I'm trying to run an analysis, in which my variables are sampled with the latin hypercube function lhdesign.
I'd like to stop my simulation when a stopping criteria is satisfied, therefore I'd like also to sample only the necessary points instead of preparing all the combinations before.
From what I know, random number generation ins't actually random, but it follows a kind of pattern (I'm not an expert).
My question is: if I stop the sampling and then I restart it, does the sampling restart from the last point, or also the generation algorithm restarts?
To be clearer, suppose I need to sample N points (N is not determined, since it depends on when the stopping criteria is satisfied), but I want these points to be sampled in groups of 100. So I sample the first 100 points, the I the computation I need to do, check the criterion, sample the next 100 points and so on.
After the first 100 sample, is the first sample of the second group a kind of 101th sample, or the sampling pattern restarts such that the first sample of the second group is the same of the first sample of the first group?
I know it's not explained ina very clear way, but thank you anyway.