9

Looking for the proof of Eberlein-Smulian Theorem.

Searching for the proof is what I break with this morning. Some of my friends recommend Haim Brezis (Functional Analysis, Sobolev Spaces and Partial Differential Equations). After I search the book, I only found the statement of the theorem, is the proof very difficult to grasp? Why is Haim Brezis skip it in his book?

Please I need a reference where I can find the proof in detail.


Theorem:(Eberlein-Smul'yan Theorem) A Banach space $E$ is reflexive if and only if every (norm) bounded sequence in $E$ has a subsequence which converges weakly to an element of $E$.


  • 11
    -1: I quote from Brézis right after stating the theorem as Theorem 3.19: The proof of Theorem 3.19 is rather delicate and is omitted; see e.g. R. Holmes [1], K. Yosida [1], N. Dunford-J.T. Schwartz [1], Diestel [2], or Problem 10. What more do you want? – t.b. Jun 03 '12 at 09:55
  • 2
    @t.b.: Nothing. – Hassan Muhammad Jun 03 '12 at 09:59

3 Answers3

12

I made this answer CW, so that other people can add further references if they think it's suitable.

1

Kôsaku Yosida, Functional Analysis, Springer 1980, Chapter V, Appendix, section 4. (This appears to be the 6th edition).

0

This is a direct consequence of Banach-Alaoglu's theorem.

First note that if $X$ is reflexive then so is $X^{\ast}.$ Hence weak and weak$^{\ast}$-topology on $X^{\ast \ast}$ coincide. So by Banach-Alaoglu's theorem it follows that any bounded closed ball in $X^{\ast \ast}$ is weakly compact. Consequently any bounded sequence in $X^{\ast \ast}$ has a weakly convergent subsequence. Now since $X$ is reflexive, it follows that $X$ is isometrically isomorphic to $X^{\ast \ast}$ and hence any bounded sequence in $X$ has a weakly convergent subsequence as well.

ACB
  • 3,068
  • It's not quite so easy, compact spaces aren't always sequentially compact. You need to take it into a separable setting. And that's one direction. The other direction, that the closed unit ball being weakly sequentially compact implies reflexivity isn't obvious either. – Dermot Craddock Jun 10 '25 at 19:37
  • Sorry I did the "only if" part. The forward direction might not be that easy. – ACB Jun 10 '25 at 20:05