Incorrect association?

[ddobie]

https://dev.pipeline.vast-survey.org/sources/44717001/

This source has two measurements from a completely unrelated source included. The declination is roughly consistent, but the RA is off by >160 deg.

I have no idea what would cause this, especially since there's two other measurements of the same coordinates associated with a different (correct) source: https://dev.pipeline.vast-survey.org/sources/44332471/

[ajstewart]

This looks like the other way around, the association is good between those points, but somehow the forced fits has been done in the wrong location and attached to that source.

My guess is a bug in the ideal source coverage calculation, or a bug in the forced fits process that has got ID's crossed.

Edit: Suspicious that it's around RA 0, RA wrapping handling maybe?

[ddobie]

It seems to only occur for 3 sources out of the ~2M in the run: 43885418, 43891486,44717001.

All three are close to RA 0 for their detections, but all have forced fits measurements that are nowhere near where the detections are (74deg, 335deg and 164deg).

[ddobie]

Quick update on this. I've put the latest extragalactic data through (extragalactic-postprocessed-20241124) and this issue persists.

Bad indices are: 61916648, 61918524, 61923192, 61952747.

All but 61918524 have a corresponding source at almost exactly the same declination (to within 2").

[ddobie]

I cannot reproduce this on a small scale.

I've downloaded all images associated to the above sources and run those through the pipeline and the errant forced fits are not present.