Skip to main navigation Skip to search Skip to main content

Lens Model Accuracy in the Expected LSST Lensed AGN Sample

  • The LSST Dark Energy Science Collaboration
  • University of Illinois at Urbana-Champaign
  • Kavli Institute for Particle Astrophysics and Cosmology
  • SLAC National Accelerator Laboratory
  • Swiss Federal Institute of Technology Zurich
  • University of Oxford
  • University of Portsmouth
  • Stanford University
  • Stony Brook University
  • Fermi National Accelerator Laboratory

Research output: Contribution to journal › Article › peer-review

Abstract

Strong gravitational lensing of active galactic nuclei (AGN) enables measurements of cosmological parameters through time-delay cosmography (TDC). With data from the upcoming LSST survey, we anticipate using a sample of (Formula presented) O(1000) lensed AGN for TDC. To prepare for this dataset and enable this measurement, we construct and analyze a realistic mock sample of 1300 systems drawn from the OM10 catalog of simulated lenses with AGN sources at z  < 3.1 in order to test a key aspect of the analysis pipeline, that of the lens modeling. We realize the lenses as power-law elliptical mass distributions and simulate 5 yr LSST i-band coadd images. From every image, we infer the lens mass model parameters using neural posterior estimation. Focusing on the key model parameters, θE (the Einstein radius) and γlens (the projected mass density profile slope), with consistent mass–light ellipticity correlations in test and training data, we recover θE with less than 1% bias on average, 6.5% precision per lens, and γlens with less than 3% bias on average, 8% precision per lens. We find that lens light subtraction prior to modeling is only useful when applied to data sampled from the training prior. If emulated deconvolution is applied to the data prior to modeling, precision improves across all parameters by a factor of 2. Finally, we combine the inferred lens mass models using Bayesian hierarchical inference to recover the global properties of the lens sample with less than 1% bias under consistent train–test mass–light correlations.

Original languageEnglish
Article number223
JournalAstronomical Journal
Volume172
Issue number4
DOIs
StatePublished - Oct 2026

Keywords

  • Convolutional neural networks (1938)
  • Hierarchical models (1925)
  • Strong gravitational lensing (1643)

Fingerprint

Dive into the research topics of 'Lens Model Accuracy in the Expected LSST Lensed AGN Sample'. Together they form a unique fingerprint.

Cite this