The Perfect Theory (26 page)

In 1974, two American astrophysicists, Joe Taylor and Russell Hulse, discovered not one but two neutron stars orbiting each other in a very tight orbit. One of the neutron stars was a pulsar, emitting bursts of light every few thousandths of a second, and could be easily followed as it orbited its silent companion. As these neutron stars orbited each other, Taylor and Hulse could measure their positions incredibly accurately. They had discovered a new, perfect laboratory for general relativity. Einstein had claimed that two objects orbiting each other would lose energy to the surrounding spacetime and their orbits would shrink until, ultimately, they would fall into each other. Although he abandoned this claim in later life, the calculation was there, ready to be tested. And the Hulse and Taylor millisecond pulsar could be used for exactly that.

In 1978, at the ninth Texas Symposium, held in Munich, Joe Taylor announced a new result. Having followed the millisecond pulsar for four years, he could safely say that the orbit was shrinking and that it was doing so as Einstein had predicted. As the two neutron stars orbited each other, they were losing energy through gravitational radiation. The evidence for gravitational radiation was indirect, but it was definitely there. It agreed beautifully with theory, and the measurements were clean and unambiguous. Gravitational waves were real.

Out of the debris of Weber's detection, a new field of experimental science was emerging. Different groups throughout the world were building their own detectors. Some were tweaking Weber's original design, cooling the cylinders dramatically so that they wouldn't vibrate at room temperature. Others changed the shape of the receptors, building spheres that would be sensitive to waves coming from all directions. But the signals they were looking for were so minute and so elusive that a bigger and better receptor was needed, one that would have the obscene sensitivity required to pick up ripples in spacetime. There was one approach that stood out from the others, hugely more powerful but also hugely more expensive: laser interferometry.

A laser interferometer makes use of the best tools of modern physics. For a start, it uses a laser beam, an incredibly focused ray of light that has been amplified and focused onto a very tiny target. Properly done, you can shine a laser for miles and it will hit its target, lighting up a pencil tip. Joe Weber had actually been one of the first people to come up with the concept of the laser, in his life before gravitational waves. He did so at the same time as Charles Townes at Columbia University but was never fully credited for his contribution and wasn't one of the awardees honored with the Nobel Prize in 1964 for its discovery.

Laser interferometry also makes use of another property of light, the fact that it behaves like a wave. Imagine waves in the ocean. When two waves with the exact same wavelength meet, they interfere. This means that if the waves meet when they are both at a crest, they add up constructively, and the resulting wave will have a much higher crest (and much deeper trough). But if they meet and one of them is at a crest and the other one at a trough, they will cancel each other out and interfere destructively. There is, of course, a whole range of behaviors between these two extremes.

These two properties of laser light can be used to detect minute motions of objects that have been affected by gravitational waves. The instruction manual is as follows: Suspend two masses at a distance from each other and shine a laser beam onto each of them. Each of the beams will reflect off the masses and interfere with the other; the resulting interference pattern will depend on the wavelength and the exact distance traveled. If one of the masses shifts ever so slightly, the interference pattern will shudder and change. By monitoring the movement in the interference pattern, it should be possible to detect the microscopic motions induced by gravitational waves. And it should be possible to do so with far more precision and accuracy than with Weber's bars.

Laser interferometry involved a completely new way of doing science, at least for relativists. Relativity had been a pencil-and-paper operation, with experiments few and far between. There were some laboratory setups and a few sparse collaborations between universities and institutions. It wasn't like particle and nuclear physics with the huge accelerators and reactors. But now, a new culture was necessary, one that could support spending tens or even hundreds of millions of dollars to construct experiments. Instead of teams with a handful of people, large organizations with hundreds of scientists and technicians would be required.

 

This time, it had to be done properly. This time, they had to know what they were looking for. It was clear that the gravitational waves had to come from something that pushed the theory to its limits. Hulse and Taylor's millisecond pulsars appeared quite benign, just two very compact stars orbiting each other. Yet they seemed to be able to spew out waves, enough of them to visibly suck out energy from their orbits. Neutron stars were stars almost on the brink of implosion that warped space and time enough to bring out the full glory of Einstein's theory.

One possible source of copious gravitational waves might be a supernova. Supernovae are imploding stars that for a few seconds shine more brightly than the billions of stars in our galaxy put together before becoming neutron stars or black holes. At any given time, a supernova is the brightest thing in the sky. Just as supernovae are a strong source of electromagnetic waves, astrophysicists speculated that they might be energetic enough to gnarl and shake spacetime into action, sending out a burst of gravitational waves. In 1987, a supernova went off in the nearby Large Magellanic Cloud, about 160,000 light-years away, and was observed in its full glory through normal telescopes. To everyone's embarrassment, not a single bar or other form of detector was running at the time to attempt to pick up the gravitational waves, except for Joe Weber's. Unsurprisingly, he claimed to have seen something, and as had become the custom, he was ignored.

The problem with supernovae is that they are too unpredictable, and while these huge explosions might indeed send a burst of energy, by the time a supernova's gravitational waves reached a detector on Earth they would be a mere blip. They could be confused with any other spurious bit of noise that might make its way to the instrument. No, what was needed was a clean signal that, even though it might be faint, would have a definite, perfectly known shape and form, like looking for a familiar face in a crowd.

There was something out there that might just do the job. The gravitational wave signal from the orbiting neutron stars Hulse and Taylor had observed could, in principle, be calculated with enough accuracy to be of use. Unlike the mess of waves coming out of a cosmic explosion, the gravitational wave signal should be regular and periodic, like a siren, and it would slowly change with time as the neutron stars lost energy and approached each other. The signal was simple, easy to describe, maybe even easy to detect.

But why stop there? Why not go for the big prize? A neutron star orbiting and plunging into a black hole would give a far stronger signal, and, of course, a binary system made up of two black holes would bring out the warpedness of Einstein's space and time in all its glory. Two black holes orbiting each other would send out a regular hum of gravitational waves. As they got closer and closer to each other, the pitch of the hum would get higher and higher until, just as they would be about to merge, they would send out a chirp and then a burst of waves that would evanesce as the black holes collapsed into one. This waveform is what the instruments would look for: the inspiral, the chirp, and then the ringdown. These relativistic binaries were like gems buried in the firmament. And the gravitational wave detectors would find them.

While it seemed straightforward—just look for the inspiraling neutron stars and black holes—a crucial piece of information was missing. What would the gravitational wave detector actually
see?
How exactly would the inspiral, chirp, and ringdown look once they arrived at the instrument? The observers, the new breed of gravitational wave astronomers, would need to know what kind of signal to expect, not roughly but exactly, if they were to be able to pull it out of the mess of noise that invariably polluted the data. And to be able to get a precise, exact answer to these questions it would be necessary to go back to the age-old problem of solving Einstein's field equations, this time to find precise mathematical solutions to describe what the gravitational waves would look like. Decades of experience showed that Einstein's equations seemed to turn and bite whoever tried to tame them. The only way forward was to solve the equations on a powerful computer and see what would happen when two black holes circled each other and ultimately collided.

Charles Misner, one of John Wheeler's students and collaborators, had already warned of the equations' treachery at the Chapel Hill meeting in 1957. You had to be careful trying to solve the gnarled, nonlinear beasts that Einstein had bequeathed, for, as Misner put it, there were only two possible outcomes:
“either the programmer will shoot himself, or the machine will blow up.” And the latter is exactly what had happened. In 1964, when one of Wheeler's ex-students, Robert Lindquist, tried to run the model, the program blew up. As the black holes got closer and closer to each other, the errors in the solutions became larger, and very quickly the computer was spewing out garbage—numerical diarrhea. The errors were so intractable that Lindquist gave up.

In the 1970s it was Bryce DeWitt's turn to try to find what would happen when two black holes collided on a computer. While his passion had always been for quantum gravity, he had learned how to simulate complicated equations on the computer during his work on the bomb project with Edward Teller at Lawrence Livermore National Laboratory in California. At Texas, he set one of his students, Larry Smarr, the task of working out how much gravitational radiation would be emitted if two black holes collided. They ran their code on the big University of Texas computer and were able to get a rough guess of what the gravitational waves would look like. And then the errors would blow up and garbage would come out. It was a glimpse of the waveform, but too rough to be useful. The singularities of spacetime would rear their ugly heads and kill the result.

For the next three decades, teams of programmers would work on trying to simulate the binaries and fail to get it right. Their work was advancing, but as Frans Pretorius, a relativist based at Princeton, recalls, “Naive things weren't working, no one exactly knew why, and people were sort of flailing about in the dark. And what made the problem so insidious was the computational expense of the full problem.” In the 1990s the black hole collision problem was even considered one of the grand challenges of computational physics in the United States, with millions of dollars given to groups all across the country to buy supercomputers and run their programs. Every now and then there would be an improvement, and the results could progress a little bit further before the errors crept in. It became a field in its own right—numerical relativity.

Solving the equations for
colliding black holes was difficult, unforgiving, as hard as detecting gravitational waves themselves and emblematic of Einstein's field equations. Young relativists would be sucked into trying to solve Einstein's equations on the computer and would spend their—often short—careers getting a small improvement on what had been done before. It was like playing an incredibly elaborate computer game, often on one's own, with no intermediate rewards, no passed levels, and no epic wins.

For some, general relativity came to mean numerical relativity. A general relativity group would not be complete without one or more relativists trying to solve the problem of black hole collisions on the computer with an eye on gravitational waves. There were conferences and meetings on the problem where everyone could get together to show off their new tricks and their plots and graphs. But the equations wouldn't give in. And with the waveforms that would come out of their simulations of binaries, there wasn't a hope in hell of finding them with the detectors.

Looking back at those dark times, Pretorius says,
“There was a serious possibility that this was sufficiently difficult that it wouldn't be solved to a degree by the time [the gravitational wave detector] came online.” The data could very easily precede a useful prediction of what the computer simulations might reveal.

Yet there was another side to the battle for numerical relativity that would have a surprising impact on the wider world. Through the late 1970s and early 1980s, Larry Smarr developed ever more elaborate numerical codes that he would attempt to run on the largest computers to which he could gain access. Based in the United States, Smarr found that he was doing many of his numerical runs in Germany, and his frustration grew at being unable to run his computer codes back home. By the mid-1980s, Smarr had successfully convinced the US government to fund a network of supercomputer centers to service all branches of science in need of “data crunching.” Smarr would end up directing one of these new centers, the National Center for Supercomputing Applications in Illinois, and it was his research group that in the 1990s came up with the first graphical Web browser, Mosaic, that allowed them to visualize remote data over the Internet. And so, in the midst of the battle to conquer black holes, it was numerical relativity that gave birth to the Web-browsing culture that is such an integral part of our lives today.

 

While the numerical relativists flailed around, the plan to build an effective gravitational wave instrument was under way. This time, there could be no false discoveries exceeding the instrument's capabilities—the era of Weber was past. The interferometer was the method of choice, but the requirements for such a device were extreme. The laser light would have to travel far enough that a tiny deflection of the masses due to gravitational waves would be detectable in the interference pattern. Even with an interferometer that was kilometers long, the laser light would have to bounce back and forth, reflecting off mirrors tied to the masses, over a hundred times. The mirrors had to be perfect and perfectly aligned. And still the deflection would be tiny. A burst of gravitational waves coming from an inspiraling binary would lead to a deflection a minute fraction of the width of a proton.