Mr Cottrell....thank you for your
reply given below.
For what it is worth, please may I suggest to
all students/researchers, particularly
those studying time series, that at any time a
simulation is utilised from which
inferences are drawn, then always have
detailed knowledge about the data produced
by the RNG utilised. As you know an ARMA or an ARFIMA simulation relies
upon
the production of random numbers. So when using
a propriety Stats Package, its proper to know
the manner by which the RNG gives its
output and to be able to justify the distribution
that feeds the ARMA generation.
Mr Cottrell's package below is the very best
I've read.
Hudson
----- Original Message -----
Sent: Saturday, January 15, 2011 11:20
AM
Subject: Re: [Gretl-users] the gretl
RNG
On Sat, 15 Jan 2011, Dr RJF Hudson wrote:
> Mmm
..random number generation is a massive subject
> never too far from a
practitioners mind, is it.
> Its always attached to a question mark with
me.
> Full marks are surely due to the wizards behind the
reseach.
> In one sentence or two, do you know, please what tests
are
> used to demonstrate randomness
> of the output observations
?
> Just asking, is all.
The currently canonical test suite seems
to be TestU01, devised by
P. L'Ecuyer and R. Simard. See
http://www.iro.umontreal.ca/~simardr/testu01/tu01.html
The
elements of the test are derived from various sources,
including several
from Donald Knuth of TeX fame.
Allin
Cottrell
_______________________________________________
Gretl-users
mailing list
Gretl-users@lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-users