6
$\begingroup$

Does there exist $f \in L^1(\mathbb{R}^n)$ s.t., $\displaystyle \widehat{f}(\xi) = \frac{\sin |\xi|}{|\xi|}$ in case of dimension $n \ge 3$?

It is known that for $n = 2$, the function $\displaystyle f(x) = \frac{\chi_{\{|x| < 1\}}}{\sqrt{1-|x|^2}}$ curiously has some constant multiple of $\dfrac{\sin |\xi|}{|\xi|}$ as Fourier transform. Or at least is there a way of finding the Fourier transform $\frac{\sin |\xi|}{|\xi|}$ in sense of tempered distributions for $n \ge 3$?

$\endgroup$
  • 4
    $\begingroup$ For $n=3$, see mathoverflow.net/questions/315536 $\endgroup$ – YCor Nov 18 '18 at 16:46
  • 3
    $\begingroup$ Just to expand very slightly on @YCor's comment/answer: because the Fourier transform is injective from $M({\bf R}^n)$ to $C_b({\bf R}^n)$, there is no $f\in L^1({\bf R}^3)$ with the properties that you desire, but instead you need to take the Fourier transform of a certain probability measure that is singular w.r.t. Lebesgue measure on ${\bf R}^3$. $\endgroup$ – Yemon Choi Nov 18 '18 at 16:52
  • $\begingroup$ There is of course a (radial) distribution that has the desired FT, by Fourier inversion. The only meaningful question you can ask along these lines is how much regularity this distribution has. $\endgroup$ – Christian Remling Nov 18 '18 at 17:00
  • 4
    $\begingroup$ Check Volume 1 of Gelfand and Shilov's book Generalized functions. You will find and explicit description of a distribution, concentrated along the unit sphere, whose Fourier transform is $(\sin|\xi|)/|\xi|$ $\endgroup$ – Liviu Nicolaescu Nov 18 '18 at 17:06
  • 1
    $\begingroup$ @LiviuNicolaescu Gelfand and Shilov's book covers the $n = 2m+3$ case. How'd we argue for the even dimensions? $\endgroup$ – sciona Nov 18 '18 at 17:39
9
$\begingroup$

Here's a formula from Duistermaat and Kolk book Distributions: Theory and Applications, Chapter 17, Eq, (17.13). We denote by $\newcommand{\eF}{\mathscr{F}}$ $\eF$ the Fourier transform. Then $\newcommand{\ii}{\boldsymbol{i}}$ $\newcommand{\ve}{\varepsilon}$

$$ \eF^{-1}\left(\frac{e^{\ii t\Vert \xi\Vert}}{\Vert\xi\Vert}\right)=p_n\lim_{\ve\searrow 0} \Big(\; \Vert x\Vert^2-(t+\ve\ii)^2\;\Big)^{-\frac{n-1}{2}}, $$ where $p_n$ is a certain universal constant,

$$p_n=\frac{\Gamma\big(\frac{n+1}{2}\big)}{(n-1)\pi^{(n+1)/2}}. $$

$\endgroup$
  • $\begingroup$ Thank you very much for the reference. $\endgroup$ – sciona Nov 18 '18 at 19:26

Your Answer

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

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