[Haskell-cafe] Numerics (was: Re: Trouble with asinh)

[Carter Schonwald]

Hey David,
Create a tracking ticket on the ghc for the set of changes / issues you
wanna address and how, and @ me on it  (I’ll then make sure to at other
applicable Dolan )and I’ll try to help you get oriented and make sure it
has the right visibility so we can help you out!

On Sat, Sep 18, 2021 at 12:00 PM David James <davjam at live.com> wrote:

> Hi thanks for the comments. I actually have draft rewrites of the Haskell
> complex functions with (I think) the correct behaviour, including branch
> cuts. But I discovered the errors in the underlying real functions while
> testing them. I’d like to try fixing all of these, but will probably need
> some help.
>
> Are there instructions somewhere on the process to fix a bug (presumably
> forking in GitHub, fixing code, adding test cases, running some CI, both
> for Linux and Windows, etc)? (And is there something similar for fixing
> mingle-w64 bugs?)
>
> I’ve been developing more test cases (esp for infinities +/-0, etc), but
> if there are any ideas/references on how to thoroughly test real or complex
> functions, that would also be useful.
>
> I’m currently away (and struggling to type on a phone) but will send more
> details (and pictures based on those in Common Lisp The Language 2nd
> edition) when I’m back.
>
> Thanks! David.
>
>
>
> > On 17 Sep 2021, at 22:06, Barak A. Pearlmutter <barak at pearlmutter.net>
> wrote:
> >
> > I suspect that most implementations of Common Lisp just call the C
> > standard library catan(3) etc, which are well tuned.
> >
> > $ clisp
> > Welcome to GNU CLISP 2.49.92 (2018-02-18) <http://clisp.org/>
> > [1]> (atan #c(0 1d-40))
> > #C(0 1.0d-40)
> >
> > In this particular case, the problem is that the Haskell Data.Complex
> > code has its own implementation of atan, which uses a log(1 + x) in
> > calculating the imaginary part. A foreign function call to the
> > appropriate libm routine would robustly address this, but that would
> > be difficult because it's trying to be generic over RealFloat a =>
> > Complex a, instead of special casing Complex Float / Complex Double.
> > Anyway, the Standard Prelude code for this is naïve: it should call
> > log1p, at the very least—which it actually goes to the trouble of
> > defining correctly, but not exporting.
> > _______________________________________________
> > Haskell-Cafe mailing list
> > To (un)subscribe, modify options or view archives go to:
> > http://mail.haskell.org/cgi-bin/mailman/listinfo/haskell-cafe
> > Only members subscribed via the mailman list are allowed to post.
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.haskell.org/pipermail/ghc-devs/attachments/20210924/b9bdd8fe/attachment.html>