Astrophysics Milestones For Pulsar Timing Array Gravitational Wave Detection

TL;DR

This paper assesses the timeline and key milestones for detecting gravitational wave backgrounds using pulsar timing arrays, based on simulations of the NANOGrav dataset and implications for future observations.

Contribution

It extends previous analyses by quantifying data collection milestones for gravitational wave detection and distinguishes between different models of the background signal.

Findings

Robust detection expected after 15-17 years of data collection.

Initial detection will constrain the GWB spectral slope within 40%.

GWB amplitude can be measured with 44% uncertainty at initial detection.

Abstract

The NANOGrav Collaboration reported strong Bayesian evidence for a common-spectrum stochastic process in its 12.5-yr pulsar timing array dataset, with median characteristic strain amplitude at periods of a year of $A_{\rm yr} = 1.92^{+0.75}_{-0.55} \times 10^{-15}$. However, evidence for the quadrupolar Hellings \& Downs interpulsar correlations, which are characteristic of gravitational wave signals, was not yet significant. We emulate and extend the NANOGrav dataset, injecting a wide range of stochastic gravitational wave background (GWB) signals that encompass a variety of amplitudes and spectral shapes, and quantify three key milestones: (I) Given the amplitude measured in the 12.5 yr analysis and assuming this signal is a GWB, we expect to accumulate robust evidence of an interpulsar-correlated GWB signal with 15--17 yrs of data, i.e., an additional 2--5 yrs from the 12.5 yr dataset; (II) At the initial detection, we expect a fractional uncertainty of $40\%$ on the power-law strain spectrum slope, which is sufficient to distinguish a GWB of supermassive black-hole binary origin from some models predicting more exotic origins;(III) Similarly, the measured GWB amplitude will have an uncertainty of $44\%$ upon initial detection, allowing us to arbitrate between some population models of supermassive black-hole binaries. In addition, power-law models are distinguishable from those having low-frequency spectral turnovers once 20~yrs of data are reached. Even though our study is based on the NANOGrav data, we also derive relations that allow for a generalization to other pulsar-timing array datasets. Most notably, by combining the data of individual arrays into the International Pulsar Timing Array, all of these milestones can be reached significantly earlier.