Ludvig af Klinteberg

Sure, no reason not to simplify. Seems to be a merge conflict at the moment though

Merging stale PR

This still appears to be an issue, even after switching to GSL 2.5. Interestingly, the error estimate provided by GSL is also completely off ```julia using GSL, PyPlot, PyCall x...

That's too bad. There actually seems to be quite a few bug reports on gsl_sf_hyperg_2F1. Let's just hope they get fixed in a future version.

I don't think GSL has hypergeometric functions for complex arguments.

> I don't think GSL has hypergeometric functions for complex arguments. Sorry, that was wrong, there's actually `sf_hyperg_2F1_conj`, which could be called from `hypergeom` for the right arguments.

I've never tried out the `multilarge_nlinear` interface, so I'm afraid I can't be of much help here. The high-level driver might not have made it into this interface because the...