Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

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

This is a question about the logic of Theorems 2.41, 2.42 in Rudin's 3rd Ed, which deal with the Heine-Borel and Weierstrass properties of sets of $R^k$, respectively.

A quick version of my question is: doesn't 2.41 (together with 2.40, on which it depends) moot 2.42?

2.41. (Heine-Borel+) If a set E in $R^k$ has one of these three properties, it has the other two: (a) E is closed and bounded; (b) E is compact; (c) Every infinite subset of E has a limit point in E.

2.42. (Weierstrass) Every bounded infinite subset of $R^k$ has a limit point in $R^k$.

The text explains that 2.41(a) implies (b) implies (c) implies (a). So the only step to be supplied for 2.42 is that a bounded subset of $R^k$ is compact (and therefore closed, to bring it into the ambit of 2.41).

But in both theorems Rudin resorts to an extrinsic proof (his 2.40) to show that that "k-cells" are compact.

I think (am not sure) there is a theorem stating that Heine-Borel implies Weierstrass (and conversely) but H-B consists of 2.41(a) and (b), according to Rudin's note preceding 2.41. I wonder if, with the addition of 2.41(c), he needs to prove Weierstrass separately?

This is a sort of fussy question but an answer might help me understand the relationship between these ideas better. Thanks. EDITED: so the quick version of the question includes the ref. to 2.40.

share|cite|improve this question
Condition (c) is very close related to Weierstress theorem but the result does not directly applies since there are bounded no closed subsets. – azarel Mar 11 '12 at 20:11
I see your point. If $T$ is infinite and bounded, get a closed ball $E$ that contains it. The ball is closed and bounded so use $(a)\to (c)$ from $2.41$. – Patrick Mar 11 '12 at 20:32
up vote 1 down vote accepted

2.41 does not trivialize 2.42, as it is not necessarily true that a bounded subset of $\mathbb{R}^k$ is compact. Why? Consider the following set:

$$\left\{\frac{1}{n} | n\in \mathbb{N}\right\}\subset \mathbb{R}$$

This is bounded but not compact. Why? $$\bigcup_{i\in\mathbb{N}} \left(\frac{1}{i+1/2},\frac{1}{i-1/2}\right)$$ is an open cover with no finite subcover. So you really do need both closed and bounded for compact. As a result, you have to supply an extra argument to get 2.42 from 2.41- an outline of such an argument is as follows:

Every bounded infinite subset $A\subset\mathbb{R}^k$ has a closure $A\cup \{\mathrm{limit points of A}\}=\overline{A}\subset\mathbb{R}^k$, which is also a bounded infinite subset of $\mathbb{R}^k$. But $\overline{A}$ is closed, so since it is also bounded, it is compact by 2.41 (a)$\Rightarrow$(b). Then we can apply 2.41 (b)$\Rightarrow$(c) to see that such a set has a limit point in $\overline{A}\subset\mathbb{R}^k$ and thus every bounded infinite subset of $\mathbb{R}^k$ has a limit point in $\mathbb{R}^k$.

share|cite|improve this answer
I'm sorry, I don't have a copy of Rudin with me. Could you perhaps post the statement of 2.40? – KReiser Mar 11 '12 at 20:24
I believe my answer still holds even if we know 2.40. There are lots of subsets of $\mathbb{R}^k$ which are infinite and bounded but not $k$-cells. You still do need something extra. – KReiser Mar 11 '12 at 20:37
Yes, and this is the assertion that makes the whole thing work, more or less. I use this in the first sentence of my last paragraph- explicitly, if $A$ is bounded, then $A$ is a subset of some $k$-cell, and thus the closure of $A$ is a subset of this same $k$-cell. – KReiser Mar 11 '12 at 21:22
Everything you have said, to include your answer above, is correct. But 2.40 is precisely the "other argument" you present in the last paragraph--it gets us compactness for 2.42. But it was used to prove compactness in 2.41(b). I am upvoting this answer (which is essentially azarel's) because it is correct, but I will wait to accept it, because...well for the reason I just gave. 2.40 is the extra argument. – daniel Mar 12 '12 at 20:50

The connection between the topics is strong, but not as strong as you currently believe. The problem is that a bounded set of $\mathbb{R}^k$ is not necessarily compact: take the open unitl ball in $\mathbb{R}^k$, or more generally any bounded set in $\mathbb{R}^k$ that is not closed. The k-cells are invoked to say that our closed and bounded set $E$ is a subset of some k-cell $I$, as it is bounded. It would be enough to take $E$'s closure, but this approach also works: as k-cells are compact, 2.41 tells us that every infinite subset of $I$ (which includes the infinite subsets of $E$) has a limit point in $I \subset \mathbb{R}^k$ (and not necessarily $E$).

Further examining Rudin's setup, both the given proofs for Heine-Borel and Weierstrass use the general observations that infinite subsets of compact sets have a limit point in that compact set, as well as the compactness of k-cells and that closed sets of compact sets are compact.

share|cite|improve this answer
In fairness, I would upvote this answer if you remove the phrase "extremely related" from the second paragraph. – daniel Mar 13 '12 at 9:41
too pedantic? In any case, it's gone. – JakeR Mar 13 '12 at 14:23

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.