific pair types
3. Completeness:
   • Orbifold fundamental group conjecture only partially resolved (reduced but not fully proved)
   • Optimality of some results not discussed

Impact Assessment

1. Theoretical Contribution:
   • Important contribution to birational geometry and minimal model theory
   • Provides new tools for research on log Calabi-Yau pairs
   • Expected to be widely cited in subsequent work in the field
2. Practical Value:
   • Provides inductive framework for concrete problems
   • Applicable to classification of log Calabi-Yau pairs
   • Related to fundamental groups, moduli spaces, etc.
3. Reproducibility:
   • As mathematical proof, logically rigorous with strong verifiability
   • All cited results have explicit references
   • Technical details are sufficient (though complex)

Applicable Scenarios

1. Theoretical Research:
   • Classification and moduli space research for log Calabi-Yau pairs
   • Fibration problems in birational geometry
   • Fundamental group and topological property research
2. Related Fields:
   • Mirror symmetry theory
   • Gromov-Witten theory
   • Derived categories and stability conditions
3. Generalization Directions:
   • Birational geometry of high-dimensional algebraic varieties
   • Analogous theory in positive characteristic
   • Analogous problems in noncommutative geometry

References

The paper cites important literature in the field, with key references including:

1. Kaw85 Kawamata: Minimal models and the Kodaira dimension of algebraic fiber spaces
2. Iit76 Iitaka: Logarithmic forms of algebraic varieties
3. Fuj24 Fujino: On quasi-albanese maps
4. GKKP11 Greb-Kebekus-Kovács-Peternell: Differential forms on log canonical spaces
5. Bir12 Birkar: Existence of log canonical flips and a special LMMP
6. BFPT24 Bernasconi-Filipazzi-Patakfalvi-Tsakanikas: Strong counterexamples to log canonical Beauville-Bogomolov decomposition

Summary

This is a high-quality algebraic geometry theory paper that systematically studies geometric properties of quasi-Albanese morphisms for log canonical Calabi-Yau pairs. The paper employs profound techniques and yields important results, providing new tools and perspectives for the field. Main contributions include three theorems on codimension-one properties, construction of crepant birational contractions, and establishment of an inductive framework. While the technical threshold is high, this is essential reading for researchers in birational geometry and minimal model theory. The paper's partial resolution of the orbifold fundamental group conjecture also points directions for future research.