🔢 mathematics

Logarithmic Fulton--MacPherson configuration spaces

This paper utilizes logarithmic geometry to construct a logarithmic analogue of Fulton--MacPherson configuration spaces and a corresponding logarithmically smooth degeneration, establishing a degeneration formula that describes the special fibre's components as proper birational modifications of products of these logarithmic spaces.

Original authors: Siao Chi Mok

Published 2026-05-22

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Siao Chi Mok

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 hosting a party where $n$ guests are arriving at a beautiful, smooth garden (the variety $X$ ). In the ideal world, everyone stays in their own spot, and no two people ever bump into each other. Mathematicians call this the "configuration space."

However, in the real world, guests might get crowded, or the garden itself might start to crumble or split apart (a "degeneration"). The famous Fulton–MacPherson construction is a way of expanding the garden so that even if guests crowd together, the space "blows up" to create a new, temporary bubble (a sphere) for them to stand on, keeping them separated. This creates a complete, tidy map of every possible way the guests can arrange themselves, even when they are squished together.

This paper, by Siao Chi Mok, takes that idea and adds two major new ingredients: Logarithmic Geometry and Grid Expansions. Here is the breakdown in simple terms:

1. The Problem: Crumbling Gardens and Crowded Guests

The author asks two main questions:

Question 1: What if the garden isn't a perfect, smooth place to begin with? What if it has a "fence" or a boundary (a divisor $D$ ) that is jagged or has corners? How do we map the guests if they can't just stand on the smooth grass, but might be near the fence?
Question 2: What if the garden itself is falling apart? Imagine the garden is a cake that is slowly cracking into pieces as time passes. How does the map of the guests change as the cake breaks?

2. The Solution: The "Grid" and the "Bubble"

The author builds a new mathematical tool called the Logarithmic Fulton–MacPherson Configuration Space. Think of it as a master blueprint that handles both the jagged fences and the crumbling cake.

Step A: The "Grid" (Separating Guests from the Fence)

First, the author deals with the jagged boundary.

The Analogy: Imagine the garden has a fence made of several intersecting walls. If a guest gets too close to a wall, they don't just stop; they step onto a "grid" of new, temporary platforms.
The Math: The author creates a "moduli space" (a map of all possibilities) called $(X|D)[n]$ . This map forces guests to step onto these new platforms (called grid expansions) if they get too close to the boundary. It's like a traffic control system that ensures no guest ever touches the jagged fence directly; they always stand on a smooth, expanded path.

Step B: The "Bubbles" (Separating Guests from Each Other)

Once the guests are safely away from the fence, they might still bump into each other.

The Analogy: This is where the original Fulton–MacPherson idea comes back. If two guests try to stand on the same spot, the ground splits open, and a new "bubble" (a sphere) pops up between them. The guests stand on the bubble, separated.
The Math: The author takes the "grid" map from Step A and performs a series of "blow-ups" (splitting the space). This creates the final space, $FM_n(X|D)$ . It is a complete, tidy map where every possible arrangement of guests—whether they are near the fence or hugging each other—is accounted for.

3. The "Crumbling Cake" (Degeneration)

The second half of the paper tackles the garden falling apart over time.

The Analogy: Imagine the garden is a cake that is slowly cracking into distinct pieces (a "degeneration"). The author constructs a "degeneration formula."
The Result: They show that as the garden cracks, the map of the guests doesn't disappear. Instead, the final map breaks apart into a collection of smaller, simpler maps.
- Think of a complex puzzle. When the puzzle box breaks, the pieces don't scatter randomly; they fall into specific, predictable piles.
- The author proves that the "special fiber" (the state of the garden when it is fully broken) is made of pieces that look like products of the simpler "grid" maps we built in Step 1.
- This allows mathematicians to calculate complex things about the broken garden by looking at the simpler pieces.

4. The "Rubber" and the "Trees"

To make all this work, the author uses some clever visual tools:

Rubber Action: Imagine the guests are on a rubber sheet. If they move slightly, the sheet stretches, but the relative order stays the same. The author uses this "rubber" concept to handle the symmetries of the space, ensuring the map is unique and well-defined.
Planted Forests: The author describes the structure of the space using "trees" and "forests."
- If guests are far apart, they are like trees with no branches.
- If guests cluster, the trees grow branches.
- The "combinatorial type" of the space is essentially a picture of a forest. This allows the author to categorize every single possible arrangement of guests by looking at the shape of the forest.

Summary of the Achievement

In short, this paper builds a universal, robust map for arranging points (guests) in a space (garden) that might have rough edges (fences) or might be breaking apart (crumbling).

It creates a Logarithmic Fulton–MacPherson space that handles rough boundaries.
It creates a degeneration formula that shows how this space breaks down into simpler, predictable pieces when the underlying space breaks.
It uses tropical geometry (thinking in terms of cones, grids, and forests) to describe the structure of these spaces, making it possible to visualize and calculate things that were previously too messy to handle.

The paper claims this construction works for any smooth projective variety and any "simple normal crossings" boundary, providing a powerful new toolkit for understanding how points behave in complex, changing environments.

Technical Summary: Logarithmic Fulton–MacPherson Configuration Spaces

Problem Statement
The paper addresses two fundamental questions regarding the Fulton–MacPherson (FM) configuration spaces, $FM_n(X)$ , which are simple normal crossings (SNC) compactifications of the ordered configuration space of $n$ distinct points on a smooth projective variety $X$ .

Relative Compactification: Given a non-proper variety $U$ with an SNC compactification $(X, D)$ , how can one construct an SNC compactification of the configuration space $Conf_n(U)$ ? Previous work by Kim and Sato [KS09] addressed the case where $D$ is smooth; this paper seeks a solution for the general case where $D$ is a general SNC divisor.
Degeneration: Given a degeneration $\omega: W \to B$ of $X$ over a curve $B$ with an SNC special fibre, how does the space $FM_n(X)$ degenerate? While Routis [Rou14] answered this for smooth boundaries, the general case requires new techniques.

Methodology
The author employs techniques from logarithmic and tropical geometry, specifically utilizing Artin fans, cone stacks, and tropicalisation, to generalize the classical FM construction. The approach proceeds in two main stages:

Construction of Grid Expansions:
- The author first constructs a moduli space $(X|D)[n]$ of stable $n$ -pointed "grid expansions."
- This relies on the tropicalisation $\Sigma_X$ of the pair $(X, D)$ , which is a rational polyhedral cone complex.
- The construction involves defining "marked grid subdivisions" of $\Sigma_X$ . These are polyhedral subdivisions induced by $n$ points on the tropicalisation, arranged in a grid-like structure (intersections of subdivisions of rays).
- Using the equivalence between cone stacks and Artin fans, the author constructs a universal family of grid expansions over a stack $En(X|D)$. The moduli space $(X|D)[n]$ is then defined as a logarithmic modification of $X^n$ obtained by pulling back this universal family.
- This space $(X|D)[n]$ parametrizes points on degenerations of $X$ along $D$ that separate points from the boundary.
Iterated Blow-ups for Point Separation:
- Following the original Fulton–MacPherson strategy, the author performs an iterated blow-up of $(X|D)[n]$ along diagonals (and their dominant transforms) to separate the $n$ points from each other.
- This yields the Logarithmic Fulton–MacPherson space, denoted $FM_n(X|D)$ .
- A universal family $\pi_{FM}: FM_n(X|D)^+ \to FM_n(X|D)$ is constructed, which parametrizes stable $n$ -pointed FM grid expansions (FM degenerations of the grid expansions).
Degeneration of FM Spaces:
- For a degeneration $\omega: W \to B$ , the author constructs a moduli space $W[n]_\omega$ of stable $n$ -pointed "simplex-lattice expanded degenerations."
- This involves "simplex-lattice subdivisions" of the tropicalisation of the degeneration, adapted to the base curve $B$ .
- An iterated blow-up of $W[n]_\omega$ yields the degeneration $FM_n(W/B) \to B$ .

Key Contributions and Results

Theorem A (Logarithmic FM Compactification): For an SNC pair $(X, D)$ , the scheme $FM_n(X|D)$ is a simple normal crossings compactification of $Conf_n(X \setminus D)$ . It represents the moduli stack of stable $n$ -pointed FM grid expansions. The space admits a stratification by combinatorial types, which correspond to "planted forests" (a tropical analogue involving grid subdivisions and rooted trees).
Theorem B (Logarithmically Smooth Degeneration): Given an SNC degeneration $\omega: W \to B$ , the morphism $FM_n(W/B) \to B$ is a proper, flat, logarithmically smooth degeneration of $FM_n(X)$ with reduced fibres. It carries a universal family of stable FM expanded degenerations.
Theorem C (Degeneration Formula): The irreducible components of the special fibre of $FM_n(W/B)$ are in bijection with "rigid combinatorial types" (types where the tropical moduli is a vertex). Each such component is a proper birational modification of a product of logarithmic FM spaces associated with the components of the underlying expansion. Specifically, if $\rho$ is a rigid type, the component $FM_n(W/B)(\rho)$ maps to $\prod_{v} FM_{I_v}(Y_v | D_v)$ , where $Y_v$ are components of the special fibre expansion.
Canonical Choice of Expansions: Unlike previous works (e.g., Maulik–Ranganathan) where expansions depend on tropical types of maps, this paper establishes a canonical choice of expansions ("grid expansions" and "simplex-lattice expansions") derived from polyhedral subdivisions. This allows for a well-defined stack of expansions representable by an algebraic stack.
Tropicalisation: The tropicalisation of $FM_n(X|D)$ is identified as the moduli space of planted forests satisfying specific conditions. When $D$ is empty, this recovers the space of rooted trees described in [FM94].

Significance and Claims
The paper claims to provide a systematic framework for handling configuration spaces on varieties with SNC boundaries and their degenerations, extending the classical Fulton–MacPherson theory to the logarithmic setting.

Generalization: It generalizes the constructions of Kim–Sato [KS09] (for smooth boundaries) and Abramovich–Fantechi [AF17] (for expansions) to the case of arbitrary SNC divisors and degenerations.
Canonical Structure: By utilizing a canonical choice of subdivisions (grid and simplex-lattice), the author avoids the ambiguity of choosing expansions based on map types, enabling a more granular description of the moduli spaces.
Applications: The author notes potential applications in:
- Defining logarithmic unramified Gromov–Witten theory for pairs $(X, D)$ (work in progress [Mok26]).
- Deriving a degeneration formula for unramified Gromov–Witten invariants, potentially linking them to Gopakumar–Vafa invariants (as conjectured by Pandharipande and proved by Nesterov [Nes25b] in the Fano case).
- Investigating the intersection theory of classical and logarithmic FM spaces, suggesting that generating series of integrals of tautological classes may be cobordism invariants expressible via universal polynomials in Chern numbers.

The paper remains modest regarding these applications, framing them as "potential" and "work in progress," while emphasizing the structural results regarding the construction and properties of the configuration spaces themselves.