Given a nonempty subset S of a metric space $(M,d)$, define $f:M\rightarrow \mathbb{R}$ by $f(x)=\inf \{ d(x,s):s\in S\}$

(a) Show that $x\in $ Clos$(S)$ if and only if $f(x)=0$.

(b) Show that $|f(x)-f(y)|\le d(x,y)$ for all $x,y\in M$ and conclude that $f$ is continuous.

Clos$(S)$ = $\{ x\in M \mid$ Every ball $B(x,r)$ intersects $S\}$

$B(x,r) = \{ y\in M \mid d(x,y)<r\}$

Working on (a) I am first trying to show that $x\in$ Clos$(S)$ implies $f(x)=0$. So assuming $x\in$ Clos$(S)$ then ...

is it correct to say that: there exists a ball, $B(x,r)$ that intersects $S$. Then $d(x,y)<r$ for all $r$ so $f(x)=0$.

Then my attempt at the converse. Assume $f(x)=0$. Then $d(x,s)=0$ for some $s\in S$. This implies that $x=s$ so $x\in S$ since $s\in S$. So $x\in$ Clos$(s)$.

Are my statements correct?

As for part (b) I am not entirely sure where to start, but I wanted to be sure I had part (a) first in the event that I can make use of it for (b).


I am afraid what you wrote is not much enough for me to evaluate or say something. Instead, I directly tackle the problems.

If $y \in M$ and if $\varepsilon > 0$, then $d(x,y) < \varepsilon$ only if $$ |f(x) - f(y)| \leq d(x,y) < \varepsilon; $$ hence continuity follows. This solves the second problem.

For the first question, we prove the "only if" part first. If $x \in \overline{S}$, then for every $\varepsilon > 0$ we have $B_{M}^{x}(\varepsilon) \cap S \neq \varnothing$, where $B_{M}^{x}(\varepsilon)$ denotes the open ball over $M$ of center $x$ and radius $\varepsilon$. If $f(x) \neq 0$, then $\inf_{s \in S}d(x,s) \neq 0$; hence if $p := \inf_{s \in S}d(x,s)$, then $B_{M}^{x}(p/2) \cap S = \varnothing$, a contradiction. Conversely, suppose $x \in M$ and $f(x) = 0$. Then $p = 0$, implying that for every $\varepsilon > 0$ there is some $s \in S$ such that $p = 0 \leq d(x,s) < p + \varepsilon = \varepsilon$; hence $B^{x}_{M}(\varepsilon) \cap S \neq \varnothing$, so $x \in \overline{S}$.


Your first statement is correct: by definition, $x \in \text{Clos}(S)$ iff for any $r > 0$, $B(x,r) \cap S \neq \emptyset$, so you can take $s$ from balls with $r$ arbitrarily small. For the converse, it may be the case that no $s \in S$ satisfies $d(x,s) = 0$. Rather, you can find a sequence $\{s_n\}_n \subset S$ such that $d(x,s_n) < \frac{1}{n}$, say. Do you see how to use the definition of closure I just stated to conclude?

Also, for b), you should show that for each $s$, $d(x,s) \leq d(x,y) + d(y,s)$, so this is also true for $s$ that "minimizes" $d(y,s)$, i.e., find a sequence $\{s_n\}_n$ such that for $n$ sufficiently large, $d(y,s_n) < f(y)+\epsilon$. Thus, $d(x,s_n) < f(y) + \epsilon + d(x,y)$. You can do the same with $x$ first by symmetry. Do you see how to use the definition of $f$ to conclude in this problem?


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.