Hi Ken,
just to repeat Prof Ripley's answer:
- R-devel has already increased itrmax (from 100 to 1000),
and if you don't want to compile R-devel yourself, there are precompiled
R-devel versions for Windows {maybe MacOS X too}.
[Also, there's an E-mail on Nov 3 from Thomas Lumley to you and R-help
where he replied to your posting from Oct 24, and told you
exactly this; did you somehow not get that mail? (please
answer offline; this gets *way* off-topic!)]
Indeed, we are interested to hear if the change has been
sufficient for all cases. Personally, I think the real
solution would have to be a different formula / algorithm
for "large" ncp values.
- If you want to change R's internals, do so and recompile R;
that's considerably less work (for you) than trying to put the
changes into a package {particulary since you should also
do the vectorization that R's internal code does for you}.
- BTW: This topic has been quite techncal from the start, and I
think the posting guide advises you to use the "R-devel"
mailing list instead of R-help { __ next time __ }
BTW, thank you for your orginal report about the pf(*, ncp = "large")
misbehavior!
Regards,
Martin <[EMAIL PROTECTED]> http://stat.ethz.ch/people/maechler
Seminar fuer Statistik, ETH-Zentrum LEO C16 Leonhardstr. 27
ETH (Federal Inst. Technology) 8092 Zurich SWITZERLAND
phone: +41-44-632-3408 fax: ...-1228 <><
______________________________________________
[email protected] mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html
