Difference between revisions of "User:Harold/Charting RP^n"
Line 35: | Line 35: | ||
** In general the transition maps have the form: {{M| (\phi_j \circ \phi_i^{-1}) : \phi_i(U_i \cap U_j) \to \phi_j(U_i \cap U_j)}}. | ** In general the transition maps have the form: {{M| (\phi_j \circ \phi_i^{-1}) : \phi_i(U_i \cap U_j) \to \phi_j(U_i \cap U_j)}}. | ||
** We obtain the transition map by [[case analysis]], as follows: | ** We obtain the transition map by [[case analysis]], as follows: | ||
− | **# {{M|i\eq j}} - in this case, the transition map is the identity map {{M| | + | **# {{M|i\eq j}} - in this case, the transition map is the identity map {{M| (\phi_i \circ \phi_i^{-1}): \underbrace{\phi_i(U_i \cap U_i)}_{\eq \R^n} \to \underbrace{\phi_i(U_i \cap U_i)}_{\eq \R^n} }} given by {{M| (\phi_i \circ \phi_i^{-1}) : x \mapsto x }}. Note that this is just the identity, i.e., we have {{M| (\phi_i \circ \phi_i^{-1}) \eq \id_{\R^n} }}. |
**# {{M|i<j}} - we obtain the following map: | **# {{M|i<j}} - we obtain the following map: | ||
**#* {{MM| \phi_j \circ \phi_i^{-1} : \phi_i(U_i \cap U_j) \to \phi_j(U_i \cap U_j)}} given by {{MM| (\phi_j \circ \phi_i^{-1})(x_0, \dotsc, x_{i-1}, x_{i+1}, \dotsc, x_n) \eq \frac{(x_0, \dotsc, x_{i - 1}, 1, x_{i+1}, \dotsc, x_{j-1}, x_{j+1}, \dotsc, x_n)}{x_j} }}. | **#* {{MM| \phi_j \circ \phi_i^{-1} : \phi_i(U_i \cap U_j) \to \phi_j(U_i \cap U_j)}} given by {{MM| (\phi_j \circ \phi_i^{-1})(x_0, \dotsc, x_{i-1}, x_{i+1}, \dotsc, x_n) \eq \frac{(x_0, \dotsc, x_{i - 1}, 1, x_{i+1}, \dotsc, x_{j-1}, x_{j+1}, \dotsc, x_n)}{x_j} }}. |
Latest revision as of 16:50, 19 February 2017
\newcommand{\R}{\mathbb{R}} \newcommand{\RPn}{\R P^n} \newcommand{\Ztwo}{\mathbb{Z} / 2 \mathbb{Z}} \newcommand{\id}{\mathrm{Id}} This article contains information on possible charts for the real projective space of dimension n, denoted by \RPn.
Definition of \RPn
We shall first define \RPn. Let S^n = \left\{ (x_0, \dotsc, x_n) \middle\vert \sum_{i = 0}^n x_i^2 = 1 \right\} be the n-sphere. Define a group action \{-1, 1\} \cong \Ztwo on S^n by mapping (\epsilon, x) \mapsto \epsilon x with \epsilon \in \{-1, 1\} and x \in S^n. This group action is "nice enough" so that the quotient space S^n / \left( \Ztwo \right) is actually a real smooth compact Hausdorff manifold.
Construction of the charts
We now construct (the) (smooth) charts on \RPn. First we introduce some notation: if x \in \RPn, we write x = [x_0 : \dotsc : x_n] if (x_0, \dotsc, x_n) is a representative of the equivalence class x. Define the subsets U_i \subset \RPn for 0 \leq i \leq n as U_i := \{ [x_0 : \dotsc : x_n] \in \RPn \vert x_i \neq 0 \}. This is well-defined, because the choice of representative only depends on a sign or a non-zero scalar multiple (if the definition of lines in \R^{n+1} is chosen; see Real projective space). Now introduce maps
\begin{align*} \phi_i: U_i & \to \R^n \\ [x_0 : \dotsc : x_{i - 1} : 1 : x_{i+1} : \dotsc : x_n] & \mapsto (x_0, \dotsc, \widehat{x_i}, \dotsc, x_n) \end{align*}
where \widehat{x_i} denotes that the i-th coordinate is omitted. These maps are well-defined, and homeomorphisms if one takes the quotient topology on \RPn, and actually define a smooth structure on \RPn, as the transition maps \phi_j \circ \phi_i^{-1} are diffeomorphisms (where defined).
On the transition maps
We obtain the following transition maps:
- Let i,j \in\{1,\ldots,n\}\subseteq\mathbb{N} be given.
- In general the transition maps have the form: (\phi_j \circ \phi_i^{-1}) : \phi_i(U_i \cap U_j) \to \phi_j(U_i \cap U_j).
- We obtain the transition map by case analysis, as follows:
- i\eq j - in this case, the transition map is the identity map (\phi_i \circ \phi_i^{-1}): \underbrace{\phi_i(U_i \cap U_i)}_{\eq \R^n} \to \underbrace{\phi_i(U_i \cap U_i)}_{\eq \R^n} given by (\phi_i \circ \phi_i^{-1}) : x \mapsto x . Note that this is just the identity, i.e., we have (\phi_i \circ \phi_i^{-1}) \eq \id_{\R^n} .
- i<j - we obtain the following map:
- \phi_j \circ \phi_i^{-1} : \phi_i(U_i \cap U_j) \to \phi_j(U_i \cap U_j) given by (\phi_j \circ \phi_i^{-1})(x_0, \dotsc, x_{i-1}, x_{i+1}, \dotsc, x_n) \eq \frac{(x_0, \dotsc, x_{i - 1}, 1, x_{i+1}, \dotsc, x_{j-1}, x_{j+1}, \dotsc, x_n)}{x_j} .
- i>j - we obtain the following map:
- \phi_j \circ \phi_i^{-1} : \phi_i(U_i \cap U_j) \to \phi_j(U_i \cap U_j) given by (\phi_j \circ \phi_i^{-1})(x_0, \dotsc, x_{i-1}, x_{i+1}, \dotsc, x_n) \eq \frac{(x_0, \dotsc, x_{j-1}, x_{j+1}, \dotsc, x_{i - 1}, 1, x_{i+1}, \dotsc, x_n)}{x_j} .
- This completes our case analysis
- Since i,j were arbitrary we have shown this for all.
- These explicit expressions makes it obvious that the transition maps are smooth, and they are obviously invertible, with smooth inverse. As such, they are diffeomorphisms from \phi_i(U_i \cap U_j) \to \phi_j(U_i \cap U_j).