Package for Bayesian inference?

We had a Gibbs sampler about 15 years ago. But we deleted it from subsequent public library versions in favor of the more general methods. Perhaps it’s time to revive the Gibbs sampler in the library. Regardless, millions of parameters might be too many for any sampler. I suspect the resulting model target density will be degenerate unless on the order of billions of input data is fed to the model.

If you have any comments to make the library more useable and user-friendly, we’d tremendously appreciate your feedback here or here.

I tested it with a small example, and it was really easy and straightforward!

1 Like