We recall the notion of a $k$-space (or compactly generated space) to fix our notations. For every topological space $X$, we can define a category $\mathfrak{M}_X$. The class of objects of $\mathfrak{M}_X$ is the class of continuous mappings $u: K \to X$ from a compact Hausdorff space $K$ to $X$. Let $v : C \to X$ be another object in $\mathfrak{M}_X$, the set of morphisms from $u$ to $v$ is the continuous mappings $f: K \to C$ such that $u = v\circ f$. Let $\tau$ be the collection of open sets in $X$. Define $k(\tau)$ to be the final topology on $X$ with respect to the class $\mathrm{Obj}(\mathfrak{M}_X)$. In other words, for any subset $U$ of $X$, $U \in k(\tau)$ if and only if $u^{-1}(U)$ is open in $K$ for any object $u : K \to X$ in $\mathfrak{M}_X$. Let $k(X)$ be the space $X$ equipped with the topology $k(\tau)$ which is finer than $\tau$ by definition. Then we call $X$ a $k$-space if $k(X) = X$, or more precisely, $\tau = k(\tau)$.

If we adopt the approach of the Grothendieck's axiom on universe to approach the above categorical setting. We can fix a universe $\mathcal{U}$ with $X \in \mathcal{U}$. The category $\mathfrak{M}_X$ is redefined as the category of $\mathcal{U}$-small continuous mappings $u : K \to X$ from a compact Hausdorff $K$ which is also $\mathcal{U}$-small. To emphasize this dependence on our universe $\mathcal{U}$, we denote this category by $\mathfrak{M}_{X,\mathcal{U}}$. Similarly, the final topology on $X$ with respect to $\mathrm{Obj}(\mathfrak{M}_{X, \mathcal{U}})$ is denote by $k_{\mathcal{U}}(\tau)$.

My question is, if we have another universe $\mathcal{V}$ with $X \in \mathcal{V}$, does $k_{\mathcal{U}}(\tau) = k_{\mathcal{V}}(\tau)$?

A sufficient condition would be there is a small (small is for any universe containing $X$) final subcategory $\mathfrak{S}_X$ of $\mathfrak{M}_X$, as is the case when $X$ is assumed to be weakly Hausdorff (we can take $\mathfrak{S}_X$ to be the full subcategory of $\mathfrak{M}_X$ generated by all embeddings $K \hookrightarrow X$ where $K$ is a closed compact Hausdorff subspace of $X$). But for general topological space $X$, can we find a small final subcategory $\mathfrak{S}_X$ of $\mathfrak{M}_X$?


The following part of 5.9.1 of Topology and Groupoids shows for a particular space $X$ how to reduce the role of a universe.

If $X$ is a k-space, there is a set $\mathcal C_{X}$ of maps $t : C_{t} \to X$ for compact Hausdorff spaces $C_t$ such that a set $A$ is closed in $X$ if and only if $t^{-1}(A)$ is closed in $C_{t}$ for all $t \in \mathcal C_{X}$.

The proof goes as follows: for each non-closed subset $B$ of $X$ there is a compact Hausdorff space $C_{B}$ and map $t: C_{B} \to X$ such that $t^{-1}[B]$ is not closed in $C_B$. Choose one such $C_{B}$ and one such $t$ for each non-closed $B$, and let $\mathcal C_{X}$ be the set of all these $t$. That this set has the required property is clear.

This leads to standard properties of such k-spaces.

The following paper uses various classes of compact Hausdorff spaces to construct and study convenient categories:

"Monoidal closed, Cartesian closed and convenient categories of topological spaces" Peter I. Booth and J. Tillotson, 88 (1980), No. 1, 35–53 DOI: 10.2140/pjm.1980.88.35

  • $\begingroup$ This solves my problem so elegantly. I guess the trick is to consider how to detect the complement of $\tau_X$ in $2^X$ instead of $\tau$ itself. $\endgroup$ – Rick Sternbach Nov 11 '18 at 12:29

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.