R-alpha: Re: Inf et al.

[Martin Maechler]

>>>>> "Ross" == Ross Ihaka <ihaka@stat.auckland.ac.nz> writes:

    Ross> I am having a look at the computations of p-values and quantiles
    Ross> which depend on the incomplete beta function at present.  I hope
    Ross> to have a better set of functions in place by weeks end.  I may
    Ross> need a tester or two to really exercise the new ones.

I'd volunteer for this..

BTW: The phyper 'inaccuracy' is not a pbeta problem however.
I think that AS 152 (which has been improved, but is still available as
		     file  ...statlib.../apstat/152 )
should be used in place of  phyper.c
		

    Ross> I'm also
    Ross> looking at the whole math library to see exactly how much work a
    Ross> proper handling of Inf, NaN and NA might involve. 

I think that this is the most important of the current TASKS.
Even though it doesn't seem to be used so often,
I've been realizing lately that in several places in my code, I just rely
on things like  1/Inf  -> 0.
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-devel-request@stat.math.ethz.ch
=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-