Classifying Large-Scale Structure Galaxies Using Machine Learning

Matias Carrasco Kind, University of Illinois at Urbana-Champaign

Usage Details

At very large scales, the cosmic landscape is dominated by web-like large-scale structure (LSS), consisting of clusters, filaments, and voids. Observations and computational simulations indicate that LSS traces the distribution of dark matter. An understanding of the distribution of LSS will provide key insight into the fundamental properties of dark matter/energy; however, current methods for classifying galaxies in terms of their LSS structural identity are limited due to its complexity.

We propose a novel method of LSS galaxy classification using machine learning (ML). Using algorithms, we constructed, we will train an ML algorithm to classify LSS galaxies using a simple model. Currently, we have demonstrated the effectiveness of our algorithm on small test samples. Using Blue Waters, we can scale up our training model to include 10⁹ galaxies, allowing us to classify observed data from large surveys.