12

I'm trying to find connections between Stone–Čech compactifications of $\mathbb{N}, \mathbb{Q}$ and $\mathbb{R}$, all with the euclidean topology. So, are there any ? e.g. is $\beta \mathbb{Q} = \beta \mathbb{R}$ ? I know that the cardinality of $\beta \mathbb{N} = \beta \mathbb{Q} = \beta \mathbb{R}$ thanks to Wikipedia, but how does one prove it ?

Potato
  • 40,171
KevinDL
  • 1,816
  • 2
    Yes, I know, but I thought it isn't necessarily true that if the cardinality of 2 sets is the same, their Stone-Cech compactifications have the same cardinality. – KevinDL Jan 03 '12 at 12:59

2 Answers2

19

Corrected Version: Let $f$ be any map of $\mathbb{N}$ onto $\mathbb{Q}$. Then $f$ is a continuous map of $\mathbb{N}$ into $\beta\mathbb{Q}$, so it extends to a map $\beta f:\beta\mathbb{N}\to\beta\mathbb{Q}$. Since $f$ is a surjection, so is $\beta f$, and it follows that $|\beta\mathbb{N}|\ge|\beta\mathbb{Q}|$. Similarly, the inclusion map $i$ of $\mathbb{Q}$ into $\mathbb{R}$ has a continuous extension to $\beta i:\beta\mathbb{Q}\to\beta\mathbb{R}$, and since $i[\mathbb{Q}]$ is dense in $\mathbb{R}$, $\beta i$ is a surjection, and $|\beta\mathbb{Q}|\ge|\beta\mathbb{R}|$. Finally, $\mathbb{N}$ is $C^*$-embedded in $\mathbb{R}$, so $\beta\mathbb{N}$ embeds in $\beta\mathbb{R}$, and $|\beta\mathbb{N}|\le|\beta\mathbb{R}|$. Thus, $|\beta\mathbb{N}|=|\beta\mathbb{Q}|=|\beta\mathbb{R}|$, and I recently gave a proof that $|\beta\mathbb{N}|=2^\mathfrak{c}$ in this answer.

Note that we knew from the beginning that these cardinalities were at most $2^\mathfrak{c}$: $\mathbb{N}$, $\mathbb{Q}$, and $\mathbb{R}$ are separable and dense in their respective Čech-Stone compactifications, so $\beta\mathbb{N},\beta\mathbb{Q}$, and $\beta\mathbb{R}$ are also separable, and a separable Hausdorff space cannot have more than $2^{2^\omega}=2^\mathfrak{c}$ points.

To see that if $X$ is a separable Hausdorff space, then $|X|\le 2^\mathfrak{c}$, let $D$ be a countable dense subset of $X$. For each $x\in X$ let $\mathscr{D}(x)=\{V\cap D:V\text{ is an open nbhd of }x\}$; $\mathscr{D}(x)$ is a subset of $\wp(D)$, and $|\wp(D)|=\mathfrak{c}$, so there are only $2^\mathfrak{c}$ possible families $\mathscr{D}(x)$. Hausdorffness of $X$ implies that $\mathscr{D}(x)\ne\mathscr{D}(y)$ when $x\ne y$, so $|X|\le 2^\mathfrak{c}$.

Brian M. Scott
  • 616,228
  • 2
    I'm sorry if these are stupid questions, but why does $\beta \mathbb{Q}$ embed in $\beta \mathbb{R}$ (the other one I can see, since $\mathbb{N}$ *-embeds in $\mathbb{Q})$ ? What is the theorem that states this bound on points in a seperable Hausdorff space ? – KevinDL Jan 03 '12 at 13:35
  • Thank you, I already knew that $|\beta \mathbb{N}|$ has cardinality $2^{\mathfrak{c}}$, forgot to mention that, sorry. But why does $\beta \mathbb{Q}$ embed in $\beta \mathbb{R}$ ? Is $\mathbb{Q}$ *-embedded in $\mathbb{R}$ ? – KevinDL Jan 03 '12 at 13:47
  • @Kevin: I’ve included the proof that $|X|\le 2^\mathfrak{c}$ if $X$ is separable and $T_2$. I slipped up on the other part and am about to fix it. – Brian M. Scott Jan 03 '12 at 13:48
  • 1
    Do my eyes deceive me, or did you write $|\mathbb Q|\geq|\mathbb R|$ there? :-) – Asaf Karagila Jan 10 '13 at 10:26
  • @Asaf: Damned betas; always wandering off when you need them. – Brian M. Scott Jan 10 '13 at 16:14
8

In answer to the first part of your question, it's not true that $\beta\mathbb Q=\beta\mathbb R$. For suppose $g$ is some autohomeomorphism of $\mathbb R$. This extends to an autohomeomorphism $\beta g$ of $\beta\mathbb R$, and if we had $\beta\mathbb Q=\beta\mathbb R$, it would follow that $\beta g$ is an autohomeomorphism of $\beta\mathbb Q$. Since the restriction of $\beta g$ to $\mathbb Q$ is then an autohomeomorphism of $\mathbb Q$ (see the paper cited below), it would follow that every autohomeomorphism of $\mathbb R$ maps $\mathbb Q$ to itself, and this is clearly not true (e.g., $x \mapsto x +\sqrt2$).

Re. comparison of the compactifications, it's potentially worth looking at the remainders. It turns out, for example, that $\mathbb Q$* is separable but $\mathbb R$* isn't. See: M. P. Stannett, "The Stone-Čech Compactification of the Rational World", Glasgow Math. J., vol. 30 (1988) pp. 181-188.

Stefan Hamcke
  • 27,733
Mike
  • 121