Home » Groups » STAR Protected

Checking Recent Gamma Maker Changes

Updated on Sat, 2008-07-05 08:48 by testadmin. Originally created by jwebb on 2008-06-29 16:33.
Abstract: Recently Pibero checked some changes into the offline/StGammaMaker code. I noticed some discrepancies in event yields between old and new versions of the code. While there are discrepancies between the two codes, they are smaller that what I was seeing.  Nevertheless, it is important to start tracking code changes.
 
Method:
 
1. Check out the "current" and "old" versions of the code, defined as
  • old: cvs co -D 06/01/2008 offline/StGammaMaker
  • cur: cvs co -D 06/27/2008 offline/StGammaMaker
2. Compile the code in each directory
3. Generate gamma trees for run 7136022
4. Compare event yields for trigger ID = 137641
 

Number of triggers expected: 10449

06/01/2008 code

  1. 6102 events in gamma trees
  2. 5079 events w/ reconstructed TPC vertex
  3. 5079 events w/ gamma candidates (3x3 cluster w/ ET > 5 GeV)

06/27/2008 code

  1. 10410 events in gamma trees
  2. 9074 events w/ reconstructed TPC vertex
  3. 5072 events w/ gamma candidates (3x3 cluster w/ ET > 5 GeV)
The current version of the code differs by 7 events in 5k (1.4%) from the earlier version of the code.
Changes to the code which explain some of the above discrepancies:
  1. Old version of gamma tree maker requires a valid gamma candidate before writing the event.  New version does not.
  2. Old version of raw maker has a different (possibly incorrect?) vertexing cut.
  3. Old version of the code does no trigger filtering during tree creation.  New verion does.  Thus, old event total of 6102 (before cuts) represents the total number of events found in the MuDst which satisfies a 3x3 ET > 5.0 GeV cut, from which 5079 have a valid TPC vertex.
New and old version of the code are consistent within 1.4%.  Not clear why we lose the 1.4%.  Possibly due to the accessor methods used to get the TPC vertex.
 

07/03/2008

Data sample:
1. Fill 7863, kick run 7136046 (ppProductionLong w/ no TPC and it crashes the new code).

2. Apply the "classic" conversion method cuts

- isolation cut, 90%  R=0.3
- hadronic veto, R=0.3
- charged particle veto R=0.3
- analyzer cut R=0.3

Old version:


N triggers = 30577
N triggers w/ flag = 25626
N triggers w/ cand = 25626

N eemc candidates = 25626
N isolated candidates = 11873
N nonhadronic candidates = 2400
N neutral candidates = 135
N analyzer = 122

Current version:


N triggers = 53134
N triggers w/ flag = 46618
N triggers w/ cand = 25716

N eemc candidates = 25711
N isolated candidates = 11764
N nonhadronic candidates = 2348
N neutral candidates = 132
N analyzer = 121

There are subtle differences here, of a few % by the time we  apply the cut on EEMC candidates, and isolation cuts.

 

 

Groups: