A 25-year muon puzzle closes and a collider puzzle opens

Lattice calculations now match Fermilab's muon measurement, leaving four decades of collider data unexplained.

For 25 years the muon's wobble in a magnetic field sat about one part in a million away from prediction — a gap wide enough that physicists wondered whether unknown particles were nudging it. Brookhaven measured the excess in 2001. Its 50-foot magnetic ring was hauled by barge and truck to Fermilab in 2013 for a more precise run, which reported in April 2021.

The prediction, not the measurement, turned out to be the moving part. One camp inferred the strong force's contribution from electron-positron collision data; its June 2020 forecast clashed with Fermilab hard enough to nearly clear the threshold for claiming a discovery. The other camp, the BMW collaboration spanning Budapest, Marseille and Wuppertal, spent from 2014 building lattice QCD simulations — a grid standing in for every quark interaction — and published in Nature the same day Fermilab's result appeared. Their number matched the experiment, the updated calculations agreeing to one part in 100 billion, and independent lattice groups have since concurred. Many physicists now consider the anomaly closed, with no new particles required.

What made that possible was compute. A calculation described as unimaginable a decade ago became tractable once the grids could be made both large and fine at the same time — a clean picture of how falling computation costs convert open questions into settled ones.

The settlement moved the problem rather than removing it. The VEPP-2000 collider in Novosibirsk installed a new detector in 2010 and published a significantly different pion production rate in 2023. Lattice results and preliminary data from the collider's second detector line up with that new rate, while a 2023 reanalysis of BABAR data backs the old one. Four decades of earlier measurements point the other way.

Source: Quanta