In a groundbreaking achievement, researchers have employed the world’s then-fastest supercomputer, Frontier, to conduct the largest astrophysical simulation of the universe to date. This monumental endeavor aimed to unlock the mysteries surrounding the nature of atomic and dark matter, forging a path for new cosmological explorations.
Frontier, classified as an exascale-class supercomputer, is capable of executing a staggering quintillion calculations per second. This immense computational power was essential for simulating the intricate dynamics of the universe, including both visible and non-visible matter. The simulation reflects data obtained from extensive telescope surveys, marking a significant advancement in astrophysical research that was previously thought unattainable.
Salman Habib, director of computational sciences at Argonne National Laboratory, emphasized the importance of integrating diverse physical phenomena in the simulation. He stated, “To truly comprehend the universe, we need to simulate both gravity and all other physical processes, including the dynamics of hot gas and the formation of stars, black holes, and galaxies.” Habib described the simulation as the astrophysical “kitchen sink,” acknowledging that it needed to account for all variables in play in the cosmic arena.
Current astronomical understanding posits that the visible matter—comprising black holes, molecular clouds, planets, and moons—constitutes only about 5% of the universe’s total content. A significant portion, approximately 27%, is believed to be dark matter, a term that encapsulates various particles and phenomena that can only be inferred through their gravitational effects on atomic matter. The remaining 68% is attributed to dark energy, a mysterious component responsible for the rapid expansion of the universe.
According to Habib, simulating vast expanses of the universe, as observed by major telescopes like the Rubin Observatory in Chile, involves investigating billions of years of cosmic evolution. Previously, such simulation efforts could only approximate gravitational forces in isolation. This latest simulation showcases not only the size of the area being studied but also the heightened physical realism guaranteed by including baryonic matter and other dynamic elements into the calculations.
Bronson Messer, director of science for the Oak Ridge Leadership Computing Facility, noted the simulation’s unique capabilities, highlighting how both its scale and detail facilitate direct comparisons to modern observational data.
Located at Oak Ridge National Laboratory, Frontier comprises over 9,400 central processing units (CPUs) and over 37,000 graphics processing units (GPUs), facilitating its status as a leading exascale supercomputer. The recent universe simulations were spearheaded by researchers at Argonne National Laboratory, utilizing advanced coding techniques developed for the Hardware/Hybrid Accelerated Cosmology Code (HACC), a framework that has seen significant updates over the last fifteen years through the Department of Energy’s Exascale Computing Project.
Although Frontier achieved these remarkable simulation results while still maintaining its position as the fastest supercomputer in the world, it was soon surpassed by the El Capitan supercomputer. El Capitan boasts a verified capacity of 1.742 quintillion calculations per second and an impressive peak performance of 2.79 quintillion calculations per second, underscoring the rapidly evolving landscape of supercomputing capabilities.
As our understanding of the cosmos deepens, these groundbreaking simulations pave the way for further exploration into the universe’s composition, offering a more nuanced perspective on the intricate dance of matter and energy that defines our reality. The pursuit of knowledge in this field not only enriches scientific inquiry but also satisfies humanity’s enduring curiosity about the universe we inhabit.