Inspired by the old question I considered the following

Let $$F=\{f\in C^1([0,1]):~~ |f(x)|\leq 1 ~~\text{and}~~ 1\leq f'(x)\leq 2 ~~\text{for all }x\in [0,1]\}$$ the set of continuous, bounded by $-1$ and $1$, functions $f:[0,1]\to \mathbb R$ with bounded, by $1$ and $2$, continuous first derivative.

In this case I can not apply the trick, which is applied there, and I guess $F$ is compact in this case, but I can not prove it yet.

Is my guess right or rather $F$ is only pre-compact (subset closure of which is compact)?


Let $X$ be a compact metric space, $(Y,d)$ any metric space and $F$ a subset of $C(X,Y)$. Then $\operatorname{cl}(F)$ is compact iff the following two conditions are valid:

  • $F$ is equicontinuous
  • for each $x\in X$, the subset $F_x=\operatorname{cl}\left(\{f(x):~f\in F\}\right)$ is a compact subspace of $Y$.

where $\operatorname{cl}(A)$ is a closure of $A$.

Is $F$ closed subset of $C([0,1])$?

If $F$ is closed one can apply Arzela-Ascoli theorem to show that $F$ is compact, since boundedness of $F_x$ and equicontinuity of $F$ are clear in this case.

Lemma: If $F$ is an equicontinuous subspace of $C(X,Y)$ then so is $\operatorname{cl}(F)$.

By this lemma $F_x=\operatorname{cl}\left(\{f(x):~ f\in F\}\right)=\operatorname{cl}\left(\{f(x):~ f\in \operatorname{cl}(F)\}\right)$, am I right?

  • $\begingroup$ why is $F_x$ included in $[0,1]$? $\endgroup$ – supinf Nov 22 '19 at 12:19
  • $\begingroup$ Oh mistype, changed. $\endgroup$ – Evgeny Kuznetsov Nov 22 '19 at 12:36

Hint: It is possible to show that $$ F_x = [x-1,x] $$ explicitly. To gain intuition of this fact it might help to think about what functions are in $F$ what and functions are not in $F$.


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.