On 30 Dec 2020, at 0:11, Raul Miller wrote:
Depending on how much precision you need, integration can be
approximated using +/ (or, d._1 but I guess because that's not general
enough, d. has been removed from recent versions of J and been
replaced by deriv_jcalculus_ -- see
https://code.jsoftware.com/wiki/Vocabulary/ddot).
But I don't read R well enough to read that expression and translate
it into J.
I can point you at a J implementation of a beta function --
https://code.jsoftware.com/wiki/Essays/Beta_Function (beta=: ] %@* [
!&<: +) but I do not know how closely that's related to R's beta
distribution.... There's also the incomplete beta mentioned at
https://www.jsoftware.com/papers/jhyper.pdf
Looking at the wikipedia writeup at
https://en.wikipedia.org/wiki/Beta_distribution, I see an expression
for the probability distribution function and for the cumulative
distribution function for beta distributions ... and I would guess
that maybe what you are looking for is a mechanism to compute the CDF
from the PDF? (But, since calculus often involves infinities,
integration often devolves into symbol lookups and/or numeric
approximations...)
I don't know if any of this helps, though..
--
Raul
Hi Raul,
please excuse my late reply… However thank you very much for your
hints. I’ll study the code in the essay you provided in greater detail
and will try to make it work for me. At a first glance it looks really
promising.
Kind regards,
Thomas
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
