Let $f:X\to Y$ be a continuous map between metric spaces satisfying the Heine-Borel theorem. Show that $f$ is proper if the following condition holds:

For every sequence $x_n\in X$ such that $f(x_n)$ is bounded, $x_n$ is also bounded.

My definition of properness is that $f^{-1}(K)$ is compact for all compact set $K\subseteq Y$.

Try: Let $K\subseteq Y$ be compact. Then $f^{-1}(K)$ is compact if every sequence in $f^{-1}(K)$ has a convergent subsequence. Let $x_n\in f^{-1}(K)$. Then, $f(x_n)\in K$ which is compact, so there is a convergence subsequence of $f(x_n)$, i.e. $f(x_{n_k})$ converges for some $n_k$. Then, is $x_{n_k}$ bounded?

  • $\begingroup$ Were you asked to prove this, or to prove or disprove it? Hint: If $d$ is a metric on $X$ then there exists a metric $d'$ on $X$ which generates the same topology as $d$, and is such that $0\le d(x,y)\le 1$ for all $x,y\in X$. $\endgroup$ – David C. Ullrich Apr 30 '16 at 12:59
  • $\begingroup$ @DavidC.Ullrich maybe he uses the notion "bounded = contained in a compact set", which some texts use? Then the sequence $(f(x_n))$ is by the definition bounded. $\endgroup$ – Henno Brandsma Apr 30 '16 at 13:33
  • $\begingroup$ I think I am supposed to use the Heine-Borel theorem. $\endgroup$ – Raubeurght Hagh Apr 30 '16 at 14:02
  • $\begingroup$ Such spaces are called Borel compact and such spaces are exactly such that bounded sequences have convergent subsequences. $\endgroup$ – Henno Brandsma Apr 30 '16 at 14:12
  • 1
    $\begingroup$ @RaubeurghtHagh Now that you've added the condition that the space have the Heine-Borel property it's certainly a question about metric spaces. Also it's now true. In the original version there was no mention of H-B - if the original version were true (and if "bounded" meant bounded in the metric, as I assumed) it would show that every map from one metric space to another was proper, precisely because any metric is topolgically equivalent to a bounded metric (and the definition of "proper" is a topological property). $\endgroup$ – David C. Ullrich Apr 30 '16 at 14:34

Assuming "bounded" means totally bounded, i.e. contained in a compact set, or that $X$ is Borel compact, which also implies that bounded sets are totally bounded:

$K \subseteq Y$ compact, pick a sequence $(x_n)$ in $f^{-1}[K]$. Then $f(x_n)$ has a convergent subsequence in $K$, so there is some $y \in K$ and a subsequence $(f(x_{n_k}))$ that converges to $y$.

Then the $f(x_{n_k})$ sequence is bounded (using my interpreted definition of boundedness) so the $x_{n_k}$ are a bounded sequence so there is some convergent subsequence $(x_{n_{k_l}})$ in it that converges to some $x$. But by continuity of $f$, we know that $f(x_{n_{k_l}})$ converges to $f(x)$, but also to $y$ (as a subsequence of the sequence $(f(x_{n_k}))$. By unicity of limits (in metric spaces certainly) we have $f(x) = y$, so $x \in f^{-1}[K]$. This shows that the original sequence from $f^{-1}[K]$ has a convergent subsequence in $f^{-1}[K]$, so $f^{-1}[K]$ is compact. So $f$ is proper.

The reverse: suppose $f$ is proper. Let $(x_n)$ be a sequence such that $f(x_n)$ is bounded (hence totally bounded, so contained in a compact set $K$ (its closure will do)). Then $x_n$ all lie in $f^{-1}[K]$ which is compact, so $(x_n)$ is bounded.

  • $\begingroup$ Thanks! For example, are manifolds Borel compact? $\endgroup$ – Raubeurght Hagh Apr 30 '16 at 14:13
  • $\begingroup$ @RaubeurghtHagh It depends. If they're embeddable as closed sets in Euclidean spaces, they are, I think, giving them the inherited metric. But there are non-metrisable manifolds, but not all people consider these "manifolds". So your definition matters here. $\endgroup$ – Henno Brandsma Apr 30 '16 at 14:16
  • $\begingroup$ If I understand correctly: the definition of Borel compact depends not just on the topology but on the metric itself. Thus, my problem is true if there exists a Borel compact metric on $X$ such that $f$ satisfies the given condition. $\endgroup$ – Raubeurght Hagh Apr 30 '16 at 14:20
  • $\begingroup$ In other words, Borel compact is not a well-defined notion of metrizable spaces, but only of metric spaces? $\endgroup$ – Raubeurght Hagh Apr 30 '16 at 14:24
  • $\begingroup$ Yes, as boundedness depends on the metric. $\endgroup$ – Henno Brandsma Apr 30 '16 at 14:26

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.