Faster inverse gamma and factorial for the WP 34S

[rprosperi]

(02-08-2015 07:10 PM)Dieter Wrote:  

(02-08-2015 05:39 PM)Bit Wrote:  Awesome. I can't work on this right now but I'm glad the real experts have taken on the issue.

Real experts would have included a perfect solution for arguments at or close to the Gamma minimum. ;-) There was a similar problem when Lambert's W was implemented in XROM, but back then I could provide a solution. Let's see if the community can come up with something similar here.

Test cases:

Code:

   0,88560 31944 10888 70027 88159 00582 5887
=> error

   0,88560 31944 10888 70027 88159 00582 5888
=>  1,4616 32155 73196 15209 53068 05276 6163

Otherwise it's not perfect. ;-)

Dieter

I'm curious - and I'll be up-front here, I really know little about this topic, other than the barest familiarity of gamma, inverse gamma, digamma, etc. - how is accurarcy for such values important? I am seriously not trying to minimze it; as many people are busily testing and pushing the envelope, it clearly must have some practical application, I (and I suspect other lurkers) would like to know how it applies, and why such precision can have practical use or impact.