search

lofar-astron / factor

Facet calibration for LOFAR
http://www.astron.nl/citt/facet-doc
GNU General Public License v2.0
19 stars 12 forks source link

Galaxy cleaned away #185

Closed amisk closed 7 years ago

amisk commented 7 years ago

Hello,

After I ran factor and only used a target facet with some calibrators around it, my galaxy gets basically cleaned away, leaving only some points. galaxy

The green contours are the same galaxy in C Band.

How can this happen?

Btw: Totally unrelated: Would it not be a good a idea to setup a message board for stuff like this? Would be great, since as I understand it, github issues are supposed to be for technical issues and bugs.

Just in case: I have experience in setting up a phpBB board ;)

mhardcastle commented 7 years ago

Surely this is a technical issue or a bug? I think existing users would prefer to keep using github...

Joshuaalbert commented 7 years ago

What clean settings did you use including sample density weighting, and masking ? Is the galaxy in the center of the target facet? How many surrounding facets were you using? Did you choose to transfer solution from nearest facet if facet failed? Cleaning diffuse faint emission is always tricky. I notice a fairly uniform weighted looking background. Perhaps a more natural weighting would help in this case. As well as multiscale. If clean broke up diffuse emission then self cal could definitely need things up. My advice would be too image this by hand since it's only a single facet.

amisk commented 7 years ago

This is what came out of factor, this is the full2-MFS-image.fits image.

In order to speed up things, I only used one facet for the target, and 6 calibrators around it as facetimages, which had only a small FOV, instead of entire facets. Alle selfcalibrations on those was completed by factor.

I ran this previously in November with ~20 facets, and it ran for one and a half month. But the result was the same.

Joshuaalbert commented 7 years ago

So it's not due to bad neighbouring solutions. You have an extended source disappearing, which makes me believe it's not properly masked/perhaps too uniform beam/is multiscale turned on by default? (I'm guessing yes - if not then you should for this facet). If it's not properly cleaned then during selfcal, the model components can shift around and break up your flux. First check mask, to ensure it's all masked. Then because it looks like the flux is there (just moved around) you should look at the model components to see where clean put them.

AnnalisaB commented 7 years ago

facet_wrong 001

Not sure it is related, but in this case I have flux not entirely subtracted in the initial_subtract step, and that's what happens. It is no bug of factor. Flux is added to the double radiogalaxy and then the mask doesn't pick it up and it disappears the next steps. Maybe if you look at the facet selfcal images you can see if the source is masked and what happens step by step.

darafferty commented 7 years ago

Closing due to lack of activity -- reopen if this is still an issue.