R devel - Sep 2004 - qchisq (PR#7212)

Full_Name: David Clayton
Version: 1.8.1
OS: Linux
Submission from: (NULL) (131.111.126.242)


qchisq behaves very strangely when ncp is passed as zero (forcing internal
qnchisq to be called) when first argument is small.

Eg

> qchisq(1-1e-6, 1, ncp=0, lower.tail=TRUE)
qchisq(1-1e-6, 1, ncp=0, lower.tail=TRUE)
[1] 1024

while, if ncp is unspecified,
> qchisq(1-1e-6, 1)
qchisq(1-1e-6, 1)
[1] 23.92813

david.clayton@cimr.cam.ac.uk writes:
> Full_Name: David Clayton
> Version: 1.8.1
....
> qchisq behaves very strangely when ncp is passed as zero (forcing internal
> qnchisq to be called) when first argument is small.
> 
> Eg
> 
> 
> > qchisq(1-1e-6, 1, ncp=0, lower.tail=TRUE)
> qchisq(1-1e-6, 1, ncp=0, lower.tail=TRUE)
> [1] 1024
> 
> while, if ncp is unspecified,
> 
> > qchisq(1-1e-6, 1)
> qchisq(1-1e-6, 1)
> [1] 23.92813
As Brian once put it, we can't fix old versions retroactively. 1.9.1
has (on "Linux" as well, although that is a bit unspecific for numeric
issues). 
> qchisq(1-1e-6, 1, ncp=0, lower.tail=TRUE)
[1] 23.92813
> qchisq(1-1e-6, 1)
[1] 23.92813

Possibly, this is a side effect of

    o   qchisq(1-e, .., ncp=.) did not terminate for small e.
        (PR#6421 (PR#875))

from BUG FIXES for 1.9.0.

-- 
   O__  ---- Peter Dalgaard             Blegdamsvej 3  
  c/ /'_ --- Dept. of Biostatistics     2200 Cph. N   
 (*) \(*) -- University of Copenhagen   Denmark      Ph: (+45) 35327918
~~~~~~~~~~ - (p.dalgaard@biostat.ku.dk)             FAX: (+45) 35327907