1
$\begingroup$

Let $X$ be completely regular space, $\beta X$ be Stone-Čech compactification of $X$, and $\upsilon X$ be Hewitt realcompactification of $X$. Then $X\subset \upsilon X\subset \beta X$.

If the remainder $\beta X\setminus X$ is countably compact space, then can $% \beta X\setminus \upsilon X$ be countably compact?

Under what conditions is it countably compact?

Does anyone have an article or book recommendation on this?

$\endgroup$
7
  • $\begingroup$ Changed the notation to upsilon $\upsilon$. $\endgroup$ Commented Jan 1, 2023 at 16:35
  • $\begingroup$ Certainly it can happen. Trivially, $X$ can be compact, Lindelof., or pseudocompact with countably compact remainder (for instance $[0,\omega_1)$, which this time is not realcompact). $\endgroup$
    – Tyrone
    Commented Jan 2, 2023 at 0:42
  • $\begingroup$ @Tyrone Can $\beta X\setminus \upsilon X$ always be countably compact? $\endgroup$ Commented Jan 2, 2023 at 8:39
  • $\begingroup$ No. Take a completely metrisable space $X$ which is not locally compact (a hedgehog with $\omega_1$ spines, for instance). We have $\nu X=X$. Since $X$ is $G_\delta$ in $\beta X$, the remainder $\beta X\setminus X$ is $\sigma$-compact and hence Lindelöf. On the other hand, $X$ is not locally compact, so $\beta X\setminus X$ is not compact. Since the remainder is Lindelöf, it cannot be countably compact. $\endgroup$
    – Tyrone
    Commented Jan 2, 2023 at 9:08
  • $\begingroup$ @Tyrone But I assume that $\beta X\setminus X$ is countably compact. If the remainder $\beta X\setminus X$ is countably compact space, then is $\beta X\setminus \upsilon X$ always countably compact? $\endgroup$ Commented Jan 2, 2023 at 9:11

1 Answer 1

4
$\begingroup$

There exists a Tychonoff space $X$ such that $\beta X\setminus X$ is countably compact but $\beta X\setminus \upsilon X$ is locally compact, $\sigma$-compact and not countably compact. Such a space can be constructed as follows.

In the Stone-$\check{\mathrm C}$ech remainder $\omega^*=\beta\omega\setminus \omega$ choose a sequence $(U_n)_{n\in\omega}$ of pairwise disjoint nonempty open sets. In each $U_n$ choose a point $x_n\in U_n$ and using the Tychonoff property of $\omega^*$, find a continuous function $f_n:\omega^*\to [0,1]$ such that $f_n(x_n)=1$ and $f_n[\omega^*\setminus U_n]=\{0\}$. Then $K_n=f_n^{-1}(1)$ is a nonempty compact $G_\delta$-set in $\omega^*$ and $(K_n)_{n\in\omega}$ is a discrete sequence of compact $G_\delta$ sets in $\omega^*$ whose union $K=\bigcup_{n\in\omega}K_n$ is a $\sigma$-compact $G_\delta$-set in $\omega^*$ and $K$ is open in its closure $\overline K$ in $\omega^*$. Consider the set $Y=\omega^*\setminus K$ and observe that $Y$ is of type $F_\sigma$ and $G_\delta$ in $\omega^*$.

It is well-known that the space $\omega^*$ has exactly $2^{\mathfrak c}$ closed sets and every infinite closed subset of $\omega^*$ has cardinality $2^{\mathfrak c}$. Using these two facts, it is easy to construct (by transfinite induction of length $2^{\mathfrak c}$) a Bernstein-like set $B\subseteq Y$ such that for every infinite closed subset $F\subseteq Y$ the sets $F\cap B$ and $F\setminus B$ are nonempty. Then the subspace $X=B\cup\omega$ of the space $\beta\omega$ has the desired property.

Indeed, its Stone-$\check{\mathrm C}$ech compactification coincides with $\beta \omega$ (because every bounded continuous function $f:X\to\mathbb R$ uniquely extends to a continuous function on $\beta\omega$). The space $\beta X\setminus X=\beta\omega\setminus X=K\cup(Y\setminus B)$ is countably compact because for every countable infinite discrete subset $C\subseteq K\cup(Y\setminus B)$ either for some $n\in\omega$ the intersection $C\cap K_n$ is infinite and hence has an accumulation point $x\in K_n\subset \beta X\setminus X$ or else for every $n\in\omega$ the intersection $K_n\cap C$ is finite and the set $\overline C\setminus C$ of accumulation points of $C$ is contained in $Y$. Since the closed set $\overline C$ is infinite, it has cardinality $2^{\mathfrak c}$ and hence $(\overline C\setminus C)\setminus B$ is non-empty, which means that the set $C$ has an accumulation point in $Y\setminus B\subseteq \beta X\subseteq X$ and $\beta X\setminus X$ is countably compact.

Next, we show that the Hewitt completion $\upsilon X$ coincides with $\beta\omega\setminus K$. Indeed, for every $x\in K$ we can find $n\in\omega$ such that $x\in K_n$ and hence $K_n$ is a $G_\delta$-subset of $\beta\omega$ that contains $x$ and is disjoint with $X$, witnessing that $x\notin\upsilon X$.

On the other hand, for every $x\in Y$, every $G_\delta$-set $G\subseteq\beta\omega$ that contains $x$ has nonempty intersection with the Bernstein-like set $B\subseteq X$ (because $G$ contains a non-empty and hence infinite closed $G_\delta$-subset, which intersects $B$ by the choice of $B$). This means that $Y\subseteq\upsilon X$ and hence $\upsilon X=\beta\omega\setminus K$.

It remains to observe that the complement $\beta X\setminus\upsilon X=K$ is locally compact, $\sigma$-compact and not countably compact, even not pseudocompact.

$\endgroup$
4
  • 2
    $\begingroup$ Did you want $K_n$ to be $f_n^{-1}(1)$ rather than $f_n^{-1}(0)$? $\endgroup$
    – Anonymous
    Commented Jan 8, 2023 at 13:58
  • 1
    $\begingroup$ @Anonymous Yes! Thank you. Of course, it had to be $f_n^{-1}(1)$. $\endgroup$ Commented Jan 8, 2023 at 19:03
  • 2
    $\begingroup$ I must be missing something, but is it clear that $X$ is countably compact? The set $K$ can be assumed to be a cozero set of $\omega^*$. (That is what you get if each of the sets $U_n$ is closed as well as open and $f_n$ is the characteristic function of $U_n$.) But $K$ is contained in $\beta X \setminus X$, and a subset of $\omega$ whose accumulation points are in $K$ has no accumulation points in $X$. $\endgroup$
    – Anonymous
    Commented Jan 8, 2023 at 21:48
  • 1
    $\begingroup$ @Anonymous Yes, you are right, $X$ is not countably compact, and cannot be countably compact if $\beta X\setminus\upsilon X$. Thanks for the comment. I have removed the corresponding line from my answer. $\endgroup$ Commented Jan 9, 2023 at 3:44

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .