🔢 mathematics

A pre-triangulated category which is not triangulated

This paper constructs an explicit example of a pre-triangulated category that fails to be triangulated, utilizing the category of finitely generated projective modules over the type- $A_5$ preprojective algebra defined over $\mathbb{F}_2$ with a suspension induced by a graph-reflection automorphism.

Original authors: Xiao-Wu Chen, Jian Liu, Xue-Song Lu, Chencheng Zhang

Published 2026-08-11

📖 6 min read🧠 Deep dive

CC0 1.0

Original authors: Xiao-Wu Chen, Jian Liu, Xue-Song Lu, Chencheng Zhang

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 a master architect designing a new kind of city. In this city, the buildings are mathematical objects, and the roads connecting them are rules for how they interact. For decades, mathematicians have been building a specific type of city called a "triangulated category." It's a place where everything follows a very strict, beautiful set of laws. One of the most important laws is the "Octahedral Axiom." Think of this as a rule about how three different roads can meet at a single intersection to form a perfect, stable pyramid shape. If you have two roads meeting, the third one must fit in a very specific way to keep the whole structure from collapsing.

For a long time, mathematicians wondered: Is this "pyramid rule" just a natural consequence of the other, simpler laws? Or is it a separate, independent rule that you have to add by hand? In other words, if you build a city that follows all the basic laws of geometry, does the pyramid rule happen automatically, or could you accidentally build a city that looks perfect but has a hidden, unstable corner? This question has been a mystery for years. The paper you are about to read dives into this mystery, not by building a whole new city from scratch, but by constructing a very specific, tiny, and tricky model to see if the pyramid rule holds up.

The Story of the Broken Pyramid

The authors of this paper, Xiao-Wu Chen and his team, decided to test the limits of these mathematical rules. They wanted to see if they could build a "pre-triangulated" category—a structure that follows the first three basic laws of the city—but fails the fourth, most complex law (the Octahedral Axiom). If they could do this, it would prove that the pyramid rule is not automatic; it's a special ingredient you have to add separately.

To build their model, they used a very specific type of mathematical "Lego set" called a preprojective algebra of type A5. Imagine a chain of five nodes (like five houses in a row) connected by two-way streets. The rules for how you can travel between these houses are defined by a field called F2, which is a world where numbers only exist as 0 and 1, and adding 1 to 1 gives you 0. It's a binary, on-off universe.

In this binary world, the authors looked at a special collection of "projective modules." Think of these as the strongest, most indestructible building blocks in their city. They also introduced a "suspension" functor, which is like a magical machine that rotates the entire city or shifts every building to a new position. In their specific setup, this machine is driven by a "graph-reflection," which is like flipping the whole chain of houses upside down (House 1 becomes House 5, House 2 becomes House 4, and so on).

The team then created a "twisted" version of the rules. Usually, when you rotate the city, the buildings line up perfectly with their old positions after a few turns. But the authors found a special building block (let's call it Module M) that behaves strangely. When they applied their rotation machine to it, it didn't just line up; it got a little "twist" or a "glitch" attached to it. They used this glitch to create a new set of rules for their city, which they call $\Delta_\epsilon$ .

This new city follows the first three laws perfectly. If you take any two roads and try to connect them, you can always find a third road to complete the triangle. The buildings are stable, and the rotations work as expected. It looks like a perfect "pre-triangulated" city.

But then, they tried to build the "four-by-four" pyramid. This is a test where you take two rows of buildings and two columns of buildings and try to fill in the rest of the grid so that every single row and column forms a perfect triangle. In a normal, fully "triangulated" city, this is always possible. You can always find the missing pieces to complete the puzzle.

However, in the authors' twisted city, the puzzle breaks. They showed that no matter how they tried to fill in the missing pieces, the final row of buildings would always be "wrong." It would be a row that fits the first three laws but doesn't match the specific "twisted" rule they created. It's like trying to force a square peg into a round hole, but the hole is made of rubber that stretches just enough to look like it fits, until you try to put the lid on, and then it snaps back.

The key to this failure was a specific mathematical object they constructed, which they called Module M, and its relationship with two other modules, A and B. They proved that if you try to complete the grid, the third vertical arrow (the missing piece) is forced to be a specific type of connection. But because of the "twist" they introduced earlier, this connection creates a contradiction. The math forces the connection to be both "twisted" and "untwisted" at the same time, which is impossible.

The Verdict

The paper proves, with absolute mathematical certainty, that the answer to the big question is no. A structure can follow the first three laws of a triangulated category and still fail the fourth. The "Octahedral Axiom" is not a free bonus; it is a separate, independent rule.

The authors didn't just guess this; they built an explicit, concrete example using the preprojective algebra of type A5 over the field with two elements ( $F_2$ ). They showed that in this specific setup, the "four-by-four" property fails. This means that the structure they built is "pre-triangulated" but not "triangulated."

This is a big deal because it settles a long-standing conjecture (Conjecture 1.1) proposed by Beligiannis, which was attributed to other famous mathematicians Keller and Neeman. For a long time, people wondered if the fourth axiom was just a hidden consequence of the first three. This paper says: "No, it's not. You have to state it explicitly, or your mathematical city might have a hidden crack."

The authors even used a bit of help from an AI system called Eureka to help search for the right mathematical building blocks, showing that even in the most abstract corners of math, new tools are helping to solve old puzzles. But the final proof is a rigorous, human-written argument that leaves no room for doubt: the pyramid rule is a rule you must add, not one that appears by magic.

Technical Summary: A Pre-Triangulated Category Which Is Not Triangulated

Problem Statement
The paper addresses a fundamental question in homological algebra regarding the relationship between pre-triangulated and triangulated categories. While Verdier's axioms (TR1–TR4) define triangulated categories, the fourth axiom (TR4), known as the octahedral axiom, governs the interaction between morphism compositions and mapping cones. If only axioms TR1–TR3 are satisfied, the structure is termed pre-triangulated (or Puppe-triangulated).

For decades, it remained an open question whether every pre-triangulated category is necessarily triangulated. This question was formalized by Beligiannis (attributed to Keller and Neeman) as Conjecture 1.1, which posits the existence of an abelian Frobenius category $B$ with a $\Sigma$ -stable structure such that the induced pre-triangulated category of projective objects is not triangulated. The paper aims to construct an explicit counterexample to this conjecture, thereby demonstrating that TR4 is not a formal consequence of TR1–TR3.

Methodology
The authors construct a counterexample using the representation theory of preprojective algebras over the field $\mathbb{F}_2$ . The methodology proceeds through the following steps:

Algebraic Setup: The study focuses on the preprojective algebra $\Lambda = \Pi(A_5)$ of type $A_5$ over $\mathbb{F}_2$ . This algebra is self-injective, making its category of finitely generated modules, $\text{mod}\,\Lambda$ , an abelian Frobenius category. The stable category $\underline{\text{mod}}\,\Lambda$ is naturally triangulated.
Stable Structures and Heller's Construction: The authors utilize Heller's theorem, which establishes a correspondence between $\Sigma$ -stable structures on a Frobenius category and pre-triangulated structures on its subcategory of projective objects $P(\Lambda)$ . A $\Sigma$ -stable structure is defined by an auto-equivalence $\Sigma$ and a natural isomorphism $\theta: \Sigma \xrightarrow{\sim} \Omega^{-3}$ satisfying specific compatibility conditions.
Twisting the Comparison:
- The standard triangulated structure on $P(\Lambda)$ arises from the graph-reflection automorphism $\nu$ (where $\Sigma = (-)^\nu$ ) and a canonical isomorphism $\theta_0$ .
- The authors identify a specific indecomposable module $M$ with dimension vector $(2, 2, 2, 1, 0)$ such that $\Omega^2(M) \cong M$ .
- Using Auslander–Reiten theory, they construct a non-trivial natural transformation $\Delta$ supported on the orbit $\{M, \Omega(M)\}$ , satisfying $\Delta^2 = 0$ and commuting with the suspension.
- This transformation is used to "twist" the canonical comparison $\theta_0$ , creating a new isomorphism $\theta_\epsilon = \epsilon \circ \theta_0$ , where $\epsilon = \text{Id} + \Delta$ . This yields a new pre-triangulated structure $(P(\Lambda), \Sigma, \triangle_\epsilon)$ satisfying TR1–TR3.
Obstruction Analysis (The Four-by-Four Property): To prove that TR4 fails, the authors demonstrate that the structure $\triangle_\epsilon$ lacks the "four-by-four property," a necessary condition for triangulated categories derived from the octahedral axiom. They construct a diagram involving two distinguished rows ( $T_A, T_B$ ) and two distinguished columns derived from Auslander–Reiten sequences involving modules $A$ and $B$ .
Rigidity Argument: The core of the proof involves analyzing the possible completions of this diagram. The authors show that any valid third vertical arrow in the diagram is rigidly constrained by the automorphisms of the second row. Specifically, the "twisted" nature of the comparison $\theta_\epsilon$ on the module $M$ (which appears as the cokernel in the third row) creates a contradiction: the required third row must simultaneously satisfy the twisted comparison (due to the structure of the columns) and the untwisted comparison (due to the rigidity of the diagram's completion).

Key Contributions and Results

Construction of a Counterexample: The paper explicitly constructs a pre-triangulated category $(P(\Pi(A_5)), \Sigma, \triangle_\epsilon)$ that satisfies TR1–TR3 but fails TR4.
Disproof of Conjecture 1.1: The result confirms Beligiannis's conjecture (attributed to Keller and Neeman) that there exists a $\Sigma$ -stable structure on an abelian Frobenius category inducing a pre-triangulated category that is not triangulated.
Failure of the Octahedral Axiom: The paper proves that the octahedral axiom (TR4) is independent of the first three axioms in the context of pre-triangulated categories derived from Heller's construction.
Specific Algebraic Counterexample: The counterexample is derived from the type- $A_5$ preprojective algebra over $\mathbb{F}_2$ , utilizing the specific properties of the module $M$ and the graph-reflection automorphism.

Significance
The paper establishes that the class of pre-triangulated categories is strictly larger than the class of triangulated categories. This clarifies the logical independence of the octahedral axiom (TR4) from the other axioms defining the triangulated structure. The result resolves a long-standing open question in the theory of triangulated categories, showing that the "formal" consequences of TR1–TR3 do not automatically guarantee the existence of the octahedral axiom. The authors note that their construction relies on the specific arithmetic of $\mathbb{F}_2$ and the specific representation theory of the $A_5$ preprojective algebra, providing a concrete, computable instance where the distinction between pre-triangulated and triangulated structures manifests.

The Story of the Broken Pyramid

The Verdict

Technical Summary: A Pre-Triangulated Category Which Is Not Triangulated

More like this