**Résumé**

**Measuring the Masses of Galaxies: Tests of Bayesian mass estimation**

#### Gwendolyn Eadie (McMaster University)

The total mass and mass profile are the most fundamental properties of a galaxy, driving its evolution from earliest stages. Thus, mass measurements of galaxies place strong constraints on galactic evolution models and cosmological simulations of dark matter halos in the universe. A powerful method to measure the mass profile is through the velocities of tracer particles distributed through its halo (such as halo stars or globular clusters), but transforming this kind of data accurately to a mass profile M(r) is not a trivial problem. In particular, I will discuss the effects that limited or incomplete data have on the analysis --- for example, what biases occur when complete 3D space motions for the particles are not available? And is there an optimum way to deal with incomplete data? To investigate such biases, I analyze simulated kinematic data sets in an isotropic Hernquist model. To deal with unknown velocity components, I create a hybrid-Gibbs sampler that treats the unknown motions as parameters in the model. The software I developed for this work will soon be available as an open source package as part of the R Project for Statistical Computing.

(doit être confirmé par le SOC)