This question came up in my homework (but was not a homework problem). The problem was proving one direction of the Heine-Borel theorem. As with all proofs of compactness, one begins with, "Suppose $A$ is closed and bounded, and $\mathcal{U}$ is an open cover ..." My proof, which I believe is typical, constructed a convergent sequence of points and derived a contradiction. The problem: to construct a sequence, one needs the cover to be countable.

So I invoked that second-countability implies Lindelöf. This is not controversial, of course, but I included a little argument along the lines of the answers in this question: Every open cover has a countable subcover (Lindelöf's lemma).

In particular, for each $x\in A$ there is some $U$ in $\mathcal{U}$ with $x\in U$; second-countability implies for each $x\in A$ there is some basis element $Q$ with $x\in Q\subset U$; the set of all such $Q$ covers $A$ and each $Q\subset U$ for at least one $U\in\mathcal{U}$, so picking one $U\supset Q$ for each $Q$ gives a countable cover of $A$.

I noticed that the answers in that thread included similar phrases: "Now for each $B\in\mathcal{B}U$ choose some $U(B)\in U$ such that $B\subseteq U$", "for each element $O$ of $\Sigma$ choose an element $U$ of $\Omega$ containing it"

My argument and the arguments suggested in the thread linked both seem to assume the Axiom of Choice. I'm not sure that I understand the axiom of choice, though. So, two questions:

1) Are we actually using the axiom of choice in these arguments?

2) Does (second countable) $\Rightarrow$ (Lindelöf) require the axiom of choice?



The results requires some choice. It is consistent that $\mathbb R$ is not Lindelöf but still second countable (without the axiom of choice).

The axiom of choice for countable families of subsets of $\mathbb R$ is in fact equivalent that second countable implies Lindelöf, and to a few other interesting assertions:

H. Herrlich and G. E. Strecker, When is $\mathbb{N}$ Lindelöf?, Comment. Math. Univ. Carolinae 38,3 (1997), 553-556.

  • $\begingroup$ Beautiful! Thank you! $\endgroup$ – Neal Dec 10 '11 at 17:57
  • $\begingroup$ @Neal: No problem, I would have answered sooner, but my system crashed ;-) $\endgroup$ – Asaf Karagila Dec 10 '11 at 18:05

I think it's worth pointing out that, although second-countability implies Lindelöf requires some choice, the Heine-Borel theorem itself doesn't. To prove that a closed and bounded subset of $\mathbb{R}$ is compact, consider first the case $[0,1]$.

Given an open cover of $[0,1]$, consider the set $S = \{x \in [0,1]\mid\text{there is a finite subcover of }[0,x]\}$. Then $0 \in S$, $S$ is open, and the least upper bound of $S$ is an element of $S$ (check each of these facts). It follows that $1 \in S$ is the least upper bound, so there is a finite subcover.

To prove that any closed and bounded set is compact, realize that it is a closed subset of $[-a,a] \cong [0,1]$ for sufficiently large $a$.


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.