2
$\begingroup$

It is well known that, given a Banach space $Y$ and its dual $Y^*,$ a subset $G$ of $Y^*$ is $w^*-$compact iff is $w^*-$closed and norm- bounded.

Could you show an example of a normed space $Y$ and a $w^*-$compact set $G\subseteq Y^*$ such that $G$ is unbounded in the dual norm? Of course, $Y$ cannot be Banach.

$\endgroup$
5
  • $\begingroup$ Depends on what you mean by "bounded". Every weak$^\ast$ compact set is weak$^\ast$ bounded. And if $Y$ is not a normed space, there is no norm topology on $Y^\ast$. $\endgroup$
    – MaoWao
    Feb 26, 2018 at 12:36
  • $\begingroup$ @MaoWao I will edit the question. You have to assume that Y is a normed space. Boundedness of $G$ is understood in the sense of the dual norm $\endgroup$
    – John D
    Feb 26, 2018 at 12:55
  • $\begingroup$ If $Y$ is a normed space, then $Y^*$ is (isometrically isomorphic to) the dual of the completion of $Y$. Hence the anwer for normed spaces does not differ from the one for Banach spaces. $\endgroup$
    – MaoWao
    Feb 26, 2018 at 12:57
  • 2
    $\begingroup$ @MaoWao: I first thought so too. But the point is that the weak-$\ast$-topologies induced by $Y$ and by its completion differ. See my answer if you are interested. $\endgroup$
    – PhoemueX
    Feb 26, 2018 at 12:59
  • $\begingroup$ @PhoemuX: Yes, I realized that just as you posted your answer. $\endgroup$
    – MaoWao
    Feb 26, 2018 at 12:59

1 Answer 1

3
$\begingroup$

Take $Y = \ell_0 (\Bbb{N})$ to be the space of all finitely supported sequences, equipped with the $\ell^\infty$-norm.

Take $\varphi_N : Y \to \Bbb{C}, (x_n)_n \mapsto N \cdot x_N$. It is easy to see $\varphi_N (x) \to 0$ for each fixed $x \in Y$ (in fact, $\varphi_N (x) = 0$ for $N \geq N(x)$ large enough).

But since $x_N = \delta_N \in Y$ with $\|x_N\|_{\ell^\infty} = 1$, we have $\|\varphi_N\| \geq |\varphi_N (x_N)| = N$, so that the set $G := \{\varphi_N : N \in \Bbb{N}\} \cup \{0\}$ is not bounded.

Finally, since $\varphi_N \to 0$ in the weak-$\ast$-topology induced by $Y$, it is not hard to see that $G$ is indeed weak-$\ast$-compact. In fact, if $(U_i)_i$ is an open covering of $G$, then $0 \in U_i$ for some $i$. Since $\varphi_N \to 0$, this implies $\varphi_N \in U_i$ for all but finitely many $N$. From here, it is not hard to extract a finite subcover.

$\endgroup$

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .