GlueX Offline Meeting, December 14, 2011
GlueX Offline Software Meeting
Wednesday, December 14, 2011
1:30 pm EDT
JLab: CEBAF Center F326
- 1 Agenda
- 2 Communication Information
- 3 Minutes
- Review of minutes from the last meeting: all
- A CLARA-based analysis system: Gagik Gavalian
- CCDB update: Dmitry
- Reconstruction sub-group reports
- Memory leak in the reconstruction?: Paul M.
- Proton reconstructions studies: Sascha
- Action Item Review
- Review of recent repository activity: all
- ESNet: 8542553
- Meeting URL: http://evo.caltech.edu/evoNext/koala.jnlp?meeting=ete8eavBvvanaaIiaiIt
- Phone Bridge ID: 11 6471
Talks can be deposited in the directory
/group/halld/www/halldweb/html/talks/2011-4Q on the JLab CUE. This directory is accessible from the web at https://halldweb.jlab.org/talks/2011-4Q/ .
- CMU: Will Levine, Paul Mattione
- FSU: Nathan Sparks
- IU: [attempted to connect via EVO, connection made by chair too late!]
- JLab: Eugene Chudakov, Hovanes Egiyan, Gagik Gavalian, Nerses Gevorgyan, Mark Ito (chair), David Lawrence, Dmitry Romanov, Sascha Somov, Simon Taylor, Elliott Wolin, Beni Zihlmann
- Elliott announced that documents on the configuration database and the conditions database are soon to appear on the DocDB.
Review of minutes from the last meeting
We went over the minutes from the November 30th meeting.
- David and Will reported that we are ready to make the 1-2-3-4 scheme for BCAL read-out the default. One of them will do the deed.
- David has a working system for reading EVIO format in the JANA framework. Find it in the FCAL beam test area of the Subversion repository. It creates JANA objects out of FADC data. He has also been working on an event viewer and a histogramming plug-in to support the test. There is some documentation linked from the FCAL test wiki page.
- The too-many-photons thing is still an outstanding issue.
A CLARA-based analysis system
Gagik gave us an overview of his system for doing physics analysis from a HDF5-based DST data set using Hall B's CLARA system. He gave a presentation and a live demonstration using an installation of the system at Old Dominion University. Multiple processes can work in parallel to create data objects, an n-tuple in his demo, and ship said objects over the network to the user. Python-based tools can receive the data and create graphical output for the user.
Dmitry gave us an update.
- CCDB has been installed on the ifarm at JLab
- installation is fully functional, including a JANA plugin
- documentation is coming!
- the web application is held up by system software compatibility issues
Reconstruction sub-group reports
Simon described a proposal for monitoring performance of tracking on single particles. He has developed a set of scripts and ROOT macros that will allow us to see how the tracking is doing in the simplest situation, as time goes on and the code continues to develop. We will likely run the package twice a week, like the "b1pi" job already implemented as a cron job.
He calculates and displays particle reconstruction efficiency and momentum resolution in various views. Other plots can be added to the package; please send your suggestions to Simon.
While looking at the plots, we discussed some of the features in the tracking results that have been known for a long time, in particular why a "1%" confidence level cut has poor efficiency in some regions of phase-space. Eugene pointed out that we need to find out whether these are for physics reasons in the reconstruction (e. g., large single-scatters), problems in the pattern recognition (hits being missed), or in the track-fitting itself. We will convene a smaller group to discuss a strategy for answering these questions.
Paul gave us an update on particle identification using drift chamber dE/dx. He has added kaons to the study and has gone on to plot the various means and sigma's as a function of momentum so that one can judge the degree of π/K/p separation. Confidence levels for all quantities look extremely flat.
Eugene wondered if delta ray production was simulated as a separate electron track. He thought that it was important to do so for this type of study. Paul will check on this and look for differences if this was not done before.
Memory leak in the reconstruction?
Will agreed to summarize the problem, first report at the morning GlueX meeting, and send email to the group.
Proton reconstructions studies
Sascha showed results of studies where he looked at reconstruction efficiency and kinematic resolutions and biases for the recoil proton in both γp→pη and γp→pπ0 events. He sees rather large inefficiencies for protons when a χ2 cut is applied. These are likely due to non-Gaussian tails in the reconstructed quantities. The source of these tails will have to be studied.