Thanks for the replies! I am even more confused now though.
I am working on a Linux cluster, in a conda environment with python 3.7. For me the output looks like this:
“Sample initial stage: …
Stage: 0 Beta: 0.000 Steps: 5 Acce: 1.000”
It prints these lines until it finishes and it gives me a trace of shape (n_draws,). So just the line with the number of chains is missing, that I saw in the example.
I tried setting up a fresh environment, but I still have problems (see below). My further questions are:
-
Since the example I linked in my original post uses pymc 3.9 as well, but they can still sample multiple chains with SMC, I thought that should be possible with pymc 3.9.2. Is there any keyword apart from n_chains that I a missing?
-
As described above, in the source code of the newest pymc 3 version on github I can see the parameter n_chains and e.g. the IMH kernel in the sample_smc.py file. The source code of pymc 3.11.4 and older versions does not contain these updates however. So, are these not included in any packaged pymc version yet?
-
I tried installing pymc directly from github to get the newest features, following the instructions given at Installation Guide (Linux) · pymc-devs/pymc Wiki · GitHub in a fresh environment. But, then the installed version is 4.0, which I do not want. It then says: “You are running the v4 development version of PyMC which currently still lacks key features. You probably want to use the stable v3 instead which you can either install via conda or find on the v3 GitHub branch: GitHub - pymc-devs/pymc at v3”, which is precisely where I got the installation from…
So, I am not sure what to do here, I just want to have an SMC sampler which can sample multiple parallel chains and uses the IMH kernel…