Re: [Blitz-support] MT19937 question/bug?

[Patrik J. <pa...@uc...>]

Hey again,

Never mind my earlier message. I worked through it and the
implementations are equivalent. I also confirmed by generating a
sequence with the reference implementation and comparing, and it's
identical (accounting for the slightly different way of seeding).

Btw, MN have changed the seeding algorithm in their current
implementation due to possible bad behavior with the old one (see
http://www.math.sci.hiroshima-u.ac.jp/~m-mat/MT/MT2002/emt19937ar.html).
As I go ahead with this change, I'll also update the seeding to their
current method unless anyone objects.

cheers,

/Patrik