Hi Regina,
Regina Henschel wrote:
Eike Rathke schrieb:
Hi Regina,
I postponed this reply far too many times.. sorry.
Too many questions at once ;-)
It seems to be one of the most difficult functions. To get an idea of
that function, have a look at the file toms708.c of the R-project.
There is some hidden beauty in that file. ;)
Without being of much help, I just want to draw attention to some
features I spotted during a peak in the code:
- often, (1-x) is decomposed into more stable terms,
- from the simple 0.5 - x -0.5
[and x - 0.5 - 0.5 (for x-1)],
- to more complex terms like:
return w * (0.5 - 1.0 / w + 0.5);
[this is actually w - 1]
The code was well written by numerical analysts, and optimised for
accuracy over a large domain of values. Is it feasible to use this
library or a similar library (if one is found)?
Rewriting this code isn't going to be easy.
Sincerely,
Leonard
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
