2025 2 polarization

Busy week Feb 2025

Analysis by Shane O'Sullivan, Vijay Mahatma (with input from Reinout, George H, Etienne)

Goals

• Understand the effect of our calibration (fulljones solves on the assumption of no linear polarization over the field and no linear polarization on the in-field delay calibrator) on circular polarisation
• Understand how well the strategy supresses leakage by comparing results with data that has no fulljones calibration
• Understand which tools we should use to be integrated with the LOFAR-VLBI pipeline going forward

Results

3C345

• Leakage strongly supressed, and very little residual circular polarization, compared to calibration without fulljones solve (bottom). Leakage in QU supressed from p=1.3% to p=0.3%, and the same values for Stokes V
• Delay calibrator has very little or no correlated flux in Stokes V

However, there are still some oscillations in the Stokes V flux as a function of frequency, the interpretation of which still needs looking into for robust information on systematic errors.

"BLLAC"

Similarly large improvement in leakage with full-jones (leakage <0.5%), no mirrored RM peaks, both jet “knots” with similar RM and consistent with LoTSS 20”, problems with V remain (but at low level, <0.5%)

Forward outlook

• We need to develop a catalogue analogous to LBCS but which also gives information on their polarized intensities. This will be useful to know if calibration referencing can be done for polarization on a science target which itself is polarized. We should discuss how to best approach this seemingly processing-heavy task. Ideal delay cal: unpolarized and unresolved in LoTSS (common) & unresolved at 0.3" (rare?)
• Process more sources? to find new issues or not.
• (investigate) Use of RMTools in the LOFAR-VLBI pipeline https://github.com/CIRADA-Tools/RM-Tools. Already available in FLOCs.
• We would first need to edit the configuration files in the pipeline that tells facetselfcal what to do -- i.e. adding "fulljones" to the soltype for the in-field calibrator, and not doing any phase-up of Dutch stations.
• Then we need to add an extra WSClean command that happens after the residual self-cal on the target, to spit out I,Q,U,V images for >200 channels from the calibrated data, concatenate them into Q and U cubes, and then run them through RMTools to produced polarized intensity images
• Beam depolarization: we should be seeing a higher degree of polarization of unresolved hotspots at 0.3". To what extent is this the case in reality?