How accurately do I need to compute the integrand in a multiple
integral to be estimated by the
Vegas algorithm? The integrand is expensive to compute (a Fourier
integral along a radius vector)
in a 6-dimensional integral over infinite space.
My idea is to first compute the integral with a small number of
points, and then use that first approximation
to estimate suitable values for absolute and relative accuracy when
doing the Fourier integration part of the problem.
(Note: For my problem I'm using a 6-dimensional generalisation of
Spherical Coordinates.
This makes the interation region infinite in the Radius alone, which
turns my problem into:
For five angles: Integrate the value of the (Fourier integral along
Radius Vector) with respect to the angles)
My basic problem is that I don't know enough about the Vegas algorithm
to compute good esimates of
how accurately I need to compute the Fourier Integrals.
Any ideas??
---------
See the amazing new SF reel: Invasion of the man eating cucumbers from
outer space.
On congratulations for a fantastic parody, the producer replies :
"What parody?"
Tommy Nordgren
[EMAIL PROTECTED]
_______________________________________________
Help-gsl mailing list
[email protected]
http://lists.gnu.org/mailman/listinfo/help-gsl
