4
$\begingroup$

Can anyone give an example of an injective map from $\mathbb{R}^2$ to $\mathbb{R}$? Clearly, such a map cannot be continuous (for instance by Borsuk-Ulam Theorem).

Thanks in advance.

$\endgroup$

4 Answers 4

11
$\begingroup$

How about this construction: Express $(x,y)\in\mathbb{R}^2$ in decimals ($x=\sum a_k 10^k$, $y=\sum b_k 10^k$) and define the image of $(x,y)$ as the real number which you obtain by interlacing the decimals (i.e. take $c_{2k} = a_k$ and $c_{2k+1} = b_k$).

$\endgroup$
3
  • $\begingroup$ Very good idea, thanks! $\endgroup$ Oct 23, 2011 at 14:36
  • 2
    $\begingroup$ I think a bit of care needs to be taken in that you'll want to take decimal expansions that don't end in a repeated trail of nines, otherwise $(1,1)$ will go to 1, but $(1,0.\overline{9})$ would go to $1.\overline{90}$. $\endgroup$
    – user5137
    Nov 4, 2011 at 20:31
  • 1
    $\begingroup$ I rolled back because: (1) this edit was far from being an actual edit. It was a comment at best, I personally found it content altering; (2) the notation in the post was that $(a,b)$ is an ordered pair, adding $(0,1)$ as an interval makes it harder and much less clearer. $\endgroup$
    – Asaf Karagila
    Jun 30, 2012 at 21:49
3
$\begingroup$

Hopefully this answer complements Dirk's direct answer. I use the fact that $\mathbb R$ has the same cardinality as $2^{\mathbb N}$ (and hence $\mathbb R^{2}$ has the same cardinality as $2^{\mathbb N} \times 2^{\mathbb N}$). Our goal, then, is to establish an injective function from $2^{\mathbb N} \times 2^{\mathbb N}$ to $2^{\mathbb N}$.

Given $S, T \subseteq \mathbb N$, consider the set: $$ R := \{2x \,:\, x \in S\}\ \cup \ \{ 2y+1 \,:\, y \in T \}. $$ Given the set $R$, it is easy to "decode" the sets $S$ and $T$. (Hint: Consider the odd and even elements of $R$.) Therefore the mapping $(S,T) \mapsto R$ as above is injective.

$\endgroup$
1
  • 5
    $\begingroup$ This map is also surjective, and continuous at least under the trivial topology :-) $\endgroup$
    – Asaf Karagila
    Oct 23, 2011 at 15:21
1
$\begingroup$

One can also consider space-filling curves. Since there's an onto map from $[0,1]$ to $[0,1]\times [0,1]$, it follows that $\vert[0,1] \vert \geq \vert [0,1] \times [0,1] \vert$. Of course, the other inequality clearly holds (via the mapping $x \mapsto (x,0)$), hence $\vert [0,1] \vert = \vert [0,1] \times [0,1] \vert$. Finally, since $\vert [0,1] \vert = \vert \mathbb{R} \vert$, we have $\vert \mathbb{R} \vert = \vert \mathbb{R}^2 \vert$.

$\endgroup$
0
$\begingroup$

let x_n be the n'th term any unique continued fractional expansion for x.

Then $z(a,b)=\sum (1/(2n)^{1/a_n}+1/(2n+1)^{1/b_n}$)

bonus:
By the Lindemann–Weierstrass theorem
$z=\sum (e^{-pi/a_n}+\pi^{-e/b_n}$)

$\endgroup$
2
  • 3
    $\begingroup$ I am unable to follow the example. Can you explain why this function is injective? $\endgroup$
    – Srivatsan
    Oct 23, 2011 at 14:38
  • $\begingroup$ @SrivatsanNarayanan because these quantities are linearly independent over the rationals $\endgroup$
    – GM2001
    Oct 23, 2011 at 14:53

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .