Demystifying impossible supernovae using astrophysics, statistics, and machine learning
Michigan State University
Type Ia supernovae are the explosions of white dwarfs. The lack of hydrogen in the spectra, copious amounts of intermediate mass elements in the spectra, and the light curve evolution point to a thermonuclear explosion in degenerate CO matter. However, there is no known mechanism for single white dwarfs to explode. Thus the community has come up with several scenarios involving binary interaction. None of the proposed scenarios have been definitively proven or disproven. The lack of understanding of the origin of these explosions has serious repercussions on the important roles that Type Ia supernovae play. In this talk, I will present how my group approaches this problem from a variety of angles. The binary evolution makes a unique prediction: a surviving companion. I will highlight our current effort on finding such a companion in Galactic remnants. The scenarios also differentiate themselves by either progressing via a deflagration or detonation. I will discuss our machine learning approach to reconstruct the explosion from time-series spectra gaining insights in the flame propagation. Finally, we are also using the different predictions on the nucleosynthesis by the proposed scenarios and compare them to data. The nucleosynthetic predictions are tested with our radiative transfer code called TARDIS. I will conclude by discussing how the proposed machine learning techniques can be adapted and used in several other astrophysics problems.
Date: Thursday, 25 February 2021 Time: 11:30 Where: Université de Montréal Zoom Contact: Laurence Perrault Levasseur