[Scilab-users] Possible error in manual

classic Classic list List threaded Threaded
1 message Options
fmiyara fmiyara
Reply | Threaded
Open this post in threaded view

[Scilab-users] Possible error in manual

Dear All,

In the help information for the FFT function I find that, regarding the "inverse" FFT,  fft(x,1), the formula is

x(k) = SUM(a(m)*exp(2*%i*%pi*(m-1)*(k-1)/n)

But this is not the inverse FFT since it lacks the required 1/n factor; however, fft(fft(a,-1),1) indeed returns a. So the procedure seems to be mistakenly described in the help information.

Interestingly, the fftw3 documentation, http://www.fftw.org/fftw3.pdf, gives, on page 42, the same formula but it indicates it is an unnormalized transform.

Would you please clarify?


Federico Miyara


users mailing list
[hidden email]