Socorro: Jeff, Lindsey, Tak, George, Kumar, Susan, Martin, Jim
CV: Mark, Akeem, Ville, Dave
News / Meetings / Visitors
Departure of Joe Depa
HPC meeting in Socorro June 22 - 27
Build, Release, Testing
Status: OSX 10.9 and 10.10 build exist.
Need to get the tests to run
Failing Tests: ListVis, 3C219D
PlotMS: Some timings have improved
Still need address issues with the counting
EVLA pipeline / CASA 4.3.1 commissioning: intent and display issues resolved, user testing resumed
Pipeline / CASA 4.4. testing: plotms averaging bug resolved by Pam, some timings improved after fix, timings better, further timing tests TBD
Pipeline HPC testing: open file issues in EVLA pipeline resolved, time sinks under investigation
Pipeline development: frequency region lists for imaging, error handling improvements
improved calibration failure error handling in several tasks
pipeline testing: plotms functionality and efficiency
investigated problems with bad channel flagging task hif_bpflagchans
Meeting with Justo, George and Jeff to discuss plotms performance w.r.t. determining memory requirements; started coding plotms to implement the interface Justo suggested
Started to investigate plotms table lock issue reported by Sandra (CAS-7428)
Investigated plotms averaging issue reported by Lindsey, shown in pipeline plots. Committed fix for this (no ticket).
Closed all open ticket for release
Memo for multi-thread gridding sent to Juergen
Note for MS-UVBin
Justo González Villalba
Vacation and "Spring Break" Holiday
CASA Users Survey Results and CASA Users Committee
SKA SDP PDR Draft
4 day vacation
CARTA customizable toolbar
4 days in office, one day NRAO holiday
(CAS-7268) importuvfits creates inconsistent MS: Concluded that this uvfits dataset was inconsistent according to Eric's document in that, in the SU table, it had NO_IF = 1 but had two rest frequencies per row. Resolved as "won't fix" after modifying code to emit more helpful error message as well as to remove bogus MS it has created after these sorts of unrecoverable exceptions are thrown.
(CAS-5329) IMMATH cannot parse strings with "_" on Mac OS: updated par.mask description to explicitly note that image names with underscores must be escaped as per the very long-standing LEL syntax.
(CAS-7135) clean task mishandles field parameter when multiple datasets refer to same field by different ids: Reviewed and tweaked Kumar's MS metadata-related code modifications.
(CAS-4231) imcontsub returns and image full of NaNs for fitorder=10 or higher: Concluded that in general, as for this case in particular, calibrated spectral baselines with large numbers of channels are not well modeled by arbitrarily high order polynomials. Suggested that more appropriate models be investigated by scientists and, when convergence occurs, provided to the CASA team for implementation.
(CAS-4283) remove ia.continuumsub in favor of ia.fitprofile(): Investigated necessary code modifications and refactored some ImageProfileFitter code in preparation for implementing these changes.
(CAS-7408) immoments crashes with underscores in mask names: Detailed discussions with Juergen and Ger about what characters need to be escaped in lattice expressions, and which do not, and why.
Investigated gaincal performance issues noted by James Robnett in pipeline testing. Appears to be tile cache thrashing when non-trivial channel selection is used.
Imager parallelization talk/discussion for HPC folk - part 1
Working on getting imaging weight calcs to work correctly for parallel runs. Basic plumbing done and working. Need to add details for multi-term. Discussed with SB about adding a nearest-neighbour gridder FTM to use in a first pass. This simply replaces the existing VisImagingWeight function, but will have access to correct data to image channel mapping (this should fix cas-7210 and give a version of multi-field for free). It is also a place to put weight accumulators for correct image sensitivity calcs.
Still working on cube data/image partition related modification to tclean and done some testing