🔢 mathematics

Toric Exoflops and Categorical Resolutions

This paper establishes sufficient criteria under which toric exoflops, a process involving the partial compactification and birational transformation of gauged Landau-Ginzburg models, yield crepant categorical resolutions or equivalences of derived categories for specific complete intersections in toric stacks.

Original authors: Tyler L. Kelly, Aimeric Malter

Published 2026-04-13

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Tyler L. Kelly, Aimeric Malter

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 design a perfect, smooth building (a mathematical object called a "Calabi-Yau variety"). Sometimes, your blueprint leads to a structure with a jagged, broken corner—a singularity. In the world of math, these broken corners are annoying because they break the rules of smooth geometry.

This paper introduces a clever construction technique called an "Exoflop" to fix these broken buildings without actually tearing them down and starting over. It's like a magical renovation that turns a broken house into a perfect one, or swaps one perfect house for a different-looking but mathematically identical one.

Here is the breakdown of how this works, using simple analogies:

1. The Problem: The "Broken" House

In mathematics, we often study shapes defined by equations. Sometimes, these shapes have "kinks" or sharp points (singularities). If you try to do calculus or physics on a kinked shape, things break.

The Goal: We want to find a smooth version of this shape that behaves exactly the same way mathematically, even if it looks different. This is called a Categorical Resolution.

2. The Tool: The "Gauged Landau-Ginzburg" Model

Instead of looking at the broken house directly, the authors use a special "shadow" or "projection" of it called a Gauged Landau-Ginzburg (LG) model.

The Analogy: Imagine the house is a complex 3D sculpture. Instead of looking at the sculpture, you look at its shadow cast on a wall. The shadow is easier to manipulate. In this paper, the "shadow" is a mathematical system involving a space, a group of symmetries, and a special function (the "potential").

3. The Two-Step Renovation: The "Exoflop"

The authors propose a two-step process to fix the broken house using its shadow. They call this an Exoflop (a play on "Exotic Flop").

Step A: The "Exo" (Partial Compactification)

What it is: You take your shadow (the LG model) and you expand the room it lives in. You add a little bit of extra space around the edges, but you do it carefully so the "broken" part of the house gets smoothed out by the new walls.
The Metaphor: Imagine your broken sculpture is sitting on a small table. You move it to a huge, empty warehouse. By adding this extra space, you can "fill in" the cracks of the sculpture using the new room's geometry. The authors prove that if you pick the right warehouse, the broken sculpture becomes a "Crepant Categorical Resolution"—a fancy way of saying "a perfect, smooth version that keeps all the original energy and properties."

Step B: The "Flop" (Variation of GIT)

What it is: Once you are in this new, expanded warehouse, you can rearrange the furniture. In math terms, you change the "stability rules" (Geometric Invariant Theory) that define how the space is built.
The Metaphor: Imagine you have a room with a movable wall. You slide the wall to a new position. Suddenly, the room looks completely different! One side of the room might look like a garden, and the other like a library. But mathematically, the total amount of space and the rules of the room haven't changed. You just shifted the perspective.
The Result: This "flop" can transform your original broken house into a completely different-looking house (a different Calabi-Yau variety). The amazing part is that these two different houses are mathematically identical (Derived Equivalent). They are like two different languages describing the exact same story.

4. Why This Matters: The "Mirror" Connection

In physics and math, there is a famous idea called Mirror Symmetry. It suggests that for every complex universe (A-model), there is a "mirror" universe (B-model) that behaves differently but is secretly the same.

The Breakthrough: This paper shows that the "Exoflop" is a machine that can find these mirrors.
The Analogy: If you have a weird, broken mirror, the Exoflop tells you exactly how to reshape it so it reflects a perfect image. Furthermore, it shows you that there are multiple ways to build this perfect mirror, and they are all interchangeable.

5. The "Recipe" for Success

The authors didn't just guess; they found a specific recipe (combinatorial criteria) to know when this trick will work.

The Recipe: If your shape is built from specific types of "polytopes" (mathematical shapes like cubes or pyramids) that fit together in a very specific, symmetric way (called "reflexive Gorenstein cones"), then the Exoflop is guaranteed to work.
The Payoff: If you follow this recipe, you can take a singular (broken) shape and turn it into a smooth one, or swap it for a different smooth shape, and you know for a fact they are mathematically equivalent.

Summary in One Sentence

This paper provides a magical construction manual (the Exoflop) that allows mathematicians to take a broken, jagged geometric shape, expand its universe, and rearrange the rules to reveal a smooth, perfect version of itself—or a completely different shape that is secretly its twin—solving long-standing problems in geometry and mirror symmetry.

1. Problem Statement

The paper addresses the challenge of establishing derived equivalences and constructing crepant categorical resolutions for singular Calabi-Yau (CY) complete intersections within toric varieties.

Context: In mirror symmetry and algebraic geometry, singular varieties often require resolution. While geometric resolutions (blowing up singularities) exist, they can be computationally heavy and may not preserve the derived category structure in a way that reveals mirror symmetry relationships.
The Gap: Previous work by Aspinwall introduced "exoflops" to find categorical resolutions for specific singular K3 surfaces. However, these results were largely example-based, relying on case-by-case computations. There was a lack of a general, combinatorial framework to determine when an exoflop yields a categorical resolution or a derived equivalence for general Complete Intersections in Toric (CICY) varieties.
Goal: To formalize the exoflop operation, provide sufficient combinatorial criteria for its success, and establish a "dynamic web" of derived equivalences between different CY varieties (including their mirrors).

2. Methodology

The authors utilize the framework of Gauged Landau-Ginzburg (LG) models and Factorization Categories. The methodology proceeds in three main stages:

A. The Exoflop Operation

The exoflop is defined as a two-step process acting on a gauged LG model $(X, G, W)$ :

Exo (Partial Compactification): The space $X$ $X$ (often a vector bundle over a toric variety) is partially compactified to a larger space $\bar{X}$ $\overset{ˉ}{X}$ . The group action $G$ $G$ is extended to $\bar{G}$ $\overset{ˉ}{G}$ , and the superpotential $W$ $W$ is extended to a global function $\bar{W}$ $\overset{ˉ}{W}$ .
- Key Insight: If the original complete intersection $Z$ is singular, its critical locus in the LG model is not proper. A "good" partial compactification can remedy this, turning the factorization category $D^{abs}[\bar{X}, \bar{G}, \bar{W}]$ into a crepant categorical resolution of the singular variety.
Flop (Variation of GIT): The authors apply Variation of Geometric Invariant Theory (VGIT) to the compactified model. By varying the stability parameter (linearization), they perform a birational transformation (a flop) to a new LG model $(X', G', W')$ $(X^{'}, G^{'}, W^{'})$ .
- Result: This often maps the original singular variety to a different smooth variety $Z'$ (or a different orbifold), establishing a relationship between their derived categories.

B. Toric and Convex Geometry Framework

To make this rigorous for CICYs, the authors translate the problem into convex geometry:

Cayley Polytopes: They utilize the structure of Cayley polytopes and reflexive Gorenstein cones. A CICY is defined by a partition of the anticanonical divisor, corresponding to a Cayley cone.
Saturated Sets: They define "saturated" sets of lattice points to ensure that the chosen linear systems for the superpotential are sufficiently generic to avoid extra singularities.
Secondary Fans: They use the secondary fan of the toric variety to parameterize the GIT quotients, ensuring that the partial compactification and the subsequent flop are well-defined geometric operations.

C. Categorical Tools

Factorization Categories: They work with $D^{abs}[X, G, W]$ , the absolute derived category of factorizations, which is equivalent to the derived category of coherent sheaves on the zero locus of $W$ (via Knörrer periodicity/Orlov's theorem).
Crepant Resolutions: They employ the definition of categorical resolutions (Kuznetsov) where a functor $F: \tilde{D} \to D$ and its adjoint $G$ satisfy specific properties (fully faithful, isomorphism on perfect complexes).

3. Key Contributions

1. Generalization of Exoflops

The paper moves beyond specific examples to provide a general theory for CICYs in toric Fano varieties. It formalizes the exoflop as a mechanism to relate derived categories of different toric complete intersections.

2. Combinatorial Criteria for Resolutions

The authors provide explicit, checkable combinatorial conditions (Theorem 1.2, Corollary 5.7) under which an exoflop yields a crepant categorical resolution.

Condition: The dual cone $\sigma_W^\vee$ associated with the superpotential must be a completely split reflexive Gorenstein cone.
Implication: If this condition holds and the coefficients of the superpotential are generic, the resulting variety $Z'$ is smooth, and the functor between $D^b(\text{coh } Z)$ and $D^b(\text{coh } Z')$ is a derived equivalence (if both are smooth) or a crepant resolution (if $Z$ is singular).

3. Connection to Mirror Symmetry

The paper demonstrates that the varieties $Z'$ obtained via exoflops are precisely those that admit Batyrev-Borisov mirrors.

This provides a constructive method to find the "mirror A-model" of a smooth CY complete intersection $Z$ by finding a $Z'$ derived equivalent to it.
It unifies different mirror constructions (Batyrev-Borisov and Berglund-Hübsch-Krawitz) under the umbrella of derived equivalence via exoflops.

4. New Examples and Generalizations

Aspinwall's Example: The authors re-prove Aspinwall's result on singular K3 surfaces using their general framework, showing how the partial compactification resolves the singularity categorically.
Libgober-Teitelbaum Family: They generalize the Libgober-Teitelbaum mirror construction for higher-dimensional CYs ( $n \geq 4$ ), proving that the Batyrev-Borisov construction provides a categorical resolution for singular members of this family.
Derived Equivalences without Birationality: They exhibit cases where $Z$ and $Z'$ are derived equivalent but not obviously birational, highlighting the power of the exoflop in discovering hidden symmetries.

4. Main Results

Theorem 1.1 / Corollary 5.2: If Assumption 5.1 holds (related to the cone structure of the partial compactification) and $Z'$ is smooth, there exists a crepant categorical resolution $F: D^b(\text{coh } Z') \to D^b(\text{coh } Z)$ and a fully faithful functor $G: \text{Perf } Z \to D^b(\text{coh } Z')$ . If both are smooth, they are derived equivalent.
Theorem 1.2 / Corollary 5.7: If the cone $\sigma_W$ and its dual are completely split reflexive Gorenstein cones and the superpotential coefficients are generic, then a crepant categorical resolution exists. If $Z$ is smooth, this is a derived equivalence.
Corollary 4.13: Summarizes the chain of relations:
$Z \xrightarrow{\text{LG}} (X, G, W) \xrightarrow{\text{exo}} (\bar{X}, \bar{G}, \bar{W}) \xrightarrow{\text{flop}} (X', G', W') \xrightarrow{\text{LG}} Z'$
This chain establishes the derived equivalence or resolution between $Z$ and $Z'$ .

5. Significance

Unification of Mirror Symmetry: The paper provides a rigorous bridge between the geometric operation of exoflops and the Homological Mirror Symmetry (HMS) conjecture. It suggests that "multiple mirrors" (different constructions for the same symplectic manifold) are derived equivalent, as predicted by HMS.
Algorithmic Approach: By reducing the problem to checking properties of polytopes and cones (integrally closed, reflexive, Gorenstein), the authors provide a pathway to algorithmically generate categorical resolutions for singular CYs without performing complex geometric blow-ups.
Handling Singularities: The framework offers a robust method to handle singular CY complete intersections, which are common in string theory and algebraic geometry, by replacing them with smooth, derived-equivalent models.
Future Directions: The paper opens questions about generalizing these constructions to higher codimensions and exploring "f-duality" (iterative mirror constructions) in the context of complete intersections.

In summary, Kelly and Malter successfully formalize the "exoflop" as a powerful tool in modern algebraic geometry, providing a combinatorial toolkit to navigate the landscape of derived categories of Calabi-Yau varieties and their mirrors.