[Help-gsl] Vegas, how accurate to compute integrand

[Tommy Nordgren]

I have a complicated Integral, in five+one dimensions. The integrand  
have a trigonometric dependance
on five angles, and the sixth dimension is a radius going from 0 to  
infinity.
I use vegas to integrate across the angular dependencies, where the  
function to be integrated is computed
by computing the fourier integral with the angles held constant.
How precise should I set the relative and absolute error bounds in  
the fourier integral, to allow vegas to return
as accurate results as feasible. Currently I use 1e-4 as both  
relative and absolute error bounds.
-------------------------------------
This sig is dedicated to the advancement of Nuclear Power
Tommy Nordgren
address@hidden