0
$\begingroup$

Can you please help me solve the question below? I have no idea how to prove this one.

Define the set $$X:=\{K\subset\mathbb C:K\text{ is bounded and closed}\}$$ Define a function $d\colon X \times X \to \mathbb{R}$ via $$ d(K_1,K_2)=\inf\{\delta>0:K_1\subset N_\delta(K_2)\text{ and }K_2\subset N_\delta(K_1)\}$$ where $$ N_\delta(K):=\bigcup_{y\in K}N_\delta(y)=\{x\in\mathbb C:\exists y\in K\text{ with }|x-y|<\delta\}.$$

i) Show that $d$ defines a metric on $X$.

ii) Is $d$ still a metric if $X$ contains all bounded sets in $\mathbb C$? All closed sets?

$\endgroup$
  • $\begingroup$ $d:X \to X$ or $d:X\times X \to \mathbb{R}$ $\endgroup$ – Kunnysan Jul 14 '13 at 21:16
  • $\begingroup$ Where are you stuck? Have you tried to show any of the metric axioms? Most of them should be fairly easy (non-negativity, non-degeneracy, and symmetry). $\endgroup$ – icurays1 Jul 14 '13 at 21:17
  • 1
    $\begingroup$ Oh! My bad. I solved i). Part ii is the problem! $\endgroup$ – InfimumMaximum Jul 14 '13 at 21:21
3
$\begingroup$

This is called Hausdorff distance, by the way.

i) The problem statement is wrong. You must exclude the empty set from $X$., ie. I shall work with $X:=\{K\subset\mathbb C:K\text{ is nonempty and compact}\}$. Also, $d$ is not a map $X\to X$ but rather $X\times X\to\mathbb R$ as required for a metric.

You need to show the properties of a metric:

  • $d(K_1,K_2)\ge 0$: Since both sets are bounded and nonempty, the set we take the infimum of is nonempty, hence the infimum is $<\infty$. As all $\delta$ in the set are $>0$, the infimum is also $\ge 0$.
  • $d(K_1,K_2)=0\Rightarrow K_1=K_2$: Assume $K_1\ne K_2$ and wlog $x\in K_2\setminus K_2$. Since $K_2$ is closed, there is an $\epsilon$ neighbourhood of $x$ such that $B_\epsilon(x)\cap K_2=\emptyset$. Then $x\notin N_\delta(K_2)$ if $\delta<\epsilon$ and hence $d(K_1,K_2)\ge \epsilon$.

  • $d(K_1,K_3)\le d(K_1,K_2)+d(K_2,K_3)$: Note that $N_{\delta_1+\delta_2}(K)\subseteq N_{\delta_1}(N_{\delta_2}(K))$. Use this and write down the exlicit condition we want to show and you are done.

ii) Let $A=\{z\in\mathbb C:|z|<1\}$ and $B=\{z\in\mathbb C:|z|\le 1\}$. These are bounded sets with $d(A,B)=0$ and $A\ne B$. Hence this is no longer a metric.

Let $A=\mathbb R$ and $B=i\mathbb R$. These are closed sets with $d(A,B)=\infty$, hence this is no longer a metric.

$\endgroup$
  • $\begingroup$ I'm not sure about how you proved d(K1, K2) >= 0. Can you explain it one more time please? $\endgroup$ – InfimumMaximum Jul 17 '13 at 5:12
  • $\begingroup$ Maybe if K1 is not equal to K2, then K2-K1 is nonempty. Then for x in K2-K1, we can say d(K1,K2) >= d(x,K1) > 0. Is this correct? (though I'm not sure if I can just say d(x,K1) > 0...) $\endgroup$ – InfimumMaximum Jul 17 '13 at 5:14

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.