New in this release:
- A
--resume-sampling
argument. This allows you to resume a Markov chain from a past session. So if you don't achieve convergence in your first attempt, you can continue collecting samples without having to start all over again. - We enable multithreaded parallelism with a
--multithread
flag. This parallelizes each step of the Markov chain and allows SSPS to accumulate more samples in less time.