"Novel Forward and Inverse Approaches to Modelling the Observable Universe"
ORIGINS Guest Lecture
- Datum: 29.04.2026
- Uhrzeit: 11:00 - 12:00
- Vortragende(r): Chris Lovell (Kavli Institute for Cosmology, Cambridge)
- Ort: Universitäts-Sternwarte München (USM), Scheinerstraße 1, 81679 Munich, Germany
- Raum: 133 (1st floor, main building at Scheinerstr.1)
- Gastgeber: ORIGINS Excellence Cluster
- Kontakt: odele.straub@origins-cluster.de
Chris Lovell presents new forward and inverse approaches to simulating the observable Universe, combining zoom simulation frameworks, galaxy emission modelling, and deep learning methods to bridge cosmic-scale structure and individual galaxy evolution. The lecture is also accessible online.
Summary:
Just as people living in busy cities experience life differently from those in rural villages, galaxies evolve dramatically differently depending on their surroundings. Some galaxies live in crowded cosmic metropolises, massive clusters containing thousands of galactic neighbors, while others exist in nearly empty cosmic countryside. New observatories, such as Euclid, are currently mapping billions of galaxies, offering an unprecedented opportunity to understand these cosmic environments. We rely on simulations to understand and interpret data from these modern instruments, but current models face a fundamental limitation: they can either simulate small cosmic regions in exquisite detail, or vast volumes of space, but not both simultaneously. It's like trying to understand New York by either examining individual streets with a magnifying glass or viewing the entire city from space; neither approach alone tells the complete story.
In this talk I'll describe new forward and inverse approaches to this challenge. Our "zoom” simulation framework, combined with detailed models of galaxy emission and deep learning methods, generates detailed mock Universe's that, like Google Earth seamlessly transitioning from satellite view to street level, captures both cosmic-scale structure and individual galaxies simultaneously. At the same time, new inference methods are rapidly accelerating the process of inverse modelling, allowing us to comprehensively explore the impact of our modelling choices on derived physical parameters, providing more robust constraints on the process of galaxy evolution, and, ultimately, the underlying cosmological model.
Just as people living in busy cities experience life differently from those in rural villages, galaxies evolve dramatically differently depending on their surroundings. Some galaxies live in crowded cosmic metropolises, massive clusters containing thousands of galactic neighbors, while others exist in nearly empty cosmic countryside. New observatories, such as Euclid, are currently mapping billions of galaxies, offering an unprecedented opportunity to understand these cosmic environments. We rely on simulations to understand and interpret data from these modern instruments, but current models face a fundamental limitation: they can either simulate small cosmic regions in exquisite detail, or vast volumes of space, but not both simultaneously. It's like trying to understand New York by either examining individual streets with a magnifying glass or viewing the entire city from space; neither approach alone tells the complete story.
In this talk I'll describe new forward and inverse approaches to this challenge. Our "zoom” simulation framework, combined with detailed models of galaxy emission and deep learning methods, generates detailed mock Universe's that, like Google Earth seamlessly transitioning from satellite view to street level, captures both cosmic-scale structure and individual galaxies simultaneously. At the same time, new inference methods are rapidly accelerating the process of inverse modelling, allowing us to comprehensively explore the impact of our modelling choices on derived physical parameters, providing more robust constraints on the process of galaxy evolution, and, ultimately, the underlying cosmological model.