Here’s the work that I’ve done on validating the soft muon tagger.  I’ve taken four samples from data: muons from J/Psi, pions from K_s, kaons from D^0, and protons from Lambda.  The soft muon ID seems to work as expected, although the fake rate is higher than I’d like at low Pt.  Perhaps we could improve it by adding more likelihood variables (E_cal/P_trk ?) but that might be too much work for very little gain.

Here’s a plot of the efficiency and fake rate for varied likelihood cuts.  As expected, the pions are the largest contributors to the fake rate.  However, this may be because the Pt distributions are different for the three hadron types:  I have very few kaons or protons below Pt=10 GeV, which is the region where most of the fakes seem to come from.

Here are two plots of the Pt dependence of efficiency and fake rate for likelihood cuts of 0.5 and 0.9.  The kaons and protons aren’t included in this plot because of the low statistics at low Pt.

-Scott