Let $\mathbb{I}$ denote the unit interval. Call a topological space $X$ $\mathbb{I}$-compact iff every continuous function $f : X \rightarrow \mathbb{I}$ attains its maximum at some point in $X$.

Question. Is every $\mathbb{I}$-compact topological space compact?

We can replace $\mathbb{I}$ with either of $[0,\infty]$ and/or $[-\infty,\infty]$, of course, without changing the meaning of the above definition.


If $X$ is $\mathbb{I}$-compact then $X$ must be pseudocompact. This can be seen by a contrapositive: suppose $X$ is not pseudocompact. Then there is a continuous unbounded function $f: X \to \mathbb{R}$. We can assume that $f[X] \subseteq [1,\infty)$ WLOG. But then $1-\frac{1}{f} : X \to [0,1]$ is also continuous and does not assume its supremum $1$, so $X$ is not $\mathbb{I}$-compact..

Reversely, every pseudocompact space is $\mathbb{I}$-compact, as a continuous image of a pseudocompact space is pseudocompact, and in metric spaces pseudocompactness is equivalent to compactness and compact subsets of $[0,1]$ have a maximum. (this is true in any ordered space.)

So the answer is no for e.g. $X= \omega_1$ in the order topology. There are many more examples of non-compact pseudocompact spaces, like Mrówka's $\Psi$ space, $\Sigma$-products of uncountably many copies of $[0,1]$ etc.

So I think that $X$ $\mathbb{I}$-compact is just equivalent to pseudocompactness of $X$. So it is equivalent to compactness in the realm of metric spaces, e.g.


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.