Matt Haberland

I don't mind keeping it. I just wonder whether it's something we want to put a lot of development time into. gh-7819 is a pretty big PR, and I thought...

I'm not aware of a table. There was a table of local optimizers in the SciPy paper, but I don't think there's anything for local or global in the documentation.

> Does the benchmark include problems with lots of variables? I did not write the benchmarks. I ran them so that we could merge this PR, but I don't really...

@tupui tried to resolve merge conflicts. Hope I did it correctly. (Oops, probably shouldn't have skipped CI. )

@NeilGirdhar could you extend this list by looking for similar problems in other distributions? For instance, the first item has been reported for `invwishart` in gh-6474. Of course, that issue...

@NeilGirdhar I removed `vonmises` from the list because it's a univariate distribution. The univariate distribution infrastructure does not really support circular distributions, and we're trying to address that, but it's...

@h-vetinari I meant that gh-15928 linked to gh-4598 and gh-6038. Now it links only to gh-6038, but I added some commentary.

Copying @NeilGirdhar's comment from gh-15928 over here: > Still, if you're going to clean up the interface, it might be nice from a user point of view to provide the...

> It used to be basically a single rst tutorial per submodule, do we have a better way now? Very good question. Currently stats is broken out; see `stats.rst` and...

@ev-br I'll merge this, but I wanted to check with you - in stats these days, we typically squash merge (noting co-authors in the commit message) unless the separate commits...