I'm reading Royden Chap 15. The last remark in section 1 reads:

For $X = l^{\infty}$, $B^*$, the closed unity ball of $X^*$, is not weak-* sequentially compact. Indeed the sequence $\{\psi_n\}\subset B^*$ defined for each $n$ by $$\psi_n(\{x_k\})=x_n \qquad \forall \{x_k\}\in l^{\infty}$$ fails to have a weak-* convergent subsequence.

I am having trouble seeing how $\{\psi_n\}$ has no weak-* convergent subsequence. I'm having trouble even stating what I want to show. I think what I want to show is: $$\exists (c>0 \text{ and }\hat{x}\in X^{**}) \, \forall(n,m\in\mathbb{N})\text{ s.t. }c<|\hat{x}(\psi_n)-\hat{x}(\psi_m)|$$


Let's start by writing down what it would mean to have a weak$^*$-convergent subsequence of $\{\psi_n\}_{n \geq 1}$.

We would then have an increasing sequence $n_k$ and $\psi \in X^*$ such that for every $x \in \ell^\infty$, $\psi_{n_k}(x) \to \psi(x)$ as $k \to \infty$. Note that $\psi_{n_k}(x) \to \psi(x)$ if and only if $x_{n_k} \to \psi(x)$.

It is then not too hard to see that this cannot happen, since you can take $x$ to be the sequence $$x_n = \begin{cases} (-1)^k \qquad n = n_k \\ 0 \qquad \text{otherwise} \end{cases}$$ and then $\psi_{n_k}(x) = (-1)^k$ does not even converge.


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.