A new agreement between Los Alamos National Laboratory (LANL) and the National Science Foundation's Artificial Intelligence Institute for Advances in Optimization (AI4OPT) at Georgia Tech is set to propel research in applied artificial intelligence (AI) and engage students and professionals in this rapidly growing field.

"This collaboration will help develop new AI technologies for the next generation of scientific discovery and the design of complex systems and the control of engineered systems," said Russell Bent, scientist at Los Alamos. "At Los Alamos, we have a lot of interest in optimizing complex systems. We see an opportunity with AI to enhance system resilience and efficiency in the face of climate change, extreme events, and other challenges."

The agreement establishes a research and educational partnership focused on advancing AI tools for a next-generation power grid. Maintaining and optimizing the energy grid involves extensive computation, and AI-informed approaches, including modeling, could address power-grid issues more effectively.

AI Approaches to Optimization and Problem-Solving

Optimization involves finding solutions that utilize resources effectively and efficiently. This research partnership will leverage Georgia Tech's expertise to develop "trustworthy foundation models" that, by incorporating AI, reduce the vast computing resources needed for solving complex problems.

In energy grid systems, optimization involves quickly sorting through possibilities and resources to deliver immediate solutions during a power-distribution crisis. The research will develop "optimization proxies" that extend current methods by incorporating broader parameters such as generator limits, line ratings, and grid topologies. Training these proxies with AI for energy applications presents a significant research challenge.

The collaboration will also address problems related to LANL's diverse missions and applications. The team's research will advance pioneering efforts in graph-based, physics-informed machine learning to solve Laboratory mission problems.

Outreach and Training Opportunities

In January 2025, the Laboratory will host a Grid Science Winter School and Conference, featuring lectures from LANL scientists and academic partners on electrical grid methods and techniques. With Georgia Tech as a co-organizer, AI optimization for the energy grid will be a focal point of the event.

Since 2020, the Laboratory has been working with Georgia Tech on energy grid projects. AI4OPT, which includes several industrial and academic partners, aims to achieve breakthroughs by combining AI and mathematical optimization.

"The use-inspired research in AI4OPT addresses fundamental societal and technological challenges," said Pascal Van Hentenryck, AI4OPT director. "The energy grid is crucial to our daily lives. Our collaboration with Los Alamos advances a research mission and educational vision with significant impact for science and society."

The three-year agreement, funded through the Laboratory Directed Research and Development program's ArtIMis initiative, runs through 2027. It supports the Laboratory's commitment to advancing AI. Earl Lawrence is the project's principal investigator, with Diane Oyen and Emily Castleton joining Bent as co-principal investigators.

Bent, Castleton, Lawrence, and Oyen are also members of the AI Council at the Laboratory. The AI Council helps the Lab navigate the evolving AI landscape, build investment capacities, and forge industry and academic partnerships.

As highlighted in the Department of Energy's Frontiers in Artificial Intelligence for Science, Security, and Technology (FASST) initiative, AI technologies will significantly enhance the contributions of laboratories to national missions. This partnership with Georgia Tech through AI4OPT is a key step towards that future.