4
$\begingroup$

Let $\{K_n\}_n$ be a sequence of compact subsets of a metric space $X$, and $K\subset X$ be compact. If $K_n$ Hausdorff converges to $K$, i.e.: $$ \lim\limits_{n\to\infty} d_{\mathrm H}(K_n,K) = \max\left\{\,\sup_{x \in K_n} d(x,K),\, \sup_{y \in K} d(K_n,y) \,\right\} = 0 , $$ then what can be said of the convergence of $I_{K_n}$ to $I_K$?

What I know: Indeed, since Hausdorff convergence strictly implies convergence of the upper Kuratowski limits then, by Convergence in compact-open topology on the Sierpiński space, we know that $I_{K_n}$ converges to $I_K$ in $C(X,\{0,1\})$ for the compact-open topology if $\{0,1\}$ has the Sierpiński topology.

What I expect: Since the Hausdorff topology is strictly finer than the Sierpiński topology then, is there a topology $\tau$ on $C(X,\{0,1\})$ for which: $$ I_{X-K_n} \overset{\tau}{\to} I_{X-K} \Leftrightarrow K_n \overset{d_H}{\to} K. $$

$\endgroup$
6
  • $\begingroup$ Note that if you use the functions $D_K(x):=\text{dist(x,K)}$ in place of $I_K(x)$, then the uniform norm $\|D_K-D_L\|_\infty$ is the Hausdorff distance $d_H(K,L)$ $\endgroup$ Commented Apr 23, 2021 at 8:39
  • $\begingroup$ @PietroMajer Thanks, actually, I did notice this, but I'm not sure how this would translate to indicator functions. $\endgroup$ Commented Apr 23, 2021 at 8:43
  • $\begingroup$ I'm a bit confused. According to the linked post, the Sierpinski topology is $\{ \emptyset, \{ 1 \}, \{ 0, 1 \} \}$. Wouldn't this mean that $\mathbf{1}_K$ is not actually continuous when $K$ is compact? (The preimage $\mathbf{1}_K^{-1}(\{ 0 \}) = X \setminus K$ is not closed, if $X$ is connected for example.) $\endgroup$
    – Leo Moos
    Commented Apr 23, 2021 at 11:22
  • $\begingroup$ @LeoMoos Yes, the identification is with $X-K_n$ and $X-K$. Thanks for noticing the typo :) $\endgroup$ Commented Apr 23, 2021 at 11:57
  • $\begingroup$ @ChristianRemling Yes, but I was looking for a metric directly on $C(X,\{0,1\})$ with a "intrinsic meaning"; in the sense that I shouldn't push-forward+pull-back via the homomorphism $C(X,\{0,1\})-> Comp(X)$. You know? $\endgroup$ Commented Apr 23, 2021 at 12:58

0

You must log in to answer this question.