4
$\begingroup$

Let $n\geq 3$ be a positive integer and $\kappa=(k_1, \dots, k_n)\in \mathbb{Z}^n$. Denote by $B_n$ the braid group on $n$ strings. Consider the braid on $n+1$ strings $\sigma_\kappa:=\sigma_1^{k_1}\dots \sigma_n^{k_n}$, where $\sigma_i=\sigma_i^+$ is the generator taking the $i$-th string to the $(i+1)$-st string and $(i+1)$-st string to the $i$-th string in a single overlap (as in Fig. 1.9 on p. 16 of Kassel and Turaev, Braid groups, volume 247, Springer science and Business media).

On closing up the braid $\sigma_\kappa$ one obtains a link, which we call $L_\kappa$, with associated Alexander polynomial $\Delta_\kappa(t)\in \Lambda$, where $\Lambda=\mathbb{Z}[t^{\pm 1}]$, which is well defined up to a unit in $\Lambda$. A well known formula for the Alexander polynomial $\Delta_b(t)$ of a link associated to a braid $b$ states that $\Delta_{b}(t)=\frac{(1-t)}{(1-t^n)}\operatorname{det}\left(\operatorname{I}_n-\bar{\psi}_n(b)\right)$, where $\bar{\psi}_n: B_n\rightarrow \operatorname{GL}_{n-1}(\Lambda)$ is the reduced Burau representation (c.f. Chapter 3 of loc. cit.), which arises from interpreting $B_n$ as the mapping class group of a disk with $n$ marked points.

Using this formula, I was able to show for $n=3, 4$ that the Alexander polynomial $\Delta_\kappa(t)$ is $\prod_{i=1}^n F_{k_i}(t)$, where $F_r(t):=1-t+t^2+\dots+(-1)^{r-1}t^{r-1}=\frac{1-(-1)^r t^r}{1+t}$. When $n$ gets larger, the size of the matrices get larger and in general it is not clear to me if a formula similar to this should hold for all $n$? I wasn't able to find this computation appear in the literature, but it seems like something that should be known. Also, I was wondering if this family of braids has any special significance.

$\endgroup$

1 Answer 1

3
$\begingroup$

The closure of the braid $\sigma_\kappa$ is a connected sum of torus links $T(2,k_i)$ (which are closures of 2-braids). Since the Alexander polynomial is multiplicative with respect to connected sums, $\Delta_\kappa = \prod_i \Delta_{T(2,k_i)}$.

You can see more generally that if you have a braid $\beta = \beta_{\rm top}\beta_{\rm bot} \in B_n$ which is a product of two braids where $\beta_{\rm top} \in \langle \sigma_1,\dots,\sigma_m\rangle$ and $\beta_{\rm bot}\in \langle \sigma_{m+1},\dots,\sigma_{n-1}\rangle$, then the closure of $\beta$ is the connected sum of the closures of $\beta_{\rm top}$ and $\beta_{\rm bot}$ (viewed as an $(m+1)$-braid and an $(n-m)$-braid, respectively). The circle running (mostly) between the $(m+1)^{\rm st}$ and the $(m+2)^{\rm nd}$ strands exhibits the connected sum.

$\endgroup$
2
  • $\begingroup$ Given two links $L_1$ and $L_2$, and two distinguished knot components $K_1$ and $K_2$, is the connected sum of $L_1$ and $L_2$ along $K_1$ and $K_2$ in general well defined? Also, what's a reference for the Alexander polynomial decomposing into a product in this case, when the links are not knots themselves? I sort of see how the argument applies here, but it would be good to have a reference. Also, is it somehow possible to prove this as an application of the formula for the Alexander polynomial via the Burau representation? $\endgroup$
    – Anwesh Ray
    Dec 9, 2022 at 15:34
  • $\begingroup$ If $K_1$ and $K_2$ are oriented, the connected sum is well-defined. As for the connected sum reference, I'd go to Lickorish's book. As for proving it via the Burau representation, I really don't know, I'm sorry. $\endgroup$ Dec 9, 2022 at 17:38

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.