Speaker
Description
Over the last decade JIVE, in collaboration with NRAO, has been working on adding VLBI functionality to CASA. As a result CASA is now a feasible alternative to AIPS. A significant number of EVN and VLBA users use CASA to reduce their VLBI data and the CASA-based rPICARD pipeline is used as one of the two calibration pipelines for the EHT.
However the technology behind CASA is already three decades old. As a result it has proven difficult to do data processing with CASA at the scale that is necessary for the next generation of radio interferometers such as the SKA, ALMA after the wide-band sensitivity upgrade and ngVLA. To address these difficult the CASA teams at NRAO, ESO and JIVE have started designing a new scalable framework to do data processing, and in collaboration with the SKA a new MSv4 data model/format to support this framework. VLBI is one of the main use cases for this new framework (as ngVLA will include VLBI-scale baselines).
In this presentation I will discuss the technologies behind this new framework: the MSv4 schema and associated Zarr-based data format, the xradio reference implementation based on xarray and the GraphVIPER package that use Dask to parallelise computation. And present the effort at JIVE to implement fringe-fitting in this new framework.
