Alexander Held

The "as desired" part is arguable, I would personally find a format where the POI entry is a list of strings better in the longer term. This is a breaking...

I believe a change could maybe be introduced here https://github.com/scikit-hep/pyhf/blob/270d59d3900fdef7549015db45260e85f83e33b6/src/pyhf/pdf.py#L29-L41 or in https://github.com/scikit-hep/pyhf/blob/270d59d3900fdef7549015db45260e85f83e33b6/src/pyhf/parameters/utils.py#L10 An addition of ```python if parameter.get("bounds"): parameter["bounds"] = [tuple(parameter["bounds"][0])] ``` before the `reduce_paramsets_requirements` call works for the...

Sorry if the example was misleading, when appending `{"name": "mu", "bounds": [(0, 10)]}` the validation will fail. So from that perspective, and also from the perspective of starting out with...

@lukasheinrich's example looks nice and could clean up the workspace building in `cabinetry` a bit. Currently it is like this ```python if sample_affected_by_modifier(sample, NormFactor): modifiers.append({"data": None, "name": NormFactor["Name"], "type": "normfactor"})...

I don't understand how the first two results are the exact same, but if this is a toy vs non-toy issue with a 4 sigma effect (which, looking at the...

I missed the sqrt required to go from q0 to the significance in the previous comment, this should be a 2 sigma effect of course: sqrt(2500) = 50, so a...

The last two numbers scale with the number of signal events, so these are something else.

At the moment this doesn't run any toys as far as I can tell, so I'm assuming this is a question about the calculators, and not about asymptotic vs toy...

I'm not sure whether I understand the direction of this issue correctly. Is it specifically about calculating expected discovery significance without any reference to observed data? `cabinetry.fit.significance` does not do...

Given the comments above, I guess the source is already clear, but just in case here is a simpler reproducer: ```python import pyhf spec = { "channels": [ { "name":...