Re:Random in the wild
by conistonwater on Tuesday December 09, 2014 @03:31PM
(#8725506)
> A Monte Carlo simulation that always gets the same results isn't going to be too hot.
A Monte Carlo simulation absolutely needs to be able to produce the same numbers every time if you want to have a hope of debugging it. In a release version, yes, sure. But even there, people will still have concerns about replicating other people's results, so being careful with random seeds is important.
Re:Random in the wild
by smeyer on Tuesday December 09, 2014 @07:01PM
(#8726701)
There's a difference between "always uses the same seed" and "can be seeded for reproducibility". You probably want the latter and probably don't want the former.
A Monte Carlo simulation absolutely needs to be able to produce the same numbers every time if you want to have a hope of debugging it. In a release version, yes, sure. But even there, people will still have concerns about replicating other people's results, so being careful with random seeds is important.