Hypothesis: Let $G$ and $H$ be defined in terms of the following presentations:

$$ G \cong \left\langle a,b \mid abab^{-1}\right\rangle $$

$$ H \cong \left\langle c,d \mid c^2 d^2 \right\rangle $$

Goal: Show that $G \cong H$.

Hint: Show that $G \cong \pi_1(K) \cong H$ where $\pi_1(K)$ denotes the fundamental group of the Klein Bottle.


(1) First I tried ignoring the hint and directly showing that $abab^{-1} = e$ if and only if $c^2d^2 = e$. Here all I could obtain is that $abab^{-1} = e \implies ba^{-1} b^{-1}a^{-1} = e$ as well as that $dc^{-1}d^{-1}cd^2 = e$

(2) Using the hint, all I know at this point in my studies is that the Klein bottle $K$ is formally defined as the quotient space of $[0,1] \times [0,1]$ arising from the equivalence relations

$$ (s,0) \equiv (s,1) $$


$$ (0,t) \equiv (1,1-t) $$

From here I'm not sure how to proceed.

  • $\begingroup$ draw the standard picture of the klein bottle as a quotient of the square (this will be $abab^{-1}$). now cut it up along the diagonal and piece it back together until you get what you want... $\endgroup$ – yoyo Mar 25 '14 at 1:50
  • $\begingroup$ What do you mean "this will be $abab^{-1}$"? $\endgroup$ – user125103 Mar 25 '14 at 1:58
  • $\begingroup$ @user125103 If you label the top and bottom sides $a$ and the left and right sides $b$, then traveling around the boundary is the path $abab^{-1}$. $\endgroup$ – André 3000 Mar 25 '14 at 3:46

You want to construct an isomorphism between $G$ and $H$. Essentially, send $a$ and $b$ to elements of $H$ in a nice way. After some fiddling, we try $a \mapsto cd$, $b \mapsto d^{-1}$. We see that $e = abab^{-1} \mapsto (cd)(d^{-1})(cd)(d^{-1})^{-1} = c^2d^2 = e$, so this is a reasonable candidate.

So it's a homomorphism, but we need to show it is an isomorphism. Clearly $ab \mapsto (cd)(d^{-1}) = c$, and $b^{-1} \mapsto (d^{-1})^{-1} = d$, so it's surjective. But showing injectivity is a pain. Informally, we can use the reasoning above to show that $abab^{-1} \mapsto c^2 d^2$, and those are both zero in their respective groups. In this case it works, but in more complicated presentations, that might not suffice.

If you want a more formal argument, consider these as quotients of the free group on two generators, $F_2 = \langle a,b \mid ~ \rangle$. We consider two subgroups, $K_1 = \langle abab^{-1} \rangle$ and $K_2 = \langle a^2 b^2 \rangle$. By definition, $G \cong F_2/K_1$, and $H \cong F_2/K_2$. We wish to show $F_2/K_1 \cong F_2/K_2$.

Let $\alpha : F_2 \to F_2$ send $a$ to $ab$, and $b$ to $b^{-1}$. This can be uniquely extended to a homomorphism by the universal property of the free group. Interestingly enough, it is its own inverse, so showing it is an isomorphism is easy: $$ \alpha(\alpha(a)) = \alpha(ab) = \alpha(a) \alpha(b) = abb^{-1} = a$$ $$ \alpha(\alpha(b)) = \alpha(b^{-1}) = \alpha(b)^{-1} = (b^{-1})^{-1} = b$$

It now suffices to show that $\alpha(K_1) = K_2$.* Let $x \in K_1$; it must be $(abab^{-1})^n$. So $$\alpha(x) = \alpha((abab^{-1})^n) = \alpha(abab^{-1})^n = ((ab)(b^{-1})(ab)(b^{-1})^{-1})^n = (a^2 b^2)^n$$ Reversing this easily shows the reverse containment.

*Because then we can take the natural projection $\pi : F_2 \to F_2 / K_2$, and create $\beta = (\pi \circ \alpha)$, which maps $F_2$ to $F_2 / K_2$, and has kernel $K_1$. This means $\beta$ factors through $K_1$, and we get a well-defined homomorphism $\gamma$. from $F_2 / K_1$ to $F_2 / K_2$. Using $\alpha$'s bijectivity, it's easy to show that $\gamma$ is also bijective.

  • $\begingroup$ On your first proof regarding injectivity: we know that $abab^{-1} \mapsto c^2d^2$, but how do we know that other elements of $G$ might not also map to $e$? (I'm assuming you're here using our homomorphism is injective iff its kernel is trivial). $\endgroup$ – user125103 Mar 25 '14 at 20:28
  • $\begingroup$ I'm not entirely sure how to pin down injectivity more rigorously in the first proof, which is why I typed up the second one. I suppose it comes down to the same idea though, that $\alpha$ maps $K_1$ to $K_2$ exactly. $\endgroup$ – Henry Swanson Mar 25 '14 at 20:54
  • $\begingroup$ Is there any direct connection between your second proof and $\pi_1(K)$? $\endgroup$ – user125103 Mar 25 '14 at 20:59
  • $\begingroup$ Not at all. I know what fundamental groups are, and can justify why certain manifolds have certain ones, but unfortunately, I've never done any rigorous proofs with them. :\ $\endgroup$ – Henry Swanson Mar 25 '14 at 21:20

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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