Re: binocdf accuracy

[Dr. Alexander Klein]

Am 07.07.2013 um 13:38 schrieb CdeMills:

> there are two sources of numerical inaccuracies in the betainc function:
> 1) the use of bincoeff
> 2) the use of power
> 
> 
> binocdf-mp.txt
> <http://octave.1599824.n4.nabble.com/file/n4655283/binocdf-mp.txt>  

Pascal,

thank you for the elaborate example! As for the binomial coefficients, these 
are correct in the Erlang version, because Erlang has integer bignums, and as 
for the powers, well I don't know much, but my results are quite close to 
yours, and I really suspect that betainc is the culprit.

Yesterday, I kluged up an approximation to the Gamma function, and I hope I'll 
get the incomplete beta up and running sometime this week.

I'll keep you updated,

        Alex

-- 
Dr. Alexander Klein, Diplom-Mathematiker
TransMIT-Projektbereich für Mathematische Analysen und Feld-Simulationen
Kerkrader Straße 3
D-35394 Gießen

http://www.transmit.de/zentren/tz.cfm?N=165