The authors used very large classical field simulations to estimate the mass of the axion if it makes all of dark matter in a simple “post‑inflation” scenario. In this scenario a Peccei‑Quinn (PQ) symmetry — a proposed symmetry whose breaking produces axions — is broken after cosmic inflation. That breaking creates a network of cosmic strings that later become attached to domain walls at the QCD epoch and then annihilate, releasing axions as radiation.

The team ran simulations on 12,288^3 grids of a minimal field model: one complex scalar field that breaks the PQ U(1) symmetry at a scale f_a of order 10^11 GeV. They tracked the string network, the formation of domain walls at the QCD transition, and the final collapse that produces axions. From these runs they measured the comoving axion number density after collapse and report it as 9.48(23) f_a^2 H_*, where H_* is the Hubble rate at the time the axion mass equals H. ‘‘Comoving’’ means the density is scaled to remove the effect of the universe’s expansion.

Turning that measured number density into a particle mass requires input from lattice QCD calculations of how the axion mass turns on with temperature. Using the commonly quoted lattice results for the topological susceptibility, the authors obtain a dark‑matter axion mass of 17.46(84) μeV. That value corresponds to a symmetry‑breaking scale f_a = 3.27(15) × 10^11 GeV and to a resonant frequency for cavity haloscope searches of 4.22(20) GHz.

This result matters because it gives a concrete target for experiments that search for axion dark matter. The authors also say their estimate is lower and more precise than several recent predictions. They emphasize that the numerical simulations themselves are in good agreement across groups; the main differences in mass estimates come from how groups extrapolate the simulations to cosmological scales and from different assumptions about the long‑term evolution of the string networks.