split r-hat

[bob-carpenter]

I added

• implementation of split-R-hat (with tests)
• additional validation in basic R-hat to ensure each chain has 2 or more draws (I added an additional test for this)

This resolves issue #36.

[codecov-commenter]

Codecov Report

Merging #37 (52a5086) into main (dff8143) will increase coverage by 0.18%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##             main      #37      +/-   ##
==========================================
+ Coverage   97.91%   98.10%   +0.18%     
==========================================
  Files          11       11              
  Lines         288      316      +28     
==========================================
+ Hits          282      310      +28     
  Misses          6        6              

Impacted Files	Coverage Δ
bayes_kit/rhat.py	100.00% <100.00%> (ø)

📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more

[bob-carpenter]

I can update for 100% coverage of the tests. And apply black.

[WardBrian]

A few minor comments

[bob-carpenter]

This is ready to review. Failures are all in MCMC tests, not in the rhat stuff I added.

This is all updated with 100% coverage tests. I just added split_rhat and rank_normalized_rhat to the same rhat.py file. The doc has full definitions and pointers to the original references.

[bob-carpenter]

@jsoules and @WardBrian. Thanks for all the comments. I'll learn Python one PR at a time.

I made all the fixes I could understand and marked them as resolved.

There are a couple points I didn't know how to fix and I would really appreciate guidance.

[bob-carpenter]

A few minor comments

27 of them and counting. :-(.

There are two remaining concerns that I do not know how to address:

• split r-hat #37 (comment)
• split r-hat #37 (comment)

[WardBrian]

I found one typo, but I promise this is my last comment. Feel free to merge after fixing