Alexander Held
Alexander Held
Thanks for opening the PR! Looking at the original implementation, I was trying to remember why I wrote it in that specific way. I cannot think of any obvious issues...
Resolved by #479, thanks!
The next `pyhf` release after https://github.com/scikit-hep/pyhf/pull/2611 will remove this backend so at that point support from `cabinetry` can also be dropped.
From what I've seen it is difficult to change the `pyhf.Model` object dynamically. It is also not generally possible to build a new model that internally has samples merged but...
Here is something I wrote recently to merge samples for plotting, but it does not get the per-sample uncertainties correct because to calculate those, the merge needs to happen before...
### Reported uncertainty vs per-sample input uncertainty The per-sample contribution for the MC statistical uncertainty is obtained by varying the sample yields up and down by the relative MC statistical...
The partial coverage question is no longer relevant now, right? With the restructuring this looks fine.
Smaller reproducer: ```python import uproot with uproot.open("https://xrootd-local.unl.edu:1094//store/user/AGC/samplegame/sample1.root") as f: f["Events"] # this works f["Events"].arrays("Jet_pt") # this breaks ```
Does `events.arrays(["Electron_pt"])["Electron_pt"]` work? I imagine `events.arrays()["Electron_pt"]` is slow because it will read the full file.
Hi @marcus-vgr, how long is "a very long time" for the graph creation? This should hopefully not be more than a few minutes at most. In case your graph becomes...