Collaborative Research: MATH-DT: Mathematical Foundations of AI-assisted Digital Twins for High Power Laser Science and Engineering

Laser technology is one of the most transformative inventions of the modern era, which has become an indispensable tool for scientific research and technological innovation - revolutionizing the semiconductor industry, telecommunications, healthcare, and defense. However, current laser design and manufacturing approaches remain stagnant, stymieing further breakthroughs. Developing novel integrated systems of laser architectures, components, and techniques leveraging digital twins (DT) is imperative to expand frontiers in intensity, wavelength regime, and high average power. This project will fill this gap using state-of-the-art predictive and generative artificial intelligence (AI) coupled with physical principles and high-fidelity, close-loop, rapid feedback between digital models and physical systems. Graduate students and postdoctoral researchers will also be integrated within the research team as part of the training of the next generation of scientists required to advance the field. This project will develop theoretical foundations for AI-assisted DTs to integrate scientific data, physical models, and machine learning for complex high-power laser science and engineering (HPLSE) to enable efficient design, failure and performance prediction, operational optimization, and emerging lasing conditions. Laser technologies are extremely complex to model because they rely on a cascaded set of mode-locked laser dynamics and a manifold of architectures and configurations of chirped pulse amplification, and nonlinear optical stages, such as parametric amplification. Their architectural complexity and multi-dimensional data far exceed current modeling and analysis tools. The project will address these challenges by (1) extracting reduced representation of scientific data from experiments or high-fidelity HPLSE simulation, (2) building data-efficient and physics-aware predictive machine learning surrogate models of laser fields with uncertainty quantification, and (3) developing generative model-based rapid closed-loop control between digital models and physical high-power laser systems. The project will be AI-focused, multi-disciplinary, and involve a diverse workforce of future scientists and engineers. The project will also include an education thrust to integrate the research results into interdisciplinary education. The project will bolster AI foundations and its application curricula at both UCLA and the University of Utah. More critically, it will forge a robust collaboration among mathematics, data science, and laser technologies. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.