Yes, that is precisely the type of example I’m looking for. Related to the pymc3-hmm repository, I’ve often come across use cases where a dynamic Gaussian model / Kalman filter- centric distribution would also be of great use. Thanks for providing these links.