On fitting the morphology of simulations of interacting galaxies to synthetic data

Abstract

Gravitational interactions between galaxies represent a fundamental cosmological process. These interactions are responsible for numerous aspects of the formation and evolution of galaxies, such as enhanced and suppressed star formation rates, the development of tidal features, and the feeding active galactic nuclei. Given observational data from 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 these models can be quite difficult. The core challenges include 1) developing an objective fitness function for quantifying the similarity between model and target images, 2) understanding the inherent symmetries of the dynamical system which promote morphological degeneracies and impede optimization, 3) determining the optimal genetic algorithm operators for the problem.

In this dissertation, we show how naive implementations of fitness functions can yield unintuitive results. We then propose a novel fitness function which was developed by utilizing data from the \textit{Galaxy Zoo: Mergers} project (GZM). The human-scored models obtained from GZM were used to validate our fitness function and led to the adoption of a tidal distortion term which dramatically improved results. We also give a characterization of various geometric and dynamical symmetries inherent within the system and show how the knowledge of these symmetries can be used to reduce the volume of the parameter search space when performing optimization. Lastly, we implement a real-coded genetic algorithm with features designed to address these symmetries. Using simulated target systems with known parameters as a surrogate for observational data, we test our fitness function and genetic algorithm for robustness, accuracy, and convergence. We discuss the link between the degree of tidal distortion present in a target image and the constraints on the dynamical parameters using three different target systems with varying morphology.

As an offshoot of our development of our work on the galaxy optimization problem, we also present a kernel mixing strategy which can be applied in both stochastic optimization and adaptive Markov chain Monte Carlo contexts. The method is flexible and robust enough to handle parameter spaces that are highly multimodal. We provide results from several benchmark problems, incorporating the method into simulated annealing, real-coded genetic algorithm, and adaptive Markov chain Monte Carlo contexts. Results show a significant increase in performance in variants of these methods which incorporate the mixing strategy over those which do not.