Galaxies and Genes: Towards an Automatic Modeling of Interacting Galaxies

Abstract

The main problems in modeling interacting galaxies are the extended parameter space and the fairly high CPU costs of self-consistent N-body simulations. Therefore, traditional modeling techniques suffer from either extreme CPU demands or trapping in local optima (or both). A very promising alternative approach are evolutionary algorithms which mimic natural adaptation in order to optimize the numerical models. One main advantage is their very weak dependence on starting points which makes them much less prone to trapping in local optima. We present a Genetic Algorithm (GA) coupled with a fast (but not self-consistent) restricted N-body solver. This combination allows us to identify interesting regions of parameter space within only a few CPU hours on a standard PC or a few CPU minutes on a parallel computer. Especially, we demonstrate here the ability of GA-based fitting procedures to analyse observational data automatically, provided the data are sufficiently accurate.