Gravitational waves could settle the disagreement over the universe’s expansion

Scientists made the first reasonably accurate measurement of the Hubble constant H₀, a characterization of the universe’s expansion rate, in the early 2000s. Since then, increasingly precise measurements (see the article by Mario Livio and Adam Riess, Physics Today, October 2013, page 41 ) have uncovered a conflict: Calculations based on the combination of cosmic microwave background (CMB) measurements and the standard cosmological model give a value of about 67 km s^-1 Mpc⁻¹, whereas so-called distance ladder measurements derived from Cepheid variable stars and type 1a supernovae yield a larger value—about 73 km s⁻¹ Mpc⁻¹—that is independent of the cosmological model. Attempts to reconcile the two values using new physics, improved modeling, or identification of systematic errors have come up short. Now a multi-institute collaboration led by Stephen Feeney of the Center for Computational Astrophysics at the Flatiron Institute has proposed a way to resolve the conflict once and for all.

The researchers consider a third measurement source: gravitational waves from binary neutron star mergers (see Physics Today, December 2017, page 19 ). The technique was first proposed in the 1980s and is becoming viable now following the successes of the Laser Interferometer Gravitational-Wave Observatory (LIGO) and Virgo (see the article by Daniel Holz, Scott Hughes, and Bernard Schutz, Physics Today, December 2018, page 34 ). The collaboration has only found one such merger, though, so to test their technique Feeney and coworkers simulated 51 binary neutron star mergers and the gravitational-wave signals they would generate.

The researchers used a posterior predictive distribution (PPD) to quantify the compatibility of different H₀ measurements. The distribution gives the probability of measuring a particular value of H₀ after taking into account existing data and model parameters. If, once taken, those measurements agree with the predicted distribution, then all is well with the model; if not, either the underlying physics or the measurements are incomplete. “In order to figure out whether two measurements are in agreement, you need a robust tool that incorporates all of the physical and instrumental effects you believe are in both datasets,” says Feeney. “That’s what we feel the PPD does very well.”

The solid lines in the graph above show PPDs generated using the simulated mergers and assuming that H₀ = 68 km s⁻¹ Mpc⁻¹, whereas the dashed lines assume H₀ = 73 km s⁻¹ Mpc⁻¹. Predictions for values measured using the distance ladder are shown in orange and using the CMB-cosmological model in blue. Vertical lines indicate the current values from each technique. The PPDs yield two distinct predictions for H₀, so once the simulated data are replaced by observations from stellar mergers, the real value should become clear. LIGO and Virgo are predicted to detect the approximately 50 events needed to make that distinction in the next decade. (S. M. Feeney et al., Phys. Rev. Lett. 122, 061105, 2019. )