| Abstract: |
| Gravitational interactions between galaxies play a pivotal role in galaxy formation and evolution creating tidal distortions, new star formation, galactic mergers, and active galactic nuclei. Due to observational limitations, simulations play an integral part of galaxy research. Gravitational n-body and restricted three-body simulations, which capture underlying gravitational dynamics, can be used to create the complex morphologies observed in galaxies. However, when attempting to model observed interacting galaxies, there are over a dozen unknown dynamical parameters due to the same limitations. We are developing a method to explore the unknown dynamical parameters by using the observed morphologies. One significant challenge is accurately matching the morphology of real and simulated systems. To address this issue, we use Computer Vision and Deep Learning to create a fitness function between simulation and observational data using citizen science data. We then optimize this fitness function generating models matching the observed target better than previous models. By analyzing the best-fit models and their dynamical parameters, we can postulate values and boundaries for previously unknown values. Ultimately, this research will become a valuable tool to explore dynamic gravitational systems otherwise limited by observational data. |
|