The real line $\Bbb R $ endowed with the metric $d(x,y)=\min(1,|x-y|)$ is a bounded metric space that isn't totally bounded.

I can easily prove that this is bounded. But to prove this is not totally bounded we have to find $P>0$ such that there is no finite cover of $P$ balls. Can someone help me to prove this?

  • $\begingroup$ I assume "$P$ balls" means balls of radius $P$. $\endgroup$ – Silvia Ghinassi Nov 23 '15 at 21:46
  • 1
    $\begingroup$ Use any $P\le 1$: in that case the $P$-balls in this metric are identical to those in the usual metric. $\endgroup$ – Brian M. Scott Nov 23 '15 at 21:47
  • $\begingroup$ Just take $P=\frac12$, or anything else smaller than $1$. $\endgroup$ – 5xum Nov 23 '15 at 21:47
  • $\begingroup$ @BrianM.Scott Don't you mean $<1$? $\endgroup$ – 5xum Nov 23 '15 at 21:47
  • $\begingroup$ @5xum: No, $1$ also works. If $x$ is in the open $1$-ball about $y$, then $|x-y|<1$. $\endgroup$ – Brian M. Scott Nov 23 '15 at 21:48

If $P\leq 1$, then $B(x,P)=\{y \in \mathbb R \mid |x-y|<P\}=(x-P,x+P)$ is an open interval centered at $x$ as if $d(x,y)<1$, as the metric $d$ becomes the usual metric on $\mathbb R$.

Since $\mathbb R$ is unbounded with the usual metric (while, as you have proved, it is bounded with this metric) we can see that we can't find finitely many balls of radius $P$ that cover the entire space.

In fact, if we could, that would mean that there exist $x_1,\dots,x_N$ such that $\mathbb R \subset \bigcup_{i=1}^N (x_i-P,x_i+P)$. But the length of the union of such intervals is at most $2NP$, while $\mathbb R$ has infinite length.

  • $\begingroup$ What i dont understand is that how can you tell R is unbounded with this metric ? It becomes the usual metric when d(x,y)<1 but it is not the usual metric $\endgroup$ – Razor1692 Nov 23 '15 at 22:11
  • $\begingroup$ You are right, it is not and $\mathbb R$ is bounded. But if you take $P<1$, then the balls of this metric are the same balls of the usual metric (that is, open intervals) so you can't cover the entire space only with finitely many. $\endgroup$ – Silvia Ghinassi Nov 23 '15 at 22:13
  • $\begingroup$ Can you tell me the way to prove R with this metric is complete? do we have to show that every cauchy sequence converges or is there an easier way? $\endgroup$ – Razor1692 Nov 23 '15 at 22:24
  • $\begingroup$ You should ask a new question, if you need help with that. Anyway, I would show that every Cauchy sequence converges. $\endgroup$ – Silvia Ghinassi Nov 23 '15 at 22:28
  • $\begingroup$ Thanks.I asked a new question anyway :) $\endgroup$ – Razor1692 Nov 23 '15 at 22:32

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.