Here is a patch for more accurate gamma and erf functions. The gamma routine is Lanczos algorithm so it's valid for complex arguments too. http://winnie.fit.edu/~gabdo/gamma.txt is a nice description of the algorithm used here. The erf routine is a hand-modified version of an f2cl conversion of the erf routine in TOMS 715. I've checked these against the tables in A&S. They match well. Ray
Attached file: csimp2.patch