1) The Background Sample
...
The pixel detector views 1750 bunch crossings.  About 1% of those bunch crossings (18)  will be filled per trigger event.  If we have a triggered event sample of 20k events, we would require 360k background events in order that every bunch crossing would have an independent background event.  However, we do not need nearly this many events, since they will be randomized in time relative to the trigger bunch crossing.  So a sample of 40k events for the background should be adequate for a 20k event sample.

Geometry: y2014a
Generator: hijing
CMS energy: 200 GeV
Collision system: AuAu
Vertex: 0,0,0
Width: 0.1, 0.1, 30.0
N events: 40k (25 per job at 1600 jobs).

2) The Physics Sample
...
Geometry: y2014a
Generator: hijing
CMS energy: 200 GeV
Collision system: AuAu
Vertex: 0,0,0
Width: 0.1, 0.1, 30.0
N events: 25k (25 per job at 1000 jobs).

3) The Pileup Sample
...
Do an "ls | sort -R > minbias.list" at the start of each job.  Use minbias.list to read in the list of background event files.  i.e. we read the files in in randomized order.