🔢 mathematics

Tropical intersection homology

This paper introduces a tropical analog of intersection homology to generalize the geometric description of numerical equivalence of algebraic cycles from smooth complex proper toric varieties to suitable pairs of smooth proper varieties and divisors.

Original authors: Ryota Mikami

Published 2026-05-14

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Ryota Mikami

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are trying to understand the shape of a complex, jagged object—like a crumpled piece of paper or a rocky mountain range. In mathematics, specifically in the world of algebraic geometry, these objects are called "varieties." When these shapes are perfectly smooth, we have excellent tools to measure their holes, loops, and twists. We call these measurements "cohomology."

However, when these shapes have sharp corners, self-intersections, or singularities (the "crumpled" parts), our standard measuring tools break down. They give us answers that don't quite make sense, or they fail to respect the fundamental rules of symmetry (a rule mathematicians call "Poincaré duality").

This paper, written by Ryota Mikami, introduces a new, more robust measuring tool called Tropical Intersection Homology.

Here is the breakdown of the paper's ideas using simple analogies:

1. The Problem: The "Broken" Map

Think of a smooth variety as a pristine, flat sheet of paper. We can easily draw lines on it and count how they cross. This is "Singular Cohomology."
Now, imagine crumpling that paper into a ball. If you try to draw lines on the crumpled ball, they might get stuck in the creases or tear. The old rules for counting intersections no longer work perfectly. In the language of the paper, the "Poincaré duality" (a kind of perfect balance between different types of measurements) fails.

2. The New World: "Tropical" Geometry

The paper operates in a field called Tropical Geometry. This is a bit like looking at a complex mathematical object through a special filter that turns it into a skeleton made of straight lines and flat planes (polyhedra).

The Analogy: Imagine taking a complex 3D sculpture and replacing it with a wireframe model made of sticks. It's not the full sculpture, but it captures the essential "skeleton" and how the pieces connect.
In this "Tropical" world, the author looks at these wireframe skeletons to understand the original complex shapes.

3. The Solution: "Intersection Homology"

In the 1980s, mathematicians Goresky and MacPherson invented "Intersection Homology" to fix the "crumpled paper" problem for standard shapes. They created a set of strict rules for how to draw lines on a crumpled shape so that the lines are allowed to pass through the smooth parts freely but must be very careful when crossing the sharp creases.

Mikami's paper does the same thing, but for the Tropical wireframe skeletons.

The "Allowability" Rule: Imagine you are walking a tightrope across a canyon. On the smooth parts of the canyon, you can walk anywhere. But near the jagged, broken edges (the singularities), you have to follow a specific path. If you get too close to the edge, you might fall.
Mikami defines a set of rules (called "allowability") that tell a mathematical "walker" exactly how close they can get to the jagged edges of a tropical shape without breaking the rules. By only counting the paths that follow these rules, he creates a new measurement system that works even when the shape is broken.

4. The Big Discovery: Connecting Two Worlds

The main result of the paper is a bridge between two different ways of counting:

Algebraic Cycles: These are like counting the number of specific "loops" or "surfaces" you can draw on the original complex shape.
Tropical Intersection Homology: This is the new measurement of the wireframe skeleton using the strict "tightrope" rules.

The Claim: The paper proves that for a wide class of shapes (specifically, smooth shapes embedded in a special type of "toric" environment), the number of loops you can draw on the original shape (modulo a specific type of equivalence called "numerical equivalence") is exactly the same as the number of valid paths you can find on the tropical wireframe skeleton using the new rules.

5. Why This Matters (According to the Paper)

It Fixes the Broken Symmetry: Just as Goresky and MacPherson fixed the symmetry problem for crumpled paper, Mikami fixes it for crumpled tropical skeletons. The new measurement system restores the perfect balance (Poincaré-Verdier duality) that was lost.
It's a Generalization: Previous work showed this connection worked for very specific, perfect shapes (like toric varieties). Mikami shows it works for a much broader, more realistic set of shapes, provided they are "tropical compactifications" (a technical way of saying they fit nicely into the tropical world).
The Method: He builds this new tool using two different approaches:
1. Geometric: Drawing the paths directly on the wireframe.
2. Sheaf-theoretic: Using a high-level mathematical framework (like a sophisticated database of local rules) to define the same thing. He proves both methods give the exact same result.

Summary

In short, Ryota Mikami has invented a new way to measure the "holes" and "loops" of complex, jagged mathematical shapes by translating them into a simpler, wireframe language (Tropical Geometry) and applying a strict set of safety rules (Intersection Homology) to ensure the measurements remain accurate even when the shape is broken. He proves that this new measurement perfectly matches the traditional counting of loops in the original complex shapes.

Technical Summary: Tropical Intersection Homology

Problem Statement
The paper addresses the gap between the algebraic theory of numerical equivalence of algebraic cycles and geometric cohomology theories in the context of tropical geometry. Classically, for smooth complex proper toric varieties $T_\Sigma$ , the quotient of the Chow group by numerical equivalence, $CH^p_{\text{Num}}(T_\Sigma) \otimes \mathbb{Q}$ , is isomorphic to the singular cohomology $H^{2p}_{\text{sing}}(T_\Sigma(\mathbb{C}); \mathbb{Q})$ . Furthermore, this is isomorphic to tropical cohomology $H^{p,p}_{\text{Trop}}(\text{Trop}(T_\Sigma); \mathbb{Q})$ , as established by Itenberg-Katzarkov-Mikhalkin-Zharkov (IKMZ).

However, for a general smooth proper variety $Y$ embedded in a toric variety $T_\Lambda$ (specifically a "tropically compact" subvariety), the tropical cohomology $H^{p,q}_{\text{Trop}}(\text{Trop}(Y); \mathbb{Q})$ does not necessarily satisfy Poincaré duality. Consequently, the natural map from tropical cohomology to the numerical Chow group $CH^p_{\text{Num}}(Y) \otimes \mathbb{Q}$ is surjective but not necessarily an isomorphism. The failure of Poincaré duality in the tropical setting for singular or non-smooth varieties mirrors the failure of Poincaré duality in singular cohomology for singular complex varieties. The paper seeks to introduce a tropical analog of intersection homology to restore this duality and provide a geometric description of $CH^p_{\text{Num}}(Y) \otimes \mathbb{Q}$ for suitable pairs $(Y, D)$ .

Methodology
The author employs a dual approach, defining tropical intersection homology both geometrically (via chains) and sheaf-theoretically (via derived categories), and proving their equivalence.

Geometric Definition:
- The paper defines tropical intersection chains $IC^{p,q}_{\text{Trop}}$ on a tropical variety $X$ (regular at infinity) by imposing "allowability" conditions on singular chains.
- These conditions are analogous to those in Goresky-MacPherson's intersection homology. They restrict the intersection of chains with singular strata based on the codimension of the stratum and the "weight" of the chain coefficients.
- The coefficients used are derived from Gubler-Jell-Rabinoff's new tropical coefficients $F^{p,w}$ , which are designed to satisfy Poincaré-Verdier duality near 1-codimensional strata, a feature missing in the original IKMZ coefficients.
- The definition involves filtrations on the chain complexes based on the local geometry of the fan, specifically utilizing the dimensions of cones $\sigma_S$ and the strata $S$ .
Sheaf-Theoretic Construction:
- To establish Poincaré-Verdier duality rigorously, the paper constructs the tropical intersection complex in the derived category of sheaves.
- A key technical innovation is the introduction of locally graded sheaves. The author defines a derived category $D_{\text{LG}}$ by localizing the derived category of graded sheaves with respect to "LG-quasi-isomorphisms." This localization is necessary because the truncation functors used to define the intersection complex involve filtrations that are only defined locally on open neighborhoods of strata.
- The tropical intersection complex $\text{IC}^p_{\text{Trop,sheaf}}$ is constructed inductively by extending the sheaf of tropical forms from the smooth part $X_{\text{sm}}$ to the singular locus using truncation functors $\tau_{p, S_k}$ that satisfy specific "axioms" (local vanishing and attaching properties) analogous to Deligne's construction of the intersection complex.
Comparison and Duality:
- The paper proves that the geometric and sheaf-theoretic definitions yield isomorphic homology groups.
- It establishes a Poincaré-Verdier duality theorem for tropical intersection homology: a non-degenerate pairing between $I H^{p,q}_{\text{Trop}}(X)$ and $I H^{d-p, d-q}_{\text{Trop}, c}(X)$ .
- The author utilizes spectral sequences comparing tropical intersection homology to tropical cohomology and singular cohomology (via the tropicalization map) to analyze the structure of these groups.

Key Contributions and Results

Definition of Tropical Intersection Homology: The paper introduces $I H^{p,q}_{\text{Trop}}(X; \mathbb{Q})$ for tropical varieties $X$ regular at infinity. This theory coincides with standard tropical cohomology when Poincaré duality holds (e.g., for smooth tropical varieties) but provides a refined theory for singular cases.
Restoration of Poincaré Duality: Theorem 1.2 establishes a non-degenerate pairing for tropical intersection homology, generalizing the duality known for smooth tropical varieties to the singular setting.
Isomorphism with Numerical Equivalence: The central result (Theorem 1.2, Corollary 7.3) states that for a smooth proper variety $Y$ tropically compact in a smooth toric variety $T_\Lambda$ (satisfying specific irreducibility and surjectivity conditions), there is a natural isomorphism:
$I H^{p,q}_{\text{Trop}}(\text{Trop}(Y); \mathbb{Q}) \cong \begin{cases} CH^p_{\text{Num}}(Y) \otimes \mathbb{Q} & \text{if } p=q \\ 0 & \text{if } p \neq q \end{cases}$
Under this isomorphism, the pairing on intersection homology corresponds to the intersection numbers of algebraic cycles.
Independence of Polyhedral Structure: The paper proves that the tropical intersection homology is independent of the specific polyhedral decomposition (subdivision) of the tropical variety, provided the variety remains regular at infinity.
Technical Framework: The development of the theory of locally graded sheaves and the associated derived categories provides a robust sheaf-theoretic framework for handling the local-to-global issues inherent in tropical intersection homology.

Significance
The paper claims to generalize the geometric description of numerical equivalence from toric varieties to a broader class of smooth proper varieties via tropicalization. By introducing tropical intersection homology, the author bridges the gap between the algebraic notion of numerical equivalence and the topological notion of cohomology in the tropical setting, specifically addressing the failure of Poincaré duality in standard tropical cohomology for singular varieties. This provides a tropical analog to the classical result that the intersection homology of a singular complex variety recovers the correct cohomological invariants related to the underlying algebraic structure. The work suggests that tropical intersection homology is the correct cohomological theory to capture the numerical equivalence classes of cycles on varieties that admit a tropical compactification.