Wrapping up our investigation of the tracking efficiency, I compared the tracking efficiency between two situations: one where I only require a trigger at the particle-level (a generated charged hadron with the right pT and eta), and one where I require a trigger at the particle-level and a trigger at the detector level (a reconstructed TPC track with the right pT, eta, and our usual track qualifications).

Previously (December 2018 and before) we were seeing efficiencies around 90%, which is what we expect from others' studies. Back then, I was not requiring a trigger when I calculated the tracking efficiency and looking at all tracks in phi-space. However, from January 2019 and onwards, I have been requiring a trigger at the particle level (to enhance pi0 statistics from the embedding) and almost always looking at only recoil tracks. Since then, I've been seeing efficiencies around 80%.

So here is a comparison between requiring only a particle level trigger and requiring both a particle and detector level trigger:

Yikes! So at low pT, we reach ~90% with a detector level trigger BUT then we settle back down into ~80% at higher pT. The dip around ~10 GeV/c for the case of a detector level trigger is something I've seen before: there's an odd depletion of high pT tracks in the trigger pT region when we use a track as a trigger.

So what's going on? I suspect it has to do with the definition of recoil tracks: in the above plots, I selected the particle level recoil tracks based on their delta phi with respect to the particle level trigger phi, and the detector level recoil tracks based on their delta phi with respect to the detector level trigger phi. There's no guarantee that the detector level trigger is the particle level trigger.

Right now, I'm running a check wherein I use the particle level trigger phi to calculate both the particle level and detector level track delta-phi. And I'll post an update tomorrow.