Minutes, DSTF Meeting, July 5, 2002

Present: Toby, Seth, Tracy, Bill, Michael, Leon

The meeting was called to have a discussion about what to write out to the TDS/PDS, and in particular, Bill's criterion that (almost) all further analysis can be initiated from the recon output without requiring any geometry, calibration, etc.

• Bill amplified on his initial suggestion: by "recon output" he means the results of AcdRecon, CalRecon and TkrRecon. (We need a convenient name for this stage; Bill suggested "Pass A.") The combination of the information from recon to classify the event is presumed to happen at the later, analysis stage. (The possibility of feeding back the info from Acd and Cal to redo tracking, for example, may complicate this.)
  • A couple of examples of the kind of information that might be necessary to proceed with the analysis: minimum distances from every track found to the hit ACD tiles; list of expected energy deposits in the CAL along the extrapolated track directions.
  • The kinds of analysis envisioned: assignment of particle ID; studies of PSF; refinement of background cuts, etc.
     
• Connected to this, the DSTF recommends that both the local and global coordinates be written out for the MC hits, including the integrating hits.  
  • It has become obvious that any further analysis outside of Gaudi requires this information. Neither Heather nor Sasha were at the meeting; since then, Heather has agreed to this and we're waiting to hear from Sasha.
     
• There are (at least) two models of how to do the analysis, and we've never really written down what path we intend to take:
  • One model involves using the PDS simply as intermediate storage, which can then be read back into Gaudi to pick up where we left off.  The second involves using Root output outside of Gaudi (in Root) to do further analysis.
  • There are different levels of effort, functionality, convenience and interactivity associated with the two paths; the Root model allows for the Gaudi one, but not vice-versa. In particular, as pointed out by Toby, in the Gaudi model, the PDS can simply be an image of the TDS, and we can eventually use Gaudi to generate the PDS automatically, but then the methods of the TDS classes will not be available. We should continue to talk about this.
     
• Discussion of the ntuple was postponed to a future meeting.
  • Determining the actual contents of the ntuple will be easier once we have established the contents of the Root output, and people have started using the Root files to do analyses.
  • However, we should start developing a plan to get us there.