Re: [boost] C++ committee actions - More statistical math functions?

From: Matthias Troyer (troyer_at_[hidden])
Date: 2003-11-06 08:34:31

On Nov 5, 2003, at 11:30 PM, Paul A. Bristow wrote:

> Thanks for this encouragement.
>
> I look forward to learning the details of Walter's proposals as 
> accepted.
>
> However the meat of the algorithms (especially the dreaded incomplete 
> beta
> function) is beyond my expertise, but I'll see if I can get permission 
> to re-use
> existing work. I might be willing to 'repackage' existing C algorithms.
>
> Testing them is also a significant issue as they won't be 'fully' 
> accurate even
> as much as exp, log, pow etc are (only a few bits 'wrong').  (Did the 
> issue of
> testing the accuracy of pow, exp, log, sin, cos - and Walter's 
> functions come
> up?)
>
> But I'd be happier with some support from other Boosters.  Surely 
> someone wants
> to be able to calculate the probability of Student's t using C++?

What kind of support do you wish for or expect? We currently need to 
implement distribution functions  such as chi^2 or K-S, and similar 
ones.

Matthias