Non-parametric reconstruction of distribution functions from observed galactic discs

A general inversion technique for the recovery of the underlying distribution function for observed galactic discs is presented and illustrated. Under the assumption that these discs are axisymmetric and thin, the proposed method yields a unique distribution compatible with all the observables available. The derivation may be carried out from the measurement of the azimuthal velocity distribution arising from positioning the slit of a spectrograph along the major axis of the galaxy. More generally, it may account for the simultaneous measurements of velocity distributions corresponding to slits presenting arbitrary orientations with respect to the major axis. The approach is non-parametric, i.e. it does not rely on a particular algebraic model for the distribution function. Special care is taken to account for the fraction of counter-rotating stars, which strongly affects the stability of the disc. An optimization algorithm is devised — generalizing the work of Skilling &38; Bryan — to carry this truly two-dimensional ill-conditioned inversion efficiently. The performance of the overall inversion technique with respect to the noise level and truncation in the data set is investigated with simulated data. Reliable results are obtained up to a mean signal-to-noise ratio of 5, and when measurements are available up to 4  R _e. A discussion of the residual biases involved in non-parametric inversions is presented. The prospects of application of the algorithm to observed galaxies and other inversion problems are discussed.