I am practicing some elementary problems from functional analysis. I am trying to show when X is a normed space then the closed unit ball of its dual $ X^* $ is weak-star closed.

I can show whenever a sequence weak-star converges in the closed unit ball of X^* then the weak-star limit is also in the closed unit ball. Is this enough to show that the closed unit ball of $ X^* $ is weak-star closed?


  • $\begingroup$ You have to be a little careful using sequences in a general topological space. In general, being closed implies that the limits of a sequence from the set all lie in the set itself, but the converse does not necessarily hold. $\endgroup$ – Theo Bendit Apr 7 '18 at 3:37
  • $\begingroup$ Yeah, that is why I am a bit cautious on this approach. I'm not sure whether or not a set containing all of its weak-star limit points implies that set is weak-star closed. $\endgroup$ – Carey Washington Apr 7 '18 at 5:06

By Banach-Alaoglu Theorem, the closed unit ball in $X^{\ast}$ is weak-star compact. Since the weak-star topology is Hausdorff, the closed unit ball is weak-star closed.

  • $\begingroup$ I am aware of this way to show that the closed ball of $ X^* $ is weak-star closed but I am trying to prove this w/o Alaoglu's theorem! Thanks! $\endgroup$ – Carey Washington Apr 7 '18 at 3:28
  • $\begingroup$ Have to think about it how to avoid that. $\endgroup$ – user284331 Apr 7 '18 at 3:29

We can just write $$\tag{1}\overline{B_{X^*}(0)} = \bigcap_{x \in \overline{B_1(0)}} \{x^* \in X^* : |x^*(x)| \leq 1 \}.$$ By definition of the weak*-star topology, we know that $x^* \mapsto x^*(x)$ is continuous. So all sets are closed and thus also the intersection. Hence the $X^*$-unit ball is closed.

In order to see the equality of the sets in (1), we note that "$\subset$" is clear by definition of the operator norm on $X^*$; we awlays have $|x^*(x)| \leq |x^*| |x| \leq 1$. On the other hand, if $|x^*(x)| \leq 1$ for all $x \in \overline{B_1(0)}$, then $|x^*| \leq 1$.

For separable banach spaces it is true that any weak*-sequentially convex closed set is already closed.

However, there are sequentially closed sets in $l^\infty$, which are not closed. Example: Let $\delta_n$ denote the operator $\delta_n(f) = f(n)$, then $$C:=\{\delta_n : n \in \mathbb{N}\} \subset (l^\infty)^*$$ is sequentially closed. Let $\delta_{n_k} \rightarrow h$ in weak*-topology. If $(n_k)$ is unbounded, we may pass over to a subsequence with $n_k \rightarrow \infty$. In this case let $f \in l^\infty$ such that $f(n_k)$ is not convergent. Then $\lim_{n \rightarrow \infty} f(n_k) = h(f)$. That is a contradiction! So $(n_k)$ have to be already eventually constant.

Since $C \subset \overline{B_1(0)}$ is a subset of compact set, the sequence $(\delta_n)_{n \in \mathbb{N}}$ has limit points!


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.