Sarajedini, G. Piotto
The image captured by NASA/ESA’s Hubble Space Telescope shows the globular star cluster NGC 6652.
If we embarked on a journey through the various eras of astronomy, we would start with the ancients who wondered why some “fixed lights” populated the night sky.
We would continue through eras where it was discovered that these lights are actually celestial bodies similar to our Sun and that our planet orbits around it. Moving forward, we would witness the discovery of gravity, influenced by the structure of spacetime, and the observation of iridescent galaxies beyond the Milky Way, up to calculating the extreme limits of supermassive black holes. In the current context, we notice an intensification of collaboration between astronomers and advanced technologies such as machine learning, which significantly accelerates our ability to explore the universe. Aritra Ghosh, a postdoctoral researcher at the University of Washington, is one of the astronomers involved in these innovative researches. Recently, Ghosh confirmed that galaxies located in the densest regions of the universe can be up to 25% larger than those of similar mass and shape located in less dense regions.
The “radius” of a galaxy considered here includes 50% of its total light emission. This discovery was made possible by using machine learning to analyze a number of galaxies – precisely 2,894,716 – more than could be analyzed humanly in a lifetime. Ghosh obtained these data from an even larger set collected through GaMPEN, a tool capable of analyzing the structure of galaxies based on user-specified parameters; in this specific case, it was about the relationship between the light emitted by the outer disk of galaxies compared to their central core. The work carried out not only allowed the use of the largest catalog ever employed to study the sizes of galaxies relative to their environment but also introduced corrective mechanisms for errors almost absent in previous studies thanks to the integration of machine learning. Furthermore, the results showed how larger galaxies tend to be located in superclusters, contrary to expectations based on internal dynamics of the clusters themselves which theoretically should reduce their sizes by dispersing matter in the process. These observations open new questions about the possible influence of dark matter in keeping galaxies larger in densely populated areas of the universe or about the original formation of the same already larger or ease of growth through mergers.
The team is now setting their sights on the Rubin Observatory, expected to illuminate the skies in the very early months of 2025 with the aim of producing even larger datasets.
The research conducted by Ghosh was published on August 14 in The Astrophysical Journal, thus demonstrating how machines can be reliable in addressing questions regarding the universe.
Follow our feed
