🔢 mathematics

Defining reduction types of curves via minimal regular and minimal normal crossings models

This paper proposes a new definition of the reduction type of curves over discretely valued fields based on the special fibre of arbitrary regular models, demonstrating its equivalence to minimal regular and minimal normal crossings models while confirming compatibility with established classifications for elliptic and genus 2 curves through a novel invariant of singularities.

Original authors: Jakab Schrettner

Published 2026-07-20

📖 8 min read🧠 Deep dive

CC BY 4.0

Original authors: Jakab Schrettner

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 a detective trying to solve a mystery, but the crime scene has been completely destroyed by a storm. All you have left are a few scattered, muddy footprints and a torn-up map. In the world of mathematics, specifically a branch called arithmetic geometry, researchers often face a similar puzzle. They study "curves" (which are like smooth, looping shapes) that exist over a special kind of number system called a "discretely valued field." Think of this field as a world where numbers have a hidden "depth" or "layer," like an onion.

When these curves are examined closely, they reveal a "special fibre"—a snapshot of what the curve looks like when you peel away the layers to see the very bottom. Sometimes, this bottom layer is a clean, smooth shape. But often, it's a mess: a tangled knot of lines, broken pieces, or weird sharp points. Mathematicians call this the "reduction type." It's like trying to describe a shattered vase just by looking at the shards on the floor. For a long time, they had two different ways to describe these shards. One way looked at the "minimal regular model," which is like trying to fix the vase with the fewest possible glue jobs to make it smooth again. The other way looked at the "minimal normal crossings model," which is like arranging the shards so they only touch at clean, right angles, even if the vase is still broken. The big question was: If you describe the vase using the first method, can you perfectly reconstruct the description from the second method, and vice versa?

This paper, written by Jakab Schrettner, answers that question with a resounding "yes," but it does so by inventing a new, super-detailed language to describe the shards. The author proposes a definition of "reduction type" that works for any regular model, not just the two special ones. The secret sauce of this new definition is a clever way to measure the "sharpness" and "twist" of the broken points. Instead of just saying "this is a sharp point," the author uses a mathematical tool called a "valuation semigroup" (think of it as a unique fingerprint or a DNA sequence for the sharpness) and counts exactly how many "branches" of the curve meet at that point and how tightly they hug each other.

The paper proves that if you have the "minimal regular" description (the smoothest possible fix), you can mathematically calculate the "minimal normal crossings" description (the cleanest intersection), and you can go backward just as easily. It's like having a universal translator that can turn a description of a smooth, repaired vase into a description of a pile of intersecting shards, and back again, without losing a single detail. The author also shows that this new, fancy definition matches perfectly with older, famous classifications for simple curves (like those with one hole or two holes), proving that the new system is compatible with the old ones. Essentially, the paper builds a bridge between two different ways of looking at broken mathematical shapes, showing that they are just two sides of the same coin, and provides a step-by-step recipe (an algorithm) to switch between them.

The Core Discovery: A Universal Translator for Broken Curves

The main finding of this paper is that the "reduction type" of a curve is a single, unified concept, regardless of which specific model you use to describe it. The author defines this type as a detailed tuple of data: a list of the curve's components (the pieces), their multiplicities (how many times they appear), their shapes (genus), and a deep dive into the singularities (the broken points). For every broken point, the author records the "branches" (the lines meeting there), their "valuation semigroups" (the unique numerical fingerprints of their sharpness), and their intersection numbers (how tightly they cross).

The paper explicitly rules out the idea that these two descriptions—the minimal regular model and the minimal normal crossings model—are fundamentally different or that one contains information the other lacks. Instead, it proves they are mathematically equivalent. If you know the type of the minimal regular model, you can determine the type of the minimal normal crossings model with certainty, and the reverse is also true. This is not a guess or a simulation; it is a rigorous proof based on how these shapes change when you perform a specific operation called a "blowup" (which is like zooming in on a sharp point and replacing it with a small circle to smooth it out). The author demonstrates that every time you blow up a point, the information in the "type" changes in a predictable, reversible way.

Furthermore, the paper argues against the notion that you need to stick to low-genus curves (curves with very few holes) to classify these shapes. While previous famous classifications (like Kodaira-Néron for genus 1 and Namikawa-Ueno for genus 2) were limited to these simpler cases, this new definition works for curves of any genus. The author shows that the new definition perfectly recovers the old classifications for genus 1 and 2, proving that the new system is a generalization that fits the old pieces perfectly.

The confidence level here is very high. The paper provides explicit algorithms (step-by-step instructions) to convert between the types. It doesn't just suggest that the connection exists; it shows you exactly how to do the math to get from one to the other. The author also clarifies that while not every possible combination of numbers (semigroups) can form a valid curve singularity, the ones that do appear in these models are fully captured by this new definition.

The Story of the Shards and the Fingerprint

To understand how this works, imagine the special fibre of a curve as a pile of broken glass on a table.

The Components: These are the individual shards of glass. Some are big, some are small.
The Multiplicity: This is like a weight tag on each shard. A shard with a weight of 2 is "twice as heavy" or "doubly present" compared to a shard with a weight of 1.
The Singularities: These are the spots where the shards touch or overlap.
The Branches: If a shard is cracked, it might have two edges meeting at a point. These edges are the "branches."
The Valuation Semigroup: This is the paper's big innovation. Imagine each branch has a unique "texture" or "grain." Some branches are smooth like silk (their semigroup is just all whole numbers). Others are rough, like sandpaper, with specific gaps in their texture (like only having numbers 2, 3, 4, 5... but missing 1). The author records this exact texture as a "semigroup."
The Intersection Number: This measures how hard the branches press against each other. If they just touch lightly, the number is 1. If they are pressed together tightly, the number is higher.

The paper shows that if you have a pile of shards (the minimal regular model) where the glass is smoothed out as much as possible, you can look at the texture of the cracks and the pressure points to figure out exactly how the shards would look if you rearranged them so they only touched at right angles (the minimal normal crossings model).

The author uses a process called "blowing up" to prove this. Think of blowing up a point on a map: you zoom in so close that a single dot becomes a whole circle. In math, when you blow up a sharp point on a curve, the sharp point disappears and is replaced by a new circle (an exceptional divisor). The paper proves that you can track exactly how the "fingerprint" (the semigroup) and the "pressure" (the intersection number) change during this zoom. Because you can track these changes forward (smoothing the curve) and backward (breaking it down), you can translate between any two models.

Why This Matters for the Curious Teen

Why should you care about broken glass and mathematical fingerprints? Because this paper solves a problem of "translation." In science and math, we often have different ways of looking at the same thing. One way might be easier to draw, another easier to calculate. Before this paper, if you wanted to switch between the "smoothest" view and the "cleanest intersection" view of a complex curve, you might have been stuck, especially for curves with many holes.

This paper provides the dictionary. It says, "Don't worry about which model you are using; they all tell the same story if you know how to read the fine print." It takes the messy, complicated world of singularities (broken points) and organizes it into a neat, combinatorial list of data. This means that mathematicians can now classify curves of any complexity using this new, robust language, knowing that their results will match up with the famous, older classifications for simpler curves.

The paper also highlights that the "fingerprint" (the valuation semigroup) is the key. Just as a detective uses a fingerprint to identify a suspect, the author uses the semigroup to identify the type of singularity. This allows for a precise classification that works in any "residue characteristic" (a fancy way of saying it works in any mathematical universe, whether the numbers behave like standard integers or something more exotic).

In short, the paper takes a chaotic pile of mathematical debris and gives us a blueprint to rebuild it, proving that no matter how you choose to look at the broken pieces, the underlying structure is always the same, waiting to be decoded.

Technical Summary: Defining Reduction Types of Curves via Minimal Regular and Minimal Normal Crossings Models

Problem Statement
The paper addresses the problem of defining and classifying the "reduction type" of a smooth projective curve $C$ over a discretely valued field $K$ with an algebraically closed residue field $k$ . Traditionally, reduction types are defined via the special fibre of a "nice" model, typically the minimal regular model or the minimal regular normal crossings model.

In the case of the minimal regular normal crossings model, the reduction type is well-defined combinatorially (as seen in the work of Dokchitser), as the singularities are restricted to simple nodes.
However, for the minimal regular model (which is often the standard object of study, e.g., in the Kodaira-Néron classification for elliptic curves), a general definition of reduction type in arbitrary genus has been lacking. This is because the special fibre of a minimal regular model can contain singularities more complex than nodes (e.g., cusps, tacnodes, or higher-order singularities), and there has been no systematic classification of these singularities to form a discrete invariant in higher genus.

Methodology
The author proposes a unified definition of reduction type based on the special fibre of any regular model, with a specific focus on the minimal regular and minimal regular normal crossings models. The methodology proceeds in four main stages:

Local Analysis of Plane Curve Germs:
The paper first establishes the theory of plane curve germs over an algebraically closed field $k$ . It defines "branches" (irreducible components of the germ) and introduces the valuation semigroup $S(B)$ as the primary invariant for a branch. This semigroup, a subset of $\mathbb{Z}_{>0}$ closed under addition, encodes the equisingularity type of the branch (following Zariski).
- For reducible germs, the paper defines the intersection number between distinct branches and the notion of tangency (branches sharing the same tangent direction).
- Crucially, the paper analyzes how these invariants (semigroups, intersection numbers, tangency classes) change under blowups at closed points. Theorem 2.27 and Theorem 2.30 provide explicit formulas relating the invariants of a germ to those of its strict transform after a blowup.
Globalization to Arithmetic Surfaces:
The local results are applied to arithmetic surfaces $X \to \text{Spec } \mathcal{O}_K$ . The special fibre $X_s$ is shown to be a locally planar projective curve. The author defines an abstract type $T$ as a tuple containing:
- The set of components and their multiplicities ( $d$ ) and geometric genera ( $g$ ).
- The set of singular points and their incidence relations.
- For each singular point, the set of branches, their valuation semigroups, and the intersection numbers between branches.
Blowup Invariance and Algorithmic Translation:
The core technical result is Theorem 4.6, which proves that the type of an arithmetic surface $X$ and the type of its blowup $\tilde{X}$ at a closed point determine each other.
- The paper distinguishes between blowing up a singular point (Type 1) and a smooth point on a component (Type 2).
- Algorithm 4.8 is provided to explicitly compute the type of the blown-up model from the original type (and vice versa), utilizing the transformation rules for semigroups (via Apéry sets) and intersection numbers derived in Section 2.
Equivalence of Models:
Since any two regular models of a curve are related by a sequence of blowups and blow-downs, the author deduces that the type of any regular model determines the type of the minimal regular model and the minimal regular normal crossings model. This establishes that the "reduction type" is a well-defined invariant of the curve itself, independent of the specific regular model chosen.

Key Contributions and Results

Definition of Reduction Type (Definition 4.2): The paper introduces a rigorous definition of reduction type that captures the combinatorial configuration of the special fibre, numerical data (multiplicities, genera), and a discrete invariant for singularities (valuation semigroups and intersection numbers). This definition works for arbitrary genus and arbitrary regular models.
Equivalence of Minimal Models (Theorem 4.10): The paper proves that the minimal regular reduction type and the minimal normal crossings reduction type determine each other. Consequently, classifying curves by their minimal regular models is equivalent to classifying them by their minimal normal crossings models. An explicit algorithm allows translation between these two perspectives.
Recovery of Low-Genus Classifications:
- Genus 1: The paper recovers the classical Kodaira-Néron classification. It shows that the minimal regular reduction type of an elliptic curve corresponds exactly to a Kodaira type (e.g., $I_n, II, III, IV$ , etc.), where the singularity invariants (semigroups) match the geometric descriptions (e.g., a cusp has semigroup $\langle 2, 3 \rangle$ ).
- Genus 2: The paper recovers the Namikawa-Ueno classification. It demonstrates that the singularities appearing in the minimal regular models of genus 2 curves are limited to nodes, cusps, $(2,5)$ -cusps, and tacnodes, and that the proposed definition aligns with the existing analytic classification over $\mathbb{C}$ .
Arbitrary Characteristic: Unlike the original Namikawa-Ueno classification which relied on analytic methods over $\mathbb{C}$ , this paper's results hold over any discrete valuation ring with an algebraically closed residue field, regardless of the residue characteristic.

Significance and Claims
The paper claims that its primary contribution is providing a general, intrinsic definition of reduction type that resolves the ambiguity of singularities in minimal regular models for curves of arbitrary genus.

It bridges the gap between the combinatorial simplicity of normal crossings models and the geometric richness of minimal regular models.
It validates the use of minimal regular models for classification in higher genus by showing that the necessary singularity data (valuation semigroups) can be systematically recorded and transformed.
The author emphasizes that the definition is "discrete," making it suitable for classification purposes, and that it unifies earlier ad-hoc classifications (like Kodaira's distinction between nodes and cusps) into a single framework based on equisingularity invariants.

The paper does not propose new experimental applications or future directions beyond the immediate scope of defining and classifying these reduction types. Its significance lies in the theoretical unification of existing classification results and the provision of a robust, algorithmic framework for studying the reduction of curves over discretely valued fields.

The Core Discovery: A Universal Translator for Broken Curves

The Story of the Shards and the Fingerprint

Why This Matters for the Curious Teen

More like this