eb8680

Hi @karthikayan4u, thanks for your interest in contributing to Pyro! The immediate background information necessary to understand this proposal is in the [paper I linked to about ChEES-HMC](http://proceedings.mlr.press/v130/hoffman21a.html). Do you...

@arijc76 great, go for it!

Hi @arijc76, nice work! The reason you are seeing different values in your custom `predict` function is that it is drawing fresh samples for the random variable `obs`, as expected....

@arijc76 that should be fine - we can help you get CI tests working in the PR.

@JamesTrick sorry for the delay, that sounds great! Please let us know if you need any help getting started.

What about just making the indices `slice`s? Was there a reason we needed them to be tensors?

I guess we need tensors to do index arithmetic (e.g. `x[t-1]`). It seems like a better long-term solution would be to manage naming in markov contexts more automatically and avoid...

> But there is also a bug in our funsor code. I'll make a fix for it soon. After that, your model should run under the enumeration mechanism. Unfortunately, enumeration...

> Does it mean that, discrete variables (if enumerated) cannot be used inside the plate ? No, not exactly - it means that independent discrete variables enumerated inside of a...

> but as @eb8680 pointed out, we need a sequential plate for `xi`. This would work in the sense that enumeration would produce the correct factors, at least for `d...