1

I'm particularly interested in the case where $n=2$. The definition of $\mathbb RP^n$ as a manifold maps the point $(x_1,\ldots,x_n)$ in $\mathbb{R}^n$ by means of the function $\frac{x_i}{x_j}$ (where $i\ne j$) though it isn't specific about the range of indexes for the denominator.

So in the case of a point $p=(x,y)$ in $\mathbb R^2$ such that $(x,y)\ne (0,0)$, using the mapping we'd have $(x/y)$ and(/or?) $(y/x)$. The trouble here is that points that lie on either the x-axis or the y-axis result in division by zero.

Would a better mapping be $(x/r)$, $(y/r)$ where $r = \sqrt{x^2+y^2}$. As long as $(x,y)\ne(0,0)$, this avoids division by zero. And, for any non-zero real number $a$, it maps $(ax,ay)$ to $\frac{ax}{ar}$, and $\frac{ay}{ar} = (\frac{x}{r},\frac{y}{r})$. So am I missing something here?

Thomas Andrews
  • 177,126
Mr X
  • 907
  • 1
    Two points $(x_1,y_1)$ and $(x_2,y_2)$ are in the same projective equivalence class if $x_1y_2 = y_1x_2$. This avoids the problem of dividing by zero. – Paul Orland Mar 09 '13 at 00:14
  • 1
    I've edited your question to use $\LaTeX$. Please make sure it still represents your original intent. Specifically, I wasn't sure what you meant by $RP^n$, so I left it basically as-is. Also, the last paragraph was a little difficult to edit, so see if I did that right. :) For help with formatting in the future, please see this meta question. – apnorton Mar 09 '13 at 00:32
  • First of all, $\mathbb RP^n$ is the image of $\mathbb R^{n+1}\setminus 0$, not of $\mathbb R^n$. In particular, you are defining $\mathbb RP^1$, not $\mathbb RP^2$ – Thomas Andrews Mar 09 '13 at 00:33
  • anorton, your editing is correct. Thanks! – Mr X Mar 09 '13 at 00:33
  • More generally, it is easier to define $\mathbb RP^n$ in terms of equivalence classes, not fractions (where division by zero is an issue.) $(x_0,\dots,x_n)\sim (y_0,\dots,y_n)$ if there is a $\lambda\in\mathbb R\setminus{0}$ such that $x_i=\lambda y_i$ for all $i$. – Thomas Andrews Mar 09 '13 at 00:37
  • anorton can you put in the lateX codes there? Sorry but I'm still learning it. :p – Mr X Mar 09 '13 at 02:24
  • Thomas Andrews, what I'm asking for is a function that maps f: \mathbb{R}^n/(0) \rightarrow \mathbb{R}P^n-1 ;(\forall n \geq 2 that satisfies the property of f(\lambda x_{1},...,\lambda x_{n})=f(x_{1},...,x_{n}) ; \forall \lambda \in \mathbb{R} \ neq {0} Thus mapping antipodal and colinear points(on any line through the origin)to the same point in the projective space. – Mr X Mar 09 '13 at 02:31

0 Answers0