Faculty

Gravitational waves and strong gravity

Dark Universe

The Early Universe

Multimessenger Astrophysics

Experimental gamma-ray astronomy and high-energy astrophysics

Gravitational lensing image analysis with deep probabilistic programming

Recent breakthroughs in machine learning, in particular deep learning, have a profound impact on how data analysis and modeling in many scientific areas are done. In particular, deep probabilistic programming techniques allow to combine the universal power of deep neural networks with specific scientific predictions derived from first principles.

The nature of dark matter, which constitutes more than 80% of the mass in the Universe, remains unknown. Gravitationally lensed images of distant quasars and galaxies allow us to study the dark matter distribution in the Universe at small scales. This delivers important information about various particle physics models for dark matter.

In this project, the student will first learn advanced deep learning analysis tools (probabilistic programming, geometric deep learning, contrastive learning), as well as the physics of gravitationally lensed images. The student will then explore various ways to improve the sensitivity of current machine-learning based analysis pipelines towards dark matter substructure.

 Gravitational lensing image analysis with deep probabilistic programming

Description of thesis project with [link](christophweniger.com) to website.

Christoph Weniger

Research Areas

Masters Thesis Projects