How wild can the topology of a length space be? That is,


  1. Let $X$ be a metric space where the distance between two points $x,y \in X$ is the infinum of lengths of rectifiable paths from $x$ to $y$. What can be said about the topology of $X$? Can spaces of this form be completely characterized?

  2. What if we additionally assume that $X$ is separable, compact, connected, and locally path-connected? In this case, $X$ is a Peano continuum -- does every Peano continuum admit a length metric?

  3. What if we relax some of the conditions in (2)?

In particular, I might guess that every length space is locally path-connected, but it's not obvious that this is true. And I think the Hawaiian earring would be an example of a length space which is not locally contractible (or even semilocally simply connected).

For a bonus question:

  1. Is the topology of an Alexandrov space (in the curvature sense) significantly more controlled than the topology of a general length space?

1 Answer 1


1 Look at 3.

3 It is almost obvious from the definition that length spaces are locally path connected, and if you assume local connectedness, 2 is true, but nontrivial:

Bing, R. H. A convex metric for a locally connected continuum. Bull. Amer. Math. Soc. 55 (1949), no. 8, 812--819. https://projecteuclid.org/euclid.bams/1183514048

4 For finite dimensional Alexandrov Spaces, it is not hard to show that there is an open dense subset which is a manifold. It is nontrivial that there is an even nicer stratification, and they are locally contractible.


The topological structure results are very well surveyed in the last chapter here:

Burago, Dmitri & Burago, Yuri & Ivanov, Sergei. (2001). A Course in Metric Geometry. Graduate Studies in Math.. 33. https://www.math.psu.edu/petrunin/papers/akp-papers/bbi.pdf

  • $\begingroup$ Thanks! That Bing reference is just what I was hoping for! I'm a bit confused, though -- are you saying that length spaces are locally path connected? Or are you saying rather that it must be taken as an additional assumption? I still can't see why length spaces should be locally path connected, so if that's the claim, could you give a hint? $\endgroup$
    – Tim Campion
    Jun 15, 2019 at 0:26
  • $\begingroup$ They are: Take $p$ in a length space and an (open) ball $B$ of radius $r$ around $p$. Pick $q \in B$, then $d(p,q) < r$ and there is a curve of length less than $r$ joining $p$ and $q$. By the triangle inequality, this curve lies entirely in $B$, so $B$ is path connected. $\endgroup$ Jun 15, 2019 at 1:00
  • $\begingroup$ Thanks, I see that is indeed almost immediate! $\endgroup$
    – Tim Campion
    Jun 15, 2019 at 1:02
  • $\begingroup$ Upon closer inspection, Bing only shows that a compact, locally connected continuum has a convex metric if additionally it is finite-dimensional. I'd be curious if the finite-dimensionality hypothesis can be lifted. $\endgroup$
    – Tim Campion
    Jun 15, 2019 at 15:56
  • $\begingroup$ My bad, wrong reference projecteuclid.org/euclid.bams/1183514375 Theorem 8 $\endgroup$ Jun 19, 2019 at 4:07

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.