...
Ross prepared his (& my) first guess of systematic errors contributing to the Xsection, see last item below.
For now the outstanding item (20%) is due to the gauss shape? used in the V-scan fit. We need more details (plots, comparisons) to discuss it.
Embedding - using it would address several uncertainties at once:
- loss of true Ws due to pileup tracks or EMC hits in either of the veto-cones
- in particular beam-background tracks affect veto in the Endcap
- amount of contamination to reco Ws due to accidental pileup track pointing to trigger pi0 cluster
- contamination due to non-collision trigger messing up lumi calculation
- loss of track efficiency due to removed clusters tagged on-line as deconvoluted, here beam background also plays the role
Significant time was spent arguing w/o concrete arguments:
- one could approximate X by Y, but not sure how accurate will it be done, vs.
- I want either embedding or deconvolution, because I do not trust anything else
This is representative to expected discussion on a broader forum and affected my suggestion how to proceed.
Material used during discussion
...