Log-Conformal Projective Manifolds

1. Problem Statement

The paper investigates the classification of smooth complex projective simple normal crossing (SNC) pairs $(X, \Delta)$ of dimension $n \ge 3$ endowed with an everywhere nondegenerate logarithmic conformal tensor.

• Log-Conformal Tensor: A section $g \in H^0(X, \text{Sym}^2 \Omega^1_X(\log \Delta) \otimes N)$, where $N$ is a line bundle, inducing a fibrewise isomorphism $TX(-\log \Delta) \xrightarrow{\sim} \Omega^1_X(\log \Delta) \otimes N$.
• Core Question: To what extent does the rigidity of classical holomorphic conformal geometry (where $\Delta = \emptyset$) persist when a boundary divisor $\Delta$ is introduced? Specifically, the authors seek to classify such pairs based on the numerical properties of the log-canonical divisor $K_X + \Delta$.

The study is divided into two regimes:

1. Non-nef regime: $K_X + \Delta$ is not nef.
2. Numerically trivial regime: $K_X + \Delta \equiv 0$.

2. Methodology

The authors employ a synthesis of birational geometry, deformation theory of rational curves, and differential geometric techniques.

• Minimal Rational Curves and Extremal Rays:

  • In the non-nef case, they utilize the Cone and Contraction Theorem to identify a $(K_X + \Delta)$-negative extremal ray $R$.
  • They analyze $(K_X + \Delta)$-minimal rational curves $f: \mathbb{P}^1 \to X$ spanning $R$.
  • They apply Langer's logarithmic cone and contraction theorem to control the intersection of these curves with the boundary $\Delta$, establishing that the reduced pullback divisor $(f^{-1}\Delta)_{\text{red}}$ has degree $\delta \le 1$.

• Splitting and Nullity Dichotomy:

  • The core technical tool is the analysis of the pullback bundle $f^*TX(-\log \Delta)$ along minimal rational curves.
  • Using the bend-and-break technique and the self-duality provided by the conformal tensor, they derive a sharp dichotomy in the splitting type of the bundle and the behavior of the "nullity section" $q_f = g(df^{\log}, df^{\log})$.
  • This leads to three distinct cases based on the degree of the pullback of the line bundle $N$ and the intersection number $\delta$.

• Varieties of Minimal Rational Tangents (VMRT):

  • For the case where the nullity section vanishes identically, the authors relate the geometry to the VMRT. They use results by Hwang and Mok to show that if the VMRT is a smooth quadric, the manifold is a quadric hypersurface.
  • For the "split-null" case, they analyze the geometry of maximal isotropic subbundles to construct fibrations.

• Bochner and Holonomy Arguments:

  • In the numerically trivial case, they assume the existence of a complete Ricci-flat Kähler metric on $M = X \setminus \Delta$.
  • They apply a Bochner extension principle and assumptions on the irreducibility of the restricted holonomy to force the holonomy group to be trivial, leading to a uniformization result.

• Cartan Geometry:

  • The authors interpret the classification results through the lens of logarithmic Cartan geometry, viewing the three outcomes as manifestations of different reductions of the conformal structure group.

3. Key Contributions and Results

A. Classification in the Non-NeF Regime (Theorem 1.1)

If $K_X + \Delta$ is not nef, exactly one of the following mutually exclusive alternatives occurs:

1. The Quadric Model: $\Delta = \emptyset$ and $X \cong Q^n$ (smooth quadric hypersurface).
   • Mechanism: The nullity section vanishes, and the VMRT is the standard null quadric.
2. The Projective-Hyperplane Model: $X \cong \mathbb{P}^n$ and $\Delta$ is a hyperplane.
   • Mechanism: The splitting of the tangent bundle is rigid ($\mathcal{O}(1)^{\oplus n}$), and the curve intersects the boundary at exactly one point with multiplicity 1.
3. The Maximal Isotropic Fibration: $n = 2m$ is even, and there exists a rational fibration $\pi: X \dashrightarrow Y$ with generic fibre $(\mathbb{P}^m, H)$ (where $H$ is a hyperplane).
   • Mechanism: This is a genuinely logarithmic phenomenon with no boundaryless analogue. The tangent bundle splits into two maximal isotropic subbundles of rank $m$, one of which corresponds to the vertical direction of the fibration. The pair $(X, \Delta)$ is locally modeled on a projective bundle $\mathbb{P}(V)$ where the boundary is a relative hyperplane.

B. Classification in the Numerically Trivial Regime (Theorem 1.2)

If $K_X + \Delta \equiv 0$, and assuming:

• $M = X \setminus \Delta$ carries a complete Ricci-flat Kähler metric satisfying a Bochner extension property and irreducible restricted holonomy.
• The induced flat connection has trivial monodromy.

Then:

• The restricted holonomy is trivial ($\{1\}$).
• $M$ is a semi-abelian variety.
• $(X, \Delta)$ is a toroidal compactification of $M$.
• The logarithmic tangent bundle $TX(-\log \Delta)$ is trivial.

C. Singular Models and Boundary Contractions (Section 7)

The authors demonstrate how these smooth models relate to singular varieties. If the boundary $\Delta$ is contracted to codimension $\ge 2$, the resulting singular variety $Y$ carries a reflexive conformal structure.

• Example: Generalized cones over abelian varieties arise as singular limits of the "Maximal Isotropic Fibration" case.
• Example: Quadric cones and cubic surfaces with specific singularities are shown to be log-conformal models arising from double or triple covers of $\mathbb{P}^2$ branched along specific divisors.

4. Significance

• Rigidity vs. Logarithmic Phenomena: The paper confirms that while logarithmic conformal geometry retains significant rigidity from the classical theory (yielding the quadric and projective models), the presence of a boundary introduces a new, distinct class of geometries: the maximal isotropic fibrations. This case is specific to the logarithmic setting and relies on the even-dimensional splitting of the tangent bundle.
• Unification of Geometric Structures: By interpreting the results via Cartan geometry, the authors unify the quadric, the affine-logarithmic projective space, and the split-null fibration as different manifestations of conformal structures, providing a deeper conceptual framework.
• Bridge to Singular Geometry: The work provides a rigorous pathway from smooth log-pairs to singular compact models with reflexive conformal structures, which is crucial for the Minimal Model Program (MMP) where singular models are inevitable.
• Uniformization: The numerically trivial result extends the classical uniformization of Calabi-Yau manifolds to the logarithmic setting, linking log-conformal tensors to semi-abelian varieties and toroidal compactifications.

5. Conclusion

Corrêa and Massarenti provide a definitive classification of log-conformal projective manifolds. They show that the geometry is either rigid (resembling classical conformal models), fibred (a new logarithmic phenomenon), or uniformized by semi-abelian varieties. The work successfully integrates birational geometry, deformation theory, and differential geometry to solve a problem at the intersection of complex geometry and the theory of geometric structures.