Kreyzig proves the following lemma:

Let $(x_n)$ be a weakly convergent sequence in a normed space $X$, then the weak limit $x$ of $(x_n)$ is unique.

He begins by supposing $(x_n)$ converges weakly to two limits, say $x$ and $y$. By supposition, this means $f(x_n) \rightarrow f(x)$ and $f(x_n) \rightarrow f(y)$. $f$ is linear. So $$f(x) - f(y) = f(x-y) = 0$$

Then Kreyzig uses Hahn-Banach theorem to assert this implies $x-y = 0$. I don't understand why this doesn't just follow from $f$ being linear, don't linear maps between normed spaces need to preserve 0 as they do for inner product spaces?

  • 4
    $\begingroup$ Yes, linear maps preserve $0$, but the crux is the other direction. Not every $z$ that is mapped to $0$ must itself be $0$. But by Hahn-Banach, every $z$ that is mapped to $0$ by all continuous linear functionals must itself be $0$. $\endgroup$ – Daniel Fischer Feb 13 '17 at 11:36
  • $\begingroup$ @DanielFischer thanks! $\endgroup$ – yoshi Feb 13 '17 at 16:23

If $x-y$ is not zero, you can define on $vect(x-y)$ a linear function $g$ such that $g(x-y)=1$. Hahn Banach implies that you can extend $g$ to a bounded function $f$ defined on $X$.You have $f(x-y)=g(x-y)=1$. Contradiction.

  • $\begingroup$ What do you mean by $vect()$? $\endgroup$ – mavavilj Apr 9 '18 at 8:01

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.