3

Consider $S$ to be the subspace of $L^\infty(\mathbb R)$ consisting of all functions supported on finite measure sets and $T:S\to B$ is a linear continuous operator between $S$ and some Banach space $B$, is there always a linear continuos mapping $F:L^\infty(\mathbb R)\to B$ extending $T$?

Note that $S$ is not dense in $L^\infty(\mathbb R)$ and if $B=\mathbb R$, this is just Hahn-Banach. But what if $B$ is a general Banach space? What if $B=L^p(\mathbb R)$ for any $p\in [1,\infty]$? I believe we should use Hahn-Banach, but I don't know how, could any one hint?

Ottavio
  • 2,287
stephenkk
  • 977

1 Answers1

2

I am not sure about the general case, but suspect it impossible.

If $B$ admits a Markushevich basis $\{(f_\alpha,x_\alpha)\}_\alpha$, then, for each $\alpha$, the functional $f_\alpha T$ admits a norm-preserving extension $g_{\alpha}\in L^{\infty}(\mathbb{R})\to B$. Moreover, since $\sum_\alpha{f_\alpha T}=T$ converges in the strong operator topology, by the principle of uniform boundedness, $\sum_\alpha{g_\alpha}$ converges in that same topology. It is easy to see that said sum is the desired extension of $T$.

The above almost entirely handle the particular cases of $B=L^p(\mathbb{R})$. For $p<\infty$, the Haar wavelets on $[n,n+1]$ (for all $n$ and all frequencies) form a Schauder (and thus Markushevich) basis for $L^p$.

  • Why T bounded implies weak star to weak star continuous of T? I don't know why after reading this: here:https://math.stackexchange.com/questions/55679/weak-to-weak-continuous-operator-which-is-not-norm-continuous – stephenkk Jul 05 '22 at 01:39
  • @stephenkk: Yes, this version of my answer is mistaken. (Although I feel compelled to point out that a more apropos link is here.) – Jacob Manaker Jul 05 '22 at 04:20
  • Funnily enough, the question you linked is precisely what I needed to fix the flaw. – Jacob Manaker Jul 05 '22 at 05:03
  • Dear @Jacob Manaker, thanks for your answer. I still don't understand the use of double subsequence trick to prove g is unique. Could you please elaborate? – stephenkk Jul 07 '22 at 01:09
  • @stephenkk: Yes, I seem to have gotten my techniques mixed up. I've removed that section, since I can't seem to find a way to fix it. Sorry. – Jacob Manaker Jul 07 '22 at 19:40