🔢 mathematics

Non-smoothable curve singularities

This paper investigates non-smoothable curve singularities by applying dimension counting and Dedekind different-based invariants to study monomial curves and cones over point sets, ultimately providing new explicit examples of non-smoothable singularities, including Gorenstein ones, and proving that cones over general self-associated point sets of genus at least 11 are not smoothable.

Original authors: Jan Stevens

Published 2026-06-02

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Jan Stevens

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 an architect trying to build a perfect, smooth bridge. In the world of mathematics, specifically in the study of shapes called "curves," there are points where the bridge might get bumpy, twisted, or knotted. These are called singularities.

Most of the time, if you have a bumpy knot, you can wiggle it around (mathematically, "deform" it) until it becomes perfectly smooth. Think of it like untangling a headphone cord; with enough fiddling, it usually straightens out.

However, this paper by Jan Stevens asks a very specific question: Are there some knots that are so hopelessly tangled that no amount of wiggling can ever make them smooth?

The answer is yes. The paper explores these "un-smoothable" knots and tries to understand what the most common (or "generic") version of these impossible knots looks like.

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

1. The Two Ways to Prove a Knot is "Unfixable"

The author explains that mathematicians have two main tools to prove a knot cannot be smoothed out.

The "Too Big to Fit" Method (Dimension Counting):
Imagine you have a giant box of tangled cords. You try to fit them into a smaller box that only holds smooth cords. If your box of tangled cords is physically too large to fit inside the "smooth" box, then you know you can't turn them all smooth. The author uses this logic to show that certain families of curves are simply too vast to be part of the "smooth" family.
The "Weighted Scale" Method (The Buchweitz Criterion):
Imagine every knot has a hidden "weight" or a specific score based on its complexity. For a knot to be smoothable, its score must stay below a certain limit. The author uses a special mathematical scale (related to something called the "Dedekind different") to weigh these knots. If the score is too high, the knot is mathematically doomed to stay knotted forever.

2. The "Monomial" and "Line" Knots

To test these theories, the author looks at two specific types of knots:

Monomial Curves: These are like knots made from a very strict, repetitive pattern (like a string of beads where the pattern is always the same). They are easy to calculate but very rigid.
Lines Through the Origin: Imagine a bundle of straws all stuck together at one single point in the center, fanning out in different directions. The author studies what happens when you have a lot of these straws.

3. The Big Discovery: New "Unfixable" Knots

The paper doesn't just repeat old findings; it finds new examples of knots that cannot be smoothed.

The "Gorenstein" Surprise: For a long time, mathematicians thought that a certain type of "symmetric" knot (called Gorenstein) could always be smoothed out if you tried hard enough. The author proves this is false. They found a specific, symmetric knot that is stubbornly un-smoothable.
The "Self-Associated" Point Set: The author uses a concept called "self-associated points" (think of a group of people standing in a circle where everyone's position is perfectly mirrored by someone else's). If you have enough of these people (specifically, if the number is high enough, like 11 or more), the shape they form (a "cone") cannot be smoothed. It's like trying to flatten a specific arrangement of stars in the sky; no matter how you push, the pattern resists becoming a flat, smooth sheet.

4. The "Generic" Knot

The paper asks: "If we pick a random, typical un-smoothable knot, what does it look like?"

The Conjecture: The author suggests that the most common un-smoothable knots are actually made of smooth branches that just happen to cross each other badly.
The Analogy: Think of a bundle of smooth, straight wires. If you twist them so they all cross at one point, you get a knot. The author suspects that the "worst" knots are usually just these smooth wires crossing, rather than wires that are themselves twisted and broken.

5. The "Large Family" Argument

One of the most powerful tools in the paper is the idea of a "Large Family."

Imagine you have a family of knots. If this family is so huge that it spills over the boundaries of the "smooth" world, then the average knot in that family must be un-smoothable.
The author uses this to show that for certain numbers of lines (straws) and dimensions (directions), the "average" knot is unfixable. For example, if you have 10 lines in a 6-dimensional space, the typical arrangement is un-smoothable.

Summary

In short, this paper is a detective story in the world of geometry. The author investigates the "criminals" of the mathematical world: curves that refuse to become smooth. By using two different investigative techniques (checking the size of the family and weighing the complexity), the author:

Confirms that some knots are permanently broken.
Finds new types of broken knots, including some that were previously thought to be fixable.
Proposes that the most common broken knots are actually just smooth lines crossing each other in a messy pile.

The paper concludes that while we can find these un-smoothable knots, our understanding of them is still limited, and we need more tools to fully map out the landscape of these mathematical tangles.

Technical Summary: Non-Smoothable Curve Singularities

Problem Statement
The paper addresses the deformation theory of reduced curve singularities. While it is known that all smoothing components of the versal deformation space of a curve singularity share the same dimension, the deformation space itself may contain components of different dimensions. The central problem investigated is the nature of the "generic singularities" that appear as the general fiber over a non-smoothing component. Specifically, the author seeks to identify which singularities are inherently non-smoothable—meaning they cannot be deformed into a smooth curve—and to provide explicit examples of such singularities, including previously unknown cases of non-smoothable Gorenstein curve singularities.

Methodology
The author employs two primary general methods to establish non-smoothability, supplemented by computational algebra:

The "Large Family" Argument: This method, rooted in work by Mumford and refined by Greuel, involves exhibiting a family of singularities $\pi: C_T \to T$ with a singular section such that the dimension of the parameter space $T$ is at least the Deligne number $e(C_0)$ (the dimension of a smoothing component). If the family is irreducible and contains no smooth fibers, the general fiber cannot be smoothable, as its image in the versal deformation space would have dimension $\ge e(C_0)$ but cannot constitute a smoothing component.
The Buchweitz Criterion: This method utilizes the semicontinuity of a specific invariant related to the Dedekind different. For a curve singularity $C$ , the author defines the $k$ -th normalized Dedekind invariant $d_k$ . If $C$ is smoothable (or deforms to a Gorenstein singularity), then $d_k \le 2k\delta$ must hold for all $k$ , where $\delta$ is the singularity degree. Violation of this inequality proves non-smoothability.
Computational Algebra: The paper relies heavily on computer algebra systems (Singular and Macaulay2) to compute infinitesimal deformations ( $T^1$ ), obstruction spaces ( $T^2$ ), and the specific invariants required for the Buchweitz criterion. The analysis focuses heavily on quasi-homogeneous singularities, particularly monomial curves and cones over point sets in projective space, where $G_m$ -actions simplify computations.

Key Contributions and Results

Revisiting Known Examples: The paper systematically analyzes known non-smoothable examples, such as monomial curves and cones over lines through the origin ( $L^n_r$ ). It provides explicit calculations of the Deligne number $e$ and the dimension of the deformation space for generic configurations.
Non-Smoothable Gorenstein Singularities: A significant contribution is the construction of non-smoothable Gorenstein curve singularities. Since the Buchweitz criterion is trivial for Gorenstein curves (where $d_k = 2k\delta$ $d_{k} = 2 k δ$ ), the author uses a double cover construction based on Stöhr-Torres semigroups.
- Proposition 4.2: The author presents a specific non-smoothable Gorenstein curve singularity, $L^{13}_{26}$ , which is the cone over a self-associated point set. This singularity is shown to be non-smoothable because its deformation leads to a non-smoothable singularity ( $L^9_{13}$ ) via a specific deformation path.
Cones over Self-Associated Point Sets: The paper investigates cones over self-associated point sets (where the Gale transform is projectively equivalent to the original set).
- Theorem 4.5 & 4.7: It is proven that the general Gorenstein cone over a self-associated point set in $\mathbb{P}^{n-1}$ (specifically $L^{n-1}_{2n}$ ) is not smoothable if the genus $g \ge 11$ . For $g \le 8$ and $g=10$ , smoothability holds or is conditional. This result connects the non-smoothability to the fact that the point set cannot be a hyperplane section of a canonical curve of genus $g$ .
Explicit Non-Smoothable Monomial Curves: The paper provides new explicit examples of non-smoothable monomial curves, including cases where the Buchweitz criterion fails for $k=1$ but succeeds for $k=2$ (Example 2.7), and examples with smaller embedding dimensions (Example 2.8).
Irreducible vs. Reducible Cases: The author distinguishes between irreducible and reducible singularities. While many non-smoothable examples are reducible (e.g., unions of lines), the paper constructs irreducible non-smoothable curves (Theorem 3.9) and notes that many non-smoothable singularities deform into curves with smooth branches.
Dimensional Anomalies: The paper identifies examples where the dimension of the base space of the versal deformation is strictly less than the Deligne number $e$ (Example 1.9), challenging simple dimensional expectations.

Significance and Claims
The paper claims to advance the understanding of the structure of deformation spaces for curve singularities by:

Providing the first explicit examples of non-smoothable Gorenstein curve singularities, a class previously thought to be more amenable to smoothing or lacking known counterexamples.
Establishing a precise threshold ( $g \ge 11$ ) for the non-smoothability of cones over general self-associated point sets, linking this geometric property to the impossibility of these sets being hyperplane sections of canonical curves.
Offering a comprehensive computational and theoretical framework for determining non-smoothability, combining the "large family" dimension count with the Buchweitz invariant.

The author modestly concludes that while the knowledge of non-smoothable singularities is expanding, it remains limited, particularly regarding irreducible curves with large embedding dimensions. Based on the collected evidence, the paper proposes Conjecture 5.1: All branches of generic curve singularities are smooth. This suggests that while the global singularity may be non-smoothable, its local analytic branches are smooth, even if their tangents are not in general position.