JER splitting
The recommendation is now to split JER uncertainties in (6) pT/eta ranges: https://twiki.cern.ch/twiki/bin/view/CMS/JetResolution#Run2_JER_uncertainty_correlation
For us I think there are 3 options:
- Do this during nanoAOD postprocessing step we run. Main drawback: this adds 10 branches for jets (most of which would be useless most of the time if the jet is not within the given pt/eta range) and 10 branches for the MET. Advantange: easy to do.
- From the single
jer
up/down variations, let bamboo create 6 up/down template pairs (taking either the nominal or the variated jet pt depending on its pt/eta). A bit more involved, and pretty complicated to propagate to MET properly? - Do 1) on-the-fly as for the other variations (CP3-style). Should be straightforward too.
My question is, do you think 2) is feasible or should I go for 1)?
Edited by Sebastien Wertz