2025 workshop program
Gravitational Lensing and the Need for AI-Accelerated Lens Modeling
The work
Abstract
Gravitational lensing, nature’s own cosmic telescope, is a powerful tool that enables us to observe magnified images of distant galaxies, probe dark matter distributions from galactic to cluster scales, test galaxy evolution models, and constrain cosmological parameters. The ASTRO 3D Galaxy Evolution with Lenses (AGEL) survey has leveraged convolutional neural networks (CNNs) to identify O(10^3) strong gravitational lens candidates from DECaLS and Dark Energy Survey images, with over 100 lenses confirmed through optical to near-infrared spectroscopy. In this talk, I will present how we are using space-based imaging of galaxy-scale lenses to investigate galaxy evolutionary tracks, and how multi-source-plane lenses can provide new constraints on cosmology. A key bottleneck in this research is the modeling of lenses—a process that remains time and resource intensive. With the next-generation surveys such as Euclid and LSST expected to discover on the order of 10^5 lenses, there is a pressing need for fast, scalable, and standardized lens modeling pipelines powered by modern machine learning techniques. I will briefly discuss current efforts and challenges in this area, and highlight opportunities for integrating AI/ML approaches to advance gravitational lens science.