Researchers from the Harvard-Smithsonian Center for Astrophysics, the MIT Kavli Institute for Astrophysics and Space Research, the Flatiron Institute, and the Max Planck Institute for Astrophysics have developed a novel approach to improve the estimation of galaxy cluster properties, which could have implications for cosmological studies and energy applications.
Galaxy clusters, particularly massive ones, are crucial for understanding the universe’s structure and cosmology. Traditionally, these clusters have been modeled as spherical for simplicity, but this approach can introduce biases in estimating their properties. The research team has introduced a deep-learning method to estimate the triaxiality and orientations of massive galaxy clusters from two-dimensional observables. This method leverages the power of convolutional neural networks (CNNs) and graph neural networks (GNNs) in a multi-modal, fusion network.
The researchers utilized the MillenniumTNG (MTNG) simulations as their ground truth. Their model extracts three-dimensional geometry information from two-dimensional, idealized cluster multi-wavelength images (soft X-ray, medium X-ray, hard X-ray, and tSZ effect) and mathematical graph representations of two-dimensional cluster member observables (line-of-sight radial velocities, two-dimensional projected positions, and V-band luminosities). This approach improves cluster geometry estimation by 30% compared to assuming spherical symmetry. The model achieves an R² regression score of 0.85 for estimating the major axis length of triaxial clusters and correctly classifies 71% of prolate clusters with elongated orientations along the line-of-sight.
For the energy sector, understanding the structure and properties of galaxy clusters can aid in the development of more accurate models for dark matter and dark energy, which are believed to influence the large-scale structure of the universe. This research, published in the Monthly Notices of the Royal Astronomical Society, offers a more precise method for estimating cluster properties, which could enhance our understanding of the universe’s energy content and evolution.
This article is based on research available at arXiv.