[SciPy-user] Fourier-sine transform with SciPy

Lubos Vrbka lists@vrbka....
Fri Jun 6 12:47:58 CDT 2008


anne, thanks for detailed discussion (and sorry for accidentally 
replying directly to you instead to list)!

 > Note that the sine transform only works on odd functions, so either
 > your input data should have f_k = f_{N-k} or you are representing only
 > half the values, i.e., N is twice the number of data points you have.
 > This too is somewhere a proper sine transform would help, if we had
 > it.
well, the original function itself is neither even nor odd, since it is 
defined only in the interval <0, +inf>. at certain point in the 
derivation, the original function is multiplied by the r or k variable, 
so it might actually be odd. the sine comes from the fact that the 
original function is actually radially symmetric. i think i will have to 
have a look into that an check the result.

hopefully the last question. how is it with the 'normalization 
constants' for the dft? from theory it should be T/N for FT and 1/T for 
iFT, but different packages treat this differently (for example fftw 
does not use them and it has to be done afterwards 'manually') - how is 
it in case of scipy.fft?

best,
lubos


-- 
Lubos _@_"
http://www.lubos.vrbka.net


More information about the SciPy-user mailing list