[SciPy-user] Integration with precalculated values
Tue Jun 17 06:29:26 CDT 2008
Thanks Robert and Anne,
I will use scipy.interpolate.splrep for my precalculation.
I finally managed to keep the factors exp(x**2) under control. The
error function can be expanded in series that contain a common factor
equal to exp(-x**2). I use in particular the approximation of the
complementary error function with rational functions, as described in
"Rational Chebyshev approximations of the error function" by WJ Cody,
Math. Comp., 1969, pp 631--638. I have also had to split the
definition of the integrand into positive and negative values of the
argument, in order for the integral to go smoothly.
Thanks for your help.
More information about the SciPy-user