Continuing the investigation:

https://drupal.star.bnl.gov/STAR/blog/dmawxc/update-06072019-run-9-embeddingmatched-vs-unmatched-track-comparison-including-those-mat

We still want to understand the tracks present in the Run 9 dijet embedding sample that are not matched to MC particles (referred to as "unmatched tracks" from hereon).  Here are several plots comparing the unmatched tracks which are matched to towers, and those which are not matched to towers.

There are some differences between the two populations of tracks, namely:

• The unmatched tracks with matches to towers have more nFit on average than those without matches to towers,
• The unmatched tracks with matches to towers have a tighter DCA distribution than those without matches to towers,
• The unmatched tracks with matches to towers have a tighter eta distribution than those without matches to towers,
• And the unmatched tracks with matches to towers have a harder pT distribution than those without matches to towers.

It's worth noting a few things: there may be reconstructed MC particles which failed the nCommon cut in the matching process (though this should be a very, very small portion), and the real data that is mixed in with the generated is zero-bias.  So I would assume we would see a hard-scatter event every now-and-again (though, again, this should be a very, very small portion).