🔢 mathematics

Totally decomposable algebras with involution and R-triviality

This paper proves that the group of proper projective similitudes for totally decomposable algebras with orthogonal or symplectic involutions of index at most 2 over a field of characteristic not 2 is R-trivial, while providing a counterexample to show this property fails for index 4 in the orthogonal case.

Original authors: M. Archita, Karim Johannes Becher

Published 2026-08-19

📖 4 min read🧠 Deep dive

CC BY 4.0

Original authors: M. Archita, Karim Johannes Becher

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

Mathematics often deals with shapes that exist not in space, but in the realm of pure logic. These are algebraic structures, intricate systems built from numbers and rules that govern how those numbers interact. For decades, mathematicians have been trying to understand the fundamental nature of these systems, specifically asking whether they can be described by simple, smooth geometric forms. This question is not just about abstract beauty; it touches on the deep architecture of symmetry that underpins much of modern science. To answer it, researchers examine how these algebraic systems behave when they are stretched or twisted by changing the field of numbers they live in. A key concept in this investigation is a property called R-triviality. In simple terms, this property asks whether every possible state of the system can be reached from a starting point through a continuous, logical path. If a system is R-trivial, it is essentially flexible and connected; if it is not, there are hidden barriers that prevent movement between certain states, suggesting a more rigid and complex underlying structure.

In a recent study, mathematicians M. Archita and Karim Johannes Becher tackled a specific class of these systems known as totally decomposable algebras with involution. Think of these algebras as complex machines built by snapping together smaller, simpler blocks. An involution is a specific rule built into the machine that acts like a mirror, flipping elements in a way that preserves the overall structure. The researchers focused on machines where the blocks are arranged in a very particular, orderly way, and they wanted to know if the symmetry group of these machines—the collection of all ways the machine can be scaled or rotated while keeping its shape—was R-trivial. They investigated systems of a certain size, specifically those where the underlying complexity, measured by an index, was small. Their work confirms that for these smaller, well-ordered machines, the answer is yes: they are completely flexible, with no hidden barriers blocking movement between states. This finding provides a solid foundation for understanding when these complex algebraic forms behave like simple, rational shapes.

However, the story does not end with a simple confirmation. The researchers discovered that this flexibility has a strict limit. They proved that the rule holding true for smaller systems breaks down completely when the complexity of the machine increases to a specific higher level. By constructing a precise example of a larger, eight-dimensional machine, they showed that it is possible to build a system that is perfectly ordered and made of simple blocks, yet still possesses those hidden barriers that prevent it from being R-trivial. This counter-example is crucial because it demonstrates that the property of being built from simple parts does not guarantee that the whole system will be simple in its behavior. The study establishes a clear boundary: the rule works for systems with an index of two or less, but it fails for systems with an index of four.

The implications of this work reach beyond just these specific algebraic machines. The failure of R-triviality in the larger example suggests that other related mathematical properties, such as the ability to approximate points on the system using rational numbers, also break down in these cases. The authors provide a concrete instance of this failure, showing that for certain field extensions, the rational points on the group are not dense, meaning there are gaps that cannot be filled by simple calculations. This result challenges the idea that all well-structured algebraic groups are easy to navigate. While the paper does not claim to have solved the entire mystery of when these groups are simple, it has drawn a sharp line in the sand. It proves that total decomposability is not enough to ensure simplicity when the system grows large enough, forcing mathematicians to look for deeper conditions that determine the true nature of these algebraic landscapes.

Problem Statement
The paper addresses the rationality properties of classical adjoint linear algebraic groups, specifically the group of proper projective similitudes, denoted $PSim^+(A, \sigma)$ , associated with a central simple algebra $A$ over a field $K$ (characteristic $\neq 2$ ) equipped with an involution $\sigma$ of the first kind. While it is known that semisimple adjoint groups of type $^1A$ and $B$ are rational, this property fails for types $^2A$ , $C$ , and $D$ . A key obstruction to stable rationality is $R$ -triviality: a group $G$ is $R$ -trivial if the group of $R$ -equivalence classes $G(F)/R$ is trivial for every field extension $F/K$ .

The specific problem investigated is whether $PSim^+(A, \sigma)$ is $R$ -trivial when $(A, \sigma)$ is a totally decomposable algebra with involution. A totally decomposable algebra is defined as a tensor product of a central simple algebra with involution of the form $Ad(\psi)$ (associated with an odd-dimensional quadratic form $\psi$ ) and a tensor product of quaternion algebras with involution. Previous work by Merkurjev established that if $A$ is split, the group is $R$ -trivial (and stably rational). The authors investigate whether this holds when $A$ is not necessarily split, focusing on the constraint of the index of $A$ (the degree of the division algebra Brauer-equivalent to $A$ ).

Methodology
The authors utilize the characterization of $R$ -equivalence classes provided by Merkurjev [13], which identifies $PSim^+(A, \sigma)(F)/R$ with a quotient of subgroups of $F^\times$ related to similarity factors and norms from splitting fields. Specifically, the group is $R$ -trivial if and only if the group of similarity factors $G^+(A, \sigma)$ coincides with the subgroup generated by norms from fields where the algebra becomes hyperbolic ( $Hyp(A, \sigma)$ ).

The methodology involves:

Decomposition Analysis: Utilizing the structure of totally decomposable algebras to express $(A, \sigma)$ as $Ad(\psi \otimes \rho) \otimes (Q, \gamma)$ , where $\rho$ is a Pfister form, $\psi$ is an odd-dimensional quadratic form, and $(Q, \gamma)$ is a quaternion algebra with involution.
Factorization of Similarity Factors: Proving that for algebras of index at most 2, any similarity factor can be factored into a product of two elements, each being a norm from a quadratic extension over which the involution becomes hyperbolic. This relies on properties of Pfister forms, Witt rings, and subform criteria (e.g., Lemma 3.5 and Lemma 3.6).
Counter-Example Construction: For the case of index 4 in the orthogonal setting, the authors construct explicit examples using trialitarian triples and function fields. They relate the algebra $(A, \sigma)$ to the even Clifford algebra of an 8-dimensional quadratic form with trivial discriminant, leveraging known results on the non- $R$ -triviality of such forms (specifically referencing constructions from [1] and [9]).

Key Contributions and Results

Theorem (Corollary 4.2): The primary positive result states that if $(A, \sigma)$ is a totally decomposable algebra with involution of the first kind and the index of $A$ is at most 2 ( $ind\ A \le 2$ ), then the algebraic group $PSim^+(A, \sigma)$ is $R$ -trivial.
- This result covers cases where $\sigma$ is symplectic (where it follows directly from the Pfister Factor Conjecture) and the non-trivial case where $\sigma$ is orthogonal and $ind\ A = 2$ .
- The proof demonstrates that $G(A, \sigma) = Hyp(A, \sigma)$ under these conditions.
Counter-Example (Corollary 5.3): The authors show that the condition $ind\ A \le 2$ is necessary for the orthogonal case. They construct a totally decomposable algebra with orthogonal involution of degree 8 and index 4 over the field $K = k_0(X, Y, Z)$ , where $k_0$ is a field of characteristic different from 2.
- Specifically, $(A, \sigma) = Ad\langle\langle X \rangle\rangle \otimes (Q_1, can_{Q_1}) \otimes (Q_2, can_{Q_2})$ with specific quaternion algebras $Q_1, Q_2$ .
- For this algebra, $PSim^+(A, \sigma)$ is not $R$ -trivial. There exists a field extension $F/K$ such that the group of $R$ -equivalence classes is non-trivial.
- This example also demonstrates a failure of the Weak Approximation property for these groups.
Generalization (Corollary 5.5): The paper further shows that the tensor product of algebras with involution does not preserve $R$ -triviality. Even if one factor is $R$ -trivial, the tensor product with a non-hyperbolic algebra with non-trivial discriminant can yield a non- $R$ -trivial group.

Significance
The paper establishes a precise boundary for the $R$ -triviality of groups of proper projective similitudes associated with totally decomposable algebras. It confirms that the property holds robustly when the underlying algebra has a low index (at most 2), extending previous results known for split algebras. Conversely, it provides a definitive counter-example showing that this property fails for index 4 in the orthogonal case, thereby clarifying the limitations of total decomposability in ensuring rationality properties. The work also highlights the connection between the failure of $R$ -triviality and the failure of Weak Approximation, linking these algebraic group properties to broader questions in the arithmetic of algebraic groups. The authors note that while $R$ -triviality is a necessary condition for stable rationality, the question of whether these groups are stably rational in the index 2 orthogonal case remains open (Question 4.3).

More like this