Göttingen 2025 – wissenschaftliches Programm

Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

T: Fachverband Teilchenphysik

T 53: Neutrino Astronomy III

T 53.3: Vortrag

Mittwoch, 2. April 2025, 16:45–17:00, VG 1.105

IFT on ice: Utilizing numerical information field theory to reconstruct glacial ice parameters — •Matthias Hübl, Laurin Söding, and Philipp Mertsch — Institute for Theoretical Particle Physics and Cosmology (RWTH Aachen University)

Due to their small interaction cross-sections, the detection of high-energy neutrinos requires the use of large, natural detection volumes, like glacial ice in the case of the IceCube Neutrino Observatory. In order to extract precise information from the light that is produced by high energy neutrinos, it has to be calibrated as accurately as possible. This means in particular that the ice properties such as scattering and absorption lengths for propagating photons should be known with high accuracy and spatial resolution. Information Field Theory (IFT) adopts a Bayesian approach, building on methods from different fields of physics, especially field theory and statistical physics. The python package NIFTy (Numerical Information Field Theory) uses the concepts of IFT and implements a variational inference approach in order to reconstruct parameter fields. This is both more robust than maximum-likelihood methods and allows determining the uncertainties of the inferred fields at the same time. Here, we present two approaches for modelling the light propagation in ice that can be interfaced with NIFTy: a differentiable Monte Carlo simulation and a finite-difference code. We compare the performance of both methods and characterise the reconstruction of a mock ice model.

Keywords: Information Field Theory; IceCube