Recently some articles in popular media1, 2 informed that LIGO will be able to measure the memory effect of gravitational waves described by Demetrios Christodoulou in 1991.3

The measurement method based on accumulation over many LIGO measurements was proposed by Paul D. Lasky and colleagues.4

I do not understand most of the details in the paper4 but a mystery for me is how do they count with the effect of gravitational waves which are out of measurement range of LIGO (either by frequency or amplitude). I think we cannot make many assumptions about such waves. Is the paper4 reasonable?

  1. Charles Q. Choi "Gravitational Waves May Permanently Alter Spacetime"
  2. "Advanced LIGO Gravitational Wave Observatory Could Detect Permanent Memory of Spacetime"
  3. D. Christodoulou (1991). "Nonlinear nature of gravitation and gravitational-wave experiments". Phys. Rev. Lett. 67 (12): 1486–1489.
  4. Paul D. Lasky, Eric Thrane, Yuri Levin, Jonathan Blackman, Yanbei Chen (2016). "Detecting gravitational-wave memory with LIGO: implications of GW150914"
  • 3
    $\begingroup$ Thanks for the links to some fascinating reads. They are all couched in 'may', 'might' and 'coulds'. It also states clearly that it is 'early days' and that nothing concrete is being proposed. But it does outline some exciting ideas regarding space time memory. Is it reasonable? Yes. Very much so. I have put this as a comment rather than an answer as I cannot give a qualified answer, but they are suggesting this is within Ligo's design limits, to throw light on things that are potentially not. Perfectly reasonable IMHO. $\endgroup$
    – PaulD
    Commented Oct 15, 2016 at 14:01
  • $\begingroup$ I have a related question: a single LIGO interferometer can detect a passing gravitational wave and derive some properties of the source, including energy and distance. After the passage of the gravitational wave the memory remains. I am guessing that, even with better detectors, the memory alone wouldn't permit deriving the properties of the source (correct?). But suppose there are, say, three events and three detectors. Could the analysis of the three memory signals measured by the three detectors permit deriving some properties of the three sources? $\endgroup$ Commented Oct 17, 2016 at 5:05
  • $\begingroup$ (continued) I am the author of the popular media review in ref. 2 and I am trying to better understand the implications. Too bad the character count for comments is limited, I had to edit down my comment above, I hope the related question still makes sense. $\endgroup$ Commented Oct 17, 2016 at 5:08
  • 2
    $\begingroup$ Just asked related question on Physics Overflow, with link to this thread: physicsoverflow.org/37410/… $\endgroup$ Commented Oct 17, 2016 at 6:14
  • $\begingroup$ If this gravitational memory indeed a property of space time and gravitational waves, and work is done displacing mass, then doesn't that mean that energy was transferred from gravitational wave to mass? Or are we talking about a 'plastic' property, a hysteresis effect of spacetime itself - so no energy loss? $\endgroup$
    – docscience
    Commented Jul 5, 2018 at 15:46

2 Answers 2


The BMS (Bondi Metzner, Sachs) symmetry is a semi-direct product of the Poincare group of spacetime symmetry with an abelian group of translations. It is this latter part which contains gravitational memory. This paper illustrates some of this physics. What this means is that a gravitational wave interacting with a set of test masses will not return these masses to their original positions. This does mean the flat spacetime before and after a gravitational wave has passed through are not identical, but related to each other by this abelian translation of points.

The LISA will be able to detect gravitational memory. This is a triplet of satellites that have a their optics on strict geodesic paths. The spacecraft makes orbital adjustments to insure there are no forces on the actual optics. The optical configuration will change with the abelian translation of points. The LIGO by way of contrast has material stresses or strains that return the Michelson-Morely interferometer to its original configuration. This makes testing for gravitational memory, these abelian translations, more difficult. It is however, not entirely impossible. If the material response of the interferometry can be understood well enough, then there maybe characteristics of these abelian translations in the signature of a gravitational wave.

  • $\begingroup$ Re "The LISA will be able to detect gravitational memory" - Doesn't help much since LISA has been canceled ;-) Now there is the eLISA, but it won't be operational until the 2030s. elisascience.org $\endgroup$ Commented Oct 17, 2016 at 5:43
  • $\begingroup$ Why do you add a happy smiley after "since LISA has been canceled". That is sad and your reaction is rediculous! $\endgroup$ Commented Jan 9, 2017 at 17:07

The recent arXiv paper "Detection of the Permanent Strain Offset Component of Gravitational-Wave Memory in Black Hole Mergers" by Jeff Scargle, using LIGO data, claims to have detected for the first time the effect of Gravitational Wave Memory.

  • $\begingroup$ Greetings! Please prefer to link arxiv abstracts, rather than PDFs. Also please consider summarizing the content of the link, for the benefit of readers here. $\endgroup$
    – rob
    Commented Oct 30, 2021 at 23:45
  • 1
    $\begingroup$ I skimmed through this paper. Just as a warning against taking the claims at face value, there are some analysis choices that I find quite troubling. For example, in section 2.5, the author states that they choose one of 108 possible analysis choices for each individual observation to optimize the signal-to-noise ratio for that event. As I understand the text, the author does not allow the analysis choices to vary when estimating the statistical significances in Table 1, which if true would invalidate the significance estimates. Also note that the paper is not published, at least as of today. $\endgroup$
    – Andrew
    Commented Oct 31, 2021 at 5:00
  • 2
    $\begingroup$ To put this in stronger terms... the author comes very close to admitting to p-hacking his results in Section 2.5. Whether or not it actually is p-hacking depends on exactly how the significance estimates described in Section 3.4 are done, but there's not enough detail there to understand exactly what the author did. The best case scenario is that the author did not feel it was important in writing the paper to make it abundantly clear that the significance estimates on which the detection claims are based avoided p-hacking. $\endgroup$
    – Andrew
    Commented Oct 31, 2021 at 5:28

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.