Choose timezone
Your profile timezone:
OpenMAPP informal teams catch-up
→
Europe/Zurich
OpenMAPP informal teams catch-up
Zoom Meeting ID
69450692292
Host
Andy Buckley
Alternative host
Sabine Kraml
Useful links
Join via phone
Zoom URL
HS3
We discussed ongoing slow evolution of the measurement-uncertainties format. Andy will prod the discussion again, since it's been a couple of weeks since requesting a non-covariance example in the latest proposed structure. See https://github.com/hep-statistics-serialization-standard/hep-statistics-serialization-standard/discussions/88
This isn't a blocker on the majority of our projects/tools that operate at event-yield level, though: Jack to discuss with CutLang, CheckMATE, MA5 (himself!), HackAnalysis etc. about conversion of SR yields to a common intermediate HS3 representation that can be standardly merged/patched into HS3-based likelihood models for reinterpretation. Spey already has features that can help, maybe avoiding duplication, and is open to feedback. Also noted that PyHS3 currently can't load non-simple HS3 files, so Spey=HS3 is currently dependent on ROOT's apparently more robust implementation -- Jack would like validation examples if possible. Jack to report the issue to Giordan etc.
Krysztof reported an issue with round-trip HS3 in ROOT 6.40: models saved as HS3 can be re-loaded but are then unable to be minimised. Kryzstof to make a bug report to ROOT.
EM Creator and static ADL analysis
Nice progress reported by Ekin. Presentation, feedback and questions answered re. details of Rivet implementation. Ekin/Kagan to update asap on implementation / readiness for testing. Ekin to investigate necessity/possibility of each tool having incompatible dependencies in Conda; Andy's impression was that the conda-forge package set should be / has to be consistent: definitely easiest/best if we can avoid incompatibilities that would mean the OpenMAPP tools can't co-exist! The mapping between different tools' analysis names should be built automatically from our published JSON files used by HepData, using Inspire ID as primary key: action on WHO?
More updates on static analysis results across multiple analyses with non-identical cut variables in next monthly meeting.
Overall strategy/progress
EMCreator/Bakery needs to be publicly usable asap, so we can start setting up the final combined-tools physics study. HS3 intermediate yield/observable implementation needed for next time. To be used as input to Spey / likelihood HS3 combination. HS3 measurement design to continue being pushed forward. Thanks, everyone!
Heads-up: in-person meetings anticipated around/before the next CERN Reinterpretation subgroup meeting, and a closing meeting in early 2027. Sabine to provide more detail when possible.
There are minutes attached to this event.
Show them.