search

cms-analysis / flashgg

20 stars 158 forks source link

Fix for dumping systematic variation of theory weights for ultra-legacy samples #1268

Closed threiten closed 1 year ago

threiten commented 3 years ago

Only tested with ultra legacy 2017 sample. Compatible with ReReco as well (tested on 2017, given the right settings) mc2hessian csv file should be confirmed to be correct here: https://github.com/cms-analysis/flashgg/blob/dd6661a55448c403b46d1155510c67a313cd44a8/MetaData/data/MetaConditions/Era2017_legacy_v1.json#L240

This is only a quick fix and it should be reviewed in the future how this information can be accessed in a more consistent way

threiten commented 3 years ago

The PR should now be in a state that produces the correct pdfWeight variations for RR16, UL17, RR18. It was tested for those datasets with the differential analysis. But it should be tested again in the main branch with one of the standard analyses.

youyingli commented 1 year ago

PR #1285 PR #1287 had fixed the PDF issue for UL and ReReco. Close this PR. Thanks for this contribution.