A recent study has demonstrated that galaxies with more neighbors often have a larger size compared to those that share similar shape and mass but exist in less crowded areas. By employing a machine-learning algorithm to scrutinize millions of galaxies, the research team found that galaxies in denser regions of the universe can be up to 25% larger than their isolated counterparts. These results illuminate an ongoing debate among astrophysicists regarding the correlation between a galaxy’s size and its surroundings, while also posing fresh inquiries into the processes of galaxy formation and development over the course of billions of years.
For many years, scientists have recognized that certain galaxies inhabit environments dense with neighboring galaxies, while others float through space largely alone, with few or no companions in their part of the universe.
A recent study has identified a significant distinction among galaxies in these differing environments: Galaxies that have more neighboring galaxies tend to be larger compared to their similar-shaped and similarly-massed counterparts that find themselves in less crowded spaces. In a paper published on August 14 in the Astrophysical Journal, researchers from the University of Washington, Yale University, the Leibniz Institute for Astrophysics Potsdam, Germany, and Waseda University, Japan, indicated that galaxies located in denser areas of the cosmos could be as much as 25% larger than those that are isolated.
This research, which used a newly developed machine-learning tool to analyze millions of galaxies, sheds light on a long-disputed question among astrophysicists concerning the interplay between a galaxy’s size and its surrounding environment. Moreover, the findings invite new inquiries into how galaxies are originated and develop over vast timescales.
“Current theories on galaxy formation and evolution fail to sufficiently account for the observation that galaxies in clusters are larger than those in more isolated regions,” explained lead author Aritra Ghosh, a postdoctoral researcher in astronomy at UW and an LSST-DA Catalyst Fellow at the UW’s DiRAC Institute. “This aspect of astrophysics is particularly fascinating; there can be discrepancies between what theories predict and what we actually observe. When that occurs, we revisit our theories to align them with these observations.”
Previous investigations examining the correlation between galaxy size and its environment yielded mixed results. Some studies suggested that galaxies within clusters were smaller than isolated ones, while others reached the opposite conclusion. Generally, these earlier studies operated on a much smaller scale, focusing on hundreds or thousands of galaxies.
In contrast, Ghosh and his team employed data from a survey of millions of galaxies obtained using the Subaru Telescope in Hawaii. This project, known as the Hyper Suprime-Cam Subaru Strategic Program, produced high-resolution images of numerous galaxies. The researchers selected around 3 million galaxies with the most reliable data and applied a machine learning algorithm to estimate the size of each one. They then defined a circle with a 30 million light-year radius around each galaxy to represent its immediate environment, exploring the simple question: How many neighboring galaxies lie within this circle?
This analysis revealed a clear trend: Galaxies with more neighbors tended to be larger on average.
Multiple explanations could account for this phenomenon. It’s possible that galaxies in densely packed regions simply start off larger or are more inclined to merge efficiently with nearby galaxies. Dark matter, the elusive substance comprising most of the universe’s mass, which is not directly observable by current methods, might also play a role. Galaxies are formed within individual “halos” of dark matter, and the gravitational influence of these halos significantly impacts the evolution of galaxies.
“To definitively understand why galaxies that are more clustered tend to be larger, theoretical astrophysicists will need to conduct comprehensive studies with simulations,” stated Ghosh. “Presently, we can confidently assert that a correlation exists between a galaxy’s environment and its size.”
The extensive dataset provided by the Hyper Suprime-Cam Subaru Strategic Program facilitated the team’s clarity regarding their findings. However, this was only part of the equation. The innovative machine-learning tools they utilized to determine each galaxy’s size effectively addressed the uncertainties inherent in size measurements.
“One crucial lesson we learned before this study is that resolving such questions requires not just surveying large numbers of galaxies, but also performing detailed statistical analyses. Part of this involves using machine learning tools capable of accurately assessing measurement uncertainties in galaxy characteristics,” noted Ghosh.
The machine-learning system they employed is named GaMPEN, which stands for Galaxy Morphology Posterior Estimation Network. Ghosh spearheaded the development of GaMPEN during his doctoral studies at Yale, and it was introduced in papers published in 2022 and 2023 in the Astrophysical Journal. This tool is accessible online and can potentially be adapted for analyzing other extensive surveys, according to Ghosh.
While this study concentrates on galaxies, it also hints at the type of research that will soon revolutionize the field of astronomy—research rooted in intricate analyses of massive datasets. Once a new generation of telescopes equipped with powerful cameras, such as the Vera C. Rubin Observatory in Chile, becomes operational, they will gather enormous volumes of cosmic data nightly. To prepare for this, scientists have been developing innovative tools like GaMPEN that can leverage these large datasets to tackle significant astrophysical questions.
“Shortly, large datasets will be commonplace in astronomy,” stated Ghosh. “This study exemplifies what can be achieved with them when the right analytical tools are in place.”
Co-authors of the study include Meg Urry, professor of physics and astronomy at Yale; Meredith Powell, a research fellow at the Leibniz Institute; Rhythm Shimakawa, associate professor at Waseda University; Frank van den Bosch, a Yale professor of astronomy; Daisuke Nagai, professor of physics and astronomy at Yale; Kaustav Mitra, a doctoral student at Yale; and Andrew Connolly, a professor of astronomy at UW and a faculty member of the DiRAC Institute and the eScience Institute. Funding for the research came from NASA, the Yale Graduate School of Arts & Sciences, the John Templeton Foundation, the Charles and Lisa Simonyi Fund for Arts and Sciences, the Washington Research Foundation, and the UW eScience Institute.
