October '21 in Grenoble (SModelS Fest '21)
Participants: Andre (virtual), Gael, Humberto (Oct 6-13), Jan (Oct 24-26), Sabine, Timothee, Wolfgang, Jamie Y. (virtual)
Topics targeting v2.2:
- Full and simplified pyhf likelihoods
- Likelihoods from obs+exp limits (improved truncated Gaussian)
- Analysis combination
- Database extension
- new ATLAS and CMS results
- new home-grown efficiency maps
Moreover (likely longer timescale):
- ML database
- ML cross section computer
- ...
SModelS 2.1.1:
- improved truncated Gaussians in likelihoodsFromLimits (but leave as experimental feature)
- caching weight matrix in simplified likelihoods
- print-all-datasets option, needed for aggregation
- tiny fix in mybinder link (see https://pypi.org/project/smodels/)
- small fix in testPrinter, make testSLHAPrinter and testTextPrinter tests fuzzy
- small change with pyhf expected llhds (posterior expected -> a priori expected)
Envisaged for SModelS 2.2:
- make truncated Gaussians a normal feature (requires more documentation)
- expected combined likelihoods
- possibly some more work on combinations (e.g. make combos between 8 and 13 Tev and/or ATLAS and CMS possible)
- possibly both versions of 'expected'
