🔢 mathematics

A smooth projective counterexample to Bondal-Polishchuk's conjecture

This paper presents a counterexample to the 1993 Bondal-Polishchuk conjecture by demonstrating that the braid group action on full exceptional collections is not transitive for a specific smooth projective weak Fano threefold, thereby providing the first such counterexample within the derived category of a smooth projective variety.

Original authors: Anya Nordskova

Published 2026-07-29

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: Anya Nordskova

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 standing in a vast, magical library where every book represents a different way of looking at a geometric shape. In this library, mathematicians have discovered that you can rearrange the pages of these books in specific, rhythmic patterns called "braid moves." For decades, a famous rule of thumb suggested that no matter how you started arranging the pages, you could always reach any other arrangement just by performing enough of these braid moves. It was like saying that if you have a deck of cards, you can shuffle them into any order you want using a specific set of moves. This idea, proposed by two mathematicians named Bondal and Polishchuk, became a guiding star for researchers exploring the hidden structures of shapes in higher dimensions. They believed that the "braid group" (the set of all possible moves) was powerful enough to connect every single possible arrangement of these mathematical books.

But what if that rule isn't true? What if there are two arrangements that are so fundamentally different that no amount of shuffling can ever turn one into the other? This is the question that Anya Nordskova tackles in a new paper. She doesn't just guess; she builds a specific, intricate mathematical shape—a "smooth projective threefold"—and proves that for this shape, the old rule breaks. She shows that there are two distinct "islands" of arrangements that the braid moves cannot bridge. This isn't just a small correction; it's a counterexample that shatters a long-held belief, proving that the landscape of these mathematical shapes is more complex and fragmented than anyone previously thought.

The Great Shuffle That Didn't Work

To understand the paper's discovery, let's picture the mathematical world as a giant, multi-dimensional playground. In this playground, mathematicians study "derived categories," which are like super-complex instruction manuals for shapes. Inside these manuals, there are special lists of objects called "full exceptional collections." Think of these lists as the perfect, complete set of building blocks needed to reconstruct the entire shape.

For a long time, mathematicians wondered: If you have two different perfect lists of building blocks for the same shape, can you always transform one list into the other by swapping pieces around? The swapping process is governed by something called the "braid group," which acts like a set of magical dance moves. Bondal and Polishchuk conjectured in 1993 that these dance moves were so versatile that you could dance your way from any starting list to any ending list. They thought the dance floor was one big, connected room.

The Counterexample: A Shape That Breaks the Dance

In this new paper, the author, Anya Nordskova, says, "Not so fast." She constructs a very specific, smooth, three-dimensional shape (a "weak Fano threefold") that acts like a trap for these dance moves.

Here is how she builds her trap:

The Setup: She starts with a standard 3D space (like the world we live in, but mathematically perfect) and draws a special curved line on it—a "rational sextic curve." She also picks two specific lines in this space that cross the curve in exactly four places.
The Twist: She performs a mathematical operation called "blowing up" along that curved line. Imagine taking a piece of paper and inflating the line into a tube. This creates a new, slightly more complex 3D shape called $X$ .
The Two Lists: On this new shape $X$ $X$ , she identifies two different lists of building blocks (exceptional collections). Let's call them List A and List B.
- List A is the "standard" list, built from the original space and the curve.
- List B is created by taking List A and applying a special "spherical twist" (a specific type of mathematical surgery) to one of the pieces.

The Proof: The Mirror That Doesn't Match

To prove that List A and List B are truly different and cannot be turned into each other by the braid moves, the author uses a clever trick involving a "mirror" (an automorphism).

She finds a symmetry in her shape $X$ —a way to flip the shape over (an involution) that leaves the shape looking exactly the same.

The Test: She checks what happens to List A when she looks at it in this mirror. The mirror reflects List A perfectly; every piece maps to itself.
The Trap: She then checks what happens to List B in the mirror. Because of the way she constructed List B, the mirror doesn't reflect it back to itself. Instead, the mirror swaps the pieces of List B in a way that makes it look different from the original.

Here is the crucial logic: If List A and List B were connected by the braid moves (the dance), then the mirror would have to treat them the same way. If you can dance from A to B, and the mirror leaves A alone, the mirror should also leave B alone. But the mirror doesn't leave B alone. Therefore, it is mathematically impossible to dance from A to B.

The Verdict

The paper concludes with a definitive "No." The author proves that for this specific shape $X$ , the braid group action is not transitive. This means there are at least two different "orbits" of lists that can never meet.

This finding is a direct counterexample to the 1993 conjecture by Bondal and Polishchuk. While previous mathematicians had found similar breaks in the rules for abstract algebraic systems or "Fukaya categories" (which are related to physics and symplectic geometry), this is the first time a counterexample has been found for a "smooth projective variety"—a classic, well-behaved geometric shape.

The author is very sure of this result. She doesn't just suggest it; she constructs the shape, defines the lists, and uses rigorous algebraic proofs to show that the symmetry argument holds up. Notably, the paper explicitly acknowledges that OpenAI's ChatGPT 5.6 played a significant role in this discovery. The AI helped by finding numerous mistakes in earlier attempted constructions, leading to the successful modifications that made the counterexample work, and even located the specific reference containing the threefold used in the proof. It's a solid, proven break in the pattern, showing that the mathematical universe of these shapes has hidden corners that the old rules simply couldn't reach.

Technical Summary: A Smooth Projective Counterexample to Bondal-Polishchuk's Conjecture

Problem Statement
The paper addresses a conjecture proposed by Bondal and Polishchuk in 1993 regarding the structure of the derived category $D^b(X)$ of a smooth projective variety $X$ . Specifically, the conjecture posits that the action of the braid group $B_n$ (generated by mutations) on the set of full exceptional collections of length $n$ is transitive, up to independent shifts of the objects. While this transitivity was previously established for del Pezzo surfaces, the second Hirzebruch surface, and certain Fano threefolds, it remained an open question whether a counterexample could exist within the specific context of $D^b(X)$ for a smooth projective variety. Previous counterexamples by Chang, Haiden, and Schroll were constructed in the realm of partially wrapped Fukaya categories (or derived categories of graded gentle algebras), but no counterexample of the form $D^b(X)$ was known.

Methodology and Strategy
The author employs a strategy suggested by Chang, Haiden, and Schroll, which relies on distinguishing braid group orbits via the action of a fixed autoequivalence. The core logical framework is established in Lemma 2.1:

Let $X$ be a smooth projective variety and $\mathcal{E} = (E_1, \dots, E_n)$ a full exceptional collection.
Suppose there exists an automorphism $f \in \text{Aut}(X)$ that fixes every object $E_i$ in the collection (up to isomorphism).
Suppose there exists a spherical object $S \in D^b(X)$ such that the spherical twist $T_S$ applied to the collection yields a new collection $T_S(\mathcal{E})$ .
If $f^*(T_S(\mathcal{E}))$ is not isomorphic to a shift of $T_S(\mathcal{E})$ , then $\mathcal{E}$ and $T_S(\mathcal{E})$ cannot lie in the same braid group orbit.

The proof proceeds by constructing a specific variety $X$ , an exceptional collection $\mathcal{E}$ , an automorphism $f$ , and a spherical object $S$ that satisfy these conditions.

Construction
The counterexample is constructed through the following geometric steps:

Base Surface: The author begins with six points $p_1, \dots, p_6$ in general position on $\mathbb{P}^2$ . Blowing up these points yields a del Pezzo surface $S$ of degree 3, which embeds into $\mathbb{P}^3$ as a smooth cubic surface.
Involution: An involution $\phi \in \text{PGL}(3, \mathbb{C})$ is defined on $\mathbb{P}^2$ by $[x:y:z] \mapsto [x:y:-z]$ . This involution preserves the set of six points, the conic $C = \{x^2+y^2+z^2=0\}$ , and swaps two specific lines $\ell_1$ and $\ell_2$ in the embedding of $S$ .
The Variety $X$ : The variety $X$ is defined as the blow-up of $\mathbb{P}^3$ along the strict transform of the conic $C$ (denoted $C \subset S \subset \mathbb{P}^3$ ). $X$ is a smooth projective threefold. It is noted that $X$ is weak Fano but not Fano.
The Exceptional Collection $\mathcal{E}$ : Using Orlov's blow-up formula, a 6-term full exceptional collection $\mathcal{E}$ is constructed on $X$ . This collection is formed by combining the standard Beilinson collection on $\mathbb{P}^3$ and the collection on the curve $C \cong \mathbb{P}^1$ .
The Spherical Objects: The strict transforms of the lines $\ell_1$ and $\ell_2$ in $X$ , denoted $\tilde{\ell}_1$ and $\tilde{\ell}_2$ , are shown to be smooth rational curves with normal bundle $\mathcal{O}_{\mathbb{P}^1}(-1) \oplus \mathcal{O}_{\mathbb{P}^1}(-1)$ . Consequently, the objects $S_j = \mathcal{O}_{\tilde{\ell}_j}(-1)$ are 3-spherical objects in $D^b(X)$ .

Key Results

Invariance of $\mathcal{E}$ : The lift of the involution $\phi$ to $X$ , denoted $\phi_X$ , fixes every object in the exceptional collection $\mathcal{E}$ term-wise (up to isomorphism).
Non-invariance of the Twisted Collection: The author demonstrates that $\phi_X$ swaps the spherical objects $S_1$ and $S_2$ (i.e., $\phi_X^*(S_1) \cong S_2$ ). By analyzing the cohomology sheaves of the spherical twist $T_{S_1}$ applied to the object $\mathcal{O}_X(H)$ , it is shown that $T_{S_1}(\mathcal{O}_X(H))$ is not isomorphic to a shift of $T_{S_2}(\mathcal{O}_X(H))$ .
Non-transitivity: Applying the strategy from Lemma 2.1, the paper concludes that the collection $\mathcal{E}$ and the twisted collection $T_{S_1}(\mathcal{E})$ are not in the same orbit under the action of the braid group $B_6 \ltimes \mathbb{Z}^6$ .

Significance and Claims
The paper provides the first known counterexample to the Bondal-Polishchuk conjecture within the category of derived categories of smooth projective varieties. By answering the question posed by Chang, Haiden, and Schroll affirmatively, the work demonstrates that the transitivity of the braid group action on full exceptional collections does not hold universally for smooth projective varieties. The author explicitly notes that the construction relies on identifying a specific weak Fano threefold where the action of a fixed autoequivalence distinguishes between braid orbits, thereby refuting the conjecture in the geometric setting.

The Great Shuffle That Didn't Work

The Counterexample: A Shape That Breaks the Dance

The Proof: The Mirror That Doesn't Match

The Verdict

More like this