Measuring gravitational-wave memory in the first LIGO/Virgo gravitational-wave transient catalog

Abstract

Gravitational-wave memory, a strong-field effect of general relativity, manifests itself as a permanent displacement in spacetime. We develop a Bayesian framework to detect gravitational-wave memory with the Advanced LIGO/Virgo detector network. We apply this algorithm on the ten binary black hole mergers in LIGO/Virgo's first transient gravitational-wave catalog. We find no evidence of memory, which is consistent with expectations. In order to estimate when memory will be detected, we use the best current population estimates to construct a realistic sample of binary black hole observations for LIGO/Virgo at design sensitivity. We show that an ensemble of O(2000) binary black hole observations can be used to find definitive evidence for gravitational-wave memory. We conclude that memory is likely to be detected in the early A+/Virgo+ era.

Forecasts for detecting the gravitational-wave memory effect with Advanced LIGO and Virgo

“…Note: While this work was being completed, a preprint by Hübner et al [33] appeared that estimated the number of BBH observations required to detect the nonlinear GW memory effect in BBH populations. There were several differences in methodology between this paper and [33].…”

“…Note: While this work was being completed, a preprint by Hübner et al [33] appeared that estimated the number of BBH observations required to detect the nonlinear GW memory effect in BBH populations. There were several differences in methodology between this paper and [33]. First, [33] computed evidence ratios for signal hypotheses including and omitting the GW memory effect and the Bayes factor (BF) for the presence versus the absence of the memory effect in the population of BBHs (rather than computing the total SNR for the memory effect, as was done in this paper).…”

“…There were several differences in methodology between this paper and [33]. First, [33] computed evidence ratios for signal hypotheses including and omitting the GW memory effect and the Bayes factor (BF) for the presence versus the absence of the memory effect in the population of BBHs (rather than computing the total SNR for the memory effect, as was done in this paper). Second, they used the higher-mode model rather than the quadrupole model as the fiducial waveform model for the GW memory effect.…”

“…Second, they used the higher-mode model rather than the quadrupole model as the fiducial waveform model for the GW memory effect. Third, they use a different model for the population of BBHs: namely, model B of [32] for a specific set of parameters given in [33]. Fourth, they do not exclude events for which the sign of the GW memory effect is not well determined.…”

“…A direct comparison of our results will require additional future work. As a rough comparison, we computed the number of detections needed to reach a total memory SNR ρ tot = 3 for the same population as in [33] using their same waveform model for the GW memory effect. We find we need 1488 +725 −879 (errors are 90% confidence intervals) to reach our SNR threshold.…”