Algebraic subgroups of the group of birational transformations of ruled surfaces

Basic Information

• Paper ID: 2111.09697
• Title: Algebraic subgroups of the group of birational transformations of ruled surfaces
• Author: Pascal Fong (Universität Basel)
• Classification: math.AG (Algebraic Geometry)
• Published Journal: Épijournal de Géométrie Algébrique, Volume 7 (2023), Article No. 13
• Paper Link: https://arxiv.org/abs/2111.09697

Abstract

This paper provides a complete classification of maximal algebraic subgroups of the group of birational transformations Bir(C×P¹) when C is a smooth projective curve of positive genus.

Research Background and Motivation

Problem Context

1. Historical Development: The study of algebraic subgroups originated with Enriques (1893) on the classification of maximal connected algebraic subgroups of Bir(P²), which was later completed by Blanc for all maximal algebraic subgroups of Bir(P²).
2. Core Problem: How to classify maximal algebraic subgroups of Bir(C×P¹) when the base curve C has positive genus? This would complete the classification for surfaces of Kodaira dimension -∞.
3. Technical Challenges:
   • Traditional equivariant compactification results of Sumihiro apply only to linear algebraic groups and cannot be directly applied
   • Complexity arising from non-rational base curves must be addressed
   • The automorphism groups of conic bundles require detailed study
4. Research Significance: Completing the classification of maximal algebraic subgroups for all surfaces of Kodaira dimension -∞ is a fundamental problem in birational geometry.

Core Contributions

1. Complete Classification Theorem: Provides a complete classification of all maximal algebraic subgroups of Bir(C×P¹) (Theorem A)
2. Technical Innovations:
   • Extends equivariant compactification theory to non-linear and non-connected algebraic group actions
   • Develops fundamental techniques of equivariant minimal model program
3. Structural Analysis: Provides detailed analysis of automorphism group structures for various types of conic bundles (ruled surfaces, exceptional conic bundles, (Z/2Z)²-conic bundles)
4. Comparative Results: Demonstrates fundamental differences from the rational case—when C has positive genus, not every algebraic subgroup is contained in some maximal subgroup (Corollary B)

Detailed Methodology

Problem Setup

Classify all maximal algebraic subgroups G ⊂ Bir(C×P¹), where C is a smooth projective curve of genus g ≥ 1.

Main Strategy

Classical three-step approach:

1. Regularization: Make G act on some smooth projective surface
2. Equivariant Compactification: Find a G-equivariant compactification
3. Equivariant Minimal Model Program: Obtain a G-minimal fibration

Technical Framework

1. Equivariant Compactification (Proposition 2.5)

For a surface X and algebraic subgroup G ⊂ Bir(X), there exist a smooth projective surface Y and a birational map ψ: X⤏Y such that ψGψ⁻¹ ⊂ Aut(Y).

Key Innovation:

• Does not depend on assumptions that G is linear or connected
• Uses Brion's results combined with fundamental blowing-up techniques
• Handles finiteness of base point sets via Lemma 2.4

2. G-Equivariant Minimal Model Program (Proposition 2.6)

If (G,X) is minimal and X is birationally equivalent to C×P¹, then X is a conic bundle over C.

3. Conic Bundle Classification Strategy

Reduces the study to automorphism groups of three classes of conic bundles:

• Ruled surfaces
• Exceptional conic bundles
• (Z/2Z)²-conic bundles

Technical Innovations

1. Application of Segre Invariants: Utilizes Segre invariants S(X) = min{σ² | σ is a section} for refined classification of ruled surfaces
2. Determinant Analysis: Distinguishes different types of conic bundles by analyzing the determinant det(f) ∈ k(C)/（k(C)）² of automorphisms
3. Normalizer Techniques: Uses normalizers of elements in PGL(2,k(C)) to analyze (Z/2Z)²-actions

Main Results

Theorem A: Classification of Maximal Algebraic Subgroups

When char(k) ≠ 2 and C is a smooth projective curve of genus g ≥ 1, the maximal algebraic subgroups of Bir(C×P¹) fall into the following six classes:

1. Trivial Case: Aut(C×P¹) ≅ Aut(C) × PGL(2,k)
2. Exceptional Conic Bundles: Aut(X), where X is an exceptional conic bundle satisfying specific conditions
   • Exact sequence: 1 → Gₘ⋊Z/2Z → Aut(X) → H
3. Singular (Z/2Z)²-Conic Bundles: Aut(X), where X is a (Z/2Z)²-conic bundle with at least one singular fiber
   • Exact sequence: 1 → (Z/2Z)² → Aut(X) → H
4. (Z/2Z)²-Ruled Surfaces: Aut(X), where X is a (Z/2Z)²-ruled surface (hence S(X) > 0)
   • Exact sequence: 1 → (Z/2Z)² → Aut(X) → Aut(C)
   • When g = 1, there exists a unique A₁ satisfying S(A₁) = 1
5. A₀ over Elliptic Curves: When g = 1, Aut(A₀), where A₀ is the unique indecomposable ruled surface with Segre invariant 0
   • Exact sequence: 1 → Gₐ → Aut(A₀) → Aut(C) → 1
6. Decomposable Ruled Surfaces: Aut(X), where X ≅ P(O_C(D)⊕O_C) is a non-trivial decomposable ruled surface with deg(D) = 0
   • When g ≥ 2, additional conditions apply: if 2D is principal, then G = Gₘ⋊Z/2Z, otherwise G = Gₘ

Corollary B: Differences in Containment Relations

Let k be an algebraically closed field of characteristic not equal to 2, and X a surface of Kodaira dimension -∞. Then every algebraic subgroup of Bir(X) is contained in some maximal subgroup if and only if X is rational.

Technical Details Analysis

Segre Invariant Theory for Ruled Surfaces

• Definition: S(S) = min{σ² | σ is a section of π}
• Classification Significance:
  • S(S) < 0: unique negative section exists, allowing construction of infinite increasing chains
  • S(S) = 0: distinguishes decomposable and indecomposable cases
  • S(S) > 0: indecomposable, automorphism group is finite

Characterization Conditions for Exceptional Conic Bundles

Equivalent characterizations of exceptional conic bundles (Lemma 3.3):

1. Exactly 2n singular fibers and two disjoint sections with self-intersection -n
2. Existence of birational morphism to a decomposable ruled surface with S(S) = -n
3. Existence of birational morphism to a decomposable ruled surface with S(S) = 0, obtained by blowing up 2n points in a specific manner

Structure of (Z/2Z)²-Conic Bundles

• Definition: AutC(X) ≅ (Z/2Z)², each non-trivial involution fixes an irreducible curve
• Characterization: All non-trivial elements have non-trivial determinant (Lemma 3.15)
• Maximality: Always maximal (Lemma 3.16)

Comparison with Related Work

Contrast with the Rational Case

• Blanc's Results: Every algebraic subgroup of Bir(P²) is contained in some maximal subgroup
• Present Findings: This no longer holds when C has positive genus; there exist algebraic subgroups not contained in any maximal subgroup

Development of Technical Methods

• Traditional Approach: Sumihiro's equivariant compactification applies only to linear algebraic groups
• Present Innovation: Extension to general algebraic groups using Brion's results and fundamental blowing-up techniques

Critical Evaluation

Strengths

1. Completeness: Provides exhaustive classification with no omitted cases
2. Technical Innovation: Develops new techniques for handling non-linear algebraic group actions
3. Clear Structure: Systematically addresses the problem through conic bundle classification
4. Deep Comparison: Reveals fundamental differences between positive genus and rational cases

Technical Contributions

1. Generalization of Equivariant Compactification: Equivariant compactification results without linearity assumptions
2. Systematic Application of Segre Invariants: Effective use in automorphism group classification
3. Determinant Method: Distinguishes different types of conic bundles through determinant analysis

Theoretical Significance

• Completes the classification of maximal algebraic subgroups for surfaces of Kodaira dimension -∞
• Provides important structural results for birational geometry
• Reveals profound influence of base curve genus on the structure of birational transformation groups

Limitations

1. Characteristic Restriction: Requires char(k) ≠ 2
2. Technical Complexity: Proofs involve multiple intricate technical lemmas
3. Abstract Constructions: Examples in certain cases are constructed abstractly

Conclusion and Prospects

This paper completes the full classification of maximal algebraic subgroups of birational transformation groups of ruled surfaces over positive genus curves, representing significant progress in the theory of birational transformation groups in algebraic geometry. The results reveal fundamental differences from the rational case and provide important theoretical foundations and technical tools for further study of higher-dimensional cases.

Future research directions may include:

1. Generalization to higher-dimensional fibrations
2. Investigation of other Kodaira dimension cases
3. Development of more general equivariant birational geometry theory