🔢 mathematics

The classification of smooth 3-fold flops

This paper establishes the analytic classification of smooth, simple 3-fold flops and provides both an intrinsic characterization and a classification of their associated contraction algebras.

Original authors: Gavin Brown, Michael Wemyss

Published 2026-10-23

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Gavin Brown, Michael Wemyss

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

In the vast landscape of modern mathematics, there is a branch dedicated to understanding the shapes of space itself, specifically those that exist in three dimensions. Imagine a smooth, continuous surface that can be stretched, bent, or twisted without tearing, yet retains its essential character. Within this world, there are special points where the geometry becomes delicate, like a thin neck or a sharp pinch. Mathematicians study how these spaces can change form through a process called a "flop." Picture a smooth curve, like a tiny loop of string, sitting inside a three-dimensional space. In a flop, this loop is contracted, squeezed down until it vanishes into a single point, and then the space is pulled back out in a different direction, creating a new, slightly altered version of the original shape. This transformation is not just a mathematical curiosity; it is a fundamental mechanism that links different versions of the same universe, playing a crucial role in both pure geometry and theoretical physics, particularly in theories describing the fabric of reality.

For decades, mathematicians have known that these flops come in different "lengths," a measure of how complex the contraction is. The simplest cases were understood long ago, but as the length increased, the possibilities seemed to multiply into an endless, chaotic variety. The question remained: is there a hidden order to this chaos? Can every possible smooth three-dimensional flop be sorted into a neat, finite list, much like the periodic table sorts chemical elements? This is the central puzzle that Gavin Brown and Michael Wemyss have now solved.

The researchers have established a complete and final classification of these smooth three-dimensional flops. They discovered that despite the apparent complexity, there are only six distinct families of these shapes, corresponding to lengths one through six. For the first time, they have provided a precise recipe for every single possible flop that can exist. Their work reveals that the universe of these shapes is far more structured than previously imagined. While the simplest flops are well-known, the more complex ones, starting from length three, possess a specific kind of flexibility. They contain continuous parameters, meaning that within a single family, there are infinitely many variations that cannot be transformed into one another, similar to how a complex machine might have dials that can be turned to create distinct, non-identical configurations.

To achieve this, the authors developed a powerful new method that translates the geometric problem of shapes and spaces into the language of non-commutative algebra. Instead of trying to visualize the twisting and turning of three-dimensional space directly, they converted the problem into studying specific algebraic structures called contraction algebras. These algebras act as a unique fingerprint for each type of flop. By analyzing these fingerprints, the team was able to prove that any smooth flop must belong to one of their six identified families. They demonstrated that if a shape does not fit into one of these categories, it simply cannot exist as a smooth flop. This approach allowed them to bypass the immense difficulty of describing the shapes directly and instead focus on the underlying algebraic rules that govern them.

A significant part of their discovery involves ruling out the existence of certain other types of geometric contractions. While smooth flops can exist up to length six, the researchers proved that a related but different type of contraction, known as a divisor-to-curve contraction, stops abruptly at length three. They showed that no smooth examples of this specific type exist for lengths four, five, or six. This finding corrects previous assumptions in the field and highlights a sharp boundary in the behavior of these geometric objects. The proof relies on the fact that the algebraic structures required for longer lengths simply do not exist in the necessary form, effectively closing the door on those possibilities.

The classification also resolves a long-standing question about numerical invariants, which are numbers used to count and distinguish these shapes. The team found that for the more complex flops of length four, five, and six, the numerical invariants are entirely determined by the length itself. In other words, once you know the length of the curve being contracted, you automatically know all the other key numerical properties of the shape. This static nature of the invariants for higher lengths was a surprising result, suggesting that the complexity of the shape is captured entirely by its length, rather than by a myriad of other fluctuating numbers.

The final result is a comprehensive list that describes every possible smooth three-dimensional flop. The list includes the equations that define the base of the shape and the specific rules for how the space is glued together. For the most complex cases, the list includes families of shapes that depend on a finite number of adjustable coefficients, confirming that these shapes have a built-in moduli, or a space of their own variations. The work confirms that the classification is not just a theoretical possibility but a concrete reality, providing a full map of the terrain. By combining deep geometric insight with sophisticated algebraic techniques, Brown and Wemyss have brought order to a chaotic corner of mathematics, showing that even in the most intricate folds of three-dimensional space, there is a strict and beautiful logic waiting to be discovered.

Technical Summary: The Classification of Smooth 3-Fold Flops

1. Problem Statement

The paper addresses the analytic classification of smooth, simple 3-fold flops. These are birational morphisms $X \to X_{\text{con}}$ that contract a single nonsingular rational curve $C \cong \mathbb{P}^1$ within a smooth 3-fold $X$ , satisfying the condition $K_X \cdot C = 0$ . Such contractions are fundamental in both algebraic geometry and theoretical physics (linking birational Calabi–Yau 3-folds).

The central challenge is that these flops possess an inherent symmetry: the birational map $X \dashrightarrow X^+$ is a flop, and so is its inverse $X^+ \dashrightarrow X$ . This symmetry makes it impossible to distinguish or prefer one side over the other. Consequently, any classification up to analytic or complete local isomorphism must account for this symmetry. The authors aim to provide a "full and final classification" of these smooth, simple flops (specifically those contracting a single curve), analogous to the classical lists of Arnold, Du Val, and Klein for surface singularities.

2. Methodology

The authors employ a noncommutative approach, utilizing the theory of contraction algebras to translate the geometric problem into an algebraic one.

2.1. The Meta-Strategy

The classification proceeds via a chain of inclusions involving three sets:

$\mathcal{F}$ : The set of all smooth 3-fold flops (up to isomorphism) contracting a single curve.
$\mathcal{A}$ : The set of finite-dimensional contraction algebras associated with these flops.
$\mathcal{B}$ : A set of algebras defined by specific intrinsic axioms (algebraic length).
$\mathcal{C}$ : A concrete, finite list of algebraic forms (normal forms).

The strategy establishes the equality $\mathcal{A} = \mathcal{B} = \mathcal{C}$ through the sequence:
$\mathcal{A} \subseteq \mathcal{B} \subseteq \mathcal{C} \subseteq \mathcal{A}$
This implies that the set of all smooth, simple flops (contracting a single curve) is in bijection with the set $\mathcal{C}$ , providing a complete classification within this specific context.

2.2. Key Technical Tools

Contraction Algebras ( $A_{\text{con}}$ ): Defined via the deformation theory of exceptional curves or noncommutative singularity theory. For a flop contracting a curve $C$ , $A_{\text{con}} \cong \text{Jac}(f)$ , the Jacobi algebra of an element $f \in \mathbb{C}\langle\langle x, y \rangle\rangle$ .
Algebraic Length ( $\ell$ ): A new invariant defined for an element $f$ . An element has algebraic length $\ell$ ( $1 \le \ell \le 6$ ) if there exists a central element $z \in Z(\text{Jac}(f))$ such that $\text{Jac}(f)/((z)) \cong e\Pi e$ , where $\Pi$ is an ADE preprojective algebra and $e$ is a primitive idempotent associated with the number $\ell$ in the highest root data.
Necklace Polynomials: The authors utilize "necklace polynomials" ( $N_{i,j}$ ), which are cyclic sums of monomials, to construct the normal forms. These serve as the noncommutative analogues of commutative monomials.
Tschirnhausen Transformations: A sequence of unitriangular automorphisms is used to "funnel" arbitrary power series into specific normal forms by eliminating lower-degree terms and pushing them into higher degrees, compatible with cyclic equivalence.
Standard Bases: Noncommutative standard bases (local Gröbner bases) are used to bound the dimension of the Jacobi algebras and determine finiteness conditions.

3. Key Contributions and Results

3.1. Intrinsic Characterization of Contraction Algebras

The paper provides an intrinsic characterization of contraction algebras without reference to the underlying geometry.

Theorem 1.2: A $\mathbb{C}$ -algebra $B$ is isomorphic to the contraction algebra of a smooth, simple 3-fold flop (or crepant divisor-to-curve contraction) if and only if $B \cong \text{Jac}(f)$ for some $f$ with an algebraic length and $\dim_{\mathbb{C}} B < \infty$ .
This resolves the problem of identifying which abstract algebras arise from geometry.

3.2. Classification of Algebraic Length (The List $\mathcal{C}$ )

The authors derive a complete list of normal forms for elements $f$ that possess an algebraic length.

Theorem 1.3: If $\text{Jac}(f)$ $Jac (f)$ is finite-dimensional, $f$ $f$ is isomorphic to one of the following forms (indexed by length $\ell$ $ℓ$ ):
1. $\ell=1$ : $x^2 + y^n$ ( $n \ge 2$ )
2. $\ell=2$ : $x^2y + y^{2n} + \epsilon y^{2m-1}$
3. $\ell=3$ : $x^3 + xy^3 + y^n + y^{n+1}q$ ( $n \ge 4$ )
4. $\ell=4$ : $x^3 + y^4 + N_{2,3} + \sum \alpha_{i,j} N_{i,j}$
5. $\ell=5$ : $x^3 + N_{2,2} + y^5 + xy^5p + y^6q$
6. $\ell=6$ : $x^3 + y^4 + N_{3,3} + \lambda N_{2,4} + \sum \alpha_{i,j} N_{i,j}$ ( $\lambda \neq 0$ )
The paper also classifies the infinite-dimensional cases (associated with divisor-to-curve contractions), showing they stop at length 3.

3.3. Geometric Classification of Flops

Translating the algebraic results back to geometry, the paper classifies the local structure of the flopping contraction.

Theorem 1.4: The completion of a smooth, simple 3-fold $X$ along the flopping curve $C$ is isomorphic to the completion of a variety defined by gluing two copies of $\mathbb{A}^3$ via a specific rule involving a polynomial $E \in \mathbb{C}[a^{\pm 1}, z]$ . The form of $E$ corresponds exactly to the algebraic normal forms listed above.
This provides a uniform, elementary geometric description of all smooth, simple flopping neighborhoods (contracting a single curve).

3.4. Numerical Invariants and Moduli

Gopakumar–Vafa (GV) Invariants: The paper calculates the GV invariants for all lengths. A significant finding is that for $\ell \ge 4$ , the GV invariants are static and determined solely by the length $\ell$ . For example, for $\ell=6$ , the invariants are $(6, 6, 4, 3, 2, 1)$ . This suggests that numerical curve invariants alone are insufficient to distinguish flops of higher length; the full algebraic structure is required.
Moduli: The classification reveals that for $\ell \ge 3$ , flops possess continuous parameters (moduli). The coefficients in the polynomials $Q_\ell$ in the geometric gluing rule represent these moduli.

3.5. Divisor-to-Curve Contractions

The paper extends the classification to crepant divisor-to-curve contractions (Type III contractions).

Corollary 1.9: Smooth crepant divisor-to-curve contractions of length $\ell \ge 4$ do not exist.
The only such contractions occur for $\ell=1, 2, 3$ . The $\ell=3$ case is realized by a specific example involving a singular curve germ, correcting previous assumptions in the literature that such contractions always contract to smooth curves.

4. Significance and Claims

The authors claim this work provides a "full and final classification" of smooth, simple 3-fold flops contracting a single curve.

Completeness: The classification is exhaustive for smooth, simple flops contracting a single curve, covering all lengths from 1 to 6. It subsumes previous classifications for $\ell=1$ (Atiyah/Pagoda) and $\ell=2$ , and includes the four known examples for $\ell \ge 3$ while revealing new families with moduli.
Intrinsic Nature: The work establishes that the contraction algebra is the "minimal classifying object." The algebraic length is shown to be equivalent to Kollár's geometric length invariant, bridging noncommutative algebra and birational geometry.
Non-existence Results: The paper definitively rules out the existence of smooth crepant divisor-to-curve contractions for lengths greater than 3, attributing this obstruction to noncommutative algebraic constraints.
Methodological Shift: The paper demonstrates that noncommutative techniques (contraction algebras, necklace polynomials) are not just auxiliary tools but are essential for achieving a complete classification where purely geometric methods (like analyzing the base ring $R$ ) become intractable due to the complexity of the defining polynomials.

The paper concludes that while numerical invariants (like GV numbers) are useful, the true classification of smooth, simple flops (contracting a single curve) relies on understanding the isomorphism classes of their associated noncommutative algebras.