Shi arrangements and low elements in Coxeter groups

Matthew S. Dyer, Christophe Hohlweg, Susanna Fishel, Alice Mark

TL;DR

This paper provides a unified geometric and combinatorial framework for "Shi arrangements" across all Coxeter groups. The authors prove that the minimal elements of regions in these arrangements are exactly the "low elements" used to solve word problems in braid groups, settling two major conjectures regarding their uniqueness and convexity.

What Problem Does This Paper Address?

Coxeter groups are mathematical abstractions of reflection groups—think of the symmetries of a kaleidoscope or the tiling of a floor. Within these groups, researchers study hyperplane arrangements, which are collections of "mirrors" that divide space into regions.

A specific type, the Shi arrangement, was originally developed to study affine Weyl groups (a subset of Coxeter groups). It was known that each region in a Shi arrangement has a unique "simplest" (minimal length) element. Separately, researchers developed the concept of low elements to solve the "word problem" (determining if two sequences of reflections represent the same group element) in Artin-Tits groups.

The gap addressed here is twofold:

  1. Generalization: Most previous results only applied to specific "affine" types of groups. The authors wanted to see if these properties hold for any Coxeter system, including "indefinite" ones that describe hyperbolic spaces.
  2. Conjectures: Two major conjectures remained unproven: one stating that the minimal elements of Shi regions are precisely the low elements, and another regarding the "convexity" (the geometric "niceness") of these sets.

Methods

The authors employ a mix of geometric group theory and combinatorics. Key techniques include:

  • Chambered Sets: To ensure their results apply to various geometric realizations (like the Tits cone or Davis complex), they define a generalized structure called a "chambered set." This allows for a uniform treatment of hyperplanes and convex subsets.
  • Short Inversion Posets: The authors introduce a poset (partially ordered set) structure on "short inversions." Their core technical breakthrough is the Sandwich Theorem, which proves that any short inversion is bounded between a "left-descent" and a "right-descent" root.
  • Dominance Order: They utilize the dominance order on roots (introduced by Brink and Howlett) to characterize $m$-small roots, which are the building blocks of the $m$-Shi arrangement.

Datasets

As a theoretical mathematics paper, it does not use traditional datasets. Instead, it validates its theorems against various types of Coxeter groups:

  • Finite types: (e.g., $H_3$)
  • Affine types: (e.g., $\tilde{B}_2$)
  • Indefinite/Hyperbolic types: The authors provide a specific counterexample for indefinite types to show where the property of convexity breaks down (see Figure 7 in the paper).

Key Insights and Ideas

  • The Geometric-Combinatorial Link: The paper proves that the geometric regions of a Shi arrangement are "gated." This means that for any region, there is a unique "gate" (the minimal element) through which you must pass to enter the region from the starting point.
  • Bipodality: They show that the set of $m$-small roots is "bipodal." This property ensures that the resulting set of low elements forms a Garside shadow, a structure that is highly efficient for navigating the group's internal logic.
  • The Sandwich Property: The discovery that short inversions are always "sandwiched" between descent roots provides a new way to visualize how complex reflections are built from simpler ones.

Key Results

  • Settling Conjecture 1: The authors prove that for any Coxeter system and any integer $m \ge 0$, each region of the $m$-Shi arrangement contains a unique minimal length element, and the set of these elements is exactly the set of $m$-low elements.
  • Settling Conjecture 2: They prove that for the classical case ($m=0$), the union of the chambers corresponding to these elements forms a convex set.
  • Quantitative Enumeration: For affine types, they confirm the formula for the number of regions: $|L_m| = ((m + 1)h + 1)^n$, where $h$ is the Coxeter number and $n$ is the rank.
  • Indefinite Counterexample: They demonstrate that while the minimal element property holds for all groups, the convexity property fails for "indefinite" groups when $m > 0$.

Limitations and Future Work

The primary limitation identified is that the convexity of the union of chambers (Theorem 1.2) does not hold for all $m$ in indefinite Coxeter systems.

Future directions include: * Classification: Identifying exactly which Coxeter graphs allow for convexity when $m \ge 1$. * Enumeration: While formulas exist for affine types, counting $m$-small roots and $m$-low elements for hyperbolic and indefinite types remains an open and difficult problem. * Medium Roots: The authors propose a new category of "m-medium roots" as a potential way to extend convexity results to indefinite systems.

Summary updated

Abstract

Given an arbitrary Coxeter system $(W,S)$ and a nonnegative integer $m$, the $m$-Shi arrangement of $(W,S)$ is a subarrangement of the Coxeter hyperplane arrangement of $(W,S)$. The classical Shi arrangement ($m=0$) was introduced in the case of affine Weyl groups by Shi to study Kazhdan-Lusztig cells for $W$. As two key results, Shi showed that each region of the Shi arrangement contains exactly one element of minimal length in $W$ and that the union of their inverses form a convex subset of the Coxeter complex. The set of $m$-low elements in $W$ were introduced to study the word problem of the corresponding Artin-Tits (braid) group and they turn out to produce automata to study the combinatorics of reduced words in $W$. In this article, we generalize and extend Shi's results to any Coxeter system for any $m$: (1) the set of minimal length elements of the regions in a $m$-Shi arrangement is precisely the set of $m$-low elements, settling a conjecture of the first and third authors in this case; (2) the union of the inverses of the ($0$-)low elements form a convex subset in the Coxeter complex, settling a conjecture by the third author, Nadeau and Williams.
BibTeX
@inproceedings{dyer2024shi,
  author = {Matthew S. Dyer and Christophe Hohlweg and Susanna Fishel and Alice Mark},
  title = {{Shi arrangements and low elements in Coxeter groups}},
  booktitle = {Proceedings of the London Mathematical Society},
  year = {2024},
  doi = {10.1112/plms.12624},
}
Figure