2

Let $X$ be the complex obtained by gluing two discs to $S^1$ via maps of relatively prime degree, say by $f:\partial D_1 \rightarrow S^1$ of degree $p$, and $g:\partial D_2 \rightarrow S^1$ of degree $q$.

I would like to show that $X$ is simply connected. (It is easy to see that $H_1(X) \cong \{e\}$ using cellular homology).

Pick points $x_i \in D_i - \partial D_i$, and let $U_i = X - x_i$.

Then $U_1$ is homotopy equivalent to the space formed by attaching a disc to $S^1$ via a degree $q$ map and $U_2$ is homotopy equivalent to the space formed by attaching a disc to $S^1$ via a degree $p$ map. We also have $U_1 \cap U_2 \simeq S^1$.

Then $\pi_1(U_1) \cong <a\ | \ a^q>, \pi_1(U_2) \cong <b \ | \ b^p >, \pi_1(U_1 \cap U_2) \cong < c >$.

By Van Kampen's theorem, we have

$\pi_1(X) \cong <a, b \ | \ a^q =e,\; b^p=e,\; b^p = a^q \ >$

since including $c$ into $U_1$ is $b^p$ and $c$ into $U_2$ is $a^q$.

This group is not trivial, but I am unable to see what's wrong (I think it's in the relations from inclusion.)

Thanks in advance!

Functor
  • 179
  • 1
    The group is trivial that you calculated. Use the Euclidean algorithm. – Cheerful Parsnip May 14 '18 at 04:39
  • @CheerfulParsnip isn't $aba$ a nontrivial word? – Functor May 14 '18 at 04:47
  • Oops, I didn't notice that your presentation isn't quite right. – Cheerful Parsnip May 14 '18 at 04:48
  • $a,b$ are the same here since they both generate the fundamental group of the intersection. – Cheerful Parsnip May 14 '18 at 04:49
  • @CheerfulParsnip But if we take a non-trivial loop in the boundary of disc ($c$ in the above notation) and include it into $U_2$, then I do not understand why $c \neq a^q$. Sorry for being dense! – Functor May 14 '18 at 04:55
  • One way to visualize this is as a circle (labeled $c$) which has two disks attached to it. Subdivide the boundary of the first disk into $p$ line segments labeled $c$. Subdivide the second disk into $q$ line segments labeled $c$. This shows you how to attach the disks. So the generator of the annular edge of one of the disks is actually $c^q$ and the other one is $c^p$. So you can conclude $c^p=e$ and $c^q=e$ since they bound disks. – Cheerful Parsnip May 14 '18 at 05:08
  • @CheerfulParsnip I agree, but isn't that adding the relation $a^q=b^p$ in the amalgamation? – Functor May 14 '18 at 16:14
  • No, it's more like $a^p=e,b^q=e,a=b$. – Cheerful Parsnip May 14 '18 at 18:19

1 Answers1

1

How do you know the group is nontrivial? (Just because a group has a non-trivial presentation doesn't imply the group itself is.)

Let's get the van Kampen right. The inclusion of $S^1$ into $U_1$ and $U_2$ induce the maps $c\mapsto a$ and $c\mapsto b$, rather than the powers of $a$ and $b$ you have. Thus, the amalgamated product is $\langle a,b\mid a^q=1,b^p=1,a=b\rangle\cong\langle a\mid a^q=1,a^p=1\rangle$.

Since $p,q$ are relatively prime, there are $c,d\in\mathbb{Z}$ such that $cp+dq=1$. Then, $a=a^{cp+dq}=(a^p)^c(a^q)^d=e$.

Kyle Miller
  • 19,353
  • Thank you! I agree that if $a=b$, then it's trivial. But why is that the case? – Functor May 14 '18 at 04:46
  • Hatcher explains in Algebraic Topology how the 1-skeleton gives the generators and how the 2-skeleton gives the relations. Make sure you understand the inclusions of the intersection correctly. – Kyle Miller May 14 '18 at 05:00