Publication Date

Summer 2025

Degree Type

Thesis

Degree Name

Master of Science (MS)

Department

Physics and Astronomy

Advisor

Kassahun Betre; Curtis Asplund; Aaron Romanowsky

Abstract

The Vera C. Rubin Observatory is expected to discover ∼ 103 galaxy-scale strongly lensed AGN systems during its 10 year operation conducting the Legacy Survey of Space and Time (LSST). I use a machine learning neural posterior estimation method to automatically model these systems with the intention of using them for time-delay cosmography, to constrain the Hubble constant and, ultimately, the properties of Dark Energy. In the work shown here, the neural network that I utilize is trained on simulated lens systems with simple mass distributions that consist of a single main deflecting galaxy. I am testing the robustness of the posterior inferences of the lens system parameters made by the neural network when adding sub-galactic structure to the main deflector in the form of dark subhalos and a single luminous pertrubing mass. To test these predictions I simulate test sets of 100 mock Hubble Space Telescope (HST) quality lenses and add complex mass to the plane of the main deflector in the form of either a single luminous perturbing mass or a realistic population of dark subhalos. The presence of the luminous perturber introduces a bias in the Fermat potential of ∼ 6%, indicating a similar bias in H0, while reducing the precision of the model by a factor of 1.25. This indicates the need to train our network with a more complex, realistic mass model that includes small-scale mass structures.

Download

Included in

Physics Commons

Share

COinS