Modeling Galaxy Formation using Robust Machine Learning Techniques

Robert J. Brunner, University of Illinois at Urbana-Champaign

Usage Details

This project, will conduct a computational, non-parametric study of the extent of the importance of ΛCDM model in the final result of small-scale gaseous interaction that form a galaxy by using cutting-edge cosmological simulations. Developing and using a variety of sophisticated machine learning algorithms is computationally expensive on the large data sets that are going to be involved. This will be an exploratory effort to move our existing machine learning efforts (as applied to the Millennium simulation) to larger scales by using Blue Waters. We also will extend this initial effort to explore how the combination of n-body and hydrodynamic simulations, as provided by the recently released Illustris simulation, can improve our models.

Our proposed approach has the potential to revolutionize the analysis of massive cosmological simulations by allowing significantly larger and more detailed mock galaxy catalogs to be generated for a given dark matter simulation. We also may be able to quantify how machine learning can be used to acerbate and guide future cosmological simulations.