6
$\begingroup$

Let $Top_1$ be the category of topological spaces which are $T_1.$

I am curious as to whether there is a categorical definition of what a closed embedding is in this environment. With a categorical definition, I mean something along the lines that in the category $Top_2$ , consisting of topological spaces which are $T_2,$ the closed embeddings are precisely the extremal monomorphisms.

In $Top$ (and $Top_1$) the extremal monomorphisms are precisely the embeddings. In $Top$ we can use the Sierpinski space to single out the ones with closed image, but this is not possible in $Top_1$ since the Sierpinski object is not $T_1.$

Any comments or thoughts would be welcome.

$\endgroup$
1
  • $\begingroup$ @Zhen Lin: I think all extremal monomorphisms are closed embeddings in Hausdorff spaces. If $f:X \rightarrow Y$ is an extremal mono ,consider $Z = \overline{f(X)}$ (the closure of the images) and let $g:X \rightarrow Z$ be the obvious map and $i: Z \rightarrow Y$ the inclusion. $g$ is an epi in the category of Hausdorff spaces,since it has dense image, so is an iso since f is supposed to be extremal. In the category of Hausdorff spaces, closed embeddings are regular monomorphisms, and since every regular monomorphism is extremal, the claim should follow. $\endgroup$
    – user44591
    Commented Sep 22, 2015 at 18:46

1 Answer 1

9
$\begingroup$

As partial motivation, let me start with the observation that in $\mathrm{Set}$, if $i: A \to B$ is a monomorphism, then $i$ is retrieved as the pullback of the monomorphism $\ast \cong A/A \hookrightarrow B/A$ along the canonical map $q: B \to B/A$ (where $B/A$ is the pushout or cofiber product of $i$ and the unique map $A \to 1$). This is one of the exactness properties singled out by Freyd in a categories-list discussion that are common to pretoposes and abelian categories. The square

$$\begin{array}{ccc} A & \to & 1 \\ i \downarrow & & \downarrow \\ B & \stackrel{q}{\to} & B/A \end{array}$$

which is simultaneously a pushout and pullback is sometimes called, affectionately, a "dolittle square" (after the pushmi-pullyu in the Doctor Dolittle story). So not knowing quite what to call such monomorphisms in general, I'll call it a dolittle monomorphism.

In other words, a monomorphism $i$ in a category with finite limits and pushouts will be called dolittle if it is the pullback of the inclusion $1 \to B/A$ in the pushout square above. Then in $\mathrm{Top}_1$, closed embeddings $i: A \to B$ are precisely dolittle monomorphisms. (Notice that the pushout construction $B/A$ in $\mathrm{Top}_1$ is constructed as the pushout $B/\bar{A}$ in $\mathrm{Top}$.)

$\endgroup$
2
  • $\begingroup$ This is very nice! Do you know if there is a similar characterization for $T_0$-spaces? Maybe I should ask it as a separate question. $\endgroup$
    – user44591
    Commented Sep 22, 2015 at 19:19
  • 1
    $\begingroup$ There is a nasty trick that lets us recognise the Sierpiński space in $\mathbf{Top}$ as an abstract category. Once we have that, we can identify the closed embeddings. See here. $\endgroup$
    – Zhen Lin
    Commented Sep 22, 2015 at 19:39

You must log in to answer this question.

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