Abstract :
[en] Any mass concentration in the Universe, luminous or dark, from vast galaxy clusters to stars
within galaxies, can be studied through its gravitational deflection of light rays from background
sources. This phenomenon, in its most impressive regime, is known as Strong Gravitational
Lensing (SGL). It has several cutting-edge applications, for example: measuring the Hubble
constant and shedding more light into the apparent tension between early and late Universe,
detecting the presence of massive subhalos within distant galaxies that can constrain different
dark matter models, and studying a galaxy’s mass partition between baryons and dark matter
with direct implications on galaxy evolution.
Extracting information from SGL data requires the careful analysis of images of gravitational
lenses, a process referred to as lens modeling, in order to generate an image of the lens based
on models of mass and light distributions of the different physical objects in play (e.g., galaxies,
quasars). In this paper we call a lens model the full set of model components, including
all mass and light models as well as the point spread function (PSF) model. Over the past
twenty years, several lens modeling codes have been developed and used in published works.
Unfortunately, there is currently no efficient and systematic way to access these published
results and use them directly for new studies, which slows down new research and causes a
waste of research time. The reason is simple: these modeling codes being based on different
methods and conventions, bridging the gap between them is a challenging task.
Here we introduce COOLEST—the COde-independent Organized LEnsing STandard—to the
lensing community, which allows researchers to, independently of the original modeling code:
• store lens models in a JSON format that is lightweight and easy to read and manipulate;
• group together all necessary data, model and inference files (such as images and arrays
in standard FITS and pickle formats);
• compute a set of key lensing quantities, such as the effective Einstein radius and mass
density slope;
• compare models by generating standardized figures using a Python API.
Any lens modeling code can adhere to this standard via a small interface that converts code-
dependent quantities to the COOLEST conventions. The documentation and all Python
routines incorporated in the API serve to keep development time to a minimum for code
developers. Figure Figure 1 below gives a concrete example of panels generated with the
plotting API, alongside quantities computed with the analysis API.
