Matt Haberland

Thanks @AtsushiSakai. To provide further evidence of the improvement, I used the `_mwu_state` object to compute all values of the survival function (that is, at all values within the support...

@lsorber happy new year! how does this relate to gh-19652? Please coordinate with @xuefeng-xu to create a unified PR. In the meantime, I need to finish reviewing gh-19631 to finish...

Marking this as a draft pending agreement in gh-19016. @xuefeng-xu please coordinate with the others to agree on a plan for the two functions. I don't need to be part...

Thanks for all your help on `rankdata` and `mannwhitneyu` @AtsushiSakai! Do you ever use `wilcoxon`? While working on this PR, I noticed some things about Wilcoxon I think should be...

@chrisb83 can I interest you in this PR? It speeds up `wilcoxon` and allows passing a `PermutationMethod` instance for `method` to improve accuracy in the presence of ties.

Thanks @chrisb83. I added numbers to the points about changes so that it's easier to refer to each. Thoughts on how we go about making these changes? What can be...

> I guess changing the return format would be difficult and not worth it. what is your view? I think it's something that should be changed in the long run....

@chrisb83 maybe let's set aside all of the interface changes to a separate PR, since I'm not sure the changes will be popular even if they are improvements. ~~I've updated...

Thanks @chrisb83. Corrected the null distribution mean edge case, removed `_wilcoxon_1d`, and addressed the new comments.

OK @chrisb83 CI is finally green. I had a bit of back and forth with the `test_present_but_private` tests but I think that's all set. (Hopefully after 1.14 branches we can...