Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I am going through the chapter on compactness and completeness from Sternberg's Advanced Calculus and trying to build an intuition for what many of this topological properties mean, and which imply which. The book defines these concepts in the setting of metric spaces, but most of what I found online is in the about topology, and from what I see (correct me if I am wrong) it doesn't change the general picture much.

I've made this diagram to see whats the relationship of the different concepts and have examples of each. I don't know if it is correct. For example, Is it true that a bounded complete metric (sub)space is compact (and therefore totally bounded)? Then why bother defining total boundedness? If you don't think anything else I wrote is a valid question stick to answering that, although pointing out any misconceptions I might have is appreciated.

metric space properties

For intuition about compactness I've found this posts really helpful. It helps me to think that there are (at least) two different kinds of infinity: one in the sense of largeness (of which boundedness is the opposite), and another in the sense of denseness (of which discreteness is the opposite).

share|cite|improve this question
I looked for a question in this post, but the only one I see is "why bother defining total boundedness"? Perhaps you should edit the Question to make what you want to ask more explicit. – hardmath Sep 13 '13 at 22:50
I knew someone would say that. I just wanted someone to tell me if there is something wrong with the relationships I have assumed in the diagram. – spelufo Sep 13 '13 at 23:08
And there is. As noted below, there are metric (sub)spaces which are complete, bounded and are not totally bounded. – spelufo Sep 14 '13 at 0:12
up vote 2 down vote accepted

I have not read Sternberg's Advanced Calculus but I think that it might be wrong about "boundedness + completeness implies compactness".

For example take any infinite set, say the unit interval $I=[0,1]$. Now consider the discrete metric on it; that is, consider the metric space $(I,d)$ where $d(x,y)= 1$ if and only if $x=y$ and $d(x,y)= 0$ otherwise.

By definition, $I$ is compact if every open cover of it admits a finite subcover. Cover $I$ with the collection of open balls $\mathcal{C} = \{U_x\}_{x \in I}$, where $U_x = B_{1/2}(x) = \{x\}$. It is clear, by the infinitude of $I$, that $\mathcal{C}$ does not admit a finite subcover and then $I$ isn't compact. However $I$ is bounded, as the maximum value of $d$ is $1$, and $(I,d)$ is complete, as every cauchy sequence is eventually constant(to see this take $\epsilon = 1/3$ in the definition of cauchy sequence).

The last paragraph proves that boundedness and completeness does not imply compactness. Furthermore, the same example (with the same cover $\mathcal{C}$) shows that boundedness and completeness together does not imply the Lindelöf property. Lindelöf property can be thought of as a weakening of compactness requiring every open cover to have a countable subcover. See this article.

In contrast with the previous counterexample, it's true that total boundedness and completeness together imply compactness. Thus, the point of defining total boundedness will be to have a "metric space charactrization" of compactness.

share|cite|improve this answer
Total boundedness + completeness $\iff$ compactness in metric spaces. – Pedro Tamaroff Sep 13 '13 at 23:43
That wasn't the book, that was me. I couldn't find an example of a complete, bounded yet not totally bounded set. Thanks for providindg one. – spelufo Sep 14 '13 at 0:09

Just to clear up an apparent misconseption. If we have the same definition of "closed subset", then whether a topological (or metric) space is closed or not is not a well defined property without specifying which superspace the property is measured against. For example, every space is a closed subset of itself. So just saying that $\mathbb{Q}$ is closed is meaningless.

Also, "in an infinite dimensional normed vector space" is a seemingly trivial property. Whatever set (of cardinality less than or equal to $\mathbb{R}$) you give me, I can find an infinite dimensional normed vector space to embed it in.

share|cite|improve this answer
Thanks. All the other properties are intrinsic to the metric space, right? To your second observation: That is the part I find most challenging, finding an intuition for the distinction between boundedness and total boundedness. – spelufo Sep 13 '13 at 23:20
I don't think these concepts are very intuitive. The distinction between total boundedness and boundedness is quite technical to my eye. It is simply a stricter form of boundedness. Wikipedia gives the example of an infinite decrete metric space as bounded space which is not totally bounded. I don't know if that helps. And yes, the other properties are intrinsic, as far as I can see. – Espen Nielsen Sep 13 '13 at 23:28

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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