I have nearly finished an implementation of the Error functions. Following
functions I have worked on:
erf(z) - Error function Erf
erfc(z) - Complement of the Error function Erfc
erfi(z) - Imaginary Error function Erfi
erf_generalized(z1, z2) - Generalized Error function Erf(z1,z2)
The numerical algorithm in float and bigfloat precision are implemented via the
Incomplete Gamma function which is already implemented. We have a slatec routine
for erf. At this time I have not tested which routine might be better.
We have already some code for the Erf-function in the file csimp2.lisp and some
definitions in suprv.lisp and at other places. In a few places the symbols $erf
and $erfc are used.
I suggest to add the code for the Error functions to the file gamma.lisp and cut
out the old code in csimp2.lisp. Furthermore we have to look at the places in
the code where the symbols $erf and $erfc are used. Erf is already implemented
as a simplifying functions, but not erfc. So the meaning of the symbol $erfc
would be changed.
So what do you think?
Dieter Kaiser