@joelkuiper: just a quick notice that multivariate MCMC has been added to Bayadera, and the first preview works quite well (and fast!). It also creates histograms of the marginal distributions, and sorts them on the fly, so you can easily identify 95% HDI even for multimodal marginals. It still needs a lot of polish and testing, but I've already created a few examples of hierarchical models from the Doing Bayesian Data Analysis book.

Nice!

I’ll definitely check it out 🙂

@joelkuiper: A quick guide to help you see it from the user's perspective: 1. Specifying the kernel of the (hierarchical) model: https://github.com/uncomplicate/bayadera/blob/master/test/clojure/uncomplicate/bayadera/examples/dbda/ch09-1mint-1coin.h 2. Loading the model: https://github.com/uncomplicate/bayadera/blob/master/test/clojure/uncomplicate/bayadera/examples/dbda/ch09_hierarchical_models_test.clj#L45 3. Running the analysis (several options are called here, normally you do not need them all): https://github.com/uncomplicate/bayadera/blob/master/test/clojure/uncomplicate/bayadera/examples/dbda/ch09_hierarchical_models_test.clj#L51 Those parts are pretty much simple (or not much complicated) already. The rest of the file is a visualization boilerplate that I haven't have time to simplify yet, but I am going to.