[Rd] Length of seed for l'Ecuyer-CMRG

[Prof Brian Ripley]

On 22/01/2013 23:22, Marius Hofert wrote:
> Dear expeRts,
>
> ./src/library/base/man/Random.Rd says that L'Ecuyer requires a seed of length 6.
>
> ./src/library/parallel/man/RngStream.Rd also mentions this, but only in the text
> part; In the "Arguments"-part, it says that "seed" has to be of length 7
>
> Also:
>
> ,----
> | > RNGkind("L'Ecuyer-CMRG")
> | > length(.Random.seed)
> | [1] 7
> `----
>
> Is the docu old? Some clarification on the two .Rd files would be great.
>
> Cheers,
>
> Marius
>
> ______________________________________________
> R-devel at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-devel

What the help page actually says is

     seed: An integer vector of length 7 as given by ‘.Random.seed’ when
           the ‘"L'Ecuyer-CMRG"’ RNG is in use.  See ‘RNG’ for the valid
           values.

.Random.seed is not the seed per se, it is an indicator of the random 
number generators in use, followed by the seed.  So for a generator with 
seed of length 6, .Random.seed is indeed of length 7.

-- 
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