I try to understand the proof of the embedding theorem:
"An $m-$dimensional differentiable manifold can be embedded as a closed subset of the Euclidean space $\Bbb R^n$, if $2m<n$."
The proof is given as follow: We know that we can find for a manifold a proper function $f:M^m \to \Bbb R^n$ for an $n>0$ and that we can approximate this by an injective immersion $g:M^m \to \Bbb R^n$, so that $\vert g - f \vert \leq 1$ and $A= \emptyset$. If $K \subset \Bbb R^n$ is compact, then $K \subset K(r)$ for some radius $r$, hence $g^{-1}(K)$ is closed in the compact set $f^{-1}(K\overline{(r+1)})$, hence compact. Therefore $g$ is proper, hence an embedding.
Now I have some questions in this proof:
Why we have that if $K \subset \Bbb R^n$ is compact, then $K \subset K(r)$ for some radius $r$?
Why it follows that $g^{-1}(K)$ is closed in the compact set $f^{-1}(K\overline{(r+1)})$?
Why we can now conclude that a proper is an embedding?
EDIT:
- I guess, he claims that $g^{-1}(K) \subset f^{-1}(K\overline{(r+1)})$ to conclude that $g^{-1}(K)$ is compact. But Why is this inclusion right? I guess that it comes from the fact, that $\vert g - f \vert \leq 1$. But I am not sure.
Many thanks for your help!