Rapthor vs. DDF-pipeline

[jurjen93]

We currently subtract sources outside our 2.5 × 2.5 degree field of view using DDF solutions and models. However, this approach has led to cases of over- and under-subtraction, and the underlying cause has been difficult to identify. We therefore would like to investigate whether performing prediction and subtraction using Rapthor solutions yields more consistent and reliable results.

[jurjen93]

Andre is running Rapthor on our test dataset. This will be completed at the end of the day.
After completion, we will perform a subtraction outside our 2.5x2.5 deg FoV using the model images + solutions and the model component file. Both results will be compared with the DDF subtraction quality.

[jurjen93]

Rapthor needed just about 1200 CPU hours, which is about twice faster than DDF.

[jurjen93]

Few images quality comparisons (left rapthor, right DDF):

In some of the cases the calibration looks better from Rapthor and other times in DDF.

[jurjen93]

See: https://github.com/LOFAR-VLBI/lofar-vlbi-pipeline/wiki/Feb2026--Rapthor-vs-DDF