Consider the following subset of $\mathbb{R}^{3}$ \begin{equation} C=\{(x,y,z)\in\mathbb{R}^{3}\:|\:0\leq x\leq 1,\:0\leq y\leq 1,\:z=x^{2}+y^{2} \}. \end{equation} Intuitively, this looks like a differentiable $2$-manifold with boundary, the boundary being the points at $z=1$.

I am looking to parametrize $C$ using the least possible charts. I think that I'll need two.

I have tried over and over to find such charts, but since I have to get the boundary, the problem got a little hard. I could come up with something like (for most of the manifold, missing a point) \begin{equation} f_{1}\colon[0,1)\times[0,1)\to\mathbb{R}^{3}\qquad\qquad f_{1}(u,v)=(1-u,\:1-v,\:1-u^{2}-v^{2}) \end{equation}

But this one has the issue that $[0,1)\times[0,1)$ is not an open set of $\mathbb{H}^{2}=\{(x,y)\in\mathbb{R}^{2}\:y\geq 0\}.$

Any tips? Thank you all.


I feel like my failure could be in assuming that I could do that with only two charts. Since the boundary is kind of circular, it doesn't really look like $\mathbb{H}^{2}$.

  • $\begingroup$ I do not think your boundary is correct. There should be three parts and the one you already have is incorrect. $\endgroup$ – Jean-François Gagnon Nov 12 '15 at 2:03
  • 1
    $\begingroup$ I am very doubtful that this is a manifold with boundary. It looks like a manifold with corners to me. $\endgroup$ – user99914 Nov 12 '15 at 2:21
  • $\begingroup$ @Jean-FrançoisGagnon Yes, as I wrote in the post, $f_{1}$ is not a chart since the set is not open. What should the boundary be? $\endgroup$ – Shoutre Nov 12 '15 at 2:27
  • $\begingroup$ @JohnMa What is a manifold with corners? Why you do not think it has boundary? It looks like neighborhoods of points at height $z=1$ locally resemble $\mathbb{H}^{2}$. $\endgroup$ – Shoutre Nov 12 '15 at 2:28
  • $\begingroup$ It is a bit unclear what you are referring to. At $z = 1$, some $(x, y) \in C$ such that $x^2 + y^2 =1$ are lying in the interior: for example $(1/\sqrt 2, 1/\sqrt 2)$. $\endgroup$ – user99914 Nov 12 '15 at 2:31

One chart is enough. Take $I=[0,1]$ and $$f:I^2\rightarrow\mathbb{R}^3:(u,v)\mapsto (u,v,u^2+v^2)$$.

  • $\begingroup$ This set $I^{2}$ is not open neither in $\mathbb{R}^{2}$ nor in $\mathbb{H}^{2}$. $\endgroup$ – Shoutre Nov 12 '15 at 2:24
  • $\begingroup$ Can't you parameterize manifolds with boundary with sets with boundary? $\endgroup$ – Jean-François Gagnon Nov 12 '15 at 2:34
  • 1
    $\begingroup$ I believe that the standard definitions ask for the domains of the charts to be open in $\mathbb{H}^{n}$ or $\mathbb{R}^{n}$, where $\mathbb{H}^{n}$ is given the subspace topology. The set $[0,1)$, for instance, is open in $\mathbb{H}^{1}$, because $[0,1)=(-1,1)$$\cap$$\mathbb{H}^{1}$. And you can tell intuitively that the left side of the set $[0,1)$ looks like the "end" of a "curve with boundary". So open sets of $\mathbb{H}^{n}$ ,even not always they look open in $\mathbb{R}^{n}$ . Boundary is a bad word because it has also a topological meaning which is not the same as the one I mean here. $\endgroup$ – Shoutre Nov 12 '15 at 2:43
  • $\begingroup$ OK so you could use 4 charts (not optimal) for the atlas. Each one with a "boundary" and cover from the outside to almost the opposite side? $\endgroup$ – Jean-François Gagnon Nov 12 '15 at 16:55
  • $\begingroup$ Of course I would like to delete my answer given this information. But on my cellphone this is not doable. $\endgroup$ – Jean-François Gagnon Nov 12 '15 at 16:58

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.