Were you able to rewrite your model as a custom distribution? I am facing a similar issue with a model. I was looking at trying to use an approach similar to modelling a non-homogeneous HMM to get around this issue, but with no success.
Were you able to rewrite your model as a custom distribution? I am facing a similar issue with a model. I was looking at trying to use an approach similar to modelling a non-homogeneous HMM to get around this issue, but with no success.