Cove Geary

Happy to incorporate the decision either way in the PR. I can get back to it early this coming week. A bit of feedback as a user; the fact that...

Also, not sure how important this is, but it seems that nearly all of the tests that do posterior predictive sampling currently use `return_inferencedata=False`. Is this fine left as-is or...

Came across this as it would be helpful for a use-case with pymc. Is there much interest in seeing this realized? My motivation is: it would be nice for me...

Hi folks, I was interested in this so I took a look with the latest codebase. As @flo-schu was finding, it's not sufficient to simply add the dim name to...

N.B. the tensorconstant name gets disappeared even if there is only one dim, a la ```python >>> pt.as_tensor_variable((pt.constant(5, name="disappears"),)).name None ``` Because normalize_size_parameter will always receive a tuple. A couple...

Forgot to mention: As a qualitative test, I also checked that the "monolithic" model in test_printing.py renders correctly in both MathJax and KaTeX.

I pushed some changes. Mainly, they clarify the exceptions that are thrown, add a number of printing tests, and fix an edge-case bug or two. I've also updated the original...

@larryshamalama of course, thank you for the review! I understand this PR ballooned a good deal and also recognize that I don't have the full big-picture vision that y'all have....