1

I've just read that the projective plane $\Bbb RP^2$ is a surface equivalent to the set of $(x,y,u,v,w) \in \Bbb R^5$ such that $$u^2=xy,v^2=x(1-x-y),w^2=y(1-x-y)$$

How is it true? I don't see how to get this result, even looking at questions like these : Give an explicit embedding from $\mathbb{R}P_2$ to $\mathbb{R}^4$, Injective map from real projective plane to $\Bbb{R}^4$, embedding projective plane in 4-space?.

Thank you!

Alphonse
  • 6,342
  • 1
  • 19
  • 48
  • Where did you read about this? – Jason DeVito - on hiatus Oct 25 '16 at 18:39
  • I saw it on slides (for explaining Poincaré's conjecture). The first page gives the link http://milnor.math.ucl.ac.be/plwiki/Olympiades2007/ but it seems to be dead now. – Alphonse Oct 25 '16 at 19:24
  • More precisely, I would like to know: where do these equations come from? If we look at this question, the equations are "proved" to be "correct"… But what about the ones here? The problem is that I don't know exactly what is meant by "equivalent". – Alphonse Oct 26 '16 at 05:26
  • 1
    I tried to "prove" the equations were correct using "obvious" choices of maps from $S^2$ to $\mathbb{R}^5$ (similarly to how was one in the questions you linked). After failing several times, I decided to find a reason I was failing, hence my answer. – Jason DeVito - on hiatus Oct 26 '16 at 12:55

1 Answers1

2

I'm not sure what is meant by "equivalent", but it is not as smooth manifolds. More specifically, the set $$X=\{(x,y,u,v,w): u^2 = xy, v^2 = x(1-x-y), w^2 = y(1-x-y)\}$$ is not a smoothly embedded submanifold of $\mathbb{R}^5$.

To see this, consider the point $p = \left(0,\frac{1}{2},0,0,\frac{1}{2}\right)\in X$. I claim that $T_p X$ contains the three vectors $(0,1,0,0,0)$, $(0,0,1,1,0)$ and $(0,0,1,-1,0)$. Thus, if $T_p X$ is a vector space, it's a least 3-dimensional. On the other hand, we will find a path $\alpha(t)$ in $X$ connecting $p$ to another point $q\in X$ where the tangent space is $2$-dimensional. Thus, on a single connected component of $X$ we find it's $2$ dimensional and at least $3$ dimensional, a contradiction.

So, let's argue that $T_p X$ contains those three vectors.

To do that, let $\gamma(t) = (0,t,0,0,\sqrt{t-t^2})$. Then one can easily verify that $\gamma\left(\frac{1}{2}\right) = p$, $\gamma(t)\in X$ for all $t$ it's defined, and $\gamma'\left(\frac{1}{2}\right) = (0,1,0,0,0)$.

For the other two, consider $\alpha_{\pm}(t) = \left( 2t^2, \frac{1}{2}, t, \pm t\sqrt{1-4t^2}, \sqrt{\frac{1}{4}-t^2}\right)$. Then $\alpha_{\pm}(0) = p$, and $\alpha_{\pm}(t)\in X$ whenever it's defined, and $\alpha'(0) = (0,0,1,\pm 1,0)$.

Finally, for either choice of $\alpha$, note that for small non-zero $t$, all components are non-zero. So $\alpha$ is a path in $X$ connecings $p$ to a point where are coordinantes are non-zero.

Now, one can compute the jacobian: One gets $\begin{bmatrix}-y & -x & 2u & 0 & 0\\ 2x + y-1 & x & 0 & 2v & 0 \\ y & 2y+x-1 & 0 & 0 & 2w\end{bmatrix}$. In particular, if $u$, $v$, and $w$ are all non-zero, this matrix has full rank. Thus, at such points, the tangent space of $X$ is $2$-dimensional.

  • I suspect that $X$ isn't even a topological manifold at $p$, so I think $X$ isn't homeomorphic to $\mathbb{R}P^2$ (or any other manifold), but I'm not sure how to prove it. – Jason DeVito - on hiatus Oct 25 '16 at 23:38
  • Thank you for your answer! I agree that the word "equivalent" they used is vague. But is there a way to "see" $\Bbb{RP}^2$ in $\Bbb R^5$ (maybe only as topological space) using these equations? I mean where do these equations come from? – Alphonse Oct 26 '16 at 05:24
  • 1
    @Alphonse:I don't know the MSE rules on bounties, but if there is some way you can save some of your reputation points by not awarding me, please do so. In my own opinion, I have not satisfactorily answered your original question. – Jason DeVito - on hiatus Nov 01 '16 at 17:59
  • I think that the reputations points will be lost… Even if your answer is very interesting (thank you for your effort!), I agree with you: I think that I won't award you the bounty. – Alphonse Nov 01 '16 at 19:07