Let $S_+$ be the closed upper hemisphere of the $2$-sphere $S^2$. We can define an equivalence relation $\sim_+$ on $S_+$ as follows:

$x\sim_+ y\:\:\Leftrightarrow\:\:\begin{cases}x=x^\prime,&\text{ or}\\x,\,x^\prime\in\partial S_+\text{ and }x=-x^\prime.\end{cases}$

In other words, we identify antipodal points on the equator of $S_+$. We also have the equivalence relation $\sim$ on $S^2$ in which $x\sim y$ iff $x=y$ or $x=-y$. I want to show $S^2/\sim$ is homeomorphic to $S_+/\sim_+$. Here is what I have done so far:

First, we observe there is an inclusion $i:S_+\hookrightarrow S^2$ of the upper hemisphere $S_+$ into the whole $2$-sphere $S^2$. We also have the canonical maps to the quotient spaces given by $p:S_+\to(S_+)/\sim_+$ and $\pi:S^2\to S^2/\sim$. The map I am going to use is the map $\pi\circ i:S_+\to S^2/\sim$.

Now, we observe that any line through the origin intersects $S^2$ at two points. If we are in the upper hemisphere, then there is one unique intersection, while on the equator there are two points in the (closed) upper hemisphere. In $S^2/\sim$ we are identifying antipodal points, and for each pair of antipodal points, at least one belongs to $S_+$. It is therefore clear that $\pi\circ i$ is surjective. It is also continuous, since it is the composition of continuous functions.

Next, suppose $x\sim_+ y$, then either $x=y$, or $x,\,y$ are antipodal points on the boundary. In either case, $\pi\circ i(x)=\pi\circ i(y)$. Conversely, if $\pi\circ i(x)=\pi\circ i(y)$, then $x,\,y$ are antipodal points, or equal. Since both belong to the (closed) upper hemisphere, we therefore conclude $x\sim_+ y$. So $\pi\circ i$ respects $\sim_+$. By the universal property, there exists a bijection $\varphi:(S_+)/\sim_+\to S^2/\sim$. It therefore remains to show $\varphi$ has a continuous inverse. I suppose I could just do a similar argument, but starting from the map $q:S^2\to S_+$, but I thought it might be neater to show $\pi\circ i$ is an open map, i.e. sends open maps to open maps.

With this in mind, let $U$ be open in $S_+$. Then am I right in saying that $U$ is simply the open ball of a point of $S_+$ that does not lie on the boundary (the equator)? If so, then $i(U)=U$ is also open in $S^2$ and hence $\pi\circ i(U)$ is open in $S^2/\sim$. It is then certainly true that $\varphi$ is a homeomorphism. Is this all alright?


There's an inclusion map $i:S_+\to S^2$. This induces a continuous map $i':{S_+/\sim}\to {S^2/\sim}$. As you point out, this is bijective. But both sides are compact Hausdorff spaces, and a continuous bijection between compact Hausdorff spaces is necessarily a homeomorphism (prove that it's a closed map).

  • $\begingroup$ Ah yes, of course. That would be much nicer. As an aside, am I correct in saying $\pi\circ i$ is an open map? $\endgroup$ – BillScroggs Apr 11 '18 at 11:41
  • $\begingroup$ @BillScroggs It's an open map all right. It's not true that each open subset of $S_+$ is an open ball though. $\endgroup$ – Lord Shark the Unknown Apr 11 '18 at 11:48
  • $\begingroup$ Ah I see, do you have any suggestions for how I show it is an open map? I would like to have experience in showing these kind of maps are open or closed. $\endgroup$ – BillScroggs Apr 11 '18 at 12:28
  • $\begingroup$ @BillScroggs What am I saying? $\pi\circ i$ is not an open map, just as $i$ isn't. But it is a closed map. $\endgroup$ – Lord Shark the Unknown Apr 11 '18 at 12:52

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.