iris-hep/analysis-grand-challenge

W+jets scale variation systematic changed after correctionlib implementation

ekauffma opened this issue · 0 comments

The up and down scale variations were originally +/- 2.5%, but this was accidentally changed to +/- 3% in #119

This effect can be seen when comparing output histograms (restricted to "scale_var_up")
Before:
Hist( Regular(25, 50, 550, name='observable', label='observable [GeV]'), StrCategory(['4j1b', '4j2b'], name='region', label='Region'), StrCategory(['wjets'], growth=True, name='process', label='Process'), storage=Weight()) # Sum: WeightedSum(value=170748, variance=7.36607e+06) (WeightedSum(value=182741, variance=7.88344e+06) with flow)

After:
Hist( Regular(25, 50, 550, name='observable', label='observable [GeV]'), StrCategory(['4j1b', '4j2b'], name='region', label='Region'), StrCategory(['wjets'], growth=True, name='process', label='Process'), storage=Weight()) # Sum: WeightedSum(value=169919, variance=7.29472e+06) (WeightedSum(value=181854, variance=7.80709e+06) with flow)