cufft normalization

Accelerated Computing CUDA CUDA Programming and Performance
#1

hi,

the docu says:
IFFT(FFT(A))= length(A)*A

but I’m not quite clear about what to do when I want to perform calculations on the result of the fft and do the ifft on the modified values.

in my code, I want to extract the magnitude and phase from the fft result, modify the magnitude and compile the modified magnitude and original phase into the complex input array for the ifft.

do I have to do normalization on the fft-result or can I wait untill after ifft?

thanks for any hints,
downforme

#2

Since FFT is a linear transform, it does not matter when you apply the scaling.

i.e. IFFT(FFT(kX)) == IFFT(kFFT(X)) == k*IFFT(FFT(X))

#3

great, thanks for your answer