2

I am having trouble with the following easy question: consider a quarter of an open disk. Is there a diffeomorphism between it and half of an open disk? I think not since this would make the square $[0,1]\times[0,1]$ a manifold with boundary.

The obvious map I could think of which takes a point and doubles the angle it makes with the $x$-axis, that is, $\begin{bmatrix} \cos(2(\arccos(\frac{x}{\sqrt{x^2+y^2}})& -\sin(2(\arccos(\frac{x}{\sqrt{x^2+y^2}})\\ \sin(2(\arccos(\frac{x}{\sqrt{x^2+y^2}})& \cos(2(\arccos(\frac{x}{\sqrt{x^2+y^2}})\end{bmatrix}$ fails to be continuous at the origin.

But, i don't know what the problem would be in general in finding such a map. Thanks.

added Quarter of an open disk is the open disk centred at the origin intersection the unit square.

Apoha
  • 145
  • No: you can't create (or remove) a sharp corner using a diffeomorphism. As you say, if they were diffeomorphic, then the quarter disc would be a manifold with boundary, so it suffices to show this is not the case. (I did this once, but my differential geometry skills are a bit rusty which is why I'm only posting this as a comment.) – Clive Newstead Jan 04 '13 at 12:33
  • By the way, an easier statement of your "obvious map" is the function $z\mapsto z^2$ defined on $\mathbb C$ (a.k.a. $\langle x,y\rangle\mapsto\langle x^2-y^2,2xy\rangle$). This also does some squashing of points, but this version is differentiable everywhere, although the Jacobian is $0$ at the origin. – Mario Carneiro Jan 04 '13 at 12:36
  • The vague idea I have is that suppose it was a smooth manifold then how does one go about defining a tangent space at the corner... But I don't know how to make this rigorous. – Apoha Jan 04 '13 at 12:40
  • 2
    What exactly is "a quarter of an open disk"? Are the straight parts of the boundary included in the set or not? If not, I don't see any problems with the origin. – Hans Lundmark Jan 04 '13 at 12:42
  • 1
    @HansLundmark Yes, the boundaries are included in the quarter disk. – Apoha Jan 04 '13 at 12:45
  • Isn't it sufficient to point out that the half-disk is a manifold-with-boundary and a quarter disk is not? Just use the definitions to prove these facts, and if you know that diffeomorphisms take manifolds-with-boundary to other manifolds-with-boundary, then you are done. – Mario Carneiro Jan 04 '13 at 12:49
  • @MarioCarneiro In fact i am trying to prove that the quarter disk is not a manifold with boundary. – Apoha Jan 04 '13 at 12:54

1 Answers1

2

Hint: Use the fact that the tangent space at any point of the half disk contains some vector $v$ together with its opposite $-v$; the "tangent space" at the corner of the quarter disk does not. Conclude that there cannot be a linear map between the two tangent spaces of full rank.

Willie Wong
  • 73,139