logo CRAQ
 

Measuring dark matter halos in strong gravitational lenses with machine learning


Adam Coogan


Université de Montréal



Strongly-lensed galaxies are a unique laboratory for probing small-scale dark matter halos and thus testing the fundamental assumptions of the Lambda-CDM paradigm. However, extracting information about these halos from observations is extremely difficult: their signatures are subtle, the variation between images is large, and inferring halos? properties requires marginalizing over numerous uncertainties in the lens and source galaxies. In this talk I will present a new analysis strategy that leverages simulation-based inference (SBI) to address these challenges. I will explain the advantages of SBI over likelihood-based methods for high-dimensional inference problems such as those present in lensing. I will then show how bringing together several additional machine learning techniques enables the application of SBI to realistic lensing images, paving the way towards application of this pipeline to existing and upcoming datasets.

Date: Jeudi, le 10 mars 2022
Heure: 11:30
Lieu: Université de Montréal
  Zoom
 

Ce site a été optimisé pour les fureteurs Microsoft Internet Explorer, version 6.0 et ultérieures, et Netscape, version 6.0 et ultérieures.