More Sophisticated models for foreground subtraction

[rbmetcalf]

Thus far we have standard profiles for the lens galaxy subtraction. We might think about a model that just tries to interpolate the surface brightness across the region where the arc is. We are not really interested in whether the lens galaxy model fits the centre of the lens galaxy well, just in what it is doing where it intersects with the arc. This could both be more flexible and not weigh inflate the chi^2 by fitting pixels that we are not really interested in.

[ntessore]

I think that could be achieved by masking the central part of the observation. We could try this for the horseshoe image @fabiobg83 just to see what happens.

[rbmetcalf]

Yes, but it would still have perfect elliptical symmetry.

[ntessore]

In the case of single strong lenses with clear Einstein ring, I suppose we could have a special annulus mask/source combination that takes care of the background sky and the central halo near the arc. This would free up > 10 parameters, which we can use to model probably all reasonable asymmetric cases.