🔢 mathematics

Birational properties of word varieties

This paper establishes that word varieties in $SL(2)\times SL(2)$ are closely related to smooth conic bundles over trace surfaces, demonstrating that such varieties can be irrational over non-algebraically closed fields and satisfy weak approximation with the Brauer–Manin obstruction over number fields.

Original authors: Tatiana Bandman, Boris Kunyavskii, Alexei N. Skorobogatov

Published 2026-06-16

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Tatiana Bandman, Boris Kunyavskii, Alexei N. Skorobogatov

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 a master puzzle solver working with a specific set of rules. In this paper, the "puzzle" involves two special 2x2 matrices (think of them as complex, rotating grids of numbers) that we'll call A and B.

The authors are investigating a specific type of equation: $w(A, B) = \alpha$ .

Here, $w$ is a "word," which is just a recipe for mixing A and B together using multiplication and inversion (like a mathematical dance step). $\alpha$ is the target result you want to achieve. The big question the paper asks is: Can we always find a pair of matrices (A, B) to solve this puzzle, and what does the "shape" of all possible solutions look like?

Here is the breakdown of their findings using everyday analogies:

1. The "Trace" Map: Turning 3D Puzzles into 2D Surfaces

Solving for two full matrices is like trying to navigate a complex 3D maze. The authors use a clever trick called the "trace." The trace is just a single number you get by adding up the diagonal numbers of a matrix.

They discovered that if you take the "trace" of your matrices A, B, and their product AB, you can map the entire 3D problem onto a simpler 2D surface (a curved sheet floating in 3D space).

The Analogy: Imagine you have a complicated 3D sculpture. Instead of studying the whole sculpture, you shine a light on it to see its shadow on the wall. The authors found that the "shadow" (the trace surface) tells you almost everything you need to know about the original sculpture.

2. The "Commutator" Puzzle: When Order Matters

The most famous "word" in this field is the commutator: $ABA^{-1}B^{-1}$ . This measures how much the order of operations matters. If you do A then B, is it different from doing B then A?

The Discovery: When the target result ( $\alpha$ ) is a specific type of matrix, the set of all solutions forms a shape that looks like a bundle of conic sections (like a stack of circles or ellipses) wrapped around that 2D "trace surface" (which they call a Markoff surface).
The Twist: Sometimes, this bundle of solutions is "rational" (easy to describe and navigate), and sometimes it is "irrational" (twisted and impossible to flatten out).

3. The Big Surprise: It's Not Always Solvable (in the way we thought)

For a long time, mathematicians wondered if these solution shapes were always "rational" (meaning you could easily list all the solutions using simple formulas).

The Result: The authors proved that no, they are not always rational. If you are working with numbers from a specific field (like the rational numbers, not the complex numbers), the shape of the solutions can be twisted in a way that makes it impossible to describe them simply.
The Metaphor: Imagine a knot. Sometimes you can untie it and lay it flat (rational). Sometimes, no matter how hard you pull, it remains a complex knot (irrational). The authors found a specific condition where the knot cannot be untied. This answers a question that had been open for decades.

4. The "Brauer-Manin Obstruction": The Invisible Wall

When working with number fields (like fractions), the authors looked at whether you can find solutions that are "close" to any set of local clues you might have.

The Finding: They found that the only thing stopping you from finding a solution is something called the Brauer-Manin obstruction.
The Analogy: Imagine you are trying to park a car in a garage. You have a map of the neighborhood (local clues) that says the garage is empty. However, there is an invisible force field (the obstruction) that prevents the car from entering, even though the map looks fine. The authors proved that this invisible force field is the only reason you might fail to park the car. If the force field isn't there, you can always find a spot.

5. Special Cases: The "Tricky" Numbers

The paper also looked at specific, tricky targets:

The Identity Matrix: If you want the result to be the "do nothing" matrix, the solutions form a nice, smooth 4D shape that is easy to understand.
The "Negative Identity": If you want the result to be the "negative identity" matrix, the solution only exists if your number system allows you to write -1 as the sum of two squares (like $1^2 + 2^2$ in some systems). If your number system doesn't allow this, the puzzle has no solution at all.

Summary

In short, this paper takes a very abstract algebraic problem about matrix equations and translates it into geometry. They showed that:

These equations can be visualized as bundles of curves wrapped around a specific surface.
Sometimes these bundles are twisted knots that cannot be simplified (irrational).
When looking for solutions in number systems, the only thing that can stop you is a specific, well-understood mathematical "force field" (the Brauer-Manin obstruction).

They didn't invent a new machine or cure a disease; they simply mapped out the hidden geometry of these mathematical puzzles, proving that some of them are inherently more complex and "knotted" than previously believed.

Technical Summary: Birational Properties of Word Varieties

Problem Statement
The paper investigates the birational geometry of word varieties $S_{w,\alpha}$ , defined as the closed subvarieties of $SL(2) \times SL(2)$ over a field $k$ (characteristic $\neq 2$ ) satisfying the matrix equation $w(A, B) = \alpha$ . Here, $w(x, y)$ is a non-trivial word in the free group $F_2$ , and $\alpha \in SL(2)(k)$ . The primary motivation is to determine the rationality properties of these varieties and to analyze the existence and distribution of $k$ -rational points, particularly when $k$ is a number field. A specific focus is placed on the commutator word $w = xyx^{-1}y^{-1}$ , addressing a question posed by Rapinchuk, Benyash-Krivetz, and Chernousov regarding whether such varieties are always $k$ -rational.

Methodology
The authors employ a systematic reduction of the structure group of the word variety to a norm 1 torus, utilizing the classical theory of trace polynomials. The core methodological steps include:

Trace Surfaces: Using the Cayley–Hamilton theorem, the trace of the word evaluation, $\text{tr}(w(A, B))$ , is expressed as a polynomial $P_w(s, t, u)$ in the traces $s=\text{tr}(A)$ , $t=\text{tr}(B)$ , and $u=\text{tr}(AB)$ . The "trace surface" $H_{w,a}$ is defined by $P_w(s, t, u) = a$ , where $a = \text{tr}(\alpha)$ .
Torsor Reduction: For a non-central semisimple element $\alpha$ , the variety $S_{w,\alpha}$ is identified as a torsor over the trace surface $H_{w,a}$ for the norm 1 torus $R^1_{k'/k}(\mathbb{G}_{m, k'})$ , where $k' = k(\sqrt{a^2-4})$ .
Conic Bundles: By analyzing the simultaneous similarity of matrices, the authors relate the word variety to a smooth conic bundle (a Severi–Brauer scheme of relative dimension 1) over the trace surface. Specifically, they show that $S_{w,\alpha}$ is a dense open subset of this conic bundle.
Brauer Group Analysis: The rationality and approximation properties are determined by computing the unramified Brauer group of the associated surfaces (specifically Markoff surfaces) and analyzing the Brauer–Manin obstruction.

Key Contributions and Results

Geometric Structure of Word Varieties:
The paper establishes that for any non-trivial word $w$ and non-central semisimple $\alpha$ , the variety $S_{w,\alpha}$ is a dense open subset of a smooth conic bundle over the trace surface $H_{w,a}$ . If $w$ belongs to the commutator subgroup $[F_2, F_2]$ , the trace surface is contained within the complement of the Cayley cubic (the "Markoff surface" context), ensuring the conic bundle is well-behaved.
The Commutator Variety and Rationality:
For the commutator word $w = xyx^{-1}y^{-1}$ , the trace surface is the Markoff surface $M_d$ defined by $s^2 + t^2 + u^2 - stu = d + 2$ , where $d = \text{tr}(\alpha) - 2$ .
- Irrationality: The authors prove that the commutator variety $S_\alpha$ is $k$ -rational if and only if the Markoff surface $M_d$ is $k$ -rational. This occurs if and only if either $d$ or $d(d+4)$ is a square in $k$ .
- Negative Answer to Rapinchuk et al.: Consequently, for "general" $\alpha \in SL(2)(k)$ , the commutator variety is irrational over $k$ . This provides a negative answer to the question raised by Rapinchuk, Benyash-Krivetz, and Chernousov regarding the universal rationality of these varieties.
Brauer Group and Unramified Classes:
The paper computes the Brauer class of the associated conic bundle. For the commutator case, the class in the Brauer group of the Markoff surface is given by the quaternion algebra $(t^2-4, d)$ . This class generates the unramified Brauer group of the generic Markoff surface modulo $\text{Br}(k)$ under general conditions.
Rational Points and Weak Approximation:
When $k$ is a number field, the paper demonstrates that the Brauer–Manin obstruction is the only obstruction to weak approximation on smooth and proper models of the commutator variety.
- If $\text{tr}(\alpha) + 2$ is a square in $k$ but $\text{tr}(\alpha) - 2$ is not, weak approximation holds unconditionally (no Brauer–Manin obstruction).
- In other cases, the obstruction is precisely described by the Brauer group element derived from the conic bundle structure.
Flatness of the Commutator Map:
As a byproduct of their method, the authors show that the commutator map $SL(2) \times SL(2) \to SL(2)$ is faithfully flat over the complement of the identity element in $SL(2)$. This result is derived without relying on the recent work of Larsen and Lu, though it aligns with their findings for prime $n$ .

Significance
The paper's significance lies in its ability to reduce the complex problem of word equations in $SL(2)$ to the study of rational points on explicit affine surfaces (trace surfaces) and their associated conic bundles. By linking these varieties to the geometry of Markoff surfaces and utilizing the Brauer–Manin obstruction, the authors provide a complete classification of the birational type of the commutator variety. This resolves a specific open question in the field by demonstrating that these varieties are not universally rational, thereby refining the understanding of the arithmetic and geometric properties of word equations in algebraic groups. The work also offers a unified framework for analyzing $k$ -rational points on word varieties over number fields.