Let $X$ be a topological space, and define a path as a continuous map $\gamma : [a,b] \rightarrow X$. Two paths $\gamma : [a,b] \rightarrow X$ and $\phi : [c,d] \rightarrow X$ are equivalent ($\gamma \sim \phi$) iff there exists an increasing homeomorphism $\psi: [a,b] \rightarrow [c,d]$ such that $\phi \circ \psi = \gamma$. The equivalence class of a path is denoted by $[\gamma ]$.

Now define the space of paths $P(X) = \lbrace [\gamma]\ \vert\ \gamma : [a,b] \rightarrow X\ \text{is a path} \rbrace$.

I am wondering: is there is a useful or a natural topology that can be put on $P(X)$, generated by $X$?

Usually topologies are chosen to make a certain type of function continuous, but I can't think of anything in particular that would be a natural type of function on paths.

  • 4
    $\begingroup$ I don't know if it will help in your application, but the compact-open topology is a useful topology to put on spaces of maps. $\endgroup$ Dec 8 '11 at 4:44
  • 4
    $\begingroup$ Useful in what way. Intuitively for me the "obvious" choice would be to topologize the set of maps $[a,b]\to X$ with the compact-open topology, and then let $P(X)$ just have the quotient topology. Is this meaningful? I'm not quite sure, but it's the first thing that "makes sense". $\endgroup$ Dec 8 '11 at 4:45
  • $\begingroup$ Yeah using the compact open topology with a quotient map would certainly give a topology on $P(X)$. I haven't really thought about what that topology is like to be honest, but it sounds interesting. $\endgroup$ Dec 8 '11 at 4:52
  • $\begingroup$ Another idea I had would be to maybe come up with a topology that makes a certain type of map continuous. For example, one which makes linear functionals on smooth paths in $\mathbb{C}$ into continuous maps $X \rightarrow \mathbb{R}$. That didn't seem to be all that interesting of a topology though when I worked it out. $\endgroup$ Dec 8 '11 at 4:53
  • 8
    $\begingroup$ The compact-open topology doesn’t seem to play well with $\sim$. Clearly we may as well consider only maps from $I=[0,1]$ to $X$. Take one of the simplest cases, the point $[1_I]\in C(I,I)/\sim$, the set of all strictly monotone increasing maps from $I$ onto $I$: if $\alpha\in C(I,I)\setminus[1_I]$ is such that $\alpha(0)=0$ and $\alpha(1)=1$, $\alpha$ cannot be separated from $[1_I]$ by an open set in the compact-open topology, and therefore $C(I,I)/\sim\;$ isn’t $T_1$. $\endgroup$ Dec 8 '11 at 5:16

Your equivalence relation seems well-suited to studying just the image of the path, so in a metric space, you could use the Hausdorff metric on the collection of images of the paths. The Hausdorff metric is where the distance between two compact sets $A,B$ is the supremum of all distances $d(a,B)$ and $d(b,A)$ for points $a$ in $A$ and $B$ in $B$.

In this topology, two equivalence slashes are close if their images are almost the same.

However, this doesn't work completely well since some paths have the same image without being equivalent.


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.