Reconstructing the Subhalo Mass Function from Strong Gravitational Lensing

Constraining the distribution of small-scale structure in our universe will allow us to probe alternatives to the cold dark matter (CDM) paradigm. Strong gravitational lensing offers a unique window into small dark matter halos (10⁹ M_sun) because these halos impart a gravitational lensing signal even if they do not host luminous galaxies. However, the tens of thousands of free parameters in gravitational lensing by a substructure population makes directly evaluating the likelihood intractable. Simulation-based inference techniques can return posterior estimates without access to the likelihood, but they require a representative set of data simulations. To that end, we introduce the package paltas which builds on the lensing package lenstronomy to create large datasets of strong lensing images with realistic substructure, observational effects, and galaxy light pulled directly from the Hubble Space Telescope's COSMOS field. We use this simulation pipeline to train a neural posterior estimator of the subhalo mass function (SHMF) parameters and place constraints on populations of lenses generated using a disjoint set of galaxy sources. We find that by combining our networks with a hierarchical inference framework, we can both reliably infer the SHMF across a variety of configurations and scale efficiently to large lens populations. To our knowledge, our work is the first to constrain the SHMF on simulations with fully realistic sources and substructure, demonstrating the potential of strong-lens imaging to probe dark matter at small scales.