Large Scale Modelling of the Milky Way - An Application of New Goodness of Fit Techniques in Inverse Regression Models

Thursday, April 25, 2002 - 1:30pm - 2:30pm
Keller 3-180
Nicolai Bissantz (Georg-August-Universität zu Göttingen)
Joint work with Axel Munk.

Models of the large-scale Milky Way near-infrared luminosity distribution are of great interest in Astronomy because they trace the distribution of (luminous) mass. Thus they allow the derivation of the gravitational potential due to luminous mass, and subsequently of models of gas dynamics and stellar kinematics. We show that reconstruction of the luminosity distribution is a noisy ill-posed problem. In particular, recovering the spatial density of the Milky Way from the two-dimensional surface luminosity on the sky causes additional difficulties.

In our talk we first introduce the astrophysical background of the problem. Then we present a non-parametric model of the luminosity distribution derived by a penalized maximum likelihood approach. In the third part of our talk we introduce a new method which allows to decide whether a non-parametric model improves over given parametric models, which might be of interest by its own. This is applied to the reconstrution of the MW and we show that our non-parametric model is superior to parametric models. Moreover, our method can be modified to analyse the spiral structure of the Milky Way in the near-infrared.

Our talk closes with a brief overview on gas dynamical models of the Milky Way based on our presented model of the luminosity distribution. We comment on the substantial implications of these dynamical models for cosmology.