Curves on the product of two $K-$trivial surfaces

This paper investigates curves on the product of two $K$-trivial surfaces and establishes that the minimal genus of a non-trivial curve on the product of two very general abelian surfaces is 6.

The Gist

Imagine you have two very different, complex worlds. Let's call them World K (a K3 surface, which is like a perfectly smooth, intricate 4D sphere with special geometric properties) and World A (an Abelian surface, which is like a 4D donut or a torus).

The paper by Federico Moretti and Giovanni Passeri asks a fundamental question: How hard is it to build a "bridge" between these two worlds?

In mathematics, a "bridge" is called a correspondence. It's a third shape (let's call it the "Bridge") that sits in the space between World K and World A. This Bridge must touch both worlds completely, and you must be able to walk from the Bridge onto either world without getting stuck.

The authors are trying to find the Covering Genus. Think of "Genus" as the number of holes in a shape (a sphere has 0, a donut has 1, a pretzel has 3).

• The Covering Genus is the minimum number of "holes" the Bridge needs to have to successfully connect the two worlds.
• If the genus is low, the bridge is simple. If it's high, the bridge is incredibly twisted and complex.

Here is the breakdown of their findings using simple analogies:

1. The Main Discovery: The "Pretzel" Bridge

The authors prove that to connect a generic K3 surface and a generic Abelian surface, you cannot use a simple bridge (like a sphere or a donut). You need a 3-holed pretzel.

• The Result: The minimum complexity (genus) required is 3.
• The Analogy: Imagine trying to connect a smooth marble (K3) and a rubber donut (Abelian) with a rope. You can't use a straight string (genus 0) or a simple loop (genus 1). You need a rope that is twisted into a complex knot with three loops to make the connection work. If you try to use a simpler knot, the geometry just doesn't allow the rope to touch both worlds properly.

2. Connecting Two "Donuts"

What if you try to connect two different Abelian surfaces (two different 4D donuts)?

• The Result: The bridge needs to be even more complex. The minimum genus is 6.
• The Analogy: Connecting two donuts is harder than connecting a donut to a marble. You need a bridge that looks like a 6-holed pretzel. The geometry of these "donut worlds" is so rigid that a simpler bridge would slip off or fail to cover the whole surface.

3. The "Irrationality" Puzzle

The paper also looks at a related concept called "irrationality." In this context, it's a measure of how "twisted" a shape is.

• They prove that the "cost" (degree) of building a bridge between two Abelian surfaces is exactly the product of their individual "twistedness."
• The Analogy: If World A1 is "3 units of twisted" and World A2 is "4 units of twisted," the bridge connecting them must be "12 units of twisted." It's a perfect multiplication rule, showing that these shapes are incredibly independent; you can't cheat the system by finding a shortcut.

How Did They Figure This Out?

The authors used a mix of high-level detective work and geometry:

1. The "Too Simple" Trap: They first tried to imagine a bridge with only 1 or 2 holes. They realized that if the bridge were that simple, it would force the two worlds to look too much like each other (mathematically, they would have to be "isogenous"). But since the authors are looking at generic (randomly chosen) worlds, they are too different for a simple bridge to work.
2. The "Counting" Game: They counted the number of ways these shapes can wiggle and deform. They found that a bridge with fewer than 3 holes (for K3/Abelian) or 6 holes (for Abelian/Abelian) simply doesn't have enough "wiggle room" to fit the mathematical requirements.
3. The "Fiber" Argument: They imagined the bridge as a family of curves (like a stack of pancakes). They proved that if the pancakes were too simple, the stack would collapse or fail to cover the target worlds.

Why Does This Matter?

In the world of algebraic geometry, we often want to know how "close" two shapes are.

• If the covering genus is low, the shapes are "close" (they share simple structures).
• If the covering genus is high, the shapes are "far apart" (they are fundamentally different).

This paper tells us that K3 surfaces and Abelian surfaces are quite far apart, and two different Abelian surfaces are even further apart. You need a very complex, multi-holed structure to link them. It's a bit like realizing that to connect a cloud to a mountain, you can't just throw a ladder; you need a massive, winding elevator shaft.

In short: The paper calculates the "minimum complexity" required to build a mathematical bridge between these special 4D shapes, proving that these bridges must be surprisingly complex (3-holed or 6-holed) to work.

Technical Summary

Based on the provided text, here is a detailed technical summary of the paper "Covering Genus of Two K-Trivial Surfaces" by Federico Moretti and Giovanni Passeri.

1. Problem Statement

The paper investigates correspondences between algebraic varieties, specifically focusing on K-trivial surfaces (K3 surfaces and abelian surfaces).

• Definition: A correspondence between two varieties $X$ and $Y$ is a subvariety $Z \subseteq X \times Y$ that dominates both $X$ and $Y$ with finite degree.
• Invariant: The authors study the covering genus, denoted $\text{cov.gen}(X, Y)$, defined as the minimal genus $g \geq 0$ such that there exists a family of curves of genus $g$ that dominates a correspondence between $X$ and $Y$. Equivalently, it is the least $g$ for which there exist dominant maps $C \to X$ and $C \to Y$ from a family of curves $C \to B$.
• Goal: Determine the exact values of the covering genus for pairs of very general K3 surfaces, abelian surfaces, and mixed pairs (K3 and abelian).

2. Methodology

The authors employ a combination of moduli theory, Hodge theory, deformation theory, and geometric invariant theory.

• Moduli Spaces and Dimension Counts: They analyze families of curves mapping to abelian surfaces and K3 surfaces. By estimating the dimensions of Hilbert schemes and moduli spaces (e.g., $\mathcal{M}_g$, $\mathcal{A}_g$), they derive lower bounds for the genus required to dominate these surfaces.
• Torelli Maps and Jacobians: A central tool is the Torelli map $T_g: \mathcal{M}_g \to \mathcal{A}_g$. The authors study the differential of this map and the tangent spaces of moduli spaces to understand deformations of curves whose Jacobians split into products of abelian surfaces and elliptic curves.
• Multiplication Maps: To prove lower bounds, they analyze the rank of specific multiplication maps on cohomology groups (e.g., $m: H^0(\Omega^1) \otimes H^0(\Omega^1) \to H^0(\omega^{\otimes 2})$). If the rank is high, the dimension of the family of curves is restricted, forcing the genus to be higher.
• Rationality and Kummer Surfaces: In the proof of Theorem C, they utilize the geometry of Kummer surfaces ($K(A) = \text{Bl}_{2\text{-torsion}}(A/\pm 1)$) and the classification of rational maps from abelian surfaces to surfaces. They leverage the fact that very general abelian surfaces do not admit non-trivial maps to certain rational surfaces or Kummer surfaces without specific constraints.
• Baire Category Theorem: Used in the construction of dominating families to ensure the existence of a dominant map from a generic member of a parameter space.

3. Key Contributions and Results

The paper establishes three main theorems regarding the covering genus and correspondence degree of K-trivial surfaces:

Theorem A: Mixed Case (K3 and Abelian)

• Statement: Let $S$ be a very general K3 surface and $A$ be a very general abelian surface. Then:
  $$\text{cov.gen}(S, A) = 3$$
• Significance: This establishes that a family of genus 3 curves is sufficient to dominate both a generic K3 and a generic abelian surface simultaneously, but genus 2 is insufficient.
• Proof Sketch:
  • Upper Bound: Constructed via a non-isotrivial pencil of genus 3 curves on $A$ (induced by polarization) and a non-isotrivial family of elliptic curves covering $S$. Using relative Hilbert schemes, they construct a genus 3 family dominating both.
  • Lower Bound: Proved by contradiction. If genus 2 sufficed, the curve would map birationally to both $S$ and $A$. However, a very general abelian surface is simple (no elliptic quotients), and genus 2 curves mapping to it cannot map non-trivially to an elliptic curve. Dimensional arguments on the moduli space of such maps show this locus is a countable union of proper closed subsets, excluding very general pairs.

Theorem B: Two Abelian Surfaces

• Statement: Let $A_1, A_2$ be very general abelian surfaces of any polarization. Then:
  $$\text{cov.gen}(A_1, A_2) = 6$$
• Significance: This is a significant result showing that connecting two generic abelian surfaces requires curves of genus 6.
• Proof Sketch:
  • Lower Bound ($\geq 5$): They rule out genus $\leq 4$.
    • Genus 4 is ruled out by analyzing the Hilbert scheme of curves on $A_1 \times A_2$. Dimensional constraints on the base of the family and the fibers of the map to the moduli space $\mathcal{A}_2$ lead to a contradiction.
    • Genus 5 is ruled out by analyzing the deformation space of curves $C$ where $\text{Jac}(C) \approx A_1 \times A_2 \times E$. They compute the rank of a mixed multiplication map $m$. They prove $\text{rank}(m) \geq 7$. Using the formula $\dim(B) \leq 12 - \text{rank}(m)$, they show the dimension of the family is too small to dominate the product of two 2-dimensional moduli spaces ($\dim = 4$) plus the elliptic curve moduli.
  • Upper Bound: While the text focuses heavily on the lower bound, the theorem asserts the value is 6, implying a construction exists (likely related to Prym varieties or specific covers).

Theorem C: Correspondence Degree

• Statement: Let $A_1, A_2$ be very general abelian surfaces. Then the minimal degree of a correspondence between them is:
  $$\text{corr}(A_1, A_2) = \text{irr}(A_1) \cdot \text{irr}(A_2)$$
  where $\text{irr}(A)$ is the degree of irrationality (the minimal degree of a dominant rational map $A \dashrightarrow \mathbb{P}^2$).
• Context: It is known that for very general abelian surfaces, $\text{irr}(A) \in \{3, 4\}$.
• Significance: This confirms that the "complexity" of a correspondence is multiplicative for very general abelian surfaces.
• Proof Sketch:
  • They assume a correspondence $Z$ has degree less than the product and derive a contradiction.
  • They utilize Theorem 5.1, which states that any generically finite rational map from a very general abelian surface to a smooth projective variety $X$ (with specific holomorphic 2-form conditions) must factor through a rational surface.
  • By analyzing the image of the correspondence and using the properties of Kummer surfaces and the degree of irrationality, they show that a lower degree would imply the existence of a map violating the known bounds on irrationality or the structure of rational maps from abelian surfaces.

4. Significance and Impact

• Quantifying Geometric Complexity: The paper provides precise numerical invariants (covering genus) for the relationship between K-trivial surfaces, moving beyond qualitative existence results.
• Rigidity of Very General Varieties: The results highlight the rigidity of "very general" abelian surfaces and K3 surfaces. They cannot be connected by low-genus curves or low-degree correspondences unless specific geometric structures (like isogenies or special pencils) are present, which are absent in the very general case.
• Methodological Advancement: The paper demonstrates a powerful technique of combining tangent space analysis of moduli spaces (via the Torelli map) with dimensional counting in Hilbert schemes to bound the genus of curves dominating products of varieties.
• Connection to Irrationality: Theorem C links the concept of covering genus/correspondence directly to the degree of irrationality, a central topic in the study of rationality problems in higher dimensions.

In summary, Moretti and Passeri have successfully determined the minimal genus of curves required to link generic K3 and abelian surfaces, revealing that these connections are significantly more complex (requiring genus 3 and 6) than previously might have been assumed, and establishing a multiplicative law for correspondence degrees.