1

I'm doing Ex 3.1 in Brezis's book of Functional Analysis.

Let $(E, | \cdot |)$ be a normed space and let $A \subset E$ be a subset that is compact in the weak topology $\sigma\left(E, E'\right)$. Prove that $A$ is bounded.

I have found a proof by Principle of Uniform Boundedness here. I'm lucky to find my own proof :v

Could you have a check on my attempt?

I posted my proof separately so that I can accept my own answer and thus remove my question from unanswered list. If other people post answers, I will happily accept theirs.

Analyst
  • 5,637

2 Answers2

1

One can prove a stronger result without resorting to the UBP (but an appeal to Baire's category theorem is needed): let $X$ denote a lc tvs and let $K$ be a $\sigma(X,X')$-compact in $X$. Then $K$ is bounded in the original topology (let us call it $\tau$)

To prove the result, we proceed as follows: Given $V$ a $\tau-$neighbourhood of $0$, let $I$ be a absolutely convex $\tau$-neighbourhood of $0$ such that $\overline{I}\subset V$. By the bipolar theorem, $(I^\circ)_{\circ}=\overline{I}$. Since $K$ is $\sigma(X,X')$ compact every linear functional is bounded on $K$. Moreover $I^\circ$ is $\sigma(X',X)$ compact by Banach-Alaoglu so an application of the Baire category theorem implies that $$M:=\sup_{f\in I^\circ} |f(K)|<\infty$$ so now we must have $$\frac 1M K\subset {I^\circ}_\circ=\overline I\subset V$$ proving the claim.

0

We have $| \cdot|$ is l.s.c. in norm topology and convex, so $| \cdot|$ is l.s.c. in weak topology. Then $|\cdot|$ is bounded on the weakly compact set $A$.

Analyst
  • 5,637
  • @daw Ah it's a typo. It should be $| \cdot|$. – Analyst Feb 21 '22 at 14:54
  • 1
    Imho lsc only implies boundedness from below. – daw Feb 21 '22 at 14:56
  • @daw You're right. Thank you so much! I made a silly mistake :v – Analyst Feb 21 '22 at 14:58
  • 1
    but one learns from mistakes :) – daw Feb 21 '22 at 14:58
  • 1
    I think (but I might be wrong) that any proof of this result must use, at some point, some form of the Baire category theorem (the UBP or some other result), so this kind of straightforward approch should not work. Of course, I am not sure of this and I would be very happy to be proved wrong! –  Feb 21 '22 at 16:14