On the Wikipedia page for "morphism (category theory)" it says that:

In the category of topological spaces, morphisms are continuous functions and isomorphisms are called homeomorphisms.

In what sense would this make sense? Is there some abstract or technical reason why "the" morphisms "are" continuous functions, rather than something else, like functions that map open sets into open sets?

  • 4
    $\begingroup$ That's the definition of this category: you define the objects and the morphisms. Your question is similar to asking why "the" field operations in R are ordinary addition and multiplication, rather than something else. Part of the way R is defined as a field is through the usual notions of adding and multiplying. It doesn't make a whole lot of sense to ask "why" we use those field operations on R, or rather you should not expect some kind of profound answer. $\endgroup$ – KCd Feb 24 '16 at 2:52
  • $\begingroup$ Oh OK, so the category of topological spaces having maps that take open sets into open sets as morphisms, would be an equally valid category as the usual one. $\endgroup$ – John Smith Feb 24 '16 at 2:55
  • $\begingroup$ It's also worth noting that to specify a category you need to specify both itsobjects and morphisms, but is usually (for better or worse) named after its objects only, with the morphisms implicit. The most common category of topological spaces is the one where morphisms are continuous functions, but yours (where the morphisms are now open maps) is a category too. $\endgroup$ – pjs36 Feb 24 '16 at 2:56
  • 1
    $\begingroup$ @JohnSmith, yes it is an equally valid category, but it sure is not an equally useful one. $\endgroup$ – KCd Feb 24 '16 at 12:10

It's a bit like asking "Why in the set $\mathbb{N}$ the elements are $0$, $1$, $2$...?" The answer is: because we define $\mathbb{N}$ like that. If we chose different elements it would be an equally "valid" set, but it would be a different set and we wouldn't call it $\mathbb{N}$.

For many applications, we want to consider the category with objects topological spaces and morphisms continuous functions. And it's so prevalent that people decided to call this category "the category of topological spaces" without more qualifiers, simply because it's shorter than "the category of topological spaces and continuous functions". If you manage to cook up a category where the objects are topological spaces but the morphisms are something else (maps that aren't necessarily continuous, open maps...), then it's a "valid" category too, but it's not the one that most people call "the category of topological spaces".


If open maps (functions that map open sets to open sets) are taken as the morphisms, then there are very few continuous (in the usual sense) functions that belong to the category. There are quotient maps onto open subsets of the target space, but no embeddings of a space into one with "thicker" open sets.

Some of the things this excludes are: parametric curves $[0,1] \to \mathbb{R}^2$, constant functions on any nontrivial space, the diagonal map $Y \to Y \times Y$, and embeddings $A \to A \times B$.

Maybe if you restrict the discontinuities of the open maps there are some interesting jigsaw puzzle-like functions in the category but basically the only things that have a connection to the usual topological category are (generalized) coverings of open sets.


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.