The CASA team has begun exploring options for a new generation of software to meet the growing demands of current and future radio telescopes. This
CASA Next Generation Infrastructure (CNGI) prototype package is a demonstration of the current state of our research efforts. Its primary purpose is to showcase new
data structures for MeasurementSet and Image contents built entirely in Python atop the popular technology stack of numpy, dask, and xarray.
The
CNGI prototype documentation contains
Visibility and
Image overview sections that describe a selection of core mathematics, manipulation, middleware and analysis functions to demonstrate the simplicity and scalability of the technology choices. Notional examples of
Calibration,
Flagging and
Imaging are provided to illustrate future design and implementation direction. A detailed explanation of technology choices, including the xarray and dask frameworks, the zarr storage format, and the functional design architecture can be found in the
Development section.
Finally, the most compute intensive areas of CASA imaging are implemented and
benchmarked to demonstrate the parallel scalability and raw performance now possible from a pure-Python software stack.
We invite anyone interested to explore the new cngi_prototype package. We welcome your feedback at
casa-feedback_at_nrao.edu.