Modern observations describe galaxies as images, spectra, and data cubes of extraordinary complexity. Yet before any physical inference is made, we choose how that information is represented: as pixels, apertures, radial profiles, parametric components, or bins. These choices are useful, but they also impose a geometry and a scale on the data, determining which structures remain visible and which are averaged away.
I will discuss a different approach in which representation is treated as part of the scientific problem itself. Spatial regions can be defined adaptively from spectral or photometric coherence rather than fixed geometry; multiscale transforms can separate structures before they are grouped; and classical radial descriptions can be generalized into families of paths and trajectories, borrowing ideas from geometry and fluid mechanics. The same path-based viewpoint can also be applied in spectral space, where the ordered shape of an emission line becomes a geometric object carrying kinematic information.
The broader aim is to develop representations that follow the structure of the data rather than forcing the data into a predetermined coordinate system. I will show how these ideas can connect morphology, stellar populations, ionized gas and kinematics, and argue that scale, geometry and information loss offer new ways of thinking about spatially resolved galaxies.
Horarios: 22 Sep 2026
Publicado por: Gijs Mulders