Relating model parameters

[rbmetcalf]

It would be nice if some parameters can be fixed to others. For example, you might want the lens galaxies to be entered on the lens masses. Or one could imaging relating the mass to the light using some function. This would reduce the effective number of parameters, but might be hard to code.

[fabiobg83]

As I noted in issue #66, there is no such thing as lens mass or lens/source redshift at the moment. The einstein radius only fixes a relation between these quantities.

[rbmetcalf]

I realise this. The Einstein radius is proportional to the velocity dispersion squared and the velocity dispersion is related to the luminosity and size of the galaxy. If all the lens galaxies are assumed to be at the same distance one could find a relation between the luminosity and size of the galaxy and the Einstein radius of the NSIE with a free normalisation that is not dependant on knowing the redshift. One could then interpret the normalisation later with a particular cosmology.

I am more interested in fixing the centres of light to the centres of mass.

[ntessore]

I think something similar to what we had in evilense can be implemented with reasonable effort:

[priors]
lens.x = copy central.x

This needs derived parameters #27 to work though, so it's a bit more involved.

[ntessore]

Derived parameters are there now with #204, but it's still not easily possible to add a copy prior. The prior system would need access to the full list of parameters for this to work. Copied parameters would also necessarily have to come after the parameters they reference, which could be derived as well.