Numerical tropical line bundles and toric b-divisors

This paper establishes a natural injective correspondence between numerical tropical line bundles on a very affine variety and toric b-divisors on its tropicalization, demonstrating that this map restricts to a bijection between the tropical nef cone and tropically nef b-Cartier divisors, thereby generalizing Baker's specialization to higher dimensions and clarifying the birational nature of tropical line bundles.

The Gist

Imagine you have a complex, beautiful sculpture made of glass and light (this is your algebraic variety, a shape defined by equations). Now, imagine you want to study this sculpture, but you can only see its shadow cast on a wall when the light is very specific. This shadow is the tropicalization.

In the world of mathematics, "tropical geometry" is like studying these shadows. It turns complex, curved shapes into simpler, angular shapes made of straight lines and flat planes (polyhedral complexes).

This paper by Carla Novelli and Stefano Urbinati is about a specific problem: How do we translate the "clothing" (line bundles) of the original glass sculpture into the "clothing" of its shadow?

Here is a simple breakdown of their journey:

1. The Problem: The Shadow Loses Details

When you cast a shadow, you lose some information.

• The Original: A line bundle on the sculpture is like a specific pattern of fabric wrapped around it. It has exact measurements and continuous curves.
• The Shadow: When you project this onto the tropical wall, the fabric turns into a piecewise linear function (a折线, or a series of straight segments).
• The Issue: Many different patterns of fabric on the original sculpture can cast the exact same shadow. If you only look at the shadow, you can't tell which specific fabric was used. The "continuous" details (the exact moduli) are lost.

2. The Solution: Counting the "Slopes"

The authors realized that while we lose the exact fabric, we don't lose the slope of the fabric.

• Imagine the fabric draped over the shadow. It goes up, down, or stays flat.
• The authors decided to stop looking at the specific fabric and instead focus on the numerical data: "How steep is the slope here?" and "How much does it rise there?"
• They call these Numerical Tropical Line Bundles. It's like saying, "I don't care if the fabric is silk or cotton, I only care that it rises 3 units for every 1 unit you move right."

3. The Bridge: The "b-Divisor" (The Universal Blueprint)

To connect the original sculpture to the shadow perfectly, they used a mathematical tool called a b-divisor.

• Analogy: Think of a b-divisor as a "Universal Blueprint" or a "Master Recipe."
• Usually, a blueprint is drawn for one specific building. But a b-divisor is a blueprint that exists for every possible version of the building, from the rough sketch to the finished masterpiece. It keeps track of how the building changes as you refine the details.
• The authors showed that every "Numerical Tropical Line Bundle" (the slope data of the shadow) corresponds to exactly one of these Universal Blueprints (a toric b-divisor).

4. The Big Discovery: A Perfect Match

The main result of the paper is a "One-to-One" match (a bijection) between two things:

1. The Tropical Nef Cone: This is a fancy way of saying "The set of all slopes that are never negative." Imagine a fabric that never hangs down; it always slopes up or stays flat.
2. The b-Cartier b-Divisors: These are the Universal Blueprints that are "well-behaved" (they don't have weird kinks or tears).

The Magic: The authors proved that if you have a "positive" slope pattern on the tropical shadow, there is exactly one corresponding "well-behaved" Universal Blueprint. And vice versa.

5. Why the "Schön" Assumption Matters

The paper has a strict rule: The original sculpture must be "schön" (a German word meaning "beautiful" or "nice").

• What it means: The sculpture must be "well-behaved" enough that when it casts its shadow, the edges of the shadow align perfectly with the grid lines of the wall.
• The Warning: If the sculpture is "ugly" (not schön), the shadow might get messy. Parts of the fabric might hide in the corners of the wall where the shadow doesn't reach. In this case, two different fabrics could cast the same shadow, and the math breaks down. The authors show that for "nice" shapes, the map is perfect; for "messy" shapes, it fails.

6. The "Curve" Connection

The paper mentions that for 1-dimensional shapes (curves), this is a known result (Baker's specialization).

• Analogy: If your sculpture is just a wire (a curve), the shadow is a line graph. It's easy to see that the slope of the wire matches the slope of the graph.
• The Breakthrough: This paper takes that simple idea for wires and proves it works for complex, multi-dimensional sculptures (higher dimensions). They generalized the rule from 1D to 3D, 4D, and beyond.

Summary in a Nutshell

Imagine you are trying to describe a complex 3D object to someone who can only see its 2D shadow.

• Old way: You try to describe the exact texture of the object. (Impossible, the shadow hides it).
• This paper's way: You describe the angles and steepness of the shadow.
• The Result: The authors built a dictionary that translates "Steepness of Shadow" directly into "Universal Blueprint."
• The Catch: This dictionary only works if the object is "well-behaved" (schön). If the object is twisted and messy, the dictionary fails.

This work is significant because it gives mathematicians a rigorous way to study the "positivity" (how things curve and grow) of complex shapes by looking at their simpler, tropical shadows, bridging the gap between the messy real world and the clean, combinatorial world of tropical geometry.

Technical Summary

1. Problem Statement

The central problem addressed is understanding the relationship between algebraic line bundles on a very affine variety $Y$ and their "tropicalizations" on the associated tropical variety $\text{Trop}(Y)$.

• The Challenge: Tropicalization is a process that maps algebraic geometry to combinatorial geometry. However, it is not injective regarding isomorphism classes of line bundles because it "forgets" continuous moduli (parameters related to the complex structure).
• The Gap: While Baker's specialization map provides a correspondence for curves, a higher-dimensional generalization for varieties of arbitrary dimension was lacking. Specifically, there was no rigorous framework to characterize which tropical objects correspond to algebraic line bundles and how to handle the non-uniqueness of extensions of line bundles to tropical compactifications.
• The Goal: To construct a natural map from the group of numerical tropical line bundles on $Y$ to the space of toric b-divisors on $\text{Trop}(Y)$, and to characterize the image of this map, particularly regarding positivity (nefness).

2. Methodology and Setup

The authors employ a synthesis of tropical geometry, toric geometry, and valuation theory (specifically the theory of b-divisors).

• Assumptions: The variety $Y$ is assumed to be very affine (a closed subvariety of an algebraic torus $T \cong (\mathbb{C}^*)^n$) and schön (in the sense of Tevelev). The "schön" condition is crucial as it guarantees the existence of "tropical compactifications" where the intersection of the variety with torus orbits behaves well (proper intersections and smooth strata).
• Tropical Compactifications: The authors utilize tropical compactifications $(Y_\Sigma, P_\Sigma)$, where $P_\Sigma$ is a toric variety associated with a fan $\Sigma$ whose support is $\text{Trop}(Y)$.
• Numerical Equivalence: To bypass the loss of continuous parameters, the authors focus on numerical equivalence classes of line bundles ($N^1$) rather than isomorphism classes.
• b-Divisors: They utilize the language of b-divisors (birational divisors), which are compatible collections of divisors across all birational models (refinements of the fan). This allows for a consistent definition across different tropical compactifications.
• Minkowski Weights: The construction relies on the correspondence between line bundles and Minkowski weights (balanced functions on the codimension-1 cones of the fan), derived from intersection numbers with toric strata.

3. Key Constructions

A. Numerical Tropical Line Bundles ($N^1_{\text{trop}}(Y)$)

The group of numerical tropical line bundles is defined as the direct limit of the Neron-Severi groups of all tropical compactifications $Y_\Sigma$ under refinement:
$$N^1_{\text{trop}}(Y) := \varinjlim_{\Sigma} N^1(Y_\Sigma)$$
An element is represented by a pair $(\Sigma, [L])$, where $[L]$ is a numerical class on a specific compactification.

B. Tropical Weights

For a line bundle $L$ on a compactification $Y_\Sigma$, the authors define tropical weights $w_L(\tau)$ for each codimension-1 cone $\tau$ in the fan $\Sigma$. These weights are the degrees of $L$ restricted to the curves $C_\tau = Y_\Sigma \cap \text{orb}(\tau)$.

• Lemma 5.5: These weights satisfy the balancing condition, making them a Minkowski weight of codimension one.

C. The Map to Toric b-Divisors

Using the balancing condition, the weights determine a piecewise linear (PL) function $\phi_\Sigma$ on the fan. This defines a toric divisor $D_\Sigma$ on $P_\Sigma$.

• Construction 5.6: By taking the limit over all refinements, the collection $(D_\Sigma)$ forms a toric b-divisor $D_{[L]}$ on the toric Zariski–Riemann space $X_{\text{tor}}$.
• This yields a map $\Phi: N^1_{\text{trop}}(Y) \to \text{Div}(X_{\text{tor}}) / \sim_{\text{lin}}$.

4. Main Results

Theorem 5.9: Injectivity and Nefness

The paper establishes the following fundamental results:

1. Injectivity: The map $\Phi$ from numerical tropical line bundles to toric b-divisors (modulo linear equivalence) is injective.
   • Significance: This proves that numerical data is fully preserved under tropicalization for schön varieties. The "continuous moduli" lost in tropicalization do not affect the numerical class.
2. Characterization of the Nef Cone: A class $[L] \in N^1_{\text{trop}}(Y)$ is nef if and only if its associated b-divisor $\Phi([L])$ is:
   • b-Cartier: It becomes a Cartier divisor on some refinement.
   • Tropically Nef: The associated PL function is convex (has non-negative slopes across codimension-1 walls).
3. Bijection: $\Phi$ restricts to a bijection between the tropical nef cone of $Y$ and the set of b-Cartier, tropically nef toric b-divisors.

Corollary 5.10: Recovery of Baker's Specialization

In the case where $Y$ is a curve, the group $N^1_{\text{trop}}(Y)$ is isomorphic to the group of tropical divisors (Baker-Norine group), and the map $\Phi$ recovers Baker's specialization map. This confirms the paper as a higher-dimensional generalization of Baker's work.

5. Significance and Implications

• Higher-Dimensional Generalization: The paper successfully extends the theory of tropical line bundles from curves to arbitrary dimensions, providing a rigorous combinatorial characterization of algebraic positivity.
• Clarification of Birational Nature: By using b-divisors, the authors clarify that tropical line bundles are not just objects on a single fan but are equivalence classes of divisors across all birational models. This resolves the issue of non-uniqueness in extending line bundles to compactifications.
• Role of the "Schön" Condition: The paper demonstrates that the "schön" assumption is necessary. In Section 6.1, they provide an example of a non-schön surface where the map $\Phi$ fails to be injective because the tropicalization cannot "see" certain numerical classes that lie entirely in the interior of the compactification.
• Connection to Valuation Theory: The work links tropical geometry to the theory of b-divisorial valuations (Shokurov's theory). Numerical tropical line bundles correspond to equivalence classes of toric b-divisorial valuations, unifying tropical positivity with general birational geometry.

6. Conclusion

Novelli and Urbinati provide a robust framework for studying line bundles on tropical varieties. By shifting focus to numerical equivalence and utilizing the machinery of b-divisors, they establish a precise, injective correspondence between tropical line bundles and toric b-divisors. This result not only generalizes Baker's specialization map but also offers a new perspective on the birational geometry of tropical compactifications, showing that the "discrete" tropical data fully captures the "numerical" algebraic geometry of the underlying variety.