Matthew Feickert

@lukasheinrich PR #1694 will use `tensorlib.percentile` over `np.percentile` to avoid the conversion. What loop in particular were you referring to? The

Not sure if @danielskatz has general thoughts in this area, given that we're trying to get a bit of a [standard form of `CITATION.cff` going for Scikit-HEP and IRIS-HEP projects](https://gitter.im/Scikit-HEP/community?at=629e7ceb06a77e1e18737c63).

We can look into adding this into an upcoming release once [`v0.4.2`](https://github.com/scikit-hep/pyhf/projects/7) and [`v0.5.0`](https://github.com/scikit-hep/pyhf/projects/8) are shipped — which means soon!

> ... it's mainly figuring out a consistent API that doesn't throw off a lot of users. @jagrundy @swertz @mswiatlo what would be helpful is to try to understand what...

> > Should we maybe have this test in addition to the old test instead of replacing it? > > We probably need the old test, but change the bounds...

@kratsg @lukasheinrich it has been a long time so can you help remind me of the plan? Was the idea to make a new `expected_result_1bin_lumi` fixture and model to support...

> Basically just have a fixture similar to one of them, but using the qtilde test stat instead of the q one -- with the values, etc... we need to...

> Only need to copy out the spec for one of the existing ones, add "qmu" to the name instead -- and change the bounds to `[-10, 10]` so that...

> FYI, typer is completely broken due to click 8.1, see [pallets/click#2226](https://github.com/pallets/click/issues/2226) or the typer issue. Yeah I think this PR is going to be in stasis for quite some...

We can start by migrating all the PUB notes from the [ATLAS Machine Learning Forum public results listing](http://atlasml.web.cern.ch/atlasml/papers/public/) that aren't already listed. We should also invite CMS to do the...