Delay Calibration: generate automated settings for in-field calibrator

[jurjen93]

The default settings are not always right - we need to work out better input settings for facetselfcal.py given all the development and updates that have gone on.

[jurjen93]

I have made a script that generates config files:
https://git.astron.nl/RD/VLBI-cwl/-/blob/auto_dical/scripts/make_config_dical.py

I ran 33 self-calibration with the default settings from the LOFAR VLBI pipeline. I compared this with the phasediff scores (measure for S/N) and Roland's manually adjusted settings to compare these. Based on this comparison, I wrote the above mentioned script to set different solution intervals and smoothness constraints for sources based on their S/N. I am currently running the delay-calibration for the same sources with the newly automated settings such that I can compare this with the default settings.

[jurjen93]

Hi @r-timmerman , are the examples that you provided for the hackathon in Leiden special cases (https://public.spider.surfsara.nl/project/lofarvwf/jdejong/DIcal_examples/)? These infield calibrators are very challenging to calibrate and I am struggling to find some automatic settings that work for all cases. If they were specifically chosen for being challenging, it would help me better understand where to focus my efforts.

[r-timmerman]

No that's just literally kind of everything I had. I appreciate that you think I could be so productive to have a cherry-picked sample of 45 DI calibrators though :D

The majority of them are relatively low declination, so that's likely at least partially why you're running into some challenges. Also, the "rej" ones are indeed rejected by me, but I still wanted to include them for completeness. An issue that you may also run into with the ILTJ ones is that they were not run with a smoothness constraint in LINC, so the Dutch stations will be suboptimal.

Edit: oh and a good fraction are Cycle 20, which was an absolute ionospheric disaster.

You have the parameters I used for quite a few ones, right?

[jurjen93]

Ah, thanks, good to know!
Yes, I have your parameters. I am trying to search for an automatic set of parameters based only on the phasediff score, but perhaps its not enough, although I get often better results than the currently default settings. But I am still seeing some diverging behaviors. If you find anymore examples, I am happy to add them to the test selection :)

[r-timmerman]

Ah yes I don't think that's enough. I'm not very surprised that you can improve upon the defaults quite easily, but I tweaked my optimal parameters also based on how the ionosphere behaves, the source structure and the environment of the DI calibrator. I think the phasediff score is a very useful ingredient, but I don't think you can ignore the rest.

[jurjen93]

Yeah, the phasediff score is a measure for S/N but it also captures a bit of calibratability (due to ionospheric effects) if the phasediff solutions are already noisy, which results in more conservative parameters. I hoped that as what I saw for DD calibration of 25 different ELAIS observations that we can already find good starting parameters purely based on phasediff scores. But seems that for DI calibration, we need indeed something more advanced for challenging cases with for instance wild ionospheres

[r-timmerman]

Ah yes I can imagine that with DD calibration as a reference, the DI solve is less straight forward.

[lmorabit]

spider
| ILTJ123441.23+314159.4_141MHz_uv.dp3-concat | default |
| ILTJ123441.23+314159.4_143MHz_uv.dp3-concat | default |
| ILTJ125911.17+351954.5_143MHz_uv.dp3-concat | redone |
| ILTJ131028.61+322045.7_143MHz_uv.dp3-concat | redone |
| ILTJ132716.06+343235.3_143MHz_uv.dp3-concat.flagged | redone |
| ILTJ135630.44+294119.3_141MHz_uv.dp3-concat | default |
| ILTJ192340.71+661146.1_143MHz_uv.dp3-concat | redone |

cosma
| ILTJ125723.87+364418.9_143MHz_uv.dp3-concat | default |
| ILTJ131417.86+364914.4_143MHz_uv.dp3-concat | default |
| ILTJ133845.16+365344.5_141MHz_uv.dp3-concat | redone |
| ILTJ135759.65+364949.9_142MHz_uv.dp3-concat | default |
| ILTJ192340.71+661146.1_143MHz_uv.dp3-concat | default |
| L329188_142MHz_uv.dp3-concat | .... |