Oral Presentation
Machine-assisted Exploration of Simulation Parameters
Presenter: Boon Kiat Oh (Seoul National University)
Numerical simulation is a key component used to study the formation and evolution of large-scale structures in the universe. The predictive capability of the simulations is realized when they are calibrated to match observations. However, it is near impossible to fully explore the parameter space of a numerical simulation. In this talk, I will present a novel method of using machine learning to obtain a set of parameters that produces a simulated galaxy matching observed properties at z=0. I will discuss the improvements in the accuracy of the results and demand in computational resources with this approach. Also, I will demonstrate the robustness of the algorithm by applying it to a different problem faced in numerical simulations.
asiaa.sinica.edu.tw