[Octave-bug-tracker] [bug #58800] BIST for rng sometimes fails

[Michael Leitner]

Follow-up Comment #8, bug #58800 (project octave):

Markus, both your suggestions won't do: with respect to this bug here, it is
about initialization, not about generation random numbers. And a random number
generation framework typically does not deal with generating its seed on its
own from thin air, you have to provide some initial randomness.

I think that it is good to have a private implementation of random number
generation in octave, for reasons of reproducibility between versions. 

Yes, it is definitely important to have different seeds if one explicitly
calls for a re-initialization of the random number generator. But if the
generator has any worth, this should not be a problem: as I see it, it should
suffice to initialize the state when octave is started with the clock time,
and then, whenever a reset is called, to do some fixed operation on the state
(in the simplest case, just add 1). As I said, if the generator is good, this
should give you a new stream that has no recognizable relation to the previous
one. If you initialize at startup with the clock time plus what would be the
PID in linux, you would have different streams even if you started a number of
octave processes in parallel.

    _______________________________________________________

Reply to this item at:

  <https://savannah.gnu.org/bugs/?58800>

_______________________________________________
  Message sent via Savannah
  https://savannah.gnu.org/