51ÁÔÆæÈë¿Ú

Abstract:

An algorithm is developed and tested for the problem posed by photometric observations of the bulge of the Milky Way. The latter subtends a non-trivial solid angle on the sky, and we show that this permits inversion of the projected brightness distribution under the assumption that the bulge has three orthogonal mirror planes of specified orientation. A serious error in the assumed orientation of the mirror planes should be detectable.

Abstract:

Axisymmetric density distributions are constructed which are invisible when viewed from a range of inclination angles $i$. By adding such distributions to a model galaxy, it can be made either disky or boxy without in any way affecting its projected image. As the inclination of a galaxy decreases from edge-on to face on, the range of `invisible' densities, the uncertainty in the deprojection, and the sensitivity of the deprojection to noise all increase. The relation between these phenomena is clarified by an analysis of Palmer's deprojection algorithm. These results imply that disk-to-bulge ratios are in principle ill-determined from photometry unless the disk is strong or the system is seen precisely edge-on. The uncertain role of third integrals in galaxies makes it unclear to what degree this indeterminacy can be resolved by kinematic studies.

Abstract:

A scheme is developed which enables one to trace backwards in time the cosmic density and velocity fields, and to determine accurately the current-epoch velocity field from the current-epoch density field, or vice versa. The scheme implements the idea of Giavalisco \etal\ (1993) that the principle of least action should be used to formulate gravitational instability as a two-point boundary-value problem. We argue that the Eulerian formulation of the problem is to be preferred to the Lagrangian one, on grounds of computational simplicity, of ease of interfacing with observational data, and of internal consistency at early times. The scheme is successfully tested on an exact solution in one dimension, and on currently Gaussian fields in one and two dimensions. The application of the scheme to real observational data appears to be eminently feasible, though computationally costly.