yayami3

I want to try to contribute. I have a question. `# event dimensions will be decided by init_dist` How is it possible ? init_dist decides batch dimensions, right ?

can be close by #1423

@fehiepsi Thanks. But still confusing. I made draft PR. I assume `init_dist` has shape `batch_shape` and `event_shape` is defined by t. Could you comment on PR ?

I think it does not make sense to consider the log density of beta under `do`. beta behaves like a constant. I do not understand in detail how numpyro handles...

@fehiepsi Absolutely! But I think there seems to be no established way to describe it. If you look at this [journal](https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2836213/pdf/ijb1203.pdf) Pearl represents the nodes as dots, not sure about...

@stergiosba Hi. Are you working on this issue ? Or else I want to do it.

@stergiosba Thanks for the offer. Which distribution are you targeting? I wrote a draft about the foundational classes and tests [here](https://github.com/yayami3/numpyro/tree/add_entropy)

@stergiosba Thanks for your comment ! I think it's a good idea, but there are other modules using `xlogy` as well, and it feels like there is a lack of...