The shape of the track efficiency (as a function of pT) looked odd when I calculated it from the Run 9 dijet embedding sample (see the blog post below).  I calculated it using all detector and particle level tracks satisfying the relevant QA cuts (described in the same blog post).  However, the embedding MuDst's store the index of the particle level track that each detector level track was matched to, and they store all of the particle level branches.

https://drupal.star.bnl.gov/STAR/blog/dmawxc/update-02172018-run-9-pp-efficiency-embedding

I've recalculated the efficiency (in pT) using only detector tracks which were successfully matched to a particle level track, and all particle level tracks (in acceptance).  The efficiency still looks weird when using matched tracks.  This is only for 1 bin of partonic pT (pTparton = 9 - 11 GeV/c), and that may be why it still looks weird...

Below are some plots comparing various detector level quantities (labeled "match") to their matched particle level counterparts (labeled "particle").  Again, this is only for pTparton = 9 - 11 GeV/c.

Note [02.23.2018]: In the right panel of the 1st plot, the efficiency was calculated using detector level tracks with 'idTruth > 0' (indicating a successful match to a particle level track) and all particle level tracks stored in the 'StMuMcTrack' branch of the MuDst which were in our acceptance (pTtrk > 0.2 GeV/c and |eta| < 1).  In the left panel of the 1st plot, the efficiency was calculated using all detector level tracks in our acceptance and all charged particle level tracks stored in the '*.geant.root' file which were in our acceptance.