Dual involutions in finite Coxeter groups

Marcus Zibrowius

TL;DR

This paper provides a complete classification of "dual involutions" in finite Coxeter groups. It identifies exactly which conjugacy class results from multiplying a given involution by the group's longest element, providing a comprehensive "lookup table" for all finite Coxeter types (e.g., $A_n, D_n, E_8, H_4$).

What Problem Does This Paper Address?

In the study of symmetry groups known as Coxeter groups (which include the symmetries of regular polyhedra and reflection groups), "involutions"—elements that are their own inverse ($w^2 = 1$)—play a central role. Mathematicians have long known how to classify these involutions into conjugacy classes based on subgraphs of the group's Coxeter diagram.

However, every finite Coxeter group has a unique "longest element" ($w_o$). In many cases, multiplying an involution $w$ by $w_o$ produces another involution. The central question of this paper is: If you know the conjugacy class of $w$, what is the conjugacy class of the "dual" involution $w_o w$?

While this sounds like a simple calculation, doing it by "brute force" for large groups is computationally exhausting and provides little insight. This paper seeks a general, combinatorial rule to identify these dual classes across all finite Coxeter groups.

Methods

The author avoids brute-force computation by employing three elegant geometric and algebraic strategies:

  1. Eigenspace Analysis: By viewing the Coxeter group as a reflection group in Euclidean space, the author looks at the dimensions of the $+1$ and $-1$ eigenspaces of the involutions. Since $w_o$ often acts as $-id$ (negating everything), multiplying by $w_o$ simply swaps these eigenspaces.
  2. Sub-root Systems (Borel–de Siebenthal Theory): When dimensions alone aren't enough to distinguish two classes, the author examines the specific types of root systems contained within the fixed-point spaces. This allows for finer differentiation between classes that might otherwise look identical.
  3. Steinberg Folding: This is a sophisticated technique where a larger, more symmetric root system is "folded" onto a smaller one via an automorphism (e.g., folding the $A_{10}$ system to study $C_5$). This allows the author to transfer known results from one family of groups to another.

Datasets

As a pure mathematics paper, it does not use external datasets. Instead, it operates on the complete classification of finite Coxeter groups (the "ADE" classification and its extensions), which is a standard foundation in algebraic Lie theory.

Key Insights and Ideas

  • The Dimension Rule: For a "standard" involution $c_I$ (defined by a subset of nodes $I$ in the Coxeter graph), the dimension of its $-1$-eigenspace is exactly the number of nodes in $I$. This provides a quick way to narrow down the possible "dual" classes.
  • Centrality Matters: The problem is most straightforward when $w_o$ is central (meaning it commutes with everything). The author identifies exactly which Coxeter groups satisfy this (Fact 1), including types $E_7, E_8, F_4, H_3,$ and $H_4$.
  • Handling Non-Centrality: For groups where $w_o$ is not central (like $A_n$ or $E_6$), the author shifts the focus to the subgroup of elements that do commute with $w_o$, allowing the duality logic to still apply.
  • Graphical Calculus: The paper translates complex algebraic properties into a "graphical calculus," where conjugacy classes are represented by coloring nodes on a Dynkin diagram.

Key Results

The paper provides a definitive set of results for every finite Coxeter group, summarized in Section 2:

  • Classical Families: For types $A_n, BC_n,$ and $D_n$, the author provides explicit formulas. For example, in $W(BC_n)$, an involution defined by $k$ and $l$ marked roots is mapped to one defined by $n - 2k - l$ roots.
  • Exceptional Groups: The paper includes detailed diagrams for the "exceptional" cases ($E_6, E_7, E_8, F_4, G_2, H_3, H_4$), showing exactly how the duality operation permutes the different conjugacy classes.
  • D_n Parity: The author resolves a tricky case in type $D_n$ (even dimension), where two specific classes ($c_-$ and $c_+$) are either fixed or swapped by $w_o$ depending on whether $n+1$ is a multiple of 4.

Limitations and Future Work

  • Finite Scope: The results are strictly for finite Coxeter groups. The author does not address affine or infinite Coxeter groups, where the concept of a "longest element" is different or non-existent.
  • Applications in Topology: The author mentions that these results were motivated by work on the real K-theory of homogeneous spaces. While the algebraic work is complete here, the specific topological implications are left for other publications (e.g., [HZ21]).

Summary updated

Abstract

There is a well-known classification of conjugacy classes of involutions in finite Coxeter groups, in terms of subsets of nodes of their Coxeter graphs. In many cases, the product of an involution with the longest element is again an involution. We identify the conjugacy class of this product involution in terms of said classification.
BibTeX
@inproceedings{zibrowius2023dual,
  author = {Marcus Zibrowius},
  title = {{Dual involutions in finite Coxeter groups}},
  booktitle = {Communications},
  year = {2023},
  doi = {10.1080/00927872.2023.2255663},
}
Figure