Hi @jessegrabowski, I replicated your solution to perform GRW by hand and indeed I get the correct posterior for Z, but weird inferences with a bunch of divergences for the standard deviation parameters (see image below). Do you have any idea why this is the case?
@ricardoV94’s answer solved the GaussianRandomWalk. This is what I get when I put the time in the last dimension.
However, I need to implement a custom RW that depends on other variables not only the previous one in the same chain, so I will need to construct an RW by hand as you did in your example.
Thank you!