[Rd] Bug in pbinom? (PR#5727)

[Prof Brian Ripley]

Oh, come on: have you never heard of the Poisson approximation?

> ppois(1, 1.95*seq(1.02,1.03,10^-3), lower=F,log=T)
 [1] -0.5259247 -0.5250276 -0.5241321 -0.5232383 -0.5223462 -0.5214557
 [7] -0.5205669 -0.5196798 -0.5187943 -0.5179104 -0.5170282

Probably you believe that computers never make errors too.

On Fri, 12 Dec 2003 liebscher at gsf.de wrote:

> Dear colleagues,
> 
> the command 
> 
> pbinom(size=1.95*10^10,prob=seq(1.02,1.03,10^-3)*10^-10,q=1,lower=F,log=T)
> 
> produced the output 
> 
>  [1] -46.0120973 -46.0101369 -46.0081784 -46.0062239 -46.0042692 -46.0023165
>  [7]  -0.5205666  -0.5196801  -0.5187944  -0.5179102  -0.5170278
> 
> 
> what seems strange. Has this survived in newer versions?
> 
> Best regards,
> Volkmar Liebscher
> 

-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595