A Practical Guide to MCMC Part 2: Further Resources

If you are coming here Part 1 then I'm sorry to tell you that this is not the fully fleshed out post you were hoping for. If you are coming here directly, good for you going straight to the "Part 2", but alas you may be disappointed.

Part 1 was written when I was finishing up my postdoc career and was still fresh in my MCMC techniques. Since then I got a new job in Data Science and haven't had a lot of time to get back to blogging and have not done any MCMC since my postdoc days. However, I have gotten a fair number of requests for more information about advanced MCMC techniques and this post is my weak attempt to disseminate some further resources for you, though they may be a bit more academic than the last post.

MCMC Resources

• Slides on some more advanced MCMC techniques: This talk and the code in the next point are part of a summer-school and also has some exercises if you want to try them out.

• Simple Differential Evolution and Parallel Tempering Implementation: This contains some basic classes (in outdated python 2) to do differential evolution and parallel tempering. Can be useful to see how these work in practice instead of just through equations.

• Brief yet technical intro to adaptive metropolis, custom jump proposals, and parallel tempering: This is from an academic paper which you are probably not interested in, but the appendix starting on page 17 has some useful information and more references if you are really interested.

  Well thats all folks, hopefully this is some help. I will mention that there are some great packages out there like emcee and pymc3 that may do a lot of this for you although I haven't tried them out.