Comment by NunoSempere

That seems like the kind of problem that would be easily done through monte-carlo approximation? How hard is it to get 1M random rows in a postgres database?