Two-stage Bayesian regression enforcing a fixed distribution (not Just Hierarchical regression)

I agree tend to agree. But I don’t know the internal workings of PyMC/graphs/derivatives ect.

My “hacky method” works but only for toy examples, any nonlinear function in the hierarchical model breaks the MCMC sampler.

How would you @ricardoV94 @junpenglao suggest inputting a fixed distribution in PyMC? Surely two-stage/sequential regression should be possible?