From: John Maddock (john_at_[hidden])
Date: 2005-12-20 05:43:16
> The now usual method is by Thompson IJ and Barnett AR (1986) Jo
> computational Physics, vol 64, pp 490-509.
That's also given a mostly-incomprehensible numerical recipies treatment,
but appears to be specific to the modified Bessel functions? (see page
247).
John.