Stellar dynamics provide a fossil record of galaxy formation. Orbit-based dynamical modelling is therefore a powerful tool which can be used to infer a galaxy’s intrinsic structure and its mass accretion history. We use this orbit-based modelling to recover the internal structure of spectroscopically observed galaxies and disentangle individual components in the orbit space of galactic radius and angular momentum, separating dynamically distinct features which link back to different origins. These features, however, are sensitive to noise.
I am using results from TNG50, cosmological simulations with high enough resolution to analyze the stellar components of individual galaxies. These allow us to construct similar orbit distributions as with the dynamical models of observed galaxies. As part of the SFB, I am working with mathematicians on how we can use these simulated orbit distributions to inform regularization on our dynamical inference to yield robust uncertainties. Additionally, I use the same set of simulated galaxies to explore the link between present-day stellar dynamics and past merger events to improve astrophysical interpretation of orbit distributions. In this talk, I will present my work on constructing orbit distributions from simulations similar to those extracted from observations by dynamical modelling and our efforts o construct physically motivated regularisation schemes. I will discuss how this and my findings on tentative correlations between the satellite dynamics and the stellar debris originating from the merger event will inform the recovery and interpretation of future dynamical models of observed galaxies.