Minutes from 13 May 2009

2009/05/13 CSEP Meeting

Attendees: John Y., Masha L

1. Status of CSEP V9.4.0 on csep-op server:
   • Masha discovered a problem introduced in V9.4.0: plots for alarm-based evaluation tests are not published to the web server:
     • FIXED: Trac ticket #150: Alarm-based tests are missing test identifier within filename
2. Tasks for intermediate CSEP V9.4.1 release:
   • FIXED: Trac ticket #147: Assumption in R-test simulation violated
   • FIXED: Trac ticket #146: Avoid overwritting of modules by testing region distribution during installation
   • Fabian reported a problem:
     1. FIXED Trac ticket #151: GMT fatal errors don't cause acceptance test failure
   • Matt reported a problem with New Zealand installation on their new development FC10 server:
     1. FIXED: GeoNetNZDataSource missing fileFormat() method
     2. FIXED: Masha discovered python import recursion problem that would only show up for New Zeland and Japanese distributions
   • Masha tagged release candidate with fixes to described above problems:
     1. SVN: csep/tags/Version_pre9_4_1_05_13
   • John installed CSEP generic, California, Global and Western Pacific packages for release candidate on csep-cert
   • Masha ran acceptance tests successfully
   • Will install on operational server next week if no problems show up on certification server for a week
3. Matt reported 5 failed acceptance tests for installed release candidate V9.4.1:
   • John is looking at matplotlib installation: does not support svg format of the image file
   • Masha traced down a warning message reported by scipy that breaks Matlab parsing of returned matrix of random numbers by cseprandom.py:
     1. python /opt/local/csep-9.4.1/src/generic/cseprandom.py -r true -f ./uncertainty.1-randomSeed.txt -n 40 /usr/lib64/python2.5/site-packages/scipy/sparse/linalg/dsolve/linsolve.py:20: DeprecationWarning: scipy.sparse.linalg.dsolve.umfpack will be removed, install scikits.umfpack instead
        ' install scikits.umfpack instead', DeprecationWarning )
     2. John is looking for a fix to get rid of the message
4. Tasks for CSEP V9.7 release:
   • Peter Shebalin's EAST three-month alarm-based model for California
     1. Peter sent reference data for M4 version of the model
     2. Masha sent email to Peter asking if we can abandon M5 version or should we test both versions of the model
   • Alvaro Gonzalez's one-day alarm-based model for California
     1. Might be part of CSEP V9.10.0 if there will be not enough time before code freeze for V9.7.0
   • Yan's Global long-term and short-term models using CMT:
     1. Yan's forecast is using 0.5 degree grid vs. CSEP 1.0 degree grid:
        • Yan provided conversion routine for short-term forecast file
        • Will ask Yan to provide routine for long-term forecast file since it has different file format
   • Trac ticket #148: Introduce acceptance tests for the viewer of testing center results
     1. John and Masha will work together on setting up a simple test case using ClientForm.py, HTML parser (BeautifulSoup, may be) and Python's unittest testing framework
     2. Masha started testing ClientForm.py module to submit requests to the web server
5. John set up new server csep-beta with cross-mounted new storage to replace current operational server:
   • Masha set up nightly crontab Dispatcher jobs to mimic operational server for testing
   • Faster than current operational server
   • There were no more problems of stuck processes on the server
   • John will start rsync of operational data to the server, and switch the hardware once it completes
6. Jeremy distributed mini_CSEP
   • Danijel suggested to re-write Matlab evaluation codes in Python
     1. Phil, Danijel, Jeremy and Masha had a meeting on Monday to discuss re-writing of Matlab codes
     2. Phil and Danijel will decide
7. Jeremy asked for S and M tests for 2.5 years of RELM experiment:
   • Masha processed RELM Mainshock and Mainshock/Aftershock groups, posted results on motion.usc.edu /home/csep/forJeremy/RELM-2008-07-01-SMTests