Study of correlation between AL(+eta) and AL(-eta), both determined from the same 8 statisticaly independent spin sorted yields.
Method: generate & reco AL 5000 times, produce correlation plot, compute correlation.
Code snipets below show key parts of both steps.
Simplifications: lumi[spin]=const, ALL=0, P=1
Attached code allows to run full simu in ~10 seconds.
Table 1 shows how correlation changes depending on assumed asymmetries.
Conclusion:
Double-sided reco of AL
Single-sided reco of AL determined from the same 8 statisticaly independent spin sorted yields.
Reminder of the meaning of correlation coefficient (from Wikipedia)
Example of W events, more like this and also QCD events are in attachments
In the background estimation for the W yields we account for W -> tau decays where the tau decays semi-leptonically to and electron and two neutrinos. In PYTHIA the tau polarization is not considered in the decay, which results in an underestimation of the W -> tau yield which satisfies our requirements. The suggestion in the PYTHIA manual is to use a program called Tauola, as an afterburner to correctly treat the polarized tau decay. I inserted Tauola in starsim to process the tau decay before the full GEANT simulation of the detector response. Attached is a summary of the study of polarized tau decays with the Tauola program.
The conclusion is that the results of the Tauola afterburner in the PYTHIA simulation is consistent with the expectation of a simple model of the Michel decay spectrum (provided previously by Carl). The Tauola afterburner is intended to be used in our background estimation to replace the ad-hoc correction for polarized decays used for the Run 9 publications.
Tauola is referenced in the Pythia manual, and a reference for it can be found at:
http://arxiv.org/pdf/hep-ph/0312240v1.pdf