-
Notifications
You must be signed in to change notification settings - Fork 21
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
fix: compatibility with pyhf 0.6.3 (parameter label API, auxdata) #248
Conversation
scikit-hep/pyhf#1560 is probably going to affect this PR, given that this PR is in response to scikit-hep/pyhf#1536, but I haven't looked over it so you might be fine. |
3898ed0
to
c00332a
Compare
Thanks for the heads-up! scikit-hep/pyhf#1560 does not interfere with this, as now the actual parameter naming format is irrelevant to |
20b189e
to
094088b
Compare
094088b
to
a94fde5
Compare
Codecov Report
@@ Coverage Diff @@
## master #248 +/- ##
=========================================
Coverage 100.00% 100.00%
=========================================
Files 19 19
Lines 1798 1787 -11
Branches 279 275 -4
=========================================
- Hits 1798 1787 -11
Continue to review full report at Codecov.
|
f506cdc
to
58614f5
Compare
scikit-hep/pyhf#1536 (see also discussion in scikit-hep/pyhf#867) added
model.config.par_names()
to thepyhf
model config API. This provides the same functionality ascabinetry.model_utils.parameter_names
, but the parameter names are now also passed through to MINUIT. A fix is included to pass the correct names when using MINOS (they used to bex0
,x1
etc.).cabinetry.model_utils.parameter_names
is removed, since the functionality is now available directly inpyhf
.The parameter naming format has changed:
parameter[bin_0]
->parameter[0]
etc., and the index is now included even for parameters like shapesys/shapefactors that only have a single component (because there is only a single bin in the region, see scikit-hep/pyhf#1560).scikit-hep/pyhf#1562 introduces an API change for the auxdata kwarg naming, from
with_aux
toinclude_auxdata
.cabinetry
usedwith_aux
also in its own API, and this now switches toinclude_auxdata
for consistency withpyhf
.Due to a
typing-extensions
requirements conflict that results in thetensorflow-probability
installation being incompatible with the installedtensorflow
version (see also scikit-hep/pyhf#1595, tensorflow/tensorflow#51743), thepyhf[backends]
installation is removed from thecabinetry
test
setup extra. It is now available via a newpyhf_backends
setup extra and installed only when needed for the backend tests. That allows installing a workingtensorflow
setup (and picks up a conflict withblack
that does not matter at that point in the CI, asblack
already ran earlier).Breaking changes:
cabinetry.model_utils.parameter_names
in favor of thepyhf
implementationmodel.config.par_names()
model_utils.model_and_data
andmodel_utils.asimov_data
: renamedwith_aux
kwarg toinclude_auxdata
for consistency withpyhf
test
setup extra no longer includes allpyhf
backends, now moved topyhf_backends
setup extra