Muons (elementary subatomic particles, similar to electrons) are the center of a mystery that implicates our very understanding of the universe. Decades ago, muon measurements taken at Brookhaven National Laboratory produced a troubling disagreement between the Standard Model (the commonly understood physical foundations of the universe) and real-world measurements. In the intervening years, researchers have failed to conclusively resolve the discrepancy. Now, a collaborative research team has returned to the muon experiment, leveraging Argonne National Laboratory’s Mira supercomputer in an attempt to pin down the elusive physical mystery behind the behavior of the muon. 

The culprit is the muon’s “magnetic moment” – how it moves when it interacts with a magnetic field. In theory, the muon’s magnetic moment is supposed to be fairly large, but in the Brookhaven experiment, the moment was negligible. “If you account for uncertainties in both the calculations and the measurements, we can’t tell if this is a real discrepancy or just a statistical fluctuation,” said Thomas Blum, a physicist at the University of Connecticut and co-author of the paper, in an interview with Brookhaven’s Christina Nunez. “So both experimentalists and theorists are trying to improve the sharpness of their results.”

The researchers focused on the strong force affecting the muons (as distinct from weak, electromagnetic or gravitational forces), which produces substantial uncertainties in muon analysis through “hadronic contributions.” The team tackled these uncertainties by applying a theory called quantum chromodynamics (QCD). 

“To do the calculation, we simulate the quantum field in a small cubic box that contains the light-by-light scattering process we are interested in,” said Luchang Jin, also a co-author and University of Connecticut physicist. “We can easily end up with millions of points in time and space in the simulation.”

To process these millions of points, the researchers turned to Mira, Argonne’s IBM supercomputer equipped with BlueGene/Q Power 16C 1.6GHz processors. Mira, which was rated at 8.6 Linpack petaflops, was decommissioned at the end of 2019. “Mira was ideally suited for this work,” said James Osborn, a computational scientist with Argonne’s Computational Science division. “With nearly 50,000 nodes connected by a very fast network, our massively parallel system enabled the team to run large simulations very efficiently.”

This work went on for four years, after which, at long last, the team produced the first-ever result for that difficult light-by-light scattering process. Alas, it was not the result they were hoping for. 

“For a long time, many people thought this contribution, because it was so challenging, would explain the discrepancy,” Blum said. “But we found previous estimates were not far off, and that the real value cannot explain the discrepancy. As far as we know, the discrepancy still stands. We are waiting to see whether the results together point to new physics, or whether the current Standard Model is still the best theory we have to explain nature.”

Of course, the researchers working on the muon problem know that they’re in it for the long haul. Already, work is underway at Fermi National Accelerator Laboratory to reduce experimental uncertainty by a factor of four.

“Physicists have been trying to understand the anomalous magnetic moment of the muon by comparing precise theoretical calculations and accurate experiments since the 1940s,” said Taku Izubuchi, a physicist at Brookhaven who co-authored the paper. “This sequence of work has led to many discoveries in particle physics and continues to expand the limits of our knowledge and capabilities in both theory and experiment.”

Header image: An illustration of the hadronic light-by-light scattering process with Mira in the background. Image courtesy of Luchang Jin.