@golosio there is a serious bug in the randmt library: init_randmt_auto_r doesn't actually set the generated seed to the rng instance, and instead uses it for the global rng...
I fixed the bug, and removed all code involving the global rng, to ensure this kind of problem cannot occur again.
I also warned the author of randmt about this bug.
I will also push a commit with a different mechanism to generate high quality seeds as I am not happy with the algorithm that is used here: the seeds are all very similar...
@golosio there is a serious bug in the randmt library: init_randmt_auto_r doesn't actually set the generated seed to the rng instance, and instead uses it for the global rng...
I fixed the bug, and removed all code involving the global rng, to ensure this kind of problem cannot occur again.
I also warned the author of randmt about this bug.
I will also push a commit with a different mechanism to generate high quality seeds as I am not happy with the algorithm that is used here: the seeds are all very similar...