2
$\begingroup$

If $(X,d_X)$ is a compact metric space, and $(Y,d)$ is another metric space. Moreover, suppose that the metric capacity of $(Y,d)$ is at-least that of $(X,d_X)$, that is $$ \kappa_X(\epsilon)\leq \kappa_Y(\epsilon) ; (\forall \epsilon \in (0,1] . $$

Here the metric capacity of a metric space $(X,d_X)$ is defined by $$ \kappa_X(\epsilon)\triangleq\sup\left\{ k : \exists x_0,\dots,x_k \in X : \exists r>0, \sqcup_{i=1}^k B(x_i;\epsilon r)\subseteq B(x_0;r) \right\} $$

Does there necessarily exist a continuous injection of $(X,d_X)$ into a metric ball in $(Y,d)$ of some (finite) radius?

Note:

  • I'm not looking for Lipschitz, just continuous inclusion.
  • $X,Y\neq \emptyset$.
$\endgroup$

1 Answer 1

4
$\begingroup$

The answer is no. You can have $X$ path connected and $Y$ totally disconnected satisfying your condition. Then any continuous map from $X$ to $Y$ is necessarily constant so there are no injective maps from $X$ to $Y$.

Here is another example: $X=S^1$ is a unit circle of with the geodesic metric and $Y=\mathbb{R}$. Since $X$ and $Y$ are locally isometric, it follows that $$ \kappa_X(\epsilon)\leq \kappa_Y(\epsilon) ; (\forall \epsilon \in (0,1]). $$ Indeed, if $r\le\pi$ in the definition of $\kappa$ the corresponding quantities are equal (balls of radius $r$ are isometric balls). If $r>\pi$, then the quantity in the definition of $\kappa_X(\epsilon)$ is less than the corresponding one in the definition of $\kappa_Y(\epsilon)$ (ball in $Y$ is bigger - segment of length $2r$ than the ball in $S^1$ - the unit circle and they both are locally isometric).

However there is no injective map from $X=S^1$ to $Y=\mathbb{R}$.

$\endgroup$
3
  • $\begingroup$ Good point, I noticed this this morning also ... In that case if Y is path connected, say Y was a larger vectorspace, then would it work? $\endgroup$
    – ABIM
    Commented Jan 19, 2019 at 8:41
  • $\begingroup$ @AIM_BLB You might still have a positive answer is you take $Y=\mathbb{R}^n$ and replace your inequality with $\kappa_X(\epsilon)\leq C\kappa_Y(\epsilon)$. I need to think about it. I suggest that you ask another questions along this line and let me know (just leave a comment here with a link to a new question) then I can try to answer it. $\endgroup$ Commented Jan 19, 2019 at 16:45
  • $\begingroup$ Here's is the link to the part 2 of the question, with the suggested modifications, mathoverflow.net/questions/321367/… $\endgroup$
    – ABIM
    Commented Jan 21, 2019 at 9:29

You must log in to answer this question.

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