---
title: Bonded Braid Monoid Overview
url: https://www.emergentmind.com/topics/bonded-braid-monoid
type: topic
---

# Bonded Braid Monoid Overview

Searching arXiv for recent papers on bonded braid monoids and closely related braid monoid structures.
arxiv_search(query="bonded braid monoid bonded braids Markov theorem", max_results=10)
arxiv_search(query="Bonded braids and the Markov theorem", max_results=5)
The bonded braid monoid is a braid-theoretic monoid designed to encode **bonded knots** or **bonded links**, namely classical knots or links endowed with additional embedded arcs called **bonds** whose endpoints lie on the underlying knot or link and become trivalent vertices. In the recent literature, bonds model intramolecular interactions such as disulfide bridges in topological protein modeling, and the resulting algebraic structure extends ordinary braid theory by adjoining bond generators to the usual braid generators. The theory now includes topological and rigid variants, Alexander- and Markov-type theorems, and Burau-type linear representations; at the same time, the monoid is shown to be algebraically isomorphic to the singular braid monoid, even though its geometric interpretation is formulated in terms of embedded bond connections rather than singular crossings [2507.04565] [2507.15086].

## 1. Geometric setting and motivating objects

A **topological bonded knot** is defined as a pair \((L,\mathbf b)\) where \(K\subset S^3\) is an oriented knot and
\[
\mathbf b=\{b_1,\dots,b_n\}
\]
is a collection of pairwise disjoint embedded intervals with endpoints on \(K\), each bond meeting the knot exactly in its endpoints. These endpoints are trivalent vertices, so bonded knots may be viewed as edge-colored spatial graphs. The same geometric picture underlies bonded links and their braid representatives [2507.04565].

Two isotopy regimes are distinguished. In the **topological** setting, twisting at a bond vertex is not rigid; in the **rigid-vertex** setting, twisting at vertices is restricted. A useful simplification is the passage to **isolated bonds**, meaning that bonds appear locally trivial and crossing-free in the diagram. In the topological setting, one may furthermore distinguish **parallel** and **non-parallel** bonds according to the local orientations of the adjacent knot arcs, and the available moves allow every oriented topological bonded knot to be replaced by an equivalent one with only parallel bonds [2507.04565].

A broader 2025 synthesis organizes bonded links into **long**, **standard**, and **tight** categories according to the type of bonds, and also distinguishes **topological vertex isotopy** from **rigid vertex isotopy**. For the bonded braid monoid itself, that account explicitly restricts attention to the **standard** and **tight** categories undergoing topological vertex isotopy. In that framework, a bonded braid on \(n\) strands is a pair \((\beta,B)\), where \(\beta\) is a classical braid on \(n\) strands and \(B\) is a set of disjoint embedded horizontal simple arcs called bonds; the endpoints of the bonds are nodes lying on braid strands and not coinciding with braid endpoints or with other nodes [2507.15086].

## 2. Algebraic presentations

Two notational conventions currently coexist in the literature.

| Source | Topological bonded braid monoid | Larger companion structure |
|---|---|---|
| [2507.04565] | \(M_n\) | \(RM_n\), \(BB_n\), \(RB_n\) |
| [2507.15086] | \(BB_n\) | \(EB_n\) |

In the topological presentation, the bonded braid monoid on \(n\) strands is generated by the usual braid generators
\[
\sigma_1^{\pm1},\dots,\sigma_{n-1}^{\pm1}
\]
together with bond generators
\[
b_1,\dots,b_{n-1},
\]
where \(b_i\) is a bond joining the \(i\)-th and \((i+1)\)-st strands. The defining relations are the ordinary braid relations
\[
\sigma_i \sigma_i^{-1} = Id_n,
\qquad
\sigma_i \sigma_j = \sigma_j \sigma_i \ \ (|i-j|\ge 2),
\qquad
\sigma_i \sigma_{i+1} \sigma_i = \sigma_{i+1} \sigma_i \sigma_{i+1},
\]
together with
\[
b_i b_j = b_j b_i \ \ (|i-j|\ge 2),
\]
and the mixed braid-bond relations
\[
\sigma_i b_j = b_j \sigma_i \ \ (|i-j|\ge 2),
\qquad
\sigma_i b_i = b_i \sigma_i,
\]
\[
\sigma_{i+1}\sigma_i b_{i+1} = b_i \sigma_{i+1}\sigma_i,
\qquad
\sigma_i \sigma_{i+1} b_i = b_{i+1} \sigma_i \sigma_{i+1}.
\]
The resulting object is a **monoid**, not a group: the braid generators are invertible, but the bond generators are not assumed invertible [2507.04565].

The rigid version enlarges the generator set by adding **kink generators**
\[
k_1,\dots,k_{n-1},
\]
producing the rigid bonded braid monoid \(RM_n\). Besides the braid relations and the bond relations above, one imposes
\[
k_i k_j = k_j k_i \ \ (|i-j|\ge 2),
\qquad
b_i k_j = k_j b_i \ \ (|i-j|\ge 2),
\]
as well as mixed braid-kink relations parallel to the braid-bond relations:
\[
\sigma_i k_j = k_j \sigma_i \ \ (|i-j|\ge 2),
\qquad
\sigma_i k_i = k_i \sigma_i,
\]
\[
\sigma_{i+1}\sigma_i k_{i+1} = k_i \sigma_{i+1}\sigma_i,
\qquad
\sigma_i \sigma_{i+1} k_i = k_{i+1} \sigma_i \sigma_{i+1}.
\]
The topological monoid embeds as a submonoid,
\[
M_n \subset RM_n.
\]
A second 2025 account gives the same reduced elementary-bond presentation, denoting the monoid by \(BB_n\) and referring to that reduced version as the **tight bonded braid monoid** [2507.04565] [2507.15086].

## 3. Reduction to elementary bonds and structural comparisons

Besides the elementary generators \(b_i\), the geometric theory also uses bonds \(b_{i,j}\) joining nonadjacent strands \(i\) and \(j\). A central reduction lemma states that a **standard bond is a word of the classical braid generators and an elementary bond**. Geometrically, one contracts a long bond to an elementary one by braided vertex slides; algebraically, this yields a presentation with only the ordinary braid generators and the elementary bonds \(b_1,\dots,b_{n-1}\) [2507.15086].

The same source proves that all elementary bonds are conjugates of a single bond generator. Explicitly,
\[
b_2 = (\sigma_2 \sigma_1)^{-1}  \, b_1\, (\sigma_2 \sigma_1),
\]
\[
b_3  = (\sigma_3 \sigma_2)^{-1}  (\sigma_2 \sigma_1)^{-1}  \, b_1\, (\sigma_2 \sigma_1) (\sigma_3 \sigma_2),
\]
and in general
\[
b_i = (\sigma_i \sigma_{i-1})^{-1} \cdots (\sigma_2 \sigma_1)^{-1}
\, b_1\, (\sigma_2 \sigma_1) \cdots  (\sigma_i \sigma_{i-1}).
\]
This produces an irredundant presentation with generators \(\sigma_1^{\pm1},\dots,\sigma_{n-1}^{\pm1},b_1\) [2507.15086].

The classical braid group sits inside the bonded theory by restricting to the braid generators:
\[
B_n \le M_n.
\]
Conversely, the bonded theory admits natural maps back to \(B_n\): one account notes surjections
\[
BB_n \twoheadrightarrow B_n
\]
obtained by sending either \(b_i\mapsto \sigma_i\) or \(b_i\mapsto id\) [2507.04565] [2507.15086].

A major structural fact is the algebraic identification with the singular braid monoid. In the notation of the topological presentation, \(M_n\) is isomorphic to the singular braid monoid \(SB_n\); in the alternative notation, \(BB_n\cong SB_n\). Under this correspondence,
\[
b_i \longmapsto \tau_i,
\qquad
\sigma_i \longmapsto \sigma_i,
\]
where \(\tau_i\) is the singular generator. The relation is not merely formal: one paper interprets a bond geometrically as a **tangential singularity**, while the other uses the isomorphism to import a faithfulness result for three strands [2507.04565] [2507.15086].

## 4. Closure, braiding theorems, and Markov-type equivalence

The closure operation is defined exactly as for ordinary braids: corresponding top and bottom endpoints are connected, and the bonds remain part of the resulting diagram. For \(\beta\) in the bonded braid monoid, the closure \(\hat\beta\) is therefore a bonded knot or bonded link [2507.04565].

The bonded analogue of Alexander’s theorem is established in both topological and rigid settings. In the topological case, every topological bonded knot can be represented as the closure of an element of the bonded braid monoid \(M_n\). The proof follows the classical Alexander strategy: isolate the bonds, assume they are parallel, contract each isolated bond to a short segment, rotate it inside a small disk so that it becomes colinear with the braiding axis, and then replace each non-braided segment by braided segments using \(\Delta\)-moves. The rigid version similarly shows that every rigid bonded knot can be represented as the closure of an element of \(RM_n\) [2507.04565].

A parallel braiding theorem for the standard and tight categories states that **every oriented topological standard bonded link can be represented isotopically as the closure of a standard resp. tight bonded braid**. That version formulates the theorem directly in terms of topological standard and tight bonded links, reflecting its choice of ambient categories [2507.15086].

The Markov theory introduces the new feature absent from the classical case: bonds can move cyclically around the closure. In the topological formulation, two bonded braids have equivalent closures if and only if they are related by a finite sequence of
\[
\beta \rightarrow \sigma_i^{\pm1}\beta\sigma_i^{\mp1},
\qquad
\beta b_i \leftrightarrow b_i \beta,
\qquad
\beta \leftrightarrow \beta \sigma_n^{\pm1}.
\]
Thus, in addition to conjugation and stabilization, one must allow **cyclic permutation of bonds**. In the rigid version, one likewise allows cyclic permutation of the kink generators,
\[
\beta k_i \leftrightarrow k_i \beta.
\]
The topological proof adapts Morton’s threading argument and analyzes the additional move in which the threading curve passes through a bond [2507.04565].

A second formulation replaces Markov moves by bonded \(L\)-moves. There are \(L_o\)- and \(L_u\)-moves, depending on whether the newly created strands run entirely over or entirely under the rest of the braid, including the bonds. In the tight setting one uses **resolved \(L\)-moves**, because any new crossing with a bond must be removed by braided vertex slide moves to recover tight form. The corresponding equivalence theorem adds **bond commuting**
\[
\alpha\, b_{i,j} \sim b_{i,j}\,\alpha
\]
for standard bonds, or
\[
\alpha\, b_i \sim b_i\, \alpha
\]
for elementary bonds in the tight setting. The associated Markov theorem has three moves:
\[
\alpha \sim \sigma_i^{\pm 1}\, \alpha\, \sigma_i^{\mp 1},
\qquad
\alpha\, b_i \sim b_i \, \alpha,
\qquad
\alpha \sim \alpha\, \sigma_n^{\pm 1}.
\]
That account explicitly remarks that there is **no separate bond-stabilization move**, since such a move would produce a vertical bond rather than a legitimate bonded braid [2507.15086].

## 5. Bonded Burau representations and faithfulness

A principal algebraic construction is the **bonded Burau representation**, extending the classical Burau representation to the bonded setting. The braid generator is sent to the usual Burau block
\[
\sigma_i \mapsto A_i =
\left(
\begin{array}{c:cc:c}
I_{i-1} & 0 & 0 & 0 \\
\hdashline
0 & 1-t & t & 0 \\
0 & 1 & 0 & 0 \\
\hdashline
0 & 0 & 0 & I_{n-i-1}
\end{array}
\right),
\]
while the bond generator is sent to
\[
b_i \mapsto B_i =
\left(
\begin{array}{c:cc:c}
I_{i-1} & 0 & 0 & 0 \\
\hdashline
0 & 1-tz & tz & 0 \\
0 & z & 1-z & 0 \\
\hdashline
0 & 0 & 0 & I_{n-i-1}
\end{array}
\right).
\]
This yields a homomorphism
\[
\psi: M_n \to \mathrm{GL}_{n}(\mathbb{Z}[t,t^{-1}, z]),
\]
and, after adjoining inverses for the bonds, a group-valued version
\[
\psi': BB_n \to \mathrm{GL}_{n}\!\big(\mathbb Z[t,t^{-1},z,(1-z-tz)^{-1}]\big).
\]
The inverse block for \(b_i\) is explicitly computed, which is why the localization by \((1-z-tz)^{-1}\) appears [2507.04565].

In the rigid setting, the representation extends further by sending the kink generator \(k_i\) to
\[
C_i =
\left(
\begin{array}{c:cc:c}
I_{i-1} & 0 & 0 & 0 \\
\hdashline
0 & 1 - t\check z & t\check z & 0 \\
0 & \check z & 1 - \check z & 0 \\
\hdashline
0 & 0 & 0 & I_{n-i-1}
\end{array}
\right),
\]
giving
\[
\psi_R: RB_n \to \mathrm{GL}_{n}(\mathbb Z[t,t^{-1}, z, \check z]).
\]
A localized group representation \(\psi_R'\) is also written down; the source notes a notational inconsistency between \(\check z\) and \(\hat z\) in the denominator [2507.04565].

The **reduced bonded Burau representation** is obtained in exact analogy with the classical reduced Burau construction by conjugating with
\[
C_n=
\begin{pmatrix}
1 & 1 & \cdots & 1 \\
0 & 1 & \cdots & 1 \\
0 & 0 & \ddots & \vdots \\
0 & 0 & \cdots & 1
\end{pmatrix}
\]
and taking the \((n-1)\times(n-1)\) upper-left blocks. The resulting representation
\[
\psi_n^r:M_n\to \mathrm{GL}_{n-1}(\Lambda),
\qquad
\Lambda=\mathbb Z[t^{\pm1},z],
\]
is reducible for all \(n\ge 2\), with invariant line \(\langle Ce_n\rangle\) [2507.04565].

The faithfulness picture is partly parallel to the classical one. The reduced bonded Burau representation is faithful for \(n=1\) and \(n=2\); for \(n=2\),
\[
\psi_2^r(\sigma_1)=-t,
\qquad
\psi_2^r(b_1)=1-z-tz,
\]
and injectivity follows from the fact that \(-t\) is a unit but \(1-z-tz\) is not. It is also faithful for \(n=3\), via the isomorphism with the singular braid monoid and the three-strand faithfulness result of Dasbach and Gemein. The case \(n=4\) is stated to be unknown, while for \(n\ge 5\) the representation is not faithful because it extends the classical Burau representation [2507.04565].

## 6. Related constructions, distinctions, and open directions

The bonded braid monoid belongs to a broader landscape of braid-like monoids with additional strand-connecting structure, but the different constructions are not interchangeable. A close relative is the theory of **tied monoids**, where one starts with a monoid \(M\), an idempotent commutative monoid \(P\) of set partitions, and an action \(\rho\), and defines the tied monoid as the semidirect product
\[
T^{\Gamma}\!M = P\rtimes_{\rho} M.
\]
In type \(A\), this produces the tied braid monoid with generators \(\sigma_i^{\pm1}\) and tie generators \(\eta_i\), subject in particular to
\[
\eta_i^2=\eta_i,
\qquad
\eta_i\eta_j=\eta_j\eta_i,
\]
together with mixed braid-tie relations. The tied construction is therefore partition-based, idempotent, and semidirect-product in nature, whereas the bonded braid monoid models actual embedded bond connections and does not impose idempotence on the bond generators [2001.00625].

A related Coxeter-level development studies **ramified** and **tied** monoids attached to the symmetric, Brauer, and Jones monoids. There the extra generators again represent ties rather than bonds, and the paper identifies the tied symmetric monoid with the ramified monoid of the symmetric group:
\[
TS_n \cong \mathcal{R}S_n.
\]
That framework is diagrammatically very close to bonded strands, but algebraically its ties encode partition classes and extended tie data rather than the noninvertible bond generators of the bonded braid monoid [2107.04170].

Miyatani’s **braid \(\mathbb{P}M\)-monoid** is different again. Its additional generators
\[
e_{k_1,\dots,k_{m-1}}
\]
encode ordered interval partitions and geometric layers; the monoid is built from a matched-pair decomposition with ordered set partitions and is realized by geometric layered braids. The extra structure is explicitly described there as neither singular crossings nor local bond generators, but rather as a global partition/layer decomposition of the full set of strands. For that reason, the braid \(\mathbb{P}M\)-monoid is conceptually adjacent to grouped or layered braids, not to the bonded braid monoid in the usual sense [1906.09398].

Several issues remain open or explicitly incomplete in the present bonded theory. The representation theory of \(M_n\), \(RM_n\), \(BB_n\), and \(RB_n\) is described as only initiated. Faithfulness of the reduced bonded Burau representation for \(n=4\) remains unknown. In addition, when inverses for bond generators are adjoined, one obtains “anti-bonds,” but these are said to lack a clear physical interpretation. A plausible implication is that the current theory is algebraically well formed but still separating its geometric, biological, and representation-theoretic motivations from one another [2507.04565].

Source: https://www.emergentmind.com/topics/bonded-braid-monoid