Gravitationally Imaging Dark Matter
Strong gravitational lensing is a powerful tool to investiagete a wide range of astrophysical phenomena. Our group focuses on deriving observational constraints on the properties of dark matter, the interplay between dark matter physics and baryonic processes and its effect on the physics of structure formation. By constraining the detailed physical properties of high-redshift lensed galaxies we also aim to quantify star formation processes and AGN activity at cosmologically-interesting epochs on sub-kpc scales. We constantly develop new approaches to gravitational lens modelling and the analysis of high-angular resolution data from optical imaging and radio/mm/sub-mm interferometeric observations.
My research involves differentiable programming, deep generative modelling and its applications in physical sciences. Traditional computational methods often fall short when faced with large amounts of data. For example, it is computationally infeasible to model all gravitational lens systems that will be discovered from upcoming surveys in the near future. Supplementing and enhancing existing algorithmic methods with data-driven machine learning can help to solve this problem. Deep generative models can be used to efficiently learn a prior data distribution. I am interested in integrating differentiable deep generative models into gradient-based Bayesian modelling frameworks. In order to draw reliable scientific conclusions from such methods, it is necessary to study their robustness and systematic uncertainty.
My work focuses on the analysis of galaxy-galaxy strong lensing observations from the Hubble Space Telescope and Keck II, that were collected by the SHARP program. The aim of the project is to gravitationally detect low-mass dark matter haloes with masses in the regime between 10^8 and 10^9 solar masses. My goal is to deliver observational constraints on the low-mass end of the halo mass function and test predictions from different dark matter models.
My work aims at using strong gravitational lensing observations to measure the properties of magnetic fields in distant galaxies. To this end, I am developing a lens modelling code that simultaneously reconstructs the source surface brightness distribution in its four polarisation states and the magnetic field of the lens via differential Faraday rotation of the lensed images. By analysing VLBI and LOFAR data, and comparing the results with magnetohydrodynamical simulations of galaxy formation, I aim to understand the evolution of magnetic fields in galaxies and their influence on the properties of the galaxies themselves.
My work generally examines the abilities and limits of strong lensing to tell us about the universe, using a combination of theory and computation. In the near future the number of known strong lenses will increase dramatically. In this context, I am developing a machine learning tool that selects the lenses from survey data which provide the best chance of finding substructure in follow-up observations. I am also interested in the constraints that strong lensing can provide on galaxy mass profiles. Previously, I developed a theory to explain the constraints on a power-law profile in typical high S/N strong lens observations. More generally I am interested in Bayesian statistics, high-performance computing and machine learning.
My work focuses on the search for low-mass dark matter haloes, with the goal of constraining dark matter particle models via the halo mass function. The main component of my work is to develop a gravitational lens modelling code to handle very long baseline interferometric (VLBI) radio data. This is advantageous because our ability to measure the abundance of low-mass dark matter haloes depends strongly on the angular resolution of the observation. While VLBI provides some of the best angular resolution in astronomy these datasets are large and therefore present unique computational challenges for modelling.
I work on Bayesian modelling of strong gravitational lensing by galaxy clusters. After having finished my master's degree at the Ludwig Maximilian University, where I developed a Bayesian imaging algorithm for combining different measurement methods in the radio regime, I joined Simona Vegetti's group at the Max Planck Institute for Astrophysics in Garching. My current work focuses on the use of galaxy clusters as cosmic telescopes to study the detailed properties of high-redshift galaxies. Here I am primarily interested in finding an efficient description of the underlying inference problem.
My research focuses on radio/mm/sub-mm interferometric imaging of gravitational lenses. The interplay between AGN and star formation is thought to be important to understand how galaxies evolve. By studying gravitationally lensed galaxies with telescopes such as the Atacama Large sub-Millimetre Array (ALMA) we can resolve structures in distant galaxies at the peak of cosmic star formation and black hole accretion, which can help understand this interplay. The high-resolution data we can obtain with interferometry can also be useful to investigate small-scale mass structures in the lens, which can help us test models for dark matter.