CASA is slow, CASA is unstable, and CASA is unpredictable. [Last update: Sep 30, 2018]

Acknowledgements: Hiroshi Nagai, Kouichiro Nakanishi, Patricio Sanhueza, Yanett Contreras.

Please let me know if you find something wrong or you have better solutions to these issues, thanks!

References:

1. The official CASA homepage
2. CASA guides and tutorials
3. Analysis utilities
4. Yclean

General Issues

• CASA imregrid may introduce a 1-arcsec offset when converting between J2000 and Galactic frames. Use montage_wrapper in astropy!
• Pay attention to the primary beam shape, especially when mapping a large area outside of the FWHM of the primary beam (e.g., pblimit=0.05 in tclean). This issue may have been fixed in v5.4.1+.
• For low frequency (e.g., VLA C band) and large mosaics (e.g., more than 10 pointings), enabling w-projection/facets in clean or tclean is necessary, or else phase errors show up for pointings far away from the imaging center and positions of sources are drifted.
• Combination of ALMA 12m array and ACA data: watch out for the flux consistency?