Optimizing function calls in GSL Vegas

[Tommy Nordgren <tommy dot nordgren at comhem dot se>]

I want to compute a complicated integral with Vegas. The integral is  
6-dimensional, and have
a sinusoidal factor in a linear combination of the integration  
variables. The integration variables vary
from negative to positive infinity. I can't solve this integral by  
simply using a coordinate substitution
that will turn each integration range to a finite range, because then  
the sinusoidal factor will become pathologial.
Instead, I use a generalized spherical coordinate transformation,  
which will turn the integration interval into a
finite interval, for 5 of the integration variables, where the  
function called by vegas is computed as a Fourier integral
with respect to the radial parameter.
	My problem is choosing good heuristics, for estimating to what  
relative and absolute accuracy to use as parameters
to the Fourier integral.

------------------------------------------------------
"Home is not where you are born, but where your heart finds peace" -
Tommy Nordgren, "The dying old crone"
tommy.nordgren@comhem.se