3

Can the following assignment on objects be made into a functor from the category of topological spaces into the category of groups? Each topological space $X$ gets mapped to $\mathrm{Aut}(X)$, its automorphism group (group of all homeomorphisms from $X$ to $X$). I know that one necessary condition of functors $F\colon \mathcal C\to\mathcal D$ is that whenever $X, Y\in\mathcal C$ are isomorphic, then $F(X)$ and $F(Y)$ are isomorphic. This should be true for $F = \mathrm{Aut}$.

I read that in algebraic topology one is interested in so-called algebraic "invariants". Is $\mathrm{Aut}$ such an invariant (after all, it assigns to each topological space a group)?

  • It is unclear to me what group-homomorphism $\operatorname{Aut}X\to\operatorname{Aut} Y$ is supposed to be "$\operatorname{Aut}f$", for a generic continuous map $f:X\to Y$. –  May 02 '20 at 14:16
  • 1
    If you consider the subcategory of Top where the morphisms are homeomorphisms then it is a functor in an obvious way. I'm not sure if there's any natural way of extending it to larger subcategories. – William May 02 '20 at 17:08
  • I'm just gonna leave this as a comment but neither the accepted answer nor the post that is a "duplicate" of this one answer @user762130 's original question. – Noel Lundström May 02 '20 at 19:56

1 Answers1

2

First off, it is true that the automorphism group is an invariant in the sense that if two spaces $X$, $Y$ are homeomorphic, then their automorphism groups are isomorphic.

To see this, for a homeomorphism $f: X \rightarrow Y$ consider $\Phi(f): \text{Aut}(X) \rightarrow \text{Aut}(Y)$ via $\varphi \mapsto f \circ \varphi \circ f^{-1}$.

However, in order to make the assignment $F: X \mapsto \text{Aut}(X)$ into a functor you of course need to specify what $F$ does on morphisms $f: X \rightarrow Y$. The naive choices would be pre- or post- composition, but neither of those work since that would not give back an automorphism, but a continuous map between the spaces.

So, maybe there is a way to make it into a functor, but the obvious candidates do not work.

Edit regarding comment: the functor that Kevin Arlin is referring to is defined as follows: Let $\text{Core}(\mathbf{Top})$ be the category with objects topological spaces and maps the homeomoprhisms. Then define the functor $\Phi: \text{Core}(\mathbf{Top}) \rightarrow \mathbf{Grp}$ on objects as $\Phi(X) = \text{Aut}(X)$ and on morphisms $f: X \rightarrow Y$ as $\Phi(f): \text{Aut}(X) \rightarrow \text{Aut}(Y)$ via $\varphi \mapsto f \circ \varphi \circ f^{-1}$.

G. Chiusole
  • 5,594
  • No, there's no way to make it into a functor defined on all maps. But it is a functor defined on the category of spaces and homeomorphisms, under the definition you've just given. – Kevin Carlson May 02 '20 at 17:24
  • @KevinArlin Do you have a reference for this? I've been thinking about this problem for a few hours, it seems obvious but I can't make progress. Thanks! – Noel Lundström May 02 '20 at 20:10
  • @NoelLundström It can’t even be done for the full subcategory spanned by a 3- and a 4-point discrete space. Since you’re interested in the question, maybe I’ll leave that as a hint for now, but I’m happy to explain more fully. – Kevin Carlson May 03 '20 at 02:36