Subgroups of braid groups generated by Birman-Ko-Lee generators

We define a Young subgroup of the braid group as a subgroup generated by an arbitrary subset of the Birman-Ko-Lee generators. We give an intrinsic description of such subgroups which yields, in particular, an easy criterion to decide membership. We also give an algorithm to write an element of a Young subgroup as a product of the generators. Our methods are based on analyzing the Hurwitz action on tuples over free groups via a diagrammatic approach.

💡 Research Summary

The paper investigates subgroups of the n‑strand braid group Bₙ that are generated by arbitrary subsets of the Birman‑Ko‑Lee (BKL) generators aᵢⱼ (1 ≤ i < j ≤ n). These subgroups are called “Young subgroups” B_Q, where Q = {Q₁,…,Q_m} is a partition of the index set {1,…,n}. For each block Q_s, all generators aᵢⱼ with i, j ∈ Q_s are included; generators connecting different blocks are omitted. This definition generalises the classical situation where subgroups generated by subsets of the Artin generators σ_i are products of smaller braid groups.

The authors’ main contribution is an intrinsic description of B_Q in terms of the Hurwitz action of Bₙ on n‑tuples of a free group F_m = ⟨x₁,…,x_m⟩. For a given partition Q they construct a specific tuple
t = (t₁,…,tₙ) ∈ (F_m)ⁿ,
where each t_k has the form C_k x_{s(k)} C_k⁻¹ with s(k) indicating the block of Q containing k and C_k ∈ F_m arbitrary. The Hurwitz action is the usual one: σ_i swaps the i‑th and (i+1)‑th entries and conjugates the new i‑th entry by the former (i+1)‑th entry. Theorem 2.6(2) proves that the Young subgroup B_Q coincides exactly with the stabiliser of t under this action:  B_Q = Stab_{Bₙ}(t).
Consequently, membership testing reduces to a simple check: an element b ∈ Bₙ belongs to B_Q iff b·t = t.

In addition to the stabiliser description, the paper characterises the orbit of t (Theorem 2.6(1)). The orbit consists of all n‑tuples (g₁,…,gₙ) such that after applying a permutation τ ∈ Sₙ, each component g_{τ(i)} lies in the conjugacy class C(t_i) and the product g₁⋯gₙ equals the product t₁⋯tₙ. This description is purely combinatorial and does not depend on the specific choice of the C_k’s.

A substantial part of the work is devoted to a diagrammatic method that visualises the Hurwitz action. The authors introduce “arc diagrams” associated to reduced words in the free group. A reduction sequence for a word w ∈ (S^{±})* (where S = {x₁,…,x_m}) is represented by arcs connecting each generator to its cancelling partner, or to a point at infinity if it survives. By concatenating the words t₁,…,tₙ into a single word ˜t and choosing a reduction sequence, one obtains a planar diagram consisting of upper arcs (encoding the reduction) and lower arcs (encoding the conjugating elements C_k and C_k⁻¹). Mutations of these diagrams, depicted in Figure 4.2, correspond precisely to the elementary Hurwitz moves σ_i and σ_i⁻¹. This visual language makes the otherwise intricate braid relations among the aᵢⱼ transparent, especially in the “crossing” case where arcs intersect and generate free subgroups (Lemma 2.2).

The paper also provides an explicit algorithm (Section 7.2) for writing any element b ∈ B_Q as a product of the BKL generators aᵢⱼ. Starting from the equality b·t = t, one applies inverse Hurwitz moves to reduce the tuple step by step, recording at each step which σ_i or σ_i⁻¹ was used. Because each move corresponds to a known expression in the aᵢⱼ, the algorithm yields a concrete factorisation of b. This algorithm is more efficient than brute‑force manipulation of the defining relations and is well suited for implementation.

Overall, the authors achieve a unified framework that links three perspectives: (1) the combinatorial data of a partition Q, (2) the algebraic data of a stabiliser under the Hurwitz action, and (3) a geometric picture via arc diagrams. This framework works uniformly for both non‑crossing partitions (where B_Q decomposes as a direct product of smaller braid groups) and crossing partitions (where non‑commuting generators generate free subgroups). The results have immediate applications to the study of braid‑group actions on exceptional collections and mutation theory, as hinted in the motivation section. The paper thus advances our understanding of braid‑group substructures and provides practical tools for further algebraic and geometric investigations.