6

Given the Artin braid groups on $n$ and $m$ strands $Br_n$ and $Br_m$, there are cabling operations $\circ_k:Br_n\times Br_m\to Br_{n+m-1}$ that take a braid $\beta\in Br_m$ and replace the $k$th strand of a braid in $Br_n$ with $\beta$. See the following picture for the operation $\circ_2\colon Br_4\times Br_2\to Br_5$:cabling in the second strand

Does anyone know of an explicit description of these operations in terms of generators? Writing them down in low degrees, I seem to be able to work out a rough idea of how they should operate, but it seems likely to me that someone has already figured this out?

  • 1
    Is it possible to reduce to the case where $m=2$? – Exit path Mar 31 '18 at 01:06
  • @leibnewtz I'm not sure about $m$, but maybe $n=2$? – Jonathan Beardsley Mar 31 '18 at 01:35
  • I might be wrong, but I don't think you can reduce to $n=2$. I'm imagining having $>2$ strands and exchanging the first with the last – Exit path Mar 31 '18 at 03:17
  • @JonathanBeardsley Let me see if I understand: $\alpha \circ_k \beta = (i_k \alpha) (D_k^n \beta) = (D_k^n \beta) (i_{\beta(k)} \alpha)$ where $i_k$ is the inclusion starting at the $k$th strand and $D_k$ doubles the $k$th strand. Right? – tcamps Mar 31 '18 at 21:04
  • @tcamps Ummmm, I think so? that notation is somewhat opaque to me. Basically the first $k-1$ strands do nothing, then you do $\beta$ on the next $m$ strands, then you have $n-k$ strands do nothing, and then you have to figure out how the trivial $m$-strand braid modifies $\alpha$ as it goes through on the $k$th strand. At least that's one possible algorithm. – Jonathan Beardsley Apr 01 '18 at 21:17

1 Answers1

3

Conventions:

  • I'll think of braids as running from an "input" side at the top to an "output" side at the bottom.

  • The strands in $B_n$ can be numbered $1,\dots,n$ on either the input side (top) or output side (bottom).

  • If $\gamma \in B_n$, then I'll write $\gamma(k)$ for the action of $\gamma$ on the numbering, i.e. if you take the $k$th strand on the input side, and trace it through, it winds up as the $\gamma(k)$th strand on the output side.

  • I'll write multiplication as follows: $\gamma \delta$ means $\gamma$ first, then $\delta$, i.e. $\gamma$ on top of $\delta$.


Unless I misunderstand the operation, we have

$$\alpha \circ_k \beta = (i_k \alpha) (D_k^n \beta) = (D_k^n \beta) (i_{\beta(k)} \alpha)$$

where

  • $i_j : Br_n \to Br_{n+m-1}$ is the inclusion at the $j$th strand: $i_j \gamma$ is a copy of $\gamma$ with $j-1$ identity strands to the left and $n+m-j$ identity strands to the right,

  • $D_k^n$ is the $n$-fold iterate of the "doubling the $k$th input strand" operator.

So it suffices to understand $D_k$. We have $D_k(\gamma \delta) = D_k(\gamma) D_{\gamma(k)} (\delta)$, so it will suffice to describe $D_k$ on generators. So let $g_j$ be the positive crossing of the $j$th input strand over the $(j+1)$st input strand. Then

$D_k g_j = \begin{cases} g_{j+1} & k < j \\ g_{j+1}g_j & k=j \\ g_j g_{j+1} & k=j+1 \\ g_j & k > j+1 \end{cases}$

The same formulas also work if we set $g_j$ to be the negative crossing at the $j$th strand.

I'm not sure if something nice comes out when you put this all together...

tcamps
  • 6,231
  • Wow Tim, I just came back to this after a while of not thinking about it. This is really nice, thanks for writing it! – Jonathan Beardsley Feb 07 '19 at 19:06
  • I guess it would be nice to have a really explicit formula like "if you write $\beta$ with these generators and $\gamma$ with these generators, then the cabling of $\beta$ into the $k^{th}$ strand of $\gamma$ is given by this string of generators." But that may be too much to ask for... – Jonathan Beardsley Feb 07 '19 at 19:08
  • @JonathanBeardsley No problem! The expression for $i_k(\alpha)$, of course, is not bad in terms of generators -- just shift the indices. But the doubling operator is complicated in terms of generators, and I guess my sense is that an algorithm like this is the best description one can hope for. Also beware, I'm not sure how standard my conventions are, or whether, once you translate them into standard conventions, I have the definition of "positive" correct or flipped. – tcamps Feb 07 '19 at 19:13
  • Yeah right, $D_k$ (and its iterates) is the tricky one. I'm trying to figure out how to use Sage to code this up right now. Sage actually does some good stuff with braids and braid groups. – Jonathan Beardsley Feb 07 '19 at 19:15
  • Not that it's actually probably of any use to anyone, but I've managed to write up some code implementing this algorithm in Sage. It was pretty fun! – Jonathan Beardsley Feb 10 '19 at 08:53