Ma · Mathematics

What Restrictions Does a Knot Complement’s Fundamental Group Satisfy?

Published By the 9to5Science editorial team

At a glance

  1. 1. Fix the setting

    Take a knot K in the three-sphere and its exterior, formed by removing an open tubular neighborhood.

  2. 2. Check the abelianization

    The group’s abelianization is always the infinite cyclic group Z.

  3. 3. Check meridians

    A meridian normally generates the whole group; Wirtinger presentations make this visible.

  4. 4. Separate general from special

    The unknot has group Z; every nontrivial knot has a nonabelian group.

Advertisement

Let KK be a knot in S3S^3, and let X=S3∖int⁡(N(K))X=S^3\setminus\operatorname{int}(N(K)) be its exterior, where N(K)N(K) is a tubular neighborhood of the knot. The knot group is π1(X)\pi_1(X). Using the exterior rather than deleting just the knot itself gives a compact space with torus boundary and the same fundamental group.

Several strong restrictions hold for this group. The most accessible ones come from homology and from the standard Wirtinger presentation. They are necessary properties of knot groups, not a complete test for whether an abstract group comes from a knot.

The abelianization is always Z\mathbb{Z}

For every knot, the first homology group of its exterior is

H1(X;Z)≅Z.H_1(X;\mathbb{Z})\cong\mathbb{Z}.

The abelianization of a fundamental group is its first homology group, so

π1(X)/[π1(X),π1(X)]≅Z.\pi_1(X)\big/\big[\pi_1(X),\pi_1(X)\big]\cong\mathbb{Z}.

In particular, a knot group is never perfect: it has a nontrivial abelian quotient. A generator of this infinite cyclic homology is represented by a meridian, a small loop that goes once around the knot.

One way to see the homology calculation is Alexander duality: the complement of a circle embedded in S3S^3 has first homology Z\mathbb{Z}. Equivalently, in a Wirtinger presentation, all the arc generators become equal after abelianizing the crossing relations, leaving one infinite cyclic generator.

A meridian normally generates the group

A meridian does more than generate the abelianization. Its normal closure is the entire knot group. If mm is a meridian, this says

⟨ ⁣⟨m⟩ ⁣⟩=π1(X),\langle\!\langle m\rangle\!\rangle=\pi_1(X),

where the left side consists of products of conjugates of mm and m−1m^{-1}.

The Wirtinger presentation explains why. Choose a diagram of the knot and assign a generator to each arc; each generator is represented by a meridian. At a crossing, the relation says that one arc meridian is conjugate to another (with the precise conjugation depending on the crossing convention). Since the diagram follows a single component, these relations express all arc generators as conjugates of one chosen meridian. Thus the normal closure of that meridian contains every generator.

This is normal generation, not necessarily ordinary generation. A meridian by itself usually does not generate the whole, generally nonabelian group.

Finite presentation and torsion

A knot diagram with finitely many arcs gives a finite Wirtinger presentation. For a diagram with nn arcs, it has nn meridional generators and crossing relations, with one relation redundant; in the usual presentation one can use n−1n-1 relations. This is a useful structural constraint, though different diagrams and presentations can give very different-looking presentations of the same group.

Knot groups are also torsion-free: no nonidentity element has finite order. This follows from the 3-manifold structure of the knot exterior, rather than just from the abelianization or the Wirtinger relations. The group is not, in general, abelian.

Example: the trefoil group

One presentation of the trefoil knot group is

G=⟨a,b∣aba=bab⟩,G=\langle a,b\mid aba=bab\rangle,

where aa and bb are meridional generators. In the abelianization, write the group operation additively. The relation becomes

2a+b=a+2b,2a+b=a+2b,

so a=ba=b. There are no further relations forcing this common generator to have finite order, and the abelianization is Z\mathbb{Z}, as it must be for any knot.

To verify that this group is nonabelian, map aa to the transposition (12)(12) and bb to (23)(23) in the symmetric group S3S_3. Both sides of the defining relation map to (13)(13), so this is a valid homomorphism. Its image contains (12)(12) and (23)(23) and is all of S3S_3, which is nonabelian. Therefore the trefoil group itself cannot be abelian.

By contrast, the unknot exterior is a solid torus, so its fundamental group is Z\mathbb{Z}. A standard theorem in knot theory says that a knot group is abelian exactly when the knot is the unknot. Thus an abstract group isomorphic to Z\mathbb{Z} cannot be the group of a nontrivial knot.

What these restrictions do—and do not—tell you

The conditions above quickly rule out many proposed groups: for example, a finite group, a group with trivial abelianization, or a group that cannot be normally generated by one element cannot be a knot group. But satisfying these tests does not prove that a group is a knot group. There are additional constraints arising from the topology of the exterior and from how the meridian and longitude sit on its boundary torus.

Also, the group alone does not always identify the knot. The group together with distinguished peripheral information—especially the meridian and longitude—retains more information. The exterior itself determines the knot in S3S^3, but the bare abstract group need not record all of that structure.

Common mistakes

  • Confusing the group with its abelianization. Every knot group has abelianization Z\mathbb{Z}, but most knot groups are nonabelian.
  • Calling a meridian a generator. A meridian normally generates the group; that does not mean the group consists of its powers.
  • Treating the restrictions as a classification. A group meeting these necessary conditions is not automatically known to be a knot group.
  • Forgetting the unknot case. The boundary torus of the unknot exterior does not inject on fundamental groups, unlike the boundary of a nontrivial knot exterior; the unknot’s group is the exceptional cyclic case.

References

  1. Knots and Links — Dale Rolfsen
  2. Algebraic Topology — Allen Hatcher

Spotted an error? Send a correction. Read how we write and check articles in our editorial policy.