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

If every rational $r_i$ in $[0,1]$ lies in the interior of an open interval $I_i=(a_i,b_i)$, $a_i,b_i$ real numbers, is there a finite subset of $\{I_i\}$, such that every $r_i$ lies in an interval in this set?

Does it make any difference if the $a_i$, $b_i$ are rationals? Does it hold for every dense countable set?

What are necessary conditions on subsets of $[0,1]$ for it to hold?

share|cite|improve this question
If subset of intervals is finite it is necessary bounded – Norbert Feb 9 '12 at 20:23
Is your notation meant to imply that we've listed all the rationals in $[0,1]$ as a sequence $(q_i)$, and that also $q_i \in (a_i,b_i)$ for each $i$ (or only for each $i$, there is some $j$ with $q_i\in (a_j,b_j)$)?? Hint: what if the sum $\sum_i b_i-a_i$ converges to something very small? – Matthew Daws Feb 9 '12 at 20:31
Yes. The latter. Can you expand your hint please? – topologycalisticalist Feb 9 '12 at 20:34
Sorry, my hint was pretty poor. But you have 2 good answers now... – Matthew Daws Feb 9 '12 at 20:48
up vote 3 down vote accepted

For the final question: Heine-Borel really says that closed bounded intervals of $\mathbb R$ are compact. So a subset of $[0,1]$ will have your property if and only if it is compact. As Heine-Borel says that $[0,1]$ is compact, it's a result that subsets of $[0,1]$ are compact if and only if they are closed.

A countable set with this property is $\{ 1/n : n\in\mathbb N \}\cup\{0\}$; of course this isn't dense. As Brian Scott notes, no proper dense set can be closed, and so won't have this property.

share|cite|improve this answer

The answer to the first question is no: consider an irrational number $r$ in $[0,1]$. Let $\left\{b_k\right\}_{k\in\mathbb{N}}$ be a sequence of irrational numbers converging monotonically to $r$ (say the sequence is increasing). Let $I_k = (b_{k-1}, b_k)$, for $k \geq 2$. Let $I_1 = (-1, b_1)$ and $I_{\infty} = (r, 2)$ (I am using $\infty$ here just as a label). Then you obtain a countable sequence of open intervals, $\left\{I_k\right\}_{k\in\mathbb{N}}\cup \left\{I_\infty\right\}$ which covers every rational in $[0, 1]$, yet no finite subset of this countable cover covers $\mathbb{Q}\cap [0,1]$.

Reason through the other questions in a similar fashion.

share|cite|improve this answer

The answer is no. Let $\alpha$ be an irrational in the interval $[0,1]$, say $\alpha=1/\sqrt{2}$.

Look at the set of all open intervals of the form $(-1, \alpha -\frac{1}{n})$ or $(\alpha+\frac{1}{n},2)$, where $n$ ranges over the positive integers. Every rational in $[0,1]$ is in one of these intervals. However, this set of intervals cannot be replaced by a finite subset, since a finite subset would miss all the rationals that are close enough to $\alpha$.

Remark: We can replace the intervals $(-1,\alpha-\frac{1}{n})$ by intervals of the shape $(-1, a_n)$, where $(a_n)$ is a sequence of rationals that approaches $\alpha$ from below, and deal similarly with the $(\alpha+\frac{1}{n},2)$. Thus, in your question, replacing reals by rationals does not change the answer.

share|cite|improve this answer

In general there need not be such a finite subcollection. Let $x\in[0,1]$ be irrational, let $\langle p_n:n\in\omega\rangle$ be a strictly increasing sequence of rationals in $[0,1]$ converging to $x$, and let $\langle q_n:n\in\omega\rangle$ be a strictly decreasing sequence of rationals in $[0,1]$ converging to $x$. For $n\in\omega$ let $I_n=(-1,p_n)$ and $J_n=(q_n,2)$, let $\mathscr{I}=\{I_n:n\in\omega\}$, and let $\mathscr{J}=\{J_n:n\in\omega\}$. Then $\mathscr{I}\cup\mathscr{J}$ is a family of open intervals with rational endpoints, every rational in $[0,1]$ is contained in a member of $\mathscr{I}\cup\mathscr{J}$, and no finite subfamily of $\mathscr{I}\cup\mathscr{J}$ covers $\mathbb{Q}\cap[0,1]$.

This construction can clearly be generalized to any dense proper subset of $[0,1]$.

share|cite|improve this answer

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.