2
$\begingroup$

How to decide if the metric spaces $((0,1)$, $d(x,y)=|x^2-y^2|)$ and $((-\frac{\pi}{2},\frac{\pi}{2})$, $d(x,y)=|\tan x-\tan y|)$ are complete or not?

For the first metric, I let any Cauchy sequence $(x_n)$ in $((0,1),d)$, then by definition I have for any small number $s>0$, there is $N>0$ such that for all $n,m>N$, $d(X_n,X_m) = |(X_n)^2-(X_m)^2| < s$. Then I have the sequence $((X_n)^2)$ being Cauchy in $((0,1)$, $d(x,y)=|x-y|)$. Then what?

$\endgroup$
  • 3
    $\begingroup$ This is a must read. $\endgroup$ – Julien Apr 16 '13 at 18:45
  • $\begingroup$ You have to figure out if such a Cauchy sequence would necessarily have a limit or not. $\endgroup$ – ferson2020 Apr 16 '13 at 18:54
3
$\begingroup$

Looks like you've plunged into epsilons, deltas and abstract reasoning from the very start. But it may be a good idea to simply check out one or two concrete sequences first, and see if you can come up with a counterexample to completeness right away.

Hint: if you have an open interval, weird things will usually happen near its endpoints.

| cite | improve this answer | |
$\endgroup$
  • $\begingroup$ Thank you for your hint. For the first metric, can I use (Xn)=1/n which is Cauchy in the metric but doesn't converge to anypoint in (0,1) beacuse 0 is not in the metric space to show that this metric is not complete? $\endgroup$ – Elvis Apr 17 '13 at 4:06
  • $\begingroup$ @Elvis Basically yes, although I would explain why $(x_n)$ diverges in different words, without mentioning $0$ explicitly. I have this feeling that whenever possible, properties of a metric space should be proved internally, i.e. using only points of that space. $\endgroup$ – Dan Shved Apr 17 '13 at 4:16
  • $\begingroup$ Cool, Thanks a lot. should the second metric space be complete? $\endgroup$ – Elvis Apr 17 '13 at 4:30
  • $\begingroup$ Hint for the second metric space: $((-\pi/2,\pi/2),d)$ is isometric to $(\mathbb{R}, d_{\mathbb{R}})$ where $d_\mathbb{R}$ is the standard distance in $\mathbb{R}$. $\endgroup$ – Dan Shved Apr 17 '13 at 4:33
  • $\begingroup$ For any s>0, there is N>0 such that for all n, m>N we have d(Xn,Xm)=|tan(Xn)-tan(Xm)|<s. Then I conclude that (tan(Xn)) is cauchy in (R,d(x,y)=|x-y|) which is complete. Then (tan(Xn)) converges to some X0 in R. Then I use this fact and get d((Xn),arctan(X0))=|tanXn-X0|<s. So the second metric is complete. Is this the right way to do it? $\endgroup$ – Elvis Apr 17 '13 at 4:40

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.