Thanks for the reply, and I indeed saw that DEMetropolis was implemented in 3.3, but some weird bugs on 3.3 showed up that prompted me to stay on 3.2 a few months back. Maybe I’ll give it a try again. (It is also not adaptive, but another variation of MCMC mixed with DE, but if it works, why not).
I didn’t realize that this method was also from 2006 like DRAM that I mentioned, so I guess it’s just that nobody was interested in implementing such methods before recently. Is the lack of time/interest from developers the answer to my question also (the question being why adaptive methods are not yet implemented)? I just wanted to make sure it wasn’t some other deeper mathematical reason that makes them not good in the end or something along those lines.