What’s up with LIGO?

The Nobel-Prize winning figure.
We don’t know exactly what it shows.
[Image Credits: LIGO]

Almost four years ago, on September 14 2015, the LIGO collaboration detected gravitational waves for the first time. In 2017, this achievement was awarded the Nobel Prize. Also in that year, the two LIGO interferometers were joined by VIRGO. Since then, a total of three detectors have been on the lookout for space-time’s subtle motions.

By now, the LIGO/VIRGO collaboration has reported dozens of gravitational wave events: black hole mergers (like the first), neutron star mergers, and black hole-neutron star mergers. But not everyone is convinced the signals are really what the collaboration claims they are.

Already in 2017, a group of physicists around Andrew Jackson in Denmark reported difficulties when they tried to reproduce the signal reconstruction of the first event. In an interview dated November last year, Jackson maintained that the only signal they have been able to reproduce is the first. About the other supposed detections he said: “We can’t see any of those events when we do a blind analysis of the data. Coming from Denmark, I am tempted to say it’s a case of the emperor’s new gravitational waves.”

For most physicists, the 170817 neutron-star merger – the strongest signal LIGO has seen so-far – erased any worries raised by the Danish group’s claims. That’s because this event came with an electromagnetic counterpart that was seen by multiple telescopes, which can demonstrate that LIGO indeed sees something of astrophysical origin and not terrestrial noise. But, as critics have pointed out correctly, the LIGO alert for this event came 40 minutes after NASA’s gamma-ray alert. For this reason, the event cannot be used as an independent confirmation of LIGO’s detection capacity. Furthermore, the interpretation of this signal as a neutron-star merger has also been criticized. And this criticism has been criticized for yet other reasons.

It further fueled the critics’ fire when Michael Brooks reported last year for New Scientist that, according to two members of the collaboration, the Nobel-prize winning figure of LIGO’s seminal detection was “not found using analysis algorithms” but partly done “by eye” and “hand-tuned for pedagogical purposes.” To this date, the journal that published the paper has refused to comment.

The LIGO collaboration has remained silent on the matter, except for issuing a statement according to which they have “full confidence” in their published results (surprise), and that we are to await further details. Glaciers are now moving faster than this collaboration.

In April this year, LIGO started the third observation run (O3) after an upgrade that increased the detection sensitivity by about 40% over the previous run.  Many physicists hoped the new observations would bring clarity with more neutron-star events that have electromagnetic counterparts, but that hasn’t happened.

Since April, the collaboration has issued 33 alerts for new events, but so-far no electromagnetic counterparts have been seen. You can check the complete list for yourself here. 9 of the 33 events have meanwhile been downgraded because they were identified as likely of terrestrial origin, and been retracted.

The number of retractions is fairly high partly because the collaboration is still coming to grips with the upgraded detector. This is new scientific territory and the researchers themselves are still learning how to best analyze and interpret the data. A further difficulty is that the alerts must go out quickly in order for telescopes to be swung around and point at the right location in the sky. This does not leave much time for careful analysis.

With the still lacking independent confirmation that LIGO sees events of astrophysical origin, critics are having a good time. In a recent article for the German online magazine Heise, Alexander Unzicker – author of a book called “The Higgs Fake” – contemplates whether the first event was a blind injection, ie, a fake signal. The three people on the blind injection team at the time say it wasn’t them, but Unzicker argues that given our lack of knowledge about the collaboration’s internal proceedings, there might well have been other people able to inject a signal. (You can find an English translation here.)

In the third observation run, the collaboration has so-far seen one high-significance binary neutron star candidate (S190425z). But the associated electromagnetic signal for this event has not been found. This may be for various reasons. For example, the analysis of the signal revealed that the event must have been far away, about 4 times farther than the 2017 neutron-star event. This means that any electromagnetic signal would have been fainter by a factor of about 16. In addition, the location in the sky was rather uncertain. So, the electromagnetic signal was plausibly hard to detect.

More recently, on August 14th, the collaboration reported a neutron-star black hole merger. Again the electromagnetic counterpart is missing. In this case they were able to locate the origin to better precision. But they still estimate the source is about 7 times farther away than the 2017 neutron-star event, meaning it would have been fainter by a factor of about 50.

Still, it is somewhat perplexing the signal wasn’t seen by any of the telescopes that looked for it. There may have been physical reasons at the source, such that the neutron-star was swallowed in one bite, in which case there wouldn’t be much emitted, or that the system was surrounded by dust, blocking the electromagnetic signal.

A second neutron star-black hole merger on August 17 was retracted. 

And then there are the “glitches”.

LIGO’s “glitches” are detector events of unknown origin whose frequency spectrum does not look like the expected gravitational wave signals. I don’t know exactly how many of those the detector suffers from, but the way they are numbered, by a date and two digits, indicates between 10 and 100 a day. LIGO uses a citizen science project, called “Gravity Spy” to identify glitches. There isn’t one type of glitch, there are many different types of them, with names like “Koi fish,” “whistle,” or “blip.” In the figures below you see a few examples.

Examples for LIGO's detector glitches. [Image Source]

This gives me some headaches, folks. If you do not know why your detector detects something that does not look like what you expect, how can you trust it in the cases where it does see what you expect?

Here is what Andrew Jackson had to say on the matter:

Jackson: “The thing you can conclude if you use a template analysis is [...] that the results are consistent with a black hole merger. But in order to make the stronger statement that it really and truly is a black hole merger you have to rule out anything else that it could be.

“And the characteristic signal here is actually pretty generic. What do they find? They find something where the amplitude increases, where the frequency increases, and then everything dies down eventually. And that describes just about every catastrophic event you can imagine. You see, increasing amplitude, increasing frequency, and then it settles into some new state. So they really were obliged to rule out every terrestrial effects, including seismic effects, and the fact that there was an enormous lightning string in Burkina Faso at exactly the same time [...]”
Interviewer: “Do you think that they failed to rule out all these other possibilities?

Jackson: “Yes…”

If it was correct what Jackson said, this would be highly problematic indeed. But I have not been able to think of any other event that looks remotely like a gravitational wave signal, even leaving aside the detector correlations. Unlike what Jackson states, a typical catastrophic event does not have a frequency increase followed by a ring-down and sudden near-silence.

Think of an earthquake for example. For the most part, earthquakes happen when stresses exceed a critical threshold. The signal don’t have a frequency build-up, and after the quake, there’s a lot of rumbling, often followed by smaller quakes. Just look at the below figure that shows the surface movement of a typical seismic event.

Example of typical earthquake signal. [Image Source]

It looks nothing like that of a gravitational wave signal.

For this reason, I don’t share Jackson’s doubts over the origin of the signals that LIGO detects. However, the question whether there are any events of terrestrial origin with similar frequency characteristics arguably requires consideration beyond Sabine scratching her head for half an hour.

So, even though I do not have the same concerns as were raised by the LIGO critics, I must say that I do find it peculiar indeed there is so little discussion about this issue. A Nobel Prize was handed out, and yet we still do not have confirmation that LIGO’s signals are not of terrestrial origin. In which other discipline is it considered good scientific practice to discard unwelcome yet not understood data, like LIGO does with the glitches? Why do we still not know just exactly what was shown in the figure of the first paper? Where are the electromagnetic counterparts?

LIGO’s third observing run will continue until March 2020. It presently doesn’t look like it will bring the awaited clarity. I certainly hope that the collaboration will make somewhat more efforts to erase the doubts that still linger around their supposed detections.