3
$\begingroup$

Consider $\mathbb{R}P^n$ as the quotient space of $S^n$ with antipodal points identified. Prove that $\mathbb{R}P^n$ is a manifold of dimension $n$.

(I'd like to clarify that I've seen the solution to this exersice when we see $\mathbb{R}P^n$ as the quotient of $\mathbb{R}^{n+1}$ with lines identified. I'd like to know if the same answer would solve the problem when it is quotient of $S^n$.)

I already proved $\mathbb{R}P^n$ is $T_2$ and second countable. By doing the same thing when $\mathbb{R}P^n$ is quotient of $\mathbb{R}^{n+1}$, I consider the open set:

$$V_i=\{x\in S^n:x_i\neq 0\},\quad \quad i=1,...,n+1,$$

and $$F_i:V_i\to\mathbb{R}^n,\quad F_i(x_1,...,x_{n+1})=\dfrac{1}{x_i}(x_1,..,x_{i-1},x_{i+1},...,x_n).$$

And then one should prove that $\phi_i:\pi(V_i)\to\mathbb{R}^n$ given by $\phi_i(\pi(x))=F_i(x)$ is a homeomorphism (where $\pi$ is the projection, which is open).

I already proved $\phi _i$ is injective and continuous, but I can't prove that it is surjective. If we take any $(x_1,..,x_n)\in\mathbb{R}^n$, the natural choice would be: $$F_i(x_1,...,x_{i-1},1,x_{i},...,x_n)=(x_1,...,x_n).$$

But $(x_1,...,x_{i-1},1,x_{i},...,x_n)$ need not be in $S^n$. Also, what would be $\phi^{-1}$? (In order to prove the inverse is continous...)

Or maybe $\phi_i$ is not surjective and we need another function. Any help?

Thank you.

$\endgroup$

1 Answer 1

2
$\begingroup$

$\newcommand{\Reals}{\mathbf{R}}\newcommand{\Proj}{\mathbf{P}}$Note that $V_{i}$ consists of two open hemispheres, so $F_{i}$ isn't injective, but $2$-to-$1$. (Specifically, $F_{i}(-x) = F_{i}(x)$ for all $x$ in $V_{i}$. Consequently, $F_{i}$ induces the mapping $\phi_{i}:U_{i} = \{[x] \in \Reals\Proj^{n}: x_{i} \neq 0\} \to \Reals^{n}$, which is continuous and injective, as you've shown.)

To work with the sphere, it may be convenient to introduce the hemispheres $$ V_{i}^{\pm} = \{x \in S^{n} : \pm x_{i} > 0\}. $$

An inverse $G_{i}^{\pm}:\Reals^{n} \to V_{i}^{\pm}$ of $F_{i}^{\pm}:V_{i}^{\pm} \to \Reals^{n}$ is $$ G_{i}^{\pm}(x_{1}, \dots, x_{n}) = \pm\frac{(x_{1}, \dots, x_{i-1}, 1, x_{i+1}, \dots, x_{n})} {\|(x_{1}, \dots, x_{i-1}, 1, x_{i+1}, \dots, x_{n})\|}. $$ Surjectivity of $\phi_{i}$ follows from invertibility of $F_{i}^{\pm}$.

$\endgroup$
1
  • 1
    $\begingroup$ Oh, thank you so much. It works perfectly. $\endgroup$
    – JonSK
    Feb 27, 2016 at 22:12

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.