6
$\begingroup$

In a Hilbert space, let a sequence $(x_n)$ be weakly convergent to $x$ and be satisfied $\|x_{n+1}-x_n\|\to 0.$ I wonder if we can deduce that $(x_n)$ strongly converges to $x$, or at least it contains a subsequence strongly converges to $x$.

Thank you in advance for your help.

$\endgroup$
5
$\begingroup$

This is not true. In $\ell^2$, you can consider the following sequence: $$ e_1, \frac12 e_2, e_2, \frac12 e_2, \frac13 e_3, \frac23 e_3, e_3, \frac23 e_3, \frac13 e_3, \frac14 e_4, \frac24 e_4, \frac34 e_4, e_4, \frac34 e_4, \frac24 e_4, \frac14 e_4, \ldots$$ It is easy to check that this sequence has all desired properties.

Now, without a convergent subsequence: $$ e_1, e_1 + \frac12 e_2, e_1 + e_2, \frac12 e_1 + e_2, e_2, e_2 + \frac13 e_3, e_2 + \frac23 e_3, e_2 + e_3, \frac23 e_2 + e_3, \frac13 e_2 + e_3, e_3, \ldots $$

$\endgroup$
  • $\begingroup$ Thank you very much for your answer. So the sequence may not converge, however in your counterexample it still contains a convergent subsequence. $\endgroup$ – KTU Apr 23 '17 at 12:06
  • $\begingroup$ @KTU: Now I have written down a similar sequence without a convergent subsequence. $\endgroup$ – gerw Apr 23 '17 at 20:12
  • $\begingroup$ Very nice example +1 $\endgroup$ – JJR Apr 23 '17 at 21:28
  • $\begingroup$ @grew: very clear and interesting example. Many thanks! $\endgroup$ – KTU Apr 24 '17 at 1:53
0
$\begingroup$

Edited heavily using In Hilbert space: $x_n → x$ if and only if $x_n \to x$ weakly and $\Vert x_n \Vert → \Vert x \Vert$. and the fact that weak convergence is unique: If $$||x_{n+1}-x_n|| $$ converges to zero, then $(x_n)$ is Cauchy in H. So, $x_n$ converges to some $y$ in H. Using the other post, we have that $x_n$ converges weakly to $y$. But, weak convergence is unique. So, $x=y$ implying that $x_n$ converges to $x$ in H. Thus, $||x-x_n||$ tends to zero trivially giving us strong convergence.

$\endgroup$
  • $\begingroup$ earlier post: math.stackexchange.com/questions/2027861/… $\endgroup$ – Sean Nemetz Apr 23 '17 at 4:39
  • $\begingroup$ I don't see how you obtain the inequality $\|x-x_n\| \leq \|x-x_m\|$. $\endgroup$ – Jacky Chong Apr 23 '17 at 4:40
  • $\begingroup$ Oh, I see the problem, I dropped a limit. We had that $||x-x_n|| = ||x-x_{n+1} +x_{n+1} -x_n|| \leq ||x-x_{n+1}|| +||x_{n+1}-x_n||$ I took a limit and disregarded later. $\endgroup$ – Sean Nemetz Apr 23 '17 at 4:46
  • $\begingroup$ You still haven't said anything. I don't see how your inequality helps with your claim. $\endgroup$ – Jacky Chong Apr 23 '17 at 4:56
  • 1
    $\begingroup$ Thanks for your answer. However, $(x_n)$ is not a Cauchy sequence with just the property $\|x_{n+1}-x_n\|\to 0$. So the proof should be revised. $\endgroup$ – KTU Apr 23 '17 at 7:22

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.