Faster inverse gamma and factorial for the WP 34S

[Dieter]

(02-08-2015 11:06 PM)BarryMead Wrote:  To be honest I don't know what one would use a high accuracy inverse gamma function for. I have never needed one.

Squeezing out the best possible, most accurate result is the fun of it. Who needs a real world application? ;-)

(02-08-2015 11:06 PM)BarryMead Wrote:  I could not have written the algorithms that Bit and Dieter submitted. I know just enough to begin to appreciate some of the techniques used to keep the errors small,

That's about the level I'm working on, as far as the Gamma/Digamma function is concerned. ;-)

The tricky part still remains unsolved: getting accurate results for arguments extremely close to the Gamma minimum at 1,46163... etc. The current algorithm could do this, provided the Digamma function is sufficiently accurate and the working precision is about 53 (!) digits (required for Gamma of a 34-digit argument). So there still is a lot to do.

Dieter