The pm.Potential requires a scalar as the input, which can be the .sum() of your vector.
If you want to track the vector elements in the trace/InferenceData, you can wrap the vector into a pm.Deterministic
Does that help?
The pm.Potential requires a scalar as the input, which can be the .sum() of your vector.
If you want to track the vector elements in the trace/InferenceData, you can wrap the vector into a pm.Deterministic
Does that help?