Matt Haberland

Thanks @melissawm. I think the "R types" refer to the "Types" listed in the documentation of [R's `quantile`](https://www.rdocumentation.org/packages/stats/versions/3.6.2/topics/quantile) function. I just removed this because it is confusing, and anyone already...

Since this has been close to its present state for a few months, I'd ask whether we can merge as-is, and if there are additional changes to be made, they...

I see the Windows CI failure in another PR (gh-26386); seems unrelated. After fixing the lint issue I squashed to clean up the commit message (since not everyone squash-merges), thinking...

@xuefeng-xu @lsorber would you be ok with that (https://github.com/scipy/scipy/issues/19016#issuecomment-1887982497)? Are there other remaining changes to be made to BC to set the precedent for YJ? (Separately, would someone be willing...

1. Yes, that's essentially what I suggested with the addition of support for negative lambda. Looks good. 2. I'm not sure if it really needs to adjust line 12. TBH,...

That's fine @lsorber. You just need to convince @xuefeng-xu, @lorentzenchr, and then the SciPy mailing list that we need another parameter now rather than [waiting until it's requested by users](https://github.com/scipy/scipy/issues/19016#issuecomment-1887982497)...

Sure. In that case, I think just @xuefeng-xu and @lorentzenchr would need to be on board.

@lsorber sounded like @xuefeng-xu and @lorentzenchr were on board with increasing the buffer. Want to open a PR? I'll review and merge after they approve it.

For the numbered problems, the objective function value from COBYQA is always better than that of COBYLA, and it beats Nelder-Mead on 44/53 problems. The histogram of function evaluation counts...

Ah. It's tough to compare the performance without knowing the objective function values. I can create an algorithm that runs pretty quickly but doesn't solve the problem : ) ~~But...