On Jul 7, 2:53*pm, spop...@speedymail.org (Steve Pope) wrote:
> Clay *<c...@claysturner.com> wrote:
> >On Jul 4, 12:55*am, Nicholas Kinar <n.ki...@usask.ca> wrote:
> >> I am trying to generate a Maximum Length Sequence (MLS) using a Linear
> >> Feedback Shift Register (LFSR).
> >When I think about generating a MAXIMUM length sequence, I woudn't
> >even consider something so short as just 4095 states before repeating.
> >I often use a Mersenne Twister, which has a period of 2^19937. Yes
> >that's correct. And it is easy to implement and free code exist on the
> >web for it.
>
> Turns out that engineering requirements come in all forms
> and sometimes you do not want, or cannot use, the more
> elaborate solution.
>
> For example, with a maximal-length sequence formed from a
> polynomial over GF(2), it's often practical to jump to
> a prescribed point in the middle of the sequence, or to
> examine the register contents and figure out where you
> are in the sequence. *Can a Mersenne twister do that?
>
> Steve

Hello Steve,

Yes you can start the twister anywhere you want in the sequence.
Likewise a small number (compared to the length) of observations can
tell you the starting state. It is not for cryptographic use because
of these properties. I'm not sure how one would measure the distance a
current state is away from a defined starting state without clocking
it through all of the intervening states. Here its long period makes
it take a while!

Clay