3
$\begingroup$

I have a Lemma from a text that states that if $g\in W^{1,2}(\mathbb{R})$ ($W^{k,p}$ a Sobolev space) and the weak derivative $Dg\in L^2(\mathbb{R})$ then the Fourier transform $\mathcal{F}g\in L^1(\mathbb{R})$. Assuming there a higher-dimensional analogue to this, should I check the condition for each of the partial weak derivatives?

Cheers...

$\endgroup$
3
$\begingroup$

The higher dimensional analog is somewhat different. The idea of the result you quoted is that the transform of $Dg$ is a multiple of $\xi \hat g(\xi)$, and once we know that $\int_{\mathbb R} (1+|\xi|^2) |\hat g(\xi)|^2 <\infty$, it follows from Cauchy-Schwarz that $$\left(\int_{\mathbb R} |\hat g(\xi)| \right)^2 \le \int_{\mathbb R} (1+|\xi|^2)^{-1} \int_{\mathbb R} (1+|\xi|^2) |\hat g(\xi)|^2 <\infty$$ Trying to do the same in dimensions $n\ge 2$, we encounter a problem: $$\int_{\mathbb R^n} (1+|\xi|^2)^{-1} =\infty$$

Solution: take more derivatives. If the weak derivatives of order $k$ are square integrable, then following the above we get $$\left(\int_{\mathbb R^n} |\hat g(\xi)| \right)^2 \le \int_{\mathbb R^n} (1+|\xi|^{2k})^{-1} \int_{\mathbb R} (1+|\xi|^{2k}) |\hat g(\xi)|^2 <\infty$$ provided that $2k>n$. Thus, the $n$-dimensional analog involves derivatives of order $\lceil (n+1)/2 \rceil$. And yes, you should check each of the weak partial derivatives.

| cite | improve this answer | |
$\endgroup$

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.