Frederik De Ceuster

FrederikDeCeuster_2.png

Office 200D - 04.37

Institute of Astronomy

KU Leuven, Belgium

I’m a postdoctoral research fellow of the Research Foundation - Flanders (FWO), leading the computational research division in the group of Prof. Dr. Leen Decin at the Institute of Astronomy (KU Leuven).

Our research revolves around modelling evolved stars as they shed their outer layers, enriching our Universe with the chemicals they produce, providing the building blocks for future generations of stars, planets, and perhaps even life. This involves 3D hydrodynamics, chemical kinetics, and radiative transfer modelling. As observations reveal ever more intricate structures (see e.g. Decin et al. 2020) our models become ever more complex, and it becomes increasingly difficult to compare them to observations. In our computational research team, we tackle this problem is several different ways, with:

  • Probabilistic 3D reconstruction of observations, to create data-augmented models that can account for the observed complexities;
  • Fast surrogate models, using machine learning to train fast and quantifiably approximate models to by-pass expensive modelling steps, such as chemical kinetics and radiative transfer;
  • Efficient use of high-performance computing resources, developing solutions that can leverage multi-node many-core systems as well as (multi-)GPU systems.

My personal research interests encompass a broad range of topics at the interface between mathematics, physics, and engineering. Currently, I mainly focus on developing efficient methods for 3D reconstruction of astronomical observations to compare with our complex theoretical models, using variational Bayesian methods and probabilistic numerics.

I obtained a PhD in computational astrophysics from University College London, working with Revd. Dr. Jeremy Yates, Prof. Dr. Peter Boyle, and Prof. Dr. James Hetherington. Here, I started developing Magritte: a software library for 3D radiative transfer and synthetic obsevations.

For a gentle introduction to my research, see e.g. this interview with Kennismakers (Dutch only) or this #ThesisThread.

Prior to that, I obtained a BSc and MSc in Physics from KU Leuven.

selected publications

  1. MNRAS
    Radiative Transfer as a Bayesian Linear Regression problem
    De Ceuster, F., Ceulemans, T., Cockayne, J., Decin, L., and Yates, J.
    Monthly Notices of the Royal Astronomical Society Feb 2023
  2. JOSS
    3D Line Radiative Transfer & Synthetic Observations with Magritte
    De Ceuster, F., Ceulemans, T., Srivastava, A., W., Homan, Bolte, J., Yates, J., Decin, L., Boyle, P., and Hetherington, J.
    Journal of Open Source Software 2022
  3. MNRAS
    MAGRITTE, a modern software library for 3D radiative transfer - II. Adaptive ray-tracing, mesh construction, and reduction
    De Ceuster, F., Bolte, J., Homan, W., Maes, S., Malfait, J., Decin, L., Yates, J., Boyle, P., and Hetherington, J.
    Monthly Notices of the Royal Astronomical Society Dec 2020
  4. Science
    (Sub)stellar companions shape the winds of evolved stars
    Decin, L., Montargès, M., Richards, A. M. S., Gottlieb, C. A., Homan, W., McDonald, I., El Mellah, I., Danilovich, T., Wallström, S. H. J., Zijlstra, A., Baudry, A., Bolte, J., Cannon, E., De Beck, E., De Ceuster, F., de Koter, A., De Ridder, J., Etoka, S., Gobrecht, D., Gray, M., Herpin, F., Jeste, M., Lagadec, E., Kervella, P., Khouri, T., Menten, K., Millar, T. J., Müller, H. S. P., Plane, J. M. C., Sahai, R., Sana, H., Van de Sande, M., Waters, L. B. F. M., Wong, K. T., and Yates, J.
    Science Sep 2020
  5. MNRAS
    MAGRITTE, a modern software library for 3D radiative transfer: I. Non-LTE atomic and molecular line modelling
    De Ceuster, F., Homan, W., Yates, J., Decin, L., Boyle, P., and Hetherington, J.
    Monthly Notices of the Royal Astronomical Society Feb 2020