Let P be the real projective plane obtained by identifying antipodal points on the unit sphere of $R^3$.

How to prove that P is nonorientable in a rigorous and elementary way? I do not want mere intuition.

My idea is to consider the closed curve $a(t)=(\cos t,\sin t, 0)$ , $0 \leq t \leq \pi$

Then the tangent vectors $a'(0)$ and $a'(\pi)$ are identical.

However, for a vector field V on $a(t)$ defined by $V(a(t))=(0,0,1)$, the tangent vectors at $a(0)$ and $a(\pi)$ differ by a sign. Therefore the normal vector cannot be continuously defined.

Is the constant vector field V continuous on the curve? Are my arguments right?


Preliminary Result
If $M$ is an orientable manifold and if $(U,\phi),(V,\psi)$ are two charts with $U,V$ connected, then the jacobian determinant of the change of coordinates $\psi\circ \phi^{-1}$ has constant sign (even if $U\cap V$ is not connected)

Let us apply this to projective space and to its two charts $$\phi : U=\{x\neq 0\}\to \mathbb R^2:[x:y:z]\mapsto (y/x,z/x)$$

$$\psi : V=\{y\neq 0\}\to \mathbb R^2:[x:y:z]\mapsto (x/y,z/y)$$
The change of coordinates $\psi\circ \phi^{-1}$ is the morphism $$\mathbb R^2\setminus \{u=0\}\to \mathbb R^2\setminus \{u=0\}:(u,v)\mapsto (\frac 1u,\frac vu)$$
whose jacobian determinant is $\frac {-1}{u^3}$.
Since this jacobian determinant changes sign on $\mathbb R^2\setminus \{x=0\}$, $\mathbb P^2$ is non-orientable as a consequence of the Preliminary Result.

The Preliminary Result seems not to be often mentioned in books on differential geometry.
It is a pity since as a consequence non orientability of manifolds is often treated in a hand-waving fashion.

  • $\begingroup$ Do you think that my arguments are correct? $\endgroup$ – noot Oct 14 '13 at 8:18
  • $\begingroup$ Dear noot, I'm not sure I understand it. $\endgroup$ – Georges Elencwajg Oct 14 '13 at 8:24
  • $\begingroup$ I wanted to show that there cannot be 2 form on the projective plane that is nonzero on any two linearly independent tangent vectors. I wanted to show such 2 form must be zero on two linearly indepent tangent vectors somewhere. $\endgroup$ – noot Oct 14 '13 at 8:34
  • $\begingroup$ But you didn't prove that every two-form vanishes on $\mathbb P^2$. Anyway, I wrote a proof answering the question in your second sentence and that's final as far as I'm concerned. I'll let other interested users discuss your arguments. $\endgroup$ – Georges Elencwajg Oct 14 '13 at 8:46

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.