Computational & AI Breakthroughs

Across its national-security mission space, LANL relies on the development and application of innovative computational methods, including advances in high-performance computing, artificial intelligence (AI), and quantum information science. These capabilities are foundational for our efforts ranging from simulation of complex phenomena with fast, accurate physics; to the fusing of large, disparate datasets; to the development of advanced sensing to monitor the behavior of complex systems.

This Critical Outcome is focused on ensuring a vibrant capability by pushing the forefront of our existing platforms while exploiting developments in new platforms. This Critical Outcome supports three of the Strategic Objectives in accomplishing the Laboratory mission: Nuclear Deterrence, Trustworthy Operations, and Technical Leadership. This Critical Outcome also leverages the Signature Institutional Commitment in AI, which builds on our recent investment in the Venado platform.

In FY23, we achieved the performance target in acceptance testing for Crossroads; developed facilities plans for ATS-5 and ATS-7; deployed new data-centric computing infrastructure; and continued to have a vibrant portfolio of fundamental and applied R&D efforts in computation and data science.

For FY24, we will build on these successes through 11 Supporting Initiatives.

Supporting Initiatives

1. Develop and gain Laboratory approval for a decadal plan to ensure competitive, robust, and sustainable energy and facilities computing infrastructure to meet ATS-5 and ATS-7 deployment timeframes by the end of calendar year 2025.
2. Demonstrate computational performance gain over ATS-1 of a 3D integrated code on problems of interest to the weapons program on Crossroads/ATS-3 by the end of calendar year 2024.
3. Demonstrate the capabilities of the Laboratory’s modern multi-physics capabilities on El Capitan by 2025.
4. Demonstrate the unique capabilities of the Venado platform to conduct calculations ranging from integrated 3D multi-physics simulations to machine learning (ML) by the end of calendar year 2024.
5. Develop and document a strategy to achieve AI-assisted integrated weapons simulation workflows within a human learning cycle, aligned with the timeframe and technology path of the ATS-5 platform by the end of calendar year 2024.
6. Explore, develop, and incorporate high-fidelity computational exploration of narrative uncertainties into a programmatic mission-critical assessment by the end of calendar year 2026.
7. Develop a comprehensive data stewardship strategy across the portfolio of scientific data to enable effective improvement of user workflows and application of AI/ML techniques by the end of calendar year 2025.
8. Develop and apply computational methods that accelerate design for manufacturing by the end of calendar year 2025.
9. Accelerate fundamental and applied computational and data science R&D with AI/ML technologies.
10. Strengthen and diversify the talent pipeline supporting all these initiatives across the LANL information science and technology community via innovative partnerships and programs.
11. Develop and apply quantum, or hybrid quantum-classical, algorithms to security mission problems.