Strong Gravitational Lensing Posterior Sampling in Pixel Space Using Diffusion Models and Recurrent Inference Machines
Abstract Modeling galaxy–galaxy strong gravitational lenses to infer the brightness of the source galaxy and the mass distribution of the foreground galaxy is computationally challenging, particularly for high-resolution, high signal-to-noise ratio observations. In this regime, high-dimensional representations of both the source and the foreground mass distribution are necessary to model the data down to the noise level. This inference problem has been challenging for both traditional and machine learning-based methods because of its high dimensionality and its nonlinearity in the foreground mass distribution. We present a method to generate joint posterior samples of the source galaxy and foreground mass distribution as pixelated images conditioned on observations. The method combines diffusion-based generative modeling and recurrent inference machines. It can model realistic gravitational lensing simulations with background and foreground galaxies drawn from cosmological hydrodynamical simulations down to the noise level.
Authors
- Gabriel Missael Barco (ORCID: https://orcid.org/0009-0008-5839-5937)
- Laurence Perreault-Levasseur (ORCID: https://orcid.org/0000-0003-3544-3939)
- Guillaume Payeur (ORCID: https://orcid.org/0000-0002-0997-4827)
- Yashar Hezaveh (ORCID: https://orcid.org/0000-0002-8669-5733)
Institutions
- Perimeter Institute (CA)
- Cielo Institute (US)
- Mila - Quebec Artificial Intelligence Institute (CA)
- Flatiron Institute (US)
- McGill University (CA)
- Université de Montréal (CA)
Publication Details
- Journal
- The Astrophysical Journal
- Published
- 2026-11-16
- DOI
- https://doi.org/10.3847/1538-4357/aeb41d
- Primary Topic
- Galaxies: Formation, Evolution, Phenomena
- Type
- article
- Field-Weighted Citation Impact
- 0.00
Funders
- Canada Research Chairs
- Fonds de recherche du Québec
- Alliance de recherche numérique du Canada
- Natural Sciences and Engineering Research Council of Canada
- Fonds de recherche du Québec – Nature et technologies