Probabilistic cataloging reloaded

In cosmology, we have the unfortunate luxury of having a standard model, i.e., Lambda Cold Dark Matter (LCDM). Although LCDM really deserves its unique place since it precisely agrees with many independent observational probes, it is often helpful to remind ourselves that most of the small-scale structure it predicts is not visible (through the electromagnetic interaction). Given the current lack of observational probes of these small matter clumps, everyone entertains his or her favorite small-scale resolution to LCDM. This is all fine, but at the end of the day, if structure formation is really bottom-up, the Universe can be full of invisible structure (anywhere from Earth-sized up to subgalactic scales) and the only way to measure them may turn out to be via their gravitational lensing of light rays from background galaxies.

In a new paper we outfit PCAT with a lens model to perform transdimensional inference of substructure in such strong lens systems. This framework revisits the fundamentals of strong lens modeling and allows across-model covariances to be accounted for when inferring the properties of substructure.