Argonne Researchers Advance New Tech Through Re-Envisioned SciDAC Institutes
Researchers will develop advanced tools to address complex problems on leadership-class computing systems and assist application scientists in implementation.
May 14, 2026 — The changing landscape of scientific computing has prompted a new direction for two U.S. Department of Energy (DOE) Scientific Discovery through Advanced Computing (SciDAC) Institutes — and Argonne researchers are helping to lead the way.
The SciDAC program has supported research in mathematics and computer science for 25 years. As part of that effort, the RAPIDS and FASTMath institutes have helped researchers develop advanced algorithms, software and mathematical tools for faster and more capable scientific campaigns using DOE supercomputers like Aurora.
Recent funding from DOE supports new research directions in both institutes, reflecting both the success of the institutes and the changing landscape of scientific computing. In particular, it highlights the growing role of artificial intelligence (AI) in mathematics, computer science and domain science applications and the important role RAPIDS and FASTMath continue to play in helping domain scientists integrate AI into how they work.
Complementary Institutes with Shared Goals
The two new institutes share several important features.
Both bring together researchers from universities and national laboratories. Both work closely with application developers to study complex physical phenomena. And both include strong outreach programs, offering workshops and tutorials to help train the next generation of scientists.
At the same time, the institutes take different but complementary research approaches.
FASTMath focuses on developing and deploying robust mathematical methods and numerical algorithms. RAPIDS works with domain scientists to apply AI techniques in their work, improve how scientists manage and extract knowledge from data, and maximize the utility of DOE’s powerful supercomputers in their science. Together, the institutes help enable new scientific discoveries.
RAPIDS: Helping Application Teams Manage Complex Workflows
The RAPIDS3 Institute for Computer Science, Data, and Artificial Intelligencefocuses on helping scientists address challenges in leveraging next-generation computing systems, managing and understanding the massive volumes of data produced by modern simulations and experiments, and accelerating discovery through thoughtful use of AI technologies.
Extreme-scale supercomputers, advanced experimental facilities and emerging software technologies create new opportunities for scientific discovery. At the same time, new technical challenges arise in making use of these cutting-edge resources.
For example, how can scientific software get the best “bang for the buck” in terms of the energy it consumes? What tools can manage the massive data volumes produced by simulations and experiments and help us glean new insights? And how can artificial intelligence be used in complex scientific endeavors to help accelerate scientific discovery in ways we can trust?
“This moment in time has us simultaneously adopting exascale supercomputing systems, exploiting the massive data volumes generated by new high-fidelity experimental devices and exploring the potential of AI for science,” said Robert Ross, a senior scientist at Argonne and director of the RAPIDS Institute. “Our goal is to help researchers overcome these challenges, including taking full advantage of DOE’s world-class computing platforms, like Aurora.”
To address these issues, the RAPIDS team is pursuing four research areas:
- Enhancing the performance, energy efficiency, portability and productivity of scientific codes.
- Lowering data access overheads, orchestrating data movement in complex workflows and providing insightful views of scientific data.
- Accelerating science through physics-aware surrogates, innovative foundation models, agentic AI and partnerships with industry.
- Advancing, productizing and integrating key software packages to ensure their availability and quality to the SciDAC community.
Community engagement is also an important part of RAPIDS.
“We work directly with DOE scientists and facilities to deploy these technologies on current and emerging supercomputing platforms,” Ross said.
Other Argonne staff involved in RAPIDS are Akash Dhruv, Paul Hovland, Jan Hückelheim, Rob Latham, Youngjun Lee, Sandeep Madireddy, Tanwi Mallick, Swann Perarnau, Tom Peterka, Krishnan Raghavan, Orcun Yildiz and Xingfu Wu.
FASTMath: Providing Scalable Algorithms and Software for Scientific Simulations
The FASTMath (Frameworks, Algorithms and Scalable Technologies for Mathematics) Institute, led by DOE’s Lawrence Livermore National Laboratory, focuses on developing robust mathematical techniques and energy-efficient software for advanced computing architectures at major DOE facilities.
Working with application developers is a central part of FASTMath’s mission.
“We collaborate closely with domain scientists,” said Todd Munson, an Argonne senior scientist and deputy director of the FASTMath Institute. “This helps us apply our expertise in mathematics and scientific AI/ML to large-scale modeling and simulation codes.”
Munson will also help coordinate activities with SciDAC application partnerships. There are currently more than 30 partnerships spanning eight DOE program offices. These multiyear collaborations bring together domain scientists, applied mathematicians and computer scientists to address large-scale scientific challenges important to DOE missions. Current Argonne projects include cosmology and nuclear imaging at the exascale.
FASTMath researchers have already demonstrated practical results. For example, computer scientists developed a method for identifying optimal placements of blocking devices that can help protect electrical grids. They also created a strategy that helps researchers identify materials for designing electromagnetic cloaks more quickly.
“Our researchers have had enormous success,” said Jeffrey Larson, the Argonne institutional lead for FASTMath. “For example, we’ve developed numerical algorithms for optimization and structured meshes. These methods are now part of the PETSc suite of scalable parallel solvers and the libEnsemble toolkit for running large collections of related simulations. Both are widely used by application developers.”
Other Argonne staff on the FASTMath team include Ahmed Attia, Emil Constantinescu, Stephen Hudson, Sven Leyffer, Matt Menickelly, Richard Tran Mills, Jean-Luke Navarro, Hong Zhang and Junchao. Zhang.
The Next Wave of Scientific Discovery
FASTMath has been supported by the SciDAC program since 2011 and RAPIDS since 2017.
“DOE’s continued support ensures that scientists have state-of-the-art mathematical tools and computational software at their disposal on current and next-generation supercomputing facilities, along with the expertise to make best use of these tools,” Ross said. “This support will help drive the next wave of scientific discovery.”
SciDAC funding comes from the DOE Office of Science’s Advanced Scientific Computing Research program.
Source: Gail Pieper, Argonne National Laboratory
Related
Yann LeCun’s AMI Secures $1B Seed to Develop AI World Models
Turing Award–winning AI researcher Yann LeCun has spent years arguing that large language models, at…
Deloitte’s State of AI 2026: Why Enterprise Execution Is Falling Behind Adoption
Deloitte’s latest State of the AI report shows that AI adoption continues to accelerate rapidly,…
Anthropic Unveils ‘Project Glasswing’ as Claude Mythos Targets Software Vulnerabilities
April 9, 2026 — Anthropic has announced Project Glasswing, a new initiative that brings together Amazon…
OpenAI Shutters Sora, Shifts Business Strategy Ahead of IPO
Back in 2022, OpenAI set off a chain reaction in the tech world when it…
Gartner Forecasts 90% Drop in LLM Inference Costs by 2030
STAMFORD, Conn., March 25, 2026 — By 2030, performing inference on a large language model (LLM)…
Google Unveils Gemini Enterprise Agent Platform, Expands Vertex AI into Full Agent Stack
At Google Cloud Next 2026 in Las Vegas this week, the company announced major changes…
FAU and P&G Host dAIper Hackathon Focused on AI and Computer Vision
May 15, 2026 — The dAIper Hackathon, held from April 24 to 26, 2026, was a…
Accenture Federal Services Expands OpenAI Collaboration for Federal AI Adoption
ARLINGTON, Va., May 14, 2026 — Accenture Federal Services and OpenAI today announced a strategic collaboration…
Argonne Researchers Advance New Tech Through Re-Envisioned SciDAC Institutes
Researchers will develop advanced tools to address complex problems on leadership-class computing systems and assist…
LLNL’s ‘STEM with Phones’ Program Brings AI-Powered Physics Research to Students
May 14, 2026 — Forget spreadsheets. At Lawrence Livermore National Laboratory’s (LLNL) STEM with Phones student workshop,…
Graphon AI Emerges from Stealth with $8.3M to Build ‘Pre-Model’ Intelligence Layer for Enterprise AI
SAN FRANCISCO, May 14, 2026 — Graphon AI emerged from stealth today with $8.3 million in…
Equinix Puts Enterprises in Control of Data Sovereignty Across Hybrid Multicloud Environments
REDWOOD CITY, Calif., May 14, 2026 — Equinix, Inc. today announced the global expansion of…
