High-intensity and high-repetition lasers emit powerful bursts of light in rapid succession, capable of firing multiple times per second. Commercial fusion energy plants and advanced compact radiation sources are common examples of systems that rely on such laser systems. However, humans are a major limiting factor as the human response time is insufficient to manage such rapid-fire systems.
To address this challenge, scientists are looking at different ways to leverage the power of automation and artificial intelligence that have real-time monitoring capabilities for high-intensity operations.
A team of researchers from Lawrence Livermore National Laboratory (LLNL), Fraunhofer Institute for Laser Technology (ILT), and the Extreme Light Infrastructure (ELI ERIC) are conducting an experiment at the ELI Beamlines Facility in the Czech Republic to optimize high-power lasers using machine learning (ML).
The researchers trained an ML code developed by LLNL’s Cognitive Simulation on laser-target interaction data allowing researchers to make adjustments as the experiment progresses. The output is fed back into the ML optimizer to allow it to fine-tune the pulse shape in real time.
The laser experiments were conducted for three weeks, with each experiment lasting around 12 hours, during which the laser shot 500 times, at 5-second intervals. After every 120 shots, the laser was stopped to replace the copper target foil and to inspect the vaporized targets.
"Our goal was to demonstrate robust diagnosis of laser-accelerated ions and electrons from solid targets at a high intensity and repetition rate," said LLNL’s Matthew Hill, the lead researcher. "Supported by rapid feedback from a machine-learning optimization algorithm to the laser front end, it was possible to maximize the total ion yield of the system."
Using the power of the state-of-the-art High-Repetition-Rate Advanced Petawatt Laser System (L3-HAPLS) and innovative ML techniques, the researchers have made significant progress in understanding the complex physics of laser-plasma interactions.
Until now researchers have relied on more traditional scientific methods, which required manual intervention and adjustments. With the ML capabilities, scientists have been able to analyze vast datasets with greater accuracy and make real-time adjustments as the experiment ran.
(NicoElNino/Shutterstock)
The success of the experiment also highlights the capabilities of the L3-HAPLS, one of the most powerful and fastest high-intensity laser systems in the world. The experiment demonstrated L3-HAPLS’s excellent performance repeatability, focal spot quality, and extremely stable alignment.
Hill and his LLNL team spent about a year preparing for the experiment in collaboration with the Fraunhofer ILT and ELI Beamlines teams. The Livermore team used several new instruments developed by the Laboratory Directed Research and Development Program, including a rep-rated scintillator imaging system and a REPPS magnetic spectrometer.
The lengthy preparation has paid off as the experiment has been successful in generating robust data that can serve as the foundation for advancements in various fields including fusion energy, material science, and medical therapy.
GenAI technology has been at the forefront of scientific innovation and discovery. It is helping researchers push the boundaries of what is scientifically possible. Last week, researchers from MIT and the University of Basel in Switzerland developed a new machine-learning framework to uncover new insights about materials science. Last week, AI proved to be highly instrumental in drug discovery.
Related Items
The Future of AI in Science
NVIDIA Powers AI Supercomputing, Scientific Innovations, and HPC Research
2024 GenAI Predictions
