Replicable Monte Carlo simulation with OpenMP

For a Monte-Carlo simulation, isn’t a set of variable results a good thing.

This could require a different random number generator for each thread, although to demonstrate a thread-safe RNG is a good outcome.

Perhaps you could do some basic stats tests for RNG values in each thread ?
Possible values could be
avg,
stdev and
consecetive values are independent ( would stdev (Xi-Xi-1) = 0 pass?)