2
$\begingroup$

This is indeed meant to be a homework question but I just figured out the answer I prepared was false all this time. Kind of panicking, could anyone give me a lift?

$T:\ell^2 \rightarrow \ell^1: Tx = \big( (2^{-k})x_k \big)_{n=1}^{\infty} $

Thanks!

Edit: Here are my thoughts:

Observe that for all $ x \in \ell^2 $ in the unit ball we can attend the bounds $ ||Tx||_{2} = \frac{1}{2}, \frac{1}{4}, \frac{1}{8},... $ Another such bound can be achieved by the series $ (\frac{1}{2^{k/2}})_{j \in \mathbb{N}} $ which goes to $1$. However looking for an upper bound we have

$$ || Tx ||_{1} = \sum_{k=1}^{\infty} | \frac{x_k}{2^k} | $$ which I think naturally should be applied Holder's inequality. I have tried various but non of them would coincide with the lower bounds I gave above.The best I could do was $ 1/\sqrt{3} $ for which I cannot get a lower bound for.

$\endgroup$
  • $\begingroup$ No. show your thoughts $\endgroup$ – mathworker21 Mar 28 '18 at 8:42
  • $\begingroup$ Attached are my thoughts. @mathworker21 I made a small mistake for which I thought the bounds would coincide some days ago and I just noticed its wrong in the very last seconds before it is due lol $\endgroup$ – Meagain Mar 28 '18 at 8:55
  • $\begingroup$ see my answer below $\endgroup$ – Guy Fsone Mar 29 '18 at 14:14
2
$\begingroup$

About Hölder and your upper bound, OK: $$ \|Tx\|_1 = \sum_{k=1}^\infty\left|\frac{x_k}{2^k}\right|\le \left(\sum_{k=1}^\infty 2^{-2k}\right)^{1/2} \left(\sum_{k=1}^\infty x_k^2\right)^{1/2} = \frac1{\sqrt 3}\|x\|_2. $$ About finding $x$ with $\|Tx\|_1\approx\frac1{\sqrt 3}\|x\|_2$ (equality can be impossible), see the conditions for equality in the Hölder inequality.

$\endgroup$
  • $\begingroup$ Never remembered the equality condition. Learned simething new, quiet happy, thanks! $\endgroup$ – Meagain Mar 28 '18 at 11:17
  • $\begingroup$ @Meagain you check the answer below $\endgroup$ – Guy Fsone Mar 29 '18 at 14:14
  • $\begingroup$ @GuyFsone, is OK. Equality iff linear dependence. $\endgroup$ – Martín-Blas Pérez Pinilla Mar 29 '18 at 14:27
0
$\begingroup$

Using Cauchy-Schwarz inequality we have

$$\|Tx\|_1=\sum_{k=1}^\infty\left|2^{-k}x_k\right|\leq\left(\sum_{k=1}^\infty 4^{-k}\right)^{1/2}\cdot \left(\sum_{k=1}^\infty|x_k|^2\right)^{1/2}\ = \frac1{\sqrt3}\|x\|_2.$$

Hence $$\|T\| = \sup_{\|x\|_2=1}\|Tx\|_1\le \frac{1}{\sqrt 3}$$

Now let $x = (x_k=\frac{1}{2^k})_k$. then $$\|x\|_2= \left(\sum_{k=1}^\infty 4^{-k}\right)^{1/2} = \frac{1}{\sqrt 3}$$

and $Tx = (\frac{1}{4^k})_k$ $$\|Tx\|_1= \sum_{k=1}^\infty 4^{-k}=\left(\sum_{k=1}^\infty 4^{-k}\right)^{1/2} \left(\sum_{k=1}^\infty 4^{-k}\right)^{1/2} = \frac{1}{\sqrt 3}\|x\|_2$$

It readily follows that $$\|Tx\| = \frac{1}{\sqrt 3}$$

$\endgroup$

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.