| Name: Khushi Lalit |
| Affiliation: Inter University Centre for Astronomy and Astrophysics |
| Conference ID: ASI2026_122 |
| Title: Fast Bayesian Modelling of Group-Scale Strong Lenses with GIGA-Lens |
| Abstract Type: Poster |
| Abstract Category: Galaxies and Cosmology |
| Author(s) and Co-Author(s) with Affiliation: Khushi Lalit(Inter-University Centre for Astronomy and Astrophysics, Pune - 411007, India), Anupreeta More(Inter-University Centre for Astronomy and Astrophysics, Pune - 411007, India), Surhud More(Inter-University Centre for Astronomy and Astrophysics, Pune - 411007, India) |
| Abstract: Strong gravitational lensing is a powerful probe of the mass distribution in the universe, but galaxy group-scale lenses remain relatively unexplored despite being more common than clusters. These systems are particularly important because baryonic matter and dark matter contribute comparably to the lensing signal, making them sensitive probes of galaxy–halo interactions. A major limitation in studying such systems is the high computational cost of traditional Markov Chain Monte Carlo (MCMC) based lens modeling, which can take many hours or days for a single group/cluster system. This becomes prohibitive given the large number of lenses already discovered and the ∼10^5 new strong lenses expected from the Rubin Legacy Survey of Space and Time (LSST).
In this work, I investigate the applicability of GIGA-Lens, a GPU-accelerated Bayesian lens modelling framework, for fast and scalable modelling of group-scale strong lenses. Realistic mock lenses are generated using GLAFIC and analyzed using a three-stage inference pipeline consisting of maximum a posteriori optimization, stochastic variational inference, and Hamiltonian Monte Carlo sampling. The statistical reliability of the inferred parameters is assessed using probability–probability (P–P) plots.
I model a range of simulated group-scale lens systems with varying complexity. The most complex system analysed so far includes one central galaxy and four satellite galaxies, with an effective Einstein radius of ≈ 2.5″, for which the Einstein radius is recovered without significant bias. I also develop a method to estimate an equivalent Einstein radius for group lenses, enabling direct comparison with observations.
|
