🔢 mathematics

Minimal resolutions of toric substacks by line bundles

This paper constructs minimal resolutions of pushforwards of structure sheaves of toric substacks by line bundles as strong deformation retracts of cellular resolutions, utilizing the homological perturbation lemma and Moore-Penrose inverses to provide a canonical, combinatorial description of their differentials.

Original authors: Zengrui Han

Published 2026-04-17

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Zengrui Han

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 trying to understand a complex, multi-layered sculpture made of glass and wire. This sculpture represents a mathematical object called a Toric Substack (a specific shape living inside a larger, smooth mathematical universe).

To understand this sculpture, mathematicians usually build a "scaffold" around it. This scaffold is a resolution: a chain of simpler building blocks (line bundles) that, when put together, perfectly mimic the shape of the sculpture.

For a long time, mathematicians had a way to build this scaffold, but it was bloated. It had too many extra struts, redundant wires, and unnecessary layers. It worked, but it was messy and hard to read. It was like trying to describe a simple house by listing every single brick, nail, and speck of dust, even the ones hidden inside the walls.

The Problem:
How do you strip away all the junk to find the minimal scaffold? The one with the absolute fewest pieces necessary to describe the shape? And even harder: How do you write down the exact instructions (the "differentials") for how these pieces connect, without just guessing?

The Solution (The Paper's Big Idea):
Zengrui Han, the author, has built a new, super-efficient machine to strip away the junk and find the perfect, minimal scaffold. He does this using two main tools:

1. The "Map" (Bondal Stratification)

First, imagine the space around your sculpture is covered in a giant, repeating grid of lines (like a honeycomb or a city map). This is the Bondal Stratification.

Every cell in this grid corresponds to a specific building block.
The original "bloated" scaffold (called the HHL resolution) connects these blocks based on how they sit on this map.

2. The "Scissors" (Homological Perturbation Lemma)

Han uses a powerful mathematical tool called the Homological Perturbation Lemma. Think of this as a pair of magical scissors that can cut away the redundant parts of a structure while keeping the shape intact.

Usually, to use these scissors, you need to make arbitrary choices (like deciding which wire to cut first). If you choose differently, you might get a different result.
Han's innovation is finding a way to use the scissors without making any arbitrary choices.

3. The "GPS" (Moore-Penrose Inverses)

This is the secret sauce. To cut the wires perfectly, Han uses something called the Moore-Penrose Inverse.

The Analogy: Imagine you have a tangled knot of strings. You want to pull one end to straighten it out. Sometimes, pulling one way tightens the knot; pulling another way loosens it.
The Moore-Penrose Inverse is like a perfect GPS for the knot. It doesn't just guess which way to pull; it calculates the exact mathematical "average" of all possible ways to untangle it. It gives a single, unique, and "canonical" answer.
Because this method is so precise and requires no guessing, the resulting minimal scaffold is canonical. It's the only true minimal version, determined entirely by the shape of the map itself.

How the New Scaffold is Built

Han's method works like a relay race:

Type I Runners: These are the standard connections between different parts of the map (like moving from a big district to a small neighborhood).
Type II Runners: These are connections within the same district. Here, the "GPS" (Moore-Penrose) takes over to figure out the most efficient path.
The Path: To get from point A to point B in the minimal scaffold, you might have to zig-zag: take a Type I step, then a Type II step, then another Type I step.
The Weight: Han calculates a "weight" (a number) for every possible path. He adds them all up. The final connection in the minimal scaffold is the sum of all these weighted paths.

The Result

The paper provides a combinatorial recipe (a step-by-step instruction manual using only counting and geometry) to build this minimal scaffold.

Before: You had a messy, non-minimal scaffold and had to guess how to clean it up.
Now: You have a clean, minimal scaffold with a clear, unique set of instructions on how every piece connects.

Why Does This Matter?

In the world of algebra and geometry, "minimal" is beautiful. It reveals the true essence of the object without the noise.

For Computer Scientists: It means algorithms can run faster because they don't have to process redundant data.
For Physicists: In theories like "Mirror Symmetry" (which relates different shapes in physics), having a clean, minimal description helps them see the underlying physics more clearly.
For Mathematicians: It solves a decades-old problem of how to systematically find these minimal structures without relying on "magic" or mirror symmetry tricks.

In a nutshell: Han took a messy, over-engineered mathematical blueprint, used a "perfect GPS" to untangle the knots, and produced the cleanest, most efficient, and uniquely defined blueprint possible. It's like turning a pile of scrap metal into a sleek, aerodynamic car.

1. Problem Statement

The paper addresses the problem of constructing minimal free resolutions for the pushforward of structure sheaves of toric substacks ( $i_* \mathcal{O}_Y$ ) embedded in smooth toric stacks ( $X$ ).

Context: While resolutions of diagonals and structure sheaves are well-studied in toric geometry, existing constructions often yield non-minimal resolutions with redundant summands.
The HHL Resolution: Hanlon, Hicks, and Lazarev (HHL) previously constructed a canonical cellular resolution using the Bondal-Thomsen collection of line bundles. However, this resolution is generally not minimal; it contains many terms that cancel out in the minimal version.
The Gap: While algorithmic methods (e.g., discrete Morse theory) exist to reduce non-minimal resolutions, they often lack a canonical, purely combinatorial description of the resulting differentials. Previous attempts to find Betti numbers relied on homological mirror symmetry, lacking a direct algebraic proof.
Goal: To provide a canonical, combinatorial construction of the minimal resolution and an explicit formula for its differentials without relying on mirror symmetry.

2. Methodology

The author employs a combination of homological algebra and combinatorial topology, specifically utilizing the Homological Perturbation Lemma (HPL) and Moore-Penrose inverses.

A. The Setup

Bondal Stratification: The paper utilizes the Bondal stratification $S$ on a real torus $(S^1)^k$ , derived from the embedding of the substack. This stratification partitions the torus into cells $\sigma$ .
Bondal-Thomsen Collection: Each cell $\sigma$ is associated with a line bundle $\mathcal{O}(\sigma)$ on $X$ . These bundles form a poset (partially ordered set) based on the existence of non-zero morphisms between them.
HHL Complex: The starting point is the HHL complex $(C^\bullet, d)$ , a cellular resolution where differentials are defined by weights $HHL_{\sigma, \tau}$ associated with adjacent cells.

B. The Core Strategy: Homological Perturbation

The construction treats the minimal resolution as a deformation of the associated graded complex of the HHL resolution.

Filtration: The HHL resolution is filtered by the Bondal-Thomsen poset.
Associated Graded Complex: The associated graded complex $(gr_F C^\bullet, gr_F d)$ is obtained by setting "non-invertible" terms (those strictly decreasing in the poset) to zero. This complex decomposes into a direct sum of cellular chain complexes $C^{BM}_\bullet(S[a])$ for each isomorphism class $[a]$ of line bundles.
Strong Deformation Retract (SDR): The author constructs an SDR datum between the homology of the graded complex (which has trivial differentials) and the graded complex itself.
- Key Innovation: Instead of choosing an arbitrary homotopy map (which would make the result non-canonical), the author uses the Moore-Penrose inverse of the differentials in the cellular chain complexes. This choice is canonical and depends only on the stratification.
Perturbation: The difference between the original HHL differential $d$ and the graded differential $gr_F d$ is treated as a perturbation $\delta$ .
Application of HPL: The Homological Perturbation Lemma is applied to transfer the SDR structure from the graded complex to the full HHL complex. This yields a new differential $d_{min}$ on the homology complex, which constitutes the minimal resolution.

3. Key Contributions

A. Canonical Construction of Minimal Resolutions

The paper provides a constructive proof that the complex $(H^\bullet, d_{min})$ is the minimal resolution of $i_* \mathcal{O}_Y$ . Unlike previous methods that might require arbitrary choices (like Morse matchings), this construction is canonical because the homotopy map is defined via the Moore-Penrose inverse.

B. Combinatorial Description of Differentials

The most significant contribution is the explicit combinatorial formula for the differentials $d_{min}$ . The author decomposes the morphisms between strata into two types:

Type I Pairs: Cells $(\sigma, \tau)$ where $\mathcal{O}(\sigma) > \mathcal{O}(\tau)$ . The weight is the standard HHL weight ( $HHL_{\sigma, \tau}$ ).
Type II Pairs: Cells $(\sigma, \tau)$ where $\mathcal{O}(\sigma) = \mathcal{O}(\tau)$ . The weight is the Moore-Penrose (MP) weight, derived from the Moore-Penrose inverse of the cellular boundary maps.

The differential $\Sigma_{\sigma, \tau}$ is defined as a sum over all zig-zag paths $\lambda$ from $\sigma$ to $\tau$ :
$\Sigma_{\sigma, \tau} = \sum_{\lambda \in P(\sigma, \tau)} (-1)^{l(\lambda)} \prod_{\text{edges in } \lambda} \text{weight}$
where the path alternates between Type I (blue) and Type II (red) pairs. The MP weights are computed using a "Hedge Formula" involving determinants of submatrices of the boundary maps.

C. Virtual Resolutions in Commutative Algebra

The paper interprets the result in the context of the Cox ring $S$ of the toric variety. It shows that the minimal resolution corresponds to a minimal virtual resolution of the graded $S$ -module $S/I_Y$ (where $I_Y$ is the defining ideal of the subvariety). This provides a purely algebraic proof of the Betti number formula previously derived by Favero and Sapronov using mirror symmetry.

4. Results

Theorem 1.1: The complex constructed via the homological perturbation lemma (using Moore-Penrose inverses) is the minimal resolution of the pushforward of the structure sheaf.
Theorem 1.2: In the language of the Cox ring, this complex is a minimal virtual resolution of the defining ideal of the toric subvariety.
Explicit Example: The paper computes the minimal resolution for a specific embedding in the weighted projective plane $\mathbb{P}(3,1,1)$ . It demonstrates how to calculate the MP weights and sum the path contributions to obtain explicit polynomial entries in the differential matrices.
Comparison with Morse Theory: The example illustrates that while a perfect Morse matching can yield a simpler (non-canonical) minimal resolution, the Moore-Penrose approach yields a unique, canonical result determined solely by the geometry of the stratification.

5. Significance

Independence from Mirror Symmetry: The proof is entirely algebraic and combinatorial, removing the need for homological mirror symmetry to establish the Betti numbers or the existence of the minimal resolution.
Canonical Nature: By utilizing Moore-Penrose inverses, the construction avoids arbitrary choices (like basis selection or Morse matchings), ensuring the resulting resolution is intrinsic to the toric data.
Computational Utility: The explicit combinatorial formulas for differentials allow for the algorithmic computation of minimal resolutions for toric substacks, which was previously difficult due to the complexity of reducing the HHL resolution.
Bridging Fields: The work connects toric geometry, commutative algebra (virtual resolutions), and linear algebra (Moore-Penrose inverses), offering a new toolkit for studying syzygies in toric settings.

In summary, Zengrui Han successfully solves the problem of constructing minimal resolutions for toric substacks by introducing a canonical method based on the homological perturbation lemma and Moore-Penrose inverses, providing both a theoretical existence proof and a practical combinatorial algorithm for computing the differentials.