A new study compares predictions from a large cosmological simulation to recent pulsar‑timing array measurements of a low‑frequency gravitational‑wave background (GWB) and finds no statistically significant mismatch. The authors built a statistical framework to measure how much theoretical predictions and data disagree. Applied to the FABLE cosmological simulation and the NANOGrav 15‑year dataset, their fiducial model gives tensions of about 1σ to 2.5σ, which is not strong evidence of disagreement.

Pulsar‑timing arrays (PTAs) search for correlated timing shifts from many millisecond pulsars and have reported a common signal whose angular pattern matches the expected quadrupolar form (the Hellings–Downs curve). That signal likely comes from many inspiralling supermassive black‑hole binaries (SMBHBs) emitting at nanohertz frequencies. Earlier theoretical models often predicted a slightly lower GWB amplitude than some PTA inferences, so the authors set out to quantify that difference carefully and to test how model choices affect the comparison.

The team used the FABLE cosmological hydrodynamical simulation to extract a population of merging supermassive black holes and to generate the expected GWB using the Holodeck tool. FABLE places black holes in dark matter haloes above 5×10^10 h^−1 solar masses with seed masses of 10^5 h^−1 solar masses. The simulation follows black‑hole growth by mergers and by gas accretion modeled with an Eddington‑limited Bondi‑Hoyle‑Lyttleton–like rate. The authors also investigated key astrophysical ingredients that shape the GWB, such as how quickly binaries shrink (the “hardening” processes) and the distribution of mass ratios in mergers.

They show two important practical points. First, assuming a simple power‑law model for the GWB spectrum in observational inference can bias estimates of tension and may overstate discrepancies with theory. Second, physically motivated changes to the simulated black‑hole population can raise the predicted GWB amplitude. For example, boosting black‑hole masses at high redshift or favoring more equal‑mass mergers increases the signal and brings predictions closer to the PTA measurements.