1
$\begingroup$

I am reading a paper on PDEs and I has been struggling trying to understand some specific identities (that should be very easy). Let me introduce the main notation of the book, which says that each time a sum like $\sum_\lambda f(\lambda)$ appears, then the $\lambda$-variable must to be understood as a dyadic variable, that is, the previous sum has to be understood as $$ \sum_{k=0}^{\infty} f(2^k). $$ This will also apply to $\mu$ below. Now, just after that, in the proof of a basic lemma the author writes the following sequence of equalities: $$ \sum_\lambda \lambda^s\sum_{\mu\geq \lambda/8} a_{\mu}d_{\lambda}=\sum_\lambda \lambda^s\sum_{\substack{k=-3,\\ 2^k\lambda\geq1}}^{\infty}a_{2^k\lambda}d_{\lambda}=\sum_{k=-3}^\infty\sum_{\lambda\geq 2^{-k}} a_{2^k\lambda}d_\lambda $$ where $\{a_\lambda\}$ and $\{d_\mu\}$ are two sequences of real numbers indexed by dyadic integers and $s>0$. I don't really understand these identities for two reasons. First, I don't see how to change the $\mu$-variable to the index $2^k\lambda$ appearing in $a$ (I would just put $a_{2^k}$ withouth the $\lambda$). But then, it seems confusing to me that the inequality in the inner sum becomes $2^k\lambda\geq 1$, particularly because in the left-hand side $\lambda$ is bounded from above, and in the other sums $\lambda$ seems to be bounded from below. Does anyone has any explanation?

Edit: The idea of the lemma is actually to prove the following inequality: $$ \sum_\lambda \lambda^s\sum_{\mu\geq \lambda/8} a_{\mu}d_{\lambda} \leq C \left(\sum_\lambda \lambda^{2s}a_\lambda^2\right)^{1/2}\left(\sum_\lambda d_\lambda^2\right)^{1/2}, $$ for some constant $C>0$ (whenever the right-hand side makes sense). Now I am feeling that the proof might be wrong, so if it is the case, I am wondering if anyone has a correct proof for it.

$\endgroup$

1 Answer 1

2
$\begingroup$

We write $\mu=2^k \lambda$ with $k \ge -3$. The standing assumption $\mu \ge 1$ becomes $2^k \lambda \ge 1$. The RHS of the sequence of equalities you reproduce is missing a factor $\lambda^s$; is this typo in the original paper?

The sequence of equalities you included should thus be written (adding $L$ on the left) as $$ L:= \sum_\lambda \lambda^s\sum_{\mu\geq \lambda/8} a_{\mu}d_{\lambda}=\sum_\lambda \lambda^s\sum_{\substack{k=-3,\\ 2^k\lambda\geq1}}^{\infty}a_{2^k\lambda}d_{\lambda}=\sum_{k=-3}^\infty \sum_{\lambda\geq 2^{-k}} \lambda^s a_{2^k\lambda}d_\lambda \,. $$ Thus $$ L= \sum_{k=-3}^\infty 2^{-ks} \sum_{\lambda\geq 2^{-k}} (2^k\lambda)^s a_{2^k\lambda}\, d_\lambda \,. $$ Apply Cauchy-Schwarz to the inner sum: $$ L \leq \sum_{k=-3}^\infty 2^{-ks}\left(\sum_{\lambda\geq 2^{-k}} (2^k \lambda)^{2s} a_{2^k\lambda}^2\right)^{1/2}\left(\sum_\lambda d_\lambda^2\right)^{1/2} $$ Then substitute $\mu=2^k \lambda$: $$ L \leq \sum_{k=-3}^\infty 2^{-ks}\left(\sum_{\mu} \mu^{2s} a_{\mu}^2\right)^{1/2}\left(\sum_\lambda d_\lambda^2\right)^{1/2} \, , $$ which yields the inequality you asked about $$ \sum_\lambda \lambda^s\sum_{\mu\geq \lambda/8} a_{\mu}d_{\lambda} \leq C \left(\sum_\lambda \lambda^{2s}a_\lambda^2\right)^{1/2}\left(\sum_\lambda d_\lambda^2\right)^{1/2} \, , $$ with $C:=\sum_{k=-3}^\infty 2^{-ks}$.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.