Let $\mathcal{T}$ be the family of all open sets in $\mathbb{R}$. Show that $| \mathcal{T}|=2^{\aleph_0}$

$\textbf{My Attempt:}$

I know that $\forall A \in \mathcal{T}$. $A$ is the countable union of open intervals with rational end points.

I want to use the Cantor-Bernstein Theorem. That is I need to find injective functions $f$ and $g$ such that $f: 2^{\aleph_0} \to \mathcal{T}$ and $g: \mathcal{T} \to 2^{\aleph_0}$.

I know each $A \in \mathcal{T}$ is of the form $A = \bigcup_{x \in A} (r_x,s_x)$ where $r_x,s_x \in \mathbb{Q}$. How can I use this fact to find the injective function $f$ and $g$?

  • $\begingroup$ sorry was editing. I mentioned ' isn't it unproveable that $2^ω$ has the same cardinality as the continuum?' I might be cconfusing my ordinals and cardinals $\endgroup$
    – user136920
    May 8, 2014 at 16:44
  • $\begingroup$ No, but it is unprovable (in ZFC) that there is no set of cardinality strictly between $\omega$ and $2^\omega$. $\endgroup$
    – user134824
    May 8, 2014 at 16:46
  • 2
    $\begingroup$ I much prefer writing $2^{\aleph_0}$ for the cardinality of $\mathbb{R}$ (or $\mathcal{P}(\mathbb{N})$ or whatever), since $2^{\omega}$ looks like ordinal exponentiation. (And with ordinal exponentiation, $2^{\omega} = \omega$.) These concerns are probably irrelevant to the question though. $\endgroup$ May 8, 2014 at 17:03

2 Answers 2


It is often useful to replace $2^{\aleph_0}$ with another set of the same cardinality for the purpose of finding injective functions; I'll do so frequently, and I recommend you learn to think about sets in this way.

For one direction, finding an injective function $\mathbb R\to\mathcal T$ should be easy. I leave it to you. (Why does $|\mathbb R|=|2^{\aleph_0}|$?)

The other direction is harder. Your observation is not quite right: I don't see how you're associating $x$ to $r_x$, since $x$ could be irrational. I think the fact you're referring to -- a version of it, at least -- is the following.

Let $A\subseteq\mathbb R$ be open. For each rational number $q\in A$ there exists a rational radius $r_q\in\mathbb Q$ such that $(q-r_q,q+r_q)\subseteq A$. Then $A=\bigcup_{q\in A}(q-r_q,q+r_q)$.

Since $A$ is completely characterized by the set $\{(q,r_q)\,:\,q\in A\cap\mathbb Q\}$ (here I mean $(q,r_q)$ as an ordered pair, not an interval!), why don't we make a function $\mathcal T\to 2^{\mathbb Q\times\mathbb Q}$ given by $A\mapsto \{(q,r_q)\,:\,q\in A\cap\mathbb Q\}$. The map is injective (why?) and $|2^{\mathbb Q\times\mathbb Q}|=|2^{\aleph_0}|$ (why?). So we're done!

Edit: The statement I gave is not correct as written, as user48481 Mirko Swirko pointed out. The problem is that if I pick the $r_q$ to be too small then I might not get all the points in the open set. For instance, if $A=(0,2)$ and I choose $r_q$ so that $\sqrt2\notin(q-r_q,q+r_q)$ for any $q$, which is possible because $\sqrt2$ is irrational, then the union of the $(q-r_q,q+r_q)$ will miss the point $\sqrt2$.

A corrected version of the statement goes like this.

Let $A\subseteq\mathbb R$ be open. There exists an assignment $q\mapsto r_q$ of rational numbers to rational radii (so $q,r_q\in\mathbb Q$) such that $A=\bigcup_{q\in A} (q-r_q,q+r_q)$.

To prove the statement, it suffices to consider the case where $A$ is an interval $(a,b)$, since every open subset of $\mathbb R$ is a countable union of such intervals, as you observed. You can do this by letting $(q_n)_{n=1}^\infty$ being an enumeration of the rationals in $(a,b)$ and choosing $r_{q_n}$ so that, e.g., if $q_n$ is closer to $a$ than $b$ then $q_n-r_{q_n}-a<2^{-n}$.

  • $\begingroup$ You're absolutely right: I was sloppy. I'm pretty sure I've made some version of this mistake three or four times, at least. I'll fix it. $\endgroup$
    – user134824
    Dec 1, 2014 at 14:41
  • $\begingroup$ Thank you for the correction, I only got aware in this inaccuracy after @user196665 posted a related question math.stackexchange.com/questions/1046029/… $\endgroup$
    – Mirko
    Dec 2, 2014 at 2:16

On the one hand, it is clear that $|\mathbb{R}|\le|\mathcal{T}|$, since for each $r\in\mathbb{R}$, we can consider the open set $(r,+\infty)$; it is clear that this assignation is injective.

On the other hand, for each open subset $A\subseteq\mathbb{R}$ there exists an assignation $q\longmapsto r_q$ between rational numbers with $r_q>0$ such that $A=\bigcup_{q\in A\cap\mathbb{Q}}(q-r_q,q+r_q)$ (because the set of open intervals with rational numbers as their extremes is a basis of the usual topology on $\mathbb{R}$).

Since $A$ is completely characterised by the set $\{\langle q,r_q\rangle|\;q\in A\cap\mathbb{Q}\}\subseteq\mathbb{Q}\times(\mathbb{Q}\setminus\{0\})$, the function $f:\mathcal{T}\longrightarrow^{\mathbb{Q}}\mathbb{Q}$ defined by: for each $A\in\mathcal{T}$:

$$f(A)=\{\langle q,r_q\rangle|\;q\in A\cap\mathbb{Q}\}\cup\{\langle q,0\rangle|\;q\not\in A\cap\mathbb{Q}\}$$

Is clearly injective, according to our previous observation, and thus $|\mathcal{T}|\le|^{\mathbb{Q}}\mathbb{Q}|=|\mathbb{R}|$

From the Cantor-Bernstein theorem we conclude that $|\mathcal{T}|=|\mathbb{R}|=2^{\aleph_0}$


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.