🔢 mathematics

Conic modules, secondary fans and non-commutative resolutions

This paper establishes a connection between conic modules and Bondal-Thomsen line bundles to simplify the combinatorics of non-commutative crepant resolutions, providing necessary and sufficient conditions for incomplete sums of conic modules to form such resolutions and classifying their existence for almost simplicial Gorenstein cones.

Original authors: Aimeric Malter

Published 2026-03-26

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: 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

The Big Picture: Fixing a Cracked Pot

Imagine you have a beautiful, intricate ceramic pot (a mathematical object called a Toric Variety). Unfortunately, it has a crack or a sharp point where the clay didn't smooth out properly. In math, we call this a singularity.

Usually, to fix a cracked pot, you try to reshape it into a smooth, perfect sphere. This is called a resolution. However, sometimes there isn't just one way to smooth it out; there might be several different shapes that work. Mathematicians want to know: Do all these different smooth shapes share a secret common language?

Enter Non-Commutative Crepant Resolutions (NCCRs). Instead of physically reshaping the pot, this paper suggests we can "fix" the pot by changing the rules of how we interact with it (using algebra instead of geometry). It's like saying, "We don't need to melt the pot down; we just need to invent a new set of instructions that makes the crack disappear in our calculations."

The Main Characters

To solve this puzzle, the author introduces three main tools:

Conic Modules (The Building Blocks):
Think of these as specific types of Lego bricks. Each brick has a unique shape and color. In the math world, these bricks are called "conic modules." The authors of previous papers showed that if you take every single type of brick and snap them together, you can build a structure that fixes the pot. But that's a huge, heavy structure.
The Goal: Can we build a fix using only a few specific bricks? (This is called an "incomplete sum").
The Secondary Fan (The Map):
Imagine you have a giant map of a city. The "Secondary Fan" is a special map that shows you all the possible ways to arrange your Lego bricks. It divides the map into different districts (called chambers).
The Innovation: The author realized that instead of trying to build the whole Lego structure brick-by-brick in 3D space, you can just look at this 2D map. If you can find a path on the map that connects certain points without getting stuck, you know you have a valid fix.
The Bondal-Thomsen Collection (The Translator):
This is a dictionary that translates between the "Lego bricks" (conic modules) and the "Map districts" (points on the Secondary Fan). It turns out that every brick corresponds to a specific dot on the map. This translation is the key to simplifying the problem.

The Core Discovery: "Lockable" and "Incredulous" Sets

The paper asks: Which specific collection of bricks (an incomplete sum) will successfully fix the pot?

The author introduces two fun concepts to answer this:

Lockable Sets: Imagine you have a set of Lego instructions. If you follow them, you might need to use a brick that isn't in your set. If you can "substitute" that missing brick with a combination of other bricks you do have, and you can keep doing this until you never need a brick outside your set, your set is Lockable. It's like a self-sustaining ecosystem; you never need to go outside the group to finish the job.
Incredulous Sets: This is a Lockable set with a special bonus. Not only is it self-sustaining, but the final structure is perfectly balanced. In math terms, the "length" of the instructions is exactly right to fix the crack without over-engineering it. If a set is Incredulous, it creates a perfect NCCR (a perfect fix).

The Big Breakthrough:
The author proves that you don't need to check the complex 3D Lego instructions to see if a set is Incredulous. You just need to look at the Map (Secondary Fan).

If you can find a set of dots on the map that are "Incredulous" (they form a perfect, balanced path), then the corresponding bricks will fix the pot.
This reduces a massive, complex 3D calculation to a much simpler 2D puzzle.

The "Almost Simplicial" Case: The Special Cone

The paper focuses on a specific type of pot called an Almost Simplicial Gorenstein Cone.

Analogy: Imagine a pyramid. A "Simplicial" cone is a perfect pyramid with triangular sides. An "Almost Simplicial" cone is a pyramid that has exactly one extra side, making it slightly wobbly or irregular.
The author classifies exactly which of these slightly wobbly pyramids can be fixed using this "Incredulous" method.

The Result:
It turns out that most of these wobbly pyramids cannot be fixed this way. Only very specific shapes work. The author found three specific patterns of "bricks" (mathematically described as collections of numbers) that allow for a perfect fix:

A specific pattern involving the numbers 2, 1, -1, -1, -1.
A pattern involving 1, 1, 1, -1, -1, -1.
A shape that looks like a Trapezoid (a four-sided shape with one pair of parallel sides).

If your pot's shape doesn't match these specific patterns, you can't fix it using this specific "incomplete sum" method.

Why Does This Matter?

Simplification: Before this paper, checking if a pot could be fixed required solving incredibly difficult 3D puzzles. Now, mathematicians can just look at a 2D map (the Secondary Fan) and check if the dots line up. It's like going from solving a Rubik's cube in your head to just looking at a solved picture on the box.
Efficiency: It tells us exactly when we can use a "smaller" fix (incomplete sum) rather than a "massive" fix (complete sum). This is important because smaller fixes are computationally cheaper and often more elegant.
Classification: It draws a clear line in the sand. We now know exactly which types of "wobbly" mathematical shapes can be smoothed out using this specific algebraic trick and which ones cannot.

Summary in One Sentence

The author discovered a "map" (the Secondary Fan) that lets us easily determine if a specific, smaller set of mathematical building blocks can perfectly repair a broken geometric shape, proving that this only works for very specific, symmetrical shapes like those based on trapezoids.

1. Problem Statement and Context

The Core Problem:
The paper addresses the existence and construction of Non-Commutative Crepant Resolutions (NCCRs) for affine Gorenstein toric algebras. An NCCR of a normal domain $R$ is an algebra $\Lambda = \text{End}_R(M)$ , where $M$ is a finitely generated reflexive $R$ -module, such that $\Lambda$ has finite global dimension and is a maximal Cohen-Macaulay (MCM) $R$ -module.

Conjecture 1.1 (Van den Bergh): Every affine Gorenstein toric algebra admits an NCCR.
Previous Work: Faber, Muller, and Smith (FMS) [FMS19] proved that for simplicial cones, a complete sum of conic modules yields an NCCR. However, for non-simplicial cones, a complete sum often fails to be crepant (i.e., the global dimension is finite, but the algebra is not MCM).
The Gap: FMS provided an example where an incomplete sum of conic modules (excluding certain isomorphism classes) yields an NCCR for a non-simplicial cone. This raises the fundamental question: Under what conditions does an incomplete sum of conic modules $B = \bigoplus_{v \in I} A_v$ yield an NCCR?

2. Methodology

The author develops a computational framework linking conic modules to the Bondal-Thomsen collection of line bundles on toric Deligne-Mumford (DM) stacks, utilizing the geometry of the secondary fan (or GKZ fan).

A. Linking Conic Modules to the Bondal-Thomsen Collection

Conic Modules ( $A_v$ ): Defined for $v \in M_{\mathbb{R}}$ as specific graded submodules of the toric algebra. They correspond to chambers of constancy in $M_{\mathbb{R}}$ .
Bondal-Thomsen Collection ( $\Theta_X$ ): A collection of line bundles on a toric DM stack $X$ associated with a simplicial refinement of the cone.
Correspondence: The paper establishes that conic modules correspond to elements of the Bondal-Thomsen collection. Specifically, the endomorphism algebra of a sum of conic modules is isomorphic to the endomorphism algebra of the corresponding sum of line bundles in the Cox category.
Reduction to Lattice Points: The collection $\Theta_X$ is in bijection (up to torsion) with lattice points inside a zonotope $Z_X$ in the class group. This allows the problem to be translated from the geometry of chambers in $M_{\mathbb{R}}$ to the combinatorics of lattice points in $Z_X$ .

B. The Secondary Fan and Valid Paths

Secondary Fan ( $\Sigma_{GKZ}$ ): A fan in the space of divisors whose maximal cones correspond to different simplicial refinements of the original cone $\sigma$ .
Vectors $\beta_\rho$ : The map from the divisor group to the class group sends ray generators $D_\rho$ to vectors $\beta_\rho$ . These vectors generate the secondary fan and define the zonotope.
Complexes $K^\bullet_v$ : FMS constructed minimal projective resolutions for simple modules using complexes of conic modules. The author reinterprets the appearance of a conic module $A_w$ in the complex $K^\bullet_v$ via paths in the zonotope.
Valid Paths: A path from lattice point $P_1$ $P_{1}$ to $P_2$ $P_{2}$ is a subset of vectors $\{\beta_\rho\}$ ${β_{ρ}}$ summing to $P_2 - P_1$ $P_{2} - P_{1}$ . A path is valid if it satisfies specific interval conditions related to the coefficients of the lattice points.
- Key Result: $A_w$ appears in $K^\bullet_v$ if and only if there is a valid path from $f_\sigma(-d(w))$ to $f_\sigma(-d(v))$ .

C. Substitution and Lockability

Substitution: If a complex $K^\bullet_v$ contains a conic module $A_j$ (where $j$ is not in the chosen sum), one can "splice" the complex $K^\bullet_j$ into $K^\bullet_v$ to eliminate $A_j$ .
Lockable Sets: A set of indices $I$ is lockable if a finite sequence of substitutions eliminates all conic modules not in $I$ from the complexes associated with $I$ . This ensures the resulting endomorphism algebra has finite global dimension (an NCR).
Incredulous Sets: A lockable set is incredulous if, after substitution, all resulting complexes have length exactly equal to $\dim \sigma$ . This ensures the algebra is MCM (crepant).

3. Key Contributions

Theoretical Unification: The paper rigorously links the combinatorial construction of conic modules (FMS) with the geometric Bondal-Thomsen collection (Ballard et al.), providing a unified language for studying NCCRs.
Torsion Reduction: The author proves that the existence of an NCCR via conic modules depends only on the combinatorics of the secondary fan and the zonotope, effectively reducing the problem to the case where the class group has no torsion. This significantly simplifies computations by ignoring torsion elements in the class group.
Characterization of NCCRs:
- Theorem 3.6: An incomplete sum of conic modules gives an NCR if and only if the set of indices is lockable. It gives an NCCR if and only if the set is incredulous.
- Theorem 4.22: The existence of an incredulous set of conic modules is equivalent to the existence of an incredulous set of lattice points in the zonotope.
Algorithmic Approach: The paper provides a method to compute the necessary complexes using valid paths in the secondary fan, replacing the complex task of analyzing cell decompositions in $M_{\mathbb{R}}$ with linear algebra on the zonotope.

4. Main Results

General Classification

The paper answers the question of when an incomplete sum yields an NCCR by classifying almost simplicial Gorenstein cones (cones where the number of rays is $\dim \sigma + 1$ ).

Theorem 5.10: An almost simplicial Gorenstein cone $\sigma$ admits an NCCR via an incomplete sum of conic modules if and only if the collection of vectors $\{\beta_\rho\}$ associated with the rays satisfies one of the following:

The set is (up to sign flip) $\{2, 1, -1, -1, -1\}$ .
The set is $\{1, 1, 1, -1, -1, -1\}$ .
The cone is 3-dimensional and lattice-equivalent to a cone over a trapezoid (specifically, the collection is $\{1, 1, -1, -1\}$ with specific geometric constraints).

Counterexamples

The paper demonstrates that not every Gorenstein cone admits an NCCR via conic modules.

Example 4.9: A specific Gorenstein cone in $\mathbb{R}^4$ is shown to have no incredulous set, proving that NCCRs via conic modules do not exist for all Gorenstein toric varieties. This provides a partial negative answer to the general conjecture in the specific context of conic module constructions.

Specific Cases

Proposition 5.19: For 3D cones over quadrilaterals, an NCCR exists if and only if the quadrilateral is a trapezoid.
Propositions 5.21 & 5.22: Explicitly verify the existence or non-existence of incredulous sets for various configurations of $\beta_\rho$ vectors (e.g., $\{2, 1, -1, -1, -1\}$ works; $\{2, 1, 1, -2, -2\}$ fails).

5. Significance

Computational Efficiency: By shifting the focus to the secondary fan and valid paths, the paper reduces the complexity of checking for NCCRs. It avoids the need to explicitly construct and analyze the high-dimensional cell decompositions of chambers of constancy.
Clarification of Minimality: The work formalizes the intuition that NCCRs are "minimal" resolutions. It shows that one does not need the full Bondal-Thomsen collection (which corresponds to a complete sum of conic modules) to get a crepant resolution; specific subsets (incredulous sets) suffice.
Limitations of Conic Modules: The paper establishes that while conic modules are a powerful tool, they do not provide a universal solution for the NCCR conjecture. The existence of cones with no incredulous sets suggests that if the general conjecture (Conjecture 1.1) is true, the NCCR for those cases must be constructed via methods other than direct sums of conic modules (e.g., tilting bundles on different stacks or non-conic modules).
Geometric Insight: The connection between the geometry of the zonotope (specifically the existence of valid paths) and the homological properties of the endomorphism algebra provides a deep geometric interpretation of the algebraic conditions for crepancy.

In summary, Malter's paper provides a robust combinatorial framework for analyzing NCCRs of toric algebras, successfully classifying the almost simplicial case and demonstrating the precise limitations of the conic module approach.