Brief description of class still missing. More...
Brief description of class still missing.
Partial fast Fourier transform, transform a sequence of n samples into a sequence of m samples after a Fourier transform. m*q=n, where q is an integer.
The algorithm is based on some comments in Bailey and Swarztrauber 1995 "The fractional Fourier Transform and Applications, SIAM Review, vol. 33 no. 3 (Sept. 1991, pg. 389-404)"
F(k+s)[x]=sum j from 0 to q-1 {exp(-2*pi*i*j*(k+s)/n)*F(k)[z(j)] }, o<=k<m
Where z(j)=x(j+l*q)*exp(-2*pi*i*l*s/m), 0<=l<m
The current type() determine the direction of the transform. If it is currently in time domain: spectrum is computed from min Hz to max Hz. On the contrary, if it is in frequency domain: the waveform is computed from min s. to max .s rather than from 0 to 1/df with the classical fastFourierTransform().
If the total input number samples is not a multiple of the output range (in terms of number of samples), some samples at the end of signal can be ignored. If the observed q is less than 10, a normal fast Fourier tranform is started.
This class is obsolete, because no significant time improvement can be achieved compared to a complete FFT transform.
