[Rd] small bug on qchisq (PR#6442)

maechler at stat.math.ethz.ch maechler at stat.math.ethz.ch
Tue Jan 20 10:34:59 MET 2004

>>>>> "MM" == Martin Maechler <maechler at stat.math.ethz.ch>
>>>>>     on Mon, 19 Jan 2004 19:03:29 +0100 (CET) writes:

>>>>> "Remy" == Remy Drouilhet <Remy.Drouilhet at upmf-grenoble.fr>
>>>>>     on Mon, 19 Jan 2004 18:10:04 +0100 (CET) writes:

    Remy> qchisq(1,10) works well but qchisq(1,10,ncp=0) does
    Remy> not work whereas ncp=0 is the default value of the
    Remy> function qchisq(1,10). (of course, 10 will be replaced
    Remy> by any integer value).

    Remy> Let us notice that this bug occurs only when applying
    Remy> probability one.  (qchisq(seq(0,.9,.1),10,ncp=0) works
    Remy> very well).

    MM> Merci, Remy, for the report.

    MM> Note that I've been working on improvements to "pnchisq"
    MM> (C) and "qnchisq" for R-devel for several days now.

    MM> You will see fixes there pretty soon, I hope.

As a matter of fact, the bug you mention isn't new.
That's been PR#6421  which has been fixed several weeks ago in
both "R-patched" and "R-devel". 
(The other things I said above are true in any case.)


More information about the R-devel mailing list