2
$\begingroup$

Background: I have a function $g(\omega)\in C^{\infty}(\mathbb{R})$, which vanishes like $O(|\omega|^{-\beta})$ at infinity for some $\beta>0$.

This answer states that functions that decays "too slowly" failes to be in $\mathcal{F}L^1(\mathbb{R})$, and exercise VI.1.7 of Kaznelson's "An Introduction to Harmonic Analysis" states that $C^{2}_{c}(\mathbb{R})\subset\mathcal{F}L^1(\mathbb{R})$.

Problem: Can we deduce $g\in \mathcal{F}L^1(\mathbb{R})$ from these conditions? Heuristically, the decay condition means that $\hat{g}$ should not have singularities of "order $\geq1$" , and therefore $\hat{g}$ should be locally integrable. Can this intuition be made precise?

$\endgroup$

1 Answer 1

7
+50
$\begingroup$

This does not follow. There are compactly supported, finite, purely singular measures $\mu$ whose Fourier transform has power decay: $|\widehat{\mu}(x)|\lesssim (1+|x|)^{-\beta}$; for more background, you can search for Fourier dimension, for example on this site.

This function $g=\widehat{\mu}$ satisfies all your conditions, but its Fourier transform $\mu$ is not an integrable function.

$\endgroup$

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.