Apologies. Snapshot from
rsync --stats -z -rC --delete rsync.r-project.org::r-devel R
yesterday which contains the fragment pow(1-p,...) which is problematic
for small p.
(If the snapshot I get with the above command has an official name I'll
be happy to use that in future reports.)
Mea culpa.
Chalk this up to incorrect mucking with the trunc function on my
side, a very old bug here. Apologies.
Morten
__
[EMAIL PROTECTED] mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
Make that 30400 orders of magnitude (natural logs y'know)...
Right. (/me raises hands showing 2.7 fingers.)
What the devil are you calculating? The probability that a random
configuration of atoms would make up the known universe?
Not quite. Where you see a cdf for the gamma distribution
A little code study, formula study and experimentation reveals that the
situation is mostly fixable:
1. Get rid of the explicit alpha limit. (A form of it is implicit in
(2) and (3) below.)
2. Use the series formula when
(x alph + 10 x 0.99 * (alph + 1000))
This guarantees
The real problem is that pbeta can take forever. That's bug #7153 and a fix is
within reach.
Morten
__
R-devel@stat.math.ethz.ch mailing list
https://stat.ethz.ch/mailman/listinfo/r-devel
For the real problem, the R source (in C), It's simple
to work around the fact that
qcauchy(0, log=TRUE)
for Morten's code proposal gives -Inf instead of +Inf.
Ouch. Good catch.
Here is what happened: I reduced 1-exp(x) to -expm1(x) which is actual wrong for
x=0 because the results will