next up previous contents
Next: References Up: Conclusions Previous: A brief Summary

Suggestions for Future Work

  Even though the AVP has been studied quite thoroughly in the last ten years, there still are quite a few aspects of the method that can be improved. As seen in this study, there is a lot of work that needs to be done in finding, or even creating, an efficient numerical optimizing method which is able to locate all the allowed solutions. In addition, there are quite a few techniques that may speed up the calculation of these methods: one being the relative novel concept of parallel non-linear optimization. The improvement of the computer processors is closing in on barriers created by the speed of light and the size of atoms, and the only way of improving the speed is to run several processors in parallel, distributing the calculations between them. A discussion of the concept of doing optimization on parallel computers can be found in [Schnabel1994].

The main problem with such a rather dense system as the LG and the LN is the occurrence of multiple solutions. There are no guarantee that one solution is more accurate than the other. Even if one excludes the bad solutions and the solutions that does not recreate the radial velocities of the galaxies that have fairly accurate observed distance estimates, one will still be left with quite a few probable ones. A method which discriminates between the solutions is therefore needed.

As seen from this study, the AVP is by no means fully evolved--not optimized, so to speak. More effort put into this area can prove to be fruitful for the study of the evolution of large-scale structures in the Universe.

Trond Hjorteland
Mon Jul 5 02:59:28 MET DST 1999