Suppose I have a signal $u(x)$ where $x\in[0,2\pi]$ and the signal has following form in Fourier space,

$\hat{u}(\kappa)=\begin{cases} \frac{1+i\tan{(\varphi_u(\kappa))}}{\sqrt{1+\tan^2{(\varphi_u(\kappa))}}}\kappa^{\alpha/2},\ \ 0<\kappa\le\kappa_c \\ 0+i0,\ \ \ else \end{cases}$

where $\varphi_u(\kappa)$ is a random number for each $\kappa$, and $\alpha$ is a constant.

How can I get $u(x)$ using inverse Fourier transform?

  • 1
    $\begingroup$ Hi and welcome to the MathOverflow. I suspect that the sole knowledge that $\varphi_u$ is a random number is not of great help: do you know something about its probability distribution as a random variable? $\endgroup$ Aug 30, 2020 at 9:09
  • 1
    $\begingroup$ @DanieleTampieri $\varphi_u(\kappa)$ is a random variable which is uniformly distributed in $[-\pi, \pi]$ for each $\kappa$. $\endgroup$
    – S.C. Zheng
    Aug 31, 2020 at 14:18

1 Answer 1


The expectation value $E[u](x)$ can be obtained by inverse Fourier transform of the expectation value of $\hat{u}(k)$, since the Fourier transform is a linear operation; If I interpret the statement in the OP that the phase $\phi$ is a "random number" as saying that $\phi$ is uniformly distributed, then the integral $$\frac{1}{\pi}\int_0^\pi \frac{1+i\tan\phi}{\sqrt{1+\tan^2\phi}}\,d\phi=\frac{2}{\pi}$$ gives $E[u](x)$ as the inverse Fourier transform of $(2/\pi)\kappa^{\alpha/2}$, which is an incomplete Gamma function.

I should note that the statement in the OP that both $u(x)$ and $\hat{u}(k)$ have compact support is inconsistent.


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.