0
$\begingroup$

Consider the Hilbert space $l_2$ with an equivalent norm

$$\Vert x \Vert = \max \{2 \Vert x \Vert_1, \Vert x \Vert_2 \},$$

where $\Vert x \Vert_1 =( \sum_{n=2}^\infty x_n^2 )^{\frac{1}{2}}$ and $\Vert x \Vert_2 = ( \sum_{n=1}^\infty x_n^2 )^{\frac{1}{2}}$, for $(x_n)_{n \geq 1} \in l_2$.

The unit ball with respect to norm $\Vert \cdot \Vert$ is $B_{\Vert \cdot \Vert}= Y \cap B_{(l_2, \Vert \cdot \Vert_2)}$, where $Y=\{x \in l_2 : \sum_{n=2}^\infty x_n^2 \leq \frac{1}{4} \}$. Then $(l_2, \Vert \cdot \Vert)$ has property (H) ( here, property (H) means the weak and norm convergence coincides on the unit sphere). Now, I need to find whether the dual of the above norm satisfies property (H).

My approach: First, I took $f_n=e_1+e_n$ and $f=e_1$. Then for $x=e_1 $, $f(x)=f_n(x)=1$. Also, $f \in S_{X^*}$; but $f_n \not \in S_{X^*}$, so the example fails to verify property (H) on the dual space. Next, I tried to prove that the dual has property (H). I took $f_n$ and $f$ on the dual sphere such that $f_n \xrightarrow{w} f$. We need to show: $\Vert f_n - f \Vert \to 0$. But I cannot understand how I can visualise this on the dual space. Kindly guide me.

$\endgroup$
5
  • $\begingroup$ What you call property (H) is usually called the Kadec-Klee property. $\endgroup$ Commented Aug 5, 2023 at 19:52
  • $\begingroup$ yes sir, you are right. $\endgroup$
    – PPB
    Commented Aug 7, 2023 at 6:35
  • $\begingroup$ The space is $\mathbb{F}\oplus \ell_2$, where $\ell_2$ has its usual norm and $\mathbb{F}$ is the scalar field. Moreover, the norm of a vector $(b,x)\in \mathbb{F}\oplus \ell_2$ only depends on $|b|$ and $\|x\|_{\ell_2}$. I suspect that once you figure out what's happening from the sequence $(1,e_n)$, $n=1,2,\ldots$, that will clarify the general case. $\endgroup$
    – user495577
    Commented Aug 18, 2023 at 13:42
  • $\begingroup$ Sorry. You might need the slightly more general $(1, \lambda e_n)$, $n=1,2,\ldots$, for an arbitrary $\lambda$. $\endgroup$
    – user495577
    Commented Aug 18, 2023 at 13:48
  • $\begingroup$ I have tried every possible by taking different values for $\lambda$ or changing the sequence $f_n$, but I failed to meet the assumption. The norm may satisfy property (H), but I do not understand how to contradict it. $\endgroup$
    – PPB
    Commented Aug 19, 2023 at 9:11

0

You must log in to answer this question.

Browse other questions tagged .