LIGO's detectors
© Enrico Sacchetti
LIGO's detectors
The news we had finally found ripples in space-time reverberated around the world in 2015. Now it seems they might have been an illusion

THERE was never much doubt that we would observe gravitational waves sooner or later. This rhythmic squeezing and stretching of space and time is a natural consequence of one of science's most well-established theories, Einstein's general relativity. So when we built a machine capable of observing the waves, it seemed that it would be only a matter of time before a detection.

In point of fact, it took two days. The Laser Interferometer Gravitational-Wave Observatory collaboration, better known as LIGO, switched on its upgraded detectors on 12 September 2015. Within 48 hours, it had made its first detection. It took a few months before the researchers were confident enough in the signal to announce a discovery. Headlines around the world soon heralded one of the greatest scientific breakthroughs of the past century. In 2017, a Nobel prize followed. Five other waves have since been spotted.

Or have they? That's the question asked by a group of physicists who have done their own analysis of the data. "We believe that LIGO has failed to make a convincing case for the detection of any gravitational wave event," says Andrew Jackson, the group's spokesperson. According to them, the breakthrough was nothing of the sort: it was all an illusion.

The big news of that first sighting broke on 11 February 2016. In a press conference, senior members of the collaboration announced that their detectors had picked up the signature of gravitational waves emitted as a pair of distant black holes spun into one another.

The misgivings of Jackson's group, based at the Niels Bohr Institute in Copenhagen, Denmark, began with this press conference. The researchers were surprised at the confident language with which the discovery was proclaimed and decided to inspect things more closely.

Their claims are not vexatious, nor do they come from ill-informed troublemakers. Although the researchers don't work on gravitational waves, they have expertise in signal analysis, and experience of working with large data sets such as the cosmic microwave background radiation, the afterglow of the big bang that is spread in a fine pattern across the sky. "These guys are credible scientists," says Duncan Brown at Syracuse University in New York, a gravitational wave expert who recently left the LIGO collaboration.

gravity wave conference
© SAUL LOEB/AFP/Getty Images
The first gravitational wave discovery was announced to the world on 11 February 2016
Gravitational waves are triggered by the collision of massive objects such as black holes or neutron stars. They travel for billions of years, alternately squeezing and stretching the space-time in their path. Spreading out in all directions, they get weaker as they go, but they can be detected on Earth with a sufficiently sensitive instrument.

The LIGO collaboration built two such instruments, the Hanford detector in Washington state and the Livingston detector in Louisiana. A third, independent instrument called Virgo, located near Pisa, Italy, joined the others in 2017. These "interferometers" shoot lasers down two long tunnels, then reflect them back in such a way that the pulses should arrive at the same time. Passing gravitational waves will distort space-time, making one tunnel longer than the other, and throwing off the synchronisation.

By the time the waves wash over Earth, they are extremely weak, and the sort of change in tunnel length we expect is equivalent to about a thousandth of the diameter of a proton. That is far smaller than the disturbances that come from background seismic tremors and even the natural thermal vibrations of the detector hardware. Noise is a huge problem in gravitational wave detections.

Hence why there are detectors in different places. We know that gravitational waves travel at the speed of light, so any signal is only legitimate if it appears in all the detectors at the right time interval. Subtract that common signal, and what is left is residual noise unique to each detector at any moment, because its seismic vibrations and so on constantly vary.

This is LIGO's main ploy for extracting a gravitational wave signal from the noise. But when Jackson and his team looked at the data from the first detection, their doubts grew. At first, Jackson printed out graphs of the two raw signals and held them to a window, one on top of the other. He thought there was some correlation between the two. He and his team later got hold of the underlying data the LIGO researchers had published and did a calculation. They checked and checked again. But still they found that the residual noise in the Hanford and Livingston detectors had characteristics in common. "We came to a conclusion that was very disturbing," says Jackson. "They didn't separate signal from noise."

The Danish team wrote up their research and posted it online. After receiving no response from the LIGO collaboration, they submitted it to the Journal of Cosmology and Astroparticle Physics. The journal's editor, Viatcheslav Mukhanov of the Ludwig Maximilian University in Munich, Germany, is a world-renowned cosmologist. The editorial and advisory boards include top physicists such as Martin Rees from the University of Cambridge, Joanna Dunkley at the University of Oxford and Andrei Linde of Stanford University in California.
gravity wave
Mukhanov sent the paper for review by suitably qualified experts. Reviewers' identities are routinely kept secret so they can comment freely on manuscripts, but these were people with a "high reputation", says Mukhanov. "Nobody was able to point out a concrete mistake in the Danish analysis," he says. "There is no mistake."

A storm in a teacup, still? General relativity is one of our most well-verified theories, after all, so there is every reason to think its prediction of gravitational waves is correct. We know LIGO should be sensitive enough to detect them. The instruments are finding the waves at exactly the right rate predicted by theory. So why worry about this noise?

Seek and ye shall find

There's a simple answer to that question. Physicists have made mistakes before, mistakes that have been exposed only by paying close attention to experimental noise (see "Embarrassing noises").

The first step to resolving the gravitational wave dispute is to ask how LIGO's researchers know what to look for. The way they excavate signal from noise is to calculate what a signal should look like, then subtract it from the detected data. If the result looks like pure, residual noise, they mark it as a detection.

Working out what a signal should look like involves solving Einstein's equations of general relativity, which tell us how gravitational forces deform space-time. Or at least it would if we could do the maths. "We are unable to solve Einstein's equations exactly for the case of two black holes merging," says Neil Cornish at Montana State University, a senior figure among LIGO's data analysts. Instead, the analysts use several methods to approximate the signals they expect to see.

The first, known as the numerical method, involves cutting up space-time into chunks. Instead of solving the equations for a continuous blob of space, you solve them for a limited number of pieces. This is easier but still requires huge computing power, meaning it can't be done for every possible source of gravitational waves.

A more general approach, known as the analytic method, uses an approximation of Einstein's equations to produce templates for gravitational wave signals that would be created by various sources, such as black holes with different masses. These take a fraction of a second to compute, but aren't accurate enough to model the final merger of two black holes. This endgame is modelled in an add-on calculation in which researchers tweak the parameters to fit the results of the initial analytic solution.

LIGO location
© David Ryder/Bloomberg via Getty Images
To spy gravitational waves, LIGO's detectors need a quiet environment
This use of precalculated templates is a problem, Cornish concedes. "With a template search, you can only ever find what you're looking for." What's more, there are some templates, such as those representing the waves created by certain types of supernovae explosions, that LIGO researchers can't create.

That's why Cornish prefers the third method, which he helped develop. It involves building a model from what he calls wavelets. These are like tiny parts of a wave signal that can be assembled in various ways. You vary the number and shape of the parts until you find a combination that removes the signal from the noise. Because wavelet analysis makes no assumptions about what created the gravitational wave, it can make the most profound discoveries. The wavelets "allow us to detect the unknown unknowns", says Cornish. The downside is that they tell us nothing about the physical attributes of the detected source. For that, we have to compare the constructed signal against the templates or the numerical analysis.

The challenge with all three methods is that accurately removing the signal from the data requires you to know when to stop. In other words, you have to understand what the residual noise should look like. That is exceedingly tricky. You can forget running the detector in the absence of gravitational waves to get a background reading. The noise changes so much that there is no reliable background. Instead, LIGO relies on characterising the noise in the detectors, so they know what it should look like at any given time. "A lot of what we do is modelling and studying the noise," says Cornish.

Jackson is suspicious of LIGO's noise analysis. One of the problems is that there is no independent check on the collaboration's results. That wasn't so with the other standout physics discovery of recent years, the Higgs boson. The particle's existence was confirmed by analysing multiple, well-controlled particle collisions in two different detectors at CERN near Geneva, Switzerland. Both detector teams kept their results from each other until the analysis was complete.

By contrast, LIGO must work with single, uncontrollable, unrepeatable events. Although there are three detectors, they work almost as one instrument. And despite there being four data-analysis teams, they cannot work entirely separately, because part of the detection process involves checking that all the instruments saw the signal. It creates a situation in which each positive observation is an uncheckable conclusion. Outsiders have to trust that LIGO is doing its job properly.

Purely illustrative

And there are legitimate questions about that trust. New Scientist has learned, for instance, that the collaboration decided to publish data plots that were not derived from actual analysis. The paper on the first detection in Physical Review Letters used a data plot that was more "illustrative" than precise", says Cornish. Some of the results presented in that paper were not found using analysis algorithms, but were done "by eye".

Brown, part of the LIGO collaboration at the time, explains this as an attempt to provide a visual aid. "It was hand-tuned for pedagogical purposes." He says he regrets that the figure wasn't labelled to point this out.

This presentation of "hand-tuned" data in a peer-reviewed, scientific report like this is certainly unusual. New Scientist asked the editor who handled the paper, Robert Garisto, whether he was aware that the published data plots weren't derived directly from LIGO's data, but were "pedagogical" and done "by eye", and whether the journal generally accepts illustrative figures. Garisto declined to comment.

There were also questionable shortcuts in the data LIGO released for public use. The collaboration approximated the subtraction of the Livingston signal from the Hanford one, leaving correlations in the data - the very correlations Jackson noticed. There is now a note on the data release web page stating that the publicly available waveform "was not tuned to precisely remove the signal".

Whatever the shortcomings of the reporting and data release, Cornish insists that the actual analysis was done with processing tools that took years to develop and significant computing power to implement - and it worked perfectly.

However, anyone outside the collaboration has to take his word for that. "It's problematic: there's not enough data to do the analysis independently," says Jackson. "It looks like they're being open, without being open at all."

Brown agrees there is a problem. "LIGO has taken great strides, and are moving towards open data and reproducible science," he says. "But I don't think they're quite there yet."

The Danish group's independent checks, published in three peer-reviewed papers, found there was little evidence for the presence of gravitational waves in the September 2015 signal. On a scale from certain at 1 to definitely not there at 0, Jackson says the analysis puts the probability of the first detection being from an event involving black holes with the properties claimed by LIGO at 0.000004. That is roughly the same as the odds that your eventual cause of death will be a comet or asteroid strike - or, as Jackson puts it,"consistent with zero". The probability of the signal being due to a merger of any sort of black holes is not huge either. Jackson and his colleagues calculate it as 0.008.

Simultaneous signal

There is other evidence to suggest that at least one of the later detections came from a gravitational wave. On 17 August 2017, the orbiting Fermi telescope saw a burst of electromagnetic radiation at the same time as the LIGO and Virgo detectors picked up a signal. Analysis of all the evidence suggests that both signals came from the brutal collision of two neutron stars.

The double whammy makes LIGO's detection seem unequivocal. Even here, though, the Danish group is dissenting. They point out that the collaboration initially registered the event as a false alarm because it coincided with what's known as a "glitch". The detectors are plagued by these short, inexplicable bursts of noise, sometimes several every hour. They seem to be something to do with the hardware with which the interferometers are built, the suspension wires and seismic isolation devices. Cornish says that LIGO analysts eventually succeeded in removing the glitch and revealing the signal, but Jackson and his collaborators are again unconvinced by the methods used, and the fact there is no way to check them.

What are we to make of all this? Nothing, apparently. "The Danish analysis is just wrong," insists Cornish. "There were very basic mistakes." Those "mistakes" boil down to decisions about how best to analyse the raw data (see "How to catch a wave").

Not everyone agrees the Danish choices were wrong. "I think their paper is a good one and it's a shame that some of the LIGO team have been so churlish in response," says Peter Coles, a cosmologist at Maynooth University in Ireland. Mukhanov concurs. "Right now, this is not the Danish group's responsibility. The ball is in LIGO's court," he says. "There are questions that should be answered."

Brown thinks the Danish group's analysis is wrong, but worth engaging with. And Cornish admits the scrutiny may not be a bad thing. He and his colleagues plan to put out a paper describing the detailed properties of the LIGO noise. "It's the kind of paper we didn't really want to write because it's boring and we've got more exciting things to do." But, he adds, it is important, and increased scrutiny and criticism may in the end be no bad thing. "You do have to understand your noise."

Comment: Providing valid proof for apparently ground breaking claims is 'boring'?

Coles himself doesn't doubt that we have detected gravitational waves, but agrees with Jackson that this cannot be confirmed until independent scientists can check the raw data and the analysis tools. "In the spirit of open science, I think LIGO should release everything needed to reproduce their results."

Jackson is unconvinced that explanatory papers will ever materialise - the collaboration has promised them before, he says. "This LIGO episode continues to be the most shocking professional experience of my 55 years as a physicist," he says. Not everyone would agree - but for a discovery of this magnitude, trust is everything.
Embarrassing noises

In 2014, the operators of the BICEP2 telescope made an announcement so momentous there was talk of a Nobel prize. A year later however, far from making their way to Stockholm for the award ceremony, they were forced to admit they had been fooled by an embarrassing noise.

Situated at the South Pole, BICEP2 had been scanning the cosmic microwave background, the pattern of radiation left on the sky from light emitted soon after the big bang. The big announcement was that it had found that gravitational waves had affected the pattern in such a way that proved a core theory of cosmology. The theory in question was inflation, which says the universe went through a period of superfast growth right after the big bang. For almost four decades it had been unproven. Now, suddenly, inflation's supporters were vindicated.

Except awkward warnings emerged within weeks, suggesting that cosmic dust clouds had scattered the radiation in a way that fooled the BICEP2 researchers. In the end, the team's estimate of the amount of dust present and the analysis of the kind of noise the dust would produce both proved to be flawed. Noise can hoodwink even the smartest. That is why, despite LIGO being a highly respected collaboration, there is good reason to take questions about its noise analysis seriously (see main story).
How to catch a wave

Output from gravitational wave detectors is full of noise. Disentangling the signal requires decision-making - and poor ones could be disastrously misleading.

The best weapon in the arsenal is known as a Fourier transform. This splits a signal into various frequency components and converts it into a power spectrum, which details how much of the signal's power is contained in each of those components. This can be done with a window function, a mathematical tool that operates on a selected part of the data. Whether or not to use one is at the heart of the disagreement over LIGO's results (see main story).

Andrew Jackson's dissenting team at the Niels Bohr Institute in Denmark chose not to use a window function, a decision that LIGO's Neil Cornish describes as a "basic mistake". Jackson says they didn't use one because it subtly alters the Fourier-transformed data in a way that can skew the results of subsequent processing.

Even with the Fourier analysis done, judgements must be made about the noise in the detectors. Is it, for example, distributed in a predictable pattern equivalent to the bell-shaped Gaussian distribution? And does it vary over time or is it "stationary"? The appropriate techniques for processing the data are different depending on the answers to these questions, so reliably detecting gravitational waves depends on making the right assumptions. Jackson's group says the decisions made during the LIGO analysis are opaque at best, and probably wrong.
This article appeared in print under the headline "Wave goodbye?"