4

What is an HNN extension? What would be some elementary, intuitive examples of them and what exercises involving them would you suggest?

The Wikipedia definition is easiest to get to, since neither indexes of Magnus et al. nor Johnson's "Presentation$\color{red}{s}$ of Groups (Old Version)" indicate where they are.

Here it is for convenience:

Let $G$ be a group with presentation $G=\langle S\mid R\rangle$, and let $\alpha: H\to K$ be an isomorphism between two subgroups of $G$. Let $t$ be a symbol not in $S$, and define

$$G\ast_\alpha=\langle S, t\mid R, tht^{-1}=\alpha(h)\forall h\in H\rangle.$$

The group $G\ast_\alpha$ is called the HNN extension of $G$ relative to $\alpha$. The original group $G$ is called the base group for the construction, while the subgroups $H$ and $K$ are the associated subgroups. The new generator $t$ is called the stable letter.

The definition in Lyndon & Schupp's "Combinatorial Group Theory", despite the concept being mentioned a few times prior to it, is on page 179. It's very similar:

Let $G$ be a group, and let $A$ and $B$ be subgroups of $G$ with $\phi: A\to B$ an isomorphism. The HNN extension of $G$ relative to $A$ and $B$ and $\phi$ is the group

$$G^\ast=\langle G, t\mid t^{-1}at=\phi(a), a\in A\rangle.$$

The group $G$ is called the base of $G^\ast$, $t$ is called the stable letter, and $A$ and $B$ are called the associated subgroups.

The definition in Baumslag's "Topics in Combinatorial Group Theory", page 66, reads

Definition 2: Let $$B=\langle X\mid R\rangle$$ be a group given by a presentation and suppose $U$ and $V$ are subgroups of $B$ equipped with an isomorphism

$$\tau: U\stackrel{\sim}{\longrightarrow} V.$$

Then we term

$$E=\langle X, t\mid R\cup\{ tut^{-1}=u\tau\}\rangle\quad(\text{where }t\notin X)$$

an HNN extension of $B$ with stable letter $t$, associated subgroups $U$ and $V$ and associating isomorphism $\tau$.

These are all well & good. I can see how they are equivalent quite readily.

The definition in Stillwell's "Classic Topology and Combinatorial Group Theory (Second Edition)" is on page 286 and is much different. I shan't copy it down, for it's quite lengthy, but, again, I can sort of see what it's getting at. There are three exercises on HNN extensions starting on that page too, now that I've looked at the dreaded topology book, so I'll give them a go now.

If you have anything more to add to help me and others understand this important concept, please do so as a comment or even an answer.

I still don't get the intuition behind'm and I know not of any tangible examples yet.

Please help :)

Shaun
  • 44,997
  • Why the downvote? – Shaun Apr 18 '19 at 17:37
  • 1
    Generally speaking, a math.SE post should have a specific question included in the title/body. "Please explain X to me" or "what's the intuition behind X" are not specific questions.

    That said, I can explain what I know in a comment: the idea of HNN extensions comes from the question of whether two abstractly isomorphic subgroups must be conjugate. This isn't true, but in the HNN extension $G'$, the isomorphism $\varphi:A\rightarrow B$ is forced to become $\varphi(a)=tat^{-1}$ for some $t$. Thus $B=tAt^{-1}$ in the HNN extension.

    – Ehsaan Apr 18 '19 at 17:37
  • 5
    This seems like an eminently reasonable question to me! – Rylee Lyman Apr 18 '19 at 17:39
  • 1
    Lyndon and Schupp are better than Magnus, Karrass and Solitar for HNN-extensions. The difference is that Magnus et. al emphasised free products with amalgamation, and only mention HNN-extensions briefly. This is wrong. I met Schupp last year and his claim (which I completely agree with!) is that HNN-extensions are the "stronger" of the two constructions and deserved of more study. His example was the Magnus hierarchy for one-relator groups. However, the point is illustrated amazingly well the proof of the normal form theorem for free products with amalgamation in Lyndon and Schupp. Read it. – user1729 Apr 18 '19 at 17:40
  • (The proof uses the normal form theorem for HNN-extensions.) – user1729 Apr 18 '19 at 17:42
  • 2
    It's just a way to define a bigger group where $\phi$ is the restriction of an inner automorphism of the bigger group. – Thomas Andrews Apr 18 '19 at 17:42
  • Are you familiar with basic algebraic topology? Like $\pi_1$ stuff –  Apr 18 '19 at 17:45
  • I wrote up an answer here which you might find useful. – user1729 Apr 18 '19 at 17:46
  • I'm afraid not, @PaulPlummer, but by all means share what you have in mind. Somebody else might benefit. I'll catch up topology-wise someday. – Shaun Apr 18 '19 at 17:47
  • 1
    nLab's entry (https://ncatlab.org/nlab/show/HNN-extension) is similar to @ThomasAndrews comment –  Apr 18 '19 at 17:49

1 Answers1

6

Here's a few bits and bobs.

1) HNN extensions generalize semidirect products with $\mathbb Z$: Take $A$ to be $G$ and $\phi$ to be an automorphism of $G$. Then, $G\ast_\phi = G\rtimes_\phi \mathbb Z$

2) The inclusion of $G$ into any HNN extension $G\ast_\alpha$ is injective, so we can think of the HNN extension as "adding something" on top of our group.

3) Moreover, HNN extensions make more things conjugate in $G\ast_\alpha$ than were in $G$. This is an interesting property to play with!

4) Some of my favourite examples (some of which are downright strange):

  • The Baumslag-Solitar groups $BS(n,m) = \langle a, t \mid ta^nt^{-1} = a^m\rangle$ (which include $\mathbb Z\oplus \mathbb Z$ as $BS(1,1)$ and the Klein bottle group as $BS(1,-1)$).

  • Some automorphisms of free groups can be rewritten as HNN extensions over $\mathbb Z \oplus \mathbb Z$, for example the automorphism $a \mapsto a$ and $b \mapsto ba$ yields the HNN extension (with stable letter $b$!) $F_2\rtimes\mathbb Z = \langle a,b, t \mid tat^{-1} = a, tbt^{-1} = ba\rangle = \langle a, t, b \mid [a,t], btb^{-1} = at\rangle = (\mathbb Z \oplus \mathbb Z)\ast$

  • Thompson's group $F$ is an HNN extension of itself, despite admitting a finite presentation!

Rylee Lyman
  • 3,930