[WP34s] Regularized incomplete Beta function

[Paul Dale]

(05-05-2014 09:13 PM)Dieter Wrote:  If this really got implemented it should quit with an error message as soon as ...

I'm slightly embarrassed that this bit of code made it in and has been unnoticed for so long I don't know what I was thinking when I implemented it -- probably rushing through recoding all the distributions.

The good news is the modified Student's T code is in both the 31S and the 34S. Should be a nice speed up and bypasses the poor implementation of the one degree of freedom case.

Quote:And I would love to know how fast this might run in compiled C code, e.g. on a 31s. ;-)

I suspect not as much faster as we'd like. The incomplete beta function, which is the heart of of the CDF, is in C already. Likewise for LogGamma in the PDF. The rest is predominately shuffling and scaling values which would speed up in C but I doubt they are the bulk of the time.

C would give five more digits in intermediates. This would justify an extra iteration in DP mode I think.


- Pauli