cuc provide
play

CUC provide Perceptions amongst members and the community at large - PowerPoint PPT Presentation

CASA Users Committee Meeting Repor t (2018. 10. 25-26, NRAO, Socorro) (https://safe.nrao.edu/wiki/bin/view/Software/CASA/CASAUsersCommittee2018) 2018. 12. 26, A-Ran Lyo Japan Radio Users Meeting Aim: To get unified input on CASA from the users


  1. CASA Users Committee Meeting Repor t (2018. 10. 25-26, NRAO, Socorro) (https://safe.nrao.edu/wiki/bin/view/Software/CASA/CASAUsersCommittee2018) 2018. 12. 26, A-Ran Lyo Japan Radio Users Meeting

  2. Aim: To get unified input on CASA from the users perspective CUC provide § Perceptions amongst members and the community at large of CASA capabilities, usability, reliability, and performance § Development priorities for the coming year § Advice on strategies for supporting other telescopes

  3. Accomplishments (“CASA Overview and Statistics – Raba”) § Refinements to tclean auto-boxing § New n-sigma deconvolution stopping criteria § New calibration library to simpify encapsulation of applycal fields § Calibration tables integrated in to PlotMS § New statwt implementation § Ephemeris imaging of multiple MeasurementSets in single execution § Parallel processing in Imaging Pipeline § New telemetry recording of user community CASA activity § new task sdsidebandsplit for ALMA Band 9&10 data § new task tsdimaging which will replace sdimaging in future § higher order interpolation available for sdcal § Support of Nobeyama 45m and ASTE data - successfully run pipeline from calibration through imaging

  4. Future Work (“CASA Overview and Statistics – Raba”) § complete migration to tsdimaging § single-dish specific plots:pointing pattern,etc. § improvement of performance § fully support data reduction of Nobeyama 45m and ASTE data

  5. Stakeholder Priorities – 5.5 and Beyond (“CASA Overview and Statistics – Raba”) § Full path for polarization calibration of linear feeds § Robust outlier rejection in calibration tasks § More flexible weighting schemes in tclean § AWProjection commissioning in tclean § Statwt improvements (time binning, signal rejection) § Pointing table corrections through a caltable format § Higher order polynomial fitting in fluxscale § More flexible calibrator models (incl. polarization, time variability, spectral shape) § New implementation of uvcontsub (incl. phase shift option) § Arbitrary ephemeris support for single dish

  6. Projects (“CASA Overview and Statistics – Raba”) • CASA6 (“ CASA6 – Schiebel”) – Migrate to Python 3 – Modularize Components – tools / tasks / GUIs – Package for standard python environments • CARTA – Replacement for CASA Viewer – Scalable to very large data sizes – Capable of local and remote operation • MeasurementSet V3 : Develop a new data model for the CASA software package that will enable it to meet both SKA Telescope and the NRAO’s long term goals for scale-up and scale-out requirements.

  7. “CASA Performance Overview” - Rada (see also, “Performance-Static Benchmarks, Montesino/Castro”, “Performance-Dynamic Collection, Suoranto”) Recommendations from Last Year • Work out concise and transparent performance benchmarking metrics • Implementation of a runtime predictor, at least for selected tasks :Runtime prediction has not been possible to date, Dynamic Benchmarks may be the answer • Continue to make progress in development of plone documentation for parallel-mode CASA. Goals • Make it faster – Reduce runtime of CASA tasks • Benchmark current performance – Identify major drivers of run time is “tclean” • Measure real world performance – collect timing data of CASA execution on users machine. Performance Factor • Hardware: Cores, Nodes, Storage • Data: Access (Data should be partitioned efficiently), Driver (Data on disk must be connected to processor/core doing the execution) , Configuration (Time, channel axes and SPW selection, continuum vs. cube) • Software: Algorithms (Independent, non-sequential pieces may be processed in parallel) Memory (Shared memory allows one copy of data to be accessed by multiple processes) Layers (Software hierarchy may do parallelization in multiple layers, ex, Pipeline/CASA/casacore)

  8. “CASA Support for Polarization – Moellenbrock” Polarization Calibration Development (CASA 5.5) 1. Move critical almapolhelpers-based functions into tasks – QU estimation from gain amp ratios (replaces qufromgain) : gaintype=‘G+QU’ (linear basis only) – Automate cross-hand phase 180 degree ambiguity resolution (replaces xyamb) : Relies on Q,U determined above, and supplied as prior 2. Modularization of cross-hand phase and QU solving within polcal – poltype=‘Xfparang’ – Modularize and improve basis-specific cross-hand Q,U solvers for basis- agnostic use prior to instrumental polarization solve – Add explicit position angle solution: poltype=‘PA’

  9. “CASA Support for VLBI – Moellenbrock” VLBI Support in CASA as of 5.3 New task: accor for amplitude normalization (cf AIPS ACCOR) New task: fringefit for rudimentary fringe-fitting (cf AIPS FRING) Improvements to task: importfitsidi Improvements to task: gencal (general antenna-based Tsys) Metadata scripts: (https: //github. com/jive-vlbi/casa-vlbi) New in CASA 5.5 • Finished – Search window support in fringefit – Switch to turn off globalization fit in fringefit (FFT only) • Planned – Improved convergence criteria in fringefit – Alternative weighting in fringefit – Re-referencing support in fringefit

Recommend


More recommend