| |
 |
| Paper: |
Data-driven Fitness Functions for Optimizing Simulations of
Interacting Galaxies |
| Volume: |
532, ASTRONOMICAL DATA ANALYSIS SOFTWARE AND SYSTEMS XXX |
| Page: |
299 |
| Authors: |
West, G.; Ogden, M.; Wallin, J. |
| Abstract: |
Given observational data of systems of interacting galaxies, we seek to determine the values of various dynamical parameters through the optimization of numerical models via genetic algorithms. However, fitting such models is difficult. The core challenges include 1) developing a robust fitness function for quantifying the similarity between model and target images and 2) understanding the symmetries of the interacting system which cause morphological degeneracies that impede optimization. In this paper, we show how naive implementations of fitness functions can yield unintuitive results. We propose a novel fitness function that was developed by utilizing data from the Galaxy Zoo: Mergers project. Testing with these human-scored models led to the adoption of a tidal distortion term in our fitness function that dramatically improved results. We also give a characterization of various symmetries inherent to interacting systems and show how the knowledge of these symmetries can be used to reduce the volume of parameter space when performing optimization. |
|
|
|
|
|
