SciPy's complex erf code is not free/open-source? (Trac #1741)

[scipy-gitbot]

Original ticket http://projects.scipy.org/scipy/ticket/1741 on 2012-10-05 by trac user stevenj, assigned to @pv.

The complex error function implementation in SciPy seems to use the subroutine WOFZ published in ACM Transactions on Mathematical Software (TOMS):

https://github.com/scipy/scipy/blob/master/scipy/special/toms/wofz.f

ACM TOMS code is generally NOT free/open-source software: it falls under the ACM Software copyright (http://www.acm.org/publications/policies/softwarecrnotice) which is free-of-charge only for noncommercial use.

I came across this because I was looking for a free/open-source complex erf routine to use in our own code; we originally wanted to use the TOMS code, but reluctantly concluded that it was unusable because of licensing. As an alternative, I wrote an independent implementation of another algorithm (which seems to be a bit more accurate than WOFZ) and released it as free/open-source software: http://ab-initio.mit.edu/wiki/index.php/Faddeeva_w

[scipy-gitbot]

@rgommers wrote on 2012-10-05

Thanks for reporting this. I asked on scipy-dev if anyone remembers a discussion on the license of this function: http://article.gmane.org/gmane.comp.python.scientific.devel/16970

Some other TOMS functions were specifically relicensed as BSD after Scipy devs asked ACM, but not this one (or at least I can't find it in the list archive).

[scipy-gitbot]

@pv wrote on 2012-10-06

We can just replace it. That code was added pre-2001, so the only one knowing about this would probably be Travis.

Here's PR: #331

[scipy-gitbot]

@rgommers wrote on 2012-10-10

PR is merged. Thanks Steven, both for pointing out the issue and writing a suitably licensed replacement.

[scipy-gitbot]

Milestone changed to 0.12.0 by @rgommers on 2012-10-10