Rob Sheldon on lowering the standard for detecting gravitational waves

Recently, an article in Nature advised that:

Gravitational-wave observatories have released their latest catalogue of cosmic collisions, bringing their total number of detections to 90. The new crop of 35 events includes one featuring the lightest neutron star ever seen, as well as two clashes involving surprisingly large black holes…

Gravitational waves are ripples in the fabric of space-time that are produced when large masses accelerate. Like the detections previously reported by LIGO–Virgo, the latest ones are all attributed to pairs of dense stellar remnants spiralling into each other and merging. The vast majority, including LIGO’s first historic detection in 2015, have involved pairs of black holes, but in a few cases one or both of the objects were neutron stars.

The collaboration initially released data on only high-confidence detections, but the latest catalogue — as well as the previous one, released in October 2020 — includes any detections that have better-than-even chances of being genuine gravitational waves. The team estimates that around 10–15% of the latest candidates in the catalogue are false alarms, “caused by instrumental noise fluctuations”.

Davide Castelvecchi, “Astrophysicists unveil glut of gravitational-wave detections” at Nature (09 November 2021)

Experimental physicist Rob Sheldon comments,

We are several years into the LIGO experiment, now we have VIRGO in Italy online and soon KAGRA in Japan. The [open access] paper mentioned above has some 1570 authors. It takes 10 pages to list the authors.

In the early days of LIGO, they announced that they would provide public alerts of a merger event, so that astronomers could point their telescopes and see the afterglow. It was expected that any event energetic enough to make gravity waves would make intense lightshow as well. Much data has been collected from the afterglows associated with Gammaray Bursts (GRBs) and it was thought the same would be true of gravity mergers. For that purpose, LIGO was point to a patch in the sky and say “That’s where this gravity wave came from, look there.”

Again, in the early days, LIGO gave really big patches of sky and apologized that with only two detectors they could not localize the direction too well, but as soon as they had a third detector, as soon as VIRGO came online, we would have pinpoint accuracy for the astronomers.

Five years have gone by, what do 1570 scientists have to say?

“These public alerts enable the astronomy community to search for multimessenger counterparts to potential GW signals. There were 39 low-latency candidates reported during O3b. Of these, 18 … survive our detailed analyses to be included as potential CBC signals in GWTC-3. Additionally, GWTC-3 includes 17 candidates with p astro > 0.5 that have not been previously presented. No confident multimessenger counterparts have currently been reported from the O3b candidates.”

Surely you must be joking. Nothing? What about localizing those patches in the sky?

“the observing time includes periods when at least two detectors were observing, and the Euclidean sensitive volume is the volume of a sphere with a radius equal to the BNS inspiral range of the second most sensitive detector in the network.”

What they just said, was they don’t do localization, they do magnitude only. Large signals happened in close, weak ones further out. And they do it with single detectors, occasionally dual, but never three. Why?

Because they don’t have triple coincidences.

And that should end the whole program right there.

Individually, the detectors have a 10,000:1 Noise to Signal Ratio. They use “matched filters” to discover waves, a method invented by the radar community who say that when the noise is twice greater than the signal the method doesn’t work. But physicists say “Bosh! we can find signals so small your radar could see a candy wrapper 10,000 miles away. You engineers are just not as creative as physicists.”

Well how do you know its real? After all, the radar guys make their own signal, but you can’t, so how do you know its not something else?

“We use coincidences. We reject a signal unless both Hanford and Louisiana see something at the same time. “

How do you know some event isn’t correlating the noise at both locations?

“We’ll use a third coincidence, we’ll use VIRGO. With three coincidences, the noise just vanishes.”

Well, VIRGO has been online for at least 2 years, show me your triple coincidences.

“Harrum. They aren’t there. But that’s okay–we can get the same results by using magnitudes and, well, we really didn’t need arrival locations anyway, the astronomers can use all-sky cameras. Clumsier, but workable.”

Why aren’t they there? I thought you said the whole secret to getting 10,000:1 noise removed was coincidences?

We don’t know, but we have the Nobel Prize and now we’re getting almost two hits a week. The data set is growing, and lots of papers are getting written, and 1570 scientists are putting these publications on their resume. We can’t let a little thing like missing triple coincidences stop us. We’ll model it as if they were there, that allows us to proceed with our analysis. And someday we’ll solve that mystery, but we really haven’t the time to stop and solve every problem right now.

Oh, and we’ll stop mentioning direction of gravity waves in our announcements. We’ll pretend the magnitude is all we have.”

So to summarize, the absence of triple coincidences is being withheld from the paper, when in fact, it delegitimizes the entire data analysis pipeline. Now we have 4 Gravity wave detectors, and soon one in space. At what point does the lack of a triple coincidence become fatal? What observation can they make that would disprove the existence of gravity waves?

Maybe they can’t stop until they have potted together evidence for a multiverse.

Note: In media work, we say: It takes three to make a trend.

Rob Sheldon is the author of Genesis: The Long Ascent and The Long Ascent, Volume II .