Re: [OctDev] complex error function

[Steven G. Johnson]

Jordi Gutiérrez Hermoso wrote:
> Your actual numerical routines go in liboctave/cruft/, and the
> Octave-callable functions should be turned into mapper functions like
> the current erf is under libinterp/corefcn/mappers.cc. We should work
> in an hg clone.

I have posted an hg clone at:

        http://ab-initio.mit.edu/octave-Faddeeva

that is patched to add complex-argument erf, erfc, erfcx, erfi, and 
dawson (scaled erfi) functions via my Faddeeva package.  [There is no 
need to add an explicit Faddeeva w(z) function, since w(z) is simply 
erfcx(-iz).]  The documentation is also patched, and a few test cases 
are added to 'make check'.

Let me know what you think.

--SGJ