Previously, I had looked into the effect of requiring a trigger at the particle-level on our jet reconstruction efficiency:

https://drupal.star.bnl.gov/STAR/blog/dmawxc/update-06282018-run-9-embedding-using-phi-vs-delta-phi-and-particle-lvl-triggers-matchin

While the efficiency was drastically improved by the presence of a particle-level trigger, the shape was still off.  For some reason, the efficiency actually dropped off at high 'pT'... For reference, here's the reconstruction efficiency for R = 0.3 charged jets from the hadron-jet analysis:

Our efficiency should be close if not identical to the 60%-80% efficiency above.  After some investigation, I've determined that it was the way which I was calculating delta-phi which was driving down the efficiency at high 'pT'.  Up until now, the matching algorithm I've used goes like so:

1. Pick a particle jet, calculate its delta-phi as 'dFpar = fParJet - fParTrg'
2. For each detector jet, calculate its delta-phi as 'dFdet = fDetJet - fDetTrg'.  Then its 'qT' and 'dR' are given by 'qT = pTdetJet / pTparJet' and 'dR = sqrt(|dFdet - dFpar|^2 + |hDetJet - hParJet|^2)'.
3. If a detector jet satisfies 'dR < 0.3' and 'qT = (0.15, 1.85)', then consider it a match candidates.
4. The detector jet which has the 'qT' closest to one is declared the "best match", and the particle-jet + best-match pair is removed from the list of jets.

(Where we only consider recoil particle and detector jets which satisfy all of our QA cuts.)  When defining 'dFpar' and 'dFdet' as so, you sometimes artificially inflate 'dR' because 'fParTrg' and 'fDetTrg' don't necessarily have to match.  This produces an efficiency which looks like so:

Since we know what the actual trigger phi is, we know what the geometry of the event actually is.  And so, during the matching algorithm I tried defining the detector jet delta-phi as 'dFdet = fDetJet - fParTrg'.  This gives an efficiency of:

Which, by eye, seems to match the hadron-jet 60%-80% efficiency in shape and numerics.  Below is the old and new efficiencies on the same plot.

Note: the hadron-jet efficiency is taken from the analysis note, which can be found here:

https://drupal.star.bnl.gov/STAR/system/files/STAR_hJet_AN_Master.pdf

Note: the error bars on my efficiencies don't really mean anything.  They're just the calculated statistical uncertainties from the histograms used to calculate the efficiencies.  The uncertainty on the hadron-jet efficiency is its systematic uncertainty.

Update [07.05.2018]: just for reference, below is the response matrix after fixing the reconstruction efficiency.