Vertex reconstruction efficiency is way down in AuAu15 GeV compared to last year's pp500 GeV. The plot shows the z-vertex distribution for ~1M events of pp500 and ~350k events of AuAu15. (The latter has been scaled up by a factor of 3, so that both plots are normalized to the same number of events, to within a few percent.) The spike at -999 shows entries for events that could not recreate a good vertex. For the pp data, ~1/3 of the events couldn't recreate a vertex, and the other two thirds show up in the distribution between -200cm and 200cm. For the AuAu data, only ~2% recreate a vertex! All told I only get about 1,000 (of the original 350k) vertices that match between the East-only and West-only analyses.

It looks like the vertex reconstruction macro throws out vertices that don't have at least one track with a good ToF hit. I guess the idea is to use ToF to independently vet tracks (and hence vertices) in the TPC. Also, about 10% of TPC hits come from pileup, which would be reduced using a ToF cut. In the above plot, you see a histogram of events with a given number of ToF hits. For pp500, most of the events have 1 or more hits. For AuAu, only about ~1-2% do. The AuAu data is from the first two run days (46 and 47), so perhaps they are still figuring out ToF?

I'm going to proceed by lifting the ToF requirement to see if we can get decent results.