A research team is using astrophysical explosions to understand the mysterious forces at work in some of the smallest building blocks in nature: atomic nuclei. In new research published in Nature Communications, the team uses machine learning and artificial intelligence to decipher the data from astrophysical observations to better understand how neutrons and protons interact in dense matter at the quantum level.
“This research represents the first time in the field that we’ve been able to robustly connect the macroscopic and microscopic realms and infer the interactions among neutrons and protons directly from astrophysical data,” said Ingo Tews, Los Alamos physicist. “Using artificial intelligence and machine learning, our framework made it possible to take data from remarkable astrophysical phenomena and infer the complicated physics of nuclear forces.”
The researchers, a team including scientists at the Technical University of Darmstadt, in Germany, used data from the 2017 detection of gravitational waves from a binary neutron star merger, as well as data from a telescope that studies neutron stars and their X-ray emissions. Their work uses machine learning to enable key constraints on nuclear couplings that describe the strength of nuclear forces.
“Our approach opens a new window into the strong-force physics of neutrons and protons and its effects on neutron stars,” said Isak Svensson, scientist at the Technical University of Darmstadt and a co-lead author. “Our framework allows us to go from neutron star observations to the interactions in dense matter.”
AI connects physics large and small
Taking many models of interacting neutrons and applying them to incredibly dense neutron stars would be “computationally intractable”; solutions to one model alone could run for hours on thousands of CPU cores. Seeking a faster and more readily available method, the research team built an AI framework that could connect the nuclear interaction to neutron star properties almost instantaneously.
One machine learning algorithm the team used employs an understanding of underlying quantum physics to arrive at a fast solution for dense-matter properties. The second algorithm, a neural network trained on large amounts of data, connects dense matter to properties of neutron stars. Hoping to predict neutron star properties such as size and tidal deformations, the algorithms act as surrogates for more complex, high-fidelity calculations.
“The tools we developed performed remarkably well — much better than we anticipated,” said Rahul Somasundaram, Los Alamos scientist and a co-lead author. “For astrophysical data from recent events, our framework offers constraints that are consistent with what we know from terrestrial experiments, albeit with larger uncertainties. For future observations by next-generation detectors, such as Cosmic Explorer, our approach will provide even better constraints that will be really powerful.”
The strong force at neutron star densities
The interactions among neutrons are driven by the strong force, one of the four fundamental forces of the universe (along with electromagnetism, the weak force and gravity). The strong force binds quarks and gluons to nucleons, such as neutrons and protons, and nucleons together in a nucleus. It is a remaining challenge in physics to build a robust quantum description of this powerful force.
Neutron stars are some of the densest objects in the universe — so dense that they can be around twice the mass of the sun despite being as small as 24 kilometers in diameter. Matter at such densities exhibits properties similar to the properties of matter in the center of atomic nuclei and has to be described by modeling interactions among nucleons at the quantum level; that is, the interactions among the dense neutrons determine the properties of the whole neutron star.
In connecting neutron star properties with the quantum mechanical properties of neutrons, the team is building a way to eventually elucidate properties of the strong force at the largest densities explored anywhere in the cosmos. This might also help scientists place constraints on exotic forms of matter, such as phase transitions to quarks and gluons.
The team’s insight was especially useful for learning about three-body forces, one of the least understood aspects of nuclear interactions. Three-body forces only appear when three or more neutrons or protons are close together.
Gravitational waves and X-rays
The team used data from the 2017 merger of two neutron stars, in which gravitational waves — ripples in the fabric of spacetime resulting from the collision — were observed by the Laser Interferometer Gravitational-Wave Observatory (LIGO). That event, named GW170817, revealed the tidal deformation that occurs when two neutron stars approach each other. The team also leveraged data from NASA’s Neutron star Interior Composition Explorer (NICER), a telescope that collects X-ray data from rapidly rotating neutron stars and uses the phenomenon of light bending in gravitational fields to extract a neutron star’s mass and radius.
Drawing from several sources and types of signals in this way is called “multimessenger” astronomy. The research approach that the team developed can be directly applied as new facilities come online. Several larger-scale, next-generation detectors, including the Einstein Telescope in Europe and Cosmic Explorer in the United States, are in the planning stages.
Paper: “Inferring three-neutron couplings from multi-messenger neutron-star observations.” Nature Communications. DOI: 10.1038/s41467-025-64756-6
Funding: The work was supported by the National Science Foundation, by the Laboratory Directed Research and Development program at Los Alamos, by the U.S. Department of Energy Office of Science’s offices of Nuclear Physics; Workforce Development for Scientists and Teachers; and Advanced Scientific Computing Research. The European Research Council and the German Research Foundation also supported the work.
LA-UR-25-31936
Contact
Public Affairs | media_relations@lanl.gov
