Combine the nine dxy sideband regions in the fake estimate into one larger sideband.
Done. The fake estimates change slightly but are all within 0.5sigma (statistical) of the previous estimates. The AN is updated.
Compare the pileup distributions in ZtoMuMu, ZtoEE, and BasicSelection events. If there is a big difference, try reweighting and see how much it changes the estimate.
See the above plots. Using the ratios as weights to the fake selection (ZtoMuMu/EE + DisTrk (no d0 cut)), the overall weights applied to P_fake^raw would be:
|
ZtoMuMu |
ZtoEE |
nLayers = 4 |
0.994 +- 0.064 |
1.01 +- 0.34 |
nLayers = 5 |
1.013 +- 0.088 |
1.0 +- 1.4 |
nLayers >= 6 |
1.0 +- 0.21 |
1.02 +- 0.83 |
Despite the plots above these average weights are very consistent with one, e.g. the estimate does not depend on this.
If possible, find the justification for using dxy with respect to the origin for the track isolation pileup subtraction.
We could not find a justification and have concluded that regardless of what inspired it, calculating the track isolation with respect to the origin was/is a mistake. However, as noted in the ARC meeting, this mistake is confined to the selection of tracks to be included in track isolation sum. The effect of which is to reduce the efficacy of the track isolation requirement. Redoing this sum would require prohibitively large reprocessing but fortunately, due to the redundancy of this cut with the calorimeter isolation requirement for charged hadrons and electrons, and the muon delta R requirement for muons, the effect on the analysis is not significant.
We've found where we initially took this from several years ago, however it seems like it was taken incorrectly. We still however observe no pileup-dependence on the track isolation requirement.
V2 -- We've found where we initially took this from several years ago, applying the cuts used in the beamspot reconstruction; it seems like this wasn't the correct thing to use. The track isolation requirement is redundant with the ECalo requirement in rejecting backgrounds from pileup, so it should not be an important issue.
Suggestion: compare the nominal Gaussian fit to a flat line for NLayers5, as there's a concern the bias towards the PV changes as nLayers increases.
With a flat line the transfer factor is purely a normalization issue and has no uncertainty; it is always 0.02 / 0.45 = 0.0444. The table below is added to the AN:
ZtoMuMu NLayers5 (gaussian fit from NLayers4) |
ZtoMuMu NLayers5 (pol0 fit, finer binning) |
|
|
The flat assumption reduces the estimates by ~2/3 and the agreement is worse, especially as there would be no 40-50% fit uncertainty.