Consider the open cover $\mathscr{F}=\begin{Bmatrix}\pmatrix{0,5-\frac{1}{n}}:n\in\mathbb{N}\end{Bmatrix}$ for the set $(0,5)$.


Is $(0,5)=\bigcup^\infty_{n=1}\pmatrix{0,5-\frac{1}{n}}$? Why does $\mathscr{F}$ have no finite subcover?

  • $\begingroup$ It would seem that if you take any interval out of the infinite union you'd still cover $(0,5)$. $\endgroup$ Sep 12, 2013 at 4:18
  • $\begingroup$ But what happens if you only consider finitely many? $\endgroup$ Sep 12, 2013 at 4:19
  • $\begingroup$ OK, so since no finite subset of $\mathscr{F}$ can possibly cover the infinite set $(0,5)$, we say that "$\mathscr{F}$ has no finite subcover over $(0,5)$," right? $\endgroup$ Sep 12, 2013 at 4:22
  • $\begingroup$ @GrumpyParsnip, do you know what the TAPI pipeline is? $\endgroup$ Sep 12, 2013 at 4:23
  • 1
    $\begingroup$ Suppose $\mathscr{G}$ were a finite subcover. Let $N\in\mathbb{N}$ be the largest $N$ such that $\pmatrix{0,5-\frac{1}{N}}$ is in $\mathscr{G}$. Since $\mathbb{Q}$ is dense in $\mathbb{R}$, choose a rational $q\in\mathbb{Q}$ such that $5-\frac{1}{N}<q<5$. This contradicts that $\mathscr{G}$ covers $(0,5)$. $\endgroup$ Sep 12, 2013 at 4:28

1 Answer 1


$\mathscr{F}$ is an open cover of $(0,5)$; it is not a cover of $\{(0,5)\}$.

  1. Yes, $(0,5)=\bigcup_{n=1}^\infty\left(0,5-\frac1n\right)$. On the one hand it’s clear that $\left(0,5-\frac1n\right)\subseteq(0,5)$ for each $n\in\Bbb Z^+$, so $\bigcup_{n=1}^\infty\left(0,5-\frac1n\right)\subseteq(0,5)$. On the other hand, if $x\in(0,5)$, then $x<5$, and $5-x>0$. Thus, there is a positive integer $k$ such that $\frac1k<5-x$ and hence $x<5-\frac1k$, so that $x\in\left(0,5-\frac1n\right)\subseteq\bigcup_{n=1}^\infty\left(0,5-\frac1n\right)$. Since $x$ was an arbitrary element of $(0,5)$, it follows that $(0,5)\subseteq\bigcup_{n=1}^\infty\left(0,5-\frac1n\right)$ and hence that $(0,5)=\bigcup_{n=1}^\infty\left(0,5-\frac1n\right)$.

  2. Let $\mathscr{F}_0$ be any finite subset of $\mathscr{F}$, say $\mathscr{F}_0=\left\{\left(0,5-\frac1{n_k}\right):k=1,\ldots,m\right\}$. Let $n=\max\{n_1,\ldots,n_m\}$; then $5-\frac1{n_k}\le\frac1n$ for $k=1,\ldots,m$, so $$\left(0,5-\frac1{n_k}\right)\subseteq\left(0,5-\frac1n\right)$$ for $k=1,\ldots,m$. This implies that $$\bigcup_{k=1}^m\left(0,5-\frac1{n_k}\right)=\left(0,5-\frac1n\right)\subsetneqq(0,5)$$ and hence that $\mathscr{F}_0$ does not cover $(0,5)$. That is, no finite subset of $\mathscr{F}$ covers $(0,5)$, i.e., $\mathscr{F}$ is an open cover of $(0,5)$ with no finite subcover.

  • $\begingroup$ The professor here use $\{(0,5)\}$ to mean $(0,5)$. $\endgroup$ Sep 12, 2013 at 4:29
  • $\begingroup$ See my $\mathscr{G}$-argument in the comments below my question. $\endgroup$ Sep 12, 2013 at 4:32
  • 1
    $\begingroup$ @SecureSpace: I doubt very much that any Harvey Mudd professor makes such an elementary mistake. That lecture is more than an hour long; if you give me a time stamp for a point at which you think that he’s made that mistake, I’ll take a look. In the meantime, I strongly suspect that he’s used $\{(0,5)\}$ correctly, to mean the set whose only member is the open interval $(0,5)$. This is of course a finite open cover of $(0,5)$, though not a very interesting one. $\endgroup$ Sep 12, 2013 at 4:35
  • $\begingroup$ @SecureSpace: Yes, the argument with $\mathscr{G}$ works fine, though there’s no need to pick a rational number: any number in $\left[5-\frac1N,5\right)$ will do fine. In fact, you can simply use $5-\frac1N$. $\endgroup$ Sep 12, 2013 at 4:36
  • $\begingroup$ 00:15:14, let me know if you're still in doubt. $\endgroup$ Sep 12, 2013 at 5:04

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.