Accelerating Science Through Artificial Intelligence

Texas A&M University nuclear physicist Dr. Jonas Karthein has always enjoyed building things in the laboratory, a passion he traces back to his childhood spent with LEGO sets and two engineer parents as his constants growing up in Germany. But when it comes to shaping his eventual career destiny, he credits a transatlantic flight and a chance seating assignment.

“In 10th grade, I was flying to Oregon for a school exchange and ended up seated next to Dr. Klaus Blaum, director at the Max Planck Institute for Nuclear Physics (MPIK),” Karthein said. “We talked for most of the flight about his research, and he invited me to intern at his institute when I returned home. I'd never seen anything like it — particle accelerators, storage rings, precision ion traps and lasers, all built to probe the fundamental constituents of the universe.”

Karthein admits he likely would have ended up in mechanical engineering otherwise, save for that single conversation that set his entire trajectory: undergraduate at Heidelberg University and MPIK, and then his master's and Ph.D. at Blaum’s CERN setup in Geneva, Switzerland.

After completing a Humboldt Postdoctoral Research Fellowship at MIT in 2024, Karthein moved to Texas A&M, where he is building his own research legacy focused on quantum sensing with nuclei. As an assistant professor in the Department of Physics and Astronomy and a member of the Cyclotron Institute, he is studying radioactive atoms and molecules using the Cyclotron Institute’s two particle accelerators to gain insight into fundamental nuclear interactions and properties.

Karthein and his independent research program received a big boost this week, securing one of the nearly 300 Genesis Mission Phase I Request For Applications (RFA) awards unveiled today by the U.S. Department of Energy (DOE) at the Genesis Mission Summit in Washington, D.C. Karthein’s project is one of four led by Texas A&M researchers and one of 168 led by universities.

The Genesis Mission is a historic national initiative led by the DOE, which is building the world’s most powerful integrated science discovery platform. By uniting government, industry, academia and philanthropy, it is accelerating breakthroughs in energy, scientific discovery and national security through a new platform that combines AI, supercomputing, quantum systems and advanced scientific instruments.

The goal of the Phase I RFA awards is to identify promising pathways toward transformative scientific capabilities and establish a foundation for future investment and scale. Project teams will design and demonstrate research workflows that integrate AI with scientific investigation, while rigorously evaluating whether those approaches can accelerate discovery, improve predictive capabilities, enhance experimentation or generate new scientific insights.

"America has no shortage of bold ideas or talented scientists, and the response to the Genesis Mission proves that," said U.S. Secretary of Energy Chris Wright. "The 278 projects selected today represent the very best of our nation’s scientific enterprise. The remarkable number of high-quality proposals we received demonstrates that America’s innovation pipeline is strong, and it points to even greater opportunities for future investment and continued expansion of the Genesis Mission portfolio."

Digital Twins To The 'Data-Scarce' Rescue

Karthein’s project, "Scalable Agentic Digital Twins for Autonomous Precision Facilities," will be led by the Texas A&M Cyclotron Institute in tandem with partners at Argonne National Laboratory, MIT and the Facility for Rare Isotope Beams (FRIB) at Michigan State University, along with California-based Gravistar USA startup as an unfunded industry partner. Karthein and his collaborators — including fellow Texas A&M physicist and Cyclotron Institute member Dr. Dan Melconian and Texas A&M statistician Dr. Abhishek Roy — will work to create an AI "digital twin," a fast, living computer model of actual experiments that functions on two key levels. First, it learns the physics of the apparatus well enough to find the best settings automatically, in seconds rather than hours. Second, an AI assistant reads the machine's sensors, logbooks and manuals, and — always under human supervision — suggests adjustments, flags problems early and clearly communicates its level of confidence in each of its recommendations in real-time.

The project addresses the DOE challenge area Enhancing Particle Accelerators for Discovery in the focus area AI-driven Accelerator Facilities.

"Artificial intelligence has already transformed large scientific facilities that produce enormous amounts of data,” Karthein said. "But these precision nuclear experiments are the opposite: they are 'data-scarce,' running only a few hundred experimental campaigns over an entire lifetime with a handful of datasets per campaign, not millions per day. Because standard, data-hungry AI simply doesn't work in this setting, this corner of science has been left behind."

Karthein notes that some of the most important experiments in nuclear science study exotic forms of atoms — rare isotopes that exist for only fractions of a second and reveal how the elements of our universe were forged in stars and stellar explosions. These experiments are extraordinarily delicate, considering that a single machine may have well over 20 experimental parameters — voltages, pressures, laser tunings — that are all subject to drifts. Ensuring that each of those settings is accurate depends heavily on the hard-won intuition of a handful of senior experts. Beamtime is scarce and expensive (ranging from many hundreds of dollars to more than $10,000 per hour), and hours can be lost to manual fine-tuning before real data can even be collected.

By slashing setup time and catching problems before they ruin a run, Karthein says the system frees precious beamtime to reach physics that is currently out of reach: the extremely neutron-rich and proton-rich nuclei that hold clues to how heavy elements are created in the cosmos. In effect, he adds, this project will enable scientists to make previously inaccessible discoveries with the same facilities.

Inspired by that futuristic vision, Karthein and his Genesis Mission project teammates will spend the next nine months executing their digital twin proof-of-concept using two university-scale testbeds — Karthein’s ion trap setup at Texas A&M and his former postdoctoral advisor and co-investigator Dr. Ronald Garcia Ruiz’s laser spectroscopy setup at MIT — aimed at scaling up for a possible Phase II. Both testbeds are exact replicas of Phase II targets at Argonne National Lab, FRIB and Texas A&M, respectively led by co-investigators and physicists Dr. Bernhard Maass, Dr. Kei Minamisono and Melconian to enable direct algorithmic scaling in possible future stages of the project.

On the hardware side, Karthein will lead integration and ion-trap validation, while Garcia Ruiz will do the same for surrogate physics and laser validation. Conversely, Argonne National Lab co-investigators and computational mathematicians Dr. Ahmed Attia and Dr. Sven Leyffer will oversee the AI development and manage high-performance computer scaling, respectively, while working with Roy at Texas A&M to develop a virtual twin to the team’s hardware setups with a focus on error prediction, simulation and optimization.

A Career Forged By Connection

Professionally and personally, Karthein says he has been building to this moment since that pivotal in-flight conversation with Blaum as a teenager, the first of many personal connections leading to the next and ultimately to Texas A&M.

As a graduate student at CERN, Karthein taught himself to code, which turned into something he genuinely loved. He then wrote the first Python-based Bayesian analysis code for the mass spectrometer ISOLTRAP at CERN, replacing methods previously performed in Excel spreadsheets. He says that's where his interest in AI and machine learning took root — a curiosity-driven skill with direct relevance to his team’s current Genesis Mission project.

While at CERN, he also met his future postdoctoral advisor, Garcia Ruiz, who recruited him to MIT. At MIT, Garcia Ruiz was starting up his own new laser spectroscopy lab and provided Karthein with state-of-the-art training in laser spectroscopy as well as various opportunities to take on organizational roles in the new group. Karthein fondly recalls the time he spent with Garcia Ruiz that ultimately taught him what it takes to build a research group from scratch.

When it came time for faculty searches, Karthein says the Texas A&M Cyclotron Institute stood out immediately with a strong nuclear physics program based on its on-campus accelerators, a competitive startup package, large modern laboratory spaces for a new research direction and, just as important, staunch support of dual-career opportunities for households with two academics. As a result, Dr. Jamie M. Karthein recently joined the Department of Physics and Astronomy and Cyclotron Institute as a faculty member after completing her National Science Foundation-funded ASCENT Postdoctoral Fellowship at MIT.

Karthein says he and Garcia Ruiz will soon embark on another shared experience, courtesy of the Genesis Mission: hosting an event at each of their campuses to showcase the self-running demo of the team’s digital twin and the untapped potential that eclipses nuclear science.

For Karthein, the ultimate beauty of the team’s project is that it embodies the goal of DOE's Genesis Mission — using AI to dramatically increase the productivity of America's scientific infrastructure. In the same vein, he cites the AI assistant that captures the tacit expertise of veteran scientists, making it available to newcomers and thereby giving graduate students expert-level capability from day one while helping train the next generation of the nuclear science workforce. Because the same architecture applies to X-ray light sources, neutron sources, and fusion facilities that face similar tuning challenges, he says the benefits extend well beyond nuclear physics, given that the team plans to release its core tools as open-source software so that the entire scientific community also can build on them.

“If even half of the 15-plus U.S. precision nuclear physics facilities adopt this approach, it could recover more than $15 million per year in effective beamtime value — roughly 20 times the initial investment,” Karthein added.

See the complete list of Phase I Genesis Mission awards and learn more about the Karthein Lab at Texas A&M.