Add multiparameter auxiliary parameter priors
alizma opened this issue · 1 comments
The package currently limits auxiliary parameter prior choices to single-parameter distributions (namely the exponential and chi2). For greater generality, include multiparameter positive continuous distributions as possible priors. Here are the ones Stan supports.
Additional validation checks will be necessary, and on the Stan side, priors to the log-likelihood should be set via a vector rather than just a single vague parameter. The API change should only require that the dictionary for the aux prior is a list.
As of this commit, this has been made a feature. At the time of writing this, the supported priors for the model auxiliary parameter are: exponential, chi2, gamma, and inverse gamma. With the existing structure, it is possible to have other multiparameter priors.