Detect the nonlinear gravitational-wave memory of black hole mergers
In plain words
General relativity predicts that a passing gravitational wave leaves detector mirrors permanently displaced, the memory effect, because gravitational waves themselves act as a source of gravity. The effect is small and has not been measured.
Precise statement
The nonlinear (Christodoulou) memory produces a permanent step in the strain, mainly in the $(\ell,m) = (2,0)$ mode, with amplitude a small fraction of the peak oscillatory strain for comparable-mass binaries. Determine whether stacking binary black hole events in ground detectors, or single loud events in LISA, detects the memory at more than $3\sigma$ with the GR amplitude; Lasky et al. (2016) estimated 35 to 90 GW150914-like events for Advanced LIGO. An answer is a detection with the measured amplitude relative to the GR prediction, or an upper limit inconsistent with it.
What would settle it
A stacked memory search over a catalog of binary black hole events reaching $3\sigma$ sensitivity to the GR amplitude.
Status in the literature
Unverified note
A 2026 catalog analysis reports a cumulative log10 Bayes factor of $1.38\ \text{plus or minus}\ 0.79$ in favor of memory, short of detection (arXiv:2607.04909); hierarchical inference finds the memory amplitude consistent with GR but unconstrained (arXiv:2605.27500, 2026).