Re: [Help-gsl] QAWF integration algorithms and low frequencies

[Brian Gough]

At Mon, 15 Sep 2008 14:15:40 +0200,
Simone Pigolotti wrote:
> I just implemented the function *gsl_integration_qawf() *which I
> need to calculate some Fourier integrals. I'm testing with some
> simple example like the Fourier transform of a Gaussian function but
> it is giving me problems: the result is correct exept for very small
> frequencies, where the functions returns zero as a result (of course
> the result of the integral of the distribution X cos(omega x) for
> omega -> 0 should be 1). Is there any known problem with this
> integration scheme? Or maybe I didn't set properly the parameters?

There are no known bugs in QAWF.  

The documentation gives a brief explanation of the algorithm, it
integrates over segments of length pi/omega.  This is probably the
cause of what you are seeing because the gaussian will be almost zero
over most of that range.  I guess the function needs to be reasonably
slowly varying over each region.

If you think this could be handled better could you submit a bug
report with a small example program showing the problem to
address@hidden, thanks.

-- 
Brian Gough

GNU Scientific Library -
http://www.gnu.org/software/gsl/