Is there an example of a bounded not closed set $A\subset \mathbb R^n$ for some $n\in \mathbb N$ and an isometry $f:A\rightarrow A$
($\forall x,y\in A\space;\space d(x,y)=d(f(x),f(y))$ where $d$ is the euclidean metric) such that $f$ isn't surjective?

- 1,165
- 8
- 18
-
A simple example: consider the function $f : [0, + \infty) \rightarrow [0, + \infty)$ given by $f(x) = x + 1$. – Pedro Apr 13 '15 at 19:23
-
1Your set is unbounded and closed--he wanted bounded and not closed. – Umberto P. Apr 13 '15 at 19:24
2 Answers
Yes. Let $n=2$, so $\mathbb R^n\cong \mathbb C$ and let $f$ be multiplication with $c:=\frac 45+\frac35 i$. Let $A=\{\,c^n\mid n\in\mathbb N_0\,\}$. (The key point is that $c^n\ne 1$ for all $n$, so many different choices of $c$ with $|c|= 1$ work as well)

- 374,180
The idea here is that you need some way to "march off to infinity" without actually going to infinity. We can't do this with $n=1$ (no bounded set can map to itself via an ambient isometry), so let's turn to $n=2$ and irrational rotations. Let $$A = \{e^{i\alpha n}\ | n\in\mathbb{N},\alpha/2\pi\notin\mathbb{Q}\}.$$
Multiplication by $e^{i\alpha}$ is an isometry carrying $A$ to itself which is not surjective, since $1$ is in $A$ but is not in the image because the rotation is irrational.
You can find a more exotic example in the proof of the Banach-Tarski paradox, which as I recall involves constructing an isometry of the sphere to a proper subset of itself.

- 32,659
-
This is great answer and I upvoted it, I am just a little perplexed by the statement "with $n=1$, no bounded set can map to itself via an ambient isometry". Do you mind explaining? Thank you. – Giuseppe Negro Jan 09 '19 at 14:39