Strong Gravitational Lensing Posterior Sampling in Pixel-Space Using Diffusion Models and Recurrent Inference Machines
Guillaume Payeur, Laurence Perreault-Levasseur, Gabriel Missael Barco, Yashar Hezaveh
astro-ph.IM, astro-ph.CO, cs.CV, stat.ML
Submitted: 2026-07-21
Comments: 31 pages, 34 figures, Extension of an article accepted at the ICML 2026 Workshop on AI for Physics
Code: https://github.com/GuillaumePayeur/DiRIM
License: http://creativecommons.org/licenses/by/4.0/
The gist: 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
Terminology
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 non-linearity 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.
Sources
- Posterior samples of source galaxies in strong gravitational lenses with score-based priors
- Appa: Bending Weather Dynamics with Latent Diffusion Models for Global Data Assimilation
- Blind Strong Gravitational Lensing Inversion: Joint Inference of Source and Lens Mass with Score-Based Models
- Conditional Image Generation with Score-Based Diffusion Models
- Inpainting Galaxy Counts onto N-Body Simulations over Multiple Cosmologies and Astrophysics
- The population of galaxy-galaxy strong lenses in forthcoming optical imaging surveys
- A point cloud approach to generative modeling for galaxy surveys at the field level
- Bayesian Imaging for Radio Interferometry with Score-Based Priors
- Adam: A Method for Stochastic Optimization
- The Principles of Diffusion Models
- Mind the Information Gap: Unveiling Detailed Morphologies of z 0.5-1.0 Galaxies with SLACS Strong Lenses and Data-Driven Analysis
- Array Programming with NumPy
- Sampling-Based Accuracy Testing of Posterior Estimators for General Inference
- Denoising Diffusion Probabilistic Models
- Elucidating the Design Space of Diffusion-Based Generative Models
- Diffusion Models for Medical Image Analysis: A Comprehensive Survey
- CosmicRIM : Reconstructing Early Universe by Combining Differentiable Simulations with Recurrent Inference Machines
- Analyzing interferometric observations of strong gravitational lenses with recurrent and convolutional neural networks
- The COSMOS-Web Lens Survey (COWLS) I: Discovery of >100 high redshift strong lenses in contiguous JWST imaging
- PyTorch: An Imperative Style, High-Performance Deep Learning Library
Related papers
- A signal dedispersion algorithm for imaging-based transient searches
- EXPO: a quantum leap in fast, wide-band X-ray polarimetry for astrophysics
- Understanding cosmic explosions
- High-frequency water megamasers as precision probes of black holes, galaxies and cosmology: A case for a next-generation sub-mm/mm interferometer
- ALMA2040: Mapping the Spatially Resolved Dust and Star Formation Physics Driving High-z Galaxy Assembly
- Direct measures of black hole masses through cosmic time