Lukas
the "modified" language comes from Alex Read's "MODIFIED FREQUENTIST ANALYSIS OF SEARCH RESULTS" https://cds.cern.ch/record/451614/files/open-2000-205.pdf , so I think it's fine. And the purpose of dividing by p_b is explicitly to...
We still need to add the ability for other types of tests (see https://github.com/scikit-hep/pyhf/pull/520), but even if you fix the test type you might still be interested in getting more...
yeah the expected band probably requires some work in that regard. also on the other end of the spectrum. The +1,+2 sigma correspond to µ^ values that are > µ_test...
i'm just saying for `f_sb = f(q_mu|mu' = mu) = 0.5*delta(0) + 0.5* chi2(qmu)` and p_values are `p_sb = 1-Phi(√qmu)` so for any observed (or expected from the alt hypothesis)...
leaving this from a manual calc here for me to follow up later. With proper qtilde it's looks better but unsure yet why this is not taken into account in...
In the end I think what we do is correct: It's just that with qtilde (i.e. bunded POI) the acceptance region in the FC neyman construction shrinks close to zero...
one last thing that might be affecting this is that this all assumes that the variance of µ^ for µ' = µ and µ' = 0 are the same, i.e....
based on the work in #1162 we need to decide what to do with this. If we now switch to reporting the expected `CLs, CLsb, CLb` and quantiles of the...
Docs build fine for me locally  this is my `pip freeze` output; ``` absl-py==0.9.0 alabaster==0.7.12 ansiwrap==0.8.4 appdirs==1.4.4 appnope==0.1.0 astmonkey==0.3.6 astunparse==1.6.3 async-generator==1.10 atomicwrites==1.4.0 attrs==19.3.0 awkward==0.12.21 Babel==2.8.0 backcall==0.2.0 black==19.10b0 bleach==3.1.5 bump2version==1.0.0...
nothings obvious to me ``` absl-py==0.9.0 absl-py==0.9.0 alabaster==0.7.12 alabaster==0.7.12 ansiwrap==0.8.4 ansiwrap==0.8.4 appdirs==1.4.4 appdirs==1.4.4 argon2-cffi==20.1.0 | appnope==0.1.0 > astmonkey==0.3.6 astunparse==1.6.3 astunparse==1.6.3 async-generator==1.10 async-generator==1.10 atomicwrites==1.4.0 atomicwrites==1.4.0 attrs==19.3.0 attrs==19.3.0 awkward==0.13.0 | awkward==0.12.21 Babel==2.8.0...