Applying an MCMC Step to Sample from Distribution with KL Divergence Term

Thanks, @ricardoV94 ! pm.Flat was the concept I was missing for how to effectively use pm.Potential.

That being said, we continued to run into other issues and have ultimately decided to shelve this pymc implementation of our algorithm for now. Thanks again for the help and I hope we’re able to circle back to pymc again in the future once the algorithm itself is better understood.

1 Like