2016.10.20 -- Software Meeting

Agenda

1. Sparsifying the LCIO output: What can be dropped?          — Sebouh, all
2. Update of the ECAL driver.                                                  — Sebouh EcalRawConverterDriver.pdf
3. Skims for 2016 data                                                             — Sebouh ProposedSkims.pdf
4. What, if anything, do need to add to the DST?                    — All 
5. Updates on detector geometry: V5 detector at 1.5mm        — Bradley
6. Status of MC production                                                      — Bradley 

Notes:

LCIO Output discussion: 

• Dropping things from LCIO, what can be dropped?
  • Matt: Raw Hits from SVT - the 6 ADC values
    • Omar: some people use those. 
  • Omar: what we can change: the recon particle driver could create particles from GBL only. This should be an easy change.
  • Omar: We need the track data persisted, for DST.
• We could have a standard "slimmer file", and then have everything we have now for a fraction, say 1 out of 100, files.
• How much disk space would we really save?
  • Norman: We should be able to ask LCIO for a budget summary of how much space each collection takes.
• It only makes sense to do this if it really saves significant amount of space.
• Tim: looking forward, we do need everything that is required for debugging. That can include pulse fitting, overlapping hits etc. I am hesitant to drop things.
  
  

Update of the ECAL driver:

• Sebouh wants to split the EcalRawConverterDriver into 3 parts.
  
  • Was Nathan's idea to break it up, worked with Holly and Sebouh
    • EcalRawConverterDriver - Pulse Fitting and pedestal subtraction.
    • EcalGainDriver  - Corrects the ADC count to energy, using gains from DB
    • EcalTimeCorrections - Time walk corrections.
  • Output will be EcalCalHits - intermediate results will not be stored.
  • Sebouh will put this change on a branch first. 
  • The result will need to be tested on a number of steering files.
    • There are a lot of steering files: Analysis, monitoring, readout, recon, + user files.
• Jeremy - Make a JIRA item for this. Make a branch named after the JIRA item. 
  • I like this direction.
  • Send out a detailed email about what needs to be done.
  • There are 3 versions of EcalRawConverterDriver. => Sebouh - there should be one. 
• Kyle: The EcalOnlineRawConverter should not be touched. It is needed for emulating the HW during runtime.
• Jeremy - Doesn't this driver handle the different data modes?
  • Should that be split up as well?  A parent driver and then sub-drivers that handle each of the different modes of the ADC.
• Jeremy - Will make a release of the current state before this is merged.
• Holly - The time walk correction was different for different data. 
• Jeremy - The current structure is fairly un-maintainable. Breaking this up will make it easier to alter one of the steps.
  
  

Proposal for different Skims for next cooking run:

• Argument: Loosen the cuts to make the various skims more universally useful, without causing too much of an increase in the file size.
• No objections to this proposal. Sebouh will email the software and analysis groups for further comment. If no objections, then he can go ahead and create new skims.
  • Nathan: The pair0 trigger is different between the 2015 and 2016 data. The 2015 data would require additional filtering.

New detector from V5 detector:

• Bradley created a new detector model from V5 for 1.5 mm opening. It seems to work when running through SLIC. There was an error then running readout.
  • This is needed for MC. The MC detector does not know about the database.
  • Suggestion: Print out the position of the sensors when running readout - to make sure that this is correct.
    
    

Q: Release of HPS-Jave early next week?

• Are there issue with the current trunk?  - Not known. Jeremy will run some tests before the release.
• Omar: Will also make a dst-maker release, before or in-time, with the HPS-Java release.
  • Omar will present in 2 weeks an overview of the planned changes.