Gordon Bell Prize Finalists
Supercomputers and Researchers at Argonne Help Four Projects Become Gordon Bell Prize Finalists
Innovative research advances the fields of quantum materials science, cosmology, and weather forecasting.
The Association of Computing Machinery’s (ACM) famous 2025 ACM Gordon Bell Prize has selected four research teams as finalists, with support from the DOE’s Argonne National Laboratory. Moreover, one of these initiatives is vying for the Gordon Bell Special Prize. Outstanding accomplishments in the use of high-speed computing are recognized with this award.
Among the exascale supercomputers used in the nominated initiatives were Aurora at the Argonne Leadership Computing Facility (ALCF) and Frontier at the Oak Ridge Leadership Computing Facility (OLCF). The DOE Office of Science is home to the ALCF and OLCF. From modelling the universe and quantum materials to creating sophisticated weather forecasting models, these investigations leveraged exascale capability to push the boundaries of simulation and Artificial Intelligence (AI) for science.
You can also read IBM Dynamic Circuits: The Utility-Scale Quantum Efficiency
During the Supercomputing Conference (SC25), which will be held in St. Louis from November 16 to 21, the award winners will be revealed.
Highlights from the Projects of Finalists
AI-Powered Exascale Forecasting
In “AERIS: Argonne Earth Systems Model for Reliable and Skilful Predictions,” a billion-parameter artificial intelligence model is used to provide high-resolution forecasts that span hours to months in the future.
- Performance: The AERIS model was developed and trained on Aurora, achieving mixed-precision performance of over 11 exaflops and generating stable forecasts for up to 90 days.
- Methodology: Sequence Window Parallelism (SWiPe) is a technique that the team created. This method reduces communication across Aurora’s more than 60,000 GPUs while effectively distributing the model’s compute tasks and data among them.
- Impact: AERIS showed promise by surpassing traditional models, demonstrating how AI models may greatly enhance subseasonal-to-seasonal forecasting.
Argonne Contributors: Venkatram Vishwanath, the lead for AI and machine learning, Rao Kotamarthi, the science director for the Centre for Resilience and Decision Science, and computational scientists Eugene Yu, Jason Stock, Murali Emani, Sam Foreman, Chunyong Jung, Sandeep Madireddy, Varuni Sastry, Sam Wheeler, Huihuo Zheng, Troy Arcomano, and Venkatram Vishwanath were among the principal authors from Argonne.
You can also read NASA & JPL Are Advancing Quantum Sensors for Space Science
Creating a Trillion-Particle Universe Model
The greatest full-sky cosmological simulation to date was run on the Frontier supercomputer by a team led by Argonne. “Cosmological Hydrodynamics at Exascale: A Trillion-Particle Leap in Capability” is the title of the research.
- Scale: The simulation, which used the CRK-HACC (Conservative Reproducing Kernel Hardware/Hybrid Accelerated Cosmology Code), was more than 15 times larger than earlier attempts and modelled cosmic evolution with over 4 trillion particles. It effectively recorded the behaviour of gas and stars over a volume similar to contemporary sky surveys, as well as gravity effects.
- Performance: Within just over a week, the simulation produced over 100 petabytes of data and reached over 500 petaflops of peak performance.
- Impact: These findings set a new benchmark for fidelity and performance in universe modelling, providing a framework for next studies on dark matter, dark energy, and the universe’s evolution.
- Argonne Contributors: Salman Habib, the Division Director and Argonne Distinguished Fellow, Nicholas Frontiere, a computational scientist with expertise in numerical hydrodynamics, Katrin Heitmann, the Deputy Division Director and computational scientist, J.D. Emberson, Michael Buehlmann, Esteban Rangel, Patricia Larsen, Vitali Morozov, and Adrian Pope were among the authors.
You can also read QHDC Achieves Milestone On 156-Qubit Quantum Processor
Trillion-Atom-Scale Simulation of Light-Matter Interactions
Large-scale simulations of light interacting with quantum materials were carried out using Aurora by a team headed by the University of Southern California (USC).
- Scale and Performance: 1.87 exaflops was the team’s continuous performance. Modelling systems with over a trillion atoms using their new Multiscale Light-Matter Dynamics framework, they produced the largest simulations of their kind to date.
- Methodology and Impact: This study showed the potential of integrating AI with physics simulations. Allegro-FM is a foundation model that allows researchers to simulate atomic behaviour throughout the periodic table. By better understanding how light may change quantum materials, the results could help designers create ultrafast, low-power electronic devices.
- A computational scientist with expertise in theoretical and computational condensed matter physics, Ye Luo, contributed to this study as an Argon contributor.
You can also read IonQ Acquires Skyloom Global To Build Quantum Networking
Progress in Quantum Materials Models
Another USC team advanced the state of the art in quantum materials simulations by utilizing the Perlmutter system at DOE’s Lawrence Berkeley National Laboratory, Aurora, and Frontier.
- Scope: The team ran dynamic simulations that accurately trace the interactions between electrons and nuclei over tens of thousands of atoms by enhancing the open-source BerkeleyGW program. G, which measures an electron’s speed through a material, and W, which measures how electrons affect one another, are represented by the “GW” notation.
- Portability and Performance: Their simulations attained 0.7 exaflops on Aurora and nearly 1 exaflop on Frontier. Successfully operating on several supercomputers showed how adaptable their techniques were to various hardware configurations.
- Impact: By offering strong new instruments, this project advances the development of upcoming quantum technology.
With the help of DOE’s Office of Science, Advanced Scientific Computing Research (ASCR) program, the Argonne Leadership Computing Facility (ALCF) offers supercomputing capabilities to the scientific community to promote basic research and knowledge in a variety of fields. In practically every scientific field, Argonne National Laboratory itself carries out cutting-edge basic and applied research.
You can also read Infleqtion Partners With ORNL For Quantum HPC Integration
