Remodel Galaxy WL object structure

[caioolivv]

Currently WL analysis is done by creating GalaxyWL objects containing GalaxyWLDist and GalaxyRedshift objects. Since shape and position information are grouped in the same object, separated from redshift data, binning processes are not straight forward. It also presents a difficulty if we were to hierarchically sample information for kernel density estimation and forward modelling. These problems would be avoided if we rewrote the WL likelihood the objects in the following structure

• NcGalaxyWL object which would contain the following objects:
  • NcGalaxySampleDistributionPosition with information on image position and redshift;
  • NcGalaxySampleDistributionShape with information on intrinsic shape, observed shape and the shear mapping (this might also be capable of holding information on measuring error);
  • NcGalaxySampleDistributionZProxy with information on proxy redshift;

Mathematically, the above objects would represent the likelihood

$$\mathcal{P} (\epsilon, r, z_p) = \int dz dr \mathcal{P}(\epsilon | z, r, \theta) \mathcal{P}(z_p | z, \theta ) \mathcal{P} (r | \theta) \mathcal{P} (z | \theta),$$

with the following map

• NcGalaxyWL -> $\mathcal{P} (\epsilon, r, z_p)$;
• NcGalaxySampleDistributionPosition -> $\mathcal{P} (r | \theta) \mathcal{P} (z | \theta)$;
• NcGalaxySampleDistributionShape -> $\mathcal{P}(\epsilon | z, r, \theta)$;
• NcGalaxySampleDistributionZProxy -> $\mathcal{P}(z_p | z, \theta )$;

Also, we might also use GSD instead of GalaxySampleDistribution.

[vitenti]

The object structure is fine. The name can be GalaxySDPosition that is instead of GalaxySampleDistribution or GSD we can use GalaxySD. What do you think?