2

By the discrete Young inequality, we know that $\|a_n * b_n\|_q \leq \|a_n\|_p \|b_n\|_r$ when $1/q+1= 1/p+1/r$.

My question is that if there exists $C>0$, such that $\|a_n * b_n\|_q \leq C \|a_n\|_p $ is true for any $a_n \in \ell^p$, can we have $q\geq p$ and $b_n\in \ell ^r$?

Any idea will be helpful. Thanks a lot.

yufeng lu
  • 467
  • 2
  • 10
  • Young's inequality is truly beautiful. However, I have a problem understanding your question.
    1. I guess we fix $p\geq1$.

    2. Do you want this? If there is $C>0$ such that

    $$|(a_n)*(b_n)|{\ell^r}\leq C|(a_n)|{\ell^p}$$ for all $(a_n)\in\ell^p$ and all $(b_n)\in\ell^q$ with $|(b_n)|_{\ell^q}\leq1$ then $$\dfrac{1}{r}+1=\dfrac{1}{p}+\dfrac{1}{q}$$

    – AD - Stop Putin - Dec 01 '21 at 13:05
  • Yes, thanks for your comment. This is exactly what I mean. – yufeng lu Dec 02 '21 at 01:33

1 Answers1

0

This is false.

Take $(a_n)\in\ell^{p}$ and $(b_n)\in\ell^{q}$ where $p,q\geq 1$, then by Youngs inequality $$\|(a_n)*(b_n)\|_{\ell^r}\leq \|(a_n)\|_{\ell^p}\|(b_n)\|_{\ell^q}$$ provided $$\dfrac{1}{r}+1=\dfrac{1}{p}+\dfrac{1}{q}$$ and then $$\|(a_n)*(b_n)\|_{\ell^s}\leq\|(a_n)\|_{\ell^p}\|(b_n)\|_{\ell^q}$$ whenever $s>r$ (see How do you show monotonicity of the $\ell^p$ norms?), even though $$\frac1s +1< \frac1p+\frac1q$$

  • Sorry, maybe I misunderstood your comment before. My question is that if the convolution is true for any $a_n\in \ell^p$, what is the optimal $q$ such that $b_n \in \ell^q$. – yufeng lu Dec 09 '21 at 12:46