🔢 mathematics

Stability conditions on threefolds

This paper investigates stability conditions on $\mathbb{P}^n$ satisfying the Li condition, demonstrating that their restrictions to a smooth projective threefold coincide with those from the Bayer-Macrì-Toda double-tilt construction, thereby proving the weak BMT conjecture.

Original authors: Yiran Cheng, Soheyla Feyzbakhsh

Published 2026-07-07

📖 5 min read🧠 Deep dive

CC0 1.0

Original authors: Yiran Cheng, Soheyla Feyzbakhsh

Original paper dedicated to the public domain under CC0 1.0 (http://creativecommons.org/publicdomain/zero/1.0/). ✨ 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 the shape and structure of a complex, multi-dimensional object (like a 3D sculpture) by looking at it through different lenses. In the world of advanced mathematics, specifically algebraic geometry, these "lenses" are called stability conditions. They help mathematicians sort and categorize abstract shapes (called "objects" in a derived category) to see which ones are "stable" and which ones are "unstable" or likely to fall apart.

This paper, by Yiran Cheng and Soheyla Feyzbakhsh, is about building a very specific, reliable set of these lenses for a particular type of 3D shape called a threefold (a three-dimensional surface embedded in a larger space).

Here is the story of what they did, explained through simple analogies:

1. The Starting Point: A Perfectly Ordered Grid

The authors start with a known, well-behaved mathematical space called Projective Space ( $P^n$ ). Think of this as a giant, perfectly organized grid or a high-dimensional chessboard.

The Problem: We know how to create "stability lenses" for this grid. But what if we want to study a smaller, more complex shape (like a curved sculpture) that is sitting inside this grid?
The Challenge: If you just take the lens designed for the grid and point it at the sculpture, the view might get distorted. The "stable" objects in the grid might look "unstable" on the sculpture, or vice versa.

2. The "Li Condition": The Safety Filter

The paper focuses on a specific rule called the "Li condition."

The Analogy: Imagine you have a filter designed to sort marbles by size on a flat table. You want to use this same filter to sort marbles on a curved ramp. The "Li condition" is a guarantee that says: "If the ramp isn't too steep and the filter is tuned correctly, the sorting will still work perfectly on the ramp."
The Result: The authors prove that if you tune their specific lens (by adjusting a parameter they call ' $a$ ' to be large enough), it satisfies this safety condition. This means the lens works perfectly not just on the big grid, but also on any smooth 3D sculpture sitting inside it.

3. The "Double-Tilt" Construction: A Two-Step Sorting Machine

For a long time, mathematicians had a different way of trying to sort these 3D shapes, called the "double-tilt" construction (invented by Bayer, Macrì, and Toda).

The Analogy: Imagine you have a messy pile of toys.
1. Step 1 (First Tilt): You separate them into "heavy" and "light" piles.
2. Step 2 (Second Tilt): You take those piles and sort them again based on "color" and "shape."
  This two-step process was conjectured (guessed) to be the correct way to create a stability lens for 3D shapes, but no one had proven it matched the "Li condition" lenses.
The Breakthrough: The authors prove that for 3D shapes, these two methods are actually the same thing.
- The "Li condition" lens (which they built by restricting the grid lens) turns out to be identical to the "double-tilt" lens.
- Why it matters: It confirms that the "double-tilt" method, which was just a guess before, is actually a valid, rigorous way to organize these mathematical shapes.

4. The "Weak BMT Conjecture": A New Rule for Stability

Once they proved these two methods are the same, they could use their new, confirmed lens to prove a famous inequality called the Weak BMT Conjecture.

The Analogy: Think of this as discovering a new law of physics for these shapes. Before, people guessed that if a shape is stable, it must obey a certain mathematical formula (like a speed limit).
The Discovery: Using their lens, the authors proved that this "speed limit" formula is true. Specifically, they showed that for any stable 3D shape, a specific combination of its "volume," "surface area," and "twist" (mathematical terms for its properties) must satisfy a strict inequality. If it doesn't, the shape isn't stable.

5. A Reversal of Logic

The paper highlights a fascinating shift in how mathematicians think about these problems.

Old Way: "Let's look at the geometry of the shape, figure out the rules, and then hope we can build a lens."
New Way (This Paper): "Let's build the lens first using pure logic (categorical construction). Once the lens exists, the rules (inequalities) naturally fall out as a consequence."
It's like saying, instead of measuring the wind to build a weather vane, you build the weather vane first, and the wind direction becomes obvious.

Summary

In short, Cheng and Feyzbakhsh built a robust mathematical tool (a stability condition) for 3D shapes by borrowing it from a simpler, higher-dimensional space. They proved that this tool is identical to a previously guessed method (the double-tilt), and as a result, they confirmed a major mathematical rule (the weak BMT conjecture) that governs how these 3D shapes can exist without falling apart. They did this by showing that if you tune your "lens" correctly, the complex 3D world behaves just as predictably as the simple grid it sits inside.

Technical Summary: Stability Conditions on Threefolds

Problem Statement
The paper addresses the construction and characterization of Bridgeland stability conditions on smooth projective threefolds $X \subset \mathbb{P}^n$ . Specifically, it investigates the relationship between two distinct approaches to defining these stability conditions:

The "Li Condition" Approach: Stability conditions on $\mathbb{P}^n$ (constructed by Li [Li26] based on Liu's work [Liu21]) that satisfy a specific phase condition, allowing them to descend (restrict) to stability conditions on smooth subvarieties $X$ .
The Double-Tilt Construction: The heart of the stability condition obtained via the double-tilt procedure introduced by Bayer–Macrì–Toda [BMT14], which is conjectured to define a stability condition on $X$ when paired with a specific central charge.

The central problem is to determine whether the stability conditions induced on a threefold $X$ via the restriction of Li's conditions on $\mathbb{P}^n$ coincide with those obtained via the Bayer–Macrì–Toda (BMT) double-tilt construction. Resolving this equivalence allows for the proof of the "weak BMT conjecture" (a Bogomolov-type inequality for tilt-semistable objects) as a consequence of the existence of the stability conditions, rather than as a prerequisite.

Methodology
The authors employ a combination of categorical descent, metric analysis on the stability manifold, and geometric reduction techniques:

Metric Control on Stability Manifolds:
The authors analyze the distance between stability conditions $\sigma_{a,b}$ and $\sigma_{a, b+\Delta b}$ on products of elliptic curves $C^n$ (using Liu's construction [Liu21]). By utilizing Bridgeland's generalized metric and analyzing the derivative of the central charge with respect to the parameter $b$ , they establish an upper bound on the phase shift of objects under tensor products with line bundles. Specifically, they prove that for sufficiently large $a$ , the phase shift $\phi(E \otimes \mathcal{O}(m)) - \phi(E)$ is strictly less than 1.
Descent to Projective Space and Subvarieties:
Using the quotient map $\pi: C^n \to \mathbb{P}^n$ , the authors lift the stability conditions to $\mathbb{P}^n$ . They then apply Li's restriction theorem [Li26], which states that a stability condition on $\mathbb{P}^n$ restricts to a smooth subvariety $X$ if the "Li condition" (a bound on phase shifts under tensoring with the ideal sheaf of $X$ ) is satisfied.
To ensure this condition holds, the authors introduce an invariant $m(X \xrightarrow{\iota} \mathbb{P}^n)$ , derived from the minimal resolution of the ideal sheaf $\mathcal{I}_{X/\mathbb{P}^n}$ by split vector bundles. They show that if the parameter $a$ satisfies $a > \frac{n}{\pi} m(X \xrightarrow{\iota} \mathbb{P}^n)$ , the restriction $\sigma^X_{a,b}$ is a valid Bridgeland stability condition.
Dimensional Reduction and Heart Identification:
For the case where $X$ is a threefold, the authors analyze the heart $\mathcal{A}^X_{a,b}$ of the restricted stability condition. They utilize a hyperplane section $S \subset X$ to reduce the problem to the surface case (where the heart is known to be a single tilt of $\text{Coh}(S)$ ). By comparing the cohomological properties of objects in the restricted heart with the slope functions defined in the BMT construction, they demonstrate that the two hearts coincide.

Key Contributions and Results

Theorem 1.1 (Phase Bound): The authors prove that for $a > \frac{mn}{\pi}$ , any nonzero object $E \in D^b(\mathbb{P}^n)$ satisfies $\phi^-_{a,b}(E \otimes \mathcal{O}_{\mathbb{P}^n}(m)) < \phi^-_{a,b}(E) + 1$ . This establishes the "Li condition" for the family of stability conditions on $\mathbb{P}^n$ .
Theorem 1.2 (Equivalence of Constructions): For a smooth projective threefold $X$ and parameters satisfying $a > a(X, H)$ (where $a(X, H)$ is defined via the invariant $m(X \xrightarrow{\iota} \mathbb{P}^n)$ ), the heart $\mathcal{A}^X_{a,b}$ of the induced stability condition coincides exactly with the double-tilted heart $(\text{Coh}(X)^{Z^{(1)}_b})^{Z^{(2)}_{a,b}}$ defined by Bayer–Macrì–Toda.
Corollary 1.3 (Weak BMT Conjecture): As a direct consequence of the equivalence in Theorem 1.2, the authors prove the weak BMT conjecture for threefolds. Specifically, for any $Z^{(2)}_{a,b}$ -semistable object $E$ in the first tilt satisfying a specific equality of slopes, the inequality $\text{ch}^{bH}_3(E) < \frac{a^2}{2} H^2 \text{ch}^{bH}_1(E)$ holds.
Metric Bounds: The paper provides explicit bounds on the distance between stability conditions on $C^n$ and $\mathbb{P}^n$ in terms of the parameters $a$ and $b$ , refining previous estimates.

Significance and Claims
The paper claims a significant shift in the philosophy of proving Castelnuovo-type inequalities (such as the BMT conjecture). Traditionally, such inequalities were established using the geometry of the underlying variety to deduce the existence of stability conditions. In contrast, this work demonstrates that:

Stability conditions on $\mathbb{P}^n$ can be constructed purely categorically (via descent from products of elliptic curves).
These conditions restrict to smooth subvarieties $X$ under explicit numerical conditions on the parameters.
The existence of these stability conditions implies the BMT-type inequalities as consequences.

The authors emphasize that their result confirms the prediction that the BMT double-tilt construction yields a valid stability condition for sufficiently large $a$ , thereby validating the categorical framework for threefolds without requiring the inequalities to be proven a priori. The work relies on the foundational constructions of Polishchuk [Pol07], Liu [Liu21], and Li [Li26], extending their results to the specific context of threefolds and establishing the precise link between the "Li condition" and the "BMT heart."