maechler@stat.math.ethz.ch
2006-Feb-07 10:04 UTC
[Rd] Problems in d-p-q-r functions (was PR#8528, but unrelated)
>>>>> "BDR" == Prof Brian Ripley <ripley at stats.ox.ac.uk> >>>>> on Tue, 7 Feb 2006 08:52:43 +0000 (GMT) writes:BDR> For the record, some of these claims are untrue: >> df(0, 2, 2) BDR> [1] 1 >> df(0, 1.3, 2) BDR> [1] Inf Well, these first two I had fixed in the mean time. So Ian was right about reporting them. >> x <- 1e-170 >> pbeta(x, x, x) BDR> [1] 0.5 ( but Ian only explicitly mentioned dbeta(x,x,x) for x = 1e-162 and that *did* need to be fixed -- and I see you did fix it; thanks a lot! ) Martin BDR> ............................ BDR> ............................