Take the 2-minute tour ×
MathOverflow is a question and answer site for professional mathematicians. It's 100% free, no registration required.

If you take a (labelled, oriented) 2 component link, it has a symmetry group which is a subgroup of the 16 element group $Z2 \times (Z2 \times Z2 \ltimes S_2)$ (mirror, reverse each component, swap the components). With Parsley, Cornish, and Mastin (http://arxiv.org/abs/1201.2722) I compiled a table of the frequency with which each conjugacy class of subgroups of this group comes up as a symmetry group for links up to 14 crossings.

Interestingly, the full group never came up as a symmetry group! Can a 2 component link ever have the full 16 element group? Is there some obstruction to this? Can anyone propose an example link?

Edit: A non split example. It is easy to construct split ones!

Clarification: The traditional symmetry group is the mapping class group $MCG(S^3,L)$. The symmetry group I want is the image of that group under the homomorphism $MCG(S^3,L) \rightarrow MCG(S^3) \times MCG(L)$, which is the 16 element group above. So another way to ask the question would be: can this homomorphism ever be surjective?

share|improve this question
you might also want to add "non-split" to the hypotheses, to avoid the example of a split union of two copies of a reversible, amphicheiral knot. –  Paul Mar 18 '12 at 4:21
do you mean its symmetry group surjects a subgroup of this group? Otherwise, I don't see why it couldn't have larger symmetry group (e.g. Hopf link has a T^2 subgroup of its symmetry group in S^3). –  Ian Agol Mar 18 '12 at 12:29
I edited to clarify these points. Thanks! –  Jason Cantarella Mar 18 '12 at 13:43
add comment

2 Answers

Hi Jason,

I think there is a non-hyperbolic link that does the job.

The link that I'm thinking of could be called the splice of two Bing doubles of a figure-8 knot. Another way to describe this link is to start with this 4-component link: Thistlethwaite's link and splice two figure-8 knots against it. You'd splice the knots against the red and purple components, respectively.

(above image taken from Morwen Thistlethwaite's homepage)

Below is a picture of the link I'm referring to:

symmetric link

The JSJ-decomposition of the complement of this link has 3 incompressible tori, separating the link complement into two Borromean ring complements, and two figure-8 knot complements. There is an orientation-reversing diffeomorphism of $S^3$ that preserves the link, since the Borromean rings admit one, and the figure-8 complements do, too.

As you can see from the way I've drawn it, there is a symmetry that permutes the two components, preserving the orientation of $S^3$, and from the Thistlethwaite diagram you can see a symmetry that reverses the orientation of either component, also preserving the orientation of $S^3$ -- this needs that the figure-8 knot is invertible.

share|improve this answer
add comment

This is more a comment on Ryan's nice answer than an independent answer itself, but it's too long for a comment box.

Let $K$ be an oriented knot in the solid torus which is isotopic (in the solid torus) to its reverse. [Edit: It should also be amphichiral, as well as invariant under the map of the solid torus which reverses the $S^1$ factor.] Let $L$ be a two component unoriented framed link in $S^3$ which is isotopic to its mirror image (without exchanging the two components) and also has an isotopy which exchanges the two components. Then replacing each component of $L$ with a copy of $K$ yields a two component oriented link in $S^3$ which has the full order 16 symmetry mentioned in the question.

In Ryan's example, $K$ is what I would call a Whitehead double with a funky figure-8 clasp, and $L$ is the Hopf link.

It's relatively easy to think of $K$ and $L$ satisfying the above conditions, so one can construct a large family of examples this way.

share|improve this answer
Thank you for the helpful comments! I thought of something like this, but I didn't think it was easy to come up with K satisfying those three conditions in the solid torus. I completely believe that this will work if you can come up with such a K, but a question for both you and Ryan is this: how do you come up with (families of) such K? –  Jason Cantarella Mar 19 '12 at 14:25
add comment

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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