I have to prove that a closed unit ball in $C[0,1]$ is not weak-compact. The hint is that I should consider sets: $$V_t=\{f\in C[0,1]:|f(t)|>1/3\}$$ and $$U_t=\{f\in C[0,1]:|f(t)|<2/3\}$$ Now I should show that $$\{V_t:t\in \mathbb{Q}\cap[0,1]\} \cup \{U_t:t\in (\mathbb{R} \setminus \mathbb{Q})\cap[0,1]\}$$ is an open cover of a closed unit ball in weak topology such that we can't choose a finite subcover.

Weak topologies are sth new to me. Can anyone help me? I did only manage to show, that $\phi_t:C[0,1]\ni f \rightarrow f(t)\in K$ are bounded linear functionals.


As you said, $\phi_t$ is a bounded linear funcional, so as

$$V_t = \phi_t^{-1} \big( \{ |x| > \frac 13 \}\big), $$

then $V_t$ is open in the weak topology. Similarly $U_t$ is also open. It is quite obvious that

$$\{V_t:t\in \mathbb{Q}\cap[0,1]\} \cup \{U_t:t\in (\mathbb{R} \setminus \mathbb{Q})\cap[0,1]\}$$

is an open cover of the unit ball in $C[0,1]$ (Actually, it is an open cover of the whole $C[0,1]$).

Now we show that: for any finite set $t_1, \cdots t_m \in \mathbb Q\cap [0,1]$, $s_1, \cdots s_n \in \mathbb R\setminus \mathbb Q \cap [0,1]$, there is a function $f \in C[0,1]$ such that $||f||\leq 1$ and

$$f\notin \bigcup_{j} V_{t_j} \cup \bigcup_{k} U_{s_k}$$

Well, it just say that I can find a continuous function $f$ such that $|f(t_j)|\leq 1/3$ (so not in $V_{t_j}$) and $|f(s_k)| \geq 2/3$ (so not in $U_{s_k}$) and $||f|| \leq 1$. Such a function can be found by setting

$$f(t_j) =1/3, \ \ \ \ f(s_k) = 2/3$$

and joining line along adjacent $t_j$, $s_k$. This shows that we cannot pick a finite subcover to cover the closed unit ball in $C[0,1]$, thus the unit ball is not weakly compact.


Although, you want to prove directly, it easily follows from Kakutani's theorem which states that

For a Banach space $X$,

$X$ is reflexive$\iff$ The closed unit ball of $X$ is weakly compact.

But if $X$ is infinite, $C(X)$ is not reflexive. As a result, $C([0,1])$ is not reflexive, So its closed unit ball can not be weakly compact.

  • $\begingroup$ Should not we require $X$ infinite compact (as a metric space)? $\endgroup$ – Federico May 21 '14 at 21:09
  • $\begingroup$ @Federico I can not remember anything right now! should be? $\endgroup$ – Fermat May 21 '14 at 21:22
  • 1
    $\begingroup$ I think yes. I remember a result in my lecture notes that says "If $(K, d)$ is a compact infinite metric space, then $C(K)$ is not reflexive". I'm not sure, however, this is the most general rule. Now it's late for me, but I'll try to check out tomorrow! $\endgroup$ – Federico May 21 '14 at 21:29
  • 1
    $\begingroup$ I have not been successful in finding the answer to the question "what if we drop compactness?", so I asked a question: math.stackexchange.com/questions/810425/… Hope could be useful! $\endgroup$ – Federico May 27 '14 at 13:44

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.