0
$\begingroup$

I have 2 real time series $x(t)$ and $y(t)$, after fft it should become $\tilde{X}(f)$ and $\tilde{Y}(f)$. Then I need to normalize $\tilde{X}(f)$ with $\tilde{Y}(f)$ : $\tilde{X}(f)/\tilde{Y}(f)=\tilde{Z}(f)$

However fft gives me $\tilde{Z}^*(f)$ instead (the sign of the imaginary part is inverted).

What could have caused this ? I know $x(-t)$ and $y(-t)$ can cause $\tilde{Z}^*(f)$, but $x(t)$ and $y(t)$ have the correct physical meaning.

$\endgroup$
  • 1
    $\begingroup$ Well then something about your vector manipulation or division is wrong. A common trap in Matlab is that the ' operator and .' operator are different. The first is conjugate transpose while the second is just transpose. $\endgroup$ – Andy Walls Jul 22 at 9:55
  • 1
    $\begingroup$ I would be easier for us if you could include your matlab code. $\endgroup$ – AlexTP Jul 22 at 10:04
2
$\begingroup$

Taking the inverse FFT (IFFT) instead of FFT would cause a phase inversion.

| improve this answer | |
$\endgroup$
0
$\begingroup$

If we assume that the Fourrier Transform formula is :

$${{\hat {x}}(\nu )=\int _{-\infty }^{+\infty }x(t)\,\mathrm {e} ^{-2{\rm {i}}\pi\nu t}\,\mathrm {d} t}.$$

Are you sure you didn't forget the minus (-) sign in the exponential ? In this case, because $x(t)$ is real, You will effectively find the conjugate of ${\hat {x}}(\nu )$ at the end.

| improve this answer | |
$\endgroup$
  • $\begingroup$ Please check my updated question. I used Matlab fft function so there is no way the fft itself is wrong. $\endgroup$ – 7E10FC9A Jul 22 at 9:14

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.